We accept this thesis as conformingto the required standardSTUDIES OF THE ORIENTATIONAL ORDER AND BILAYERTHICKNESS IN BIOLOGICAL AND MODEL MEMBRANES.Myrna A. MonckB. Sc. (Math/Comp Sci.) St. Francis Xavier UniversityM. Sc. (Biochemistry) University of OttawaA THESIS SUBMITTED IN PARTIAL FULFILLMENT OFTHE REQUIREMENTS FOR THE DEGREE OFDOCTOR OF PHILOSOPHYinTHE FACULTY OF GRADUATE STUDIESDEPARTMENT OF BIOCHEMISTRYTHE UNIVERSITY OF BRITISH COLUMBIAApril 1993© Myrna A. MonckIn presenting this thesis in partial fulfilment of the requirements for an advanceddegree at the University of British Columbia, I agree that the Library shall make itfreely available for reference and study. I further agree that permission for extensivecopying of this thesis for scholarly purposes may be granted by the head of mydepartment or by his or her representatives. It is understood that copying orpublication of this thesis for financial gain shall not be allowed without my writtenpermission.(Signature)Department of 8/0CHE)-TWYThe University of British ColumbiaVancouver, CanadaDate DE-6 (2/88)AbstractThe hydrocarbon order of a membrane bilayer in the hydrocarbon region is suggestedto play a fundamental role in maintaining functional integrity in the biological mem-brane system, Acholeplasma laidlawii strain B. The hydrocarbon order profile, whichcan be considered to be linearly related to hydrophobic thickness, has been measured by2 11 NMR methods for a range of Acholeplasma laidlawii membranes containing exoge-nously incorporated perdeuteriated palmitic acid and a second fatty acid of increasingunsaturation. The microorganism was grown under conditions where de novo fatty acidbiosynthesis was suppressed. At 37°C, the growth temperature, there exists a range ofhydrocarbon order compatible with good growth characteristics of the microorganismand outside of which the organism grows poorly or not at all. When grown in thepresence of cholesterol which is known to increase the orientational order in the hydro-carbon region, it appears that a small fraction of the cholesterol is solubilized by themembrane. A significant fraction of membrane cholesterol, however, is excluded fromthe lipid bilayer of the microorganism or is in a membrane domain separate from therest of the lipid. 211 NMR measurements show that this pool of cholesterol is solid -likeor crystalline but is available for membrane incorporation using several solubilizationmethods. It is suggested that the range of hydrocarbon order compatible with good A.laidlawii growth characteristics is maintained even in the presence of cholesterol.Using model membrane systems, T2 relaxation anisotropy measurements have beenmade for multilamellar vesicles and macroscopically aligned multibilayers by 2 11 NMRquadrupolar echo techniques. The dominant relaxation mechanism in the multilamellarsystem is found to be fundamentally different from that in the macroscopically alignediimultibilayers as suggested by the T2 anisotropy found for each. The multilamellarvesicles show an orientation dependence consistent with such mechanisms as collec-tive lipid motions or surface undulations. The macroscopically aligned multi-layers,however, appear to damp out several of these motional modes as suggested by theanisotropy and also an increase in the magnitudes of the relaxation times obtained. Aphenomenological theory developed on the basis of similar experimental results sug-gests that fluctuations in bilayer thickness could be the mechanism responsible for T2relaxation.iiiTable of ContentsAbstract^ iiList of Tables^ vList of Figures^ viList of Abbreviations^ viiAcknowledgement^ viii1 Introduction 11.1 Membrane Structure^ 11.2 Proteins ^ 61.3 Cholesterol 71.4 Lipid Phases ^ 101.5 Lipid Polymorphism ^ 111.6 The Acholeplasma laidlawii Membrane ^ 161.7 Membrane Hydrocarbon Order ^ 191.8 Bilayer Thickness ^ 211.9 2 11 NMR Theory 231.9.1^Energy Transitions ^ 241.9.2^Powder Spectra in the Absence of Motion ^ 271.9.3^The Effect of Lipid Motion ^ 271.10 DePakeing ^ 29iv1.11 Motivation and Thesis Outline ^ 332 Materials and Methods^ 352.1 Isolation of A. laidlawii Membranes ^ 352.2 Preparation of Samples for 2 11 NMR Spectroscopy ^ 352.3 Extraction of A. laidlawii Lipids and Preparation of the Polar LipidFraction ^ 362.4 Thin Layer Chromatography ^ 372.5 Column Chromatographic Separation of A. Laidlawii Membrane Lipids 372.6 Determination of Polar Headgroup Composition ^ 382.7 Determination of Cholesterol ^ 382.8 Differential Scanning Calorimetry. 382.9 2 H NMR Measurements ^ 392.10 T 1 Measurements ^ 392.11 Derivation of Order Profiles ^ 393 Influence of Lipid Composition on in A1B Membranes^413.1 Summary ^ 413.2 Introduction 423.3 Results ^ 433.4 Discussion 564 Evidence for Two Pools of Cholesterol in the A. laidlawii B membrane 614.1 Summary ^ 614.2 Introduction 624.3 Results ^ 644.4 Discussion 77v5 Anisotropy of T2 Relaxation in Pure and Mixed Lipid Systems^825.1 Summary ^ 825.2 Introduction 845.3 Materials and Methods ^ 865.3.1 Samples ^ 865.3.2 Sample Preparation ^ 865.3.3 Nuclear Magnetic Resonance ^ 875.3.4 Measurement of T2 ^ 885.4 Results ^ 895.5 Discussion 1016 Future Directions^ 1067 Bibliography^ 108A^ 119B Phenomenological Theory of Transverse Relaxation in Membranes 120viList of Tables1.1 Polymorphic Preferences of Membrane Lipids from Eukaryotes . ^ 153.1 A. laidlawii B Membrane Lipid Composition ^ 483.2 Hydrophobic Thickness calculated ^ 594.1 Cholesterol levels in A. laidlawii Membranes of Given Fatty Acid Com-position ^ 654.2 Average Order Parameter vs Temperature for Intact A. laidlawii B Mem-brane Preparation with Cholesterola ^ 695.1 Ratios of T2 (F2)fit coefficients for POPCd 31 and POPC/POPEd31 bi-layers and A02 values ^ 96A.1 Values of AV, B(n) and C(n) for POPCd31 and POPC/POPEd31 fits . . 119viiList of Figures1.1 Hypothetical Erythrocyte Membrane Bilayer. . ^21.2 Various Lipid classes. ^41.3 Calorimetric tracings and Lipid Phases. ^91.4 Lipid Composition of the Erythrocyte Membrane. ^ 121.5 Lipid Packing Parameters. ^ 141.6 A. laidlawii B Membrane Lipids. ^ 181.7 Acyl chain geometries^ 221.8 Deuterium nucleus energy levels and rotation of a single crystal ^ 251.9 Typical 2 H NMR Spectrum^ 281.10 Spectra of P-DPPC d2 membranes ^ 301.11 Spectra of POPC d31 membranes. 323.1 Spectra and order profiles of intact and redispersed A. laidlawii Mem-branes. ^ 453.2 Order parameter profiles of intact A. laidlawii membranes. ^ 473.3 Order Profiles of Intact A1B membranes (at limits of growth)^ 493.4 Order profiles of A. laidlawii membranes and liposomes (grown withMyristic Acid) 513.5 Order Profiles for liposomes derived from A. laidlawii membranes. 533.6 Order parameter profiles of A. laidlawii isolated lipid dispersions 554.1 2 H NMR spectra for Intact Membranes and Derived Liposomes of A.laidlawii. ^ 66viii4.2 Order Profiles for Intact Membranes and Derived Liposomes of A. laidlawii. 684.3 2 H NMR spectra for Intact Membranes of A. laidlawii at different tem-peratures^ 704.4 DSC tracings of Intact A. laidlawii membranes with and without Choles-terol. ^ 724.5 211 NMR spectra of 2,2,3,4,6-d5-Cholesterol in A. laidlawii membranes. ^ 744.6 2H NMR spectra and FID of Solid Cholesterol alone and in A. laidlawiimembranes^ 765.1 Relaxation spectrum of a-DPPC-d2 ^ 905.2 Relaxation and Pake-doublet spectra of POPCd31 ^ 925.3 Anisotropy of relaxation from oriented POPCd31 and POPC:POPEd31 ^ 945.4 Relaxation spectra of B2, B2 + Leul6 and B2 + Leu24 ^ 995.5 Relaxation spectrum of A. laidlawii membranes and powder pattern forreference^ 100ixList of AbbreviationsB.1 ALB^Acholeplasma laidlawii strain B ^ 127B.2 DGDG^diglucosyl diacylglycerol 127B.3 DMPC^1-0, 2-0 dimyristoyl phosphatidylcholine ^ 128B.4 DPPC^1-0, 2-0 dipalmitoyl phosphatidylcholine 128B.5 DSC^differential scanning calorimetry ^ 128B.6 DSPC^1-0, 2-0 distearoyl phosphatidylcholine ^ 128B.7 FID^free induction decay ^ 128B.8 FTIR^Fourier transform infrared spectroscopy ^ 128B.9 GLX^glycolipid-X ^ 128B.10 GPDGDG^glycero-phosphoryl-diglucosyl diacylglycerol ^ 128B.11 Leul6^Lys2-Gly-Leum-Lys2-Ala- Amide ^ 128B.12 Leu24^Lys2-Gly-Leu24-Lys2-Ala- Amide 128B.13 MGDG^monoglucosyl diacylglycerol ^ 128B.14 MLV^Multilamellar Vesicle ^ 128B.15 0-APG^0-amino acyl phosphatidyl glycerol ^ 128B.16 NMR^nuclear magnetic resonance ^ 128B.17 PC^phosphatidylcholine ^ 128B.18 PE^phosphatidylethanolamine ^ 128B.19 PEG^Polyethylene glycol 8000 129B.20 PG^phosphatidyl glycerol ^ 129B.21 POPC^1-0-palmitoyl 2-0-oleoyl phosphatidylcholine ^ 129B.22 POPE^1-0-palmitoyl 2-O-oleoyl phosphatidylethanolamine^129B.23 R1^Longitudinal relaxation rate ^ 129B.24 R2 Transverse relaxation rate 129B.25 T 1^Longitudinal relaxation time ^ 129B.26 T2 Transverse relaxation time 129B.27 TLC^thin layer chromatography ^ 129B.28 TR repetition time ^ 129xiAcknowledgementWith great pleasure I Thank:My supervisor, Pieter Cullis for giving me the opportunity to work with such a greatgroup of people and for stimulating discussions during the past four years. Yourcontribution to Canadian Industry has also been much appreciated.Myer Bloom, for all of the discussions, for access to Room 100 and for enriching myknowledge of NMR. The shady addition for lunch was always a treat.All Room 100 inhabitants: Cyndy, Irene, Frank L., Francois, Alex, Clare, Frank N.,Teresa, Michel R. and Jenifer T. for all your help and friendship...Cullis lab inhabitants: Austin, Nancy, Archie, Ajoy, Simon, Dave F., Richard , Troy,Sandy, Paul, Michel L., Nellie, Tom, Barb, Mike, Wendy, Sean, Jeff V., Jeff W.,Conrad, Shane, Kim, Steven, Lewis, Laval, Neil and Mick, for hikes, chats, help,advice (I hope I have named all of you).The McElhaney lab: Ron for many contributions to this work, Ruthven for all of theA. laidlawii preparations and Dave Mannock for helpful discussions.The Evan Evans lab: Dave K. for great cappucinos and the occasional Frapexperiment, Andrew, Tony and Wieslawa, who are all great mechanics.My family for their love and support over the years.Dedication:Finally, I Dedicate this thesis to two important people in my life:To the memory of Eva Princz, with whom I began graduate work and who was agreat friend.To Taun Chapman, my friend and partner in life, for all his love and encouragement.xiiChapter 1Introduction1.1 Membrane StructureThe lipid bilayer of cell membranes is a fundamental molecular assembly in biologydue primarily to its role in the maintenance of the functional integrity of the cell ororganelle it encloses. While prokaryotic species have a single membrane bilayer whichsurrounds the entire organism, eukaryotic cells have several membranes. In addition tothe plasma membrane surrounding the cell, internal membranes surround the organelles(units of cellular machinery) such as the mitochondrial membrane, nuclear envelope, theendoplasmic reticulum and golgi apparatus. The basic composition of the membraneis particular to the cell or organelle enclosed. The description of the lipid structurein membranes as a bilayer was originally proposed by Gorter and Grendel, 1925 andwas further detailed by Danielli and Dayson, 1935. The ability of the lipids to diffuselaterally within the plane of the bilayer was incorporated in the "Fluid Mosaic Model"of Singer and Nicolson, 1972. A depiction of a two dimensional sea of lipids in whichproteins are embedded, has been significantly enhanced to depict a many-componentsystem (Figure 1.1). Many different species of lipid, protein, sterol and carbohydratecomponents comprise the membranes of cells or organelles with various functions and1Chapter 1. Introduction^ 2Figure 1.1: A two dimensional model of a membrane bilayer containingproteins, lipids, cholesterol and glycosylated proteins as de-picted. Interactions with cytoskeletal proteins are also indi-cated (Reproduced from Bloom et al., 1991).Chapter 1. Introduction^ 3interact with the internal and external environments of the cell. Membranes withspecific functions include, for example, the myelin sheath which encases nervous tissueand which acts as an insulator to allow for the rapid passage of uninterrupted electricalsignals down the length of the nerve axon. The selective permeability of the plasmamembrane to large and small molecules, which largely maintains their intracellularand extracellular concentrations, is one of the major membrane functions. Part of thediversity in membrane lipid composition may exist to achieve this function of selectivepermeability while at the same time satisfying packing requirements around integralmembrane proteins.The glycerol-based, glycerophospholipids and sphingosine-based, sphingolipids makeup the largest structural lipid groups of eukaryotic (animal) membranes with the glyc-eroglycolipids and sulfolipids, in relatively small amounts, making up the rest. As apercentage of all living cells including the plant kingdom, the glyceroglycolipids areby far the largest membrane structural component (Gurr and Harwood, 1992). It isnoteworthy that the glyceroglycolipids contribute the major class of lipids in the my-coplasma species of bacteria. These will be discussed in greater detail in the appropriatesections. The basic structures of some types of these glycerol-based lipids are shown inFigure 1.2. Only glycerolipids have been used in the work presented in this thesis.The amphipathic character, hydrophilic headgroup and hydrophobic hydrocarbonchains, as portrayed in the Figure, promotes lipid assembly into a bilayer structure.The hydrophobic portions are sequestered away from the hydrophilic extracellular andintracellular spaces forming an energetically favorable structure (free energy is mini-mized) (see Figure 1.2D). The structures of both the chain and headgroup components,along with cholesterol and protein, contribute to membrane function. In the case ofglycerolipids, one usually finds an unsaturated fatty acid esterified to the sn2 positionof the glycerol backbone and a saturated fatty acid esterified to the snl position (CullisDhydrocarbonchainsChapter 1. Introduction^ 4cH31,CH2 C HC— NH 3COOI ±A HO H — —H B H 3 C — NI — CH3 I1H^—CH2OH^ CH2^ CHHO— C 0^ 0 0I I0—^ 0 = P —0—^0 = P —0—\ 0I0IH^CH2^ I^ II I H^ HCH2 CH2^H --/ C — C— 0\=--0^I 1°^ 0H — C — C— 0 / C — C — O \C=0 H2CH^ /^I^/\^H —^I/\ /^ 0 H C=0^0 C=0\ \^H / \/CH2^C=0 H2C^ 0=0 H2CH2^ /^\C\ / CH 2^/,CH 2H2C \^H2C H2C/CH2^ \^H2CH2C\CH2 / CH 2^\ CHCH22^\^H2C\ CH2CH 2 H2C^H2C\ /^H2C //CH2^/\ CH 2 H2C\H2C\^/CH2 H 2 C\/ CH 2H2C—... s^H2O^ H2C^H2C\/ CH 2^x•C'' \ H02--^ H2C/ CH 2^ \^CH2C Crs>'^.C., /CH2^)H\\CH2/CH2 /^H2C\^ C..>>' H2C\^H2CH2C^ H2C / CH 2^ /^/ CH 2^\/ CH 2/\ H2C H2C\ CH2^ / CH2 H2C\^\ \^H2C\H2C^ H2C^/CH2^ / CH2/CH2^CH2\ H 2 C\^ // CH 2^/CH2 H 2C\^ H 2 C\H2C\^H2C\/ CH2^/2H2C\^H2O^CH^/CH2CH 2^\ CH 2 H2C\^ H2C\^H2CCH2^\CH2 H2CH3C^ H2C\^H C C H2/---"2^H2O //^\ CH 2zCH3 3^C\ H3 H3C H3CFigure 1.2: A. 1-0-Palmitoyl, 2-0-01eoy1-3-0-(a-D- glucopyranosyl)- sn-glycerol B. 1- 0-Palmitoyl, 2- 0- Oleoyl Phosphatidylcholine andC. 1-0-Palmitoyl, 2-0- Stearoyl Phosphatidylserine are used toillustrate phospho- and glycoglycerolipid classes which may befound in various biological membranes in D. is represented alipid membrane segment where the hydrocarbon chains are se-questered toward the center of the bilayer and the headgroupsare in contact with the aqueous phase.Chapter 1. Introduction^ 5and Hope, 1984). This is depicted in Figure 1.2. In the case of the erythrocyte mem-brane, the most well studied species, it is usual to find phosphatidylserine and phos-phatidylethanolamine containing the most unsaturated fatty acids (Vance and Vance,1984) where other lipids would be more saturated. The effects of increasing unsatura-tion in the hydrocarbon region in the absence of other factors is to maintain the bilayerin a fluid state, where more saturated membranes are less flexible or gel-like since theytend to be more tightly packed (see section 1.4 for a description of gel and fluid states).The headgroup imparts similar properties to the bilayer. The relative size of theheadgroup appears to play a major role. For instance, under similar conditions ofacyl chain composition, ionic strength, pH and temperature phosphatidyl-choline (PC)membranes are more fluid (see section 1.4) than are phosphatidyl-ethanolamine (PE)membranes (Cullis and Hope, 1984). In addition, the PC headgroups are generallylarger than are those of PE, under these similar conditions. A smaller size allows fortighter packing of the PE molecules in the membrane.A recent example of the influence of the headgroup on the physical properties of themembrane, measured by 2 H NMR order parameter methods, is found in POPC d31 andPOPEd31 model membranes. All other factors of temperature, pH and ionic strengthbeing equal, membranes of POPEd 31 are more ordered than are POPC d31 membranes.This implies a tighter packing of acyl chains in bilayers with the PE headgroup (Lafleuret al. 1990a). The concept of membrane order is discussed in section 1.7 of the intro-duction. At this point we draw a parallel between the degree of order and the degreeof acyl-chain packing in the hydrocarbon region of a membrane.In a new study using skin model membranes (cholesterol:palmitic acid:sphingomyelin1:1:1 and cholesterol:palmitic acid:ceramide 1:1:1) the influence of the headgroup is sug-gested to manifest itself in a similar way. The major difference between the outer layerof the skin, the stratum corneum or barrier, and the underlying dermal layers is theChapter 1. Introduction^ 6presence of ceramide in the barrier derived, due to the action of sphingomyelinase, fromsphingomyelin in the dermis. The effect is to reduce the size of the lipid headgroup.It has been found (Thewalt et al., 1992) that the ceramide is much more solid-likethan is sphingomyelin which exhibits liquid-crystalline like properties in 2 H NMR mea-surements and thus ceramide would be expected to contribute to a more impermeablematerial. Good reviews of the general physical properties of lipid species can be found(Vance and Vance 1984; Marsh, 1990).1.2 ProteinsMembrane proteins often make up the largest membrane component by weight ( 50%or more in some membranes). The fluid mosaic model of Singer and Nicolson, 