Monitoring Dynamical and Structural Changes at the Lipid-Water Interface Through Chemical Shift Analysis

Permanent Link: http://ufdc.ufl.edu/UFE0041232/00001

Material Information

Title: Monitoring Dynamical and Structural Changes at the Lipid-Water Interface Through Chemical Shift Analysis A Xe-129 NMR Study
Physical Description: 1 online resource (203 p.)
Language: english
Creator: Pointer-Keenan, Caroline
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2009


Subjects / Keywords: anesthetic, binding, inverted, lipid, meyer, nonbilayer, partition, solubility, xenon
Chemistry -- Dissertations, Academic -- UF
Genre: Chemistry thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation


Abstract: The accumulation of inhalational anesthetic in lipid membranes can alter their distributive properties and energetic demands. It is hypothesized that as anesthetic molecules partition into the biomembrane, changes in the lateral pressure profile ensue, indirectly altering membrane protein structure and function. Though it is well established that this dynamic process is largely controlled by interfacial properties, the actual modifications the surface tension undergoes at this boundary is not fully understood. Here, the inhalational anesthetic xenon (Xe) is used as a nonpolar, weakly binding spin probe to investigate the anesthetic-lipid bilayer interaction by gaseous NMR spectroscopy. Fundamental kinetic and thermodynamic behaviors are studied by monitoring interaction induced chemical shift changes of thermal and hyperpolarized 129Xe in various lipidic media. The nature of xenon-phospholipid interactions and Xe exchange depend on the structure of the lipid headgroups and acyl chains, the phase state of the lipid bilayer, and the heterogeneity in both vesicle size and overall distribution of lipids with external variables. The primary focus of this work is to explore the effects of temperature and composition on solvation parameters, gas adsorption properties and molecular rearrangement at the bilayer interface via 129Xe chemical shift changes. Thermodynamic and kinetic information are extracted by monitoring the observed chemical shift with changing external variables; results are fit to a mathematical model in order to extract pertinent parameters. Partitioning behavior as related to increasing molecular stress and changing lipid morphology. Intermediate lipid phases were probed as well. Data suggests the presence of multiple binding sites as well as moderate cooperative binding. The mole fraction partition coefficient increases with temperature and behaves ideally in a single component lipid system. The presence of nonbilayer lipids has an opposite effect on the partitioning parameters.What s more, evidence suggests the addition of Xe promotes the formation of highly curved structures at elevated temperatures and pressures. Anodic Aluminum Oxide (AAO) substrates are utilized to stabilize bilayers in the magnetic field, facilitating the study of Xe diffusivity between phases using 2D-exchange NMR methods. Results are discussed in context of anesthetic action and the lateral pressure profile.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Caroline Pointer-Keenan.
Thesis: Thesis (Ph.D.)--University of Florida, 2009.
Local: Adviser: Bowers, Clifford R.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2009
System ID: UFE0041232:00001

Permanent Link: http://ufdc.ufl.edu/UFE0041232/00001

Material Information

Title: Monitoring Dynamical and Structural Changes at the Lipid-Water Interface Through Chemical Shift Analysis A Xe-129 NMR Study
Physical Description: 1 online resource (203 p.)
Language: english
Creator: Pointer-Keenan, Caroline
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2009


Subjects / Keywords: anesthetic, binding, inverted, lipid, meyer, nonbilayer, partition, solubility, xenon
Chemistry -- Dissertations, Academic -- UF
Genre: Chemistry thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation


Abstract: The accumulation of inhalational anesthetic in lipid membranes can alter their distributive properties and energetic demands. It is hypothesized that as anesthetic molecules partition into the biomembrane, changes in the lateral pressure profile ensue, indirectly altering membrane protein structure and function. Though it is well established that this dynamic process is largely controlled by interfacial properties, the actual modifications the surface tension undergoes at this boundary is not fully understood. Here, the inhalational anesthetic xenon (Xe) is used as a nonpolar, weakly binding spin probe to investigate the anesthetic-lipid bilayer interaction by gaseous NMR spectroscopy. Fundamental kinetic and thermodynamic behaviors are studied by monitoring interaction induced chemical shift changes of thermal and hyperpolarized 129Xe in various lipidic media. The nature of xenon-phospholipid interactions and Xe exchange depend on the structure of the lipid headgroups and acyl chains, the phase state of the lipid bilayer, and the heterogeneity in both vesicle size and overall distribution of lipids with external variables. The primary focus of this work is to explore the effects of temperature and composition on solvation parameters, gas adsorption properties and molecular rearrangement at the bilayer interface via 129Xe chemical shift changes. Thermodynamic and kinetic information are extracted by monitoring the observed chemical shift with changing external variables; results are fit to a mathematical model in order to extract pertinent parameters. Partitioning behavior as related to increasing molecular stress and changing lipid morphology. Intermediate lipid phases were probed as well. Data suggests the presence of multiple binding sites as well as moderate cooperative binding. The mole fraction partition coefficient increases with temperature and behaves ideally in a single component lipid system. The presence of nonbilayer lipids has an opposite effect on the partitioning parameters.What s more, evidence suggests the addition of Xe promotes the formation of highly curved structures at elevated temperatures and pressures. Anodic Aluminum Oxide (AAO) substrates are utilized to stabilize bilayers in the magnetic field, facilitating the study of Xe diffusivity between phases using 2D-exchange NMR methods. Results are discussed in context of anesthetic action and the lateral pressure profile.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Caroline Pointer-Keenan.
Thesis: Thesis (Ph.D.)--University of Florida, 2009.
Local: Adviser: Bowers, Clifford R.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2009
System ID: UFE0041232:00001

This item has the following downloads:

Full Text




2 2009 Caroline D. Pointer Keenan


3 To my father


4 ACKNOWLEDGMENTS I would like to begin by thanking my primary research advisor, Prof. Clifford R. Bowers. I am grateful for the opportunity to have worked in his laboratoryProf. Bowers enthusiasm and drive are an inspiration. It has been a real privilege working for him. I would also like to thank my secondary advisor, Prof. Gail E. Fanucci, for her assistance in getting this project off the ground both financially and conceptually. Not only did she provide the lipid samples, she graciously allowed me access to the tools required for my sample preparations. Our initial discussions with her and Chad Mair were paramount to this project. To my committee members: Prof. Vala, Prof. Vasenkov and Prof. Smith my limited interactions with each of them have been both rewarding and informative. To Dr. Ann Donnelly and SEAGEP for providing the necessary financial support needed for me to complete my dissertation. Her support enabled me to participate in conferences and meet people within the scientific community that I would not have otherwise met nor had access to. Special thanks to Marc Link in physics machine shop for his generous contributions to the many jobs I submitted over the past two years. I am also indebted to cryogenic services in the New Physics Building ( NPB): Greg Labbe and John Graham. The sheer magnitude of support and specialized knowledge they have shared since our lab moved to NPB is astounding. I am also appreciative of Joe Caruso in glass shop; he constructed our customized NMR tubes and the optica l pumping cell utilized in our experiments. His quality of work has been missed. I would also like to acknowledge the contributions of past and present group members and collaborators: Bhavin Adhyaru, Joshua Caldwell, Chi Yuan Cheng; Yuying Wei, Amrish Menjoge, Ryan Wood, and Chris Reeg. Our scientific and personal discussions have been insightful and a real pleasure. From outside the group, I would like to thank my friends that I have met while in Gainesville, whose company has kept me grounded and enrich ed my life.


5 Lastly, I thank my family (immediate and extended). Their unquestioning support, love and encouragement have helped sustain me. Special thanks to my grandparents who always stay in touch even when I seem to have disappeared off the face of the earth And to Apollon, who always encourages me to do my best I love you all.


6 TABLE OF CONTENTS page ACKNOWLEDGMENTS ...............................................................................................................4 LIST OF TABLES ...........................................................................................................................9 LIST OF FIGURES .......................................................................................................................10 A B S T R A C T ...................................................................................................................................14 C H A P T E R 1 INTRODUCTION ..................................................................................................................16 1.1 Introduction .......................................................................................................................16 1.2 Structure of Dissertation ...................................................................................................18 2 NMR FUNDAMENTALS ......................................................................................................20 2.1 Introduction .......................................................................................................................20 2.2 Basic NMR Theory ...........................................................................................................20 2.2.1 Nuclear Polarization ...............................................................................................22 2.2.2 Chemical Shift Anisotropy .....................................................................................24 2.2.3 Relaxation Mechanisms ..........................................................................................27 2.2.4 Chemical Exchange ................................................................................................29 2.2.5 129Xe 2D EXSY Exchange NMR ...........................................................................32 2.3 Introduction to 129Xe NMR ...............................................................................................37 2.3.1 The 129Xe Chemica l Shift .......................................................................................37 2.3.2 Alkali Metal Noble Gas Spin Exchange Optical Pumping (SEOP) ......................41 2.3.3 Recent Improvements to Gas Delivery System ......................................................46 3 THE LIPID ENVIRONMEN T ...............................................................................................49 3.1 Introduction .......................................................................................................................49 3.2 Lipid Structure and Assemblies ........................................................................................49 3.2.1 What Makes a Lipid? ..............................................................................................49 3.2.2 The Bilayer Phase and Self Assembly ...................................................................51 3.2.3 The Shape Concept of Lipid Polymorphism ..........................................................53 3.3 Physical and Chemical Properties of Lipid Membranes ...................................................55 3.3.1 Interfacial Tension and the Lateral Pressure Profile ..............................................55 3.3.2 Lyotropic versus Thermotropic Phase Transitions .................................................56 3.4 Experimental Methods ......................................................................................................57 3.4.1 Methods for Preparing and Characterizing Liposomes ..........................................57 3.4.2 Verification of Vesicle Size through Dynamic Light Scattering ............................59 3.4.3 Phosphate Assays of Lipid Stock Solutions ...........................................................59


7 3.5 R esults and Discussion .....................................................................................................60 3.5.1 31P NMR of Lipids ..................................................................................................60 3.5.2 Dynamic Light Scattering .......................................................................................61 3.5.3 Phosphate Assay Analysis ......................................................................................62 3.6 Conclusions .......................................................................................................................63 4 CHARACTERIZATION OF XE LIPID PARITIONING BY 129XE NMR ..........................64 4.1 Introduction .......................................................................................................................64 4.1.1 The Dissolution of 129Xe in Bio logical Media .......................................................65 4.1.2 Xenon and the Lipid Bilayer ..................................................................................66 4.2 Partition Model .................................................................................................................68 4.2.1 Membrane Buffer Partitionin g ...............................................................................68 4.2.2 Determination of Partition Coefficients by NMR ..................................................69 4.3 Results and Discussions ....................................................................................................71 4.3.1 Spectral Properties of the Xe Lipid Interaction ......................................................71 4.3.2 129Xe T1 Relaxation Rates Determined by Saturation Recovery Methods .............74 4.3.3 Pressure Dependence of Xe Partitioning ................................................................76 4.3.4 129Xe Relaxivity Utilized to Probe Partitioning and Vesicle Stability ...................82 4.3.5 Bunsen, Ostwald, and Mole Fraction Partition Coefficients ..................................87 4.4 Conclusions .......................................................................................................................93 5 THERMODYNAMIC PROPERTIES OF PARTITIONING ................................................95 5.1 Introduction .......................................................................................................................95 5.1.1 The Classic Hydrophobic Effect ............................................................................97 5.1.2 Environmental Swap Energy (ESE) .......................................................................99 5.2 The Partitioning Model as a Function of Temperature ...................................................105 5.2.1 Determining Kp( T ) by Chemical Shift Methods ...................................................105 5.2.2 Influence of Partitioning Units on Calculated Transfer Energies ........................106 5.3 Results and Discussions ..................................................................................................107 5.3.1 The vant Hoff Plot ...............................................................................................107 5.3.3 EnthalpyEntropy Compensation .........................................................................112 5.4 Conclusions .....................................................................................................................114 6 PRESSURE EFFECTS ON BINDING BEHAVIOR ...........................................................116 6.1 Introduction .....................................................................................................................116 6.1.1 Specific versus Nonspecific Binding ....................................................................117 6.1.2 Interfacial Membrane Partitioning .......................................................................118 6.1.3 Common Adsorption Types .................................................................................119 6.2 Binding Models ..............................................................................................................120 6.2.1 Adsorption Equilibrium ........................................................................................120 6.2.2 Macroscopic and Microscopic Binding Constants ...............................................124 6.3 Results and Discussion ...................................................................................................126 6.3.1 Estimation of Kinetic Parameters .........................................................................126 6.4 Conclusions .....................................................................................................................136


8 7 CHANGES IN LIPID COM POSITION: EFFECTS OF NONBILAYER LIPIDS .............138 7.1 Introduction .....................................................................................................................138 7.1.2 Nonbilayer Lipids on Biological Processes ..........................................................139 7.1.2 Energetics of the Bilayer to Inverted Hexagonal Transition ...............................142 7.2 Experimental ...................................................................................................................146 7.3 Results and Discussion ...................................................................................................148 7.3.1 Effects of Xenon Doping and Temperature ..........................................................153 7.3.2 Lamellar to Inverted Hexagonal Transition .........................................................156 7.3.3 Evidence of Kinetically Trapped Structures .........................................................163 7.4 C onclusions .....................................................................................................................165 8 LIPID NANOTUBE ARRAYS INVESTIGATED BY HYPERPOLARIZED XE 129 NMR .....................................................................................................................................167 8.1 Introduction .....................................................................................................................167 8.2 Experimental ...................................................................................................................168 8.2.1 Materials ...............................................................................................................168 8.2.2 Physical Description of AAO ...............................................................................168 8.2.3 Preparation of Lipid Nanotube Arrays .................................................................169 8.3 Results and Discussion ...................................................................................................170 8.3.1 Adsorption of Single Lipid Bilayers onto Substrates ...........................................170 8.3.2 Evidence for 129Xe inside AAO Pores ..................................................................171 8.3.3 Relaxation Rates of Dissolved Xenon ..................................................................174 8.3.4 Indication of Chemical Exchange Between Anopore and Gas Phase 129Xe ........174 8.4 Conclusions .....................................................................................................................175 9 CONCLUSIONS AND OUTL OOK ....................................................................................178 A P P E N D I X : NEWLY DESIGNED SAMPL E HOLDER FOR AAO STUDIES .....................182 LIST OF REFERENCES .............................................................................................................184 BIOGRAPHICAL SKETCH .......................................................................................................203


9 LIST OF TABLES Table page 21 Natural abundance and magnetic properties for several spin 1/2 nuclei utilized in our work; 1H and 13C are given for comparison.. ........................................................................21 22 The effects of exchange on the properties of an NMR spectrum for various exchange rates. ......................................................................................................................................31 31 Ap proximate percentage by weight of total lipid in several cell membranes. ......................50 41 Experimentally determined 129Xe relaxation times associated with the lipid free buffer solution and the lipid phase in DOPC LUVs and MLVs. .....................................................77 42 Extracted mole fraction partition coefficients obtained through a variety of methods .........86 43 The mole fraction partition coefficients of xenon in various lipidic speci es and solvents. ................................................................................................................................88 44 The dielectric constants of apolar solvents, water, and a PC lipid system. ..........................90 51 Tabulated literature values of thermodynamic parameters of transfer for 1 atm of Xe(g) to both water and hydrocarbon solutions. ...........................................................................100 52 Extracted thermodynamic parameters for the transfer of xenon from solvent water to various lipid phases and apolar solvents .............................................................................110 71 Structural and elastic properties of single component DOPC and DOPE lipids within a (1:3) DOPC/DOPE lipid mixture ........................................................................................144


10 LIST OF FIGURES Figure page 21 The placement o f the magic angle spinning rotor with respect to the external magnetic field. ....................................................................................................................27 22 Simulated NMR spectra depicting the effect of exchange and population on NMR spectra. ...............................................................................................................................30 23 The pulse s equence for 2D EXSY experiment for thermally and hyper polarized 129Xe. ..................................................................................................................................33 24 An example of 2D EXSY spectra.. ....................................................................................36 25 The energy level diagram of 87Rb showing the hyperfine structures for the D1 transition. ..........................................................................................................................42 26 Optical pumping for transitions from the ground state F = 2 level to the excited states in 87Rb .. ..............................................................................................................................44 27 A schematic illustration of the polarization of 129Xe nuclei via collision and spin exchange. ...........................................................................................................................44 28 Schematic drawing depicting the redesigned gas delivery system ....................................47 31 A schematic representation of the lateral pressure profile that exists near the water/lipid interface. ..........................................................................................................52 32 The geometric packing parameter as a function of lipid type. ...........................................54 33 The effect of 31P orientation and motion on the 31P CSA. .................................................58 34 The 31P NMR spectra of 50 mM DOPC for vari ous lipid environments.. .........................61 35 Dynamic light scattering data on a 2 mM DOPC LUV suspension.. ................................62 36 Experimentally determined v alues of phosphate assays ...................................................63 41 A series of 129Xe NMR lineshapes of xenon in MLV, LUV, and lipid free environments .....................................................................................................................72 42 129Xe NMR chemical shift and lineshape behavior of 5 atm of xeno n dissolved in 50 mM DOPC MLVs. ............................................................................................................73 43 Temperature dependence of the 129Xe chemical shift in 50mM MLVs and LUVs ...........74 44 129Xe NMR saturation recovery curves for xenon dissolve d in various solutions.. ...........75


11 45 Concentration dependence of the observed 129X e chemical shift as a function. ...............77 46 Determination of the infinite dilution mole fraction partition coefficient.. .......................79 47 The pressure dependen ce of the partition coefficient .......................................................80 48 The limiting chemical shift as a function of xenon conc entration in solution.. .................81 49 129Xe NMR spectra of xenon dissolved in 50mM DOPC MLVs, in the absence and presence of paramagnetic shift reagent. .............................................................................85 410 The variation in the NMR parameters with increas ing paramagnetic shift reagent. ........86 411 The variation in the membrane dielectric constant a s a function of membrane depth ......90 412 A graphical representation of xenonoil/lipid partition coefficients, sorted by increasing molar mass. ......................................................................................................91 51 Thermodynamic solvation parameters. ..............................................................................96 52 Graphical description of the solute transfer process. .......................................................101 53 A simple schematic depicting the differences between several types of partitioning processes. .........................................................................................................................102 54 The vant Hoff plot showing the temperature dependence of xenonmembrane partitioning for 5 atm of xenon dissolved in 50 mM of DOPC.. .....................................108 55 Graphical representation of tabulated data (Table 5 2).. .................................................109 56 The enthalpy of solute transfer as a function of temperature for 50 mM DOPC under PXe = 5 atm of overpressure. ............................................................................................109 57 Variations in the molar enthalpy and entropy values with temperature.. ........................113 61 Several examples of xenonmembrane binding models. .................................................120 62 The difference between macroscopic and mi croscopic binding constants ......................125 63 A graphic illustration of the twodimensional binding model. ........................................127 64 A Scat chard plot of the binding data. ...............................................................................130 65 A Hill plot fit to experimental data. .................................................................................132 66 Adsorption profiles of experimental data fit to several models. .....................................132 67 Langmuir isotherm an d the DArcy and Watt isotherm fit to experimental data. ..........134


12 68 Klotz affinity profi le for binding of xenon to DOPC .. ....................................................135 71 A graphical depiction of the hydrophobic mismatch and examples of common membrane deformations. ..................................................................................................141 72 Depiction of the membrane fusion intermediated according to the stalk mechanism. ....142 73 Factors that promote and inhibit the lamellar to inverted hexagonal phase transit ion. ...146 74 Schematic of the recirculation apparatus used to introduce hyper polarized 129Xe gas mixture to lipid sample.. ..................................................................................................147 75 Influence of DOPE doping on the xenonmembrane mole fraction partition coefficient. .......................................................................................................................149 76 A comparison in binding and partitioning behavior for DOPE free and DOPE containing membranes. ....................................................................................................150 77 The change in the transfer energy as a function of mole fraction of DOPE and membrane curvature. ........................................................................................................152 78 Variation in NMR parameters of DOPE containing lipids as a function of Xe overpressure. ....................................................................................................................152 79 31P NMR spectrum of the (1:1) DOPC/DOPE lipid mixture. ..........................................155 710 Temperature dependence of the observed chemical shift for various lipid compositions and pressures ..............................................................................................155 711 Variation in the observed NMR parameters of xenon dissolved in a (1:1) DOPC/DOPE with increasing temperature.. ....................................................................158 712 Potential effects solute s have on the lamellar to inverted hexagonal phase transition temperature.. ....................................................................................................................159 713 Phase diagram of various DOPC/DOPE lipid mixtures ..................................................161 714 Concentration dependence of the 129Xe chemical shift dissolved in (1:3) DOPC/DOPE MLV lipid mixture. ...................................................................................163 715 31P NMR spectra of the different lipid phases in the presence of xenon. ........................165 81 The side and top views of an AAO membrane ................................................................169 82 Various 31P NMR spectra of AAO supported DOPC bi layers ........................................171 83 31P NMR spectra of DOPC under various conditions ......................................................172 84 Illustration describing the diffusion of xenon to AAO supported bilayers. .....................172


13 85 A series of 129Xe NMR spectra of xenon dissolved/adsorbed gas in various AAO environments.. ..................................................................................................................173 86 129Xe NMR saturation recovery curves for xenon dissolved in AAO supported DOPC bilayers using continuous flow methods. .........................................................................175 87 Experimentally determined 2D 129Xe EXSY spectra. .....................................................176


14 Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy MONITORING DYNAMICAL AND STRUCTUR AL CHANGES AT THE LIPID-WATER INTERFACE THROUGH CHEMICAL SH IFT ANALYSIS: A XE-129 NMR STUDY By Caroline D. Pointer-Keenan December 2009 Chair: Clifford R. Bowers Major: Chemistry The accumulation of inhalational anesthetic in lipid membranes can alter their distributive properties and energetic demands. It is hypothesized that as anesthetic molecules partition into the biomembrane, changes in the lateral pressure profile ensue, indir ectly altering membrane protein structure and function. Though it is well es tablished that this dyna mic process is largely controlled by interfacial propertie s, the actual modifications the surface tension undergoes at this boundary is not fully understood. Here, the inhala tional anesthetic xenon (Xe) is used as a nonpolar, weakly binding spin probe to investigat e the anesthetic-lipid bilayer interaction by gaseous NMR spectroscopy. Fundamental kinetic and thermodynamic behaviors are studied by monitoring interaction induced chemical shift changes of thermal and hyperpolarized 129Xe in various lipidic media. The nature of xenon-phospholipid interactions and Xe exchange depend on the structure of the lipid headgroups and acyl ch ains, the phase state of the lipid bilayer, and the heterogeneity in both vesicle size and overall distribut ion of lipids with external variables. The primary focus of this work is to explor e the effects of temper ature and composition on solvation parameters, gas adso rption properties and molecular rearrangement at the bilayer interface via 129Xe chemical shift changes.


15 Thermodynamic and kinetic information ar e extracted by monitoring the observed chemical shift with changing external variables; results are fit to a mathematical model in order to extract pertinent parameters. Partitioning behavior as related to increasing molecular stress and changing lipid morphology. Intermediate lipid phases were probed as we ll. Data suggests the presence of multiple binding sites as well as moderate cooperative binding. The mole fraction partition coefficient increases w ith temperature and behaves idea lly in a single component lipid system. The presence of nonbilayer lipids has an opposite effect on the pa rtitioning parameters. Whats more, evidence suggests the addition of Xe promotes the formation of highly curved structures at elevated temperat ures and pressures. Anodic Alumin um Oxide (AAO) substrates are utilized to stabilize bilayers in the magnetic field, facilitating th e study of Xe diffusivity between phases using 2D-exchange NMR methods. Results ar e discussed in context of anesthetic action and the lateral pressure profile.


16 CHAPTER 1 INTRODUCTION 1.1 Introduction The membrane is the main structural elemen t of a living cell and is often viewed as a supportive substrate whose purpose is to facilitate protein function. In addition to acting as a permeability barrier, lipid membranes form the in tracellular environment and define its outer boundaries; the lack of covalent interactions within th ese assemblies provides the basis for its highly dynamic properties. Recent studies into the effect of surface tension on interfacial dynamics of a phospholipid bilayer suggest a surf ace tension mediated mechanism for membrane organization and structure. As such, lateral organization of the membrane may have consequence on protein function. In the indirect mechanism of anesthetic action, anesthetic molecules partition into the biomembrane a nd modify the lateral pressure profile. According to several modern lipid theories of anesthesia, the prefer ential location of anesth etics at the membrane interface is of prominent importance. It has be en proposed that the anesthetics intrinsic properties may mediate the redistribution of the late ral pressure profile and in this way facilitate anesthesia. Our interests lie in investigating the pot ential effect lateral heterogeneity plays in the membrane-anesthetic interaction while exploring the utility of 129Xe NMR as a probe of interfacial dynamics. 129Xe NMR has found many applications in material science and medicine within the last twenty years due to its high sensitivity to subtle changes in its local environment and ability to be hyperpolarized.[1-7] The detection of hydrophobic cavities with in proteins has prompted creative technical advancements in gas delivery and e xperimental applications and its use as a biomolecular probe of dynamical systems in solution has been recently reviewed.[1-6] 129Xe NMR spectroscopy has been employed to distinguish be tween multiple conformational states of protein


17 solutions and to monitor protein-peptide binding events in solution.[7] Whats more, it has been proven useful for probing specific and nonspecific binding to proteins in addition to detecting changes in cell morphology.[2, 8-10] The use of hyperpolarized 129Xe as a biosensor is also very appealing due to its potential to be used to detect foreign matter (e .g. tumors, lesions) in inaccessible regions in the human body.[11-13] Xenon gas is a potent anesthetic with NMR properties that ma ke it well-suited to the study of the basic nature of the anesthetic-membran e interactions. Similar to other inhalation anesthetics, it shows a high affin ity for the amphipathic region of the bilayer and is thought to interact directly with water molecules near the lipid membrane-water interface.[14] However, molecular dynamic simulations predict that a signifi cant fraction of the gas is also located within the bilayer core, a behavior comm only identified with nonimmobilizers.[15, 16] As discussed by Stimson et al., charge distribution and polari zability may be the key to understanding the distribution processes and conse quent effects of small molecules on the biomembrane. Clearly, a study of these effects is needed. Herein we study what role nonspecific, weak interactions play in membrane dynamics and how additional stress imposed by the presence of nonpolar molecules affects partitioning and binding behavior. In brief, 129Xe NMR spectroscopy is used to characterize the anestheticmembrane interaction with modified surface tensions; changes in solubility and binding parameters with temperature and membrane composition should prove useful to those developing techniques in lung imaging and bl ood perfusion, and the la rger anesthesiology community as a whole. Though rarely used due to its expense, xenon produces the lowest toxicity levels compared to all other anesthetic molecules. Several questions we hope to address in particular are:


18 What directs the distribution of xenon between water and the membrane environment? Is the thermodynamic transfer of noble ga s between phases governed by the hydrophobic effect and what membrane properties does it reflect? Is it possible to distinguish between su rface and interior membrane binding by via 129Xe chemical shift measurements? How does packing heterogeneity affect lipid membrane st ructure and dynamics with changing lipid composition and applied molecular stress? 1.2 Structure of Dissertation The work is structured as follows: Chapter 2 summarizes the basics of NMR theory and briefly describes specific NMR techniques utiliz ed within this dissert ation (e.g. magic angle spinning and 2D EXSY). Chemical shifts, chemi cal exchange nuclear relaxation phenomena are discussed in particular detail. The se cond part of this chapter introduces 129Xe NMR and key aspects of the spin exchange optical pumping process. C ontributions to the observed 129Xe chemical shift are discussed in context of xenon adsorption and dissoluti on into solid materials and bulk solvent, respectively. Recent developments in the gas delivery methods are also presented. A general overview of the lipid envi ronment is given in Chapter 3 in which we discuss the general structural and chemical prope rties of lipid molecules and bilayer membrane. The differences between bilayer and nonbilaye r lipids are highlighted. Several techniques utilized in the characterization of our lipid syst ems are explained and their experimental results provided. In Chapter 4 129Xe NMR is employed to study it s interaction with dioleoylphosphotidylcholine (DOPC) bilayers in several different vesicle morphol ogies. A new model is introduced to extract pertinent partitioning parameters based on chemical shift measurements; experimental results are compared to literature values for verification. These results provide the basis for Chapter 5 and Chapter 6, wherein th ermodynamic and kinetic properties of the xenon-


19 membrane association are evaluated, respectivel y. Solvation thermodynamics are utilized in Chapter 5 to elucidate the temperature dependence of the transfer mechanism. Thermodynamic parameters (e.g., solvation enthalpi es and entropies) are determined from chemical shift data and compared to literature values. We then treat the lipid matrix as a porous solid in Chapter 6; a number of surface models are employed to ex tract binding constants with the aim of differentiating between surf ace and hydrocarbon binding. Furthermore, we examine the effect of varying lipid composition on xenon partitioning in a two-component lipid system (Chapter 7). Incr easing the mole fraction of the nonbilayer lipid dioleoyl-phosphatidylethanolamine (DOPE) in a DOPC bilayer membrane induces changes in the membranes surface tension. The extent in which xenon solubility parameters vary with increasing DOPE content was examined using th e approach previously employed in the DOPC system. In this chapter we also investigate the effect of temperature and changing lipid composition on membrane morphology and local deform ations within bilayer structures. Lastly, preliminary data investigating the use of a nodic aluminum oxide (AAO) supported membranes are presented in Chapter 8. The main obstacle in making 129Xe NMR a useful tool in the characterization of biological a nd inorganic materials in solution within a reasonable experiment time and moderate physical conditions, is the lo w density and long relaxation time of thermally polarized 129Xe in solution. The only way to circumve nt this is through hyperpolarization ( HP ) techniques. Inorganic substrates, such as AAO, ca n be functionalized by se lf assembly of lipid bilayers on their surfaces, creati ng a convenient model of cellular membranes. In addition to verifying effective bilayer formation on the substrate via 31P magic angle spinning, we monitor at the xenon gas-to-lipid membrane exchange via 2D EXSY. Conclusions and future applications discussed in Chapter 9.


20 CHAPTER 2 NMR FUNDAMENTALS 2.1 Introduction Nuclear magnetic resonance (NMR) spectroscopy is the predominant technique utilized in this work. This chapter focuses on providi ng the background necessary for understanding measurable NMR parameters and various techniques employed in our studies. Of which include: polarization, relaxation, saturati on recovery, chemical exchange spin-exchange optical pumping and factors contributing to the chemical shift and relaxation behavior of 129Xe in solution and solid materials. In light of the many books and li terature reviews on this subject, we refer the reader to a number of texts for more basic,[17, 18] intermediate,[19-22] and advanced/theoretical approaches to NMR theory.[23-25] 2.2 Basic NMR Theory The NMR phenomenon relies on the interaction of an atomic nucleus with static magnetic field. All that is required is for nuclei of elements with an odd atomic number or odd mass to possess a nuclear magnetic moment (i.e nuclear spin). The nuclear spin is often view as a microscopic needle, having both magnitude and direction. Under normal circumstances, the nuclear spin can adopt any spatial orientation. Once a NMR active nucleus is placed into a strong magnetic field its nuclear magnetic energy levels split into multiple, non-degenerate spin states where only a discrete number of orientations are allowed. This process, called the Zeeman splitting, is caused by interactions between the nuclear magnetic moment and the magnetic field that breaks degeneracy and crea tes a slight shift in the atomic energy levels at thermal equilibrium. The distinct quantum energy for each state ( Em) is directly proportional to the strength of the applied magnetic field ( Bo) and is characterized by th e magnetogyric ratio of the spins ( I), which varies between atomic nuclei:


21 2mzoIoEBmhB (2-1) The magnetic quantum number, m denotes the total number of sp in states and is given by m = I -I +1,,+ I where I is the nuclear spin quantum number The total number of orientations is given by (2 I +1). Due to restrictions described by quantum mechanics, is unable to align exactly parallel to Bo. The torque experienced by the magnetic moment causes it to precess around Bo with an angular frequency, L, known as the Larmor frequency. -122 sLoIoB (2-2) The direction of the rotation is dependent on the sign of I while the angle between it and the magnetic field is based on the directional quantization, m The effect of I and natural abundance on the relative spectral sensitivity is gi ven in Table 2-1 for select nuclei. NMR is induced by applying a small alternating magnetic field ( B1) of appropriate frequency ( L) to flip the spin from the low energy state (| ) to the high energy state (| ). When the applied energy is equal to the energy absorbed by the nuc lear spins causes a transition between states; those in the low spin state move to the high sp in state and vice versa. The Table 2-1. Natural abundance and magnetic properties for several spin-1/2 nuc lei utilized in our work; 1H and 13C are given for comparison. The relative NMR sensitivity ( S) for equal number of nuclei at constant Bo and temperature, is dependent on the 3 I Values are IUPAC recommended valu es from Harris et al. (2001).[26] Neglecting relaxation times, the 129Xe nucleus is about 34 times easier to observe than 13Cthis is more so for 31P. Nucleus I Abundance (%) I 107 rads-1T-1 |SI / SH| 100% |SI / SC| 100% 1H 13C 31P 129Xe 1/2 1/2 1/2 1/2 99.99 1.07 100.00 26.44 26.75 6.73 10.84 -7.45 100.000 0.017 6.650 0.572 4150.6 3357.4


22 irradiation energy is in th e radiofrequency range ( RF) and is generally applied as a short pulse (on the order of microseconds). This absorpti on process induces a volta ge in the coil region which can be detected and amplified. The result ing signal is displayed as the free induction decay ( FID ). Thermal equilibrium will be restored via relaxation processes assuming no additional RF pulses are applied. According to Eq. (2-2), all the nuclei with in a sample would resonate at the same frequency if they all experience the same Bo. However, this is not the ca se since the characteristic L of each nucleus is reflective of changes in it s molecular environment; these differences are expressed as chemical shift values. The chemical shift (in parts per million, ppm) of a 129Xe nucleus is given as: 610Lref o (2-3) where ref is the resonance of our gas standard and o is the operating frequency of the spectrometer (400 MHz). The thermal standard us ed for reference consists of 5.26 atm of xenon gas contained within a flamed-sealed NMR tube with 0.26 atm of oxygen gas. 2.2.1 Nuclear Polarization At thermal equilibrium, the difference in the atomic energy levels prevents them from being equally populated. The population ra tio between states is dependent on I and Bo and can be characterized by Boltzmann statistics. As such, the occupation numbers of the of the | and | states, denoted by N and N for a I = 1/2 system, have direct consequence on the nuclear spin polarization ( P ). This in turn affects the magnetic resonance signal which is directly proportional to the z-magnetization ( Mz). Specific relations are provided in Eq. (2.2)-(2.5) for clarity:


23 exp11 0.9999777 2Io BBBN hB EE NkTkTkT (2-4) 100% NN P NN (2-5) 4100%I zIh P M (2-6) T is the temperature (in Kelvin), kB is the Boltzmann constant, h, the Planck constant, and [ I ] is the total number of spins per unit volume. N and N represent relative populations of the upper and lower energy levels, respectively; consistent with Eq. (2-4), N > N However, E is small at thermal equilibrium, so the population difference between the spin states is small (on the order of parts per million). These spin conditions generate extremely weak NMR signals and are the primary reason for the low sensitivity of NMR in standard experiments. For example, the E value for an ensemble of spin-1/2 129Xe gas in a magnetic field of 9.4 Tesla (400 Mz NMR) is approximately 7.38 10-26 J yielding an equilibrium sp in polarization of about 9 10-6 at 300 K. For NMR experiments suffering from low signa l-to-noise, the signal can be enhanced by increasing the number of detectable nuclei, or by signal averaging. As shown in Table 2-1, the 129Xe isotope is only 26.4% natura lly abundant. In the gaseous state, the amount of xenon can be increased by simply increasing th e pressure in the NMR tube. However, the use of high pressures in biological samples can be detrimental to the system being studied and the sensitivity is often limited by solubility. Use of high pressures is somewhat limited in solid samples as well since the amount of adsorbed xenon is limited by th e number of available surface sites. These limitations can be circumvented by utilizing is otopically enriched xenon. Though enrichments of > 86% are commercially available, they are more costly and only provide a threefold increase in the NMR signal. Signal averaging is the most co mmon technique used to increase the signal-to-


24 noise ratio ( SNR)though there are practical limitations to the improvements that can be gained. Nuclei with long relaxation times can require an ywhere from a few hours to several days for adequate resolution. The SNR is proportional to the square root of the number of signal transients averaged. Thus, a tenfold increase would requ ire the averaging of one hundred free-inductiondecays ( FID ). The nuclear spin polarization may also be enhanced via brute force methods (i.e. significant decrease in the temperature, or in crease in the magnetic field). Unfortunately, low temperatures have adverse effects on biol ogical samples and the maximum magnetic field currently available, 21 Tesla (900 MHz) only enhances the 129Xe polarization by one order of magnitude (P 2 10-5) under similar experimental conditio ns. Polarization enhancement may also be achieved through non-equilibrium meth ods, as in the case of hyperpolarized NMR,[27-30] dynamic nuclear polarization,[31-33] parahydrogen induced nuclear polarization,[34-36] and semiconductors.[37-39] The spin-exchange optical pumping ( SEOP ) of spin-1/2 noble gases is of particular interest.[40-43] Details of this particular technique will be discussed in more detail in subsequent sections. 2.2.2 Chemical Shift Anisotropy As mentioned previously, ch emical shifts reflect changes within the electronic environment that surrounds the nucleus. The electron density redu ces the external field at the nucleus by some factor and modifies the local magnetic fiel ds experienced by different nuclei. These local fields can be either isotropic or anisot ropic. If the electron density is identical in all directions, the local field is said to be isotropicit has no aff ect on the shielding ( ) and thus no affect on the observed chemical shift. If the shie lding is anisotropic, th e local magnetic fields generated by the nucleus will depend on the molecular orientation with respect to Bo; these


25 different magnetic fields at the nucleus can change with molecular motion, causing fluctuations in the chemical shift. The chemical shift tensor gives the orientati on dependence of the chem ical shift by relating the orientation of the magnetic field to the mo lecular frame in which the induced fields are generated: 1xxxyxz yxyyyzIo zxzyzz B (2-7) where is the shielding tensor. When placed into its principal axis system ( PAS), takes on a simplified form: 00 003 00xx PASyyisoxxyyzz zz (2-8) The principal axis system is the reference frame in which yields the diagonal form of the chemical shift tensor and the chemical shift th at would be observed if the magnetic field were along the x y or z -axis of the PAS. The resulting eigenvalues (xx, yyx, zz,) are termed the principal components. The eigenvectors are the direction cosines that relate the PAS to the original frame of reference (i.e., laboratory or molecular frame). The laboratory frame is typically defined by the direction of the field, while the molecular frame is defined by the local molecular symmetry. If the molecular motion is fast on the NMR time scale, the observed (isotropic) shift will be an average of the ei genvalues: one third the tr ace (sum of the diagonal) of the chemical shift tensor: 3isoxxyyzz (2-9)


26 A similar expression can be generate d for the shielding tensor, where xx yy zz and xx yy zz by convention. The chemical shift anisotropy ( CSA) is defined as the chemical shift difference between the isotropic and anisotr opic states and is us ually expressed as: 2zzxxyy yyxxzz (2-10) the asymmetry of the tensor () is given by: yyxxzz (2-11) Often times molecular rotations (in 2-dimensions ) causes the shielding to show axial symmetry; this results in two of the three principle com ponents to be equal. Acco rding to convention, the tensor element parallel to the symmetry axis is denoted by z, while the two equivalent elements are designated by As such, the CSA of this particular system is defined as the difference between the two (z ). This particular lineshape is called a CSA powder pattern and is commonly found in the NMR spectra of solid systems. Magic angle spinning. High resolution solid-state spec tra can be obtained one of two ways: by magic angle spinning (MAS) or to obs erve aligned molecules. MAS is a tool commonly used in solid state NMR to remove the effects of CSA and heteronuclear dipolar coupling from NMR spectra by mimicking the ra pid tumbling that would naturally occur in solution to average the anisotropic interactions in solid-state. As mentioned previously, if the rate of change in molecular orientation is fast relative to the magnitude of the CSA or dipole-dipole coupling (in frequency units), the molecular orientation dependence of the transition frequencies will result in an isotropically averaged valu e. The molecular orientation dependence of the nuclear spin interactions are disc ussed in more detail elsewhere. Sufficed to say, they are of the general form: 3cos2 -1. The orientation of the spin interact ion tensor to the magnetic field (i.e.,