1972described the lipid membrane as a fluid sea in which membrane proteins can carry outtheir functions. A common theme emphasized throughout this thesis is the functionalrole of membrane lipids in these interactions. Two main classes of membrane proteinsare usually found and are identified by their ease of separation from the lipid bilayer.Peripheral proteins, such as myelin basic protein from the myelin sheath, cyto chromeC from the inner mitochondrial membrane and spectrin of the red blood cell innermonolayer, are to a large extent electrostatically associated with the membrane surfaceand may be extracted using high salt solutions. The main interactions of peripheralproteins with lipids take place at the interface (in the headgroup region) perhaps me-diated by cations such as Ca2+. It is surmised that some peripheral proteins (myelinbasic protein, for example) may indeed have hydrophobic interactions with the bilayerbecause of the presence of stretches of hydrophobic amino acids which could penetrateinto its hydrophobic core.The second class of membrane proteins, the integral proteins, require detergentChapter 1. Introduction^ 7disruption of the hydrophobic interactions with the surrounding membrane lipids toachieve extraction (Vance and Vance, 1984). The tertiary structures of some integralproteins are well known. A group of these including rhodopsin have several trans-membrane helices which interact hydrophobically with the lipid matrix, while the cyto-plasmic and extracellular domains are more hydrophilic. Many transmembrane trans-porters, such as the passive sugar transport system in erythrocytes have a well definedtransmembrane domain whose function depends on lipid structure. In particular, it hasbeen noted that a characteristic length of hydrocarbon chains is necessary to achievethe correct arrangement of functional domains of the protein across the bilayer for op-timal protein function (Gurr and Harwood, 1992). Similar correlations between lipidhydrocarbon length and protein activity have been observed in reconstitution studiesof Na+-Mg 2+ ATPase extracted from A. laidlawii B (Silvius and McElhaney, 1980).Since most integral membrane proteins contain transmembrane regions these observa-tions are not surprising. However, most natural membranes contain many differentlipid types and cholesterol; thus the important physical properties of lipid moleculesinteracting with proteins to maintain functional integrity are not immediately obvious.This topic will be addressed further in Chapter 5.1.3 Cholesterol0-0H Cholesterol, the most common sterol form found in membrane lamellae, is presentin the plasma membranes of eukaryotic cells. Some prokaryotes such as mycoplasmacapricolum require sterol for growth and survival but characteristically are incapableof synthesizing it and must obtain it from a host organism. In eukaryotes, cholesterolis primarily found in the plasma membranes of cells and lacking , for the most part,in their organelles. The reason for this remains unknown. However, since cholesterolChapter 1. Introduction^ 8imparts mechanical strength to a lipid membrane (Evans and Needham, 1987), in-creased durability of the plasma membrane compared to the internal membranes maybe a factor. The interest here in membrane cholesterol is largely due to the physicalproperties it imparts to the overall structure. Some most striking physical proper-ties of cholesterol have come from differential scanning calorimetric measurements inwhich the main lipid melting phase transition, gel-to-liquid crystalline, decreases inmembranes containing up to ti 30% cholesterol and is abolished above ti 30% (see Fig-ure 1.3A). Similar studies encouraged the production of a phase diagram based on 2 11NMR measurements of DPPC/Cholesterol model membranes, (Vist and Davis, 1991).A remarkable feature of these systems is the presence of a "Liquid Ordered" phasecharacterized by a fluid-like 2 11 NMR spectrum (see section 1.4 and Figure 1.10 for adescription of "Liquid Ordered" and an example of a fluid 2 11 NMR spectrum, respec-tively). Other descriptions of the effects of cholesterol on the lipid bilayer, based onobservations from 2 11 NMR studies, suggested that cholesterol disorders the gel phaseand orders the fluid phase of a membrane. Thus the sterol functions as an "alloying"agent in the membrane, straightening out (orientationallly ordering) the acyl chains,reducing the average lipid cross-sectional area and thickening the bilayer (Bloom et al.,1991). It has been postulated that the Q hydroxyl group of cholesterol hydrogen bondswith the lipid phosphate group which positiOns the lipid in the bilayer environment(Boggs, 1987) but supporting evidence for this mechanism is yet to be found. Theeffects on biological function appear as a reduction in the passive bilayer permeability(10 fold for water through membranes containing 30% cholesterol, Koenig et al., 1992)while maintaining the fluid characteristics of the bilayer.In addition, it has been suggested that cholesterol played an important role in theevolution of membranes. This is supported by the observation in precursor sterols ofthe prokaryotes that the oxygen-requiring enzymatic step to convert the precursor toChapter 1. Introduction^ 9EndothermicABPure DPPC + 5 mol% cholesterol+ 12.5 mol% cholesterol+ 20 mol% cholesterol+ 32 molo/o cholesterol+ 50 mol% cholesterol290^320^350Temperature (K)i0—^M cellarOr-r,-(r00,f4Bilayer11 11Figure 1.3: A) Calorimetric tracings of DPPC dispersions (fully hydrated)with and without cholesterol. B) Lamellar and non-lamellarlipid structures, bilayer, micellar and inverted hexagonal.Chapter 1. Introduction^ 10cholesterol is lacking (Bloch, 1976).1.4 Lipid PhasesSome of the concepts introduced earlier are addressed in this section. As alreadysuggested, membranes can exist in a less-mobile gel-state where the main motionspresent are a type of axial diffusion of the lipid molecules, or in a fluid liquid-crystallinestate where rapid lateral diffusion of the lipid molecules occurs and the lipid acylchains can undergo fluctuations about individual carbon-carbon segments. These lattermotions are often referred to as trans-gauche isomerizations (see Figure 1.7). Theparticular state depends on temperature, and transitions between these states as afunction of temperature can be monitored by various techniques. Those of interest hereare 2H NMR and, probably the most direct method, differential scanning calorimetry(DSC). An example of the calorimetric behaviour of dipalmitoyl -phosphatidylcholine(DPPC) is given in Figure 1.3A.We identify certain useful features of the tracings, namely: the area under thecurve (transition enthalpy), the width of the transition at half height (cooperativity)and the transition midpoint (Tm ). The transition enthalpy relates to the numberof lipid molecules taking part in the transition while the cooperativity reflects thenumber of molecules undergoing a transition simultaneously. Cholesterol is known toabolish this phase transition, depending on the concentration; however, in the presenceof cholesterol, the lipid fatty acyl chains are considered to be highly orientationally"ordered" yet the lipid molecules retain fluid-like motions and thus the "liquid" stateof the membrane is maintained. The term "Liquid Ordered" was used to describe thesecharacteristics (Ipsen et al., 1990).Lipid structures (or phases) other than bilayer have been identified for certain lipidChapter 1. Introduction^ 11types and are often exhibited at physiological or higher temperatures for dispersions ofthese single lipid species. The common non-bilayer phases are the inverted hexagonal,micellar and cubic phases. Some of these non-bilayer structures are depicted in Figure1.3B and their significance is discussed in sections 1.5 and 1.6.1.5 Lipid PolymorphismThe large diversity of lipid types, their structural preferences and their anisotropicdistribution across the biological membrane bilayer are subjects of ongoing interest.The amounts of various lipid types are depicted in Figure 1.4 for the phospholipidsphosphatidylethanolamine (PE), phosphatidylcholine (PC), phosphatidylserine (PS),phosphatidylinositol (PI); Cholesterol (Chol) and Sphingomyelin(SM) found in the hu-man erythrocyte. Since differences in lipid characteristics are not necessarily observedin an intact biological membrane, it has been general practice to study lipid assem-blies of one or two components and derive conclusions based on results obtained fromthese simpler systems. An examination of the non-bilayer phase preferences of indi-vidual component lipid membranes, commonly referred to as lipid polymorphism, hasrevealed certain general characteristics of the individual lipid species.Specifically, unsaturated lipids having small, relatively poorly hydrated headgroupssuch as dioleoyl-phosphatidylethanolamine (DOPE), in isolation, adopt inverted hexag-onal structures at temperatures well below physiological (7-12°C for DOPE, Rilfors etal., 1984). This has been attributed to a cross-sectional area of the lipid headgroupthat is smaller than the cross-sectional area of the lipid acyl chains. In general, it isfound that lipids which tend to form lamellar structures have well-matched headgroupand acyl chain cross-sectional areas and those which tend to form micelles have a largerheadgroup area than acyl chain area. These observations are quantitatively predictedChapter 1. Introduction^ 12LIPID TYPE Percent^of^Total^LipidCHOL 23PE 18PC 17SM 18GL 7Others 13Figure 1.4: Phosphatidyl choline (PC), Phosphatidyl ethanolamine (PE)Phosphatidyl serine (PS), Cholesterol (Chol), Glycolipid (GL),Sphingomyelin (SM) and Others are expressed as weight% oftotal lipid in the erythrocyte membranes (data are from Vanceand Vance, 1984).Chapter 1. Introduction^ 13by a geometrical packing parameter that was first used by Israelachvili and coworkers(1980). The parameter is defined as P=v(a./) -1 where v is the hydrophobic volume, a isthe cross-sectional area occupied by the polar headgroup region at the water-membraneinterface and l is the hydrophobic length of the acyl chains. Aggregates with values of P< 1 /2 were predicted to form micellar structures. For cylindrical molecules with valuesof P ',----- 1, bilayers are formed and those molecules with P > 1 give rise to reversed aggre-gate structures where the polar headgroups are oriented towards the interior aqueousspace. A summary of the predicted shapes based on the packing parameter is givenin Figure 1.5. These are correlated with typical 2 11 NMR lineshapes. We note thatseveral molecular lipid species possess shapes which promote non -bilayer structures.An extensive treatise of the factors which influence these can be found in Rilfors et al.,(1984) and Cullis and deKruijff, (1979). Some of these non-bilayer structures are ofimmediate interest.To reiterate, lipid species which have a large hydrophobic volume compared to a.1,when dispersed in water, can form inverse hexagonal structures. These formations arehighly dependent upon external factors of temperature, pH, ion concentration and thepresence of other molecules such as cholesterol and some proteins. Under physiologicalconditions, these factors become important in determining the phase behaviour of thelipid species. Polymorphism of various synthetic and naturally occuring phospholipidsand glycolipids have been investigated using standard methods of X-ray diffraction, 31PNMR and 211 NMR and freeze-fracture EM (Cullis and deKruijff, 1979, Gruner et al.,1985; Lafleur et al., 1990a;b;c, Sternin et al., 1988; Rilfors et al., 1984).Table 1.1 exhibits polymorphic phase preferences of some lipids from eukaryotes. Itis interesting that a significant number of these form thermodynamically stable invertedhexagonal phases. None of these inverted assemblies have been observed in biologicalChapter 1. Introduction^ 142 H NMR spectra^Phospholipid Phases> 1Inverted^HexagonalIsotropic1.^Vesicles< 1 2.^Inverted micelles2^ 3.^Micelles4.^Cubic...-------'Packing Parameter(V/0.1)2 LamellarFigure 1.5: Typical 2 11 NMR spectra observed for the phases indicated andwith the indicated packing parameters. Spectra are from (top)POPEd31 multilamellar vesicles measured at 50°C (middle)POPEd31 dispersions measured at 75°C and (bottom) 100nmvesicles of DOPC initially dispersed with 20% perdeuteriatedpalmitic acid . The values of the packing parameter are takenfrom Israelachvili et al., 1980.Chapter I. Introduction^ 15Table 1.1: ukaryotes.Polymorphic Preferences of Membrane Lipids from lBilayer^ HexagonalPCSPMPSPGPIPACLPEPS(pH < 3)PA(d-Ca2+)PA(pH < 3)CL(H-Ca2+)(data from Vance and Vance, 1984)systems since the integrity of the bilayer would be compromised. However, in someinstances, the appearance of an isotropic line in '13 NMR spectra have been attributedto similar formations. It is suggested that these inverted structures play a role in eventssuch as membrane fusion, exocytosis and endocytosis, an idea currently maintained.Indeed Siegel et al., (1989) provided theoretical evidence to suggest that the half-livesof "inverted micellar intermediate" structures are short, in the us to ms range whichare too short to detect by NMR methods.Polymorphism is also exhibited by the glyceroglycolipids which is of particular rel-evance to this thesis. It has been suggested in the A. laidlawii strain A microorganismthat the presence of MGDG, an RH forming lipid, balanced in concentration by thebilayer forming, DGDG, PG, cardiolipin and others, provides a regulatory mechanismfor maintaining the bilayer formation under growth conditions which would favor 1-I IIor micellar formations (Eriksson et al., 1991; Wieslander et al., 1980). For instance, theintroduction of cholesterol or unsaturated fatty acids, or growth at an elevated tem-perature would favor a reduction in the molar ratio MGDG/DGDG. This is becauseChapter I. Introduction^ 16cholesterol is considered to be an 11 .11 phase promoter and unsaturated fatty acids in-crease the cross-sectional area of the lipid acyl chains preferentially over that of theheadgroup as does an increase in temperature. Although this is true to some extentin the evolutionarily related species A. laidlawii strain B, it is emphasized that, forthis organism, the maintenance of a fluid lipid bilayer at the growth temperature isthe primary reason for altering such ratios (see McElhaney, 1984 for a review). Thusthe environmental adaptation is facilitated by the use of H n- or micellar forming lipids.Extensive reviews in the area of lipid polymorphism are presented in Rilfors et al., 1984and Vance and Vance, 1984.1.6 The Acholeplasma laidlawii MembraneAlthough much work has gone into investigating the physical properties of the mem-brane bilayer, relatively few studies have been carried out on biological membranesmostly due to difficulties in their isolation and/or manipulation. Fortunately, the mi-croorganism Acholeplasma laidlawii is an exception. It provided the biological mem-brane system for much of the work presented in this thesis for two reasons. First, severalmembers of the mycoplasmas, a cell-wall-less group of prokaryotic microorganisms towhich A. laidlawii belongs, have the unique feature of being fatty acid auxotrophic.This means that it is possible to achieve a desired membrane lipid fatty acid compo-sition through addition of the fatty acids to the growth medium while suppressing denovo fatty acid biosynthesis. As is discussed in Chapter 3, this property was exploitedin determining the range of membrane order, measured by 2H NMR order parametermethods, in which the microorganism is seen to maintain good growth characteristics.Second, since the microorganism lacks a cell wall, unlike most prokaryotic cells, themembranes can be isolated by gentle lysis procedures and are easily obtained in largeChapter 1. Introduction^ 17quantities. This is useful in 2 1-1 NMR studies due to the low sensitivity of the deuteriumnucleus.A typical A. laidlawii membrane contains (by weight) 55-65% protein, 30-35% lipidand 10-15% carbohydrate as the major constituents, although the relative levels ofthese can be varied to some extent (McElhaney, 1984). In addition, low levels (< 5%)of carotenoids, free fatty acids and diacylglycerols are usually found. The predom-inant lipids found in the membranes of the A. laidlawii B strain are the glycolipids,monoglucosyl diacylglycerol (MGDG) and diglucosyl diacylglycerol (DGDG), and (usu-ally to a lesser extent) phosphatidylglycerol (PG). Membrane lipids usually present inlesser quantities are glycerophosphoryl diglucosyl diacylglycerol (GPDGDG) and an 0-amino acid linked PG. It is pertinent to mention another lipid species usually presentin trace quantities but which can be found in large amounts under particular conditionsof growth (Bhakoo et al., 1987) and which is referred to as G1X in this thesis. Thestuctures of these various lipid species are presented in Figure 1.6.Depending on the fatty acid composition, MGDG has been found to form predom-inantly inverted hexagonal phases at temperatures around 37°C (Wieslander et al.,1978; 1981), whereas DGDG and PG prefer bilayer structures. Recently, it has beenfound using 2H NMR methods that GPDGDG dispersions, which are optically clear, form some sort of "isotropic" (micellar) phase in isolation. The significance of thepresence of non-bilayer forming lipids in these membranes will be introduced in thenext section and discussed further in Chapter 3.Fatty acid compositions which support growth of the microorganism under condi-tions where de novo fatty acid biosynthesis is suppressed cover a fairly wide range fromthe relatively short, 14 carbon chains to the relatively long 18 carbon chains. Probablythe major restriction on fatty acids which will support growth is the maintenance ofChapter 1. Introduction^ 18BCH2OH)1-- 0CH2OH /oH \L_0^N^//oH \N^A OH 00 - CH2CH — RCH 2^ ROH 0H0-CH 2 A)CH 2OH/1-0 \/OH \N^1OH 0HCH ^ R1CH 2^ R^oE) CH 2 - CH - CH2 -O -0(bli OH ^oCH0/HoOH OH- 0I^ I-o=p-o -CH- CH-CH 2 0 --p-0 -CH- CH CH 2I^2^1^1II^2^ICH2^OH OH CH2^OHCH — R CH — RI ICH-- R^C14- R^ CH/I-0/OH\IOH 0- o C) D)NH 3 +0 - CH 2CH ^ RCH 2^ RFigure 1.6: A) Monoglucosyl diacylglycerol (MGDG), B) Diglucosyldia-cylglycerol (DGDG), C) Phosphatidyl glycerol (PG), D) Glyc-erophosphoryl diglucosyl diacylglycerol (GPDGDG), and E)0-amino acid linked phosphatidyl glycerol (OAPG) (see McEl-haney, 1984).Chapter 1. Introduction^ 19a fluid membrane at the growth temperature. Other aspects pertaining to the physi-cal properties of this biological membrane will be discussed further in Chapters 3 and4. Excellent reviews concerning biophysical studies of this membrane can be found inMcElhaney, 1984; 1989.1.7 Membrane Hydrocarbon OrderA variety of methods have been used to measure orientational order parameters inmodel and biological membranes, the values of which depend on the technique em-ployed. These include ESR, fluorescence anisotropy, 19F-NMR and 2 H-NMR methods.2 11 NMR methods have been identified as being particularly useful in the measurementof hydrocarbon order parameters in both model and biological membrane systems (seeSeelig, 1977 and Davis, 1983 for reviews).A deuterium NMR spectrum from a deuterium-labelled lipid membrane containstwo intense peaks (see Figure 1.10) separated by a value, Avg (kHz) = Scp x 125kHz,called the quadrupolar splitting. A description of how Avg arises is given in the 2 HNMR theory section. The degree of orientational order (or organization) of the envi-ronment that a deuterium-labelled carbon atom experiences may be described by anorder parameter, SCD = 112(3 - 1) where we define 0 as the angle between thenormal to the bilayer plane and the CD bond axis, the angular brackets denote thetime average of cos 20. Theoretically Sc .') will vary between -1/2 and 1. However, using2 11 NMR (lineshape) methods we have access to the magnitude of the order parameterand not the sign (see Ipsen et al., 1990); thus the Scp values reported hereafter in thisthesis all refer to IS cp I. The higher the value of Scp for a deuteriated carbon atom ona saturated fatty acyl chain, the greater is the degree of order. In the liquid crystallineChapter 1. Introduction^ 20phase of lipid bilayers, the greatest degree of orientational order corresponds to the all-trans conformation. In this most ordered fluid phase IScp I = 0.5 since the moleculesare rotating rapidly about the bilayer normal and the CD bond axis makes an angle of90° with the surface normal. Conformational excitations of the type described in thenext section result in a decrease of 'Sap I.It is possible to employ 2 11 NMR order parameter methods to measure the orderin a membrane system of lipid molecules containing a perdeuteriated fatty acyl chain,such as POPC d31 (see Lafleur et al., 1989). One generally finds an increase in acylchain flexibility toward the center of the bilayer which may be expressed in an orderparameter profile of S c./3 (n) = S(n) measured for each position, n. Characteristicallythe order profile exhibits a monotonic decrease in order, with a relatively flat region forthe first 8 to 10 carbon atoms termed the "plateau region" and a more rapid decreasefor the remaining carbon atoms. Such features indicate that little variation in theamplitudes of motion occurs near the headgroup region and that the amplitudes ofmotion increase toward the center of the bilayer. The methods for deriving an orderprofile from a superposition of spectra due to a perdeuteriated membrane lipid sampleare explicitly described in Chapter 2 and Lafleur et al., 1989. A typical profile for amembrane bilayer system is shown in Chapter 3, Figure 3.1C.It is convenient to compare order profiles using the mean order which, foran acyl chain having N carbon atoms, is simply the average [E„N=2 S(n)]/(N — 1) tobe described in Chapter 2. As mentioned earlier (section 1.1), membranes such asPOPEd31 which exhibit higher values than POPC d31 at the same temperature areconsidered to be more ordered (Lafleur et al., 1990b). We note that the value of may be empirically related to the thickness of the bilayer hydrocarbon region as willbe shown in the next section.Chapter 1. Introduction^ 211.8 Bilayer ThicknessAn operational indicator of the bilayer hydrophobic thickness is the length of the pro-jection of the acyl chains on the bilayer normal from the carbon 2 position of the snlacyl chain on one leaflet of the membrane to the average carbon 2 position of the acylchain on the other leaflet. Original work on the relationship between hydrocarbon orderand hydrophobic thickness by 2 11 NMR methods was presented by Seelig and Seelig,1974.A brief outline of the basis for this relationship is given here. Assuming, for a simplemodel bilayer of DPPC molecules, that an all-trans chain projects into the bilayer at anangle normal to the bilayer plane and that 1.25 A is the length of one (trans) carbon-carbon segment, then the largest hydrophobic length possible would be 37.5 A. In theevent of chain disordering by the presence of gauche bonds, this length would decreaseby an amount related to the number of chain segments, 1, 9 , whose projection is dueon average to gauche conformations, given that l ig = l icos 60° (60° is the angle of agauche bond with respect to the bilayer normal, see Figure 1.7). The molecular orderparameter S moi = 1/2(3 - 1), where 0' is the angle between carbon-carbonsegment 1 and the bilayer normal, is directly related to geometrical disordering due togauche bonds, see Figure 1.7. If pA and pB denote the probabilities of a segment beingin state A (all trans) or B (due to gauche bonds) with PA pB = 1, then the averageorder over the various orientations is given by S moi = 1/2(pA (3(cos 2 0) - 1) pB 3(cos 260 - 1)). The probability of state B is thus given by pB = (1-S mo/ )/1.125. This is relatedto the length, 1, of a particular chain segment, i, in the following manner: = p iA 1p iB 1, cos (60)=1(1 - 0.5p iB ) and the total length of hydrocarbon chain is givenby ^E,_2 < /i >. Thus the thickness is given by 2. Indeed, the thickness'Here we consider a carbon-carbon segment to be the line joining the midpoint of two consecutivec-c bondsChapter 1. Introduction^ 22\^\C C/^/C C\^\C C/^/C C\^\C C/CC/CC/CCC\C- CC\C/C\C/CCall trans tg+t jog t tg Kim
1/2 have nonspherical charge symmetry at the nucleus and,in the case of the deuterium nucleus, possess an electric quadrupole moment, Q. Theinteraction of Q with the electric field gradient, VE = V, at the position of the nucleusgives rise to the quadrupolar hamiltonian HQ with energy EQ . The electric field gradi-ent is properly represented as a symmetric, traceless tensor with principal componentsVex , Vyy and V. For an spa hybridized carbon nucleus the electric field is approxi-mately axially symmetric with Yz, coincident with the CD bond axis and is definedhere as, Vs, = eq. e is the charge on the electron and q is the second derivative ofAum=-1^ _m=0m=+1Ay +_ -Zeeman+QuadrupolarAEZeemanChapter I. Introduction^ 25+HozBFigure 1.8: Depicted are A. an energy level diagram for a spin 1 nucleusperturbed by the quadrupolar interaction at the nucleus andB. the results of rotation of a single crystal by angles 0 and 0.Chapter 1. Introduction^ 26the electric potential at the nucleus. Departure from axial symmetry at the nucleus ismeasured by an asymmetry parameter 77 defined as 77 = Vas — Vyy /Vzz . By definitionVzz > Vex > Vyy, , thus 0 < < 1. In the case of the principal axis coordinate systemwhere V„ is always coincident with the largest field gradient, V (/;'°) = V„, yr ) = 0and VV'2) _/6( -Gs — Vyy ) = 7-7V (i '0) (77 < 0.05 for a CD bond, Seelig, 1977).The Zeeman interaction at high fields (46.175 MHz for our purposes) dominates thequadrupolar interaction (;::-2, 200 kHz). One can treat the quadrupolar interaction as aperturbation, the result of which shifts the energy levels as depicted in Figure 1.8A.For a system in which VV' 2) = 0 and the magnetic field is applied along the z-axis thecorresponding energy is writtenEm = —gONHorri 41(2 e 1)Q 1) V (2 '°) [3m 2 — (-T + 1)]^(1.3)1 — In the case of deuterium, I=1, the three shifted energy levels at m=-1, 0, +1 are usuallyexpressed asE+1 =Eo =E_i =1—gONH. + —4 eQV(2,0 )2- 1eQV (2A9/3N-H. + 4 eQV(2,0 )(1.4)(1.5)(1.6)The selection rule Am = +1 restricts the allowed transitions tohv+ = E_ 1 — E0 = gi3NHo + 43 eQV (2 '°)^(1.7)hv_ = E0 — E+1 = g ONHo — 4 eQV (2 '°) (1.8)giving rise to two transition frequencies (shown in Figure 1.8A) whose separation Av gis known as the quadrupolar splitting3eQ^3 e2 qQAvg =^= 2h V(2'°) = 2 h (1.9)Chapter 1. Introduction^ 271.9.2 Powder Spectra in the Absence of Motione2 qQ/h is generally referred to as the static quadrupole coupling constant whose valuefor deuterium labelled lipid molecules is approximately 170 kHz. The Av g varies ac-cording to the angle which the CD bond makes with H. This can be best illustratedby looking at the rotation of a single crystal through the Euler angles a, 0 and -ycorresponding to a = 0, = B and 'y = 0 (see Figure 1.8b) and is determined byV(2 '°)^E D,20) ( 090) VIC2' )^(1.10)P=0,±1,±21—2 V„[(3cos20 — 1)] (1.11)for VV'°) = V„ and VV'1) = VV'2) = 0. lie(090) are the appropriate elements of theWigner Rotation Matrix (Rose, 1957). Substitution of 1.11 into 1.9 shows that thequadrupolar splitting varies with 0 as follows:3 e2 qQ 1Av9 (0) = 2 h 2 (3cos 2 9 — 1)^ (1.12)The measurement of a polycrystalline sample consists of measuring a powder whoselabelled sites are randomly oriented with respect to H. The probability of orientationof a site in the zone B + dB with respect to H o is given by the probability of findingin a zone on a great sphere or, P(8) = (1/2)sinO, where 0 = 90° is the most probableand 0 = 0 is the least probable orientation. The 2 11 NMR spectrum which arises froma superposition of doublets, separated by Av q (0) and weighted by P(9) is referred toas a Pake doublet (G.E. Pake, 1956) as shown in Figure 1.9.1.9.3 The Effect of Lipid MotionAlthough the static 2 11 spectrum of a membrane bears similarity to that of a deuteriumpowder sample, the lipids do not remain motionless. Properties of multilamellar dis-persions have been compared with lyotropic liquid crystals (Bloom et al., 1977) whereChapter 1. Introduction^ 28Figure 1.9: Simulated deuterium NMR spectrum using line broadening, cs= 0.01. The most intense peaks are separated by 125 kHz.Chapter 1. Introduction^ 29the chains can undergo angular excursions normal to the molecular long axis. This isreferred to as disordering and usually occurs by fluctuations such as trans-gauche iso-merizations (see Figure 1.7). A convenient measure of the degree of the fluctuations isthe so-called order parameter S ti2 defined as S it = 1/2(3< cos 2 Pi > - 1) (i=1,2,3=x,y,z).The term < cos 2A > denotes the time average of cos 20 . We note that since cosO, aredirection cosines cos213i = 1 and = 0. In the case of an axially symmetrictensor, the z axis coincident with the symmetry axis, it is evident that S33 = -2S11 =-2S 22 .The quadrupolar splitting measurement of a polycrystalline powder was given byequation 1.12 above. In a membrane, if we allow for disordering of the chains, then theelectric field gradient tensor is time averaged and the quadrupolar splitting measure-ment takes the following formv3 e2 qQ 1^12 h 2(3 < cos 2 ,3 > —1)-2(3cos 2 0 — 1)32 e2qhQ S3321 (3CO3 2 0 = 1)(1.13)(1.14)S33 is usually referred to as Sap (the bond order parameter) and is measured at 9 =90°. It is equation 1.14 which describes the measurements that will be presented inthis thesis.1.10 DePakeingThe complexity of overlapping broadline 2 H NMR spectra makes it difficult to extractpertinent data (Scp for example) and derive useful information from the individualpowder patterns. Thus the technique of dePakeing was developed (Bloom et al., 1981;Sternin et al., 1983). It can be described as a numerical procedure to deconvolute aPake-doublet spectrum into its oriented counterpart (depicted in Figure 1.10) whereI^I^1—20^—10 0 10Frequency (kHz)20Chapter 1. Introduction^ 30Figure 1.10: 2 H NMR spectra of A. multilamellar vesicles and B. orientedmultibilayers of DPPC deuterium labelled in the /3 position ofthe headgroup. (Spectra are courtesy of C. Morrison).Chapter 1. Introduction^ 31the 0° oriented spectrum results. As stated in the last section the orientation refersto the angle between the symmetry axis and H. Since the oriented, F o (x), and Pakedoublet, G(w), spectra are related to one another it is possible to derive an orientedspectrum from the corresponding Pake doublet. The dePakeing procedure is describedin detail in Sternin et al., (1983). An outline of "how it works" is presented to give thereader an idea of the overall method.The dePakeing procedure of Sternin et al., (1983) is an iterative method involvingthe calculation of F o (x) from G(w). The spectrum G(w) is stored in a digital formcontaining N points where G(+ n/N) denotes the spectrum at a given point and F o (x)is replaced by its discrete representation Fo (N). For a given spectrum, such as thatin Figure 1.10A, one half of the oriented spectrum is first calculated as though eachpoint contributes to an edge (90° orientation). This is an over- estimation of thecontribution since intensity is derived from the shoulder (0° orientation), which is oflower intensity, as well as the edge, which contributes greater intensity. The seconditeration on the other half of the spectrum under-estimates the contributions sinceequal edge and shoulder contributions are considered. Since the estimation of F o(N) isimproved with each iteration, the spectrum eventually converges to a good estimate ofFo(N). The algorithm was derived in this way to enable depakeing of both symmetric(e.g. deuterium) and asymmetric (e.g. phosphorous) spectra.A specific case of the use of dePakeing is shown in Figure 1.11. The spectrum of asample of POPC d31 multibilayers is presented along with its dePaked counterpart. Onecan see that the spectrum is much simplified by the dePakeing procedure since it ispossible to measure resonance frequencies contributed by a single CD 2 (CD3 ) group.In general, one can obtain a reasonable dePaked spectrum from a powder spectrumshowing a v cx P 2cos(/3) dependence and where axially symmetric motion is observed. Inthe case of fluid deuteriated membranes, the procedure works very well in determiningI^I^I^1—40^—20 0 20Frequency (kHz)—60 6040Chapter 1. Introduction^ 32Figure 1.11: A. Deuterium NMR spectrum of POPCd31 multilamellar vesi-cles and B. its dePaked counterpart.Chapter 1. Introduction^ 33the oriented spectrum, as given in figure 1.11 for 0 = 0. As will be shown, the dePakeingprocedure has been of great use in the derivation of order parameters in biologicalmembrane systems.1.11 Motivation and Thesis OutlineAn understanding of fundamental properties measurable by 2 1--I NMR lineshape andrelaxation methods which relate to the hydrocarbon order of the lipid membrane inbiological and model membranes are the major goals of this work. The Acholeplasmalaidlawii strain B membrane is an excellent biological membrane system for study dueto its fatty acid auxotrophic character. As indicated earlier this means that the de-sired composition of the A. laidlawii membrane can be controlled to a large extent.Questions which were immediately of interest related to the range of hydrocarbon or-der/hydrophobic thickness in which the microorganism remains viable. The first 2 1-1NMR studies of A. laidlawii membranes investigated the fluid character of the mem-brane (Oldfield et al., 1972, 1976), the similarity of order in A. laidlawii to other bio-logical membrane systems (Kang et al., 1981), similarity to model membrane systems,(Stockton et al., 1977; Rance et al., 1982) and the effects of cholesterol on membraneorder (Davis et al., 1980), a comparison of the orientational order in whole membranesand dispersions of A. laidlawii membrane lipids (Jarrell et al., 1982), or the T2 relax-ation behaviour of whole membranes (Rance, 1980), and did not focus on the viabilityof the species in particular.In addition, relatively little information is known about the motions of membraneswith thicknesses which vary spatially and/or temporally. Thus it was of interest tostudy such membranes using 211 NMR transverse (T 2 ) relaxation methods which aresensitive to slow motions and would be affected by these variations in thickness. Up toChapter 1. Introduction^ 34now few 2 I-I NMR investigations of relatively slow motions occurring in membranes havebeen carried out. This is probably due to the complexity of motional modes occurringon this timescale, which are poorly understood from a theoretical standpoint.In Chapter 2, common methods used in the work presented in Chapters 3 and 4are outlined. The methods given in Chapter 5 pertain only to Chapter 5 studies. Theinfluence of fatty acids on the range of order observed in A. laidlawii B membranes isdiscussed in Chapter 3. This chapter closely follows the paper by Monck et al., 1992titled "Influence of Lipid Composition on the Orientational Order in Acholeplasmalaidlawii Strain B membrane: A Deuterium NMR Study". Presented in Chapter 4is an investigation of the extent of cholesterol incorporation and the phase state ofthe bulk of the cholesterol seen by 211 NMR methods. Suggestions for the locationof the cholesterol in the A. laidlawii B membrane are discussed. This chapter closelyfollows the paper by Monck et al., 1993 titled "Evidence for Two Pools of Choles-terol in the Acholeplasma laidlawii Strain B Membrane: A Deuterium NMR and DSCStudy". The angular dependence of transverse relaxation times in 211 NMR membranemeasurements can be shown to be influenced by fluctuations in the thickness of themembrane bilayer in several different membrane systems. This type of motion showsvery different transverse relaxation anisotropy from that due to fluctuations in the lipidmolecular long axis. Such differences are described in Chapter 5 and suggestions con-cerning a phenomenological theory of thickness fluctuations and surface undulationsare discussed. The "Future Directions" (Chapter 6) section briefly describes furtherexperiments which could be used to answer specific questions on membrane systemssuch as those presented in the thesis.Chapter 2Materials and MethodsThe fatty acids used in these studies were obtained from Nuchek Prep Inc. (Elysian,Minnesota). Palmitic acid was perdeuteriated using the methodology of Hsiao et al.