27 Figure 2-1. The placement of the magic angle spinning (MAS) rotor with respect to the external magnetic field ( Bo). shielding tensor, dipolar coupli ng tensor) is given by the angle In powder samples, all molecular orientations and so all possible values of are represented. Spinning the sample at a particular angle, R, with respect to the magnetic fi eld results in an averaging of with time. So, provided that the spinning speed exceeds the am plitude of the anisotropic interactions, it is possible to average out the orientation depende nce of the interacti on anisotropy by setting R to a value of (the magic angle). Mathematical details of this NMR technique can be found elsewhere.[20, 44] 2.2.3 Relaxation Mechanisms Relaxation processes arise from interactions between the nuclear spin and fluctuating magnetic fields due to i) dipole-dipole interactions (1 D DT), ii.) molecular motions (1 SRT) iii.) chemical shift anisotropy (1 CSAT), and interactions with unpair ed electrons in paramagnetic compounds (1 paraT): 1111111111 ...c DDSRCSAparaTTTTT (2-12)


28 all of which have the potential to induce oscilla ting magnetic fields of the same frequency as L (Larmor resonance frequency). Dipo le-dipole interactions do so via the through-space interaction of magnetic dipoles of a pair of nuclei. Fluctuating fields may also be produced by molecular motions (vibrational, rotational, or translational) or by changes in chemical shielding of other surrounding nuclei. CSA describes th e orientation dependence of a ch emical shift with respect to the magnetic field. This relaxation mechanism is particularly important for nuclei with a large chemical shift scale, such as 31P. As seen in Eq. (2-12), the longitudinal relaxation time is inversely proportional to the correlation time (c) for aqueous solutions of normal viscosity. As such, slower random motions often give rise to shorter longitudi nal relaxation times ( T1) while faster molecular motions result in longer T1 values. The longitudinal relaxation time is the am ount of time required for the longitudinal magnetization to return to 63% of its original value after pertur bation. This particular relaxation is due to energy exchange between the nuclear spins and its surroundings and is characterized by an exponentially growth in th e observable magnetization: 11001tT tT zzMtMeMe (2-13) where, Mo is the equilibrium magnetization and Mz( t ) is the magnetization along the longitudinal magnetic field at time t The spin system is considered to be fully relaxed after three to five T1 periods. Herein, the longitudina l relaxation time was obtained by saturation-recovery methods this essentially equalizes the populations of the | and | spin states (i.e., saturation) and measures the recovery of the magnetization from zero as a function of time. The exponential growth is then fit to Eq. (2-13). The decay of the magnetization in the xy plane is governed by th e transverse relaxation ( T2). Similar to the longitudinal relaxation, T2 is the amount of time required for the net


29 magnetization in the xy plane to be reduced approximately 63% of its original value. This particular relaxation process is characterized by a continued de phasing of the magnetic moments in the transverse plane with time. Whats more, T2 can be monitored through changes in the NMR linewidth by the following relation: 12 21 T (2-14) It should be noted that T2 is always shorter than T1 as the spins in the transverse plane dephase faster than they align with the static magnetic field. Effect of Paramagnetic Ions. Unpaired electrons produce sign ificantly stronger magnetic fields than nuclei. As such, the presence of paramagnetic ion in solution will significantly increase the longitudinal relaxation rate (Eq. (2-12 )). In addition to their shifting ability, the use of shift reagents (i.e., contrast agents) can intr oduce severe line broadening in the NMR signal if present at high enough concentrations. This is co nsequence of the relaxation process provided by the unpaired electron. Paramagnetic shift reagents are commonly used to induce changes in both the chemical shift and relaxation behavior of se lect signals and molecular environments within liposomes. In the context of this work, it is used to confirm the chemical shift assignment of xenon dissolved within the lipid environment and to help resolve the relaxation properties of 129Xe gas within two separate compartments. 2.2.4 Chemical Exchange Chemical exchange is a dynamic process wh ich causes a nucleus to reside between multiple environments. This can have large repercussions on the observable NMR parameters, making it difficult to accurately describe the envi ronment. Here, we limit ourselves to several key points since the theory of chemical exchange effects have been treat ed in detail elsewhere.[45-


30 47] Let us consider the two-site chemical exchange of a 129Xe nucleus between sites A and B, having fractional populations fA and fB: A Bk kAB (2-15) where fAkA = fBkB at equilibrium. Two separate resonance lines will be observed at their characteristic resonant frequencies in the abse nce of chemical exchange (Figure 2-2). NMR evolution of this process can be described acco rding to the modified Bloch equations (i.e., McConnell equations), the solutions of which are composed of real and imaginary parts.[46, 48] It is the solution to the imaginary com ponent which yields the lineshape. Figure 2-2. Simulated NMR sp ectra depicting the effect of exchange and population on NMR spectra. A) Population: fA = fB = 0.50 for equally populated exchange sites. B) fB fA for unequally populated exchange sites for a two-site exchange process. intermediate0.1 s-15 s-1200 s-188.8 s-140 s-120 s-110 s-1400 s-1800 s-110,000 s-1 fast slow 020 -20kexHz intermediate0.1 s-15 s-1200 s-188.8 s-140 s-120 s-110 s-1400 s-1800 s-110,000 s-1 fast slow 020 -20kexHz (b) (a)AB AB A) B)


31 Table 2-2. The effects of exchange on the properties of an NMR spectrum for various exchange rates. Exchange Rate obs (in Hz) T2 relaxation (s -1) T1 relaxation (s -1) Experimental technique Slow exABk 2,2,11ex A Bk TT 1,1,11ex A Bk TT Linewidth measurements Intermediate exABk 2,2,11ex A Bk TT 1,1,11ex A Bk TT Lineshape analysis Fast exABk 2,2,11ex A Bk TT 1,1,11ex A Bk TT Spin-echo Slow exchange. Slow exchange gives rise to tw o, separate, broadened signals. The increase in the full-width at half height (FWHH) of 129Xe at each site ( i = A, B ) is given by: 12, 21ie xk T (2-16) Thus, for slow exchange, the exchange rate can be measured from the linewidth of the observed resonance (see Eq. (2-14)). In this regime, cha nges in the observed relative intensities are often related to modifications in relati ve population with varying solute, or ligand concentration as the positions of the resonance signals remain unchanged. Fast exchange. Spins from sites A and B experience an average local field within the fast exchange regime, resulting in a si ngle resonance represen ting a weighted average of the intrinsic chemical shifts at each site (see Eq. (2-17)). The increase in the linewidth due to chemical exchange is now given by: obsAABBff (2-17) 2 224ABAB Aff k (2-18)


32 So, the smaller the frequency difference (in Hertz) the narrower the linewidth. This regime is by far the most useful for determining binding and partitioning parameters from changes in NMR parameters. Intermediate exchange. As its name implies, intermediate exchange exists somewhere between fast and slow exchange conditions; its lineshape is ch aracterized by a broad, complex spectra that can be difficult to analyze. This ex change process can result in broadening so great that peaks disappear. Analysis can be made easier by shifting the system to either fast or slow exchange. 2.2.5 129Xe 2D EXSY Exchange NMR A number of two-dimensional (2D) NMR techniques have been developed for the study of slow exchange phenomena, the most common of which are Nuclear Overhauser Effect (NOESY) and Exchange (EXSY) Spectroscopy. [23, 49-51] Since the transfer of magnetization by the NOE is identical to the magnetization ex change resulting from the physical movement of nuclei, the pulse sequence of the NOESY and EXSY experime nts are the same. NOESY is used to provide information on through-space correlations through sp in-lattice relaxation while EXSY is utilized to detect chemical and conformational exchange Whats more, exchange rates are generally much faster than the cross-relaxation rate in 2D EXSY than for NOESY.[21] EXSY is characterized by peaks of the same sign and m easures correlation times between 10 ms 1 s.[51] The greatest amount of information is generally obtained when the exchange is slow on the NMR timescale and fast on the T1 scale. Chemical exchange is the process in which a nucleus physically moves from one molecular environment to another, resulting in a change in its chemical shift. Exchange measurements can be used to qualitatively map exchange pathways when little to no previous information is known about the dynamics of the system. It can also be utilized to quantitati vely determine rate of


33 interparticle xenon exchange All that is required is for the exchange process to be slow on the NMR timescale. However, if the exchange is too slow, all memory of the exchange process will be removed through relaxation processes. Similarly, the characteristic reso nance of each site will be unresolved if the exchange is too fast, maki ng the transfer process unobservable. Cross peaks due to magnetization transfer are based on chem ical exchange during the mixing period of the EXSY pulse sequence. Thermally polarized experiment. The standard 2D EXSY expe riment consists of three 2 x pulses. As shown in Figure 2-3A, the firs t two pulses are separated by evolution delay Figure 2-3. The pulse sequence for 2D EXSY (and NOESY) experiment for thermally and hyper -polarized 129Xe. A) The standard EXSY pulse se quence is indicated. The sample is treated with a saturating pulse train (SAT) followed by a repolarization delay (1) which reduces the acquisition times due to long T1s typical in thermally polarized 129Xe experiments. B) Continous flow pulse sequence. C) Interrupted flow pulse sequence. Figure modified from Cheng and Bowers (2007).[52] A) B) C) Flow rate 90 ml/min Flow rate 90 ml/min valve closed


34 time, t1, which is the frequency labeling period. Th e second and third pulses are separated by the mixing time, m, during which the magnetization is transferred between spins; t1 is the detection period. Like all 2D spectra, all p eaks have two frequency coordinates: f1 corresponds to the frequency experienced by the spin during the first time period ( t1) while f2 is the frequency observed during the detection period ( t2). In the simplest case, spins A and B both contribute to the z-magnetization at thermal equilibrium. The equilibrium magnetization in the Iz before the first pulse is given by:[23, 49-51] AB zzz I II (2-19) The first 2 x pulse produces transverse magnetization Iy by tipping the bulk magnetization onto the y axis on the transverse plane. Usi ng the product operator treatment, this can be described as follows: 2x A B zyy I II (2-20) Iy develops as it precesses around Bo with different Larmor frequencies; the final phase () of each precessing NMR signal increases as a function of t1. It is this phase angle that determines how much magnetization remains on the y-axis. Only the magnetization for the first spin (spin A ) will be considered for simplicity. The second arro w has no effect since it involves the operators of spin B only. The evolution of the system during t1 can be described for each spin by: 11 111 evolution time () cossinAB AzBztItI AAA yyAxA tII t I t (2-21) The second 2 x pulse then rotates the first term of Eq. (2-21) into the z-direction, creating a z-magnetization that has a cosine dependence on As shown below, the second pulse leaves the sine term unaffected:


35 22 11cos cosAB xx xxII AA yAzAIt It (2-22) 22 11sin sinAB xx xxII AA xAxAIt It (2-23) The longitudinal magnetization evolves under two effects: T1 relaxation and chemical exchange. The amplitude of the signal recorded during t2 is dependent on both m and the efficiency of magnetization transfer; the z-magnetization is proportional to the frequency and t1 via the cosine term. If spin A undergoes chemical exchange with spin B during the mixing time, it carries with it the freq uency label acquired during t1. The effect of the mixing process is given by: 11 1cos 1cos cosmixing AAB zAzAzA I tf It f It (2-24) where f is the fraction of spins of type A that are chemically exchanged with spins of type B As mentioned previously, chemical exchange induces magnetization exchange between the exchange sites due to the physical relocation of nuclei between two or more molecular environments. It is this magneti zation transfer that contributes to the formation of off-diagonal peaks and is the only compone nt affected by the last 2 x pulse: 22 11 22 111cos 1cos cos cosAB xx xx AB xx xxII AA zA yA II BB zA yA f It f It fItfIt (2-25) The final pulse creates the magnetization in the transverse plane which is ultimately detected during t2. The t1 value is sequentially increased to form the second chemical shift axis, while t2 is the acquisition time. The 2D spectrum is a re sult of a double Fourier transformation in the t1 and t2 time domains. It is the cos(t2) term that modulates the relative size of the peaks at f1 and f2. For a simple, first-order two-site exchange pr ocess with equal relaxation rates and relative populations between sites (t hermally polarized experiment); the exchange rate can be determined


36 Figure 2-4. An example of 2D EXSY spectra. A) A spin A in t1 that is transformed into a B spin during t2 will have the coordinates (A, B). A cross peak would also appear at (B, A) in the event that exchange proceeds in both directions simu ltaneously. B) A spin that remains unchanged during the experi ment will have the same frequency coordinates in both dimensions, de noted by the diagonalized peaks (A, A) and (B, B). directly by comparing the diagonal and cross-peak intensities. The cross-p eak intensities reflect cross-relaxation, while the intensities of the di agonal peaks reflect the characteristic relaxation behavior of the nuclei in each phase. They are se nsitive to an phenomena that contributes to the magnetization transfer during the mixi ng time (e.g., relaxation processes). Continuous and interrupted flow experiment. Long relaxation times and low densities of thermally polarized samples often limit the appli cability of 2D EXSY to kinetic studies due to long acquisition times. Kinetic models exist that permit the extraction of relaxation and kinetic parameters from a 2D EXSY spectrum acquired at a single mixing time.[53, 54] However, more accurate results can be obtained by acquiring a multiple spectra as a function of mixing time.[53, 55] Recent studies have provided the analytical expressions pertinent to the application and interpretation of hyper-polarized 129Xe 2D EXSY. [52, 56] Bowers and Cheng (2007) presented two modified 2D EXSY pulse sequences. The first sequence is for continuous flow experiments in which a stream of hyper-polarized gas circulates through the sample, continuously replenishing the hyper-polarized gas in the sa mple region (Figure 2-3B). The second pulse A) B)


37 sequence describes the interrupted flow techniqu e wherein the gas flow is stopped for a short time (2) right before the 2D EXSY sequence. This has been shown to significantly increase the signal-to-noise ratio. Interrupt ing the flow during the mixing time gives the probe molecule sufficient time to desorb from its adsorption site and accumulate in the coil region for detection. A schematic of the interrupted flow pulse sequence is provided in Figure 23C for visualization. Assuming a steady-state, two-site exchange process under continuous flow conditions, the theoretical values of the diagonalized ( IAA, IBB) intensities are expressed as: 101 ,expB AAmAmd RAn IIk n (2-26) 10,BBmBm dkIIe (2-27) and the cross-peak ( IAB, IBA) intensities are given by: 10 1/,d B ABmA RdAm m d Rkk n IIee kn (2-28) 10 1/,d BAmB Rdm m d Rkk IIee k (2-29) where, A is the 129Xe gas phase environment and B is the nanotube environment; nB/nA denotes the ratio of total adsorption sites to gas atoms, kd is the desorption rate constant, R, the gas residence time (inversely proportion al to the gas flow rate), IA0 and IB0 are the initial magnetizations in each respective phase at m = 0, while is the fraction of occupied sites. According to Eq. (2-28) and (2-29), the crosspeaks are likely to have unequal amplitudes. 2.3 Introduction to 129Xe NMR 2.3.1 The 129Xe Chemical Shift The chemical shift of 129Xe is approximately 2 orders of magnitude larger than of 1H (7500 ppm versus 20 ppm) and is a sensitive function of configuration of atoms that make up a physical


38 system. The observed 129Xe-NMR chemical shift is generally expressed as a weighted average of these configurations according to its relative adsorption probabili ty. As such, the intermolecular contributions to the chemical shift depend on the distance and lifetime between xenon and the individual atoms that constitute the absorption si tes. While highly fluid, lipid membranes possess mechanical properties similar to solid substrat es. It is important to understand the potential intermolecular interactions that contribute to the observed 129Xe chemical shift in porous solids and bulk solutions. Xenon adsorbed onto solid surfaces. Let us consider a solid surface that consists of chemically different target sites (Si) available for xenon adsorption. The observed 129Xe NMR spectrum will not only depend on the distribution of target sites in the sample, but the relative lifetime xenon spends at each adsorption site as well.[57] Theoretically, if the lifetime of xenon on each surface site is long, the resulting 129Xe NMR spectrum would contain as many chemical shifts as there are target types ( i); the population at each site would be proportional to its spectral intensity. However, if the lifetime of xen on at these absorption sites is short, the rapid diffusion between environments (surface and pores) w ill cause the intrinsic sh ifts of each site to coalesce, making it very difficult to extract any in formation about the system. If the distribution is homogenous, the observed chemical shift is then given by the following expression: 1 m obsii i (2-30) The characteristic shift and proba bilities corresponding to each Xe-Si are denoted by i and i respectively. According to studies by Fraissard et al. (1988) the chemical shift of the adsorbed xenon can be written as follows:[58]


39 oSXeSASEM (2-31) o is the gas reference at zero pressure; S denotes the contribution from the Xe-Si interactions and reflects the geometry of the Xe geometry at the surface; the Xe contribution reflects Xe-Xe collisions (proportional to the local xenon density); SAS arises from interactions between xenon and strong adsorption sites ( SAS); while E and M express the difference in the electric and magnetic fields due to the presence of charged io ns. Study of the different terms of Eq. (2-31) has been helpful in revealing structural def ects; determining the dimensions and internal volumes; and determining the porosit y of new mesoporous solids. Xenon dissolved into isotropic solutions. Often dubbed the medium effect, the total shielding of a solute molecule (relative to a reference shielding: o) due to solute-solvent interactions can be expressed as a sum of several terms corre sponding to various perturbations: 2owrbaE E (2-32) where, w denotes the van der Waal s dispersion interaction; r, the repulsive interaction, b, bulk susceptibility effects; a, the magnetic anisotropy of neighboring solvent molecules; E, reaction field induced by the permanent electric moment of the solute; and E 2, the contribution due to the permanent electric dipole moment of the solvent. This expression is similar to that found in Eq. (2-31). However, both the E and E 2 contributions can be neglect ed due to the lack of and relatively weak permanent dipole moments of xenon and polar solvents, respectively. Whats more, shifts arising from bulk susceptibility and solvent magnetic anisotropy have been estimated to be under 2 ppm in value; this is negligible compared to the 129Xe NMR shifts typically observed in polymer solutions (200 ppm). Thus, the dominant contributions to the 129Xe chemical shift are from van der Waals dispersion and repulsive interactions.


40 Effects of exchange. According to the fast exchange m odel proposed by Liu et al. (1994), the chemical shift can be expressed as follows:[59] obsddididggfff (2-33) for xenon adsorbed onto a least two, distinct environments (i.e., direct and indirect adsorption sites). fd, fid, and fg are the fractions of xenon adsorbed directly ( d ), indirectly (id ), and gaseous xenon ( g ), while d, id, and g are their corresponding intrinsic chemical shifts. The fgg contribution was found to be less than one percen t, resulting in the following simplification of Eq. (2-33): 1obs iddididff (2-34) Here, the fraction of indirectly adsorbed xenon (fid) is proportional to the fractional coverage of the zeolite surface, and can be expressed as: 2 12...id SSNN f NN (2-35) where 1 and 2 are the expansion coefficients, N is the number of adsorbed Xe atoms, and NS is the total number of available site s per zeolite supercage. Lui et al (1994) also intr oduced a virial expansion model in which d and id are a function of N .[59] Fitting experimental values to: 2 12...obs SNCNCN (2-36) provides information on the adsorption strength via C1 and C2; o denotes the effect of the zeolite surface on d in the absence of Xe-Xe interactions. More specifically, the two virial coefficients C1 and C2 represent coefficients of binary and three particle Xe-Xe interactions, respectively. However, the adsorption strength must be considered as a weighted average of the two adsorption sites in order to distinguish between the direc tly and indirectly adsorbed Xe atoms; there is no experimental method av ailable that allows for direct measurement of the adsorption


41 strength between the two. This particular met hod appears to provide a m eaningful quantitative description of a large number of systemsas such, we hope to use a similar formalism to differentiate between 129Xe binding at the membrane-water interface surface and the lipid core. 2.3.2 Alkali Metal-Noble Gas Spin-Ex change Optical Pumping (SEOP) Spin-exchange optical pumping is a two step process in which the angular momentum from an alkali valence electron is transferred to the nucleus of a ra re gas atom. This can result in 129Xe NMR signal intensity three or four orders of magnitude higher than its thermal equilibrium value. As mentioned previously, this is a non-equilibrium method of polarization enhancement which involves an imbalance in the atomic ground st ates via circularly po larized light. Here we present a brief, qualitative description of the opt ical pumping process necessary to understand the SEOP enhancement of the 129Xe nucleus. A more in-depth look at this process is provided in literatureit has been studied and reviewed extensively. Optical pumping. The first step in the polarization pr ocess is the optical pumping of an alkali metal, which uses the el ectronic excitation of atoms by light to transfer angular momentum from photons to atoms. With respect to hyperpolarized gas, the most successful methods make use of simple alkali metals. Rubidium ( Rb ) is the common alkali metal of choice for several reasons: its low melting point and high vapor pressure ensu res a high atom density at temperatures less than 473 K, and the wavelength of its electronic transiti on is within range of many commercially available, tunable light s ources. Rubidium has two naturally occurring isotopes: 85Rb (72.2% naturally abundant, I = 5/2), and 87Rb (27.8% naturally abundant, I = 3/2). The electronic ground state of 2S1/2 is split into two stat es, in accordance with the previously described Zeeman splitting. A give n excited level can be populated by irradiating Rb metal vapor with circularly polarized light whos e frequency corresponds to the absorption line of the atom at the ground state. The rubidium D1 line for 87Rb occurs at 794 nm and is displayed


42 Figure 2-5. The energy level diagram of 87Rb showing the hyperfin e structures for the D1 transition. F is the total angular momentum quantum number, I is the Rb nuclear spin, J is the Rb electron spin. The magnetic sublevels (mF) are separated by the Zeeman interaction. schematically in Figure 2-5. Light is scattered by only one of the S multiplet, provided that it is tuned to the lowest energy transition (2S1/2 2P1/2) and the light source is sufficiently narrow to avoid the excitation of other states. Selections rules ( mF = 1) dictate that spin-down state can only adsorb right-circula rly polarized light ( -), and spin-up states left -circularly polarized light ( +). For example, continuous irradiation of + light only excites atoms from the spin-down sublevel of 2S1/2 to the spin-up sublevel of 2P1/2. Once excited, the subl evel populations in (2P1/2)are randomized by collisional mixing (Figure 2-6). Thus the electron has equal probability of decaying into either of the ground states; excited electrons are relaxed by fluorescence or non-radiative path ways. Non-radiative F=2 F=2 F=1 F=1 mFlevels Hyperfine Structure Zeeman Splitting (by weak magnetic field) Fine Structure D1794nm F=|J+I||J-I| mF=2F+1 J=|l+s||l-s|D2780nm (removed) 2S1/2 2P1/2 2P3/2 E


43 quenching of the excited 87Rb in the presence of nitrogen and helium gas allows the electrons to return to the ground state without releasing depolarizing radiation. And so, continued irradiation of light repeats the excitation/relaxation cycl e until the electrons reach a spin state with no allowable transitions. These trapped electrons are wh at give rise to Rubidi ums net polarization. Spin-exchange. The second step of SEOP involves the transfer of spin between interacting particles. The spin excha nge process between the optic ally pumped rubidium and 129Xe is based on collisional polarization transf er and can occur in one of two ways: through simple binary collisions, or as short-lived, three body van der Waal s molecules described by Eqs. (2-37) and (2-38), respectively. RbXeRbXe (2-37) 2222 R bXeNRbXeNRbXeNRbXeN (2-38) The specific number and kind of co llisions is determined by the ga s pressure in the pumping cell; binary complexes dominate under high pressure conditions, while three body van der Waals molecules are more likely at low gas pressures (s ee Figure 2-7). Spin exchange occurs during the lifetime of the van der Waals pair and the relative lifetime of both complexes is limited by collisions with another buffer gas molecule (e.g., N2). The Fermi contact interaction describes the hyperfine coupling that exists du ring the polarization transfer a nd is responsible for the spin exchange between electron and nuclear spins. The time evolution of the 129Xe nuclear polarization can be expressed by: 111SEt SE XeRb SEPtPe (2-39) where I is the relaxation rate of 129Xe, and SE is the rubidium-129Xe exchange rate. As shown in Eq. ((2-40)), SE is dependent only on the rubidium concen tration and velocity averaged cross


44 Figure 2-6. Optical pumping for transitions fr om the ground state F = 2 level to the excited states in 87Rb The red line shows absorption of + photons while the orange, purple and blue lines show decays from each of the 5P1/2 magnetic sublevels. All ground state sublevels are depopulated by absorption, except for the m = +1/2 level where the population builds up. Figure 2-7. A schematic illustration of the polarization of 129Xe nuclei via collision and spin exchange. A) Binary complexes form at high gas pressure. B) Three-bodied complexes between Rb Xe and N2 gas occur at low pressures. Nonradiative Quench 1/2 2/3 Radiative Quench 794nm + CollisionalMixing 2S1/2 2P1/2 Relaxation 1/3 1/2 m=-1/2 m=+1/2 Xe Rb Rb Rb (a)(b) N2 N2 Rb Xe Xe Xe A) B)


45 section ( Rb-Xe) in binary spin exchange; it is dete rmined by the temperature dependent vapor pressure of the rubidium. M SE RbXe R b Xe (2-40) M is a constant that depends on the rate of van der Waals molecule formation and contact interaction strength between rubidium and 129Xe; is a constant that depends on the polarization and isotopic composition of rubidium. The averag e achievable polarization, as a function of the optical pumping (CP) and spin relaxation ( SD) rates can be written as: OP Rb OPSDP (2-41) The rate of rubidium optical pumping is depende nt on the time spent in the optical pumping cell and the temperature therein, the laser power, the rubidium adsorption lineshape, and the frequency overlap of the laser output. The spin re laxation rate in the inte rior of the pumping cell is determined by collisions between alkali-alkali metals and alkali-buffer gas molecules, as well as the strength of the external magnetic field. The spin relaxation rate, SD is the sum of the relaxation rates from all relaxation mechanisms. The main contributor to the sp in relaxation of rubidium comes from the Rb X spin rotation interaction; X denotes a specific nucleus such as Xe N2, or another Rb atom. The relaxation rates due to each relaxation collision process are given in Eqs. (2-42)-(2-44), the sum of which gives the projected value of SD. RbRb RbRbRb (2-42) 2 22RbN RbNN (2-43) 2RbXe RbNXe (2-44)


46 2SDRbRbRbNRbXeother (2-45) The brackets denote the relative con centrations of each nuclei while Rb-Xe is the velocity averaged relaxation cross section for each Rb X pair. Collisions with the cell wall and magnetic field inhomogeneity also lead to relaxation, details of wh ich are discussed elsewhere. 2.3.3 Recent Improvements to Gas Delivery System The gas handling system was redesigned due to severe leaks in the previous apparatus. The current system was designed to suit the needs of various types of hype rpolarized experiments which include continuous flow, direct expansion, and interrupted flow techniques. The schematic design is presented in Figure 2-8. The ballast tank was reincorporated to expand the total volume (2.0 Liters) of the system if needed; it is pa rticularly useful for experiments requiring the condensation of hyperpolarized 129Xe liquid or solid. Consistent with previous designs, the gas flow was generated by a magnetic ally coupled gas recirculat ion pump (Model 51429, Thomas Ind, Sheboygan, WS) while the flow rate is co ntrolled inline by a combin ation needle valve and variable area flowmeter (Model U-03217-06, Cole -Palmer, Vernon Hills, IL). The previously existing PFA tubing within the r ecirculation loop was replaced w ith copper tubing to reduce the risk of leaks in the system; copper ferrules were placed on copper tubing and PFA ferrules on PFA tubing. To begin, a gas mixture of speci fied pressure is introduced to the system from one or more of the following sources: GS 1, GS 2, GS 3. The various gas lines allow for custom mixing between different source gases, if desired. The gas output is then passed through an oxygen trap (Model 4002, Alltech Associates Inc, Deerfield, IL) and an inert gas purifier (Model 35K F-I4R, Aeronex Inc. San Diego, CA) to reduce the amount of impurities in the gas before it is loaded into the pumping cell (not shown). Numerous valves are in place to help control the


47 Figure 2-8. Schematic drawing depicting the redesigned gas deliv ery system. GS1, GS2, GS3: gas sources 1,2,3; P: pressure gauge; RV: re lief valve; V: needle valve; S: normally open solenoid valve; BP: bypass; : throughput valve; : oxy-trap; GP: gas purifier; : recirculation pump; : three-way valve; : needle valve. to Vacuum V toPumping Cell V Vacuum Line Turbo Pump ballast tank S P flow meter S GP RV V V VGS 3 GS 2 GS 1 BPBP from Pumping Cell FromProbe to Probe


48 expansion of the gas and the pressure is monitored using a Baratron pressure transducer (Model 722A, MKS Instruments, Andover, MA) near th e input of the pumping cell. In the closed-cycle, recirculation mode, three-way valves (B-42XS4, Swagelok, Solon, OH) control whether the sample probe is bypassed or not. We find them to be useful wh en building up polarizat ion via recirculation methods and when changing the NMR samples. For clarification, the gas ha ndling system is mounted onto an aluminum table built on the top of optical pumping system which consists of the optical polarizer, pumping cell and Helmholtz pairs. Six stainless steel supports were introduced to provide additional stability to the mounted apparatus (Figure 2-8) an d create a secure frame for th e laser curtain. The relative distance between the gas handling system and the 9.4 Tesla NMR magn et remained unchanged (approximately two feet) in order to pr event depolarization of hyperpolarized 129Xe gas during the transport process. Specific details of optical pumping apparatus can be found in Zook et al. (2002).[60]


49 CHAPTER 3 THE LIPID ENVIRONMENT 3.1 Introduction The cellular membrane is a semi-permeable network of glycerophospholipids, sterols and proteins. On average, a single mammalian cell is 5 nm thick and possesses an average diameter of only 20 m. Its structural integrity is maintained primarily through non-covalent interactions between proteins and lipid molecules. Despite its small size and apparent fragility, the cellular structure is highly flexible and m echanically stable. This is largely attributed to the amphipathic nature of phospholipids and the vast number and variety of lipid sp ecies that help compose it. In fact, cells are able to cont rol the molecular tr ansport propert ies through the membrane by modifying its lipid content; changes in membrane fl uidity and solvation allow the cell to regulate its functions. Biomembrane composition differs between species and can even vary between organelles within the same cell. While the actual role the lipid membrane and its heterogeneity play in common biological proce sses are unclear, it is increasingl y apparent that lipid biophysics contributes to a large number of cellular processes.[61-64] As stated in our general introduction, our specific interests lie in elucidating the role of lipid membranes in anesthetic action. The following chapter will introduce the readers to the ge neral structural and chemical properties of lipidsthe foundation on which my dissertation rese arch is based. Experimental results on the characterization of our lipid systems are also pr ovided as several techniques were employed. 3.2 Lipid Structure and Assemblies 3.2.1 What Makes a Lipid? Phospholipids are by far the most common t ype of lipid in the cell membrane, though what actually constitutes a lipid loosely defined. Easily distinguished by its structure, phospholipids typically possess i) a phosphate cont aining headgroup, ii) a semi polar backbone


50 Table 3-1. Approximate percentage by weight of total lipid in several cell membranes. Lipid types are characterized by the composition of their headgroup re gion. This is an adapted table.[65] Membrane/Lipid PC PE PS GlycolipidsSphingomyelinCholesterol Others Human erythrocyte 17 18 731823 14 Rat liver plasma 24 7 47 19 17 22 Myelin 11 17 920 8 28 7 Neurons 48 21 53 4 11 8 and iii) two fatty acid chains. Th ese three regions are illustrated for several lipid species in Figure 3-1 for visualization. Phospholipids are different from sphingomyelins and glycolipids in that they possess a glycerol in stead of a sphingosine backbone. The fatty acid chain can vary both in length (12-24 carbons) and degree of unsatur ation; modifications in either have large consequence on the membranes overall fluidity. Fo r clarification, a saturated acyl chain consists of a singly bound carbon chain while an unsatur ated lipid has one or more double bonds. It should be noted that this region is largely apolar. As mentioned previously, the lipid composition can vary c onsiderably between species (Table 3-1). A simple modificat ion in the headgroup region can give rise to distinct changes in the membranes chemical and physical environmen t. Contrary to the hy drocarbon region, this segment of the lipid molecule is distinc tly polar. Two of the most commonly found phospholipids are the zwitterionic (no net charge) phosphatidylcholine (PC) and phosphatidylethanolamine (PE). These particular lipids will be discussed in more detail in subsequent chapters. The other lipids listed in Ta ble 3-1 appear in lesser quantities within the cell membrane. And while cholesterol is a primary ingredient of the lipid membrane, it is a lipid


51 steroid whose chemical structure is unique unto itself. Unlike ot her lipids, cholesterol is unable to form membranes on its own due to its pronounced rigid structure. 3.2.2 The Bilayer Phase and Self-Assembly The greatest attribute the phospholipid brings to the cell membrane is its amphiphilicy. One part of the lipid molecule is polar (the hea dgroup) and thus water solu ble, while the other is not. The apolar hydrocarbon chain makes it char acteristically hydrophobic and only moderately soluble. When lipid molecules are placed into polar solutions they tend to aggregateforming self-assembled structures in water. This process is often referred to as the hydrophobic effect.[66] Contrary to common thought, hydrophobicity does not indicate repulsion between the hydrophobic molecule and water. Rather, the non-polar parts of a molecule aggregate with each other to exclude water molecules; the solubility properties are driven by an entropic effect. The water molecules that would surround the hydrocarbons can do so only at th e cost of decreasing the entropy of the water shell structure, thus making it more ordered. According to the second law of thermodynamics, the more favored process would be that which in creases the entropy of the water structure. This is ach ieved through the self aggregati on of the hydrocarbons, effectively freeing the water molecules. Electrostatic inte ractions, hydrogen bonding, and van der Waals and dispersion forces also contribute to the effectiveness of lipid self-assembly.[67] The prominent self assembled lipid structure is the bilayer. We have already established that it is energetically favorable for the hydroc arbon chains to aggregat e. This can be doubly effective if the acyl chains form their own do main, effectively sheltering it from the polar exterior and the hydrophilic natu re of the headgroup region. Th is double layer of lipids possesses both mechanical and electrical pr operties and is viscous and permeab le. It is for these reasons the bilayer is described as a one dimensional solid, two dimensional liquid structure. A pictorial representation of the bilayer membrane environmen t is provided in Figure 3-2. Lipid molecules


52 Figure 3-1. A schematic represen tation of the lateral pressure profile that exists near the water/lipid interface. As the purple line de viates to the right of the dotted one, repulsive interactions dominate due to in teractions in the lip id chain and headgroup regions. Deviation to the left represent the interfacial tension that exists to balance those repulsive forces (pressure). R signifi es the characteristic rotational frequency along the lipids long axis, and D is the typical rate of late ral diffusion. Lastly, to get a sense of how molecular structures and t hus translational and conformational lipid dynamics change with lipid type, three di fferent lipids are shown. The primary difference in the DOPC and POPC phospholip ids are in the chain length and degree of saturation. The first two letters of the li pid name describe the molecular nature of the hydrocarbon region. For example, (18:1 9-Cis) means two-oleoyl (DO) hydrocarbon chains, eighteen carbons in length, having a cis-double bond at C-9. When connected to a phosphocholine (PC) headgroup, the acronym becomes DOPC. As seen above, POPC also contains a PC headgroup, but has a saturated (16:0) chain (palmitoyl) instead of a second unsatur ated oleoyl hydrocarbon chain (18:1 9-Cis). And while both DOPE and DOPC possess iden tical makeup within the hydrocarbon region, the phosphoethanolamine (PE) headgroup is signif icantly smaller. For clarification, the co lor code is as follows: grey spheres-carbon, red-oxygen, orangephosphorus, blue-nitrogen and white-hydrogen.[68] diffuse laterally though the membrane plane and occasionally flip-flop between the inner and outer leaflets. As biological membranes are noto riously complex systems, artificial cells are often tailored for more controlled studies in a bilayer setting. Headgroup Region Hydrocarbon Zone BilayerPlane Aqueous Phase p= 0 tensionpressure (-)(+) D ~ 10-7cm2s-1 R ~ 109s-1DOPC DOPE POPC


53 A large variety of artificial membranes can be made from amphiphilic molecules for laboratory use. Commercial lipids normally come dissolved in an organic solvent, but once the lipid is dry, it can spontaneously assemble in water to form lipid bilayers. These model membranes, also known as liposomes, provide si mple imitations of biological membranes and cells. Liposomes have been used in the study of li pid-lipid, lipid-protein interactions as well as shape transformations, transport and elasticity.[68, 69] These model membranes form vesicles with an enclosed aqueous center. As mentioned previously, they can be tailor made to mimic the compositions of cellular membranes. The physical properties underlying th e various shapes and topological transformations of lipid membranes at variable environmental conditions often affect lipid membrane properties and functions.[70-72] 3.2.3 The Shape Concept of Lipid Polymorphism While most phospholipids possesses a bilaye r structure, some adopt alternate morphologies.[73] The subtle balance between favorable and unfavorable interactions at the hydrophilic-hydrophobic lipid interface allows for th e stabilization of all lipid membrane architectures.[74, 75] The importance of the dynamics at this boundary has been extensively reviewed and has shown that th e relationship between the mol ecular area and molecular shape contributes largely to the overall intrinsic properties of the lipid membrane.[76, 77] The shapestructure concept of lipid polymor phism is a simplified way to vi ew and predict preferred lipid structure. By comparing the cross sectional area of the lipid headgroup to that of the hydrocarbon chain region one can determine the average molecular shape via the packing parameter. The geometric packing parameter ( P ) is a dimensionless quantity used to characterize the preferred shape a lipid will possess at set environmental conditions.[78] According to Israelachvili (1991),[67] P =/aolc, where is the average volume of the entire molecule, lc is the approximate


54 length, and ao the average area of the polar head group at the lipid-water inte rface. For a bilayer structure (i.e. phase), lipids have an overall cylindrical shape, ha ving zero intrinsic curvature and a packing parameter is equal to one (e.g. PC ). However, when the average area of the phospholipid headgroup is much sma ller than that of th e acyl chains, inverted curved assemblies are preferred, consistent with P > 1 and negative intrinsic curvature (e.g. PE). When P < 1, the headgroup region is large relative to the total volume of the mol ecule and micelle -like phases are formed, possessing positive intrinsic curvature (e.g. lyso-PC). See Figure 3-2 for pictorial representation. Figure 3-2. The geometric packing parameter as a function of lipid type. The intrinsic geometry of the lipid structure changes with the packing parameter. The positive and negative curvature arising from the nonbilayer lipids are thought to play a part in membrane transport dynamics and protein function. This figure has been modified from Shearman et al. (2003).[79] The packing parameter can be used to predic t mean intrinsic curvature for various lipid structures having singular geometry.[80] It has also been shown that the mean curvature of lipid mixtures can be expressed in terms of P allowing for interpretations changing lipid membrane morphology as a function of temperature and composition.[81] To recap, each lipid has an intrinsic molecular shape. This shape-concept allows for the prediction of spontaneous curvature,


55 which is fixed for set environmental conditions However, small changes in any of those conditions (e.g., acyl chain length and/or uns aturation, headgroup t ype, pH, temperature, hydration, osmotic pressure, solution composition) can result in a transition between lipid phases as the preferred geometry and/ or curvature is altered. 3.3 Physical and Chemical Pr operties of Lipid Membranes 3.3.1 Interfacial Tension and the Lateral Pressure Profile At equilibrium, the lipid membrane is a ba lance act between the attractive and repulsive forces between its headgroups and acyl chains.[82] The lateral pressure profile describes the distribution of lateral stress in the membrane created by the conf ormational disorder and lateral repulsion existing in the in the apol ar and polar lipid regions, respectively (see Figure 3-1). It is not homogenous throughout the membrane and varies with bilayer depth.[83-85] These forces are balanced by the surface tension, which lies at the phospholipid headgroup/acyl chain interface. Its existence is a direct consequence of th e hydrophobic effectthe nonpolar solvent/acyl chain interaction is highly unfavorable. The surface te nsion is largely dependent on the lipid packing and thus changes with lipid type. This in turn can affect lipid membrane permeability,[86, 87] thus has a direct consequence on partitioning. It is this interfacial phase that makes partitioning between water-lipid systems fundamentally di fferent from typical two-phase partitioning.[88] The possible effects of lipid type on the latera l pressure profile and it subsequent role in protein dynamics was first introduced by Cantor.[84, 89] He proposed that the conformational balance of integral membrane proteins and anes thetic action are both affected by changes in the lateral pressure profile.[90] If we compare two lipid types, one with a packing parameter P > 1, the other having a P < 1, the cone shaped lipid (e.g. PE) is more likely to have its pressure dominated by stress in the hydrocarbon core, while a bilayer containing micellular lipids ( P < 1) is more likely have stress nearer the interface. Wh en incorporated in a bilayer environment, these


56 lipids introduce a variety of stress which may re sult in interdigitation at the membrane surface. Biomembranes often form self-assembled micro-domai ns (lipid rafts) as a way of balancing the additional strain.[91] 3.3.2 Lyotropic versus Thermot ropic Phase Transitions It is energetically possible to inhibit the form ation of a lipids intrinsic curvature by forcing the lipids to remain in the lamellar phase.[92, 93] The most common example of this particular kind of liposome is a PC/PE containing lip id mixture, often referred to as a frustrated bilayer. A phase transition will occur between two dynamical st ates when the preferred morphology becomes more energetically favorable than the lipid b ilayer arrangement. This can be brought about by modifying the environmental parameters of the li pid membrane, such as temperature and/or lipid composition. Lyotropic phase transitions are those stimulated by changes in temperature and concentration, while thermotropic transitions are those induced thermally.[94] Transitions that occur between phases of matter (g as, liquid, solid) take place at a specified temperature (e.g., melting point, boiling point). There are no intermediary phasesthe molecule will exist in either the gas, liquid or solid phase. This is not so for phospholipids; it is not uncommon for several phase transitions between phases. These pre-tran sition states often have complex morphologies representative of deformations displa yed in cellular functions. Nonbilayer structures have been postulated to play an important role in membrane function; the formation of local regions of nonbila yer structures (i.e. lip id rafts) within the biomembrane is one such idea.[95-98] It has been proposed that tr ansient formation of inverted structures is responsible for trans-bilayer transport of lipid s and polar solutes, a possible precursor to membrane fusion. By affecting the membrane barrier and flexibility properties, the presence of nonbilayer lipids may indirectly aff ect protein function as wellthey have been shown to increase the activity of numerous pe ripheral and integral membrane proteins.[99] In


57 addition, nonbilayer lipids can also influence protein folding and the conducting properties of channel forming peptides such as gramicidin.[100, 101] It should be noted that no experimental methods are currently available th at allow for direct measurement of the lateral pressure profile with sufficient precision. However, we can study the affect of lipid micro-domains on the partitioning of non-polar solutes. Th is issue will be addressed in more detail in Chapter 7. 3.4 Experimental Methods 3.4.1 Methods for Preparing and Characterizing Liposomes Phospholipids DOPC and DOPE were obtained from Avanti Polar Lipids and used without further purification. Two types of lipid suspensions were prepared: large unilamellar vesicles (LUVs) and multilamellar vesicles (MLVs). MLVs can vary in size from 0.5 to 5 m and often contain multiple non-spherical compartments and a non-uniform distribution of lipids, while LUVs are smaller (200-1000 nm), spherical in shape and possess a homogenous dispersal of lipids. Single component lipid suspensions were prep ared as follows: 1) Lipi ds in organic solvent of predetermined volume were dried with a gentle stream of nitrogen, forming a thin film, then placed under vacuum to remove any residual solv ent 2) a set amount of buffer solution (50 mM Hepes Buffer with 10% D2O) was added in appropriate amou nt, effectively hydrating the lipid vesicles and leading to the form ation of MLVs, 3) these hydrated vesicles were then extruded subjected to 7-40 complete passes through a 100 nm pore size polycarbonate filter, depending on the sample concentration. Once a single pass co uld be performed under a minute (and the sample appearance was as opaque as possible) one could be assured of homogen ous size and distribution of lipids within solution. The ex trusion methods lead to the fo rmation of LUVs which were 100 nm in diameter. Lipid mixtures containing vari ous mole percentages of DOPC and DOPE were produced along similar guidelines, only they were premixed in appropriate amounts while still in their organic solvent prior dehydration.