(1974), purified by column chromatography and recrystallization from ethanol.2.1 Isolation of A. laidlawii MembranesThe organism A. laidlawii B was cultured in the presence of the desired exogenouslysupplied fatty acids and cholesterol (where applicable) under conditions where en-dogenous fatty acid biosynthesis and exogenous fatty acid chain elongation have beeninhibited by the inclusion of avidin in the growth medium (see Silvius and McElhaney,1978). For all cultures in which either fatty acids only or cholesterol and fatty acidswere added to the growth medium, they were presented in the form of a mixed micelle ina small amount of ethanol. The membranes were isolated by differential centrifugationafter cell lysis by osmotic shock (see Silvius and McElhaney, 1978).2.2 Preparation of Samples for 2H NMR SpectroscopyIntact A. laidlawii membranes were prepared for 2 H NMR spectroscopy by a methodthat is essentially similar to that described by de Kruijff et al. (1976). Briefly, a freshsample was centrifuged for 20 minutes at 45,000 r.p.m. (Beckman Ti75 rotor) andthe pellet resuspended in deuterium-depleted buffered water (20mM Hepes, 100mM35Chapter 2. Materials and Methods^ 36NaC1, pH 7.5), recentrifuged as above and resuspended in the same buffer to a finalvolume of 0.7 ml. Lipid dispersions for 2 H NMR spectroscopy were prepared by thegentle vortexing of the freeze-dried lipid with either the deuterium-depleted buffer ordeuterium depleted water at temperatures above that of the gel/liquid-crystalline phasetransition temperature of the lipid sample. Under comparable conditions 2 1-1 NMRspectra of a sample dispersed in deuterium-depleted water were indistinguishable fromthose of dispersions of the same material in deuterium-depleted buffer.2.3 Extraction of A. laidlawii Lipids and Preparation of the Polar LipidFractionTotal lipid extracts were obtained by a modified method of Bligh and Dyer (1959). Theaqueous dispersion of intact A. laidlawii membranes was diluted with an equal volume ofmethanol and the mixture heated at 70°C for 45 minutes. After the mixture had cooled,it was diluted with an equal volume of water and four times its volume of chloroform.After vigorous shaking, the chloroform layer was separated and dried by filtrationthrough chloroform-wetted filter paper and concentrated by rotary evaporation. Next,the lipid concentrate was redissolved in benzene:acetone (1:1 by volume) and the finaltraces of water removed by azeotropic evaporation of the solvent. The dried lipid extractwas redissolved in chloroform and applied to a column of silicic acid (Biosil A, Biorad)in chloroform. The column was washed with at least three volumes of chloroform andthe polar lipids were subsequently eluted with methanol. The methanol fraction wassubsequently concentrated to dryness by rotary evaporation and the polar lipids werefreeze-dried from benzene, flushed with nitrogen and stored at -20°C until required.Chapter 2. Materials and Methods^ 372.4 Thin Layer ChromatographyAnalytical TLC was used to determine if there was significant lipid degradation duringthe NMR experiments and preparative TLC was used primarily in the determinationof the lipid polar headgroup composition of the A. laidlawii membrane preparations.Samples were analyzed on 0 5mm silica gel G plates which were developed with asolvent system consisting of chloroform:methanol:water (75:25:3, by volume). The lipidcomponents were visualized either by charring (analytical studies) or by staining withiodine (preparative studies).2.5 Column Chromatographic Separation of A. Laidlawii Membrane LipidsThe chromatographic separation of the MGDG, DGDG and PG present in the totalpolar lipid extract of A. laidlawii membranes was as follows. The total polar lipid frac-tion was redissolved in chloroform and applied to a column of silica gel (Davisil grade634, Aldrich) in chloroform. The column was washed with chloroform and then de-veloped sequentially with 10 column volumes of each chloroform:acetonitrile (75:25 byvolume) and chloroform:acetonitrile (1:1 by volume), 4 column volumes of acetonitrile,10 column volumes of chloroform:methanol (95:5 by volume), 4 column volumes of chlo-roform:methanol (90:10 by volume), 10 column volumes of each chloroform:methanol(85:15 by volume) and then chloroform:methanol (80:20 by volume) and finally chlo-roform:methanol (7:3 by volume). Under these conditions, the MGDG elutes in thechloroform:acetonitrile 1:1 fractions, the DGDG in the chloroform:methanol 95:5 frac-tions and the PG mainly in the chloroform:methanol 90:10 and 80:20 fractions. Thefractions containing each of the purified lipid components were separately concentratedby rotary evaporation, lyophylized from benzene and stored at -20°C until required.Chapter 2. Materials and Methods^ 382.6 Determination of Polar Headgroup CompositionAfter separation of the polar lipid components by preparative TLC, the individualcomponents were visualized and the various lipid fractions scraped quantitatively intoclean screw-capped tubes. At this stage a known amount of an appropriate phos-phatidylcholine was added as an internal standard and the entire mixture was trans-esterified with 5% H 2 SO4 in methanol. The lipid components were each quantified bygas chromatographic analysis of the methyl esters formed and from this the polar lipidcomposition of the given lipid mixture was determined.2.7 Determination of CholesterolThe cholesterol content of A. laidlawii membranes was determined by the colorimetricmethod of Watson (1960). Briefly, a known volume of membrane sample was placed ina clean glass test tube and then heated at 110° until all of the water had evaporated.Next, 200 itL of each H2 O and glacial acetic acid was added followed by 5 mL ofa chromogenic solution. The chromogenic solution contained a mixture of 1 volumeof 0.25 M 2,5-dimethylbenzenesulfonic acid in glacial acetic acid, 3 volumes of aceticanhydride, and 1 volume of glacial acetic acid. After the samples were allowed to cool(approximately 15-20 min.) 0.6 mL of concentrated H 2 SO 4 was added to each of theabove solutions which were then mixed thoroughtly, and allowed to stand for 20 minto enable completion of color development. Afterward, the absorbance was read at 620nm. (Courtesy of Dr. R.N.A.H. Lewis).2.8 Differential Scanning Calorimetry.DSC thermograms were recorded with a Microcal MC2 high-sensitivity scanning calorime-ter operating at heating rates near 11.5°C/h. (Courtesy of Dr. R.N.A.H. Lewis).Chapter 2. Materials and Methods^ 392.9 2 H NMR Measurements2 H NMR measurements were performed at 46.175 MHz on a home-built spectrometer(Sternin, 1985). The quadrupolar echo pulse sequence (Davis, 1983) was used witha 300ms recycle delay time, a 50ms interpulse spacing and a 30.5ms ring-down delay.FIDs were acquired with an 8-step phase cycle sequence. Spectral widths for all spectraacquired were either 200kHz or 500kHz. Experiments were repeated for all systemsstudied with the exception of the myristic acid-containing samples.2.10 T 1 MeasurementsA saturation recovery method was used to measure the T 1 relaxation time of the solidcholesterol. The quadrupolar echo pulse sequence (Davis, 1983) was used with a vari-able TR, a 50-ps interpulse spacing, and a 30.5-,as ring-down delay. All FID's wereacquired with an 8-step phase cycle sequence. The spectral width was 500 kHz. Theecho height was fit to an exponential function of the form S(T R) = S (o0) (1-e-TR/Ti)where S(TR ) is the signal intensity at time TR, T1 is the longitudinal relaxation time,and S( (,) is the signal intensity in the limit that TR >> 5T1.2.11 Derivation of Order ProfilesThe spectra of A. laidlawii membranes and derived liposomes recorded at 37°C aretypical of lipids bearing perdeuteriated chains in the La phase. Since the lipid systemexhibits axial symmetry, it is possible to apply a method previously discussed (Sterninet al., 1988; Lafleur et al., 1989) to derive the general shape of the order profile.Although the order parameter profile derived from using a series of specifically labelledpalmitic acid chains provides details of the order at each carbon position (see Seeligand Seelig, 1977), we chose to use the method of Lafleur et al., 1989 for two reasons.Chapter 2. Materials and Methods^ 40This latter method can be used to derive the general features of the order profile, andthe use of a perdeuteriated palmitic acid minimizes the inherent biological variabilityinevitable in using membranes derived from different cultures of the microorganismgrown using a series of specifically labelled palmitic acid chains.First, dePakeing was performed by the method of Sternin et al. (1983). ThedePaked spectrum represents the continuous probability distribution of order for thedeuteriated acyl chain.' Assuming that CD2 groups contribute equal intensity to thedePaked spectrum and that there is a monotonic decrease of order from the interfacetoward the middle of the bilayer, an average value of the quadrupolar splitting Av g wasassigned to each methylene group denoted by its acyl chain position n=2,3,...16. Theorder parameters S(n) were then calculated using equation 2.1Avg — 43 e2qhQS(n) (2.1)where e2 qQ/h. 167 kHz is the quadrupolar coupling constant (Davis, 1983). Averageorder values, mentioned in the text, were determined by calculating an arithmeticmean for the S(n), for 2 < n < 16. Corrections were made for S(16) by a linearextrapolation of S(14) and S(15) (Lafleur et al., 1989).lAs is often found in studies of this type, a small departure from the Pake doublet shape manifestsitself in a small negative-going base line just outside the main dePaked spectrum (see Chapter 3, Figure3.1). This flat negative spectral region was used to define the base line. Changes in the order profilearising from different plausible choices for the base line are of the order of the random changes resultingfrom repetition of the experiment. The errors in the derived values of all order parameters are 5-10%Chapter 3Influence of Lipid Composition on in A1B Membranes3.1 Summary2 I-I NMR techniques have recently been developed to determine the complete orienta-tional order profile of lipid bilayers employing lipids containing perdeuteriated palmiticacid (Lafieur,M., Fine,B., Sternin,E., Cullis,P.R., Bloom,M., 1989, Biophys.J. 56:1037-1041). In this work these techniques have been applied to study order profiles in intactmembranes derived from Acholeplasma laidlawii strain B. It is shown that completeorientational order profiles can be readily obtained from the intact membranes of A.laidlawii B grown on equimolar amounts of perdeuteriated palmitic acid and a non-deuterated fatty acid of varying length and unsaturation. By varying the fatty acidcomposition employing mixtures of perdeuteriated palmitic acid with myristic, elaidic,oleic or linoleic acid, a range of hydrocarbon order compatible with high rates and ex-tents of cell growth has been obtained where the average order parameter variesover the range 0.140 to 0.176. This same variation in order is seen for liposomes derivedfrom total lipids extracted from these intact membranes. 2 1-I NMR studies on liposomescomposed of individual species of the extracted lipids indicate that modulation of themembrane lipid headgroup composition has the potential to play an important role inmaintaining the membrane order within this range.41Chapter 3. Influence of Lipid Composition on in A1B Membranes^423.2 IntroductionAs indicated in Chapter 1, deuterium nuclear magnetic resonance spectroscopy ( 2 HNMR) is an important technique for characterizing order and dynamics in the hydrocar-bon region of model and biological membranes (Davis et al., 1980; Lafleur et al., 1989;Rance et al., 1982; Jarrell et al., 1982). Orientational order profiles of the hydrocarbonchains have usually been constructed using a series of membranes each containing aspecifically CD 2 labelled fatty acyl chain. Recent reports have detailed the develop-ment of convenient 2 H NMR procedures to obtain complete hydrocarbon order profilesfor lipid systems containing perdeuteriated long chain saturated hydrocarbons. Thisincludes systems containing perdeuteriated tetradecanol (Sternin et al., 1988) as wellas systems containing perdeuteriated palmitic acid (Lafleur et al.,1989,1990a). Herewe extend this type of measurement to intact biological membranes in which perdeu-teriated palmitic acid has been biosynthetically incorporated into the membrane lipids.The membranes of the mycoplasma Acholeplasma laidlawii strain B are ideally suitedfor such a study since the fatty acid composition of membrane lipids can be widely ma-nipulated by the incorporation of exogenously supplied fatty acids (see, for example,Silvius and McElhaney, 1978), and because these membranes have been extensivelystudied by many physical techniques (McElhaney 1984, 1989). The use of one prepa-ration containing perdeuteriated palmitic acid, as opposed to a series of preparationseach containing a different specifically CD 2 labelled palmitic acid, offers significant ad-vantages with respect to cost and time and minimizes the biological variability inherentin the use of different preparations. Here we show that biosynthetically incorporatedperdeuteriated palmitic acid in A. laidlawii allows the straightforward generation ofthe complete hydrocarbon order profile for a saturated chain. We have employed thisChapter 3. Influence of Lipid Composition on in A1B Membranes^43technique to investigate the range of order profiles consistent with normal growth char-acteristics of this organism. Further, we have examined the hydrocarbon order inliposomal dispersions derived from total membrane lipid extracts and from the ma-jor A. laidlawii membrane lipids. Order profiles measured for these systems indicatethat the nature of the lipid polar headgroups in the intact A. laidlawii membrane canstrongly modulate the order profile.3.3 ResultsThe first set of experiments was aimed at demonstrating that the 2 11 NMR dePakeingapproach for obtaining membrane hydrocarbon order profiles could be applied to anintact biological membrane system. Here, perdeuteriated palmitic acid was biosynthet-ically incorporated into the membrane lipids of A. laidlawii B. This was achieved byculturing the microorganism in avidin-containing media supplemented with perdeuteri-ated palmitic acid and another fatty acid. Under these conditions, endogenous fattyacid synthesis is suppressed and the growth of the microorganism is totally dependenton the exogenous supply of fatty acids (Silvius and McElhaney, 1978). It should also benoted that under these conditions the exogenous 16:0d31 is esterified almost exclusivelyto the snl position of the glycerol backbone (McElhaney and Tourtellotte, 1970; Saitoet al., 1977). The 211 NMR spectra obtained from the purified A. laidlawii membrane,as well as liposomes prepared from the total extracted lipids, is shown in Figure 3.1A.The "dePaked" spectra derived from these lineshapes are shown in Figure 3.1B. Fewof the quadrupolar splittings arising from deuterons at different positions on the chaincan be resolved; however, a smoothed orientational order profile can be achieved by as-suming a monotonic decrease in order along the acyl chain (Sternin et al., 1988; Lafleuret al., 1989). The resulting order profile derived from the dePaked spectra are shownChapter 3. Influence of Lipid Composition on in A1B Membranes^44in Figure 3.1C. Two points may be noted: first, as expected, both profiles exhibit theplateau region characteristic of lipid bilayer systems (Seelig and Seelig, 1980); second,the order profile in the intact membrane is essentially the same, within experimentalerror, as that observed for the liposomes composed of the extracted lipids.A major objective of this study was to determine the range of order profiles com-patible with growth of the microorganism A. laidlawii B. Previous studies of fattyacid-enriched and fatty acid-homogeneous cultures of A. laidlawii B have shown thatthe growth of the microorganism is inhibited when the growth temperature is morethan 50° C above the gel/liquid crystalline phase transition temperature of the mem-brane lipids, or when more than 50% of the membrane lipids are in the gel state at anygiven growth temperature (McElhaney, 1974; Silvius and McElhaney, 1978; Silvius etal., 1980). Thus the organism will grow poorly or not at all on pure palmitic or pureoleic acid, for example (vide infra). Given these limitations, there may be a range ofmembrane lipid order within which normal growth and function can occur. It is pos-sible to examine the membrane hydrocarbon order at the limits of growth by varyingthe effective unsaturation of fatty acids supplied to the A. laidlawii growth medium.We have done this in two ways: 1) by varying the number and type (cis or trans) ofdouble bonds in the fatty acids keeping the saturated/unsaturated ratios constant; and2) by varying the ratios of saturated and cis-monounsaturated fatty acids.In order to compare the variation in derived from our A. laidlawii membranepreparations with those in fluid model membranes, we have measured for DPP Cd62and for DOPC:POPCd 31 (90:10) liposomes. These values correspond to = 0.177and = 0.133 for the DPPC d62 and DOPC:POPC d31 liposomes, respectively. Thevalues of obtained for most fluid membranes lacking cholesterol normally fallwithin this range for perdeuteriated palmitic acid chains. Since DPPC d62 is only fluidat temperatures above 40°C we have measured the order parameters at 42°C for bothChapter 3. Influence of Lipid Composition on in A1B Membranes^450.30C0.24 ---o •0.18 —OO0.12 —0.06 _0.0010.^2.^4.^6.^8.^10.^12.114. 16.Carbon Number- ( n)Figure 3.1: Sample 2 H NMR spectra (A) and corresponding dePaked spec-tra (B) of A. laidlawii B. intact membranes and derived lipo-somes with a fatty acyl chain composition of 16:0d31/18:1c6.9,47:53 (mol%). (C) Order profiles derived from B as describedin Chapter 2. The ticks beneath the dePaked spectra give thefrequency assigned to a carbon position of unit area (Lafleur etal. 1989). Spectra were recorded at 37°C, using the quadrupo-lar echo pulse technique (Davis 1983). 