58 Figure 3-3. The effect of 31P orientation and motion on the 31P CSA for phosphodiester, representative of lipids in the membrane e nvironment. A) A fully static spectrum in which all components are reso lved due to limited moti on. B) A typical powder pattern, having order with rapid axial ro tation. C) The inverted hexagonal phase possessing additional disorder due to la teral diffusion around small water pores. HG denotes the oxygen (pink sphere) connected to the phospholipid headgroup region. 31P NMR is frequently used to probe the structural and dynamical properties of membranes. Its high natural abundance (100%) a nd presence in the headgroup region of each phospholipid molecule make it particularly useful in the study of lipid polymorphism. Despite its high sensitivity and large chemi cal shift range (100 ppm), the 31P linewidth is often dominated by an unusually large chemical sh ift anisotropy which ar ises from the orientation dependence of the local nuclear structure. For example, the fa st diffusion of small ve sicles gives rise to isotropically averaged lineshap es; the more the motion is rest ricted, the larger the spectral broadening. In a completely static sample lipid sa mple (Figure 3-3A), each of the chemical shift tensors can be resolved (see Chapter 1). In th e case of MLVs, the lipid rotation along its long axis (Figure 3-3B) is only partially averaged, leading to the characteristic powder pattern A) B) C)


59 associated with the bilayer phase. The formati on of nonbilayer phases is often monitored through 31P NMR as they often result in radical changes in lipid motion (Figure 3-3C). 3.4.2 Verification of Vesicle Size through Dynamic Light Scattering The particle size distribution was measur ed by dynamic light scattering on a PDDLSCool Batch +90T system (Preci sion Detectors, Bellingham, MA) tuned at a wavelength of 800 nm. All measurements were performed at 293 K using the viscosity a nd refractive index of water. The intensity autocorrelation was meas ured at a scattering angle of 90 and the distribution profiles were analyz ed using the Precision Deconvolve software package. Newly prepared LUVs were placed in a quartz cuvette and subjected to ten separate measurements of the intensity correlation function. This technique measures part icle diffusion due to Brownian motion. Differences in the solv ent-vesicle interaction produce ra ndom motion of the vesicles in solution. DLS relates the diffusive properties of the liposomes to its particle size through the translational diffusion coefficient ( D ). If the viscosity ( ) of the solution (i.e., water) is known, the Stoke-Einstein equation, Eq. (3-1), can be us ed to correlate the particle position decay; the smaller the particle, the faster the diffusion and decay. 3BkT dH D (3-1) Here, d(H) is the hydrodynamic diameter, kB is Boltzmanns constant, and T is the absolute temperature. This hydrodynamic diameter is th e diameter measured by the DLS via particle diffusion within a specified liquid. Ion concentr ation and type utilized in the experiment can potentially affect the vesicle diffusion speed if their presence change interfacial properties. 3.4.3 Phosphate Assays of Lipid Stock Solutions The concentrations of the lipid stock solutions were determined through phosphate assay analysis. The lipid assay is based on phosphate c ontent in the range of 1-100 nmole quantities of


60 lipid. This assay is dependent on the dephosphorylat ion of the lipid molecule by sulfuric acid (H2SO4). Once the phosphate ion is liberated and s ubsequently oxidized, a colored complex is formed which is quantified by absorbance spectr ophotometry. To be specific, 25, 50, and 100 nmole (estimated) quantities of lipid were placed into separate glass tubes. Then the solvent was removed by evaporation using a stream of nitroge n gas. Six phosphorus standards were prepared for comparison, ranging from 0-0.0975 moles. H2SO4 was then added to all samples in order to free the phosphate ion. The sample s were heated to 493 K to facilitate this process. Both the standards and samples were cooled to room te mperature and mixed thoroughly via vortex prior to addition of the malachite green phosphate assay solution. After a brief incubation (10 minutes), the absorbance was read at 650 nm. All samples were prepared in triplicate to help determine error. Experimental results are presente d as dilution-corrected averages of the 25, 50, and 100 nmole samples. 3.5 Results and Discussion 3.5.1 31P NMR of Lipids 31P NMR was performed on single component lipid samples of DOPC and DOPE. In addition to monitoring the lamellar-to -inverted hexagonal phase transition, 31P NMR was also used to verify the vesicle type in solution. As seen in Figure 3-4, aqueous suspensions of multilamellar vesicles gave rise to the characteristic powder pattern discussed in the previous sections. The broad, single lineshape observed in LUV samples is consequent of the higher vesicle mobility (Brownian motion) in solution. The full width at half-height (FWHH) for the LUV DOPC sample is much broader than those ty pically seen in micellular systems, which is expected. As discussed previously, the 31P NMR CSA is dependent on motional averaging of both the lipid molecule in its local environmen t and the macromolecular system in which it exists. For all intensive purposes, the DOPC lipid is confined to the bilayer, thus any averaging


61 of the // component of CSA must come from the diffusi ve nature of the vesicle. If the vesicle tumbles rapidly enough in solution, the anisotropy will reduce a single resonance according to iso= (11 + 22 + 33)/3 or ( // + )/2. The broader the resonance, the slower and larger the vesicle must be. Figure 3-4. The 31P NMR spectra of 50 mM DOPC for va rious lipid environments. A) Large unilamellar vesicle (LUVs) systems result in a broad isotropic peak. B) Multilamellar vesicles (MLVs) possess a characteristic powder pattern. The between the and components determined to be 41 ppm, consistent with expected values. The presence of xenon gas had no noticeable effect on the values or the shape of the MLV CSA. 3.5.2 Dynamic Light Scattering We performed DLS on two separate concentrations of DOPC LUV solutions. Two samples of varying concentration (2 mM and 1 mM) were prepared using the extrusion method discussed above. The average vesicle diameter wa s calculated after twenty sequential runs. As seen in Figure 3-5, this ga ve a value of 117 20 nm, which is within error of the 100 nm polycarbonate pore size used in the extruder. Also the sharp slope of th e correlation function is A) B)


62 consistent with the high degree of Brownian motion expected in vesicles of this size. Despite our interest to do so, we were unable to modify the apparatus to handle gas tight samples over the pressure range used in our studies. While modifi cations in bilayer thickness in the presence of anesthetics have yet to be obser ved, it would have been an inte resting attempt. In addition, it would have allowed us to test for vesicle aggr egation in the presence of anesthetic gas. Figure 3-5. Dynamic light sc attering data on a 2 mM DOPC LUV suspension. A) The volume distribution with respect to vesicle size. B) The correlation function associated with the 2 mM samples. 3.5.3 Phosphate Assay Analysis The standard curve shown in Figure 3-5 wa s made by plotting the calibrated absorbance of the standard samples as a function of the relative nmoles of phosphate contained in each sample. This resulted in a nonlin ear least squares fit, yielding a R2 value of 0.984. The stock bottle concentrations were measured for both DOPC and DOPE and their dilution corrected values are provided in the table embedded in Figure 3-6. Approximately, 25, 50 and 100 nmol samples were measured in triplicate for each lip id sample. The dilution correction adjusts the 50 and 100 nmol values for comparison to the 25 nmol sample. This is the source of the measured and dilution corrected values found in Figure 3-6. The corr elation ratio gives the relative Correlation coefficientTime (msec) Davg= 116.7 20.1 nm Vesicle Size (nm)Volume (%)(a) (b) A) B)


63 difference between the theoretical and experiment ally determined values. As one can see, both concentrations of the stock solutions are within error of the predicted value. Thus, we can assume that no significant solvent evaporation has taken place and that our projected lipid concentrations are correct. Figure 3-6. Experimentally determined values of phosphate assays. A) The titration curve of the phosphate standards. B) Extracted and dilu tion corrected concentration values for both DOPC and DOPE stock bottles. 3.6 Conclusions From these results we can conclude that we have uniform dispersions of large unilamellar vesicles using the extrusion method described ab ove. Both the DOPC vesicle size and the bilayer and inverted hexagonal phases have been confirmed through 31P NMR. Lastly, our projected lipid concentrations were confirmed through phosphate assay for both DOPE and DOPC stock solutions. While the phosphate assay was not pe rformed on every lipid sample made, I am confident that in our projected lipid concentratio ns. Not only were the sample volumes incredibly large (1 ml) and easy to manage, the majority of the samples were prepared during the same time frame in which the phosphate assay was performed. A) B)


64 CHAPTER 4 CHARACTERIZATION OF XE-LIPID PARITIONING BY 129XE NMR 4.1 Introduction Despite continued research on the pharmacologi cal properties of general anesthetics, there is still no generally accepted molecular level mechanism.[102-105] Two prevalent theories exist: those that focus on membrane perturbation eff ects and those that point to specific protein activation sites. What is apparent is that accumulation of inhalational anesthetics in lipid membranes can alter their distri butive properties and en ergetics. It is hypo thesized that as anesthetic molecules partition into the biomembr ane, changes in the lateral pressure profile ensue, indirectly altering membra ne protein structure and function. Though it is well established that this dynamic process is largely controlled by interfacial properties, the actual modification of the surface tension at this boundary is not fully understood. Here, the inhalational anesthetic xenon ( Xe ) is used as a non-polar, weak ly binding, spin-probe to i nvestigate the anesthetic-lipid bilayer interaction by NMR spectroscopy. We use 129Xe NMR to study the interaction between xenon with dioleoylphosphotidyl -choline (DOPC) bilayers in several different vesicle morphologies. Although 129Xe NMR has been used extensively in biological research, the basic kinetic and dynamic properties of the xenon-lipid membrane interact ion have not been fully characterized. In this chapte r, the xenon-DOPC interaction will be investigated though the Xe chemical shift. In addition to analyzing the sensitivity of 129Xe NMR to various types of DOPC vesicles, we extract the limiting chemical shift of 129Xe in DOPC using methods which are frequent in anesthetic binding studies. The limiting shift is then used to determine the buffer/lipid membrane partition coefficient. Obtaining a us eful, reliable expression to extract pertinent thermodynamic and kinetic parameters is vital. Th e details presented in th is chapter provide the


65 foundation for Chapters 5 and 6. We begin by providing a brief review of the uses of 129Xe gas in the study of biologically relevant media. 4.1.1 The Dissolution of 129Xe in Biological Media As discussed in Chapter 1, of the prevailing hypothesis for anesthetic action involves perturbation effects experienced within the membrane due to anesthetic uptake. Anesthetics have been shown previously to depr ess the gel-to-lamellar phase tr ansition temperature for various lipidic systems,[106-110] increase both membrane fluidity[111, 112] and surface area,[113, 114] and decrease acyl chain order.[115-117] While it is generally underst ood that the partitioning of small molecules has the potential to induce significant changes in the physical properties of the lipid membrane, it is still unclear if and how these translat e to anesthesia. This is in part due to the low physiological concentration needed to induce anesthesia. However, computational studies as well as several experimental observations indicate th at anesthetic molecules strongly influence the membrane surface tension.[118, 119] Thus, it has been recently pr oposed that an anesthetics intrinsic properties may mediate the redistribution of the lateral pressure profile, and in this way facilitate anesthetic action.[84, 89] Xenon is a potent anesthetic with NMR propertie s that make it well-suited to the study of the basic nature of the anes thetic-membrane interactions.[7, 120-123] Xenon gas is inherently nonpolar and chemically inert; it possesses distin ct hydrophobic character and a large chemical shift range ( 250 ppm) when dissolved in simple liquids due to the high polarizability of its electronic cloud. Its relative high so lubility in lipid rich media has led to a significant increase in its use in biological studies for lipid containing components.[120, 124-126] Enhanced sensitivity of 129Xe through spin exchange optic al pumping has prompted creativ e technical advancements in gas delivery and experi mental applications.[127-129]


66 Recent studies have focused on designing efficient transport mechanisms to enhance 129Xe NMR/MRI signals in lipid rich domains in macr omolecular biomedia (i.e ., brain, lung). Laserpolarized 129Xe gas has been successfully delivered in vivo through both inhalation and vascular injection methods[130] and has been used to study proteins in solution in an attempt to elucidate the site of anesthetic action.[131] However, the high exchange rate of the xenon between two or more environments usually yields broad resonances wit hout resolution of dis tinct binding sites. When an isolated xenon gas atom is dissolved in solution, its chemi cal shielding changes by several hundred ppm.[132, 133] This gas-to-solution chemical shift, commonly dubbed the medium effect, is dominated by van der Waals dispersive and repulsive forces between the Xe atom and its surroundings. The interpretations of solvent shifts are typically based on the Onsager reaction field model which predicts the shift to vary linearly with the squa re of the function f( n )=[( n21 )/(2n2+1)], where n is the refractive index of the pure solvent.[134] Correlating the solvent shifts with the solubility parameter is another approach wherein the shift is related to the cohesive energy of the solvent through its heat of vaporization and the molar volume. However, these models do not provide much information about the intermolecular interactions that give rise to these shifts. 4.1.2 Xenon and the Lipid Bilayer Several published studies have examined the interaction of xenon in suspensions of PC vesicles of various compositions at room temperature using 129Xe and 131Xe NMR spectroscopy, with the intention of elucidating xenons anesthetic properties.[10, 15, 123, 135, 136] It is impossible to assign a specific location of xenon within a mi celle or single bilaye r vesicle through direct observation of 129Xe NMR chemical shift due to fast exchange processes. The rapid exchange of xenon within the lipid phase and bulk solution genera lly results in a single broad resonance in the 129Xe spectrum.[10, 137] However the dissolution of 129Xe into aqueous suspensions of multi-


67 lamellar vesicles (MLVs) often results in two di stinct resonances; a broad peak representing the lipid environment is often accompanied by an add itional narrow resonance associated with the solvent phase (for example, see Figure 4-1). Th e specifics of the exchange processes and the significance of vesicle type will be discu ssed in further detail below. Almost all 129Xe NMR studies in biomembranes employ MLVs since the signals representing the two phases, water and lipid, are fully resolved Mohseni-Hosseini (1985)[136] did attempt to look at the size dependence of the vesicles on the gel-tolamellar phase transition of several saturated lipids in fully hydrated solutions.32 By using sonication methods to create increasingly homogenous distributions of vesicle size in solution, a general trend emergedif the vesicle size is too small, detection of the gel-to-lamellar is obscured. One of the most informative studies of the xenonlipid interaction wa s published by Xu and Tang (1997).[14] The intermolecular 129Xenuclear Overhauser eff ect (NOE) was utilized to determine the location of xenon in egg yolk PC (EPC). By select ively saturating different groups of protons within the lipid membrane and measuring the subsequent change in the 129Xe spectral intensity as a function of proton saturation time, the authors were able to determine the most probable location of the xenon fr om the build-up rate of the 129Xe-{1H} NOE. The effect of cholesterol was also investigated and shown to enhance xenons tendency to interact with the interfacial region of the model membrane. These results indicate that xenon interacts preferentially with the amphiph ilic interface in PC lipid membrane, not the hydrocarbon core. The result was attributed to the high polarizability of the Xe atom. It was previously thought that an anesthetics hydrophobic nature would make the interaction w ith the acyl chain region more favorable, but it is now known that only st rongly hydrophobic solutes (non-immobilizers)


68 partition into the membrane core.[138] Similar trends have been observed utilizing 19F NMR techniques when comparing th e dissolution of fluorinated anesthetics and non-anesthetics.[16, 139] 4.2 Partition Model 4.2.1 Membrane-Buffer Partitioning The ideal partition model is generally used to elucidate the small molecule-biomembrane interaction.[140] Often related to lipid solubility, partitioning describes the distribution of solute between two distinct phases. In a dilute solution, one can assume ideal mixing of the solute in each phase, allowing for the determination of thermodynamic parameters. Here, we relate partitioning theory to NMR fast exchange theo ry in order to further characterize the xenonmembrane interaction. The mole fraction parti tion coefficient is define d as the ratio of the equilibrium mole fraction of xenon in the lipid vesicle (XXe,L) to the equilibrium mole fraction of xenon in the bulk aqueous phase ( XXe,aq); i.e., Xe XeL WL p X eX e Xea L q aqLX nn K X nnn (4-1) For clarification, Wn and Ln denote the total moles of wa ter and lipid in solution and Xe L nand Xe aqn represent the moles of xenon in the lipid and aqueous phases at equilibrium. In the limit of low solute concentration,Xe aqWnn resulting in a minor simplification in Eq. (4-1). Assuming ideal solution conditions with an activity coefficient of unity the equilibrium fraction of bound xenon ( XXe,L) is expected to be directly proportional to XXe,aq. In order to determine the mole fraction of xenon in both the aqueous and lipid phas es, the partition coefficient is related to the NMR chemical shift in the fast exchange regime, providing the basis for the determination of the partition coefficients from the observed 129Xe chemical shifts in LUVs. The same approach has been utilized in the study of drug-membrane interactions.[16, 141]


69 4.2.2 Determination of Partitio n Coefficients by NMR Here we utilize the simple two-site excha nge model, assuming rapid xenon exchange between the aqueous and lipid membrane phases; the aqueous phase consists of Xe atoms located in the bulk water phase away from the vesi cle while the lipid phase consists of all Xe atoms bound to the vesicle surface and membrane in terior. In the fast exchange regime, the expression for the observed chemical shift is gi ven by the weighted average of the chemical shifts in each of the two phases. As explained in Chapter 2, if the rate at which xenon relocates between two or more environments is much faster than the reciprocal of their limiting chemical shift difference, the observed chem ical shift will occur at the weighted average shift. In this particular system, xenon is either associated with the lipid environm ent (bound) or the bulk solution (free). [16, 142-144] Thus, obsboundboundfreefreeXX (4-2) where obs is the observed chemical shift of xenon in DOPC solution, bound and free represent the intrinsic chemical shifts of xenon in the free and bound environments, while Xbound and Xfree correspond to the mole fractions of bound (/XeXe L totnn)and free (/XeXe aqtotnn) xenon. Setting the observed chemical shift of xenon in lipid-free buffer solution (buffer obs) equal to free and substituting Xfree = 1 Xbound into Eq. (4-2) yields: max buffer obsobs bound buffer boundobsX (4-3) Note that bound free ma x is the difference in the intrinsic shifts of xenon in the lipid and aqueous phases. Rearrangement and substitution of Eq. (4-1) into Eq. (4-3 ) allows for extraction of the max value, providing the basis for NMR bi nding analysis. As shown in Eq. (4-4), measuring the change in the observed chemical shift with increasing moles of lipid allows for the


70 determination of the mole fraction partition coefficient at fixed xe non overpressures. This expression is easily converted to concentrati on form by dividing by the total volume of the solution (1ml). max max[] 55.5[]pL p Xe WpLpaq pp aqKn KL nKnKnMKLKXe (4-4) Similar to previous annotations, [ L ], 55.5 M and [ Xe ]aq represent the tota l phospholipid, water, and free xenon concentrations in aqueous solution respectively. The mole fraction form of the partition coefficient is then found by plotting against the total lipid concentration. From this point forward, the total lipid concentration, [ L ], will be referred to as [DOPC]. For certain small-molecule lipid membrane systems, the two-site exchange model described above is an oversimpflication that cannot be made to fit the data, and better compliance can be obtained using a three-site exchange model in which the membrane surface and membrane interior represent distinct, chemically inequivalent environments. The Xboundbound term in Eq. (4-2) is modified to account for the additional binding site: obsSSIIfreefreeXXX (4-5) Here, XS and XI are the fractional occupancies of xenon at the membrane surface and interior, respectively, while S and I denote the theoretical limiting chemical shifts of xenon associated with each of these respective environments. C onsidering the nonspecific nature of the xenonmembrane interactions (no specific binding site s), and apparent fast exchange between the surface and lipid environments, the two-site model is deemed adequate for our purposes. As shown by Kennan and Pollack (1990)[145] the solubility of xenon in water does not change significantly within the pressu re range used in this study. [Xe ]aq is estimated assuming the solubility in buffer solution is the same as th e solubility in water, and that it behaves ideally.


71 According to Henrys law, the concentration of xenon in solution is directly proportional to the gas pressure. So, [ Xe ]aq = PXeKH where KH denotes Henrys Law constant and PXe the xenon overpressure in the gas phase above the liquid. The KH value was taken to be 0.0437 M/atm.[146] As required by conservation of mass, the total concentration of xenon in solution is given by [ Xe ]tot = [ Xe ]aq + [ Xe ]bound. 4.3 Results and Discussions 4.3.1 Spectral Properties of the Xe-Lipid Interaction It has been shown previously th at xenon typically exhibits fast exchange in and out of the hydrophobic environment in small unilamellar vesicles.[136] Spectra of MLVs in solution typically exhibit two resolved signals; a broad peak corresponding to the dissolu tion of xenon in the lipid environment and a narrower peak associated with free Xe atoms in the buffer solution.[10, 136] We observed two separate peaks in MLV samples formed in 50 mM DOPC. In this lipid, the broad peak associated with the lipid environment occurs downfield (higher ppm) of the solvent peak (Figure 4-1) contrary to the upfield sh ift observed in dimyristoylphosphatidylcholine (DMPC) and dipa lmitoyl-phosphatidylcholine (DPPC).[10, 136] Whether this observation is a consequence of factors such as partitioning behavior, lipid packing effects, vesicle fluidity, etc, is not known since only a few lipid systems have been investigated via 129Xe NMR. Nevertheless, partitioning of inhalation anesthetics have been shown to depend on the acyl chain length as well as the degree of acyl chain saturation which are likely to be reflected in the observed 129Xe NMR chemical shift. The addition of double bonds induces chain disordering while increasing the molecular order in the vicinity of the double bond. This results in an increased area per lipid molecule toward the cente r of the bilayer, which in turn reduces the packing density at the lipid/wate r interface. Recent investigations into the phase behavior of thermotropic liquid crystals (LCs) via 129Xe NMR spectroscopy demonstrates the sensitivity of


72 Figure 4-1. A series of 129Xe NMR lineshapes of xenon in MLV, LUV, and lipid free environments. A) 5atm of xenon dissolved into 50 mM Hepes Buffer solution (pH=7.4). B) The equivalent pr essure over 50 mM DOPC LUVs. C) PXe = 5 atm in 50 mM DOPC MLVs. The gas phase-to-liquid vol umes are identical for all samples. the chemical shift to changes in liquid crystall ine order, density and temperature for both the nematic and smectic phases.[147] The peak assignments to the aque ous and lipid environments of 129Xe in MLVs were verified independently through chemical shift analysis of 129Xe in lipid-free buffer solution and LUV samples (Figure 4-1). Note th at the intrinsic chemical shift of xenon in the buffer solution (Figure 4-1A) is up-field with respect to that observed in 50 mM DOP C LUVs (Figure 4-1B). Furthermore, the observed 129Xe spectrum in LUVs consists of a single peak which lies almost directly between the two phases observed in MLVs This is consistent with fast exchange between the lipid and aqueous phases in DOPC LUVs. However, the associated trends with increasing temperature differ from th ose reported by Miller et al. (1981).[10] Namely, fast exchange is not induced with temperature and the DMPC associated peak is upfield of the water A) B) C) max PPM 198 196 194 192 190 188 186 184 182


73 Figure 4-2. 129Xe NMR chemical shift and lineshape behavior of 5 atm of xenon dissolved in 50 mM DOPC MLVs. While spectral overlap is obs erved at 313 K, it is not indicative of fast exchange. This is simply a cross-over event, in which the two resonances happen to pass through a common chemical shift range. peak. While the resonances do overlap at 313 K, two distinct peaks re-emerge at higher temperatures (Figure 4-2). A summary plot describing the te mperature dependence of the chemical shifts is provided in Figure 4-3. The variation in the observed sh ift decreases monotonically with temperature for both MLV and LUV lipid dispersions. The chemi cal shift behavior of the lipid-free buffer solution appears rather insensitive in comparison. In addition, the temperature dependence of the LUV resonance appears as a weighted average between the two phases exhibited in the MLV data. Inspection of Figure 4-3 suggests that xenons affinity for the lipid phase increases with temperature; the fast exchanging LUV resonanc e shifts away from the MLV aqueous phase towards the MLV lipid phase. This will be explored in more detail in the next chapter. It should be noted that the resonance asso ciated with the lipid phase unde r slow exchange conditions was 293K 298K 308K 313K 323K 323 K 313 K 308 K 298 K 293 K PPM 200 198 196 194 192 190 188 186 184 182


74 Figure 4-3. Temperature dependence of the 129Xe chemical shift in 50mM MLVs and LUVs: ( ) and ( ) denote MLV lipid and aqueous pha se, (*) signifies LUVs environment. Inset: lipid-free buffer solution, ( ), compared to the MLV aqueous phase. Experiments performed with identic al gas-to-solution volumes under PXe = 5 atm. found to be dependent on the concentrations of both lipid and xenon in solution. For this lipid system, a single resonance was observed under the following conditions: [DOPC] < 25 mM, and PXe < 5 atm. So, the fast and slow exchange observed in PC vesicles by Saba et al. (1996)[148] and Miller (1981),[10] respectively, may be a result of both PXe / lipid concentrations utilized in each study as well as vesicle size; Miller[10] dissolved 129Xe into MLV dispersions while Saba[148] performed experiments on sonicated vesicles. 4.3.2 129Xe T1 Relaxation Rates Determined by Saturation Recovery Methods The longitudinal relaxation time (T1) was measured for 129Xe dissolved in lipid-free buffer solution by the saturation-recovery method and found to be 110 34 s at 5 atm and 298 K. The relaxation recovery of xenon in DOPC MLV soluti ons required a bi-exponen tial fitting function 188.2 188.0 187.8 187.6 330 325 320 315 310 305 300 191 190 189 188 187 186 185 129-Xe (ppm) 340 330 320 310 300 Temperature (K) Temperature (K) Temperature (K) 129-Xe (ppm)


75 for [DOPC] 25 mM, allowing for the extraction of T1 values in both the aqueous and lipid associated phases in solution assuming that the exchange is slow on the time-scale of T1. Both the mono-exponential and bi-expone ntial fits of xenon in 25 mM MLV DOPC solution are provided in Figure 4-4B for comparison; experimental results show better compliance to the biexponential function. On the other hand, the relaxation recovery in LUV solutions could be fit to a single exponential, as expected for fast exchan ge between the two environments. The extracted Figure 4-4. 129Xe NMR saturation recovery curves for xenon dissolved in various solutions. A) Lipid-free buffer solution (pH = 7.4). B) 25 mM DOPC LUV solution with a monoexponential fit (dotted line) versus bi-expone ntial fit (dashed line). C) Lipid-free buffer solution in the presence of 3.5 mM MnCl2. D) The lipid ( ) and aqueous ( ) associated phases in 50 mM DOPC MLV solution under PXe = 5 atm of overpressure. 1.0 0.8 0.6 0.4 0.2 0.0Normalized Signal 200 150 100 50 0Tau (sec) 1.0 0.8 0.6 0.4 0.2 Normalized Signal 20 16 12 8 4 0Tau (sec) 1.0 0.8 0.6 0.4 0.2 0.0Normalized Signal 12 10 8 6 4 2 0Tau (sec) 1.00 0.95 0.90 0.85 0.80 0.75 0.70 0.65 0.60 0.55 0.50 0.45Normalized Signal 600 500 400 300 200 100 0Tau (sec) a) c) b) d) A) B) C) D)


76 T1 value is comparable to those determined by Xu and Tang (1997)[14] and serves to confirm the relative strength of the xenon-membrane intera ction under fast exchange conditions (Figure 4-4C). Saturation recovery measurements were al so performed on a 50 mM DOPC MLV sample under similar experimental conditio ns in order to extract the T1 values of xenon associated with the lipid and aqueous environments through indepe ndent analysis of thei r previously assigned resonances (Figure 4-1C). As seen in Figure 4-4D, the longitudinal relaxation rates differ significantly; the dissolved gas within the lipid environment relaxes more rapidly than that within the aqueous phase. The re sults are summarized in Table 4-1. There is a significant decrease between the T1 values of xenon in the lipid-free buffer solution and the aqueous phase of the MLV solutions despite the fact that their chemical shifts ar e near identical (inset of Figure 4-3). This reduction may be attributed to Xe -surface interactions at or near the lipid/water interface and/or Xe relaxation in the lipid associated pha se due to exchange. The heterogeneous nature of the MLVs allow for xenon to diffuse freely through multiple bilayers, with limited contact with the aqueous phase. 129Xe NMR methods were used to differentiate between DMPC/DHPC bicelle phases via changes in T1 relaxation times.[135] Contrary to previous investigations,[149] no significant change was observed, suggesting that increased viscosity, changes in morphology and/or orde ring has no substantial effect on the observed relaxation time. 4.3.3 Pressure Dependence of Xe Partitioning According to the simple two-site partitioni ng model, variations in the observed xenon chemical shift can be analyzed according to Eq. (4-4). The observed variation in the 129Xe chemical shift with lipid concentrat ion is shown in Figure 4-5A, where denotes the difference in the observed chemical shift in aqueous suspensions of lipid vesicles (LUVs) and


77 Table 4-1. Experimentally determined 129Xe relaxation times ( T1) associated with the lipid-free buffer solution and the lipid phase in DOPC LU Vs and MLVs, in units of seconds (s). Literature values for aqueous dispersions of Egg PC vesicles (sonicated LUVs) are given for comparison. Errors are reported as standard errors of the estimate (SSE). [Lipid] (mM) MLVs lipid phase MLVs buffer phase LUVs Lipid-free solution DOPC 50 25 15 10 2 s 4 1 s 3 1 s 74 12 s 70 11 s 53 9 s 54 8 s 72 15 s 58 7 s 110 34 s Egg PC [14] 37 51 4 s 122 37 s Figure 4-5. Concentration dependence of the observed 129Xe chemical shift as a function. A) The lipid concentration depe ndence of the observed chemi cal shift performed over a large concentration range. Similar plots were made at various loading pressures (1-10 atm) in order to get an adequate profile. These measurements were performed at 5 atm in 1 ml of solution. Data points po ssessing significant error is a result of inaccurate loading. B) The Xe( g ) pressure dependence of the observed chemical shift for a range of lipid concentrations: 15 mM ( ), 25 mM ( ), 40 mM (), 50 mM (), and 100 mM (*). The open circles ( ) denote the pressure dependence of the lipidfree buffer solution. lipid-free buffer solution. This wa s repeated for a series of xenon gas overpressures. Figure 4-5B shows the chemical shift change as a function of [ Xe ]aq for a specified lipid concentration. Notice the distinct difference in the shape of the plot s; the chemical shift appears be approaching a 1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 (ppm) 100x10-3 75 50 25 0[DOPC] (M) 190.4 190.0 189.6 189.2 188.8 188.4 188.0 129-Xe (ppm) 40x10-3 35 30 25 20 15 10 5 [Xe]aq (M) a) b) A) B)


78 plateau with increasing lipid conc entration (Figure 4-5A) while the trend is distinctly linear with increasing [ Xe ]aq (Figure 4-5B). It has been suggested that a solutes lipid solubility is limited by its saturation of the aqueous solution.[150] According to this hypothesi s, one might expect the 129Xe chemical shift to reach a limiting value at high enough xenon overpressures. However, the linear nature of the chemi cal shift, as seen in Figure 4-5A, makes this unlikely for the pressure range utilized in this study. The change in the 129Xe chemical shift with increasing lipid concentration (Figure 4-5) is similar to the trends seen in xenonprotein studies and is characteristic of nonspecific interactions. The saturation value should be comparable to the limiting shift of xenon at a specified pressure. Xenon partitioning in lipid vesicle solutions was characterized as a function of [ Xe ]aq. The partition coefficient should be independent of the amount of dissolv ed xenon in an ideal-dilute solution. Previous studies have shown that Kp can decrease with increasing anesthetic concentration,[151] suggesting the possibility of non-ideal partitioni ng at higher anesthetic concentrations.[152] The infinite-dilute value of the partition coefficient ( Kp ) was determined by fitting y-intercept values ( ) extrapolated from Figure 4-5B to Eq. (4-4). The subsequent plot of against [DOPC], shown in Figure 46 allows for the extraction of both Kp and max values, where max is the infinite-dilution value of the maximum chemical shift difference between the lipid and buffer phases. Similar methods were applied to investigate the concentration dependence of xenon partitioning in POPC bilayers.[153] The variation of buffer obswith increasing [ Xe ]aq in the lipid-free buffer solution likely arises from xenon-solvent interactions in solution; the downfield shift obs erved with increasing temperature may reflect its decreased solubility with temperature. To account for any contribution this might have on the


79 Figure 4-6. Determination of the infinite d ilution mole fraction partition coefficient. The yintercept values of Figure 4-1 plotted as a function of lipid concentration result in a infinite dilution partition coefficien t of 390 46 when fit to Eq. (4-4). observed chemical shift associated with the lipid phase, we have subtracted this variation in order to more accurately differe ntiate between the bound and free xenon in solution. If the concentration of bound xenon is neglig ible with respect to the total lipid concentration, then the Kp[ Xe ]aq correction term in Eq. (4-4) can be safely neglected. However, this leads to an overestimation in the partition coefficient at higher xenon concentrations (Figure 4-7A). The standard ex pression utilized in the determination of the binding constant ( Ka) via NMR chemical shift data is similar to Eq. (4-4), where max[] 1[]a aKL KL (4-6) As described by Eqs. (4-7) and Ka 55.5M Kp if [ Xe ]bound is low enough; changes in the binding and/or partitioning behavior are signified by a changes in the slope. A comparison of partitioning and binding profiles is provided in Figure 4-7B. 1.1 1.0 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0 (ppm) 100x10-3 80 60 40 20 0[DOPC] (M)


80 bound a boundaqXe K L XeXe (4-7) 55.5bound p boundaqXeM K L XeXe (4-8) 55.5a bound p boundLXe KM KLXe (4-9) Note that the measured partition coeffi cient does not vary significantly with [ Xe ]aq over the pressure range used in this studythe observed slope is distinctly lin ear. As expected, the binding constant diverges at higher loading pressures. The binding be havior will be evaluated in further detail in Chapter 6. The predicted limiting shift difference, max extracted from the fits of the experimental data to both Eq. (4-4) and Eq. (4-6) are identic al, and serves to validate both equations. The linear behavior of the limiting chemical shif t difference allowed us to extrapolate the max Figure 4-7. The pressure dependence of the partition coefficient. A) Results from fits to Eq. (4-4): (blue ), Eq. (4-10): ( ), and Eq. (4-4) without the Kp[ Xe ]aq correction: (red ). B) The binding model, ( ), versus the partitioning model, ( ): the difference denoted by [ L ]-[ Xe ]bound and by [ L ]+[ Xe ]bound, respectively (Eq.(4-9)). Errors are reported as (SSE), resulting from the fit of chemical shift data to Eq.(4-4). a) b) 70 60 50 40 30 20 10 0[Xe]bound x [55.5M] /([L] [Xe]bound) 45x10-3 40 35 30 25 20 15 10 5 0[Xe]aq (M) 800 700 600 500 400 300Kp 40x10-3 30 20 10 0[Xe]aq (M) A) B)


81 Figure 4-8. The limiting chemical shift as a f unction of xenon concentration in solution. The limiting shift changes + 37 2 ppm for every M of [ Xe ]aq. The infinite dilution value of the chemical shift was found to be appr oximately 2.56 ppm. Errors are reported as (SSE). values in order to obtain [ Xe ]bound and thus Kp for all pressures. The experimentally determined max values are provided in Figure 4-8 wh ere it is plotted as a function of [ Xe ]aq. Now that and max are known for each [ Xe ]aq and specific lipid concentration, Kp can be determined using Eq. (4-10), where Xbound = / max. 55.5bound p boundbound aqXM K LLXXeX (4-10) As one can see in Figure 4-7A, the two different methods yield par tition coefficients which are in close agreement. For purposes of compar ison, we were able to refit Meiers[153] data to our model and determined the mole fraction partiti on coefficient of xen on in POPC to be Kp = 466 78. This can be compared to our experimentally determined result of 475 41 for DOPC under similar experimental conditions. Small errors in the extracted max values are propagated into 4.5 4.0 3.5 3.0 2.5max (ppm) 40x10-3 35 30 25 20 15 10 5 0[Xe]aq (M)


82 the overall errors for Kp having a large effect. The smaller error in our determination of Kp reflects the greater number of data points available for the fit. While their headgroup regions ar e identical, these DOP C and POPC differ structurally only in their acyl chain regions (Figure 3-1); the presence of the shorter, saturated acyl chain in POPC results in a reduction in the molecular volume.[154, 155] Thus, POPC has a slightly smaller interfacial area, effectively re ducing the number of water molecules bound per lipid molecule in fully hydrated lipids. According to theoretica l predictions from Cantor, partitioning depends strongly on lipid composition and degree of unsaturation.[156, 157] Recent experimental work, comparing the distributive proper ties of ethanol in fully satura ted DMPC and unsaturated DOPC, cite similar behavior.[119] It should be noted that the differe nce in our partition coefficients for DOPC and POPC are within error of each other. According to the statistical thermodynamic estimates by Cantor, the values for these tw o lipids are expected to be very close.[158] 4.3.4 129Xe Relaxivity Utilized to Probe Partitioning and Vesicle Stability Following previously established theory, NMR relaxation methods were used in conjunction with the two-site rapi d exchange model to confirm the Kp values determined by chemical shift analysis.[159] If the exchange rate of xenon molecules between the aqueous and lipid environments is fast on the NMR times cale, the observed spin relaxation rate ( R1 = 1 /T1) becomes a weighted average of the intrinsic relaxati on rates of xenon in each phase. According to this model, the average time (b) xenon spends in the bound stat e must be much less than the bound relaxation time, T1 ,bound.[160] This results in the following expression: 1, 1, 1, 1,1 1bound boun obs boun q d da obsRR RXX T (4-11) where R1 ,obs is the observed longitudinal relaxation rate of xenon in the fast exchange regime, while R1 ,bound and R1 ,aq are the spin-relaxation rates of xe non in the lipid and aqueous phases


83 respectively. Saturation recovery methods were used to determine R1 ,bound and R1 ,aq for the dissolution of 5 atm of xenon w ithin a 50 mM DOPC MLV sample. The two distinct resonances denoting the aqueous and lipid phases have markedly different relaxation times; the fit of the narrow upfield peak associated with free yields a T1 ,aq value of 74 12 s while the broader peak located downfield (bound) a T1 ,bound of 10 2 s. These relaxation times provide an estimation of R1 ,aq and R1 ,bound, while the analysis of the 50 mM DOPC LUV solution under similar experimental conditions allo ws for the approximation of R1 ,obs. Kp can now be calculated directly from Eqs. (4-10) and (4-11). The result of whic h is significantly less than expected, yielding a mole fraction partition coefficient of 66. In an attempt to make sense of this inconsistency, we reassess the initial assumption that the relaxation time of xenon within the lipid environment is much greater than its lifetime within the MLV state. This results in a slight modifi cation of Eq. (4-11) to account for the average binding time, b: 1, 1, 1,1 1obs a bound bou q bound nd bTT X X T (4-12) Here, we utilized the Xbound parameter extracted from chemical shift measurements and set the longitudinal relaxation times to prev iously said values. The extracted b value was found to be approximately 33 seconds. This relatively long lifetime within the lipi d phase is somewhat understandable in a MLV envir onment since there is a high likelihood for xenon to diffuse through multiple bilayers, with limited contact w ith the aqueous phase. Substitution of Eq. (4-11) into Eq. (4-10) also allows for an estimation of T1 ,bound under fast exchange conditions. Using Kitamura et al. (1999)[159] as a guide, we set T1 ,obs to 54 s, Kp to 475, and T1 ,aq to 110 s.


84 According to this model, the predicted longitudinal relaxation time of xe non associated with the 50 mM DOPC LUV environment under 5 atm of xenon gas is a pproximately 24 s. The NMR paramagnetic relaxation technique was employed to verify the stability of both the LUV and MLV DOPC solutions at elevated 129Xe loading pressures. The presence of paramagnetic ions has been shown to significantly increase the relaxation rates of nuclear spin systems in aqueous solution.[161-163] This method is based on assumption that paramagnetic ions are unable to pass through the lipid bilayer envi ronment. If there is a significant amount of vesicle rupture, the paramagnetic contribution to xenons relaxivity and chemical shift behavior cannot be distinguished between phases. As sh own in Figure 4-9, the paramagnetic induced chemical shift change appears to be limited to upfield resonance, while the broad resonance representing xenon dissolved with in the lipid phase remains unchanged. Thus, we can infer that R1,bound is unaffected and the paramagnetic contributi on to the longitudinal relaxation rate can be expressed as: 1, 1, 1,(1).pp boundboundboun obsaq dRXRXR (4-13) under fast exchange conditions, where 1,p obs R and 1,p aq R are the observed relaxation rates of xenon in 25 mM DOPC and lipid-free buffer solution in the presence of MnCl2, respectively. The severe broadening of the buffer resonance is an unfortunate si de effect of the paramagnetic induced relaxation process. The solubilization parameter ( Xbound) can be calculated by measuring the change in relaxation rates of xenon in both lipid-free buffer solution and LUV DO PC solutions, in the presence and absence of paramagnetic shift reagent. Together, Eqs. (4-11) and (4-13) generate a usable expression for Xbound:


85 1,1, 1,1, 2 11 1obsobs bound aqaq p pslope X s R lo Rpe R R (4-14) where slope 1 and 2 provided in Eqs. (4-15) and (416) signify the relaxa tion rate enhancement generated by the presence of pa ramagnetic ion in solution. 1, 1,1, 22 1aq aqaq pR slope Mn R Ml R ClnC (4-15) 1, 1,1, 22 2obs obs p bsoR slope MnCl R l R MnC (4-16) Consistent with theory, the paramagnetic enhancement of the longitudinal relaxation rate (R1, obs) increases with [MnCl2] for both lipid containing and lipid free solutions (Figure 4-10A). Though paramagnetic ions have the potential to impede xenon partitioning when present at sufficiently high concentrations, the linear natu re of the relaxation behavior s uggests this not to be the case Figure 4-9. 129Xe NMR spectra of xenon dissolved in 50mM DOPC MLVs, in the absence (red spectrum) and presence (blue spectrum) of MnCl2. The paramagnetic shift reagent is limited to the aqueous phase (upfield resonance). PPM 196 194 192 190 188 186 184 182


86 Figure 4-10. The variation in the NMR paramete rs with increasing paramagnetic shift reagent. A) Observed longitudinal relaxation rates of 129Xe in lipid-free buffer solution (R1 ,aq) and 25 mM LUV DOPC lipid suspension ( R1 ,obs) as a function of [MnCl2]. B) The variation in the 129Xe chemical shift with increasing [MnCl2] for 10 atm of xenon dissolved in lipid-free buffe r solution (referenced to buffer obs). Table 4-2. Extracted mole fraction partition co efficients obtained thr ough a variety of methods: chemical shift analysis ( Kp.1), direct measurement of longitudinal relaxation rates associated with the isolated lipid and aqueous phases in MLV solution ( Kp.2), and paramagnetic shift enhancement (Kp.3). Errors are reported as (SSE). PXe (atm) Kp,1 Eq (4-4) Kp,2 Eqs .(4-10)/(4-11) Kp,3 Eq .(4-17) 5 10 475 41 452 34 66 20 485 53 here. Eq. (4-14) is easily modi fied to make use of the MnCl2 titration data provided in Figure 4-10A. This allows for a more accurate dete rmination of the solubilization parameter; substitution of Eq. (4-14) into Eq. (4-10) provides a direct m easurement of the mole fraction partition coefficient, yielding a Kp value of 485 53. 1,1, 1,1,1,55.5aqobs p obsaqobs aqMRR K LRXeRR (4-17) 1.0 0.8 0.6 0.4 0.2 0.0obsbuffer (ppm) 3.5 3.0 2.5 2.0 1.5 1.0 0.5 0.0[MnCl2] (mM) a) b) 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0R1,obs (s-1) 10 8 6 4 2 0[MnCl2] (mM) 1.2 1.0 0.8 0.6 0.4 0.2 0.0R1,aq (s-1) A) B)


87 These results agree with previously determine par tition values using chemi cal shift analysis and confirms that the vesicles remain intact when exposed to high xenon overpressures. 4.3.5 Bunsen, Ostwald, and Mole Fraction Partition Coefficients The mole fractions partition coefficients of xenon dissolved in olive oil and various lipid solutions are summarized in Table 4-3 for comparison. This has been graphe d in Figure 4-12 for visualization and is listed by increasing molecula r weight of the biomedia (excluding lecithin). The Ostwald solubility coefficient of xenon in olive oil is often used to estimate its concentration in lipid bilayers and is easily converted into the mole faction partition coefficient. For clarification, the Bunsen coefficient, ( ) measures the volume of dissolved gas per volume of water or lipid sample at atmospheric pressu re and 273.15 K. We corrected for the sample temperature to allow for easy conversion to the Ostwald ( L ) coefficient, the details of which are shown in Eq. (4-18).[164] Like the Bunsen coefficient, the Ostwald solubility is a volume based measurement that relates the quantity of absorbed gas to the quantity of absorbing solvent at a given pressure and temperature. The concentr ation form of the partition coefficient, K(c) was then determined by lipid-to-water ratio of Ostwald coefficients (Eq. (4-19)), allowing us to obtain the mole fraction form of the partition coefficient, as shown in Eq. (4-20).[165] 273.15 T L K (4-18) oil c water L K L (4-19) oil W LL p c water WWLVM L KK VLM (4-20) Here, VL and VW denote the partial molar volumes, while Loil and Lwater symbolize the Ostwald coefficients of oil/lipid and water, respectivel y. When the molar volumes were not readily


88 available, we utilized molecular weight a nd density values found in the literature. The Meyer-Overton correlation between olive oil solubility and anesthetic potency suggests that the site of anesth etic action has similar physical pr operties to that of olive oil. While lipid theories assume that nonspecific anes thetic-membrane interactions is key, protein theories target specific interactions between proteins and receptors. However, neither proteins nor lipids have uniform structure; their anisotr opic nature suggests different physical properties compared to a bulk, isotropic apolar liquids A number of organic solvents have been investigated in addition to olive oil with hopes of establishing a more accurate chemical model to quantify the solubility-potency correlation. This has lead to the octanol-water partition coefficient as the most widely used model for lipophilicity.[166-168] The presence of the hydroxyl Table 4-3. The mole fraction partition coefficients of xenon in various lipidic species and solvents. LUV-large unilamellar vesicles; SUV-small unilamellar vesicles; SVsonicated vesicles; HCS-hydrocarbon solvent. Errors are reported as (SSE), resulting from the fit of chemical shift data to Eq.(4-4) Aqueous Media Vesicle type Temp. (K) PXe (atm) Kp-value Corrected Kp value max (ppm) Exp. method DOPC POPC[153] Lecithin[146] PC/PA[169] PC/PA/chol.[169] 2:1 PC/chol.[170] n-octane[171] n-octanol[103] Olive oil[172] LUV SUV SV SV SV SV SV HCS HCS HCS 298 298 297 310 299 298 293 293 298 298 1 5 6 1 1 1 1 1 1 1 461 584 597 804 636 481 445 326 205 945 446 42 475 41 466 78 2.8 0.2 3.3 0.2 3.5 0.4 This work This work NMR Ostwald Bunsen Bunsen Ostwald Ostwald Ostwald Bunsen


89 group in the octanol molecule intr oduces a small degree of polarity within the solvent, resulting in a better correlation.[173, 174] Despite the improvement, this model still fails to adequately characterize the interact ion of charged solutes, given that octanol can only interact through hydrogen bonding and hydrophobic interactions. As the focus of this thesis is on the partitioning behavior of a nonpolar solute, we refrain from discussing this fu rther and refer the reader to Razak et al. (2001)[175] and Escher et al. (1996)[176] for additional information. The primary difference between the hydr ocarbon-water model and liposome-water interaction is that the physical properties within the apolar solvents are uniform throughout while they vary as a function of memb rane depth in the lipid bilayer phase. To be more specific, the dielectric constant ( ) exists as a gradient, decreasing from a high value of approximately 76 in bulk water to about 2 in the lipid co re; the lower the numeric value of the greater the hydrophobicity.[127, 177-180] According to Ueda et al., the real meaning of the Meyer-Overton rule is that anesthetic molecules will tend to re side in regions where the dielectric constant is near that of the lipid/water and protein/water interfaces; the best correlation is when the relative permittivity is between 10-11.[181] The dielectric constants of various lipid and lipid free solutions are tabulated in Table 4-4 for comparison. An illustration is provided in Figure 4-11 for visualization. Excluding the high frequency permittivity (), values summarized in Table 4-4 were obtained at low frequencies. Molecular rotation can lead to a net loss in the permittivity which increases with frequency. As shown belo w, the dielectric constants associated with various regions within the bilayer are no l onger distinguishable at high frequencies. As mentioned previously, intermolecular 129Xe-{1H} NOEs have been utilized to quantify the xenon-membrane interaction strength for diff erent regions in the lipid bilayer; the intermolecular cross relaxa tion rate of xenon associated with th e interfacial choline protons are


90 Table 4-4. The dielectric constants of apolar so lvents, water, and a PC lipid system at 298 K, unless stated otherwise. The bound water re fers to water molecules in the immediate vicinity of the PC headgroup (HG) region. Here, denotes the permittivity perpendicular to the membrane surface. The ex istence of a dielectric constant in the xy-plane, II is consequent of the dipola rity in the HG region. Finally, represents dielectric decrement due to mo lecular motion (increases with frequency), and indicates the high frequency permittivity. Aqueous Media II Water Bulk[179, 182] Bound-to-membrane surface[183, 184] 76 20-30 6.7 ~71 ~47 1.8 Olive Oil (293 K)[127, 185] 3.1 1-Octanol (293 K)[127, 186] 10.3 Integral Membrane Protein[180] 6-8 PC lipid[179-184, 187] P-N headgroup Glycerol backbone Lipid core 30 8-10 2-4 20-35 ~10 140 0 2.2 2.2 2.2 Figure 4-11. The variation in the membrane dielectric constant as a function of membrane depth ( z) according to a simple capacitor model. The bound water (BW), lipid headgroup (HG) and hydrocarbon chain (H C) regions all serve as possible sources for the membrane dipole potential.[188] Here, denotes the relative permittivity, perpendicular to the bilayer normal (norm).[189] 2.9nm 0.8nm 0.5nm Membrane center plane ~80(z)HC Polar HG BW Layer ~11 ~2.5z


91 Figure 4-12. A graphical representation of xe non-oil/lipid partition coefficients, sorted by increasing molar mass. Bunsen and Ostwald co efficients were converted from values found in literature. References and numeric values are found in Table 4-3. The darker grey bars represent the corrected Kp values determined by 129Xe NMR chemical shift analysis. The error bars on the lighter grey regions are (SSE) when multiple values were available in literature. The Kp values were obtained within 297-298 K and at atmospheric pressure, unless stated otherwise. three times larger than that associated with the aliphatic region. The induced dipole on the Xe atom is due its close proximity to the polar lip id headgroup which makes it more adaptable to the interfacial region than the lipid core.[190] And so, according to Uedas theory, the partition coefficient should be closer to that of 1-octanol than olive oil. However, the interfacial properties of the lipid membrane can be adjusted via ch anges in the surface charge density, facilitated through modifications in lipid membrane compositi on (e.g. lipid type, cholesterol) or external variables. As such, changes in the interfacial properties may manifest in partitioning behavior. High pressure (> 50 atm) xenon-membrane studie s showed that the net effect of gas on membrane order results from a two-step process.[191] Following a sudden increase in the xenon pressure, the presence of xenon was shown to increase the relative order of the membrane n-octane n-octanol PC/PA 2:1 PC/PA POPC DOPC DOPC Lecithin Olive oil /chol PC/chol (310 K) Media Composition K p


92 environment, due to a membrane compression eff ect. However, the gas molecules were able to diffuse and dissolve into the lipid core, eff ectively decreasing th e order parameter once equilibrium is established. Whats more, de spite an initial increas e with xenon uptake, the relative permittivity was shown to gradually decr ease as the xenon diffused through the bilayer. Since high pressures are involved in this study, it is unclear if these effects are relevant under physiologically conditions. Taking a closer look at Table 43 (or Figure 4-12) we see that the lipid/water mole fraction partition coefficient of xenon dissolved in variou s compositions of lipid vesicles spans between 445 and 636, yielding maximum and minimum valu es of 945 and 205 in olive oil and octanol, respectively. This has been graphed in Figure 4-12 for visualization a nd is listed by increasing molecular weight of the biomedia (excluding lecithin). Assuming the degree of hydrophobicity is proportional to the experimenta lly obtained partition coefficients neither octanol or olive oil models fully describe the partitioning of xenon be tween the aqueous and lipid phases. And while the Ostwald and Bunsen solubility coefficients for the lipid solutions are comparable to our uncorrected NMR partitioning values under 5 atm of xenon gas, they are not entirely equivalent to the corrected quantities. Th is is in part because the B unsen and Ostwald methods for determining partitioning essentially measure the dissolution of gas into two bulk phases (aqueous solution and concentrated lipid) and compare them This is reminiscent of xenon partitioning in the MLVs, where the lipid phase is saturate d by xenon (Figure 4-1). While these methods provided a good estimation of the maximum part itioning capacity of bul k lipids, the solventsolute interactions may not necessarily be the same as those in smaller, homogenously dispersed systems.


93 Cell membranes typically contain cholesterol, whose presence contributes largely to the control of the membrane diffusive properties. Si milar to other partitioni ng studies, cholesterol appears to decrease th e partition coefficient.[192, 193] The addition of cholesterol is known to induce tighter molecular packing within the cell membrane, which has consequence on the membrane permeabilitythe higher the cholesterol content, the lower the uptake of anesthetic. According to the solubility diffusion mechanism, the solute molecules first partition into the hydrophobic phase and then diffuse across the lipi d bilayer. Assuming xe non permeates through the lipid bilayer according to this method, we can relate the decreas e in permeability to a decrease in partitioning, consistent with the obser ved trends. In contrast, elevated temperatures appear to enhance partitioning (l ecithin at 310 K). This has been verified in the absence and presence of cholesterol.92 The amount of anesthetic associat ed with the lipid phase has been shown to steadily decrease with incr eased concentration of cholesterol. 4.4 Conclusions The nature of xenon-phospholipid interactions and xenon exchange depend on the structure of the lipid headgroups and acyl ch ains, the phase state of the lipid bilayer, and the heterogeneity in both vesicle size and overall distribution of lip ids with external vari ables. Our study of the dissolution of xenon into aqueous solutions of MLVs and LUVs provide a reliable basis for understanding the 129Xe exchange dynamics within DOPC lipid membranes. What is clear is that not all lipid systems exhibit the same trends. For instance, both DMPC and DPPC MLVs have been shown to start as a single peak at room temperature then split into two phases with increasing temperature.32 Miller et al. also looked at the dissolution of xenon in DMPC and saw a change in the exchange process from sl ow to fast with increasing temperature.[10] The primary difference between these lipids those that we have studied is that DPPC and DMPC are saturated


94 lipids, while DOPC is unsaturated. The majority of studies investiga ting hydrophobic cavities of proteins in solution use a slightly more complex model to differentiate between specific and nonspecific interactions. Non-specifi c interactions take place at the protein surface, while specific interactions consist of those inside a hydrophobic pocket. This treatment may prove to be useful in the study of lipid domains (a.k.a. lipid rafts) to determine localized effects of xenon-lipid interactions. 129Xe NMR studies of non-speci fic xenon-lipid interactions in the lipid system will help to develop 129Xe NMR as a biomolecular probe of packing effects, chemical exchange, factors affecting lateral pressure, and phase tr ansitions in membrane model systems. Can we disentangle the contributions from the acyl chai n regions and those from the lipid headgroup? Which has a larger contribution to the observed chemical shift? Now that the basic theory is established and validated, it will be possible to extend this appr oach to fully characterize the thermodynamic properties in other lipid systems.


95 CHAPTER 5 THERMODYNAMIC PROPERTIES OF PARTITIONING 5.1 Introduction Historically, an anesthetics hydrophobicity was thought to dominate its interaction with biologically relevant media. Th e linear correlation between an anesthetics potency and its relative solubility in hydrocarbon solvents, as shown by Meyer and Overton,[86, 194] largely contributed to this view. However, it has beco me increasingly apparent that the predictive behavior provided by the Meyer-Overtone (MO) Rule may not adequa tely relate lipid solubility and anesthetic potency.[157, 195-198] Several exceptions include th e existence of non-immobilizers (molecules predicted by the MO rule to be anesth etics, but arent) and observed variations in solubility with temperature and lipid composition. The partitioning behavior of anesthetic molecules is traditionally thought to be driven by the hydrophobic effect. Mo re specifically, it is believed to follow a similar thermodynamic mode l of partitioning as observed between aqueous solutions and hydrocarbon solvents. Herein we hope to establish whether the Meyer-Overton Rule is applicable to xenon-bi layer interactions and determine the extent the hydrophobic effect contributes to the pa rtitioning process. Two important thermodynamic functions are the standard enthalpy and the standard entropy. The standard entropy accounts for the re lative change in membrane order, while the standard enthalpy measures the energy of the solu te-solvent interaction in solution. According to general statistical mechanical theory, the solvation energy depends on the kinetic energy, the solutes rotational and vi brational properties, and the potentia l energies due to nearest neighbor interactions.[199-201] To our benefit, these complex f unctions are simplified for a noble gas containing apolar solvent as th e rotational and vibrational cont ributions disappear due to the spherical symmetry of these small solute molecules. Unlike simple solutions, liposomes are


96 Figure 5-1. Thermodynamic solvation parameters. A) For 1 atm of xenon from the gas phase to water. B) From the gas phase to the so lvent hexane at 298 K. Numeric values provided by Hefter et al. (2003) and Boni fcio et al. (2001).[202, 203] -12 -10 -8 -6 -4 -2 0 2 4 6 kcal / mol 700 650 600 550 500 450 400 350 300 Temperature (K) H T S THTS -6 -5 -4 -3 -2 -1 0 1 2 3 kcal / mol 330 320 310 300 290 280 Temperature (K) H T S Temperature (K) A) B) Temperature (K)


97 comprised of three phases: bulk water, non-solv ent water (strongly bound water molecules at the interface), and a solvent-like hydrophobic core. It may possible to monitor solute induced structural changes at the lipid membrane-water interface using these th ermodynamic parameters. Does preferential hydration of the solute at th e lipid-water interface di rect the distributive properties of non-polar molecules with temperat ure or modifications in membrane order? In this chapter we use 129Xe NMR chemical shift analysis to directly measure the thermodynamic partitioning parameters of xeno n in DOPC. The extracted molar enthalpy, entropy and chemical potential are then compar ed with model data for xenon partitioning in several hydrocarbon solvents. We begin by presenting a more detailed picture of the hydrophobic effect as related to the solvation model. This is then expanded to include a brief summary on the theory behind partitioning thermodynamics; specifi c relations utilized in our own analysis are highlighted. Lastly, we discuss our results in the context of the recent lite rature and confirm that our analysis of experimentally determined chemi cal shifts (Chapter 4) is a valid approach in thermodynamic analysis of the xenon-membrane interaction. 5.1.1 The Classic Hydrophobic Effect The hydrophobic effect describes the anomalous behavior of non-polar solutes in aqueous solutions and their tendency to form aggregates of like molecules in solvent water. Hydrophobic effects are characterized by two distinct processes: hydrophobic hydration and hydrophobic interactions. The molecular in terpretation of the hydrophobic hydr ation is associated with structured water molecules in the presen ce of nonpolar solutes in aqueous solution.[204-206] The dissolution process depends on the capacity of the solute to s ubstitute for the hydrogen bonds lost by the water network. In the presence of a nonpolar molecule, the water molecules will reorganize in an attempt to minimize the loss of en ergy, resulting in more ordered structures, or water cages, around the solute. This leads to an enthalpic gain as well as a entropic loss. Thus,


98 the whole process is energetically unfavorable and the nonpolar solu te will move to a friendlier environment when available. This entropic effect is considered to be the dominant force behind a solutes hydrophobicity (compared to the enthalpic factor) and provi des the basis for the classical interpretation of hydrophobic interactions; non-polar groups aggregate in such a way that minimizes their contact area with water.[207] The thermodynamics of transfer for xenon from is gaseous phase Xe(g) to a bulk solvent differs substantially depending on the molecular natu re of the solvent. For example, the solvation energy of Xe(g) to an apolar solvent is spontaneous a nd dominated by a large, negative enthalpy and moderately small entropy; both thermodynamic parameters should be relatively independent of temperature. The mixing of nonpolar molecules in water exhibits markedly different behavior. The simplest thermodynamic description of the hydrophobic hydration is found in the transfer of a nonpolar solute from a gas or liquid phase, to bu lk water. This process is identified by i) a large, negative entropy (unfavorable) associated with the release of the water molecules from around the solute molecule, and ii) a temperatur e dependence of the chemical potential that yields a large, positive heat capacity.[208, 209] The solvation entropy a nd enthalpy of nonpolar solute from the gaseous phase to bulk water incr ease with temperature, consequent of the large change in the heat capacity. In comparison, th e solvation energy is large, positive and rather insensitive to changes in temperature due to the compensatory nature of the entropy and enthalpy parameters. The solvation energy profile is r eaches a maximum at a critical temperature, TS; it has been shown that the transfer process is most favorable when the entropic contribution to the solvation energy becomes negligible.[210] As such, the entropy contribution to the transfer energy is optimal at temperatures above TS. Using similar notation, TH is the temperature above which


99 the enthalpy of transfer becomes unfavorable, reflecting the weakening and/or breaking of the hydrogen bonded water network. These effects are illustrated in Figure 5-1 for clarity. The change in the heat capacity ( Cp) upon the transfer of nonpolar solute from bulk water to hydrocarbon solvent is typically large and negative in value. It increases with temperature and is thought to arise from the melting of the ordered water.[211] It was once thought that the waterordering effect was responsible for the low solu bility of nonpolar solutes in aqueous solution; however it is now known that th is occurs because it makes th e interaction more favorable.[211] The variation in the Cp value with changing molecular envi ronments has been found to be directly proportional to the number of water molecu les in the first hydration shell; the larger the size of the embedded hydrophobic surface ar ea, the higher the degree of hydrophobic interaction.[212] The negative contribution to the heat capac ity change arises from the solvation of nonpolar molecules within like envi ronments and provides a measure of the strength of solventinduced attractive forces.[213] General expressions for the variation in these thermodynamic parameters with temperature and their respective relations to the heat capacity will be provided in subsequent sections. Thermodynami c parameters for the transfer of Xe(g) from its gas phase to water, and several simple hydrocarbon so lvents are provided in Table 5-1. 5.1.2 Environmental Swap Energy (ESE) Now that we have some sense of the energies at play in gas-to-sol ution solvation, it is useful to describe the processes involved in the transfer of solu te from an aqueous to apolar phase. According to the Ben-Naim model of solvation thermodynamics, a solutes solubility is dependent on the reversible work required fo r cavity creation and the solute-induced forces upon insertion at a given temperature, pressure and chemical composition.[214, 215] As mentioned earlier, water molecules tend to orient themselv es around the solute, decreasing the entropy of

PAGE 100

100 Table 5-1. Tabulated literature values of ther modynamic parameters of transfer for 1 atm of Xe(g) to both water and hydrocarbon solutions. Here, A denotes experimentally obtained values of the solvation energies,[171, 216] while B indicates calculated Ben-Naim[217] standard enthalpy, entropy and solvation en ergy changes using stat istical methods and C denotes values determined from scaled particle theory.[216] Gaseous phase transfer -tosimple solvent Method Temp. (K) (kcal/mol) m H (kcal/mol) mS (cal/molK) Xe(g) : H2O A B [211] 273 293 298 +2.36 +1.23 +1.58 -4.05 -4.32 -4.71 -18.00 -18.90 -21.10 Xe(g) : 44 solvents A 273 -0.63-1.84 -4.10 Xe(g) : n -C8H18 Ben Naim method Molar concentration scale Mole fraction scale B C A,C 293 298 298 -0.86 -0.63 1.87 -2.12 -1.74 -2.04 -4.30 -3.57 -13.12 the system. These solvent cages are broken upon solute removal, resu lting in a loss in structured water and a positive change in the standard entropy. The removal of solute from the aqueous phase also triggers reforma tion of the water hydrogen-bonded network, yielding a negative change in the enthalpy. In the same way, work must be done in order to form a cavity of appropriate size within the apolar solvent (positive enthalpy) at which point the solute-solvent interaction yields negative contributions to the standard enthal py and entropy. Thus, the energy of transfer from the aqueous to apolar phase may be viewed as:[218, 219] solnsolnsolncavcav intinthcaq hcaq hcaq (5-1) where soln denotes the energy change of transfer from the aqueous ( aq) to hydrocarbon ( hc ) phase, cav the work required to form a cavity, and int reflects changes in both solvent reorganization and the relative strength of the solute-solvent interaction. Here, the cav

PAGE 101

101 Figure 5-2. Graphical desc ription of the solute transfer process. A loss of a solute from water is expected to be accompanied by an increase in entropy, while a negative change in enthalpy is accompanied with filling the cavity with water. Lipid bilayers are more organized structures than apolar solvents and thus would require more energy to form a suitable cavity in the lipid phase. The van der Waals interaction associated with the hydrocarbon phase would not be significantly differe nt from that observed in apolar solvents. If present at sufficiently high levels in the membrane, solute molecules may increase the average distan ce between lipid molecules, resulting in positive enthalpy and entropy values. [220-222] contribution to soln is significantly greater than the solute-solvent term ( int) as the energetic cost of cavity formation is higher than van der Waals in teractionsit depends strongly on the molecular size and density of the solvent.[223] A general scheme describing the solute transfer process is provided in Figure 5-2. While Eq. (5-1) pr ovides valuable mathematical model for the solute transfer process, we are unable to experimentally confirm the individual contributions of cav and int However, the partition coefficien t can be used as a structural Cavity creationhc cav H S hc int H S Cavity creationw cav H S Solvent-solute interactionsw int H S Solute in gas phase solute in water solute in lipid soln bulk water unperturbed lipid bilayer+ + +H, -S +H, +S H, -S H, -S + + w cav w int + hc int + hc cav soln

PAGE 102

102 tool to monitor changes in lipophilicity parameters via changes in the free energy, enthalpy and entropy of transfer and does not requ ire knowledge of the solvent density. Standard States. The partition equilibrium is influenced largely by the solutes lipophilicity which is a measure of the capacity of a solute mo lecule to solvate into the hydrophobic region of the lipid bilayer. The higher the lipophilicity, the d eeper the solute is embedded within the hydrophobic core. The differe nces in the structural properties between simple binary solutions and aqueous dispersions of lipid vesicles have been discussed previously. The inhomogeneous, anisotropic nature of the bila yer phase has the poten tial to exclude solute molecules due to packing restraints caused by th e incorporation of solute into the acyl chain region. As such, changes in the partition equilibrium ensuenow multiple equilibrium process exist between the dissolved solute, undissolved so lute near the lipid-water interface, and bulk Figure 5-3. A simple schematic depicting the di fferences between several types of partitioning processes. A) The dissolution of Xe( g ) into single component, water (aq) or apolar solvent (solv). B) The net result of the Ostwald and Bunsen solubility ratios (Loil/Lwater) between two distinct phases. C) Grap hical representation of the dispersion of LUVs. D) The heterogeneous environmen t of the MLVs that gives rise to the limiting chemical shift of xenon associated with the lipid environment: subscripts (I1) and (I2) signify the multiple interfacial phases that can exist in this kind of lipid suspension. A) B) C) D)

PAGE 103

103 aqueous solution. The differences between severa l types of partitioning processes (bulk solution, binary solution, LUVs, and MLVs) are illustrated in Figure 5-3 for clarity. When a gaseous solute (such as an inhalation anesthetic) is introduced to simple biphasic system, it is distributed over the gas (g), wa ter (w) and hydrocarbon (s) domains (Figure 5-3B). At thermodynamic equilibrium the chemical potential of the gaseous solute in each phase are equal,[224] where g ws (5-2) The individual chemical potentials of gaseous solute in each phase can be written as: ln g ggRTfP (5-3) lnww w w R TX (5-4) ln s ss s R TX (5-5) Assuming ideal gas and solution theory, the fugacity ( f ) and activity coefficients (w = s) can be approximated to unity at physiologi cally relevant temperatures and pressures. Here, the standard state of solute in the gas, aqueous and solvent phases are denoted by g w and s respectively; Pg signifies the partial pressure of the gas, while Xw and Xs refer to the mole fractions of gaseous solute in the aqueous a nd hydrocarbon solvent phases. The mole fraction form of the Ostwald solubility co efficient accounts for the relative distribution of gaseous solute between phases through population ratios. Manipulati ons of Eq.(5-3), (5-4), and (5-5) result in the following expressions: 1lnwg gwRTPX (5-6)

PAGE 104

104 2ln s gg sRTPX (5-7) where the chemical potential differences 1 and 2 represent the energies of transfer from the gaseous solute to the water and solvent phase respectively. These are also known as the solvation energies, which possess both enthalpic and entropic contributio ns. The transfer energy of gas between the water and hydrocarbon phases can now be related by Eq. (5-8), where: 21ln s w R TXX (5-8) The environmental swap (or transfer) energy is now represented by which is easily related to the mole fraction form of the partiti on coefficient and is equi valent to the Ben-Naim notation, soln As will be discussed in more detail in th e next section, the possibility exists that Pg, Xw, and Xs are temperature dependent Unlike the hydrocarbon solvent, the degree of water-lipid interaction at the bilayer interface depends largely on the phospholipid headgroup. While th ese hydrating water molecules experience less hydrogen bonding than those in bulk solution, they have been shown to possess large dynamic orie ntational order.[225] The relative degree of order depends on how close the water molecules are to the lipid-w ater interface. For PC lipids, simulation studies suggest that approximately 22 water molecules per PC lipid[226] are required for complete hydrationbut only 0.5-3 of those water molecu les are actually tightly bound.[227] Furthermore, this structured effect is only thought to extend 7 from the out ermost part of the PC headgroup. As stated previously, these waters lose thei r structural properties at elevated temperatures and eventually adapt the behavior of the bulk so lution. This effective dehydration at the lipid interface may vary between lipid systems and have an effect on the partition coefficient.