60000 transients wererecorded for each spectrum.Chapter 3. Influence of Lipid Composition on in AIB Membranes^46of the above liposome samples. Based on differences in order profile measurements ofthe POPC d31 dispersions taken at different temperatures, we estimate that the values of given above for both the DPPCd62 and DOPC:POPCd31 (90:10) are approximately5% lower than expected at 37°C.The order profiles shown in Figure 3.2 are those of A. laidlawii B membranes pre-pared from fatty acid auxotrophic cultures grown at 37°C in the presence of perdeuteri-ated palmitic acid and an equimolar quantity of either elaidic (18:1t09), oleic (18:1t09)or linoleic (18:2c09, 012) acids. As may be expected, the order decreases for more un-saturated acyl chain substituents in the sequence 16:0d31/18:1t09 > 16:0d31/18:1c09> 16:0d31/18:2c09,012 at a given temperature. This trend was not unexpected and isessentially consistent with what one would expect from an examination of the meltingpoints of the three unsaturated fatty acids concerned as well as the results of previous19F NMR studies of MacDonald et al. (1984, 1985a,b). The average order (calcu-lated as described in Chapter 2) decreased by approximately 20% over the range where=0.173 for the 16:0d31/18:1t09 system and 0.140 for the 16:0d31/18:2c09,012preparation. It should be noted that, in particular, the linoleate content of the linoleate-containing membranes (59 mol%, see Table 3.1) is close to the maximum which can beincorporated into A. laidlawii B membranes while supporting normal or near normalgrowth of the microorganism (Silvius, 1979). Thus, the = 0.140 may well beapproaching the lower limits of hydrocarbon order which can support normal growthand membrane function in A. laidlawii B at 37°C.In the next series of experiments, the order profiles were determined for A. laidlawiiB membranes in which the ratio of 16:0d31 and 18:1cL9 in the membrane were varied.It has been demonstrated previously that neither highly saturated (e.g. palmitic acid)nor highly unsaturated (e.g. oleic acid) fatty acids alone can support the growth ofA. laidlawii B when made fatty acid auxotrophic by the inclusion of avidin in the0.400.35-0.30-0.25_(JD0.20_0.15-0.10-0.05-0.00Chapter 3. Influence of Lipid Composition on in A1B Membranes^470^2^4^6^8^10^12^14^16Carbon_ Number (ri)Figure 3.2: Order profiles derived from the dePaked spectra of intact A.laidlawii membranes. All spectra were the sum of 60000 tran-sients, recorded at 37° C. Fatty acid compositions correspond-ing to individual profiles are as follows: 16:0d31/18:1t09 46:54mol% (closed rectangles); 16:0d31/18:1c09 47:53 mol% (openrectangles); 16:0d31 /18:2c09, 012 41:59 mol% (open trian-gles).Chapter 3. Influence of Lipid Composition on in AJB Membranes^48Table 3.1: A. laidlawii B Membrane Lipid Compositionfatty acidcompositionsMGDG DGDG PG OAPG+GPDGDGGLX16:0d31/14:0, 4 7 19 9 6045:54(mol%)16:0d31/18:1t09, 34 34 26 5 tr46:54(mol%)16:0d31/18:1c09, 55 5 18 1 2178:22(mol%)16:0d31/18:1c09a, 33 25 27 15 tr47:53(mol%)16:0d31/18:1c09, 34 28 28 9 tr21:79(mol%)16:0d31/18:2c09, 012, 32 21 23 24 tr41:59(mol%)a Average from analyses of several A. laidlawii (16:0d31/ 18:1c09) preparations.growth medium. However, the organism will grow normally when the growth mediumis supplemented with mixtures of a high-melting and a low-melting fatty acid (Silviusand McElhaney, 1978). These limitations can be rationalized by the suggestion thatA. laidlawii membrane lipids derived from appropriate fatty acid mixtures will havean order () within the range that can support normal growth and membranefunction. Membrane lipids derived from either of the pure fatty acids, however, willresult in either membranes that are too disordered, for the 18:1c09 lipid, or membranescontaining too much gel-phase lipid for the microorganism to function normally at thegrowth temperature, for the 16:0 lipid (vide infra).In recent experiments it was found that mixtures of palmitic and oleic acids cansupport good cell growth provided that the mixture contains no more than 80 mol%of any single conponent (Lewis et al., unpublished results). Assuming that these areclose to the limits of saturation consistent with the normal growth and membrane0.40_0.35-0.30-0.25_CID0.20_0.15_0.10_0.05_0.000 2I^I^■^I^i4 6 8 10 12Carbon_ Number (r)14 16Chapter 3. Influence of Lipid Composition on in A1B Membranes^49Figure 3.3: Order profiles of A. laidlawii intact membranes from spec-tra recorded at 37 C. The top and bottom profiles cor-respond to membranes containing the fatty acid composi-tions: 16:0d31/18:1c09 78:22 mol% and 16:0d31/18:1cA921:79 mol% respectively. Methods for deriving the order pro-files are described in Chapter 2.Chapter 3. Influence of Lipid Composition on in A1B Membranes^50function of the microorganism, we measured the hydrocarbon order in intact A. laid-lawii membranes grown in media supplemented with 16:0d31/18:1c09 (80:20) and16:0d31/18:1c09 (20:80). The observed order profiles are illustrated in Figure 3.3.The systems containing larger amounts of oleic acid are significantly more disordered,as expected. This is expressed by the average order which decreases from 0.176for the 16:0d31/18:1c09 (80:20) membranes to 0.140 for the 16:0d31/18:1c09 (20:80)containing membranes. Since DSC studies (data not presented) have shown that theupper boundary of the gel/liquid crystalline phase transition of the palmitate (80)/oleate (20) membranes falls above the growth temperature of 37°C, it seems likelythat the order parameters observed with these membranes may be close to the upperlimits that are compatible with the normal functioning of A. laidlawii membranes (at37°C). It should be noted that the 16:0d31/18:1c09 (80:20) membranes gave rise toa spectrum characteristic of a mixture of gel (approximately 50% based on enthalpyestimates from DSC measurements, data not shown) and liquid crystalline phases at37°C. Thus it was necessary to perform spectral subtraction using a method outlinedpreviously to obtain an order profile for those lipids in the liquid crystalline state (Vistand Davis, 1990). Spectra obtained at 37°C and 42°C were used in this case. Weestimate that =0.176 is probably 5% lower than what would be obtained from apure liquid crystalline membrane preparation based on the difference between orderprofile determinations of POPC dispersions at different temperatures.In an effort to further probe these limits, we have also examined the order profileof membrane preparations isolated from cultures supplemented with the fatty acids16:0d31 and 14:0. Studies on A. laidlawii membranes have previously shown that goodgrowth occurs when A. laidlawii is cultured in an avidin-free medium on single fattyacid species with chain lengths in the interval 13 to 19 carbons (Silvius et al., 1980;McElhaney, 1984). The order profiles obtained from intact membranes and liposomesChapter 3. Influence of Lipid Composition on in A1B Membranes^510.40 0.35_0.30_11^1^1^1^14 6 8 10 12Carbon_ Number (n)0.25_------Z.--0.200.15_0.100.050 ^012114^16Figure 3.4: Order profiles of intact A. laidlawii membranes (closed rectan-gles) and derived liposomes (open rectangles) with the follow-ing fatty acid composition: 16:0d31 14:0 46:54 mol%. Spectrawere recorded at 37 C. 60000 transients were recorded for eachspectrum. Methods for deriving the order profiles are describedin Chapter 2.Chapter 3 Influence of Lipid Composition on in AJB Membranes^52composed of the total extracted lipids of A. laidlawii grown on an equimolar mixture ofperdeuteriated palmitic acid and myristic acid are shown in Figure 3.4. It may be ob-served that order in the intact membrane, =0.176, is within the range given above.DSC studies of this system (not shown) indicate that a small proportion (approximately10-15%) of the membrane lipids are in the gel phase. A gel phase component was notdetected in the 211 NMR spectrum probably due to a combination of the signal-to-noiseratio obtained and the breadth of frequencies over which such a gel-phase componentwould normally be observed. The order in the derived liposomes, =0.3, was foundto be significantly higher than that in the intact membrane (=0.195), in contrastto the system grown on 16:0d31/18:1cA9 (see Figure 3.1). In an attempt to under-stand the basis of this difference, the lipid composition of these A. laidlawii membraneswas analyzed and compared to the compositions of the other preparations employed inthis study. As shown in Table 3.1, the membrane grown on the 16:0d31/14:0 mixturecontained large amounts of a glycolipid (referred to as G1X) which was synthesizedat the expense of MGDG and DGDG. The structure of this lipid has only recentlybeen characterized in A. laidlawii membranes (Bhakoo et al., 1987) and exhibits ahydrocarbon region comprised of an exogenously supplied fatty acid and a 20-carbonpolyprenyl chain. The function of G1X in these membranes is presently unknown butit has been surmised that its flexibility is similar to that of the phytanyl lipids. Furtherstudy would be necessary for a complete understanding of the function of G1X in thesemembranes.It is of interest to compare the range of order observed for the intact A. laidlawiimembranes with that observed for liposomes composed of the total extracted lipids aswell as individual lipid species. As shown in Figure 3.5, the range of order profilesfor liposomes composed of total lipids is very similar to that observed in the intactmembranes (Figures 3.2 and 3.3). The values range from 0.176 to 0.140 for0.150.100.050.00Chapter 3. Influence of Lipid Composition on in AIB Membranes^530^2^4^6^8^10^12^14^16Carbon Number (n)Figure 3.5: Order profiles otained from the dePaked spectra of derived A.laidlawii liposomes. All spectra were the sum of 60000 tran-sients, recorded at 37°C. Symbols used in profiles correspond-ing to spe-cific fatty acid compositions are as follows: 16:0d31/18:1cL978:22 mol% (closed rectangles) 16:0d31/18:1t09 46:54 mol%(open rectangles); 16:0d31/18:1cA9 47:53 mol% (open tri-angles); 16:0d31/18:2c09,Al2 41:59 mol% (closed triangles);16:0d31/18:1cA9 21:79 mol% (open circles). Methods for de-riving the order profiles are described in Chapter 2.Chapter 3. Influence of Lipid Composition on in ALB Membranes^54the 16:0d31/18:1c09 (80:20) and 16:0d31/18:1c09(20:80) systems. Further, it may benoted that, with the exception of the 16:0d31/14:0 preparation noted above, the orderprofiles of the intact membranes and the derived liposomes are practically identical,indicating that membrane protein does not significantly influence either acyl chainmotion or packing in the hydrocarbon region.From the above results and from previous studies (Seelig and Waespe-Sarcevic,1978), it is evident that the order parameters measured are dependent upon the de-gree of saturation of the fatty acid composition. However, with A. laidlawii B, it isalso known that changes in the fatty acid composition tend to be accompanied bychanges in the polar headgroup composition of the membrane lipids (McElhaney, 1984,1989; Wieslander et al., 1980). The lipids usually found in A. laidlawii B membranesare predominantly the glycolipids MGDG and DGDG, which tend to form invertedhexagonal (HIT) and bilayer phases, respectively (Wieslander et al., 1980) and PG,a bilayer-forming phosphatide present in quantities of about 30 mol% of polar lipidcontent. Since changes in the polar headgroup composition can also influence the mag-nitude of the order parameters measured, we have examined the order profiles of thesepolar lipids. Multilamellar vesicle (MLV) dispersions of MGDG, DGDG and PG ex-tracted from 16:0d31/18:1c09 A. laidlawii membranes were therefore prepared andorder profiles determined by 2 11 NMR methods. The results are presented in Figure3.6.The glycolipids MGDG and DGDG exhibit vastly different 2 H NMR spectra at 37°C.By analogy with 2 11 NMR results obtained for bilayer- and HII-preferring phospholipids(Perly et al., 1985; Lafleur et al., 1990a,b), the DGDG spectrum is characteristic ofbilayer structure whereas the MGDG dispersions give rise to a spectrum characteristicof a mixture of (predominantly) HII and bilayer phases at 37°C. Due to the smallcontribution of the bilayer component (less than 10%), these spectra are not easily0.24 —0.06 —Chapter 3. Influence of Lipid Composition on in AJB Membranes^550 300.002^4^6^9^10^12^114^16Carbon Number ( n)0 300.24-- 0.18 —0.12 —0.06 —0.00 a2 4 8^10^12 14 16Car bon Number ( n)Figure 3.6: Bilayer order profiles of "model membranes". A) PG lipo-somes (top profile) and DGDG liposomes (bottom profile). B)MGDG:DGDG:PG dispersions in 40:35:25 molar ratios (openrectangles), MGDG:DGDG dispersions in 50:50 molar ratios(closed rectangles). All spectra were recorded at 37°C as forthe above figures. Each spectrum giving rise to the order pro-files was a sum of 24000 transients. Fatty acyl chain composi-tion was 16:0d31/18:1cA9 47:53 mol%. Methods for derivingthe order profiles are described in Chapter 2.Chapter 3. Influence of Lipid Composition on in AJB Membranes^56separable, which would be necessary to apply the dePakeing and integration techniques(Lafleur et al., 1989; Sternin et al., 1988). Thus, an order profile for MGDG liposomesis not presented. As shown in Figure 3.6, both PG and DGDG form bilayers with thecharacteristic order profile shape although the magnitude of the order in PG bilayersis significantly higher than in DGDG bilayers.MGDG/DGDG/PG (40:35:25) lipid dispersions, mimicking the lipid compositionof the parent membrane, were investigated by the above methods. The hydrocarbonorder profile obtained (Figure 6B) is essentially the same as that given by the parent,intact membrane (16:0d31/18:1c09 in Figure 3.2), showing that the proportions of thecomponent lipid species are important determinants of the magnitude of the order pro-file in A. laid lawii. Further, whereas the proportion of PG is relatively unchanged overthe range of fatty acids employed, the proportion of DGDG can change considerablyalong with GPDGDG. As noted previously (Lafleur et al., 1990a and Eriksson et al.,1991) HII-preferring lipids in bilayers can exert an ordering effect. This is also exhibitedby liposomes of MGDG/DGDG (1:1). The mixture MGDG/DGDG (1:1) exhibits a20% increase in over that of the DGDG alone where =0.164 and =0.140respectively.3.4 DiscussionThe major results of this study concern the measurement of hydrocarbon order profilesin intact biological membranes, the implications of the range of order profiles which arecompatible with growth, and the roles of individual lipid components in modulatingmembrane order. With regard to the ability to measure complete order profiles forsaturated chains in intact membranes, the 2 11 NMR dePakeing technique clearly offerssignificant advantages in comparison to previous procedures (Lafleur et al., 1989; JarrellChapter 3 Influence of Lipid Composition on in AJB Membranes^57et al., 1982) which require growing A. laidlawii on a series of specifically 2 H-labelledfatty acids. As pointed out elsewhere, for model systems (Sternin et al., 1988; Lafleur etal., 1989) the procedure describes the general features of the order gradient without theneed for synthesis and, in the present use, biological incorporation of specifically labelledacyl chains. It would clearly be useful to extend these procedures to other biologicalmembranes, such as those of eukaryotes, which are not fatty acid auxotrophs. In thisregard, it has been shown that free fatty acids as well as certain long chain alcohols, suchas tetradecanol, induce little or no change in membrane order in concentrations up to20 mol% (Pauls et al., 1983; Lafleur et al. 1990c) and that there is a strong correlationbetween the magnitude of the order parameters and the shape of the order profile. Thissuggests that perdeuteriated alcohols could be used as probes of order profiles in otherbiological membranes, a possibility which is currently under investigation.The range of order profiles compatible with high rates and extents of cell growthwere determined to correspond to =0.140 to =0.176. These values appearto fall very close to the maximum and minimum values of = 0.177 ± 5% and = 0.133 ± 5% found for DPPCd 62 and DOPC:POPCd31 (90:10), respectively,and probably reflect the limits of achievable hydrocarbon order in a fluid membranelacking cholesterol at physiological temperatures. Since the incorporation of 30 mol%cholesterol in a model membrane would result in > 0.3 (see for example, Lafleuret al., 1990c) one could argue that the maximum possible value of for a "fluid"membrane is actually much higher than that obtained here and therefore the range 0.140< < 0.176 is narrow in comparison. However, cell growth is greatly inhibited forA. laidlawii preparations falling outside this range, suggesting a requirement for thehydrocarbon order to fall within this range in order to maintain normal membranefunction. The intact membrane preparation of 16:0d31/14:0 is an example of an A.laidlawii system which meets these conditions. High rates and extents of growth wereChapter 3. Influence of Lipid Composition on in AlB Membranes^58observed for the preparation and =0.176 falls within the above limits. Since themyristate-containing membranes have a component in the gel phase, the order in thissystem may well be close to the maximum which can be obtained in an A. laidlawii Bmembrane.The observation of a restricted range in hydrocarbon order may, in turn, reflect arequirement by the organism that other membrane-related parameters are maintainedwithin fixed boundaries. For example, it has been postulated (Ipsen et al., 1990) (seealso Seelig and Seelig, 1974 and DeYoung and Dill, 1988) that membrane thickness, 2d,is directly related to the average order parameter by the relationd = dl[a < S > +1)] (3.1)where d is the average projection of the acyl chain along the bilayer normal, d1=19.7 isthe length of