PAGE 105

105 5.2 The Partitioning Model as a Function of Temperature 5.2.1 Determining Kp( T ) by Chemical Shift Methods The temperature dependence of the chemical shifts of xenon in LUVs, MLVs and lipidfree buffer solution allowed us to extract thermodynamic propertie s such as enthalpy, entropy, and the heat capacity corresponding to the transfer of xenon from bulk solution to the lipid. Once it was established that the partition coefficients determined from Eq. (4.6) were comparable to values obtained from Eq. (4.8), and that the lipid phase observed in the MLVs correlates well to chemical shift of xenon saturated by lipid, we were able to manipul ate these properties in order to examine the role hydrophobic forces play in the xenon-membrane interaction. More specifically, we utilized the following equation, 55.5p aqM K XeLL (5-9) where = / max, [ L ]is the total lipid concentration, and [ Xe ]aq is the amount of xenon in aqueous solution. As mentioned previously, higher [ Xe ]aq is required in order to obtain the two distinct chemical shifts representative of the two partitioning phases in DOPC MLVs. Thus, experiments were performed at an overpressur e of 5 atm and a DOPC concentration of 50 mM. Xenons solubility in pure water decreases at el evated temperatures, so minor corrections were made to the [ Xe ]aq value in Eq. (5-9) utiliz ing solubility parameters available in literature.[171, 228] A simple relation is needed to determine the change in the chemical potential ( ) when 1 mole of xenon is transferred from bulk solution to the membrane: ln()p R TKT (5-10) mm H TS (5-11)

PAGE 106

106 ln / ()/pmm R R THT KS (5-12) where R is the gas constant, T is the temperature, and Kp( T ) the temperature dependence of Kp. The enthalpic and entropic components of pa rtitioning are related via Eq. (5-11), where m H and are the partial molar enthal py and entropy of transfer, both of which possess the capacity to be temperature dependent. The combination of Eq. (5-10) and (5-11) yields the vant Hoff equation (Eq. (5-12)). Plotting ln Kp( T ) versus 1/ T allows for the extraction of m H and mS : mTH is given by the changes in the slope. 5.2.2 Influence of Partitioning Units on Calculated Transfer Energies In order to get a sense of how our extracted thermodynamic parameters compare to similar systems, we searched the literature and tabulated changes in the chemical potential, enthalpy and entropy due to the transfer of xenon from aqueous solution to several organic suspensions. The results are summarized in Table 5-1. The molar chemical potentials at each given temperature were determined by converting the vo lume fraction partition coefficient ( Kc) to the mole fraction partition coefficient ( Kp). Details of the convers ion process can be found in section 4.3.3 of the previous chapter. The choice of partitioning units has been shown to affect the ESEs.[229-231] The principle thermodynamic transfer function of Xe(g) is dependent on whether the molar concentration, molality or mole fraction partiti on coefficients are used. According to results published by Jung Hag Park et al. (1989), the energy of transfer ( 2) for Xe(g) to n -alkanes can be either positive or negative depending on which form of the partition coefficient is used (see Table 5-1).[216] According to our results, the mole fraction form of describing the partitioning of xenon from aqueous solution to DOPC lipid vesicles, yields a numeric value of 3.6 kcal/mol at 298 K. Making use of the molar c oncentration partition coefficient yields a lesser

PAGE 107

107 value of -1.39 kcal/mol. Careful attention must be paid when comparing partitioning values in literature. We have adopted the mole fracti on partition coefficient as it does not require knowledge of the molecular volume of the lipid phase. All tabulated values have been converted to the mole fraction partition coefficient using methods presented in Sec tion 4.3.5. We have already demonstrated the potenti al error introduced when the contribution of bound xenon to the partition expression is neglected when working at elevated pressures. It is likely that overlooking the contribution of xenon to the overall volume of the lipid phase will introduce a similar error. 5.3 Results and Discussions 5.3.1 The vant Hoff Plot The vant Hoff plot describing the partitioning behavior of 5atm of xenon in 50 mM DOPC is given in Figure 5-4. First and fore most, there is a clear change in ln Kp with temperatureit increases monotonically until a maximum is reache d, at which point the slope changes. The observed curvature reflects changes in the heat capacity. Cp can be approximated by monitoring the standard molar enthalpy as a f unction of temperature (Eq.(5-13)). pmH CT (5-13) For clarification, the dashed line in Figure 5-4 is the vant Hoff (Eq. (5-12)) equation fit to the first three data points ignoring Cp( T ) effects (curvature). The solid line is the least squares fit of our partitioning data to Eq. (5-12). [203, 232, 233] We have tabulated environmental swap energies for xenon partitioning in various binary soluti ons for comparison. The extracted thermodynamic parameters are listed in Table 5-2. When co mparing the magnitude of the molar chemical potential of transfer between solutions, we see th at it is relatively consistent over the whole range of lipid/oil species found. In addi tion to being all negative, and thus spontaneous, there is very little variation in the actual numeric value of the molar chemical potential. This is can be seen

PAGE 108

108 Figure 5-4. The vant Hoff plot showing th e temperature dependence of xenon-membrane partitioning for 5 atm of xenon dissolved in 50 mM of DOPC. Here, ( ) signifies the maximum partitioning temperature and ( ) denotes the experimentally determined values from chemical shift analysis. The so lid line is the non-linear least s quares fit of the vant Hoff equation to all data points.[203, 232, 233] The dashed line reflects the fit of Eq. (5-12) to the firs t three data points. more clearly in Figure 5-5. Whats more, al l enthalpy and entropy valu es are positive, in accordance with the classical hydr ophobic effect. While we can confirm that xenon exhibits typical lipophilic thermodynamics, interpreting th e significance of the thermodynamic functions is more complicated. As seen in Figure 5-6, the molar enthalpy of tr ansfer decreases with increasing temperature; it rema ins endothermic under 317 K and is exothermic above it. The sign of ()m H T is a good quantitative measure of how the xenon distribution changes between the lipid and water phases with temperature; TH reflects the temperature at which water reorganization effects are energetic ally balanced with solute-sol vent interactions. Our results suggest that xenons affinity to the lipid phase increases even though the solubility decreases in the aqueous phase. The subsequent decrease in the relative magnitude of the molar enthalpy of 7.6 7.4 7.2 7.0 6.8 6.6 6.4 6.2 ln Kp(T) 3.35x10-3 3.30 3.25 3.20 3.15 3.10 3.05T -1 (1/K) max

PAGE 109

109 Figure 5-5. Graphical re presentation of tabulated data (Table 5-2). DOPC* indicates parameters resulting from the fit to Eq. (5-13), POPC* are parameters obtained from the vant Hoff plot using corrected Kp values, POPC (no asterisk) are explicit values obtained by Meier (2006).[153] Thermodynamic partition parameters were determined under 1atm of Xe(g) overpressure, unless stated otherwise. Figure 5-6. The enthalpy of solute transfer as a function of temperature for 50 mM DOPC under PXe = 5 atm of overpressure: the sl ope provides an approximation of Cp (see Eq. (5-13)). 10 5 0 -5 Hm(T) ( kcal/mol ) 325 320 315 310 305 300 Temperature (K) TH Temperature (K) Solvents(avg) Olive oil DOPC DOPC* POPC POPC* PC/PA PC/PA/chol T Sm Hm

PAGE 110

110 Table 5-2. Extracted thermodynamic parameters fo r the transfer of xenon from solvent water to various lipid phases and apol ar solvents: superscript a denotes experimental values taken from Figure 5-4, b refers to uncorrected partitioning values cited directly from source, while c represent vant Hoff equation fit to two points utilizing corrected Kp values (our model), and d calculated from variable temperature Bunsen and Ostwald solubility coefficients provided in literature. The standard errors of the estimate (SSE) were less than 10% (this work). H2O : DOPC 50 mM PXe (atm) Temp. (K) (kcal/mol) m H (kcal/mol) mS (cal/molK) pC (cal/molK) This work 5a 298 303 308 313 318 323 -3.64 -3.87 -4.11 -4.24 -4.33 -4.35 10.9 10.1 4.32 1.36 -2.45 -3.22 48.8 46.1 27.4 17.9 5.91 3.52 -601 This work 5c 298-3.5911.450.3 -715 H2O : POPC* H2O : POPC[153] H2O : PC/PA[169] H2O : PC/PA/Chol[169] H2O : DMPC[234], [235] H2O : olive oil[169] H2O : 44 solvents[171] 6c 6b 1d 1d 1b 1d 1d 297 297 298 298 303/0 298 298 -3.62 -4.20 -3.82 -3.66 -4.25[234]-4.06 -1.93 9.66 3.75 3.40 3.89 2.11[235]4.49 2.21 44.6 -1.51 24.1 25.3 28.7 13.9 -95 transfer with temperature mean s that xenons interaction w ith the lipid phase becomes increasingly favorable at elevated temperatures The change in the molar heat capacity due to transfer (fit of Figure 5-6 to Eq. (5-13)) was found to be -601 50 cal/mol K. The large negative

PAGE 111

111 change is not surprising as it is a thermodynamic signature of strong hydrophobic interactions. For comparison, the heat capacity of xe non in water is approximately 60 cal/mol K.[236] As can be seen from Table 5-2, xenon tran sfer from bulk water to lipid membranes is driven by entropy change and is endothermic ( H > 0) at ambient temperature. According to trends reported in literature, pa rtitioning into liposomes can be driven by either enthalpic or entropic effects.[110, 237-242] For example, partitioning into gel phase liposomes have been shown to be governed by changes in entropy, while lipi d membranes in the liquid crystalline phase tend to be driven by enthalpy change. Experimental re sults suggest that more structurally organized liposomes consisting of saturated lipids, non bilayer lipids, and/or cholesterol, possess a much larger entropy contribution to the transfer pro cesses. This can be explained on the grounds that more rigid membrane structures require more therma l energy to create a sufficient void for solute transfer. This is confirmed by the slight ch ange in the thermodynamic parameters between H2O:PC/PA and H2O:PC/PA/cholesterol partitioning systems (see Figure 5-5). Notice the significant difference between the molar enthal py and entropy values for DOPC and POPC lipid vesicles compared to other lipi d/oil systems. Recall that the thermodynamic parameters of DOPC and POPC are measured under 5 and 6 atm of xenon overpressure, respectively. The thermodynamic parameters of all other so lutions are determined at 1 atm of PXe (see Figure 5-5). While xenon is a rather small solute molecule, it may perturb the membrane structure if present at sufficiently high levels in the membrane. An increase in the average distance between lipid molecules will result in larger, pos itive enthalpy and entropy values. As discussed earlier, POPC has one shorter, sa turated chain which increases the fluidity of the bilayer. Despite this difference, the therm odynamic parameters of transfer between water and the DOPC/POPC environments are nearly iden tical (utilizing the mole fraction partition

PAGE 112

112 coefficient). The only thing that can be said for sure is that xenon partitioning appears to increase with molecular weight of the lipid/oil phase. Th ese observations are simila r to those observed by Simon et al. and their investigations of the halothane-lipid interaction.[243] When comparing the Kp( T ) behavior in a variety of lipid systems, they found the partition coefficient to be insensitive both to acyl chain length and degree of satu ration. However, a four-fold increase in Kp( T ) was observed at the dipalmitoyl-phosphocholine (DPPC) main phase tran sition temperature. This was attributed to the reduced entropic and enthalpic contributions due to tighter acyl chain packing in the gel phase which possesses more order and fever van der W aals interactions. Recent molecular dynamics simulations studying th e role of lipid membranes on anesthesia may be helpful in the interpretation of our data.[15] While these studies we re not performed at variable temperature, it does provide a sense of the effect increased anesthetic concentration has on the bilayer structure. We recognize that the thermal changes we have introduced on our system are likely to perturb the whole system, wh ile anesthetic doping is li kely to induce changes more locally. Despite this, Stimson et al. saw an increase in membrane thickness and area per lipid with increasing xenon concen tration. It also exhibited bimodal distribution between the lipid interface and the hydrophobic core. However th e marked preference for the inter-leaflet space is at odds with experimental results.[14] A slight increase in the deuterium order parameter suggests ordering of the acyl chai ns with increasing xenon concentr ation. Thus, our decrease in ()mST at higher temperature might be partially explai ned in terms of the increased partitioning. 5.3.3 Enthalpy-Entropy Compensation The compensatory effect of the enthalpy and entropy as described by Eq. (5-12) is shown clearly in Figure 5-7. The filled circles are the en ergy values extracted by combining Eqs. (5-9)

PAGE 113

113 Figure 5-7. Variations in the molar enthalpy and entropy values with temperature. Closed circles ( ) are the standard enthalpy and entropy values obtai ned from experimental data. The open circles ( ) show the relative location of TS and TH. and (5-12). As a reminder, the temperature depe ndence of the mole fraction partition coefficient is known from chemical shift data. This is then substituted into the vant Hoff equation changes in the slope and y-intercept values reflect changes in the molar enthalpy and entropy of transfer, respectively. Stronger mo lecular interactions between solute/solvent molecules result in a reduction in the configurationa l freedom of the system, which in turn decreases the entropy. Similarly, weaker interactions lead to an increase in the entr opy. Evaluation of Figure 5-4 and Figure 5-7 allowed for the determination of th e upper and lower limits of partitioning and the chemical potential, respectively. The maximum possible entropic cont ribution to the molar chemical potential, mS(317 K), is determined to be -5.18 kcal/mol. This occurs when the enthalpy is equal to zero. Accordingly, the maximum enthalpic contribu tion occurs at 324 K, with a o m H (324 K) value of -4.89 kcal/mol. Just as o m H (324 K) denotes the minimum in the 14 12 10 8 6 4 2 0 -2 -4 -6 Hm(T) (kcal/mol) 15 10 5 0 T Sm(T) (kcal/mol) TS TH

PAGE 114

114 molar chemical potential for the transfer of xenon from the wate r to the lipid phase, o mS (317 K) defines the upper partitioning limit. Similar to th e molar enthalpy of transfer, the change in the molar entropy is also positive and changes with temperature. The reduction in the overall degree of order with increasing temperature is likely two-fold. As the temperature is increased i) the structured water molecules surrounding xenon become more like bulk so lution, increasing the entropy associated with the aque ous phase, while ii) van der W aals interactions between the xenon and the acyl chains may also affect the order of xenon associated with the lipid phase. It is possible that both of these effects are manifest in o mS ( T ). 5.4 Conclusions Herein we proposed a simple method to extr act the thermodynamic parameters associated with the transfer of the noble gas xenon between two aqueous phases. By relating the temperature dependence of fast exchanging LUVs to the slower MLV system and assuming the resonance associated with the lipid phase (see Figure 4-1) is represen tative of the maximum chemical shift difference of xenon dissolved be tween phases, we manipulated the two-site exchange model to extract pertinent parameters According to the mole fraction form of the partition coefficient, the energy of transfer is characterized by a positive enth alpy change and favorable entropy change; this process appears to be driven mainly by an entropic effect for all tabulated lipid dispersions. One of the main drawbacks to this technique is that it requires direct observation of the NMR resonance associated with the lipid phase under slow exchange conditions. This creates complications because th e peak is very broad and highly dependent on the xenon overpressure and lipid co ncentration, having a direct e ffect on the maximum chemical shift difference between aqueous and lipid phases. As mentioned previously, the DOPC lipid

PAGE 115

115 system requires an overpressure of at least 5 at m in order to successfully resolve out the lipid phase; if the lipid or xenon concentrations are to o low only a single resonan ce will be observed. In analyzing our data we cannot ignore the pos sibility that higher concentrations of xenon may perturb the lipid matrix as subtle change s in membrane structure or dynamics may be reflected in the standard molar entropy and en thalpy values. For example, changes in xenons distributional volume or diffusion properties may have influence on the lipid membranes lateral or perpendicular pressures with temperatur e. Results suggest increased xenon-membrane interaction at elevated temperat ures. Furthermore, the positive en tropy change that is shown to dominate the transfer process could be a due to the release of water molecules at the lipid-water interface as well as increased diso rder of the acyl chain region w ith partitioning. So, to answer the question posed in the introduction, it is our view that bot h preferential hydration of the xenon at the membrane-water interface and modifications in membrane order direct the distributive properties of xenon with temperature. These two effects are not mutually exclusive. Whats more, the lack of significant change in the molar chemical potential s, enthalpies and entropies of transfer of xenon between solutio ns (lipid -water versus oilwater) suggests good adherence to the MO rule with temperature (at low Xe( g ) overpressures).

PAGE 116

116 CHAPTER 6 PRESSURE EFFECTS ON BINDING BEHAVIOR 6.1 Introduction The two prevailing models used to characte rize the association of small molecules to phospholipid vesicles are the adsorption isotherm (binding) and ideal partitioning models.[244-247] As described in previous chapte rs, the partition equilibrium considers a two phase problem in which a probe molecule is distributed according to favorable solvation effects. The partition coefficient can be utilized to compare the solutes lipid solubility by comparing its free energies in each phase with changing extern al variables. Binding models provide simple explanations for interactions between probe molecule s (i.e., xenon) and a target site It is the association of the anesthetic molecule to the lipid and/or protein that leads to its relative activity. The complexity of the binding model depends on whether the bind ing sites are independent and/or equivalent. And though it is generally acknowledged that an esthetic binding to lipid membranes is nonspecific in nature, solutes ar e not uniformly distributed th roughout the membrane. So, while the partitioning model provides information on the relative distribution between phases, it does not account for differences between surface and inner-membrane binding. Recent computer simulations investigating th e anesthetic potency/solubility relationship have shown that the potencies of inhalation anes thetics correlate better with their interfacial solubility.[248] This was particularly true for nonpol ar compounds, like xenon. These results are consistent with several modern lipid theories of anesthesia, which suggest that the preferential location of anesthetics at the membrane interfaci al region is of importance to anesthetic action.610 However, there is limited experimental eviden ce of a two-step adsorption mechanism between anesthetic molecules and the lipid membrane, s uggesting the presence of both a low capacity, high affinity adsorption site at the hydrophili c surface and a high capacity, low affinity site

PAGE 117

117 within the lipid core.[249-255] It is the lipid core, with its higher binding capacity that is thought to correlate well with the solubility/potency relati onship of apolar solutes in olive oil (MeyerOverton rule) as it involves part itioning into the acyl chain region.[256] Herein, we utilize extracted binding data from the 129Xe NMR chemical shift analysis (presented in Chapter 4) to differentiate betw een anesthetic binding at the membrane interface and the lipid core. If observations by Xu and Tang (1997)[14] are correct, and xenon interacts specifically with water at the lipid-water interface, we should be able to confirm it through binding analysis. The mathematical treatment of binding data is often f it according to solution and surface models. Since the lipid is consider ed to possess both solution and solid state properties, we use several isothe rms for comparison. The Langmuir,[257] BET,[258] the DArcy and Watt[259] isotherms will be reviewed in subsequent sections. Our primary goal in this chapter is to determine whether the lipid membrane surface is saturable and ascertain whether 129Xe NMR can distinguish between binding sites at the bilayer interface and lipid core. 6.1.1 Specific versus Nonspecific Binding While specific binding implies strong, locali zed interactions, non-sp ecific binding is often characterized by weaker, less localized interac tions. Hydrogen bonding and electrostatic interactions are typical exampl es of specific binding, just as dispersion forces and hydrophobic interactions epitomize nonspecific binding. It rema ins unclear whether anes thetics act indirectly through the lipid membrane, or directly through to sp ecific protein target site s. A wide variety of inhalational anesthetics ar e either diatomic or noble gases, which possess no permanent dipole or chargethey interact primarily through van der Waals forces. Xenon, for example, has a large electron cloud which is highly polar izable. This allows for nonspecific interactions at the protein surface, as well as low capacity, high affin ity binding within hydrophobic pockets. However, identifying a specific protein molecule that possesses the necessary saturable binding and

PAGE 118

118 structural stabilization associated with anesthesia has been difficult.[260] As recently discussed by Carmody (2009), the Meyer-Overton theory of anes thesia overly emphasi zes lipid solubility. While Carmody supports protein medi ated theories of anesthetic action, he readdresses the potential role structured water molecules have on both protein conformations and lipid solubility. Similar to lipid membranes, anesthetics have been shown to preferentially bind to exposed protein-water interfaces in a nonspecific manner (e.g., firefly luciferase) despite inhibitory effects at lower anesthetic concentrations.[261] If anesthetics act on all sites of intermediate polarity, regardless of protein or lipid environm ent, there may be credence to lipid-mediated action.[84, 89, 262] Or, as suggested by Ueda and Yoshida (1999),[181] anesthetic action may be due to simple changes in prot ein and lipid hydration. 6.1.2 Interfacial Membrane Partitioning Experimental results indicate that the primary difference between non-immobilizers (molecules predicted to be anesthetics based on hydrophobicity, but are not) and anesthetics are their relative locati ons within the bilayer.[263-265] Anesthetics associate strongly with the lipidwater interface, significantly affecting the dipo le potential, while non-immobilizers partition into the hydrocarbon core, having minimal perturbing eff ect. Experimental results suggest that the adsorption mechanism of anesthet ics in membrane systems change the hydrophilic properties of the membrane. Using microwave spectroscopy, Enders showed that volatile anesthetics decrease the Debye frequency of membrane-associated water molecules by a factor of three.[251] Not only was the effect completely reversible, the obs erved behavior was identical for chemically different anesthetic species. Ueda et al. (1986) also showed a low-dose absorption mechanism for the inhalation anesthetic-membrane intera ction using low freque ncy capacitance and conductance measurements.[249] The lipid membrane is thought to possess an energy barrier at its surface, which prevents anesthetics from penetrating into the lipid core. At sufficiently high

PAGE 119

119 anesthetic concentrations, the membrane loses its barrier properties, allowing penetration within the lipid core. Statistical therm odynamic models were utilized to evaluate the capacitance data and two binding modes were determined. All an esthetic molecules exhibited a similar surfaceassociated binding constant of 650 M-1 while the secondary binding affinity ranged between 1.8 and 7 M-1. Similar to recent computer simulations, the interfacial concentrations were significantly higher than thos e within the core region.[248] 6.1.3 Common Adsorption Types Adsorption can be defined as the selective accumulation of a molecular species between two phases. If the molecule of in terest is gaseous and it binds to a liquid or solid, the interface is normally between the gas and solid/liquid phase. Adsorption can be classified in one of two ways: chemisorptions (chemical adsorption) and physisorption (physical adsorption). Chemisorption is characterized by strong interactions between ad sorbate and adsorbent, while physisorption is associated with weaker van der Waals interactions (i.e., di spersion and dipole). The stronger the interaction, the longer the contact time is betw een the adsorbing molecule and the surface. This is why chemisorbed gases tend to be more difficult to remove from the adsorbent, often resulting in changes in surf ace structure. Unlike chemical adsorption, physisorption is a much longer ra nge interaction caused by nons pecific interactions. Though the molecular interaction is weak, accommodation of the adsorbate can also result in structural modifications of the surface structure including bi layer swelling, changes in the phase transition temperature, membrane fluidity, and/or the available lipid area Several of these have been discussed in previous chapters. Several factors which help to determine the strength and specificity of an intermolecular interaction are the hydrophobic a nd hydrophilic properties of the in teracting species, in addition to their respective molecular size, symmetry, degrees of freedom and polarizability. As discussed

PAGE 120

120 above, the higher the specificity, the stronger the interaction. Here we c onsider the interaction between dissolved xenon gas and the lipid me mbrane, where the inhalation anesthetic is physisorbed to the membrane. Similar to previ ous studies, we assume multilayer adsorption; each binding mode is taken to possess its ow n volume and capacity for the gas. After accommodation, the sorption is reversible when the gas overpressure is decreased. The bilayer surface will be referred as the adsorbent, while dissolved xenon in bulk solution ([ Xe ]aq ) is considered the adsorbate molecule. The sorption of xenon from bulk solution to the lipid surface is depicted in Figure 6-1. The specific models ill ustrated will be explaine d in further detail in upcoming sections. Figure 6-1. Several examples of xenon-memb rane binding models. The surface is the lipid membrane, and the adsorbent is xenon gas. A) Depicts monolayer coverage of the membrane surface, in which each binding site is discrete, identical and noninteracting. B) Assumes multilayer form ation on the surface (BET isotherm). C) suggests that second and higher level adsorption occurs in the membrane core (modified BET isotherm).[249] 6.2 Binding Models 6.2.1 Adsorption Equilibrium Due to the complex nature of the adsorpti on process, adsorption experiments are often presented in the form of binding isotherms. Thes e binding models attempt to make sense of the adsorbate-adsorbent interaction through different assumptions a nd mathematical relations. The amount of pure gaseous species adsorbed by a unit mass of adsorben t is a function of (a) (b) (c)1stlayer 2ndlayer 2ndlayer Monolayer adsorption Multilayer adsorption Multilayer adsorption A) B) C)

PAGE 121

121 temperature and pressure alone. Results are most commonly expressed as a function of pressure, at constant temperature. The Langmuir and Brunauer, Emmet, and Teller (BET) isotherms are two of the most commonly used binding models a nd are typically used to describe gas-to-surface binding. The DArcy and Watt isotherm, a solu tion based model, is less commonly used. The base of each of these models is the simple equilibrium expression which measures the affinity of adsorbate to a given adsorbent. A key aspect of the binding process is that the magnitude of the affinity between two substances is related to their interaction energy under a given set of conditions. The affinity is measured by the equilibrium constant ( Ka); the larger the Ka value, the higher the affinity. A general binding expression describing the bi nding process between the probe molecule, Xe(aq), and a lipid adsorption site, L is given in Eq. (6-1). Its specific relation to the stoichiometric equilibrium constant (Ka) is provided in Eq. (6-2). + a dk n aq knXeLLXe (6-1) n a a n d aqLXe k K k LXe (6-2) Here, [ L ] is the total lipid concentration, [ Xe ]aq, the available xenon concentration in aqueous solution, and [ LXen], the xenon-lipid binding complex in solu tion. If a macromolecule (i.e., lipid molecule) possesses many identical and non-interacting bind ing sites, the n -value shown above would be taken as 1, which is the basis of the La ngmuir isotherm. However, if the binding of one adsorbate affects subsequent binding events, th e affinity changes as a function of adsorbate concentration. Before we get ahead of ourselves let us go over the assumptions of the most simple of isotherm models, the Langmuir isotherm.

PAGE 122

122 Langmuir isotherm. The Langmuir isotherm was init ially developed to describe monolayer formation of chemisorbed processes. Once equilibrium is established between the gas phase and the partially formed monolayer, one can measure the fractional coverage ( ) of a given adsorbent surface at a specified pressure PXe. To be specific, is defined as V/Vm where V is the volume of adsorbed gas and Vm is the volume of gas adsorption sites that exist on the surface. Under this model, dynamic equilibrium is reached when the rate of desorption ( kd) is equal to the rate of condensation ( ka) of the gas molecules, as described by: 1aXe dkPk (6-3) This allows for a simple relation between the fractional surface coverage, the gas pressure, and the dissociation constant where 1Xea Xea dPk PK k (6-4) at equilibrium. If PXe is small, the Langmuir adsorption isotherm reduces to a simplified expression between and the binding constant, consistent with Henrys Lawall isotherms should reduce to zero as the pressure goes to zero For clarification, is the amount of adsorbed gas per unit of adsorbent divided by the saturati on value for monolayer coverage. As explained previously in Chapter 4 (Section 4.2), PXe is the xenon overpressure in the gas phase above the aqueous solution of dispersed vesicl es. It is easily related to xenon s solubility in water, allowing for a direct relation between PXe and [ Xe ]aq. There are two main criteria of the Langmuir isotherm. Firstly, the gaseous molecules are assumed to be adsorbed at a fixed number of well defined sites where each site has the capacity to hold a single adsorbate molecule. Furthermore, each site is considered to be energetically equivalentso, there are no lateral interacti ons between gaseous molecules adsorbed onto neighboring sites. The Langmuir mo del is excellent at identifyi ng ideal binding behavior but may

PAGE 123

123 not adequately describe the ga s-to-membrane adsorption process. We can easily confirm ideality from the n-value shown in Eq. (6-2). If there is no interaction between sites the n-value will be unity. The linearized form of the Langmuir isotherm is called the Scatchard plot and it is perhaps the most widely used expression in the analysis of binding equlibria in biological solutions. It is useful in that it can be used to identify changes in the binding behavior graphically. It will be discussed in more detail in upcoming sections. The BET isotherm. Similar to the Langmuir isotherm, the BET isotherm assumes that the surface is homogenous and that there are no late ral interactions betw een adsorbed gases. However, gas adsorption is not limited to a si ngle monolayer and dynamic equilibrium exists between each layer (i.e., the rate of adsorption to the first layer equals the rate of desorption of the second layer). Furthermore, the heat of adsorpti on of all layers above the first is equal to the heat of condensation. When the saturated vapor pr essure is reached, the gas will condense as an ordinary liquid on the surface, creating an infini te number of layers on the adsorbent surface. This is unlikely for our experime nts as the vapor pressure of xe non is approximately 58.21 atm at 298 K. The general expression for the BET isotherm consists of two parts, i) a Langmuir-type segment which accounts for monolayer of adsorp tion onto the surface, ii) a second component accounting for weakly adsorbed multi-layers. The full expression is given as: 11aq aq m aq aqKXeXe V VKXeXe (6-5) where V/Vm is the fraction of the surface covered and [ Xe ]aq is as previously defined. Notice that K is not explicitly defined as the association cons tant. Instead, it is a proportionality constant that represents both the binding affinity and the specific number of binding sites needed to accommodate a monolayer of adsorbate. A shortcom ing of the BET isotherm is that it fails to

PAGE 124

124 predict limited adsorption. Guggenheim, Anderson, de Boer (GAB) model has been proposed to correct this.[266] DArcy and Watt isotherm. Briefly, the DArcy and Watt isotherm is a three parameter model that tries to account for both strong and weak adsorption as well as the interaction of the gaseous molecule with water.[259] It can be described as a soluti on-to-solid adsorption isotherm that accounts for two types of binding interactions Similar to the BET isot herm, the first term of Eq. (6-6) deals with monolayer adsorption onto a surface. The second term describes adsorption to weaker binding sites, while the last term acc ounts for multilayer formation, similar to the BET isotherm. The equation coefficients, K and [ Xe ]aq are as previously defined, while C is a constant proportional to the number and a ffinity of weakly adsorbed sites, and D and D denote the number and affinity of the adsorbed multi-layers (Eq. (6-6)). 11aq aq aq aq aq D DXe KKXe VCXe K Xe DXe (6-6) 6.2.2 Macroscopic and Microscopic Binding Constants Now that we have a sense of the typical models used in binding analysis, let us distinguish between macroscopic and microscopic binding constant s. As noted in the models above, the explicit form of the association constant is not given. Rather, they express Ka as proportionality constants. If a macromolecule has multiple identical non-interacting binding sites, it can be shown that the binding isothe rm will be the same whether microscopic or macroscopic equilibrium constants are used. Mi croscopic binding is site specific, while macroscopic binding is not. This is seen more clearly in Figure 6-2; K1 and K2 represent sequential stoichiometric binding constants wh ich describes occupanc y by class, while k1 and k2 describe the complete adsorption/desorption cycle in a site specific manner. Distinct classes of binding sites are characterized by different binding affinities.

PAGE 125

125 Figure 6-2. The difference betw een macroscopic and microscopic binding constants. In the case of multiple identical binding sites, the relationship between the microscopic ( k1 and k2) and macroscopic binding constants ( K1 and K2) is relatively simple. When both binding sites are empty (box 1 and 2), ther e are two possible binding sites for the adsorbate, so K1 = k1+k2 = 2 K where K is the microscopic binding constant. When both sites are occupied, there are two sites from which the adsorbate can dissociate, leading to K2 = k1k2/( k1+k2) = K/ 2 .[267] As the binding scheme becomes more co mplex, statistical thermodynamic methods become more useful. For a macromolecule that possesses more than one set of independent identical binding sites (having di fferent affinities), then the de gree of binding can be expressed as a summation of binding classes. The genera l expression is given in Eq. (6-7), where r signifies the moles of bound adsorbate per mole of macromolecule, ni is the number of distinguishable sites in the i th class, M is the number of classes, and Ki is the binding constant associated with each. 1+ 1M ii aq i i aqnKXe r KXe (6-7) 11 1. 1aq aqnKXe KXe (6-8) nonspecific aq K Xe (6-9) K2 K1L1L2 +12 + 12 + 12 1 + k1 k2 k1 k2 2 or 2 XeL1 XeL2Xe Xe Xe Xe

PAGE 126

126 Notice how each term in Eq. (6-7) has the same form as the first term of the BET and DArcy Watt isotherms. Eq. (6-7) takes into account more complex binding behavior. For example, if there is another binding class that is not independent of previous binding events, an additional term can be added to account for it (e.g ., Eq. (6-8) and (6-9)). This is particularly useful when cooperative binding occurs. Taken independently, the expression in Eq. (6-8) is known as the Hill equation, where is the Hill coefficient.[268] If = 1, the term becomes equivalent to the first binding cl ass in Eq. (6-7). However, if is less or greater than unity, the term adjusts for positive or negative cooperativ e binding. Cooperativity is when one binding event promotes or inhibits s ubsequent binding. The last term accounts for nonspecific binding (Eq. (6-9)). This is similar to the linear term in the DArcy and Watt isotherm as it describes weakly, non-interacting binding processes. The BET and DArcy and Watt isotherms describe sequential binding events to account for the formation of multi-layers. The adsorbate forms a monolayer on the adsorbent, after which multi-layers begin to form. Ueda et al. (1983) utilized a modified BET isotherm to extract the equilibriu m constants arising from surface and lipid core partitioning.[249] While the first layer was taken as m onolayer adsorption, the second and higher layer adsorptions were taken to occur in the inte rior of the membrane instead of on top of the first. They assumed that surface adsorption sites were distinguishable and that an additional set of distinguishable binding sites are made availabl e in the lipid core when a surface site becomes occupied. This model was too complicated to apply to our data. However, we did make use of similar assumptions by accounting for the possibili ty multiple, distinguishable binding sites. 6.3 Results and Discussion 6.3.1 Estimation of Kinetic Parameters The membrane surface can be considered to be a two-dimensional array of lipid molecules, having distinct partition sites whic h allow for gas adsorption into the lipid core.

PAGE 127

127 While it is understood that th ere are no specific interactions between lipid and xenon, the existence of multiple adsorption sites possessing di fferent affinities is likely. The question is whether the lowest free energy regi on lies within the lipid interior or at the membrane surface. As the surface areas of both adsorbate and ad sorbent are known we can start with a twodimensional picture of the binding process. Fo llowing Ueda et al.s lead, consider a two dimensional triangular lattice on the membrane surf ace. Now let us consider the interstitial site located at the center of th is lattice to be a par tition site, allowing for xe non to penetrate into the membrane core. As shown in Figure 6-3, this ad sorption site is characterized by the surface areas of both the triangular lattice a nd the adsorbate molecule. Figure 6-3. A graphic illustra tion of the two-dimensional binding model. According to this model, the number of partition sites ( S ) determines the saturation limit. The degree of saturation is dependent on the available membrane surface area[269-271] and the relative amount of bound adsorbate. This figure was modified from Word and Smejtek (2005).[272] In calculating the total membrane surface area, both inner and outer membrane leaflets are considered available for binding, each exhibiting identical behavior It should be noted that if only the outer layer is considered, the binding affin ity increases by a factor of two. Firstly, the number of partition sites is determined according to the following relation: m L XeXeA LA S AA (6-10) AXe 602AL 722Xe Partition Site Lipid MatrixSurface AreasXe LL LPC headgroup region Xenon atom

PAGE 128

128 Am denotes the lipid membrane surface area, AXe, the area of the partition site, AL, the membrane surface per lipid molecule, and L the total number of lipid molecules in a given experiment.[272] The number of lipid molecules per partition site is given by the AXe / AL ratio. Here, the fractional coverage ( ), which was introduced in previous sections, is considered to be equivalent to the fraction of occupancy of these partition sites. The fractional coverage gives an indication of membrane saturation and provides the basis for all previously described binding models. As discussed by Word and Smej tek (2005), if lateral interactions between adsorbate and absorbent prevent nearest neighbor gas molecules from approaching each other within a distance comparable to the separation between lipid mo lecules, the number of lipid molecules associated with the partiti on site will be greater than one.[272] According to this model, is determined directly from the xenon-to-lipid ratio, where Xe Xe boundAboundXe LLnNnA SnA (6-11) This now allows us to estimate the binding paramete rs for the xenon-lipid inte raction. It must be stated that while is approximately equal to the r -value defined in Eq. (6-7). Even so, each will be specified when used. At low solute concentrations, the partition coefficient may be related directly to the association constant according to Ka = Kp / 55.5M. However, we have already observed the potential errors that can arise from this approximation. The infinite dilution value of the partition coefficient predicts an apparent binding constant ( Kp ) of 7.1 0.8 M-1 at trace levels of xenon. To determine the concentrati on dependence of xenon on the binding equilibrium, we generated isotherms for several lipid concentrations with xenon overpressures ranging from 1 to 10 atm. If ideality holds, and there is only a single part ition site, the binding constant will remain unchanged.

PAGE 129

129 Assuming a 1:1 complex, the association be tween xenon and the lipid membrane can be described by a simple equilibrium expression; aq X eLXeL where Xe( g ) is the free xenon in solution, L the membrane environment, and XeL the binding complex. Utilizing the law of mass action in conjunction with and max values obtained from spectroscopic data (see Chapter 4 for details), the apparent binding affi nity can be measured from the association constant of the complex ( Ka), according to Eq. (6-2). Assuming ideal behavior, experimentally determined values of [ Xe ]bound were extracted from the following relation: max boundaqXeXe (6-12) As a reminder, is the experimentally observed chemical shift difference between xenon in the LUVs and lipid-free buffer solution (obsobs L UVbuffer), and max is the maximum chemical shift difference between the lipid and buffer phases (F igure 4-1). The pressure dependence of the max value was shown previously (Figure 4-8). The combination of Eqs. (6-12) and (6-2) yiel ds a linearized form the Langmuir isotherm known as the Scatchard equation: max aa aqr nKrK XeL (6-13) Here, r refers to the amount of bound xenon per mole of lipid, and n is the number of bound adsorbate molecules per binding site. The Scatchard pl ot is often used to quantitatively interpret the binding behavior of biologi cally relevant media in soluti on and is generally employed to differentiate between speci fic and nonspecific binding.[273-275] Nonspecific interactions are characterized by its proportionality to the amoun t of adsorbate in solu tion, i.e. doubling the concentration of [ Xe ]aq doubles the amount of nonspeci fic binding. Thus, nonspecific

PAGE 130

130 Figure 6-4. A Scatchard plot of the binding data. r-values were determined by fitting experimental data to Eq. (6-13); the pos itive slope is characteristic of nonspecific interactions. Here, the lipid concentra tion was held constant, as the xenon overpressure was increased. The reported values are an average of 6 separate experiments (lipid concentrations be tween 15-100 mM and over pressures ranging from 1-10 atm). The error is the standard deviation of the mean. The open circle represents the extracted in finite dilution value of Ka. The solid line is to guide the eye. interactions are often indicated by straight, horizontal lines (zero slope) in the Scatchard plot. Specific, higher affinity interactions are usua lly manifest by a low initial slope (negative in value).[276, 277] Changes in the slope reflect changes in the relative binding activity. Here, a Scatchard plot were made for fixed [ L ] and variable [ Xe ]aq, (Figure 6-4). The initial slope is positive, which is an unusual feature for Scatchard plots. Positive slopes are often given as indications of positive cooperati vity and unsaturated binding.[278-281] To further elucidate the quantitative binding of xenon to lipid vesicl es, the experimental data was fit to the Hill equation to check for co operativity. It is possible that the changing affinity is due to the increased presence of xe non on the surface. The Hill plot is commonly used in the evaluation of dose-respons e curves and allows for the ex traction of both the apparent 13.0 12.0 11.0 10.0 9.0 8.0 7.0r / [Xe]aq (M-1) 0.50 0.40 0.30 0.20 0.10 0.00r

PAGE 131

131 association constant ( Ka) and the relative degree of coopera tivity. The Hill equation is shown explicitly in Eq. (6-8), while the fit of experimentally derived data to this model is provided in Figure 6-5. A fit to the Langmuir isotherm is given for comparison. As expected, the Langmuir isotherm provides a good fit at low [ Xe ]aq. This section, defined by th e initial slope from origin, is called the Henrys law region and is characte ristic of nonspecific binding. The apparent association constant resulting from the fit of th e Langmuir isotherm (Eq.(6 -7)) to the first three points was found to be 8.1 0.2 M-1; this is near our predetermi ned infinite dilution value ( Ka = 7.1 0.8 M-1). But, just as the Langmuir model fails to accurately fit binding at higher xenon doping, the Hill equation gives a poor fit for bi nding at low xenon concentrations (dashed line Figure 6-5). Combining the two mode ls yields the following equation: 11 221aq aqnKXenKXe (6-14) where n1 = n2 = 1, is the Hill coefficient, and K1 and K2 signify the association constants stemming from the Langmuir isotherm and the Hill equations, respectively. Doing so generates an adequate model for our data, th e fit of which can be seen clearly in Figure 6-5 (solid line). The extracted equilibrium constants are as follows: K1 = 10.8 0.8 M-1, K2 = 21.3 0.2 M-1, with = 3.7 0.3. The high -value suggests a large degree of positive cooperativity with increased doping, meaning that each xenon-me mbrane binding event facilitates further interaction. The dose dependency of [ Xe ]aq on the nonspecific binding isotherm (Eq. (6-9)) is shown in Figure 6-6. Similar to the Hill model, the nonspeci fic binding isotherm fails to accurately fit binding at low xenon doping (dashed line Figure 6-6). However, adjusting for cooperativity yields an apparent a ssociation constant of Knonspecific = 13.6 0.1 M-1 and a Hill coefficient () of

PAGE 132

132 Figure 6-5. A Hill plot (dashed line) fit to experimental data resulted in an apparent binding constant of 24.9 0.3 M-1 and a Hill coefficient of 2.1 0.1. The dotted line is the Langmuir isotherm (Eq. (6-4)) fit to the first three points, yielding a Ka of 8.1 0.2. solid line corresponds to the fit which combines the two models (Eq. (6-1 4)). Errors are reported as (SSE). Figure 6-6. Adsorption profiles of experimental data fit to severa l models. Experimental data fit to the Langmuir isotherm (dotted line: fit to first three points only) and a generalized non-specific model (dashed line) which yields a Ka of 8.1 0.2 and 11.5 0.2 M-1, respectively. The solid line is the fit to Eq. (6-15). Errors are reported as (SSE). 1.2 1.0 0.8 0.6 0.4 0.2 0.0 40x10-3 35 30 25 20 15 10 5 0[Xe]aq (M) 0.6 0.5 0.4 0.3 0.2 0.1 0.0r -value 45x10-3 40 35 30 25 20 15 10 5 0[Xe]aq (M)

PAGE 133

133 1.2 0.1, indicating slight positive cooperativity with increasing [ Xe ]aq. A reduced form of the Hill equation (Eq. (6-8)) accounting for cooperative nonspecific binding is then combined with the Langmuir adsorption isotherm through the following relation: 11 1. 1aq nonspecific aq aqnKXe rKXe KXe (6-15) Doing so results in an adequate fit of our bi nding data, as shown in Figure 6-6. The extracted binding parameters were: K1 = 6.9 1.3 M-1, n1 = 1, Knonspecific = 11.0 0.2 M-1, and = 1.6 0.2. The difference between the nonspecific porti on of the modified binding model shown in Eq. (6-15) and the Hill equation (see Eq. (6-7)) is th e lack of terms within the denominator. This simplification is correct assuming Knonspecific[ Xe ]aq 1. As suggested by the Scatchard plot (Figure 6-2), there appears to be two stages of bindingan ini tial Langmuir-type (i.e., identical, non-interacting sites) and another that a ppears consequence of increased doping. In order to rule out multi-layer binding on th e membrane surface, we attempted to fit our data to the BET and DArcy and Watt isotherms. While unsuccessful on reaching satisfactory fits with the BET model, accounting for an additional weak adsorption site (second term of DArcy and Watt isotherm) did allow for the extraction of the binding constants from Eq. (6-6). The fit yielded six parameters (see Figure 6-7): the affinity for xenon binding on the surface and its number of adsorptions sites ( K1 and M1, respectively), a weaker interaction ( K2) and its specified sites ( M2), and a correction term that accounts fo r xenons affinity to and formation of multilayers. The extracted parameters are as follows, M1 = 3.8, K1 = 6.5, M2 = 1.2, K2 = 11.7, D = 6.3, and D = -6.5. Consistent with Eq. (6-6) formalisms C = M2K2 14, K = M1Vm, and K = K1. Interestingly, the negative value for the D parameter changes the sign of the third term in Eq. (6-6). Though unclear why, its num eric value is identical to the K1 term, suggesting that

PAGE 134

134 Figure 6-7. Langmuir isotherm ( dotted line) and the DArcy and Watt isotherm (solid line) fit to experimental data. Note that the Langmuir isotherm is fit to all points here instead being limited to the first thr ee points, as done previously. multilayer binding on the membrane surface possesses similar affinity. If this were true, the K1 binding constant could be associated with the membrane surface. However, the DArcy and Watt model does indicate the presence of two binding m odes. Similar to the Scatchard plot, we see a higher capacity, lower affinity bi nding site as well as a lower capacity, higher affinity site. However, adsorption isotherm analysis still doe s not identify which bindi ng site corresponds to the lipid surface and which to the membrane core. Lastly, we looked at the effect of sequen tial binding to see whether the nonspecific cooperative binding is due to part itioning into the lipid core or surface binding. A universally valid equation that accounts for cooperativity effects, the possibility of multiple binding sites that may or may not possess different affinities, is the stoichiometric binding equation: 2 112123 2 1121232 1 M M aqaqaq M M aqaqaqKXeKKXeMKKKKXe rXe KXeKKXeKKKKXe (6-16) 0.4 0.3 0.2 0.1 0.0V / Vm 40x10-3 35 30 25 20 15 10 5 0[Xe]aq (M)

PAGE 135

135 Figure 6-8. Klotz a ffinity profile for binding of xenon to DOPC. Ki is the stoichiometric binding constant and i the sequential binding number. The dotted line is an approximation of ideal binding, while ( ) signifies two binding events, ( ), three, ( ), four, and (*) seven binding events. where, K1 to KM are stoichiometric binding constants. For ideal, non-intera cting sites, the macroscopic Ki (Ind) values are linearly related when placed on a Klotz affinity plot.[282, 283] If no cooperativity exists, the plot will show a linear decrea se in the stoichiometric parameter with increasing number of binding events ( i ) according to the following relation: 111iIndiKnKiK (6-17) Positive and negative cooperativity are seen graphically through changes in slope; a positive slope implies enhanced binding while negative slopes suggest some impedance. Though Eq. (6-16) provides a stepwise interpretation of the binding data it does not indicate the number of binding events that are present ( M ). In order to account for this we used a simple iterative procedure, varying the number of binding events in a sequentially, from i = 2 to i = 7. The variations in binding parameters are obtained by re peating the fit procedure, starting from initial 40 30 20 10 0iKi (M-1) 7 6 5 4 3 2 1i

PAGE 136

136 estimates. As seen in Figure 6-8, the best fit ch anges slightly when additional binding events are introduced. Our initial estimation of K1 and K2, provided from parameters acquired through Eq. (6-13), drastically improved our chances for convergence. The two binding events are shown clearly in the Klotz plot The binding constant, Ki, is the average binding affinity of xenon to the lipid matrix. The first binding event to one lipid mo lecule is rather low affinity, while the second binding even to the lipid shows so me cooperativity as well as highe r affinity. Interestingly, the third binding event shows negative cooperativity a nd almost no affinity. However, it appears that xenon will have greater affinity fo r it once the binding event has occurred. This is consistent with Uedas theory[249] of a dysfunctional lipid inte rface with increased doping. 6.4 Conclusions The goal of this chapter was to differentiate between surface and core binding via chemical shift analysis. The Scatchard plot s show atypical be havior at low r-values in the form of a positive slope. This unusual feature is of ten indicative of positive cooperativity.[278-281] Various binding isotherms were prepared from the xenon-lipid binding data with hopes of el ucidation this phenomenon, thereby providing additional informa tion about the net affinity with increasing xenon loading. Langmuir type adsorption (ideal, noninteracting binding sites) is observed at low xenon concentrations while posit ively cooperative, non-specific behavior emerges at higher concentrations. Kreishman et al. (1985) repor ted similar behavior in that higher ethanol concentrations were found to perturb the membrane in a cooperative manner.[284] Computer simulations suggest that as anesthetic molecu les bind to the membrane interface, the surface becomes partially dehydrated, increasi ng the anesthetic binding potential.[285] Interestingly, this is consistent with theoretical work investigating the distri bution of xenon in a phosphocholine environment with increased doping; the population in the membrane core was more substantial at higher xenon-to-lipid ratios.[15] This dose dependent bind ing may also explain why 129XeNOE

PAGE 137

137 experiments by Xu and Tan (1997)[14] only observed interactions at the lipid interfacetheir experiments were performed at 1atm. Lastly it should be noted that Ostwald solubility coefficients are essentially association consta nts. The typical values range between 12-20 M-1 in lipoid systems at 1 atm of Xe( g ) overpressure, which is in the range of our higher affinity adsorption site. Consistent with theory, we see that partition coefficients do not necessarily translate to net affinity when working at higher xenon overpressure s. To be clear, we understand that there are no specific bindi ng sites between xenon and the membrane adsorption sites. Our results clearly indicate that the collective interactions in the lipi d array may change with relative amounts of anesthetic gas.

PAGE 138

138 CHAPTER 7 CHANGES IN LIPID COMPOSITION: EFFECTS OF NONBILAYER LIPIDS 7.1 Introduction There are still many open questions concerni ng the mechanisms by which variations in lipid composition affect membrane function and considerable interest has been focused on elucidating the structural propertie s and behavior of nonbilayer lipids.[286, 287] A particular source of debate is whether proteins modulate its interaction with the cell membrane, or if the local dynamics and structural features of the lipid membrane prompt biological action. Though there are strong advocates for both side s of this controversial topic,[286, 288] our interests lie in gaining insight into how variations in lipid composition affect molecular interactions and transformations within lipid membrane topologies. Nonbilayer structures are thought to play an important role in membrane function[286, 289, 290]; the formation of local regions of nonbilayer st ructures within the biomembrane (i.e. lipid rafts) is one such idea. In addi tion to being responsible for tran s-bilayer transport mechanism of lipids and polar solutes, transient formations of inverted structures may be possible precursors to membrane fusion events. By affecting the barrier and flexibility properties, the presence of these lipid species may indirectly aff ect protein function as well. Not only have they been shown to increase the activity of numerous peripheral and integral membrane proteins (e.g. protein kinase C and rhodopsin), there is experimental evidence that they also influence the conductance of channel forming peptides as well.[291, 292] Not surprisingly, nonbilaye r lipids have also been suggested to play a key role in the anesthesia.[293] Chapter 7 is focused on elucidating what role, if any, the nonbilayer lipid DOP E plays in these processes by 31P and 129Xe NMR methods. We begin by studying the effect of increasing molecular strain by nonbilayer lipids with hopes of gaining insight into the s ubtle energetic changes in stressed lipid bilayers. In lieu of the

PAGE 139

139 129Xe-DOPC NMR studies, similar app lications of the partition mode l (Chapter 4) were utilized to investigate the influence of the nonbilayer lipid DOPE on the partitioning behavior of xenon between bulk solution and frustrate d bilayers. Theoretically, the lateral pressure profile should be altered with PE content and may be reflec ted in changes in the apparent mole fraction partition coefficient. By inducing the actual phase transition from L-toHII we hope to confirm the presence of transient structures possessing similar curvature exhibited in some of these membrane processes. 7.1.2 Nonbilayer Lipids on Biological Processes Though biological membranes inherently possess b ilayer structure they also tend to contain substantial amounts of nonbilaye r lipids. The sheer abundance ( 75%) of phosphotidylethanolamine (PE) in the inner membrane of Es cherichia Coli suggests so me role in the overall functionality of the membrane, yet a precise explanation for it s presence has been elusive. Numerous models describing nonbilay er behavior exist in the literature and can be explained in terms of one or more of the following factors: curvature stress, packing defects, lateral pressure profile, hydrophobic mismatch or bilayer fluidity Despite this, only a moderate mechanistic understanding is known about nonbilayer lipid function at the mol ecular level. While the physics of isolated tubular membrane structures (e.g PE) has been studied extensively and is well characterized, the mechanism by which these stru ctures are formed from planar or spherical membrane species are not. At equilibrium, the lipid membrane stability reflects a balance between the attractive and repulsive forces between the headgroups and hydr ocarbon tails, and is often described in terms of the lateral pressure profile. In the context of lipid-protein interactions, modifications in bilayer properties (i.e., intrinsic lipid cu rvature, bilayer thickness, etc. ) can decrease the free-energy difference between various protein conforma tions. As discussed by Anderson and Koeppe

PAGE 140

140 (2007),[287] the bilayer deformation energy is largely responsible for regulating the kinetic and equilibrium properties of the protein due to hydrophobic coupling betwee n protein function and bilayer properties. Deformations in the membrane via area compression, expansion, or bending processes, help facilitate morphological chan ges in lipid assemblies which can affect the interfacial properties of the bilayer. For exampl e, the protein rhodopsin has been shown to be extremely sensitive to the degree of spontaneous curvature. When reconstituted into PC lipid mixtures, this protein does not activate; its function appears to be dependent on the presence of lipids that possess the tendenc y to promote nonbilayer phases and/or negative curvature.[294-296] A graphical representation on th e effect of membrane compre ssion and bending properties as related to integral membrane proteins is show n in Figure 7-1. Evidence suggests that nonbilayer lipids localize near membrane spanning channels a nd proteins in an effort to reduce the bilayer stress associated with conformational changes. Lipid rafts are areas of me mbrane that possess different composition and physiochemical properties from the rest of the lipid matrix; they have been experimentally verified to exist in both biomembranes and model membranes.[297-300] Lateral heterogeneity has been induced by alcohol[301, 302] as well as dehydration mixtures[303] and been shown to be caused by interactions between lipids and integral membrane proteins.[304] Furthermore, it is hypothesized that these domain formations are precursors to the inverted hexagonal phase ( HII) formation.[305] The L-toHII transition is a multistep process which involves the formation of intermediate structures having similar curvature and structure as those involved in membrane transport events (e.g. vesicle trafficking, pore forma tion). The pore-stalk fusion hypothesis is the most widely accepted model for this process.[306, 307] The stalk mechanism is based on the existence of membrane defects in the lipid membrane. In order for fusion to occur i) two membranes must be brought

PAGE 141

141 into contact, ii) defects in the two membranes must align, iii) fuse and form a bridge-like configuration called a stalk, iv) internal monol ayers are expelled from the contact region as external monolayers are simultaneously brought together, resulting in an elongated stalk structure which allows v) small holes to begi n to form. An open fusion pore results when the bilayer formed by the external monolayers is brok en down. This can be seen more clearly in Figure 7-2. While the inverted hexagonal phase itself has never been directly detected under physiological conditions, local st ructures with similar curvat ure have been observed. The difficulty in creating a concrete mol ecular picture of the formation of these intermediate structures is that the structures themselves are short lived and small in size. For Figure 7-1. A graphical depi ction of the hydrophobic mismat ch and examples of common membrane deformations associated with it. A) dc = dp. B) dc < dp. C) dc dp. When the length of the unperturbed hydrocarbon region ( dc) is equal to the hydrophobic span ( dp) of an integral membrane, the deformation energy is 0. If dc < dp or dc > dp the molecular strain is manifest through modulations in membrane structure. The energy associated with expansion/co mpression and bending mechanisms are proportional to a(2 C/dc)2 and c( 2D-Ho)2, respectively. Here, C is proportional to the changes in area, and D to the membranes principle curvature.[308] This figure has been modified from Anderson and Koeppe (2007).[287] Bending (-) curvature Bending (+) curvature StretchingCompressionoo(a) (c) (b)dc> dpdc< dpdc= dp A) B) C)

PAGE 142

142 Figure 7-2. Depiction of the membrane fusion intermediated according to the stalk mechanism. When two bilayers come into close proxim ity to each other, m odifications in the surface curvature are thought to ensue, resu lting in pore formation. The process is described in detail in the text. This figure was modified from Jahn, Yang and Sudhof (2003).[309] example, the cubic phase, an intermediate structure in the inverted hexagonal phase formation, is characterized by a single isotropic peak in 31P NMR spectra and does not necessarily provide the best molecular description beyond changes in motion. Micellular phases and small unilamellar vesicles (SUVs) are also char acterized by isotropic peaks. X-ray diffraction methods were recently used to verify the existence of pore-like structures on a bilayer[310], but how these morphological changes are initiated is still not comple tely understood. Interestingly, several solute molecules have been shown to stabilize the formation of these intermediate structures.[310, 311] Assuming xenon gas behaves in a similar manner, we expect 129Xe NMR to substantially improve our understanding of the process. 7.1.2 Energetics of the Bilayer-to-I nverted Hexagonal Transition As mentioned in Chapter 3, bilayers can exis t in a variety of physic al states depending on the membrane composition and temperature. The mo st common change in states occurs between the gel and bilayer phase, termed the main phase transition. The gel phase is characterized by higher order and rigidity. When the temperature is increased the membrane core becomes more fluid due to a conformational cha nge of the acyl chains. The temperature at which this occurs is (i) (ii) (iii) (iv) (v) (i) (ii) (iii) (v) (iv)

PAGE 143

143 known as the main phase transition temperature ( Tm). This depends largely on the length and degree of unsaturation of the acyl chain; longer acyl chains an d higher saturation increase the packing efficiency, increasing the value of Tm. Conceptually, the bilayer-to-inverted hexagonal phase ( L-toHII) transition follows similar principles, albeit on a more complicated system. Unlike the gel-to-bilayer transition, the L-toHII transition involves a significant morphological change in the bilayer structure. However, the transition temperature is still dependent on the relative packing efficiency of the bilayer. If ther e is too much molecular strain at the membrane surface or hydrophobic core, the energy is released through the formation of lipid domains or even a phase transition.[312-314] In many cases the bilayer can be viewed as two separate, independent monolayers. The free lateral motion of the lipids in each leaflet enables each la yer to effectively relax surface tension (caused by changes in protein/pore conformations) through th e redistribution of nonbilayer lipids. If we consider the lipid membra ne as an incompressible, two-dimensional fluid that behaves as an elastic body when bent, the curvature elastic energy can be expressed as a summation of the bending modulus and the area expansion (see Figure 7-1).[308] As mentioned in Chapter 3, it is energetically possible to suppress a lipids intrinsic curvature, forcing the lipids to remain in the bilayer phase. Doing so introduces lateral stress in the bilayer which often manifests as changes in the membranes deformation energy (). The spontaneous curvature measures the tend ency for monolayers to bend into nonplanar geometries. If the headgroup layer prefers to be convex and bend outward into the water phase, it is assigned positive curvature Figure 7-1C. Howe ver, it the headgroup layer prefers to curl around the water in a concave fash ion, it is said to possess nega tive curvature (F igure 7-1B).

PAGE 144

144 Table 7-1. Structural and elastic properties of single component DOPC and DOPE lipids within a (1:3) DOPC/DOPE lipid mixture Property Molecular weight VL (3) dc () co dc Ao (2) A/Ao c / int b / int dc 2 DOPC 786 1292 25[315]-0.29[316]0.00820.05[317] 2.9[315] 0.12[318] DOPE 744 1235 30[319]-1.1[318]0.00650.05[317] 4.4[315] 0.10[316] When the spontaneous curvature is sufficiently negative, inverted and cubic phases are formed. A comparison of structural and elastic proper ties of DOPE and DOPC bilayer membranes is provided in Table 7-1. Despite the small differe nces in their relative molecular weights and volumes, the curvature frustration energy for main taining DOPE in the bilayer phase can range between 0.2-2 kBT / nm2. The spontaneous curvature ( co) is likely to vary between -0.0107 -1 and -0.0476 -1 for DOPC and DOPE, respectively. The memb rane properties listed in Table 7-1 are defined as follows: dc is the thickness of the membrane hydrophobic core, co, the monolayer spontaneous curvature, A/Ao, the bilayer area expansion, c, the bilayer elastic compressibility modulus, b, the monolayer bending modulus, and c / int the interfacial tension. Molecular weights and volumes were pr ovided by Chen and Rand (1998).[320] The L-toHII phase transition temperature ( Th) is largely dependent on the curvature elastic energy and various models have been presented over the years disc ussing the stabilizing factors of the inverted hexagonal phase form ation in a variety of lipid syst ems. In general, the difference of the chemical potential between phases ( ) can be expressed as a summation of the curvature elastic energy ( curv), and the interstitial chain packing en ergy of the inverted hexagonal phase ( ch):[321] II II IIHHH bilbilbil curvcurvchchcurvch (7-1)

PAGE 145

145 Where bil and I IH are the chemical potentials of the bilayer and inverted hexagonal phases, respectively. The L-toHII transition arises primarily from reductions in the membrane spontaneous curvature. An entropic need to pack the hydrocarbon moieties as uniformly as possible is what gives rise to the chain packing energyit is more pronounced in the inverted hexagonal phase than the lamellar phase, which means that ch is always positive. Interstitial domains are formed in the HII phase and the acyl chains of th e lipids to extend to different lengths to accommodate. This in turn decreases the entropy of the hydr ocarbon chains and increases the chemical potential of the membrane. 2111 2curv Aco woNA RR (7-2) 21 0 2bil coHbil A II curv curvcurvoNA R (7-3) The general expression for the curvature elastic energy (curv) is provided in Eq. (7-2) and the difference between the inverted hexagonal and bilayer phas e in Eq. (7-3); curv must be negative in order for a st able transition to occur. 25 The terms are defined as follows: NA, Avogadros number; c, the elastic bending modulus; Ao, the optimal surface area of the lipid phase; 1/ Rw = cw, curvature of the lipid monolayer; and 1/Ro = co, the intrinsic curvature of the lipid monolayer. In fully hydrated conditions 1/ Rw is approximately equal to 1/ Ro in the inverted hexagonal phase, facilitating the minimizati on of its curvature energy. Likewise, 1/ Rw = co must equal zero in the bilayer phase in order for its elastic curvature to be reduced. As shown in Figure 7-3, the i nverted hexagonal phase is charac terized by tubular structures formed around water channels and the hydrophobic interstitial spaces that exist between them. The acyl chains must stretch in order to accommod ate the structural change which reduces the

PAGE 146

146 Figure 7-3. Factors that promote and inhibit the lamellar to inverted hexagonal phase transition. An increase in headgroup size, ionization and water content favor the lamellar phase, while higher temperatures and relative degree of lipid unsaturation promotes the inverted hexagonal phase. This figure has been modified from Gruner et al. (1985). entropy and increases the chain packing energy. However, this can be lowered by adding nonpolar oils,[322] long chain alkanes,[323] or longer chained phospholipids,[324] which act to stabilize the interstices. In theory, these should stabilize fu sion intermediates as well. Th can be further reduced by increasing the PE content since the spontaneous cu rvature of a lipid mixture is a weighted average of the curvatures of the i ndividual components. This chapter focuses on the potential utility of xenon to stab ilize these intermediate struct ures and what role, if any, nonbilayer lipids play in the anes thetic-membrane interaction via 129Xe NMR. 7.2 Experimental In this study we used DOPC/DOPE mixtur es containing 0 to 75% DOPE. All samples were extruded to 100 nm LUVs, excluding the (1:3) DOPC/DOPE mixture which is already in the inverted hexagonal phase. While the majority of experiments were performed at relatively low xenon overpressures (2.5 atm), we were unable to detect the phase transition via 129Xe NMR at overpressures under 5 atm of thermally pol arized xenon. Thus, we made to use hyperpolarized 129Xe NMR to study the lamellar to inverted hexagonal phase tr ansition at low xenon Hydrocarbon Unsaturation Temperature HeadgroupSize HeadgroupIonization Water ContentLHII

PAGE 147

147 concentrations. A schematic of the hyperpolarized 129Xe NMR apparatus used is shown in Figure 7-4. The experimental procedure was a follows: 3300 Torr of 2 % gas mixture of xenon was recirculated through the pum ping cell for optimal polarization. As shown below, two valves were placed on the input and output of the sample in order to create a pressure differential over the solution. The normally closed solenoid va lve was controlled by the spectrometer through a pulse sequence. Each bubbling ev ent consisted of ten open/close cy cles of the solenoid valve, where d3 is bubbling time (25 ms), and d1 the dela y between opening and closing of the solenoid valve (2s). Each spectrum obtained through this method resulted in a net loss of 600Torr of gas mixture. Figure 7-4. Schematic of the recirculation apparatus used to introduce hyper-polarized 129Xe gas mixture to lipid sample. S: solenoid valve; V: needle valve; PC: pumping cell; d3: bubbling time; d1: delay between bubbling events; p1: 90 pulse length; aq: acquisition. d3 d1 p1 aqopenclose loopLoop : bubble 10 times with a 3 second delay between each, then acquire PC vent in [loop (open close) ] / acquire Coil region spectrometer Solenoid Valve Needle Valve recirculation bubbling w/in field within field

PAGE 148

148 7.3 Results and Discussion The primary difference between DOPE and DOPC is the headgroup moiety (see Chapter 3). The substitution of the three methyl groups by hydrogen, N ( CH3)3to NH3, has large consequences on the interfacial structure; not only does the hydrogen bonding scheme change but modifications in both the bilayer swelling limit and vesicle fluidity can occur. While the orientation of water molecules around PC membrane is characteristic of solvation shells of hydrophilic solutes (e.g. clathratelike structures), PE posses hydrophobic-like solvation in which the water oxygens oriented towards the nitrogen.[187] This may result in stronger hydrogenbonding interactions for PE headgroups as we ll as stronger inter-h eadgroup hydrogen bonding, leading to a more restricted motion.[325, 326] Thus, the PE headgroup protrudes further into the bilayer and is less fluid, allowing for the fo rmation of hydrophobic pockets at the lipid-water interface.[327] To begin, we looked at xenon partitioning as a function of the DOPC/DOPE molar ratio utilizing the NMR exchange theory presented in Chapter 4. As was done previously, the values were plotted as a function of total lipid concentration and fit to Eq. (4.2). The only difference is that the lipid composition has change d. This is shown for se veral lipid compositions in Figure 7-5A. All spectra were obtained at 298 K, a 129Xe overpressure of 2.5 atm and fixed volume. From previous results,we know that it is highly likely that xenon interacts at the membrane-water interface at low [ Xe ]aq. This is further substantiated by the DOPE data. As seen in Figure 7-5B, the partition coefficient de creases with increasing DOPE mole fraction X (DOPE). Whats more, the association constant also appears to decrease with higher DOPE content isotherm (Figure 7-6A). The associat ion constants were determined by plotting [ Xe ]bound

PAGE 149

149 /[ Xe ]aq as a function of lipid concentration, wher e the bound-to-free xenon ratio was calculated from the following equation: max.bound aqXe Xe (7-4) As mentioned previously, is the experimentally observed ( ) chemical shift difference between the fast exchanging xenon between the L UVs of various composition and the bulk water system. The predicted maximum shift difference ( max) was obtained from the fit of experimental results to Eq. (4.2). The extracted Ka values were found to be proportional to the mole fraction partition coefficient ( Ka = Kp/55.5 M ), suggesting ideal binding and the Henrys law region of the binding isotherm (Figure 7-6B). Numeric values for Ka were found to be 7.9 M-1, 5.8 M-1, 4.1 M-1, 3.7 M-1 for 0, 0.125, 0.25 and 0.50 mole fractions of DOPE in DOPC containing liposomes, respectively. Again, it shoul d be made clear, that both inner and outer vesicle leaflets were consider ed available for binding. Figure 7-5. Influence of DOPE doping on th e xenon-membrane mole fraction partition coefficient. A) Experimental data fit to Eq. (4.2) for various lipid compositions. Here, ( ) denotes pure DOPC, ( ), (7:1) DOPC:DOPE, and ( ), (3:1) DOPC:DOPE lipid ratio. B) Shows the influence of DOPE doping on the xenon-membrane mole fraction partition coefficient. Errors are reported as (SSE). 1.5 1.0 0.5 0.0 (ppm) 100x10-3 80 60 40 20 0[LIPID]total (M) 400 350 300 250 200 150Kp 0.6 0.5 0.4 0.3 0.2 0.1 0.0X(DOPE) (a) (b) A) B)

PAGE 150

150 Figure 7-6. A comparison in binding and partitioning behavior for DOPE-free and DOPEcontaining membranes. A) ( ) denotes pure DOPC and ( ) the (1:1) DOPC:DOPE lipid mixture, yielding Ka values of 7.9 M-1 and 3.7 M-1, respectively. B) A comparison between Ka values obtained from Eq. (7-4) ( ) and Kp/55.5 M (*) determined from Figure 7-5B. Recent studies have shown that the membrane spanning peptide gramicidin is particularly sensitive to lipid composition.[328] Gramicidin is shorter than th e hydrophobic thickness of either DOPC or DOPE, which results in a hydrophobic mismatch (Figure 7-1). As explained earlier, this mismatch introduces both curvature and elasti c strain. The introduction of halothane led to a 10 fold decrease in channel lifetime in DOPC bilayers; the magnitude of this effect was shown to be dependent on both lipid composition and anesthe tic concentration. The lifetime of the channel was shown to be inversely proporti onal to the anesthetic concen tration; the partitioning of halothane was reduced 3-fold in DOPE cont aining membranes compared to the single component DOPC lipid matrix. The tighter lipid packing of DOPE actually reduced the effect of halothane on the channel due to reduced par titioning. Theoretical predictions by Cantor (2001)[158] indicate that in the limit of low solute concentration the partitioning of short n alkanols should decrease in the presence of bot h DOPE and cholesterol, consistent with our observations and those of Weinrich et al. (2009).[328] 0.6 0.4 0.2 0.0[Xe]bound/[Xe]aq 100x10-3 80 60 40 20 0[LIPID]total (M) (a) (b) 8.0 7.5 7.0 6.5 6.0 5.5 5.0 4.5 4.0 3.5 3.0Ka (M-1) 0.5 0.4 0.3 0.2 0.1 0.0X(DOPE) A) B)

PAGE 151

151 The reduced partitioning may also be explained in terms of changes in the membranes curvature elastic energy (curv); modifications in the membranes chemical potential due to increased molecular strain can have repercussions on the Kp value since it is related to the difference in the chemical potential between phases. This means that curv should be directly proportional to the changes in the spontaneous curvature, co. The net membrane curvature can be expressed as a weighted average of the curv atures of the individua l lipid components. For lipid membranes containing mixtures of DOPC a nd DOPE, the curvature can be formulated as a weighted sum: 1mixture ooDOPEDOPEoDOPCDOPEccXcX (7-5) where mixture oc is the membrane curvature of the specified lipid composition, XDOPE is the mole fraction of DOPE in the system, and co,DOPE and co DOPC are the values for the intrinsic curvatures for DOPE and DOPC, respectively. Using the data from Figure 7-5B, we calculated the change in the energy of transfer ( ) of xenon between the aqueous and bilayer phase as a function of both spontaneous curvature (mixture oc) and DOPE mole fraction. As mentioned previously, Eqs. (5-9)) relates experimentally obtained chemical shift data to the mole fraction partition coefficient, while Eq. (5-10) allows for the approximation of Lewis and Cafiso showed that the energy of transfer of the membrane sp anning peptide alamethici n was linearly dependent on the membrane curvature.[329] Our results are consistent with their observations at low DOPE mole fractions. Once the DOPE mole fraction reac hes 50%, the linearity does not hold. This may be due to xenon induced curvature effects at the lipid interface. The ch ange in the energy of transfer, as fit to the linear segment of Figure 7-7, yields a slope of 1.5 kcal/mol per mole fraction of added DOPE

PAGE 152

152 Figure 7-7. The change in the transfer energy, as a function of mole fraction of DOPE and membrane curvature, mixture oc, for various DOPC/DOPE mixtures. Errors are reported as (SSE). Figure 7-8. Variation in NM R parameters of DOPE containing lipids as a function of Xe overpressure. A) The observed as a function of %DOPE for various PXe: 2.5 atm ( ), 5 atm (green), 8.5 atm ( ), and 10 (red) atm. B) Normalized Bruker peak intensity with increasing xenon overpressure s for various lipid compositions: 0 ( ), 0.125 ( ), 0.25 ( ), and 0.50 ( ) mole fraction of DOPE. No te that the total lipid concentration and sample volumes were fi xed to 50 mM and 1.5 ml, respectively. Errors are reported as (SSE), resulting from deviations in PXe. 600 500 400 300 200 100 0 (cal/mol) 0.50 0.40 0.30 0.20 0.10 0.00X(DOPE) -28x10-3 -24 -20 -16 -12comixture (Angstrom-1) 1.8 1.6 1.4 1.2 1.0 0.8 (ppm) 0.5 0.4 0.3 0.2 0.1 0.0X(DOPE) (a) (b) 1.0 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2Normalized Peak Intensiy 10 9 8 7 6 5 4 3PXe(atmospheres) A) B)

PAGE 153

153 7.3.1 Effects of Xenon Doping and Temperature In addition to molecular stress imposed by the addition of DOPE to DOPC lipid membranes, we also studied the effect of external strain brought on by xenon doping and temperature in order to better characterize the e ffect of packing frustration within the bilayer phase. The general trends have been summarized in Figure 7-8A. While a steady downfield shift is observed with xenon doping, there is a distinct change in chemical shift behavior with increasing DOPC/DOPE molar ratio. In line with experimental observations, let us assume that an upfield shift reflects decr eased interaction between xenon and the lipid environment due to reduced van der Waals interactions, and a dow nfield shift corresponds to increased xenonmembrane interactions, according to the spectr oscopic trends shown in Figure 7-5. Using previous chapters as a guide, it would seem that decreased partitioning is observed with increasing PE content when PXe 5 atm; the downfield shift at higher overpressures are indicative of increased partitioning and/or binding. As the xenon overpressure is increased above 5 atm, the ability of DOPE to inhibit partitioning seems reduced. This is further substantiated by Figure 7-8B, where the spectral intensity is plotted with respect to xenon loading pressure and percent DOPE in a 50 mM lipid mixture with DOPC at 298 K. As the peak intens ity is proportional to the area under the peak, one may assume that the amount of xenon associat ed with the lipid memb ranes increases with pressure, consistent with singl e component lipid vesicles, studi ed in previous chapters. The increased interaction at higher PE may be a resu lt of partitioning into the lipid core due to a disruption in the membranes elasticity or changes in xenon diffusion propertiesDOPE decreases the effective membrane surface area which could reduce the rate of diffusion.

PAGE 154

154 Recent investigations on th e dibucaine-membrane interac tion have shown that the anesthetic modifies the molecular packing of monolayer or bilayer su rfaces due to in teractions at the lipid headgroup region. [330] Increased anesthetic concentr ations led to longer incubation times as well as local disruptions at lipid-wate r interface. Cotta el al. (2009) described this anesthetic induced disruption as a two step pro cess. First, the lipid monolayer becomes packed with anesthetic resulting in a maximum elasticity, at which point dibucaine inserts deeper into the bilayer leading to larger local stresses and ch anges in curvature. Small angle X-ray scattering and 2H NMR experiments have also shown that higher concentrations of inhalation anesthetics i) displace the water molecules at the water-lipid interface (dehydration), ii) solubilize into the hydrophobic core and iii) encour age the loss of curvature.[331] Chloroform, for example, has even been shown to form structures with periodic curvature (e.g. cubic phase) when present at high concentrations in several types of lipid vesicles. As discussed by the authors, the driving force between these transitions is the change in th e average molecular shape of the lipid (shape concept of lipid polymorphism, see Chapter 3). In order to see whether the changing chemical sh ift behavior is due to structural changes in the lamellar phase, we looked at the 31P NMR data of xenon dissolved in (1:1) DOPC/DOPE lipid mixture. As seen in Figure 7-8A, the differe nce in the chemical shift behavior at low and high xenon overpressures is the greatest for this sample. While the 31P NMR chemical shift at low xenon concentrations resulted in a single, br oad isotropic resonance centered around 0 ppm, three distinct resonance peaks were observed in the (1:1) DOPC/DOPE lipid mixture (Figure 7-9). The upfield 31P resonance at 0 ppm (Figure 7-9A) will now be referred to as the lamellar phase as it is similar to isotropic peak obs erved in the single co mponent DOPC LUV system. The two additional resonances includes one of low intensity at3.94 ppm (Figure 7-9A),

PAGE 155

155 Figure 7-9. 31P NMR spectrum of the (1:1) DOPC/DOPE lipid mixture at 298 K and 10 atm. A) characteristic of HII phase while C) is due to extruded LUVs (0 ppm). The intermediate peak B) may represent lateral diffusion of lipids between curved and non-curved states: B) appears at th e average of resonance A) and C). Figure 7-10. Temperature dependence of the observed chemical shift (obs) for various lipid compositions and pressures; ( ) and ( ) are (7:1), (3:1) DOPC/DOPE molar ratios at 10 atm, respectively. Here, ( ) denotes (7:1) DOPC/DOPE at 2.5 atm of xenon and (*) is lipid-free buffer solution at 10 atm for comparison. 190.0 189.5 189.0 188.5 188.0 187.5 187.0 186.5 186.0 obs (ppm) 335 330 325 320 315 310 305 300 295 290 Temperature (K) A) B) C) PPM 10 6 2 -2 -6 -10 -14 Temperature (K) obs (ppm)

PAGE 156

156 characteristic of the inverted hexagonal phase, and a narrow resonance of greater amplitude at 1.91 ppm (Figure 7-9B) located di rectly between the lamellar (Figure 7-9C) and the curved phase (Figure 7-9A). This middle peak may s uggest some lateral diffusion between the two states or indicate the formation of an intermediate structure. So, while it is clear that xenon does not induce a full phase transition on its own, it does appear to promote nonbilayer structures when present at higher concentrations. The in creased xenon-membrane interaction at higher xenon overpressures may facilitate larger local st ress at the membrane interface, leading to our observed changes in curvature. Since increasing temperature has the same eff ect as dehydration, we attempted to monitor changes in the interfacial curvature thermotropically Investigations into the thermal properties of the xenon-membrane interaction are consistent with previously observed trends in single component lipid systems: all samples display a downfield shift with increased xenon loading and exhibit a maximum at 290 K before shifting monotoni cally upfield. Similar to Figure 7-8B, there is no noticeable difference between the observed chemical shifts at higher loading pressures ( PXe 10 atm), which suggests similar thermodynami c behavior at elev ated temperatures. Furthermore, the chemical shifts of all samp les seem to converge at higher temperatures, signifying faster exchange condi tions of xenon between phases. 7.3.2 Lamellar-to-Inverted Hexagonal Transition As mentioned previously, the DOPC/DOPE mixture can be described as a highly cooperative, two-component syst em; the higher temperature stru cture possessing higher entropy (DOPE) and the lower temperature structur e having lower entropy (DOPC). A macroscopic rearrangement of the lipid morphology occurs when these entropies are energetically equal. The L-toHII phase transition can be induced thermotropi cally between 323-328 K in fully hydrated

PAGE 157

157 (1:1) DOPC/DOPE mixtures and the process is char acterized by slow lipid exchange between the two structures.[332, 333] We attempted to detect this structural rearrangement via 129Xe NMR by heating the sample to the temperature region specified. In doing so, we observed the emergence of a second peak downfield from the pre-existing chemical shift trend at 318 K (Figure 7-11), likely indicating the coexisten ce of both lamellar and inverted hexagonal phase. As the temperature was increased, the 129Xe chemical shift appears to merge with the previously identified upfield trend (Figure 7-10). This experiment was repeated with 56.5% and 60% PE content under identical experime ntal conditions. The slight incr ease in the DOPE mole fraction resulted in significan t depressions in the L-to-HII phase transition temperature ( Th). Results are summarized in Figure 7-12A. It shou ld be noted that the second re sonance was not observable at loading pressures under 10 atm. In order to verify the reproducib ility of the chemical shift beha vior at low concentrations of Xe, a 2% mixture of hyperpolarized xenon was bubbled into aqueous dispersions of (1:1) DOPC/DOPE LUVs. The specifics of this process are described within th e Experimental section of this chapter. As seen in Figure 7-11B, the 129Xe NMR resonance shifts downfield at approximately the same temperature for both th ermally polarized and hyperpolarized methods. Similar studies have shown depressions in the phase transition temperature to be dependent on the concentration of anes thetic in solution. This appears not to be the case here. In studying the effect of anesthetics on the thermal transitions of DPPC vesicles via 1H NMR, Yokono et al. (1981) demonstrated that anesthetic-induced transitions do not occur simultaneously; the headgroup protons responded at lower anesthetic concentrations than the hydrocarbon region.[334] While it is likely that the presence of xenon f acilitates structural re arrangement by introducing additional stress to the lipid-wa ter interface, no notable depend ence on the xenon concentration

PAGE 158

158 Figure 7-11. Variation in the observed NMR parameters of xenon dissolved in a (1:1) DOPC/DOPE 50mM lipid mixture with increasing temperature. A) 129Xe NMR spectra of thermally polarized xenon at an overpressure of PXe 10 atm. B) Summary plot of the temperature dependence of the L-toHII transition: ( ) and ( ) signify experimental resu lts using hyperpolarized 129Xe NMR bubbling sequence and thermally polarized xenon, respectively. was observed via 129Xe NMR. However modest changes in the overall DOPE mole fraction was shown to have large effects on Th; the higher the DOPE content, th e greater the depression in the transition temperature. Changes in the transition temperature. A simple relation has been made between the packing parameter and shifts in the L-toHII transition temperature by Marsh et al. (1996).[81] According to this proportionality (Eq. (7-6)), the phase transition temperature should decrease with increasing packing parameter value. Incr easing the mole fraction of DOPE within the DOPC lipid matrix enhances the packing parameter, which in turn depresses the transition temperature. 1 21 2bil coA hV curv SS A l hh R oNA T (7-6) Since anesthetics tend to stabiliz e the higher entropy structure ( HII phase), it should not be surprising that xenon does the same.[103] When a solute is incorporated into a lipid matrix its 191 190 189 188 187 186 obs (ppm) 340 335 330 325 320 315 310 305 300 295Temperature (K) (a) (b) PPM 193 192 191 190 189 188 187 186 185 obs (ppm) 328 K 323 K 320 K 318 K 313 K 308 K Temperature (K) A) B)

PAGE 159

159 degree of stabilization is reflected in the transi tion temperature. Assuming the solute does not exhibit preference for the lamellar or inverted he xagonal phases, the enthalpic contribution of the solute ( Hs) to the transition is expressed as curvatur es stress. As discussed in detail by Janes (1996),[335] solute induced changes in the packing parameter within th e lamellar phase are relaxed and stored as potential energy in the inverted hexagonal phase. Changes in the enthalpic curvature stress imparted on the lipid matrix are manifest in increases/decreases in Th and can be quantified through changes in the transition entropy ( Sh). In accordance with Eq. (7-1), modifications in the chemical potential of the bilayer phase ( bil) due to changes in curvature or packing parameters in the presence of small molecules, has the potential to effectively shift the phase transition temperature. [335] For a pure DOPE bilayer, the thermal entropy of change from the lamellar to inverted hexagonal phase is 1.03 cal/moldeg. If the addition of a solute decreases Th by one degree, Hs Figure 7-12. Potential effects the addition of a solute can have on the lamellar-to-inverted hexagonal phase transition temperature. A) The L-toHII transition temperature ( Th) as a function of DOPE mole fraction: (*) indicates reported values in fully hydrated (1:1) DOPC/DOPE mixtures without xenon.[332] ( ) and ( ) denote experimentally obtained values via thermally polarized and hyperpolarized 129Xe NMR, respectively. B) Depression in Th as a function of 21o R Results were obtained at an overpressure of 10 atm of thermally polarized xenon. Th e solid line denotes the fit to Eq.(7-6). (a) (b) 1.04x10-3 1.00 0.96 0.92 0.88 1/Ro2 (Angstrom-2) -18 -16 -14 -12 -10 -8 TH (K) 325 320 315 310 305 300 TH ( K ) 0.75 0.70 0.65 0.60 0.55 0.50 X(DOPE) TH (K) TH ( K ) A) B)

PAGE 160

160 would amount to 1.03 cal. For example, tetradecane has been shown to depress Th by 4.5 C/mol% in DOPE-Me.[335] Thus, the enthalpic contribution of the solute to the transition is13.5 cal. Using this as a guide, we employed Eq. (7-6) to estimate Sh by plotting the change in the phase transition temperature as a function of 21o R (Figure 7-12 B). This yields an average value of 4.63 0.23 cal/mol K for Sh. Similar to the expression provided for the intrinsic curvature in Eq. (7-5), we approximate bil o A according to the following expression: 1mixture ooDOPEDOPEoDOPCDOPEAAXAX (7-7) where Ao,DOPC and Ao,DOPE are 82 2 and 65 2, respectively. The radius of the intrinsic curvature Ro was calculated from Eq. (7-5) since 1mixturemixture oocR and the mean-curvature elastic modulus, c, was set to 9.9kBT .[336] These results suggest that xe non induces 4.68 cal of enthalpic curvature stress per degree, which corresponds to 37.4 cal/mol for the (1:1) DOPC/DOPE lipid mixture with 10 atm of xenon overpressure. Looking back at the thermodynamic analysis of the DOPC lipid we are reminded that the maximum entropic contribution to the chemi cal shift occurs at 317 K and the maximum enthalpic contribution at 324 K. As discussed previously, mS (317 K) defines the maximum partitioning limit and m H (324 K) the minimum in the molar chemical potential of transfer of xenon between the aqueous and bilayer phases. A ssuming that the thermal partitioning behavior of xenon in the DOPC/DOPE lipid matrix is co mparable to the observed trends in the DOPC lipid system we can expect that the Kp increases with temperature. So, it makes sense that the xenon-induced change in the transition temperature would be n ear the maximal partition limit since it likely leads to higher degree of dehydration. Accordi ng to recent X-ray diffraction studies on fusion intermediates,[337] inverted hexagonal phase wa s formed in at (1:1)