an all-trans palmitoyl chain (Marcelja, 1974) and a and b are numerical pa-rameters satisfying 0.5a + b=1; values of a=1 and b=0.5 have been used. According tothis analysis, the variation in order profiles determined in this study would correspondto a change in hydrophobic thickness of 1.5A (Table 3.2). Other groups have postu-lated a relationship between parameters such as polymorphism and hydrocarbon order(Epand, 1990). Since the interior of a membrane may be approximated as an incom-pressible fluid, the choice of lipid cross-sectional area as a membrane coordinate, whichis used by others (Thurmond et al., 1991), is equivalent to that of bilayer thickness.However, the details of such quantitative relationships remain, as yet, unestablished.The roles of particular membrane constituents in establishing the order profile ob-served is of interest. In the exceptional case of the 16:0d31/14:0 preparation, the verydifferent value of obtained for the lipid dispersion relative to the intact mem-brane suggests that significant protein-lipid interactions may influence the observedorder profile. Alternatively, since the DSC studies indicated that more gel phase wasChapter 3. Influence of Lipid Composition on in AJB Membranes^59Table 3.2: Hydrophobic Thickness calculatedfatty acids and compositions intact^derivedmembranes A liposomes A16:0d31/14:0,45:54(mol%) 26.7 27.416:0d31/18:1t09,46:54(mol%) 26.7 26.716:0d31/18:1c09,78:22(mol%) 26.5 26.616:0d31/18:1c09,47:53(mol%) 26.2 26.216:0d31/18:1c09,21:79(mo1%) 25.2 25.516:0d31/18:2c09, Al2,41:59(mol%) 25.2 25.2present in the derived liposomes than in the intact preparation, the proteins may in-teract with the lipids to reduce the gel-phase component in the membrane to allowcontinued viability of the A. laidlawii cells during growth. The interpretation of theseresults is complicated by the presence of the G1X and will require further examinationto elucidate.With the exception of the 16:0d31/14:0 preparation, the order profile observed forthe intact membrane is effectively identical within experimental error to that for lipo-somes composed of the total extracted lipids. This indicates that membrane proteinsdo not significantly perturb acyl chain order and packing, at least for acyl chain lengthsof 16 and 18 carbons. This is consistent with previous observations (Seelig and Seelig,1980; Bloom and Smith, 1985; Bloom, 1979) and, in the spirit of the Mattress Model ofMouritsen and Bloom (1984), suggests that the hydrophobic regions of integral proteinsare matched in a manner such that the presence of the protein does not perturb themotional freedom of the lipids on the NMR timescale.The predominant lipid species in A. laidlawii, MGDG, DGDG and PG differ sig-nificantly from one another with respect to their polymorphic phase preferences andhydrocarbon order. In addition, from the results obtained, MGDG and DGDG incombination appear to potentially play an important role in establishing the orderChapter 3. Influence of Lipid Composition on in AlB Membranes^60profile in the intact membrane. It is clear that the presence of MGDG, which adoptsthe HII phase in isolation, increases the order in bilayers formed with MGDG/DGDGmixtures. The increase has been attributed to an increase in lateral pressure inducedby the nonbilayer-forming lipid on the acyl chains of the lipid mixture (Lafleur et al.,1990a). This effect has also been observed for POPE in mixtures with POPC (Lafleuret al., 1990a).Wieslander et al. (1980) have shown for the A. laidlawii strain A that the MGDGto DGDG ratio varies for different fatty acid compositions in a manner which can beinterpreted as conserving the overall polymorphic preference of the lipid bilayer. Suchregulation has not been consistently observed in A. laidlawii strain B (McElhaney etal., 1984 and 1989). As shown in Table 3.1, the only consistent effect observed in A.laidlawii strain B is a variation in the DGDG to OAPG+GPDGDG ratio as a functionof acyl chain unsaturation. Although OAPG and GPDGDG have not received detailedattention in this regard, they also may play a role in maintaining membrane orderwithin defined limits.In summary, this work demonstrates that 211 NMR dePaking methods may be usedto determine the order gradient in the hydrocarbon region of A. laidlawii membranesgrown on perdeuteriated palmitic acid. A relationship apparently exists between hy-drocarbon order and high rates and extents of A. laidlawii cell growth. Under naturalconditions, the organism may modulate this order via acyl chain and lipid headgroupcomposition. By restricting the fatty acyl chain composition while allowing for goodgrowth of the organism, the lipid headgroup composition is varied in a manner whichcan be related to the need to achieve hydrocarbon order profiles lying within a fairlywell defined range. The contribution of lipids normally present in lesser amounts in theA. laidlawii membrane (OAPG and GPDGDG) appear to be important in this regard.Chapter 4Evidence for Two Pools of Cholesterol in the A. laidlawii B membrane4.1 SummaryInvestigations presented in Chapter 2 have indicated that there exists a well definedrange of membrane hydrocarbon order compatible with good growth of the microor-ganism A. laidlawii B (Monck et al., 1992). Since cholesterol increases hydrocarbonorder in membranes, it was of interest to examine the effect of cholesterol on the hy-drocarbon order and growth characteristics of A. laidlawii B. Cholesterol is normallyabsent from A. laidlawii membranes since it is neither biosynthesized, nor requiredfor the growth or survival of the microorganism. However, cholesterol will be incor-porated into the membrane if exogenously supplied to the A. laidlawii culture. Formembranes prepared from cells grown in the presence of cholesterol, chemical determi-nations indicated cholesterol represented as much as 40 mol% of the total membranelipid. However, 2H-NMR order parameter measurements and DSC studies of the samemembrane preparation suggested that cholesterol was present at significantly lower lev-els (P-., 10-15 mol%) in the membrane lipid bilayer. Further incorporation of cholesterolinto the A. laidlawii lipid bilayer was found to occur with an increase in temperature orby lyophilization and rehydration at high temperatures, suggesting that sterol presentin a separate pool in the membrane preparation could then gain access to the bilayer.2 H-NMR spectra of A. laidlawii membrane preparations containing deuterium labelledcholesterol indicate that the bulk of the cholesterol present in this separate pool is in61Chapter 4. Evidence for Two Pools of Cholesterol in the A. laidlawii B membrane 62a solid form.4.2 IntroductionCholesterol and related sterols are major components of at least the plasma membranesof eukaryotic cells. In contrast, prokaryotes rarely synthesize or exhibit a growth re-quirement for sterols (Nes and McKean, 1977 and Rohmer et al., 1979) a finding ofconsiderable evolutionary significance (Bloch, 1976, 1983; Nes and Nes, 1980 and Bloomet al., 1991). As stated in Chapter 1, many members of the mycoplasmas, require exoge-nous cholesterol or a closely related sterol for cell growth (Razin, 1982; Rottem, 1979).It has been found that even mycoplasmas which do not require sterol will incorporatevariable but significant amounts of exogenous cholesterol into their plasma membranes(McElhaney, 1984, 1989). Since these simple organisms offer many natural advantagesin studies of membrane structure and function (Rottem, 1979), both sterol requiringand non-requiring mycoplasma species have contributed greatly to our understandingof the role of cholesterol and related sterols in the membrane (Razin, 1982; Rottem,1979 and McElhaney, 1984;1989).A number of studies on the effect of cholesterol incorporation on the structure andfunction of the membranes of the sterol-nonrequiring mycoplasma A. laidlawii B inparticular have been carried out using a wide variety of techniques. For example, DSCstudies have shown that the incorporation of cholesterol reduces the temperature, en-thalpy and cooperativity of the lipid gel to liquid-crystalline phase transition (deKruijffet al., 1972; 1973). 2 11 NMR and ESR spectroscopic studies have demonstrated thatcholesterol incorporation substantially increases the degree of hydrocarbon order formembranes in the liquid-crystalline state (Davis et al., 1980; Rance et al., 1982; Jar-rell et al., 1983, Butler et al., 1978 and Koblin and Wang, 1981). In addition, theChapter 4. Evidence for Two Pools of Cholesterol in the A. laidlawii B membrane 63incorporation of cholesterol has been shown to reduce the nonelectrolyte permeability(McElhaney et al., 1970, 1973; de Kruijff et al., 1972,;1973), the valinomycin-mediatedK+ permeability (van der Neut-Kok et al., 1974), and the rates of glucose uptake(Read and McElhaney, 1975) in A. laidlawii cells and to reduce the ATPase activityin isolated membranes (de Kruijff et al., 1973). In most of the above studies, choles-terol levels of 14-28 mol% were obtained, consistent with the general finding that thesterol-nonrequiring mycoplasmas incorporate substantially less exogenous cholesterolinto their membranes than the sterol-requiring mycoplasmas, which typically incorpo-rate cholesterol to levels approaching 50 mol% (Razin, 1982; Rottem, 1979). Moreover,as indicated by the qualitative and quantitative effects of cholesterol on membrane lipidbilayer organization and on membrane function, it appears from most of the above stud-ies that the majority of the exogenous cholesterol associated with the A. laidlawii Bmembranes is located in the lipid bilayer. However, in two studies (Davis et al., 1980;Koblin and Wang, 1981) much higher levels of cholesterol incorporation were reported,39 and 40 mol% respectively. These values seem high for A. laidlawii B, particularlysince in both of these studies the membrane lipids were enriched in saturated fattyacids, which usually results in lower levels of cholesterol incorporation (typically 12-15mol%) than with lipids enriched in primarily unsaturated fatty acids (deKruijff et al.,1972; 1973; Wieslander and Selstram, 1987; Bhakoo and McElhaney, 1988 and Rilforset al., 1987). Using 2 H NMR dePakeing methods previously developed (Sternin et al.,1988; Lafleur et al., 1989), we found (see Chapter 3) that membrane hydrocarbon orderfalls within a defined range if good rates and extents of A. laidlawii growth are to beobserved. Thus changing the membrane composition to achieve a significant increasein order would be expected to influence the growth characteristics of the microorgan-ism. In light of the effects of cholesterol on the plasma membranes of A. laidlawii Blisted above, it is of interest to observe the effects of cholesterol on the 2 11 NMR orderChapter 4. Evidence for Two Pools of Cholesterol in the A. laidlawii B membrane 64profile derived from A. laidlawii membranes. We note that several different physicaltechniques have been used to study the membrane of the organism A. laidlawii B, (seeMcElhaney, 1984; 1989); thus it has been the most well studied of the mycoplasmamembranes to date.In the present study we investigate the effect of addition of cholesterol on lipidthermotropic phase behaviour and on the orientational order/hydrophobic thicknessin intact A. laidlawii membranes and derived liposomes. The organism was grownin the presence of perdeuteriated palmitic acid and elaidic, oleic or linoleic acid (andcholesterol). 2 H NMR order parameters, DSC studies and chemical methods were usedto determine the relative locations and quantities of cholesterol associated with themembrane. Additionally, we have used deuteriated [2,2,3,4,4,6-d6] cholesterol to furtherexamine the phase state of cholesterol associated with the A. laidlawii membrane. Theresults indicate that two pools of cholesterol exist, both tightly associated with theintact A. laidlawii membrane. Some of the cholesterol is in direct contact with thelipid fatty acyl chains while the bulk of the exogenously supplied sterol is present in asolid form in close association with the membrane.4.3 ResultsIn Chapter 3 (and in a recent paper, Monck et al., (1992)), we demonstrated that thereis a fairly narrow range of hydrocarbon order/ hydrophobic thickness that is compatiblewith good growth of the microorganism A. laidlawii B. Cholesterol has a (well known)large ordering effect in the hydrocarbon region of model and biological membranes(Huang et al., 1991; Rance et al., 1982; Davis et al., 1980; Gaily et al., 1979; Brownand Seelig, 1978; Stockton and Smith, 1976). Thus, it was of interest to examine theeffect of cholesterol on the membrane hydrocarbon order and growth characteristics ofChapter 4. Evidence for Two Pools of Cholesterol in the A. laidlawii B membrane 65Table 4.1: Cholesterol levels in A. laidlawii Membranes of Given Fatty Acid Composi-tionfatty acidcompositionCholesterol content(mol% total lipid)16:0d31/18:2c09, 01216:0d31/18:1c0916:0d31/18:1t09293741A. laidlawii B. For this experiment, equimolar mixtures of cholesterol, perdeuteriatedpalmitic acid (16:0d31) and elaidic acid (18:1tA9) were presented to the growth mediumas a mixed micelle in a small volume (P.,- 500 ml per litre of culture). Progressive growthof the microorganism was observed at a rate (and to an extent) similar to that previouslyobserved (approximately 24 hours at 37°C). The 2 11 NMR spectrum obtained for theintact membranes and extracted lipids are presented in Figure 4.1. Hydrocarbon orderprofiles and average order parameters, , were derived from 21-1 NMR spectra usingthe dePakeing and integration methods described in Chapter 2.The average order parameter obtained from the intact membrane preparation, = 0.20, is slightly greater than that found for a similar (16:0d31/18:1t09) intact mem-brane preparation lacking cholesterol; and corresponds to a relatively small 13%)increase in order over = 0.176, the average order parameter for the sample lack-ing cholesterol. This is surprisingly low given the high levels of cholesterol assayed inthe membranes used for this study (41 mol%, see Table 4.1). We note that the value obtained is slightly higher than the upper limit of the range 0.176 > >0.140 compatible with good growth characteristics for A. laidlawii B. found recently(see Chapter 3 and Monck et al., 1992).The derived liposome dispersion gave rise to the spectrum shown in Figure 4.1B fromwhich an = 0.3 was derived. This contrasts strongly with the value of = 0.2I^1^i^i^i^i^I^1^1^1^1-60 -50 -40 -30 -20 -10 0 10 20 30 40 50 60Frequency (kHz)Chapter 4. Evidence for Two Pools of Cholesterol in the A. laidlawii B membrane 66Figure 4.1: Deuterium NMR spectra of A) intact A. laidlawii B membranesand B) derived liposomes containing equimolar mixtures of16:0d31 and 18:1t09. Both spectra were measured at 37°Cand were the result of 60000 scans.Chapter 4. Evidence for Two Pools of Cholesterol in the A. laidlawii B membrane 67obtained for the intact membranes. The concentration of cholesterol in this membranepreparation constituted 41 mol% of the total membrane lipid content as determined bychemical methods (see Chapter 2). Because of this large difference between valuesin the intact membrane and the extracted lipid preparations it appeared that either lessof the endogenous cholesterol was interacting with the fatty acyl chains of the intactmembrane than with those of the lipid dispersions, or the influence of cholesterol wasreduced in the intact membranes, possibly due to the presence of proteins.In order to characterize the generality of this effect, we investigated the amountof cholesterol associated with intact A. laidlawii membranes with different fatty acylchain compositions. Cholesterol and equimolar mixtures of perdeuteriated palmitic16:0d31 and either elaidic (18:1t09), oleic (18:1c09) or linoleic (18:2cL9, 012) acidswere exogenously supplied to the growth medium. 