PAGE 161

161 Figure 7-13. Phase diagram of various DOPC/DOPE lipid mixtures. A) Phase dependence as a function of relative humidity (RH) and mo le fraction of DOPE at 298 K determined by X-ray diffraction methods. B) Phase dependence of (1:1 ) DOPC/DOPE lipid mixture with increasing te mperature and hydration: L-, rhombohedral lattice (R), distorted hexagonal ( HII)-, HII -, and mixedphases. Figures modified from Yang, Ding, and Huang (2003).[337] DOPC/DOPE lipid mixture at and 53-95% rela tive humidity (Figure 7-13A). This further supports the dehydrating effect of xenon and temperature on the mixed DOPC/DOPE dispersions. Consistent with Eq.(7-6), increasing the relati ve amount of DOPE within the lipid mixture increases the chemical potential of the lamella r phase and lowers the transition temperature (Figure 7-12A). While it is expected that the phase transition temperature will decrease with increasing DOPE content, the phase diagram provided by Yang, Ding, and Huang (2003) shows that a significant reduction in the hydration is required for full inverted hexagonal phase formation in PC/PE ratios less than (1:2). According to our resu lts, the solute-induced enthalpic curvature stress increases with DOPE content; Hs was determined to be 74.8 cal/mol and 84.2 cal/mol for 0.57 and 0.60 mole fraction of DOPE, respectively. A) B)

PAGE 162

162 In an attempt to elucidate the downfield sh ift observed in Figure 7-11A (at 318 K) we dissolved thermally polarized xenon into (1:3) DOPC/DOPE lip id mixture. At low lipid concentrations a single resonance is observed near the lipid-free buffer solution, signifying fast exchange conditions. As the lipi d concentration is increased, a second peak is resolved downfield. This trend differs from those obs erved in the single component DOPC system. Consistent with 19F NMR studies of anesthetic partitioning into lipid membranes, the 129Xe NMR resonance associated with the water-lipid interfac e is near that of the lipid-free buffer solution at dilute DOPC concentrations. As the DOPC concentr ation is increased, so does the chemical shift difference between.[16] Non-immobilizers, which have b een shown to partition within the membrane core, do not exhibit the same chemical shift behavior. Since they do not reflect interactions at the membrane-bu ffer interface, these resonances a ppear at distinctly different chemical shifts. According to this logic, th e upfield resonance shown in Figure 7-14 can be associated with xenon-membrane interactions at the lipid interface, while the downfield peak can be attributed to xenon partitioning into inters titial voids produced by the inverted hexagonal phase. The high degree of packing frustration in the HII phase is a result of inefficient geometric packing of the inverted cylinders Interstitial voids form, requiring the acyl chains to stretch or compress away from their preferred morphology. For the packing energy to stabilize the inverted phase there must be a lowering of ch (Eq. (7-1)) In order to do so, xenon must populate the interstitial region and relax the acyl chain packing stress. General anesthetics have been shown to remain at the lipid-water interf ace area without losing c ontact with bulk water or penetrating into the terminal methyl region of the acyl chains.[143] The upfield resonance in Figure 7-14 suggests that xenon remains in contact with th e interfacial region. And as the HII packing efficiency

PAGE 163

163 improves with increasing lipid concentrations, so does the xenon exchange rate between the interfacial and acyl chain domains. It should be noted that while two 129Xe NMR chemical shifts were observed in the (1:3) DOPC/DOPE lipid mixt ure at a xenon overpressure of 2.5 atm, only a single resonance was exhibited under similar experimental conditions for xenon dissolved in DOPC lipid dispersions. Figure 7-14. Concentration dependence of the 129Xe chemical shift dissolved in (1:3) DOPC/DOPE MLV lipid mixture compared to the chemical shifts of a A) 100 mM DOPC LUV sample and B) lipid free buffe r solution, under similar experimental conditions. The blue spectrum is 12.5 mM, the red spectrum, 25 mM, and the black, 100 mM concentrations of (1:3) DOPC/DOP E, respectively. All spectra were obtained at 298 K and 2.5 atm of xenon. 7.3.3 Evidence of Kinetically Trapped Structures Cubic phases refer to a family of ordered nonlamellar, liquid crystalline lipid phases having cubic symmetry. These structures can be formed directly from lamellar phases as end products, or as intermediates in the formation of nonbilayer phases (eg. HII). Though these geometries are difficult to detect due to their se nsitivity to temperature, scanning rate and the A) B) PPM 190.8 190.0 18 9.2 188.4 187.6

PAGE 164

164 sheer variety of cubic phases, some of these structures can remain kinetically trapped once formed, and can only be relaxed after cooling the sample below the gel/lamellar phase transition. As mentioned previously, certain molecules can stabilize the fo rmation of these structures.[338-340] Experimental evidence suggests that the formation of the cubic pha se is accelerated between the L-toHII transition temperatures and can be induced by cooling or heating. [341] The rhombohedral phase is another example of a transition intermediate of DOPC/DOPE lipid mixtures. Evidence suggests that this is a stable phospholipid stalk phase, similar to structures shown in Figure 7-2.[310] Like the cubic phase, the rhom bohedral phase exists between the lamellar and inverted hexagona l lipid morphologies and is sens itive to both temperature and hydration levels. Molecular dynamics simulations s uggest that elongated st alk structures formed in pure DOPE lipids can drive th e lipid matrix directly into the inverted hexagonal phase.[342] Unlike the cubic phase, these stalks are orde red in a hexagonal pattern. As shown in the DOPE/DOPC phase diagram (Figure 7-13B), the pr opensity of mixed phase formation is higher at elevated temperatures. A 31P NMR spectrum of the (1:1) DOPC/DOPE LUV lipid mixture was taken after induction of the inverted hexagonal phase (after cooling) in order to confirm the relaxation of the lipid matrix out of the inverted hexagonal phase Doing so resulted in a lineshape with three distinguishable characteristics: a broad peak upfield associated with the lamellar phase, a smaller resonance at 1.10 ppm, and a third at 3.57 ppm, indicative of the inverted hexagonal phase. The xenon containing sample was then subjected to a series of freeze/thaw cycles one week after cooling; it was immersed in a bath of liquid nitr ogen five times per cycle. This resulted in an increase in the lamellar phase and a sharpening of the resonances associated with the nonlamellar

PAGE 165

165 phases (Figure 7-15C). Thus, it appears that th e presence of xenon in the lipid dispersions stabilize these nonbilayer structur es long after the inverted hexa gonal phase has been induced. Figure 7-15. 31P NMR spectra of the different lipid phase s in the presence of xenon. A) Single component lipid DOPC as LUVs, B) Single component DOPC as MLVs. C) (1:1) DOPC/DOPE lipid mixture at 328 K. D) (1:1) DOPC/DOPE mixture at 298 K, 1 week after cooling, following 2 separate freeze/thaw cycles. 7.4 Conclusions In summary, our results show that the in teractions of xenon w ith lipid membranes containing DOPE are much more complex that those in the single DOPC lipid matrix. At low xenon loading the presence of DOP E appears to inhibit the xenon partitioning, decreasing the association constant with incr easing PE content. As the conc entration of xenon in solution increases, the lipid matrix appears to lose this capability. Dispersions with high DOPE mole ratios were shown to be especially sensitive to the presence of xenon. 31P NMR of (1:1) DOPC/DOPE mixtures under 10 atm of xenon overp ressure shows the existence of nonbilayer phases prior to heating. This is likely attributed to alterations in the lateral bilayer organization A) B) C) D) PPM 70 60 50 40 30 20 10 0 -10 -20 -3 0 -40 -50 -60 -70

PAGE 166

166 due to displacement of water molecules at the me mbrane interface. Furthermore, the presence of xenon was also shown to promote changes in the thermal behavior of the (1:1) DOPC/DOPE lipid. The phase transition temperature was depr essed approximately 8 K, consistent with changes in the packing parameter. Since change s in the phase transiti on temperature indicate changes in the membrane curvature, xenon likely induces curvature defects at high concentrations and temperatures in DOPE containing lipid membranes. Interestingly, once the nonbilayer structures were formed, they were unab le to be fully relaxed into the lamellar phase. 31P NMR suggests the coexistence of three lipid phases even after being subjected to multiple freeze/thaw cycles. Regulation of the membrane pressure is esse ntial for cellular function. There is a whole family of membrane proteins called mechano-sensitive channels that depend on membrane pressure to open and close. Take for example the bacterial protein MscL. When the bacteria experiences significant stress, it opens in order to relieve the pressure. Agents that induce asymmetry in the membrane environment introduce curvature and thus pressure, which can trigger the opening of the channel. We have s hown that xenon can induce a significant defect in the bilayer structure in the pres ence of nonbilayer lip ids and that it stabil izes the formation of curved structures long after they should ha ve been relaxed. It is our view that 129Xe NMR is uniquely suited for the elucidation of the anes thetic-protein interaction in membrane-supported environments.

PAGE 167

167 CHAPTER 8 LIPID NANOTUBE ARRAYS INVESTIGATED BY HYPERPOLARIZED XE-129 NMR 8.1 Introduction The biofunctionalization of inor ganic substrates has become one of the most attractive methods to study the physiochemical properties of biomedia in confined systems. Interest in assembly and applications of nonmaterials fo r new technologies has prompted integrative research in chemistry, physics and biology.[343-347] As mentioned previously, the low density and long relaxation time of disso lved thermally polarized 129Xe prevents it from becoming a viable tool to study biological and inor ganic materials in solution within moderate physical conditions and experiment times. The only way to circumvent this problem is through hyper-polarization techniques. Inorganic substrates, such as A nodic Aluminum Oxide (AAO) membranes, can be functionalized by self assembly of lipid bilayers on their surfaces, creating a convenient model of cellular membranes.[348-350]Although these membrane systems ar e generally used for filtration purposes (e.g., lipid extrusion), its use in the fabr ication of lipid nanotube arrays has emerged as a new trend in biotechnology.[351-353] Their confined surface and increased stability reduces the likelihood of surface perturbation and contamination, making these lipid nanotube arrays more robust than mechanically aligned bilayers. Furthermore, these lipid nanotube arrays can retain water for a longer duration through capillary action; the network of tightly bound water molecules at the lipid interface is not easil y removed. And while lipid diffusion between channels does not occur, the surfaces of both l eaflets are fully accessible to aqueous solutes. The increased used of membrane filters in th e fabrication of nanostructured materials can be attributed to their i) commercial availabil ity, ii) tunability of its pore dimensions, iii) high thermal stability, iv ) relative ease of lipid a dherence v) long shelf lif e and vi) the ability to dehydrate/rehydrate with full sample viability. As such, the cylindrical lipid bilayers formed

PAGE 168

168 inside AAO nanotubes are ideal for structure-f unction studies of membrane proteins, cell signaling, and ligand-receptor in teractions within a model membrane environment. Herein, we study the effects of nano-scal e confinement on the spectra l properties of dissolved 129Xe in selfassembled phospholipid membranes. In addition to verifying effective bilayer formation on the substrate via 31P MAS, we monitor the xenon gas-to-lipid membrane exchange via 2D EXSY. 8.2 Experimental 8.2.1 Materials As mentioned in previous chapters, DOP C was purchased from Avanti Polar Lipids (Alabaster, AL). The 50 mM (pH 7.4) Hepes Bu ffer solution used for hydration was purchased from Sigma-Aldrich. Anodic Aluminum Oxide (aka, ANOPORE inorganic membranes, Aluminum Oxide Membranes, AAO) filters, havi ng 200 nm pore size and 47 mm diameter were purchased from Whatman Internat ional Ltd (England). Material s were used as supplied. 8.2.2 Physical Description of AAO ANOPORE inorganic membranes are available commercially from Whatman in a variety of pore sizes: 20 nm, 100 nm, and 200 nm. The hydrophilic nature of the membrane makes it highly compatible with most aque ous bio-material and solvents. As shown in Figure 8-1(A-B), these aligned, through-film porous structures are macroscopically homogenous and assembled in a hexagonal-like structure in which each por e represents a separate channel (60 m long). However, recent NMR studies have shown the presence of short, repetitive segments, approximately 3-5 m in length (Figure 8-1C).[354] The high heterogeneity observed in commercial AAO solid supports often complicates the interpretation of the effects of nanoporeconfined lipid bilayers, prompting researcher s to synthesize customi zed support materials. Despite this, we find the commercially available membranes to be adequate for our preliminary investigations. All reported sizes are as specified by Whatman International.

PAGE 169

169 Figure 8-1. The side and top views of an AAO membra ne. A) AAO possesses hexagonal packing structure. B) AAO with 200 nm channel diameter, from Whatman International. C) Model for lipid adsorption to the surf ace of pores in AAO; upper C): lipids adsorb as wavy tubules with water remaining trapped between these tubules and the AAO surface; lower C): the relative sh ape of the lipid bilayer tubules inside the AAO pore.[354] 8.2.3 Preparation of Lipid Nanotube Arrays Instructions on how to prepare AAO supporte d bilayers are well documented in the literature.[355-358] We followed the following guidelines for manual deposition of DOPC onto these supports. First, one side of the AAO support was exposed to an aqueous suspension of MLV which led to an immediate wetting of the A AO disc, making it semitransparent to the eye. The main phase transition temperature of DOPC is well below room temperature, so constant heating was not required. Both sides of th e AAO disc were cleaned by repetitive wiping (Kimwipes EX-L) in order to remove any excess lipid from the membrane surface. The sample was then crushed with a motor and pestle and placed into the wideline probe, which held approximately 20 mg of sample (AAO, lipid, and water). In the case of shift-reagent studies, DyCl3 was dissolved into buffer solution (50 mM) and the disc soaked in the mixture for 30 minutes before being wiped of excess lipid, crushed, and placed within the sample holder. More than one layer of substrate is often de sired for NMR studies as the bound quantity of phospholipids is proportional to the total coverage area (one bilayer per pore).[359] Generally a series of approximately 50 or more commerc ial AAO discs are stacked and oriented at a particular angle in the magneti c field. For NMR experiments, AAO discs were stacked in the (a) (b) (c) A) B) C)

PAGE 170

170 rotor and this design had previously been used to assess the degree of alignment of the lipid bilayer inside the nanoporous support; having prec ise control of the alignment with the magnetic field is imperative for more intricate st udies. A new sample holder was designed to accommodate the increase in volume required for this process, the details of which can be found in the Appendix. The diameter is approximately 2.9 times larger th an the previous sample holder, providing a significant increase in the total available sample volume. This particular design was able to hold vacuum at temperatures 208 K. 8.3 Results and Discussion 8.3.1 Adsorption of Single Lipi d Bilayers onto Substrates Substrate supported bilayers were deposited inside the 200 nm pores of AAO filters using methods described in the previous section. 31P NMR spectroscopy was employed to verify proper adhesion of bilayer to the inne r surface of the pores Consistent with published results, the 31P NMR of lipid-free AAO results in a significantly broad spectrum in both the static and MAS spectra.[358] This is due to the low levels of phosphorus contained in the AAO (7.4 wt%).[360]The blue line in Figure 8-2 is the static 31P spectra of unoriented lipid bi layers incorporated into AAO substrate; neither decoupling nor presence of pa ramagnetic shift reagent has a noticeable effect on the overall lineshape. Recent studies show that only the first bilayer adheres strongly to the inner surface of the pore. Add itional bilayers layers can be easily washed out by water. As discussed previously, paramagnetic shift reagent can significantly broaden the 31PNMR signal, effectively reducing the contribution of loose lipid molecules to the chemical shift if present at high enough concentrations. According to our results, the addition of 50 mM DyCl3 reduced the overall spectral intensity, but did not result in a noticeable shift in the resonance (Figure 8-3D). The isotropic peaks for both phosphoric acid (standard) and hydrated, AAO supported DOPC for

PAGE 171

171 Figure 8-2. Various 31P NMR spectra of AAO supported DOPC b ilayers: red, sharp line denotes the MAS spectra, (spinning rate 10 kHz) with high power decoupling; the blue line is the static 31P NMR spectra on the same sample; and the broader black line is the static spectra of the lipid-fr ee AAO powered sample. comparison. Results presented in Figure 8-3 show a number of 31P NMR spectra obtained over a period of days, in se quential order. The and II values of the 50 mM DOPC MLV 31P NMR spectra were 24.22 ppm and -15.65 ppm, respec tively. A graphical re presentation of xenon-AAO supported bilayer interaction is given in Figure 8-4. Xenon first dissolves into the water pore, and then diffuses through the lipid bilayer. It ma y be possible for lipids to become clogged in the pores. This can neither be c onfirmed nor disproven using 129Xe or 31P NMR methods. 8.3.2 Evidence for 129Xe inside AAO Pores A series of simple 1D NMR spectra of 129Xe dissolved into various lipid phases is shown in Figure 8-5. To begin we looked at the li pid-free, crushed AAO support to verify limited interaction between its surface and gas phase xenon ( Xe( g )); the temperature dependence of the xenon-AAO surface interaction is provided within th e Appendix. Note the lack of signal in the aqueous region of the spectrum. Thus it can be concluded that Xe-Xe in teractions on the AAO PPM 40 30 20 10 0 -10 -20 -30 -40 -50

PAGE 172

172 Figure 8-3. 31P NMR spectra of DOPC under various conditions. A) Chemical shift of phosphoric acid reference: iso = 0 ppm. B) DOPC bilayer on AAO substrate (crushed, 200 nm pore size): iso = -0.836 ppm. C) DOPC b ilayers rehydrated with buffer solution. D) Sample washed with 50 mM DyCl3 solution. E) Spectra after sample was soaked in excess buffer solution to remove DyCl3. Figure 8-4. Illustration descri bing the diffusion of xenon to AAO supported bilayers. A) From the gas phase to the aqueous pore. B)The subsequent diffusion of Xe( g ) in the bilayer plane. (a) (b) (c) (d) (e) Xe Xe Xe (b) (a) A) B) E) D) C) B) A) PPM 4 2 0 -2 -4 -6

PAGE 173

173 Figure 8-5. A series of 129Xe NMR spectra of xenon dissolved/adsorbed gas in various AAO environments. A) Lipid-free, crushed AAO substrate using continuous flow HP 129Xe NMR methods: 1 scan. B) 50 mM DOPC LUV solution utilizing thermally polarized 129Xe NMR: 720 scans. C) AAO supporte d DOPC using continuous flow HP 129Xe NMR: single scan. D) AAO supported DOP C employing interrupted flow HP 129Xe NMR technique: single scan. surface will have negligible contribution to the ob served chemical shift ne ar the water and lipid associated peaks. Next we used continuous flow hyperpolarized (HP) 129Xe NMR methods to verify adsorption into the lipid filled AAO channelsa 129Xe NMR spectrum of Xe( g ) dissolved into aqueous dispersions of lipid vesicles ( 50 mM DOPC LUVs) is given for comparison. This particular sample (see Figure 2-8B) had less volume in the coil region and possessed an isolated gas peak which was lined up with the 129Xe(g) of all other spectra and used as a reference. Spectra (C) and (D) of Figure 85 were obtained using continuous and interrupted flow methods, respectively. In the continuous flow mode, the HP gas mixture is continuously recirculated through the coil region at a steady fl ow rate of 90 ml/min, as recorded from the calibrated flow (d) (c) (b) (a) A) B) C) D) PPM 224 220 216 212 208 204 200 196 192 188 184 180 176 172 168 164 160

PAGE 174

174 meter (see Figure 2-8). The flow rate is briefly interrupted before the detection time in the interrupted flow technique, increasing the resi dence time of xenon associ ated with the lipid phase, yielding an increase the signal-to-noise ratio of the 129Xe dissolved within that nanotube phase. This improvement is seen clearl y in Figure 2-8C and Figure 2-8D. 8.3.3 Relaxation Rates of Dissolved Xenon While we have gained significant improvement in the experiment time (seconds opposed to hours), only a single peak is observed due to fast exchange conditions. We then performed a saturation recovery experiment to get a sense of how relaxation times of xenon changes within the lipid nanotube environment as compared to our previously studied lipid suspensions. Continuous flow HP 129Xe NMR was used to determine th e longitudinal relaxation time on confined lipid phase (Figure 2-8C). Results we re fit to both a single and bi-exponential fit (Figure 8-6), yielding values of 0.67 0.23 s-1 and 8.31 2.47 s-1 for the bi-exponential fit and 2.10 0.40 s-1 for the mono-exponential fit. It appears that the bi-e xponential fit is slightly better. The higher valued 1T corresponds to the water pore and the shorter to the supported lipid bilayer; the longer the correla tion time, the shorter the l ongitudinal relaxation time. 8.3.4 Indication of Chemical Exchange Between Anopore and Gas Phase 129Xe Lastly, a 2D EXSY experiment was performed to see whether we could resolve additional information on 129Xe exchanging between the lipid/water/ gas phases. The complete absence of cross peaks at 50 ms indicates no significant ex change between the lipi d and gaseous phase during this period (Figure 8-7A). However, as seen in Figure 8-7B, the exchange between Xe( g ) and xenon dissolved within the lipid associated ph ase gives rise to two small cross peaks at mixing times greater than 100 ms. These mixing tim es are on the same order as those observed by Tallavaara and Jokisaari (2006), who studied 129Xe dissolved into thermo tropic liquid crystals

PAGE 175

175 Figure 8-6. 129Xe NMR saturation recove ry curves for xenon dissolved in AAO supported DOPC bilayers using continuous flow met hods, with a mono-exponential fit (dotted line) versus bi-exponential fit (dashed line) Data fit to broad line in the general vicinity of the fast exchanging li pid/water peak (see Figure 8-5C). confined to a mesoporous Contro lled-Pore Glass (CPG) material.[361] Two distinct sites were resolved: xenon dissolved in the bulk liquid crystal betw een the CPG particles and xenon in the liquid crystal confined to the por es. Two cross-peaks, of equal population were observed at mixing times greater th an 80 ms; the diffusion rate was determined to be very slow. The disproportionality of our cross-peak inte nsities between the gas and adsorbed phase is consistent with Eqs.(2-28) and (2-29) 8.4 Conclusions Recent results suggest that 2D EXSY signals can be amplified under interrupted flow conditions, providing greater sensitivity. [52, 56] The interrupted flow pulse sequence is different than the continuous flow sequence in that the HP gas is halted prior to the first 2 x pulse of the EXSY sequence and is interrupted during the mixing time, changing the magnetization dynamics of 129Xe in both the gas and lipid coated nanotube s. Briefly, this interruption allows the 1.0 0.8 0.6 0.4 0.2 0.0Normalized Signal 10 9 8 7 6 5 4 3 2 1 0Tau (sec)

PAGE 176

176 Figure 8-7. 2D 129Xe EXSY spectra. A) AAO supported DOPC at m = 50 ms. B) AAO supported DOPC at m = 120 ms. Ptot = 2983 Torr (2% Xe /2% N2 /96% He), and 298 K for both. gas sufficient time to accumulate in the more energetically favorable phase by increasing the residence time of xenon in a particular phase, improving the probability of detecting xenon in the adsorbed phase during the mixing time. Our results suggest that this me thod could yield more detailed information on the kinetic behavior of xenon embedded within th e bilayer and the water Xe(g) Xe(g)Xe(PORE) Xe(PORE) Xe(PORE) Xe(g)Xe(g) Xe(PORE) Xe(g) Xe(g)Xe(PORE) Xe(PORE) A) B) PPM(F2) 180 160 140 120 100 80 60 40 20 0 PPM(F2) 180 160 140 120 100 80 60 40 20 0 PPM(F1) -20 0 20 40 60 80 100 120 140 160 180 PPM(F1) -20 0 20 40 60 80 100 120 140 160 180

PAGE 177

177 pore. Extracted T1 values suggest the presence of two distinct environments within the AAO porous tubule. Whats more, the use of supported b ilayers in anesthetic re search would be very useful. Incorporating proteins into membrane e nvironments can provide substantial information on the site of anesthetic actionmore so than studies involving disso lved proteins in bulk solution.

PAGE 178

178 CHAPTER 9 CONCLUSIONS AND OUTLOOK The nature of xenon-phospholipid interactions an d Xe exchange depend on the structure of the lipid headgroups and acyl chains, the phase state of the lipid bilayer, and the heterogeneity in both vesicle size and overall dist ribution of lipids with external variables. Our study of the dissolution of Xe into aqueous solutions of MLVs and LUVs provide the basis for Xe exchange dynamics within DOPC lipid membranes. While slow exchange was observed in emulsions of MLVs, rapid chemical exchange was observed in the homogeneously dispersed vesicles of LUVs. However, not all lipid systems exhibit the same trends. For instance, both DMPC and DPPC MLVs have been shown to st art as a single peak at room temperature then split into two phases with increasing temperature. Separate Xe-DMPC studies showed opposite behavior. The primary difference between these lipid and those th at we have studied is that DPPC and DMPC are saturated lipids, while DOPC and DOPE are unsaturated. Can we extrapolate the contributions from the acyl chain regions and thos e from the lipid headgroup? Which has a larger contribution to the observed chemical shift? Explor ing these differences in terms of Xe exchange and vesicle fluidity is of merit. As the partition coefficients of Xe in DOPC and DOPE are not known, the expression utilized in anesthetic binding studies is most useful. Not only does it allow for an estimation of the limiting shifts of Xe in the bound environment and determination of th e association constant, it also allows for a comparison of the solubility of xenon in the lipid memb rane with respect to that in water under fast exchange conditions. Our results indicate that the limiting shift of bound xenon increases substantially with xenon concen tration as does the binding constant. The majority of studies investigati ng hydrophobic cavities of proteins in solution use a slightly more advanced model to differentiate between speci fic and non-specific interactions. Non-specific

PAGE 179

179 interactions take place at the prot ein surface, while specific intera ctions consist of those inside the hydrophobic core. This treatment may prove to be useful in the study of lipid domains (a.k.a. lipid rafts) to determine localized effects of xe non-lipid interactions. Despite all this, neither method takes into account the transverse relaxation ( T2), which characterizes the motional restriction of xenon. Conventional Carr-Purcell-Meiboom-Gill (CPMG) spin-echo methods can be used for a more accurate determination of the binding constant. The present results also show that the inter actions of Xe with lipid membranes containing DOPE are much more complex than those in pur e DOPC. Whether the obs erved trends are due to the presence of hydrophobic pockets or decrea se in vesicle mobility still needs to be determined. The 129Xe chemical shift dependence on the xen on loading pressure is of significant interest. Recent molecular dynamics simulati ons studying the role of lipid membranes on anesthesia may be helpful in the interpretation of our data. Not only did the added presence of xenon lead to an increase in membrane thickness, surface area and acyl chain order, but in the occupation within the inter-leafle t space as well. This is contradictory to experimental studies which suggest that Xe is located at the lipid headgroup region Our studies indicate that Xe changes the thermal behavior of the (1:1) DOPC:DOPE 50 mM LUV mixture, inadvertently altering the lateral organization. The depression in the transition temperature is consistent with changes in acyl chain packing. As its presence facilitates the HII transition, one might assume that Xe assists by filling the in terstitial voids. A second peak n ear phase transition temperature suggests some change in the stru ctural properties between 40-45 C. This anomalous peak may due to a pre-transition state su ch as an elongated stalk. A particular limitation to the studies perf ormed was the long acquisition times. Although the increased solubility of Xe in lipid systems is substantially greater than in water, we were still

PAGE 180

180 forced to find creative ways to reduce the experiment time. T1 estimates of Xe dissolved in lipid vesicle suspensions are on the orde r of 60 14s. We also attempte d to dissolve optically-pumped 129Xe in solution by bubbling the gas mixture through the lipid em ulsions. While we were able to obtain spectra corresponding to Xe under fast exchange conditions with a single scan, modifications are still needed to allow for signal averaging. Spectral intensity was shown to decrease with temperature and broaden with increasing lipid concentration. Lastly, Anodic Aluminum Oxide (AAO) substrates are utilized to stabilize and align bi layers in the magnetic field, facilitating the st udy of Xe diffusivity between pores via two-dimensional (2-D) exchange spectroscopy (EXSY). 129Xe NMR studies of non-specific Xe-lipid interactions in the DOPC/DOPE lipid system will help to develop 129Xe NMR as a biomolecular probe of packing effects, chemical exchange, factors affecting lateral pressure, and phase tran sitions in membrane model systems. While our investigations provide a good starting point, additional st udies are needed to fully characterize observed effects. Here is a list of experiments that would further elucidate our studies to date as well as some additional systems of interest: Make use of SPINOE: enhance the polarization of the 13C within the acyl-chains to study the dynamics within the hydrophobic interior of the bilaye r with increasing stress. Membrane defects such as lipid rafts are localized domains of lipids with expressed curvature. As changes in membrane permeability are suggested to change at these boundaries, it would be a good candi date for Xe NMR studies. Low concentrations of choleste rol are thought to pref erentially partition into lipid domains and enhance lipid membrane permeability. High c oncentrations of cholesterol increase the lateral packing effectively reduc ing the free volume of the memb rane. This would be useful in the study of lipid domains. Our preliminary investigations into the us e of inorganic substr ates to support model membranes show potential in determining the Xe diffusion coefficients. Not only could this be used to validate the presence of intermediate structures and lipid domains, it can help elucidate binding kinetics at the lipid-water interface with changing lipid composition. This can be used in conjunction with magnetizati on transfer experiments, measurements of

PAGE 181

181 relaxation times ( T1, T1 T2)all of which exploit the dynamics of a system at different time scales. Lastly, the antimicrobial peptide MSI-78 has been shown to induce signif icant changes in the lipid bilayer. In addition to inhibiting the lamellar-to-inverted hexagonal phase formation it has been shown to alter the bilayer morphology such as to resemble the formation of a toroidal pore at low concentrations. We woul d like to further characterize these phenomena with Xe NMR methods.

PAGE 182

182 APPENDIX NEWLY DESIGNED SAMPLE HOLDER FOR AAO STUDIES Schematics for a sample holder custom made fo r use in wideline probe. Particular focus was made on creating a larger sample space for the use to stacked AAO supports, and allowing the use of an insert (comprised of two halves of a cylinder) to reduce the filling factor, and providing more accurate control of the alignment. As seen below, the apparatus consists of three parts labeled 1, 2, and 3. Part 1 connects the 1/8 outer diameter (O.D.) PFA tubing to the sample region. The second component (Part 2) receives th e tubing and creates a seal with a 1/8 ferrule where Part 1 and 2 meet. Part two essential acts as a reducer; it reduces the diameter of the sample region to accommodate the size of the PFA tubing. The seal between Parts 2 and 3 is created by the placement of two o-rings on either side of the sample body (Part 3). As mentioned previously, this particular design allo wed for experiments at temperatures 208 K. Figure A-1. Schematic diagram of the custom sa mple holder designed for aligned lipid bilayers on AAO supports. 1 2 3 1.590 0.280 l = 0.35 0.100 0.665 0.112 W= 0.180 0.475

PAGE 183

183 Figure A-2. 129Xe NMR stacked plot showing the xenon-AAO interaction with decreasing temperature. The exchange rate is redu ced as the temperature is lowered. Part #1: Top of cap Bottom of cap 0.235 0.200 O.D. = 0.350 I.D. = 0.235 O.D. = 0.235 I.D. = 0.140 Part #2 0.180 0.220 0.100 O.D. = 0.475 I.D. #1 = 0.235 I.D. #2 = 0.125 O.D. = 0.180 I.D. = 0.075 Top of capBottom of cap0.180 Part #3 0.180 0.220 Side viewO.D. = 0.475 I.D. = 0.180 Wall thickness: 0.035 0.475 0.475 0.350 O-Ring Groove O.D. = 0.296 O-Ring Groove O.D. = 0.296 O.D. = 0.475 I.D. = 0.075 0.112 0.79 0.350 0.140 0.100 0.235 0.125 0.075 298(K) 288 278 268 258 248 238 228 218 208

PAGE 184

184 LIST OF REFERENCES [1] R. F. Tilton, I. D. Kuntz Biochemistry. 1982, 21, 6850-6857. [2] S. M. Rubin, M. M. Spence, B. M. Goodson, D. E. Wemmer, A. Pines Proceedings of the National Academy of Sciences of the United States of America 2000, 97, 9472-9475. [3] T. J. Lowery, S. M. Rubin, E. J. Ruiz, M. M. Spence, N. Winssinger, P. G. Schultz, A. Pines, D. E. Wemmer Magnetic Resonance Imaging. 2003, 21, 1235-1239. [4] Z. I. Cleveland, H. E. Moller, L. W. Hedlund, B. Driehuys Journal of Physical Chemistry B 2009, 113, 12489-12499. [5] B. Driehuys, H. E. Moller, Z. I. Cleveland, J. Polla ro, L. W. Hedlund Radiology 2009, 252, 386-393. [6] B. M. Goodson Journal of Magnetic Resonance 2002, 155, 157-216. [7] S. M. Rubin, M. M. Spence, I. E. Dimitrov, E. J. Ruiz, A. Pines, D. E. Wemmer Journal of the American Chemical Society 2001, 123, 8616-8617. [8] E. Locci, Y. Dehouck, M. Casu, G. Saba, A. Lai, M. Luhmer, J. Reisse, K. Bartik Journal of Magnetic Resonance 2001, 150, 167-174. [9] J. Wolber, A. Cherubini, M. O. Leach, A. Bifone Magnetic Resonance in Medicine 2000, 43, 491-496. [10] K. W. Miller, N. V. Reo, A. J. M. S. U iterkamp, D. P. Stengle, T. R. Stengle, K. L. Williamson Proceedings of the National Academy of Sciences of the United States of America. 1981, 78, 4946-4949. [11] S. I. Han, S. Garcia, T. J. Lowery, E. J. Ruiz, J. A. Seeley, L. Chavez, D. S. King, D. E. Wemmer, A. Pines Analytical Chemistry. 2005, 77, 4008-4012. [12] S. M. Rubin, M. M. Spen ce, A. Pines, D. E. Wemmer Journal of Magnetic Resonance 2001, 152, 79-86. [13] M. M. Spence, S. M. Rubin, I. E. Dimitrov, E. J. Ruiz, D. E. Wemmer, A. Pines, S. Q. Yao, F. Tian, P. G. Schultz Proceedings of the National Acade my of Sciences of the United States of America. 2001, 98, 10654-10657. [14] Y. Xu, P. Tang Biochimica Et Biophysi ca Acta-Biomembranes 1997, 1323, 154-162. [15] L. M. Stimson, I. Vattulainen, T. Rog, M. Karttunen Cellular & Molecular Biology Letters 2005, 10, 563-569. [16] P. Tang, B. Yan, Y. Xu Biophysical Journal 1997, 72, 1676-1682.

PAGE 185

185 [17] J. K. M. Sanders, B. K. Hunter, Modern NMR Spectroscopy: a guide for chemists, Oxford University Press, New York, 1987. [18] R. J. Abraham, J. Fisher, P. Loftus, In troduction ot NMR Spectroscopy, John Wiley & Sons Ltd, Toronto, Canada, 1988. [19] E. D. Becker, High resolution NMR. Theory and chemical applications, Acedemic Press, Inc, Orlando, 1980. [20] M. J. Duer, Solid State NMR Spectroscopy: Principles and Applications, Blackwell Science, Ltd, Oxford, 2002. [21] M. H. Levitt, Spin Dynami cs: Basics of Nuclear Magnetic Resonance, John Wiley & Sons, Ltd, West Sussex, 2008. [22] J. Keeler, Understanding NMR Spectroscop y, John Wiley & Sons, Ltd, West Sussex, 2005. [23] R. R. Ernst, G. Bodenhausen, A. Wokaun, Principles of Nuclear Magnetic Resonance in One and Two Dimensions Oxford University Press, New York, 1986. [24] C. P. Slichter, Princi ples of Magnetic Resonance, Spinger-Verlag, Berlin, 1989. [25] A. Abragam, The principles of nuclear magnetism, Oxford University Press, New York, 1989. [26] R. K. Harris, E. D. Becker, S. M. Cabr al, C. de Menezes, R. Goodfellow, P. Granger Pure and Applied Chemistry 2001, 73, 1795-1818. [27] C. R. Bowers, V. Storhaug, C. E. Webster, J. Bharatam, A. Cottone, R. Gianna, K. Betsey, B. J. Gaffney Journal of the American Chemical Society 1999, 121, 9370-9377. [28] N. Lisitza, I. Muradian, E. Frederic k, S. Patz, H. Hatabu, E. Y. Chekmenev J Chem Phys 2009, 131, 044508. [29] C. Gabellieri, S. Reynolds, A. Lavie, G. S. Payne, M. O. Leach, T. R. Eykyn Journal of the American Chemical Society 2008, 130, 4598-+. [30] D. Raftery, H. Long, T. Meersmann, P. J. Grandinetti, L. Reven, A. Pines Physical Review Letters. 1991, 66, 584-587. [31] D. A. Hall, D. C. Maus, G. J. Gerfen, S. J. Inati, L. R. Becerra, F. W. Dahlquist, R. G. Griffin Science. 1997, 276, 930-932. [32] C. R. Bowers, D. P. Weitekamp Journal of the American Chemical Society 1987, 109, 5541-5542. [33] C. T. Farrar, D. A. Hall, G. J. Gerfen, S. J. Inati, R. G. Griffin Journal of Chemical Physics 2001, 114, 4922-4933.

PAGE 186

186 [34] J. Natterer, J. Bargon Progress in Nuclear Magnetic Resonance Spectroscopy 1997, 31, 293-315. [35] C. R. Bowers, D. P. Weitekamp Physical Review Letters 1986, 57, 2645-2648. [36] S. B. Duckett, C. J. Sleigh Progress in Nuclear Magnetic Resonance Spectroscopy 1999, 34, 71-92. [37] I. Tifrea, M. E. Flatte Journal of Superconductivity 2005, 18, 207-213. [38] M. I. Dyakonov in Spin Physics in Semi conductors, Vol. (Ed.^Eds.: Editor), SpringerVerlag, City, 2008. [39] C. R. Bowers Solid State Nuclear Magnetic Resonance 1998, 11, 11-20. [40] T. G. Walker, W. Happer Reviews of Modern Physics 1997, 69, 629-642. [41] M. S. Rosen, T. E. Chupp, K. P. Coulter, R. C. Welsh, S. D. Swanson Review of Scientific Instruments 1999, 70, 1546-1552. [42] M. G. Mortuza, S. Anala, G. E. Pavlovskaya, T. J. Dieken, T. Meersmann Journal of Chemical Physics 2003, 118, 1581-1584. [43] V. J. Storhaug, F. Liebig, C. R. Bowers Journal of Physical Chemistry B 2002, 106, 28842891. [44] J. Klinowski, New Techniques in Solid-State NMR, Spinger-Verlag, Berlin, 2005. [45] J. G. Kempf, J. P. Loria Cell Biochemistry and Biophysics 2002, 39, 187-212. [46] A. G. Palmer, C. D. Kroenke, J. P. Loria Methods in Enzymology. 2001, 339 (Part B), 204238. [47] H. Wennerstrom Molecular Physics 1972, 24, 69-80. [48] H. M. McConnell Journal of Chemical Physics. 1958, 28, 430-431. [49] J. Jeener, B. H. Meier, P. Bachmann, R. R. Ernst Journal of Chemical Physics 1979, 71, 4546-4553. [50] J.-L. Yan, X.-A. Mao, L.-F. Shen Chemical Physics Letters 1997, 272, 278-283. [51] P. T. Manoharan, G. V. R. Chandramouli Proceedings of Indian National Science Academy 1999, 65, 613-631. [52] C.-Y. Cheng, C. R. Bowers Journal of the American Chemical Society 2007, 129, 1399714002. [53] C. L. Perrin, T. J. Dwyer Chemical Reviews. 2002, 90, 935-967.