2 1-I NMR spectra were obtainedfor each of the cholesterol-containing intact membrane and derived liposome samples.Hydrocarbon order profiles were derived using the dePakeing and integration methods(see Chapter 2) and were compared with corresponding samples lacking cholesterol.The results are presented in Figure 4.2. Approximately the same fractional increase inorder was observed for each of the above systems.As shown above, a large increase in acyl chain orientational order is obtainedupon lipid extraction and redispersion of the lipid mixture at temperatures above thegel/liquid-crystalline main phase transition. It was found that heating the intact mem-brane also gave this result. Normally, an increase in temperature results in increasedmotional freedom of the acyl chains in a bilayer which, in a 2 1-1 NMR experiment, ismeasured as a reduction in the quadrupolar splitting or a decrease in the hydrocarbonorder. Thus one would expect to see a decrease in hydrocarbon order in A. laidlawiimembranes as the temperature is increased. We examined the temperature dependence0.40^0.35_0.30_0.25-0.200.15_0.10_0.05_0.00-00.400.35-0.30_0.25_I^I^I^I^I4 6 8 10 12Carbon. Number (n)I^I^I^I^i^I2 160.00i0^2^4^6^8^10Carbon Number (n)16Chapter 4. Evidence for Two Pools of Cholesterol in the A. laidlawii B membrane 68Figure 4.2: Order profiles obtained from A) intact A. laidlawii B mem-branes containing 16:0d31 and 18:2c09, 012 (rectangles),18:1cA9 (triangles), 18:1t09 (circles) and B) from lipid disper-sions of extracted (total) A. laidlawii lipids containing 16:0d31and 18:1cA9 (rectangles) and 18:1t09 (triangles). The filledsymbols represent order parameters from membrane prepara-tions in which cholesterol was added to the growth medium.The open symbols represent order parameters from membranepreparations completely lacking cholesterol.Chapter 4. Evidence for Two Pools of Cholesterol in the A. laidlawii B membrane 69Table 4.2: Average Order Parameter vs Temperature for Intact A. laidlawii B Mem-brane Preparation with Cholesterol'temp (°C) 37 0.20147 0.15457 0.17867 0.18577 0.19587 0.177376 0.294'Fatty acid composition was an equimolar mixture of 16:0d31 and 18:1t09.b Obtained after cooling from 87°C for lh.of order profiles measured for A. laidlawii membrane preparations containing choles-terol and equimolar mixtures of 16:0d31 and 18:1t09. The results are shown in Figure4.3 and Table 4.2. As can be seen in the table, from 37°C to 47°C a decrease in isobserved as would be expected. However, at higher temperatures, 57°C to 77°C, increases monotonically and then decreases again at 87°C. A second measurement at37°C after cooling showed to be approximately 50% higher than in the initialmeasurement. This is the type of result that would be obtained in model membranesif cholesterol was incorporated into the membrane bilayer upon heating.DSC studies of cholesterol-containing and cholesterol-free A. laidlawii membraneswere performed to further characterize the effects of heating on the amount of choles-terol interacting with the membrane hydrocarbon chains. DSC thermograms typicallyfound for cholesterol-free and cholesterol-containing A. laidlawii B membranes are pre-sented in Figure 4.4. With both sets of membranes, the membrane lipid chain-meltingphase transition occurs as a reversible lower-temperature thermal event, whereas the1^■^I^I^I^I^I^I^1^1^i-60 -50-40 -30 -20 -10 0 10 20 30 40 50 60Frequency (kHz)Chapter 4. Evidence for Two Pools of Cholesterol in the A. laidlawii B membrane 70Figure 4.3: obtained for an A. laidlawii intact membrane as a func-tion of temperature. For this experiment the microogranismwas grown on 16:0d31, 18:1t09 and 20mM cholesterol andspectra were measured at A) 37°C (Initial spectrum) B) 77°Cand C) 37°C (after cooling) from 87°C.Chapter 4. Evidence for Two Pools of Cholesterol in the A. laidlawii B membrane 71endothermic transitions attributable to the thermal denaturation of the membrane pro-teins are the broad (and irreversible) higher temperature events. With the cholesterol-free membrane samples, the area under the peak attributable to the gel/liquid-crystallinephase transition of the membrane lipids is usually 10-15% lower in the first heating scan,when compared with subsequent heating scans (see Figure 4.4, left panel). The smallerchange in enthalpy in the first scan probably reflects the amount of lipid whose phasebehaviour is perturbed prior to the thermal unfolding of the membrane protein. How-ever, in the case of the cholesterol- containing membranes, the thermograms shown inFigure 4.4 (right panel) clearly indicate that the enthalpy of the chain-melting phasetransition of the membrane lipids is some 50-55% higher in the first heating scan (A)than in subsequent heating scans (B). A decrease in the enthalpy of the lipid chain-melting phase transition is precisely what would be expected if the high temperatureincubation had resulted in an increase in the amount of cholesterol interacting withthe acyl chains of the membrane lipids. It is well known that the incorporation ofcholesterol into lipid bilayer model membranes reduces both the enthalpy and the co-operativity of the gel/liquid-crystalline phase transition (McElhaney, 1982; Keough,1984).Previous 211 NMR studies of A. laidlawii showed that cholesterol causes a largeincrease in hydrocarbon order. However, these studies involved sample lyophilization(Rance et al., 1982; Davis et al., 1980) in contrast to the intact wet membrane samplesused here. An intact A. laidlawii membrane (16:0d31/18:1t09/cholesterol) preparationwas lyophilized and redispersed in deuterium-depleted buffer (see Chapter 2) in orderto compare the effect of variation of sample preparation on the values of obtainedfrom 2 11 NMR dePakeing methods. Values of ‘c,-, 0.3 were observed for such samplescompared to = 0.20 for non-lyophilized intact membrane samples prepared from0 20 40^80^80 100^0^20^40TEMPERATURE, °C80 80 100Chapter 4. Evidence for Two Pools of Cholesterol in the A. laidlawii B membrane 72Figure 4.4: Differential Scanning Calorimetric tracings of an intact A. laid-lawii membrane preparation containing 16:0d31, 18:1tA9 with-out cholesterol (left panel) and with cholesterol (right panel).A. Initial heating scan (scan rate of 10 degrees per hour). B.Final heating scan after cooling for several hours (same rate asin A).Chapter 4. Evidence for Two Pools of Cholesterol in the A. laidlawii B membrane 73the same A. laidlawii cell culture. These results indicate that lyophilization and re-dispersion of intact cells containing large amounts of cholesterol gives rise to the sametype of irreversible changes in acyl chain order as the heating and/or solubilizationprocedures described earlier.Experiments involving 211 labelled cholesterol should provide more information onthe physical environment of the cholesterol "associated" with the A. laidlawii mem-branes that exhibit these irreversible effects. Cholesterol, deuterium- labelled in the2,2,3,4 and 6, ring positions was used in the growth media in place of deuteriatedpalmitate for the following experiments in which palmitate and elaidate were suppliedexogenously. All other conditions concerning the growth of the organism remained thesame as in the experiments described earlier.The 2 11 NMR spectra obtained at 37°C for the intact A. laidlawii wet membranescontaining deuteriated cholesterol and for their derived liposomes are given in Figure4.5 (A and C). Following a heating and cooling cycle over the course of 48 hours, duringwhich spectra were taken at 50°C and 70°C (not shown), the intact membrane samplewas re-measured at 37°C. This spectrum is shown in Figure 4.5B. The quadrupolarsplittings of the individual resonances for the 2,2,3,4 and 6 positions were determinedfrom the dePaked spectrum of the derived liposomes (Figure 4.5C) and were 4kHz,33kHz, 44kHz for the deuterons on the 6, 2,4 equatorial, 2 axial and the 3 positions onthe deuteriated cholesterol respectively. The splittings from the intact wet membranespectrum (Figure 4.5B) are not as easily obtained by this method due to the poor signal-to-noise ratio although some of the splittings are evident and can be estimated fromthe spectrum. In both cases the splittings correlate well with previously determinedsplittings of [2,2,3,4,4,6-d5] cholesterol in membranes measured at 35°C and are typicalof cholesterol interacting with hydrocarbon chains (see Dufourc, 1983; Kelusky et al.,1983; Dufourc and Smith, 1986; Bonmatin et al., 1990).I^I^I^I^I^i^I^I^1^1^i—60-50 —40 —30 —20 —10 0 10 20 30 40 50 60Frequency (kHz)Chapter 4. Evidence for Two Pools of Cholesterol in the A. laidlawii B membrane 74ABI^I^I^I^I^I^I^I^I^I^I—60 —50 —40 —30 —20 —10 0 10 20 30 40 50 60Frequency (kHz)Figure 4.5: Deuterium NMR spectra of A) and B) intact A. laidlawii mem-branes and C) derived liposomes containing 16:0/18:1t/.9 50:50(mol%) and [2,2,3,4,6-d5] cholesterol. Spectra were taken atA) 37°C B) 37°C acquired after cooling for an hour from 70°Cand C) 37°C. The intensity of the central peak (due to residualHDO, membrane fragments and small vesicles) was reduced inorder to show the details of the broadline spectrum. Quadrupo-lar splittings in C) were measured from the dePaked spectrum(not shown) and are as indicated in the text .Chapter 4. Evidence for Two Pools of Cholesterol in the A. laidlawii B membrane 75The integrated area under a 211 NMR spectrum reflects the number of contributingdeuterons in the sample of interest. The ratio of the relative integrated areas of the twospectra before and after heating (Figure 4.5A, B), where the spectra were normalized tothe number of scans, was 1:2.2 indicating that differences existed in the environmentsof some of the cholesterol before and after heating. This suggests that a large fractionof cholesterol does not contribute to the type of 2 11 NMR signal arising from a fluidlipid bilayer, consistent with the calorimetric and 2H NMR results above. A possibleorganization for the pool of cholesterol not associated with the acyl chains is a solid orcrystalline form. In this regard lipids in the solid phase can have long T1 relaxationtimes which are on the order of seconds, Valic et al. (1979). If the time, TR, betweensignal acquisition is not long compared with Ti, a loss in signal intensity will occur. AT1 of 4.3 seconds was measured for solid deuteriated cholesterol in a separate experi-ment (see Valic et al., 1979 for comparison with solid cholesteryl ester). A spectrum ofsolid [2,2,3,4,6-d5] cholesterol is shown in Figure 4.6A and of [2,2,3,4,6-d5] cholesterolin an intact A. laidlawii membrane preparation in Figure 4.6B. A TR = 20 secondswas used in both cases. Although the signal to noise ratio is poor the broad spectrum(Figure 4.6B) with a Avg = 127 kHz is clearly characteristic of a spectrum of solid[2,2,3,4,4,6-d6] cholesterol.The free induction decay of the spectrum in Figure 4.6B is shown in Figure 4.6C.Two major components can be identified in the figure. The magnitudes of the fastand slower decaying components are denoted by I I and 12 respectively as shown in theFigure. 1 A distortion in the spectrum, derived from 12, has been identified and isdue to finite rf pulse width effects as described previously, Bloom et al. (1980). Asdetermined from Bloom et al. (1980) the expected magnitude for the 12 component1 The measurement of I I does not originate from 0 intensity due to the presence of a slowly decayingcomponent in the FID which contributes to the isotropic peak in the spectrum. This component definesa baseline from which I I is measured.A192 224 256I^1^i^I^1^1^132 64 96^128 160Point NumberChapter 4. Evidence for Two Pools of Cholesterol in the A. laidlawii B membrane 76i^i^I^i^I^I^i^i^1-200 -160-120 -80 -40 0 40 80 120 160 200Frequency (kHz)Figure 4.6: Deuterium NMR spectra of A) Solid [2,2,3,4,6-d5] cholesterolB) an intact A. laidlawii preparation containing 16:0/18:1tA950:50 mol% and [2,2,3,4,6-d5] cholesterol and C) FID of B withsolubilized cholesterol (I i) and solid cholesterol (I 2 ) compo-nents identified. Both spectra in A and B were measured at37°C with TR=20 seconds. The maximum splitting shown is127 kHz. We note that in B a line broadening function was ap-plied to the FID prior to Fourier transformation. The appliedfunction involved an exponential decay with T 2 =0.6 seconds.Chapter 4. Evidence for Two Pools of Cholesterol in the A. laidlawii B membrane 77given byl2cor, should be, I2corr = 12/0.8 . The ratio of the heights I 1 /I2c„,., measured atthe peak of the echo gives the relative proportions of membrane associated cholesterolto solid cholesterol in the sample. We found a ratio of I1/I2corr = 0.30 in this case whichsuggests further that the bulk of the cholesterol is in the solid phase.4.4 DiscussionThe motivation for this study was largely due to the identification of a range of hydro-carbon order/hydrophobic thickness that is consistent with good growth characteristicsof the organism A. laidlawii B. (see Chapter 3 and Monck et al., 1992). It is well knownthat cholesterol increases orientational order in the hydrocarbon region of model mem-branes (Brown and Seelig, 1978; Stockton and Smith, 1976; Lafleur et al., 1990b).Cholesterol has also been shown to be effective in this regard in A. laidlawii mem-branes (Davis et al., 1980; Rance et al., 1982) and this has been thought to be truefor A. laidlawii membranes in general (Bloom et al., 1991). However, if the observedrange in order 0.176 > > 0.140, (Monck et al., 1992) is necessary for growthof the organism at 37°C, the optimal growth temperature, one would expect that theincorporation of significant amounts of cholesterol, giving rise to >> 0.176, wouldresult in diminished A. laidlawii growth rates. As outlined in the Results section, wefound that the presence of cholesterol in the culture medium did not decrease the rateor extent of cell growth compared to cultures lacking cholesterol. In addition, the rangeof order found in these membranes corresponded to = 0.20 for 16:0d31/18:1t09down to = 0.149 for 16:0d31/18:2c09,012, an increase of no more than 13% overthat of membranes lacking cholesterol in any case. In the previous study (see Chapter3), it was difficult to determine an upper bound for hydrocarbon order due to poorgrowth of the organism on long chain saturated fatty acids at 37°C. It was presumedChapter 4. Evidence for Two Pools of Cholesterol in the A. laidlawii B membrane 78that the presence of more than 50% gel state lipid, under these conditions, inhibitedA. laidlawii growth. Given that cholesterol fluidizes a gel state membrane bilayer, itis possible that the value of = 0.20 better approximates the upper bound of therange of order than does = 0.176.A striking feature of these cholesterol-containing membrane systems is the remark-ably large values observed for the derived liposomes as compared to those observedfor the intact parent membranes. Such a difference is well illustrated by the spectrain Figure 4.1 and the order profiles of Figure 4.2. From these data, it is obvious thatcholesterol, in large amounts (see Table 5.1), is not having the same effect on the mem-brane hydrocarbon order in A. laidlawii B. as was shown previously (Davis et al., 1980;Rance et al., 1982). A logical explanation is that a large fraction of the cholesterol isinitially excluded from the intact A. laidlawii membrane lipid bilayer and may becomeincorporated through perturbation of the membranes, for example, by extraction andresuspension of the total A. laidlawii lipids.Large increases in hydrocarbon order were also observed after incubation at elevatedtemperatures and/or after lyophilization and resuspension of the intact membranes,provided that resuspension was performed at temperatures well above the gel/liquid-crystalline phase transition temperature of the intact membrane. In addition, the DSCstudies of the lipid thermotropic phase behaviour showed a decrease in the enthalpy oftransition typical of that due to cholesterol in a membrane, only after the first scan to70°C (see Figure 4.4). It is possible that protein denaturation influences the increasesin hydrocarbon order and decrease in transition enthalpy seen in the 2 1-I NMR andDSC measurements, respectively. We cannot rule out this possibility. However, a morereasonable explanation is that a large fraction of the cholesterol, "associated" withthe lipid bilayer but not interacting with the lipid acyl chains, is incorporated into themembranes during heating, lyophilization and resuspension of the membranes or duringChapter 4. Evidence for Two Pools of Cholesterol in the A. laidlawii B membrane 79lipid extraction and that this is responsible for the large increase in hydrocarbon orderor decrease in transition enthalpy. This is discussed more fully below.The use of deuteriated [2,2,3,4,6-d5] cholesterol proved fruitful in determining somephysical characteristics of the cholesterol associated with the intact A. laidlawii mem-brane bilayer. The initial spectrum obtained at 37°C for A. laidlawii membranes grownon 16:0/18:1t09 and [2,2,3,4,6-d5] cholesterol contained a sharp isotropic resonancewhich, on the basis of further experiment, probably arose from a combination of somesmall membrane fragments, residual HDO and small vesicles in the sample. The ratioof 1:2.2 observed for the integrated 211 NMR signal intensities arising from deuteri-ated cholesterol before and after heating the A. laidlawii membranes to 37°C is clearlyconsistent with the presence of two pools of cholesterol. It is estimated that with aT1 = 4.3 seconds, a solid cholesterol signal would contribute approximately 7% of theobservable 211 NMR signal obtained under the experimental conditions used here (TR= 300ms). The solid cholesterol (see Figure 4.6), although absent from the membranebilayer, must be closely associated with the A. laidlawii membrane.As discussed in the Introduction, in most of the early studies of the effect of choles-terol on the structure and function of the A. laidlawii B membrane, it appeared thatmost of the exogenous cholesterol present resided in the lipid bilayer. Thus, althoughexogenous cholesterol was presented to cultures of this organism in the same manneras in the present study, the existence of two types of cholesterol in these A. laidlawiimembranes is not obvious. In the 2 11 NMR studies of Davis et al. (1980) and the ESRstudies of Koblin and Wang (1981), where quite high levels of cholesterol were reported,it is possible that these high values arose, at least in part, because of the presence of twopools of cholesterol, one solubilized in the bilayer and one absent from it. Although thevalues of hydrocarbon order observed by Davis and coworkers for palmitate-enriched A.laidlawii membranes were compatible with their reported cholesterol levels of 39 mol%,Chapter 4. Evidence for Two Pools of Cholesterol in the A. laidlawii B membrane 80we note that the membranes utilized in those 2 11 NMR studies were lyophilized andrehydrated at an elevated temperature. Thus the phenomena described in this studycould provide a rationale both for the unusually high levels of cholesterol incorporationobserved in the study of Davis et al. (1980) and for their experimental results.Although precise the location of the solid-like pool of cholesterol is not determinedby the results presented here, (in principle) the two pools appearing in our spectra maybe related to some previously published data on the rates of cholesterol exchange be-tween A. laidlawii cells or isolated membranes and egg phosphatidylcholine/cholesterolvesicles. Davis et al. (1984) reported that in intact cells, about one-half of the choles-terol associated with the cell exhanges relatively rapidly while the other exchangesmuch more slowly. These workers initially discussed the possibility that the cholesterolpresent in the outer monolayer of the membrane bilayer could exchange rapidly, whilethat present in the inner monolayer first had to undergo transverse diffusion (flip-flop)to the outer monolayer before exchange could occur. However, since cholesterol transbi-layer movement is believed to be quite rapid in model and in most biological membranes(Philips et al., 1987), this was not a completely convincing explanation for these re-sults. Further experiments by the same workers using unsealed isolated membranes alsogave two exchange rates for cholesterol, indicating that