PAGE 187

187 [54] I. L. Moudrakovski, C. I. Ratcliffe, J. A. Ripmeester Journal of the American Chemical Society 1998, 120, 3123-3132. [55] P. Tallavaara, J. Jokisaari Physical Chemistry Chemical Physics 2006, 8, 4902-4907. [56] C.-Y. Cheng, J. Pfeilsticker, C. R. Bowers Journal of the Americ an Chemical Society 2008, 130, 2390-2391. [57] J. L. Bonardet, J. Fraissard A. Gedeon, M. A. Springuel-Huet Catalysis Reviews-Science and Engineering 1999, 41, 115-225. [58] M. A. Springuelhuet, J. L. Bonardet, J. Fraissard Applied Magnetic Resonance. 1995, 8, 427-456. [59] S. B. Liu, C. S. Lee, P. F. Shiu, B. M. Fung in Zeolites and Microporous Crystals, Vol. 83 (Ed.^Eds.: Editor), Kodansha Lt d & Elsevier Science, City, 1994, pp.233-242. [60] A. L. Zook, B. B. Adhyaru, C. R. Bowers Journal of Magnetic Resonance 2002, 159, 175182. [61] M. Bogdanov, E. Mileykovskaya, W. Dowhan Subcellular Biochemistry 2008, 197-239. [62] A. T. Hammond, F. A. Heberle, T. Baumga rt, D. Holowka, B. Baird, G. W. Feigenson Proceedings of the National Academy of Sc iences of the United States of America 2005, 102, 6320-6325. [63] S. Svetina, B. Zeks Anatomical Record 2002, 268, 215-225. [64] A. van Dalen, B. de Kruijff Biochimica Et Biophysica Acta-Molecular Cell Research 2004, 1694, 97-109. [65] H. Hauser, G. Poupart in Lipid Structure, Vol. (Ed. P. L. Yeagle), CRC Press, 2004, pp.26. [66] D. E. Vance, J. E. Vance, Biochemistry of Lipids, Lipoproteins and Membranes, Elsevier, 2008. [67] Y. L. Chen, J. Israelachvili Langmuir 1991, 7, 2694-2699. [68] G. Van Meer Embo Journal 2005, 24, 3159-3165. [69] M. O. Roy, M. Pugniere, M. Jullien, J. Chopineau, J. C. Mani Journal of Molecular Recognition 2001, 14, 72-78. [70] M. J. Hope, P. R. Cullis Biochem. Biophys. Res. Comm. 1980, 92, 846-852. [71] J. M. Seddon, et al. Biochemistry 1983, 22, 1280-1289. [72] S. M. Bezrukov, R. P. Rand, I. Vodyanoy, V. A. Parsegian Faraday Discuss. 1998, 111, 173-183.

PAGE 188

188 [73] D. M. Small, The Physical Chemistry of Lipids : From Alkanes to Phospholipids Plenum Press, New York, 1986. [74] J. N. Israelachvili, D. J. Mitchell, B. W. Ninham Faraday Transactions 2. 1976, 1525-1568. [75] J. N. Israelachvili, Mitchell, D.J., Ninham, B.W. Biochimica et biophysica acta 1977, 470, 185-201. [76] C. Tanford Proceedings of the National Academy of Sciences 1979, 76. [77] S. Morein, R. E. Koeppe II, G. Lindblom, B. de Kruijf, J. A. Killian Biophysical Journal 2000, 78, 2475-2485. [78] Y. Barenholz, Cevc, G., Physical Chemistry of Biological Interfaces, CRC, 1999. [79] G. C. Shearman, O. Ces, R. H. Templer, J. M. Seddon Journal of Physics: Condensed Matter 2006, 18, S1105-S1124. [80] I. R. Cooke, M. Deserno Biophysical Journal 2006, 91, 487-495. [81] D. Marsh biophysical Journal. 1996, 70, 2248-2255. [82] J. N. Israelachvili, Intermolecular & Surface Forces, Academic Press, London, 1991. [83] R. S. Cantor in The latera l pressure profile in membranes: a physical mechanism of general anesthesia, Vol. (Ed.^Eds.: Editor), City, 1998, pp.451-458. [84] R. S. Cantor Physical Chemistry B 1997, 101, 1723-1725. [85] R. Rosso, E. G. Virga in Inhomogeneities in Biol ogical Membranes Vol. (Ed. F. Mollica, Preziosi, L., Rajagopal, K. R.), Birkhuser, Boston, 2007, pp.323-357. [86] C. E. Overton, Studies on Narcosis, Chapman and Hall, London, 1991. [87] K. Charalambous, D. Mill er, P. Curnow, P. J. Booth BMC Biochemistry 2008, 9, 31-42. [88] L. R. de Young, K. A. Dill Biochemistry. 1988, 27, 5281-5289. [89] R. S. Cantor Biochemistry. 1997, 32, 2339-2344. [90] R. S. Cantor Chemistry and Physics of Lipids 1999, 101, 45-56. [91] K. Simons, E. Ikonen Nature. 1997, 387, 569-572. [92] B. de Kruijff Nature 1997, 386, 129-130. [93] O. F. Andersen, I. Koeppe, R. E. Annu. Rev. Biophys. Biomol. Struct. 2007, 36, 107-130. [94] D. Marsh Chemistry and Physics of Lipids. 1991, 57, 109-120.

PAGE 189

189 [95] J. A. Killian, G. van Meer EMBO reports 2001, 2, 91-95. [96] A. G. Rietveld, M. C. Koorengevel, B. de Kruijff EMBO Journal 1995, 14, 5506. [97] P. R. Cullis, C. P. Tilcock in Lipid Polymorphism, Vol. (Ed. J. Wilschut, Hoekstra, D. ), CRC Press, 1990. [98] S. M. Gruner, E. Shyamsunder Annals of the New York Academy of Sciences 1991, 625, 685 697. [99] E. van den Brinkvan der Laan, J. A. Killian, B. de Kruijff Biochim Biophys Acta 2004, 1666, 275-288. [100] H. Tournois, C. H. Fabrie, K. N. Burger, J. Mandersloot, P. Hilgers, H. van Dalen, J. e Gier, B. de Kruijff Biochemistry. 1990, 29, 8297-8307. [101] J. U. Bowie Nature. 2005, 438, 581-589. [102] J. M. Sonner, J. F. Antognini, R. C. Dutt on, P. Flood, A. T. Gray, R. A. Harris, G. E. Homanics, J. Kendig, B. Orser, D. E. Ra ines, J. Trudell, B. Vissel, E. I. Eger Anesthesia and Analgesia. 2003, 97, 718-740. [103] B. W. Urban, M. Bleckwenn, M. Barann Pharmacol. Ther. 2006, 111, 729-770. [104] I. Ueda, H. Kamaya Anesthesia and Analgesia 1984, 63, 929-945. [105] R. F. Albrecht, D. J. Miletich General Pharmacology 1988, 19, 339-346. [106] T. Hata, H. Matsuki, S. Kaneshina Colloids and Surfaces B-Biointerfaces 2000, 18, 4150. [107] T. Hata, T. Sakamoto, H. Matsuki, S. Kaneshina Colloids and Surfaces B-Biointerfaces 2001, 22, 77-84. [108] M. W. Hill Biochem. Biophys. Acta 1974, 356, 117-124. [109] Y. Kaminoh, C. Tashiro, H. Kamaya, I. Ueda Biochim Biophys Acta 1989, 946, 215-220. [110] E. S. Rowe Biochemistry. 1983, 22, 3299-3305. [111] S. F., H. T., C. S. M., W. W. G. Archives of Biochemistry and Biophysics 1995, 316, 343352. [112] P. H. Rosenberg, A. Alila Naunyn-Schmiedeberg's Archives of Pharmacology 1982, 319, 95-100. [113] K. Tu, M. Tarek, M. L. Klein, D. Scharf Biophysical Journal 1998, 75, 2123-2134.

PAGE 190

190 [114] J. J. L. p. Cascales, J. G. H. n. Cifre, J. G. a. de la Torre Physical Chemistry B 1998, 102, 625-631. [115] S. E. Feller, C. A. Brow n, D. T. Nizza, K. Gawrisch Biophysical Journal 2002, 82, 13961404. [116] M. Patra, E. Salonen, E. Terama, I. Va ttulainen, R. Faller, B. W. Lee, M. Karttunen biophysical Journal. 2006, 90, 1121-1135. [117] P. H. Rosenberg Naunyn-Schmiedeberg's Ar chives of Pharmacology 1979, 307, 199-206. [118] H. Matsuki, M. Yamanaka, S. Kaneshina, H. Kamaya, I. Ueda Colloids and Surfaces BBiointerfaces. 1998, 11, 87-94. [119] E. Terama, O. H. S. Ollila, E. Salonen, A. C. Rowat, C. Tr andum, P. Westh, M. Patra, M. Karttunen, I. Vattulainen Journal of Physical Chemistry B 2008, 112, 4131-4139. [120] M. S. Albert, G. D. Cates, B. Driehuys, W. Happer, B. Saam, C. S. Springer, A. Wishnia Nature. 1994, 370, 199-201. [121] S. McKim, J. F. Hinton Biochimica Et Biophysica Acta-Biomembranes 1994, 1193, 186198. [122] A. K. Venkatesh, L. Zhao, D. Balamore, F. A. Jolesz, M. S. Albert in Hyperpolarized Xe129 MRI using gas-filled liposomes, Vol. (Ed.^Eds.: Editor), City, 2002, pp.S270-S274. [123] J. H. Chin, J. R. Trudell, E. N. Cohen Life Sciences 1976, 18, 489-497. [124] E. Brunner Concepts in Magnetic Resonance 1999, 11, 313-335. [125] A. Cherubi ni, A. Bifone Progress in Nuclear Magnetic Resonance Spectroscopy 2003, 42, 1-30. [126] A. M. Oros, N. J. Shah Physics in Medicine and Biology 2004, 49, R105-R153. [127] D. Baumer, E. Brunner, P. Blumler, P. P. Zanker, H. W. Spiess Angewandte ChemieInternational Edition 2006, 45, 7282-7284. [128] E. Brunner Magnetic Resonance in Chemistry 1999, 37, S14-S22. [129] K. Ruppert, J. F. Ma ta, J. R. Brookeman, K. D. Hagspiel, J. P. Mugler Magnetic Resonance in Medicine. 2004, 51, 676-687. [130] B. M. Goodson, Y. Q. Song, R. E. Taylor, V. D. Schepkin, K. M. Brennan, G. C. Chingas, T. F. Budinger, G. Navon, A. Pines Proceedings of the National Academy of Sciences 1997, 94, 14725. [131] M. Corda, B. Era, A. Fais, M. Casu Biochimica Et Biophysica Acta-General Subjects. 2004, 1674, 182-192.

PAGE 191

191 [132] N. Muller Journal of Physical Chemistry 1982, 86, 2109-2110. [133] A. Bagno, F. Rastrelli, G. Saielli Progress in Nuclear Magnetic Resonance Spectroscopy 2005, 47, 41-93. [134] C. J. Jameson, D. N. Sears, S. Murad Journal of Chemical Physics 2004, 121, 9581-9592. [135] X. X. Li, C. Newberry, I. Saha, P. Nikolaou, N. Whiting, B. M. Goodson Chemical Physics Letters 2006, 419, 233-239. [136] S. M. Hosseini, Univ ersity of Massachusetts, 1985. [137] J. Wolber, A. Cherubini, A. S. K. Dzik-J urasz, M. O. Leach, A. Bifone in Spin-lattice relaxation of laser-polarized xenon in huma n blood, Vol. (Ed.^Eds.: Editor), City, 1999, pp.3664-3669. [138] L. Koubi, M. Tarek, S. Bandyopa dhyay, M. L. Klein, D. Scharf Biophysical Journal 2001, 81, 3339-3345. [139] Y. Xu, P. Tang, S. Liachenko in Unifying characteristics of site s of anesthetic action revealed by combined use of anesthetics and non-anesthetics, Vol. (Ed.^Eds.: Editor), City, 1998, pp.347-352. [140] J. Sangster, Octanol-Water Partition Coeffi cients: Fundamentals and Physical Chemistry, Wiley, 1997. [141] A. A. Omran, K. Kitamura, S. Takegami, M. Kume, M. Yoshida, A. A. Y. El-Sayed, M. H. Mohamed, M. Abdel-Mottaleb Journal of Pharmaceutical and Biomedical Analysis 2002, 30, 1087-1092. [142] Y. Xu, P. Tang, L. L. Firestone, T. T. Zhang biophysical Journal. 1996, 70, 532-538. [143] T. Yoshida, K. Takahashi, I. Ueda Biochem. Biophys. Acta 1989, 985, 331-333. [144] O. Laszlo, M. Lin, C. K. Larive Analytical Chemistry 1998, 70, 1339-1345. [145] R. P. Kennan, G. L. Pollack Journal of Chemical Physics 1990, 93, 2724-2735. [146] H. L. Clever, R. Battino, B. Jaselskis, Y. P. Yampol'skii, P. Scharlin, C. L. Young Journal of Physical and Chemical Reference Data 2005, 34, 201-435. [147] B. Jagadeesh, A. Prabhakar, M. Rao, C. V. S. Murty, V. Pisipati A. C. Kunwar, C. R. Bowers Journal of Physical Chemistry B 2004, 108, 11272-11279. [148] G. Saba, M. Casu, A. Lai International Journal of Quantum Chemistry 1996, 59, 343-348. [149] J. Jokisaari Progress in Nuclear Magne tic Resonance Spectroscopy 1994, 26, 1-26. [150] E. de Paula, S. Schreier Biochimica et Biophysica Acta 1995, 1240.

PAGE 192

192 [151] A. Seelig Biochemica et Biophysica Acta 1987, 899, 196-204. [152] M. M. Sarasua, K. R. Faught, S. L. Steedman, M. D. Gordin, M. K. Washington Alcoholism: Clinical and Experimental Research 1989, 12, 698-705. [153] M. Meier, Dissertation, University of Basel, 2006. [154] S. Tristram-Nagle, H. I. Petrache, J. F. Nagle Biophysical Journal 1998, 75, 917-925. [155] H. Schmiedel, P. Jrc hel, M. Kiselev, G. Klose Physical Chemistry B 2001, 105, 111-117. [156] R. S. Cantor Biophysical Journal 1999, 76, 2625-2639. [157] R. S. Cantor Biophysical Journal 2001, 80, 2284-2297. [158] R. S. Cantor Journal of Physical Chemistry B 2001, 105, 7550-7553. [159] K. Kitamura, M. Yamamoto, S. Takegami, M. Sugiura Talanta. 1999, 49, 261-265. [160] J. R. Zimmerman, W. E. Brittin Journal of Physical Chemistry 1957, 61, 1328-1333. [161] M. Rosoff in Vesicles, Vol. (Ed.^E ds.: Editor), Markel Dekker, Inc, City, 1996. [162] L. I. Barsukov, A. V. Victorov, I. A. Va silenko, R. P. Evstigneeva, L. D. Bergelson Biochimica Et Biophysica Acta 1980, 598, 153-168. [163] M. Frohlich, V. Brecht, R. Peschka-Suss Chemistry and Physics of Lipids 2001, 109, 103112. [164] R. Battino, H. L. Clever Chemical Reviews 1966, 66, 395-&. [165] H. Kamaya, S. Kaneshina, I. Ueda Biochem. Biophys. Acta 1981, 646, 135-142. [166] M. H. Abraham, W. R. Lieb, N. P. Franks Journal of Pharmaceutical Sciences 1991, 80, 719-724. [167] A. J. Leo Methods in Enzymology 1991, 202, 544-591. [168] A. Leo, C. Hansch, D. Elkins Chemical Reviews. 1971, 71, 525-616. [169] R. A. Smith, E. G. Porter, K. W. Miller Biochemica et Biophysica Acta 1981, 645, 327338. [170] K. Y. Y. Pang, L. M. Braswell, L. Chang, T. J. Sommer, K. W. Miller Molecular Pharmacology 1980, 18, 84-90. [171] G. L. Pollack Science 1991, 251, 1323-1330. [172] P. Seeman Pharmacological Reviews 1972, 24, 583-655.

PAGE 193

193 [173] N. P. Franks, W. R. Lieb Nature. 1978, 274, 339-342. [174] A. Avdeef Current Topics in Medicinal Chemistry 2001, 1, 227-351. [175] J. L. Razak, B. J. Cutak, C. K. Larive, C. E. Lunte Pharmaceutical Research 2001, 18, 104-111. [176] B. I. Escher, R. P. Schwarzenbach Environmental Science & Technology 1996, 30, 260270. [177] J. Dilger, S. McLaughlin, T. McIntosh, S. Simon Science 1979, 206, 1196-1198. [178] I. Ueda Anesthesiology 1999, 90, 336. [179] U. Kaatze, A. Dittrich, K. D. Gopel, R. Pottel Chemistry and Physics of Lipids 1984, 35, 279-290. [180] T. Hianik, V. I. Passechnik in The electrical field in membranes and their capacitance, Vol. Springer-Verlag, New York, 1995, pp.150-152. [181] I. Ueda, T. Yoshida Chemistry and Physics of Lipids 1999, 101, 65-79. [182] R. F. Flewelli ng, W. L. Hubbell Biophysical Journal 1986, 49, 541-552. [183] B. Kloesgen, C. Reichle, S. Kohlsmann, K. D. Kramer Biophysical Journal 1996, 71, 3251-3260. [184] M. Simeonova, J. Gimsa Bioelectromagnetics 2006, 27, 652-666. [185] O. G. Martinsen, S. Grimnes, Bioimpedan ce and Bioelectricity Basics, Elsevier Science & Technology Books, Oxford, UK, 2008. [186] D. R. Lide, CRC Handbook of Chem istry and Physics, CRC Press LLC, 2003. [187] F. Zhou, K. Schulten Journal of Physical Chemistry 1995, 99, 2194-2207. [188] H. Brockman Chemistry and Physics of Lipids 1994, 73, 57-79. [189] M. Simeonova, J. Gimsa Bioelectromagnetics 2006, 27, 652-666. [190] Y. Xu, P. Tang, S. Liachenko Toxicology Letters 1998, 100-101, 347-352. [191] J. H. Chin, J. R. Trudell, E. N. Cohen Life Sciences 1976, 18, 489-497. [192] M. Mondal, K. Mukhopadhyay, S. Basak, A. Chakrabarti Biochimica Et Biophysica ActaBiomembranes 2001, 1511, 146-155. [193] K. Matsuoka, K. Nagayama, T. Hata H. Satake, H. Matsuki, S. Kaneshina International Congress Series 2005, 1283, 328-329.

PAGE 194

194 [194] K. H. Meyer Transactions of the Faraday Society 1937, 33, 1062-1068. [195] J. M. A. Grime, M. A. Edwards, N. C. Rudd, P. R. Unwin Proceedings of the National Academy of Sciences 2008, 105, 14277-14282. [196] D. D. Koblin, B. S. Chortkoff, M. J. Laster, E. I. Eger II, M. J. Halsey, D. Phil, P. Ionescu Anesthesia and Analgesia 1994, 79, 1043-1048. [197] P. H. Tonner, Scholz, J., Koch, C., am Esch, J.S. Anesthesia and Analgesia. 1997, 84, 618622. [198] N. P. Franks, W. R. Lieb Journal of Molecular Biology 1979, 133, 469-500. [199] S.-T. Lin, C.-M. Hsieh, M.-T. Lee Journal of the Chinese Institute of Chemical Engineers 38, 467-476. [200] J. H. Griffith, H. A. Scheraga Journal of Molecular Structure: THEOCHEM 2004, 682, 97-113. [201] I. Benjamin Chemical Physics Letters. 2009, 469, 229-241. [202] R. P. Bonifcio, M. F. Costa Gomes, E. J. M. Filipe Fluid Phase Equilibria 2002, 193, 4151. [203] G. T. Hefter, R. P. T. Tomkins in The E xperimental Determination of Solubilities, Vol. (Ed.^Eds.: Editor), Wiley, John & Sons, Incorporated, City, 2003. [204] B. Y. Zaslavsky Physiochemical propert ies of phases in aqueous polymer systems, CRC Press, 1994. [205] D. Ben-Amotz, Underwood, R. Accounts of Chemical Research 2008, 41, 957-967. [206] B. Y. Zaslavsky Aqueous two-phase pa rtitioning, Marcel De kker, Inc., New York, 1994. [207] H. S. Frank, M. W. Evans Journal of Chemical Physics 1945, 13, 507-532. [208] Y. A. Mikheev, L. N. Guseva, E. Y. Davydov, Y. A. Ershov Russian Journal of Physical Chemistry 2007, 81, 2119-2136. [209] K. A. Sharp, B. Madan Physical Chemistry B 1997, 101, 4343-4348. [210] K. A. T. Silverstein, A. D. J. Haymet, K. A. Dill Journal of Chemical Physics 1999, 111, 8000-8009. [211] T. E. Creighton, Protei ns: Structures and Molecular Properties, W. H. Freeman, 1992. [212] K. P. Murphy, P. L. Privalov, S. J. Gill Science 1990, 247, 559-561. [213] W. Blokzijl, J. Engberts Angewandte Chemie-Inte rnational Edition. 1993, 32, 1545-1579.

PAGE 195

195 [214] A. Ben-Naim, Solvation thermodynamics, Plenum Press, New York, 1987. [215] A. Ben-Naim, Molecular theory of solu tions, Oxford University Press, New York, 2006. [216] J. H. Park, J. E. Brady, P. W. Carr Bulletin of the Korean Chemical Society 1989, 10, 309314. [217] G. Graziano Biophysical Chemistry 2003, 105, 371-382. [218] Y. Katz, J. M. Diamond Journal of Membrane Biology. 1974, 17, 101-120. [219] F. Gobas, J. M. Lahittete, G. Garofalo, Y. S. Wan, D. Mackay Journal of Pharmaceutical Sciences 1988, 77, 265-272. [220] J. A. Dix, J. M. Diamond, D. Kivelson Proceedings of Indian National Science Academy 1974, 71, 474-478. [221] N. T. Southall, K. A. Dill, A. D. J. Haymet The Journal of Physical Chemistry B 2001, 106, 521-533. [222] T.-x. Xiang, B. D. Anderson The Journal of Chemical Physics 1999, 110, 1807-1818. [223] B. Lee Energetics of Biological Macromolecules 1995, 259, 555-576. [224] C. Tanford Science. 1978, 200, 1012-1018. [225] J. X. Cheng, S. Pautot, D. A. Weitz, X. S. Xie Proceedings of the National Academy of Sciences 2003, 100, 9826-9830. [226] A. S. Ulrich, A. Watts Biophysical Journal 1994, 66, 1441-1449. [227] G. L. Jendrasiak, R. L. Smith, W. Shaw Biochem. Biophys. Acta 1996, 1279, 63-69. [228] A. von Antropoff Proceedings of the Royal Society of London. Series A, Containing Papers of a Mathematical and Physical Character 1910, 83, 474-482. [229] A. Holtzer Biopolymers 1991, 32, 711-715. [230] E. S. Rowe, F. L. Zhang, T. W. Leung, J. S. Parr, P. T. Guy Biochemistry. 1998, 37, 24302440. [231] M. M. Sarasua, K. R. Faught, S. L. Steedman, M. D. Gordin, M. K. Washington Alcoholism-Clinical and Experimental Research 1989, 13, 698-705. [232] E. C. W. Clark, D. N. Glew Transactions of the Faraday Society 1966, 62, 539. [233] R. Dickinson, W. R. Lieb, N. P. Franks British Journal of Pharmacology 1995, 116, 29492956.

PAGE 196

196 [234] Y. Katz Biochem. Biophys. Acta 1981, 647, 119-129. [235] Y. Katz Journal of Physical Chemistry 1986, 90, 190-195. [236] B. B. Benson, D. Krause, Jr. Journal of Chemical Physics 1976, 64, 689-709. [237] B. N. Woodrow, J. G. Dorsey Environmental science & technology 1997, 31, 2812-2820. [238] A. P. van Wezel, A. Opperhuizen Chemosphere. 1995, 31, 3605-3615. [239] C. M. vila, F. Martinez Chemical & Pharmaceutical Bulletin 2003, 51, 237-240. [240] J. Seelig, P. Ganz Biochemistry 1993, 30, 9354-9359. [241] M. R. Wenk, A. Fahr, R. Reszka, J. Seelig Journal of Pharmaceutical Sciences 1996, 85, 228-231. [242] S. Takegami, K. Kitamura, T. Kitade, K. Kawamura Chemical & Pharmaceutical Bulletin 2003, 51, 1056-1059. [243] S. A. Simon, T. J. McIntosh, P. B. Bennett, B. B. Shrivastav Molecular Pharmacology. 1979, 16, 163-170. [244] H. Kamaya, S. Kaneshina, I. Ueda Biochem. Biophys. Acta 1981, 646, 135-142. [245] D. Lichtenberg Biochimica et Biophysica Acta (BBA) Biomembranes 1985, 821, 470478. [246] P. Westh, C. Trandum Biophysical Journal 2000, 78, 1620PLAT. [247] E. Kuchinka, J. Seelig Biol. Chem. Hoppe-Seyler. 1989, 370, 625-625. [248] A. Pohorille, M. A. Wils on, M. H. New, C. Chipot Toxicology Letters 1998, 101, 421-430. [249] H. Ohshima, H. Kamaya, T. Yoshida, I. Ueda Colloid and Polymer Science 1986, 264, 817-822. [250] T. Yoshida, K. Takahashi, H. Kamaya, I. Ueda Journal of Colloid and Interface Science 1988, 124, 176-185. [251] A. Enders Biochimica Et Biophysica Acta. 1990, 1029, 43-50. [252] A. S. Evers Nature 1989, 341, 766-766. [253] A. S. Evers, B. A. Berkowitz, D. A. Davignon Nature. 1987, 328, 157-160. [254] L. F. Fraceto, A. Spisni S. Schreier, E. de Paula Biophysical Chemistry 2005, 115, 11-18.

PAGE 197

197 [255] A. Zouni, R. J. Cl arke, A. Visser, N. V. Visser, J. F. Holzwarth Biochimica Et Biophysica Acta 1993, 1153, 203-212. [256] C. Chipot, M. A. Wilson, A. Pohorille Journal of Physical Chemistry B 1997, 101, 782791. [257] A. W. Adamson, A. P. Gast, Physic al Chemistry of Surfaces, Wiley, New York, 1997. [258] S. Brunauer, P. H. Emmett, E. Teller Journal of the American Chemical Society 1938, 60, 309-319. [259] R. L. Darcy, I. C. Watt Transactions of the Faraday Society 1970, 66, 1236-&. [260] R. G. Eckenhoff Molecular Pharmacology. 1998, 54, 610-615. [261] J. S. Chiou, I. Ueda in Ethanol unfolds firefly luciferase while competitive inhibitors antagonize unfolding -DSC and FTIR analyses, Vol. (Ed.^Eds.: Editor), PergamonElsevier Science Ltd, City, 1994, pp.969-975. [262] Z. H. Qin, G. Szabo, D. S. Cafiso Biochemistry. 1995, 34, 5536-5543. [263] J. S. Johansson, H. Zou Anesthesiology 2001, 95, 558-561. [264] L. Koubi, L. Saiz, M. Ta rek, D. Scharf, M. L. Klein Journal of Physical Chemistry B 2003, 107, 14500-14508. [265] L. Koubi, M. Tarek, S. Bandyopa dhyay, M. L. Klein, D. Scharf Anesthesiology 2002, 97, 848-855. [266] J. Blahovec, S. Yanniotis Food Bioprocess Technol. 2008, 1, 82-90. [267] I. M. Klotz, D. L. Hunston J. Biol. Chem. 1975, 250, 3001-3009. [268] A. V. Hill Biochem. J. 1913, 7, 471-480. [269] T. Huber J. Am. Chem. Soc. 2001, 124, 298-309. [270] Y. Takaoka, M. Pasenkie wicz-Gierula, H. Miyagawa, K. Kitamura, Y. Tamura, A. Kusumi. 2000, 79, 3118-3138. [271] S. Tristram-Nagle, H. I. Petrache, J. F. Nagle. 1998, 75, 917-925. [272] R. C. Word, P. Smejtek Journal of Membrane Biology. 2005, 203, 127-142. [273] O. Gezici, H. Kara, A. Ayar, M. Topkafa Sep. Purif. Technol. 2007, 55, 132-139. [274] C. J. Bowmer, W. E. Lindup Journal of Pharmaceutical Sciences. 1978, 67, 1193-1195.

PAGE 198

198 [275] S. M. Hartig, R. R. Greene, G. Carlesso J. N. Higginbotham, W. N. Khan, A. Prokop, J. M. Davidson Biomaterials 2007, 28, 3843-3855. [276] L. S. Clegg, W. E. Lindup J. Pharm. Pharmacol. 1984, 36, 776-778. [277] C. D. Thaler, R. A. Cardullo J. Biol. Chem. 1996, 271, 23289-23297. [278] L. S. Rosenberg, M. J. Carvlin, T. R. Krugh Biochemistry 2002, 25, 1002-1008. [279] S. LeGrue, A. Friedman, B. Kahan J Immunol 1983, 131, 712-718. [280] M. McKinney, R. Raddatz Current Protocols in Pharmacology 2006, 33, 1.3.1-1.3.33. [281] C. Gauvrit, S. B. Wilson J. Exp. Bot. 1983, 34, 367-380. [282] I. M. Klotz Accounts of Chemical Research. 1974, 7, 162-168. [283] S. A. Bobrovnik J. Biochem. Biophys. Methods 2002, 52, 135-143. [284] G. P. Kreishman, C. Grah am-Brittain, R. J. Hitzemann Biochemical and Biophysical Research Communications. 1985, 130, 301-305. [285] M. Pasenkiewicz-Gierula, T. Rog, J. Grochowski, P. Serda, R. Czarnecki, T. Librowski, S. Lochynski Biophysical Journal. 2003, 85, 1248-1258. [286] R. M. Epand Biochim. Biophys. Acta-Rev. Biomembr. 1998, 1376, 353-368. [287] O. S. Andersen, R. E. Koeppe Annu. Rev. Biophys. Biomolec. Struct. 2007, 36, 107-130. [288] B. de Kruijff Biochimica Et Biophysica Acta-Biomembranes 2004, 1666, 1-1. [289] G. Basanez, F. M. Goni, A. Alonso Biochemistry. 1998, 37, 3901-3908. [290] T. J. McIntosh, K. G. Kulkarni, S. A. Simon Biophysical Journal 1999, 76, 2090-2098. [291] E. van den Brink-van der Laan, J. A. Killian, B. de Kruijff Biochim Biophys Acta 2004, 1666, 275-288. [292] S. J. Slater, M. B. Kelly, F. J. Taddeo, C. Ho, E. Rubin, C. D. Stubbs Journal of Biological Chemistry 1994, 269. [293] S. M. Gruner, E. Shyamsunder Annals of the New York Academy of Sciences 1991, 625, 685-697. [294] P. A. Baldwin, W. L. Hubbell Biochemistry 1985, 24, 2633-2639. [295] M. F. Brown Chemistry and Physics of Lipids 1994, 73, 159-180. [296] D. S. Cline, D. S. Cafiso Biochem. Biophys. Acta 1986, 854, 151-155.

PAGE 199

199 [297] R. Welti, M. Glaser Chemistry and Physics of Lipids 1994, 73, 121-137. [298] O. G. Mouritsen, P. K. J. Kinnunen, Role of lipid organization and dynamics for membrane functionality, Birkhaeuser Boston, Inc.; Birkhaeuser Verlag, 1996. [299] D. A. Brown, E. London Annual Review of Cell and Developmental Biology 1998, 14, 111-136. [300] D. A. Brown, E. London Journal of Membrane Biology. 1998, 164, 103-114. [301] E. S. Rowe Biochemistry. 1987, 26, 46-51. [302] J. X. Mou, J. Yang, C. Huang, Z. F. Shao Biochemistry 1994, 33, 9981-9985. [303] J. Y. A. Lehtonen, P. K. J. Kinnunen Biophysical Journal 1995, 68, 525-535. [304] J. C. M. Holthuis, G. van Meer, K. Huitema Molecular Membrane Biology 2003, 20, 231241. [305] J. Wilschut, D. Hoekstra Chemistry and Physics of Lipids 1986, 40, 145-166. [306] D. P. Siegel, R. M. Epand Biophysical Journal 1997, 73, 3089-3111. [307] R. M. Epand, R. F. Epand Biopolymers 2000, 55, 358-363. [308] S. Svetina, B. Zeks, R. E. Waugh, R. M. Raphael European Biophysics Journal with Biophysics Letters 1998, 27, 197-209. [309] R. Jahn, T. Yang, T. C. Sudhof Cell. 2003, 112, 519-533. [310] L. Yang, H. Huang Science 2002, 297, 1877-1879. [311] D. P. Siegel, V. Cherezov, D. V. Greathouse, R. E. Koe ppe, J. A. Killian, M. Caffrey Biophysical Journal 2006, 90, 200-211. [312] C. Leidy, W. F. Wolkers, K. Jo rgensen, O. G. Mouritsen, J. H. Crowe Biophysical Journal 2001, 80, 1819-1828. [313] E. E. Williams, J. A. Coope r, W. Stillwell, L. J. Jenski Molecular Membrane Biology 2000, 17, 157-164. [314] O. G. Mouritsen, K. Jorgensen Curr. Opin. Struct. Biol. 1997, 7, 518-527. [315] R. P. Rand, N. L. Fuller, S. M. Gruner, V. A. Parsegian Biochemistry 1990, 29, 76-87. [316] Z. Chen, R. P. Rand Biophysical Journal 1997, 73, 267-276. [317] M. Muller, K. Katsov, M. Schick Physics Reports-Review Section of Physics Letters 2006, 434, 113-176.

PAGE 200

200 [318] S. Leikin, M. M. Kozlov, N. L. Fuller, R. P. Rand Biophysical Journal 1996, 71, 26232632. [319] R. P. Rand, V. A. Parsegian Biochimica Et Biophysica Acta 1989, 988, 351-376. [320] Z. Chen, R. P. Rand Biophysical Journal 1998, 74, 944-952. [321] S. M. Gruner Proceedings of the National Academy of Sciences of the United States of America. 1985, 82, 3665-3669. [322] D. P. Siegel Biophysical Journal 1993, 65, 2124-2140. [323] D. P. Siegel, J. Ba nschbach, P. L. Yeagle Biochemistry. 1989, 28, 5010-5019. [324] R. Lewis, D. A. Mannock, R. N. McElhaney, D. C. Turner, S. M. Gruner Biochemistry. 1989, 28, 541-548. [325] K. V. Damodaran, K. M. Merz, Jr. Biophysical Journal 1994, 66, 1076-1087. [326] T. J. McIntosh Annu. Rev. Biophys. Biomol. Struct. 1994, 23, 27-51. [327] D. P. Tieleman, S. J. Marrink, H. J. C. Berendsen Biochimica et Biophysica Acta (BBA) Reviews on Biomembranes 1997, 1331, 235-270. [328] M. Weinrich, T. K. Rostovtseva, S. M. Bezrukov biochemistry. 2009, 48, 5501-5503. [329] J. R. Lewis, D. S. Cafiso Biophysical Journal 1999, 76, A18-A18. [330] G. S. Lorite, T. M. Nobre, M. E. D. Zaniquelli, E. de Paula, M. A. Cotta Biophysical Chemistry 2009, 139, 75-83. [331] M. Baciu, M. C. Holmes, M. S. Leaver Journal of Physical Chemistry B 2007, 111, 909917. [332] M. S. Webb, T. C. Irving, P. L. Steponkus Biochimica Et Biophysica Acta-Biomembranes 1995, 1239, 226-238. [333] D. B. Fenske, P. R. Cullis Biochimica Et Biophysica Acta 1992, 1108, 201-209. [334] S. Yokono, D. D. Shieh, I. Ueda Biochimica Et Biophysica Acta 1981, 645, 237-242. [335] N. Janes Chemistry and Physics of Lipids. 1996, 81, 133-150. [336] D. Marsh Biophysical Journal 2007, 93, 3884-3899. [337] L. Yang, L. Ding, H. W. Huang Biochemistry. 2003, 42, 6631-6635. [338] S. L. Keller, S. M. Gruner, K. Gawrisch Biochem. Biophys. Acta 1996, 1278, 241-246.

PAGE 201

201 [339] F. Liu, R. N. A. H. Lewis, R. S. H odges, R. N. McElhaney Biochemistry 2001, 40, 760768. [340] S. Morein, R. E. Koeppe II, G. Lindblom, B. de Kruijf, J. A. Killian Biophysical Journal 2000, 78, 2475-2485. [341] B. G. Tenchov, R. C. MacDonald, D. P. Siegel Biophysical Journal 2006, 91, 2508-2516. [342] S.-J. Marrink, A. E. Mark. 2004, 87, 3894-3900. [343] G. M. Whitesides, J. P. Mathias, C. T. Seto Science 1991, 254, 1312-1319. [344] D. Phil, J. F. Stoddart Angewandte Chemie International Edition 1996, 35, 1154-1196. [345] S. Zhang Nature Biotechnology 2003, 21, 1171-1178. [346] Y. Zhou, T. Shimizu Chemistry of Materials 2008, 20, 625-633. [347] S. B. Lee, C. R. Martin Chemistry of Materials 2001, 13, 3236-3244. [348] A. M. Alaouie, A. I. Smirnov Biophysical Journal 2004, 88, L11-L13. [349] E. Y. Chekmenev, P. L. Gor'kov, T. A. Cross, A. M. Alaouie, A. I. Smirnov Biophysical Journal 2006, 91, 3076-3084. [350] G. A. Lorigan, P. C. Dave E. K. Tiburu, K. Damodaran, S. Abu-Baker, E. S. Karp, W. J. Gibbons, R. E. Minto Journal of the Amer ican Chemical Society 2004, 459, 9504-9505. [351] J. Kim, J. W. Grate, P. Wang Trends in Biotechnology 2008, 26, 639-646. [352] P. Kohli, C. R. Martin Current Pharmaceutical Biotechnology 2005, 6, 35-47. [353] E. Evans, H. Bowman, A. Leung, D. Needham, D. Tirrell Science. 1996, 273, 933-935. [354] K. Gawrisch, O. Soubias Chemistry and Physics of Lipids 2008, 153, 64-75. [355] E. S. Karp, J. P. News tadt, S. Chu, G. A. Lorigan Journal of Magnetic Resonance 2007, 187, 112-119. [356] P. Takmakov, I. Vlassiouk, S. Smirnov Analytical and Bioanalytical Chemistry 2006, 385, 954-958. [357] C. A. Keller, K. Glasmstar, V. P. Zhdanov, B. Kasemo Physical Review Letters 2000, 84, 5443. [358] H. C. Gaede, K. M. Lucke tt, I. V. Polozov, K. Gawrisch Langmuir 2004, 20, 7711-7719. [359] O. Wattraint, A. Arnold, M. Auger, C. Bourdillon, C. Sarazin Analytical Biochemistry 2005, 336, 253-261.

PAGE 202

202 [360] R. C. Furneaux, W. R. Rigby, A. P. Davidson Nature 1989, 337, 147-149. [361] P. Tallavaar a, J. Jokisaari Physical Chemistry Chemical Physics 2006, 8, 4902-4907.

PAGE 203

203 BIOGRAPHICAL SKETCH Caroline D. Pointer-Keenan was born in Anc horage, Alaska in 1979. She graduated from Robert Service High School (Anchorage, Alaska) in 1997 and received her B.S. degree in chemistry from Lincoln University, PA, in 200 0. She acquired her M.S. degree in physical chemistry from the University of Michigan Ann Arbor, MI, in 2003 and joined Dr. Russ Bowers research group at the University of Florida thereafter to pur sue her Ph.D. degree in physical chemistry. Since then she has worked as a full-time graduate student and teaching assistant. While a member of the Bowers gro up, Caroline has designed various experimental apparatus utilized in both gaseous and solid stat e NMR experiments. Her immediate goal is to continue her scientific development on the postdoc toral level; she is eager to put her experience into practice and enhance her knowledge of app lied physical chemistry. Her future goal is to perform research and teach in academia.