the observed differences are notdue to a particular transbilayer distribution of cholesterol. A second suggestion wasthat the slowly exchanging cholesterol is due to some sort of preferential interaction ofcholesterol with certain classes of lipids or with certain integral membrane proteins inthe A. laidlawii membrane. If the slowly exchanging cholesterol is identified with oursolid-like pool, then the interaction presumably gives rise to an immobilization of thecholesterol molecules. It is clear that additional experimental work will be required tofirmly establish any of these hypotheses.Perhaps the simplest mechanism for the production of a solid pool of cholesterol isChapter 4. Evidence for Two Pools of Cholesterol in the A. laidlawii B membrane 81crystallization of the cholesterol and its subsequent association with the mycoplasmamembrane surfaces. Behaviour analogous to this has been observed for some macrophages(see Johnson et al. (1991) for a review). Such a crystalline state should give rise toBragg peaks in X-ray diffraction that are characteristic of crystalline cholesterol. Inan attempt to observe such peaks we detected none. However, this result does notcompletely rule out the presence of crystalline cholesterol.In summary, 2 11 NMR and DSC measurements of the effect of cholesterol on A.laidlawii membranes have given two major results. First, two pools of cholesterolare associated with the membrane bilayer. One is dissolved in the membrane whilethe other is solid-like, associated with the membrane and can be incorporated in themembrane by various solubilization procedures. Finally, the level of cholesterol residingin the bilayer results in a maximum = 0.20 which is compatible with good growthcharacteristics of the organism, A. laidlawii B.Chapter 5Anisotropy of T2 Relaxation in Pure and Mixed Lipid Systems5.1 Summary211 NMR methods have been used to examine possible mechanisms of T2 relaxation inmultilamellar dispersions and oriented multibilayers of various pure and mixed phospho-lipid systems. The use of deuterium labelling in lipid membranes provides a convenientprobe for obtaining motional and geometrical information on lipids in this environment.Since T2 relaxation times are sensitive to relatively slow motions (7 10 -3 -10 -6s) onthe NMR timescale it is possible to derive information about slow motional processes(surface undulations, diffusion and/or collective motions) which can contribute poten-tial relaxation mechanisms in a lipid environment.211 NMR measurements of the T2 anisotropy of headgroup deuteriated powders (a-DPPC d2 ) show relaxation behaviour clearly consistent with motional mechanisms dueto fluctuations in O. Similar relaxation mechanisms are found for POPC d31 multilamel-lar dispersions and for POPC d31 bilayers containing cholesterol and both, cholesteroland Leu24, a hydrophobic peptide whose hydrophobic length matches the thickness ofthe bilayer in which it is solubilized. However, a very different anisotropy of T2 relax-ation was found for oriented multibilayers suggesting that a different motional modeprovides the dominant relaxation mechanism in these systems. A phenomenologicaltheory describing fluctuations in thickness suggests that random temporal fluctuationsin the membrane hydrophobic thickness or those due to lipid lateral diffusion through82Chapter 5. Anisotropy of T2 Relaxation in Pure and Mixed Lipid Systems^83neighboring regions of different thickness can provide a potent mechanism for T2 re-laxation in membranes. Comparison of the T2 anisotropy measurements (in the ori-ented systems) with a phenomenological theory (Appendix B) is consistent with sucha mechanism. In addition, measurement of the T2 relaxation behaviour in a multi-lamellar system containing POPC d31 , cholesterol and Leu16, a short peptide whichimposes a lateral variation in hydrophobic thickness throughout the membrane plane,is qualitatively consistent with the relaxation theory due to fluctuations in membranehydrophobic thickness.Chapter 5. Anisotropy of T2 Relaxation in Pure and Mixed Lipid Systems^845.2 IntroductionIn recent years a certain emphasis has been placed on the dynamical propertiesof the membrane lipid bilayer. Studies measuring the activity of membrane proteinsreconstituted in lipid bilayers have suggested that considerable motional flexibility ofthe lipid component is necessary for the normal functioning of the proteins and thusthe membrane as a whole (see McElhaney , 1984 and Mouritsen and Sperotto, 1992and the references therein for relevant reviews). Moreover, the extent of hydrophobicmatching between proteins and proteins and lipids may have fundamental significancefor such processes as protein segregation, or receptor mediated exocytosis/ endocytosis(Mouritsen and Sperotto, 1992). Thus the identification of the motions which eitherdirectly or indirectly influence these processes and the conditions of temperature andlipid composition under which they occur are crucial to the understanding of membranefunction.Techniques which have been successful in identifying the dynamics of lipids in mem-branes include, for example, FTIR spectroscopy, various fluorescence methods andNMR spectroscopy. 31 P and 2 H NMR spectroscopy have proven particularly fruitfulin the identification of various modes of lipid motions including slow motions such aslipid lateral diffusion (Fenske and Jarrell, 1991; Auger et al., 1991; Bloom and Sternin,1987), surface undulations (Bloom and Evans, 1991; Bloom et al., 1991) or other typesof collective motions (Stohrer et al., 1991) and the faster rotations or jump motionsChapter 5. Anisotropy of T2 Relaxation in Pure and Mixed Lipid Systems^85around the long axis of the lipid molecule ( Jarrell et al. 1987; Auger et al. 1990;Morrison and Bloom, 1993). The presence of membrane surface undulations, in partic-ular, have been identified in model membrane systems as potentially providing potentmechanisms for T2 relaxation (Bloom and Evans, 1991).It has become apparent that mechanical measurements normally associated withhard materials are also important in the study of soft materials such as the modelor biological membrane lipid bilayer. The mechanical properties in these systems arecharacterized by, the elastic bending modulus k, elastic area compressibility 1/Ka andshear modulus r, each of which has been measured for certain systems (Helfrich andServuss, 1984; Evans and Needham, 1987 and Bloom et al., 1991). Progress in thisarea has facilitated the development of theories describing T2 relaxation processes,particularly with regard to motions which give rise to the various types of collectivelipid motions (see Stohrer et al. 1991; Bloom and Evans 1991).In the present study, the 2 11 NMR transverse relaxation time, T2 of various lipidspecies deuteriated in the headgroup or hydrocarbon regions are determined to try tofurther characterize slow motional processes occurring in the lipid membrane. Compar-isons with effects observed in systems containing cholesterol and either of two integralmembrane peptides (Leul6 and Leu24) (Nezil, 1992) are used to gain perspective onthe motional processes responsible for T2 relaxation in these systems. It is hopedthat the extension of these studies will ultimately provide new insights into membranelipid-protein and lipid-lipid interactions.Chapter 5. Anisotropy of T2 Relaxation in Pure and Mixed Lipid Systems^865.3 Materials and Methods5.3.1 SamplesPOPC, POPCd31 and POPEd31 were purchased from Avanti Polar Lipids (Birming-ham,Alabama). Cholesterol was purchased from Sigma (St. Louis, MO). The peptides,Leul6 and Leu24 were synthesized using solid phase synthesis methods previously pub-lished (Davis et al., 1983). a-DPPC d2 was a generous gift from Dr. Michel Roux,CEN-Saclay, France.5.3.2 Sample PreparationMultilamellar suspensions containing lipid samples were made by extensive vortexing oflipids or lipid mixtures containing the appropriate dry weight of material in deuteriumdepleted water for POPC d31 and POPEd31 :POPC and in excess Hepes buffer (50 mM indeuterium depleted water, pH 7.5, 40 mM NaCl) for a-DPPC d2 , at temperatures wellabove the gel-liquid crystalline phase transition temperature. In addition the DPPCsamples were submitted to three freezing (liquid nitrogen) and thawing (50°C) cycles,centrifuged and the pellet was transferred directly to an NMR tube.Oriented multibilayers were prepared using a slight modification of methods previ-ously described (Jarrell et al., 1987). Typically, 30-60mg of lipid, dissolved in chloro-form, was smeared on 20 glass slides of average dimensions 10mm x 7mm x 0.15mm.The slides were stacked in a 10mm (O.D.) NMR tube and excess chloroform waspumped off under high vacuum overnight. The sample was then placed in a humidChapter 5. Anisotropy of T2 Relaxation in Pure and Mixed Lipid Systems^87atmosphere at room temperature (for POPC d31 ) or 10-15°C above the gel-liquid crys-talline phase transition temperature for 72 hours. A solution of 4% polyethylene glycol8000 by weight in deuterium depleted water was then added to the samples using a1.0m1 syringe until the glass slides were completely covered (Morrison, 1993). Thesample was left to equilibrate for at least 24 hours prior to measurement by NMR.Preparation of the samples containing Leul6 and Leu24 peptides was a slight mod-ification of the procedure by Huschilt et al., (1985). Briefly, the appropriate lipids andpeptides were weighed out to achieve the desired component ratios and hydrated using800/21 of buffer (50mM Hepes, 100mM NaC1 at pH 7.4 in deuterium depleted water).The sample then underwent four cycles of freeze-thaw using liquid nitrogen followedby warming to room temperature. The samples were then pelletted at 25°C using amicroBeckman centrifuge and the pellet was transferred to a 10mm (O.D.) NMR tubefor measurement by 2 H NMR methods.Intact A. laidlawii membranes were prepared as described in Chapter 2.5.3.3 Nuclear Magnetic ResonanceThe 211 NMR measurements were performed on a home-built spectrometer operating at46.175 MHz for deuterium (Sternin, 1985). The quadrupolar echo pulse sequence:90 s-7-90y-t-acq, (Davis et al. 1976) was used for performing T2 relaxation measurements.The value of r was varied to obtain the decay of the echo envelope as a function ofpulse spacing. A range of values, typically 50ys to 2ms, was used for r dependingon the signal-to-noise ratio. t was typically lOps shorter than T to enable accuratedefinition of the echo peak. Between 5000 and 12000 transients were recorded for eachspectrum. The signals were detected in quadrature using a standard 8 Cyclops phasecycle sequence (Rance and Byrd, 1983). The temperature was controlled using a Brukermodel BV-T1000 temperature controller (Bruker Instruments, Inc., Billerica, MA).Chapter 5. Anisotropy of T2 Relaxation in Pure and Mixed Lipid Systems^885.3.4 Measurement of T2The T2 or spin-spin relaxation time may be determined from the decay of the quadrupo-lar echo or of regions of the frequency spectrum as a function of time. This decay isdue to a loss of phase memory of the spins arising from local frequency fluctuationsand as such is sensitive to relatively slow motions. In this chapter, we examine theinformation associated with the dependence of T2 on the angle 0 between the localbilayer surface normal and the external magnetic field. A phenomenological model ofthe T2 relaxation processes and their dependence on 0 can be found in Appendix B. Werefer the reader to Abragam, 1961 for a general theoretical description of T2 relaxation.The angular dependence of T2 relaxation in powder samples can be examined usingmethods previously developed (Nezil et al., 1991). By fitting the decay of intensitiesat corresponding points in a series of spectra (F(7)) as a sum of a variable number ofexponentials using a non-negative least squares algorithm, a "relaxation spectrum" canbe obtained. In a frequency spectrum of a given 2H site (with a particular value of S(n))each value of 0 contributes to the magnitude of the intensity. Since a Pake doubletcontains two contributions to the intensity in the region between the 90° edges but oneelsewhere, one can expect to obtain two T2 relaxation times between the edges and onerelaxation time elsewhere for each deuterium site. The resulting relaxation spectrumis parameterized using a weighted average of the relaxation rates determined, 1/T e2ffIn the oriented samples, T2 relaxation times were calculated as a function of thepulse spacing for all resolved carbon atoms at certain values of 0. The decay of the spec-tral peak intensity, I, was adequately fitted by an exponential function of 7 accordingto 47) = Ioe -2T/T2 where I() is the peak intensity at 7 = 0.Chapter 5. Anisotropy of T2 Relaxation in Pure and Mixed Lipid Systems^895.4 ResultsA system previously selected for the study of T2 relaxation anisotropy (for the purposeof demonstrating relaxation spectra, Nezil et al., 1991) was DPPC bilayers, deuterium-labelled in the a position of the PC headgroup. The advantage in using this system wasthat a singly labelled CD 2 species would provide a clean system for study. Thus it isuseful to present this data here for comparison with the more complicated spectra dis-cussed below. Spectra of the lipid dispersions were obtained at 50°C (approximately 5degrees above the main phase transition temperature). The T2 relaxation spectrum wasdetermined for the series of spectra taken as a function of the pulse spacing (90-T-90)using methods described in the materials and methods section. There was essentiallyone relaxation rate between the shoulders (0°) and edges 90°, and two between theedges, consistent with a single exponential for each angle 0. The Pake doublet, the T2relaxation spectrum and a fit of the relaxation data to1=^Bsin20cos20 1T2 T20(5.1)are shown in Figure 5.1. The reason for the choice of the fit to an equation of this typeis outlined in the paper by (Bloom and Evans, 1991) where sin 2 Ocos2 0 is the angulardependence expected for a lipid membrane undergoing fluctuations about the surfacenormal. Such fluctuations provide local field fluctuations at the nucleus giving rise toT2 relaxation. A summary of this theory is presented in Appendix B.The coefficients in equation 5.1 were defined using experimental values of 1/7V fat 0 = 0° and 0 = 54.7°, giving values of 0.30 and 6.7 (ms -1 ), respectively. TheisT20identified with angularly independent relaxation processes. The fit of the orientationChapter 5. Anisotropy of T2 Relaxation in Pure and Mixed Lipid Systems^903.••••■w2NW 1CDH0-10^-5^0^5^10Frequency (kHz)Figure 5.1: Relaxation spectrum of a-DPPC d2 (solid line) obtained usinga NNLS algorithm with a weighted harmonic average of therelaxation rate 1/T2 to parameterize the relaxation. The re-laxation spectrum was fitted by an equation of the form 1/T 2= 1/T20 + B sin 2 Ocos 20 and plotted as a function of frequency(dashed line). The powder spectrum (top) is presented for ref-erence. All measurements were made at 50°C. (Spectra werecourtesy of C. Morrison).1Chapter 5. Anisotropy of T2 Relaxation in Pure and Mixed Lipid Systems^91dependence of T2 suggests that motions such as thermal fluctuations of the orientationof the local surface normal are dominant mechanisms for T2 relaxation in the headgroupdeuteriated system. From the coefficient B, one can determine an average value of thecorrelation time, r0 , for the motion contributing the relaxation process (cf AppendixB). For a-DPPCd2 with a Avg 7kHz, the value of B is identified with a correlationtime of ti 4 x 10's, similar to that predicted by Bloom and Evans, (1991).In order to determine if the same types of motions are dominant in the hydrocar-bon region of the bilayer, a similar investigation using chain deuteriated lipids was ofinterest. POPC d31 multibilayers were measured using the T2 methods described above.POPCd31 was chosen, in part, for reasons of availability.The powder spectrum of POPC d31 perdeuteriated in the palmitoyl chain representsa superposition of powder spectra from each of the deuterium labelled carbons resultingin T2 values which are weighted averages over angular contributions from contributingdeuterons. The T2 relaxation spectrum obtained for POPCd31 is presented in Figure 5.2.The prominent feature of this spectrum is that the longest T2 values are found at the90 and 0 degree orientations suggesting that a T2 relaxation mechanism predominantlydue to fluctuations in the local surface normal is the mechanism for all chain labelledpositions, as was the case for the headgroup-labelled powder sample of cy - DPPCd2.We thought that it would be possible to select motions such as surface undulationswhile diminishing effects due to large membrane curvature using samples where thelipids were macroscopically oriented between glass plates (oriented samples). In thecase of POPCd3 1 , one could obtain motional information at the 7 labelled acyl chainsegments with resolved splittings and the plateau region (carbons 2-8). Accordingly, theangular variation of T2 has been measured at all (resolved) deuterium-labelled carbonpositions of the palmitoyl chain in POPC d31 with the exception of the methyl deuterons.The results are illustrated in Figure 5.3 (left panel) for representative CD2-labelledChapter 5. Anisotropy of T2 Relaxation in Pure and Mixed Lipid Systems^92I^I^I^I^I^I^I^I^I—50 —40 —30 —20 —10 0 10 20 30 40 50Frequency (kHz)Figure 5.2: Relaxation spectrum of POPC d31 (solid line) obtained usinga NNLS algorithm with a weighted harmonic average of therelaxation rate 1/T 2 to parameterize the relaxation. The pow-der spectrum (dotted line) is presented for reference. Note theshoulder (the 0° orientation at r-:.1± 25 kHz) and edge (the 90°orientation at ± 25 kHz) regions which show average relax-ation times of 500ms and r-z..-1650ms respectively. The samplewas measured at 30°C. (Plot is courtesy of Dr. F. Nezil).76 —5 —rn4 —321 —Chapter 5. Anisotropy of T2 Relaxation in Pure and Mixed Lipid Systems^93carbon positions of the palmitoyl chain. All other chain positions showed a similarangular dependence of T2. Two features of the data are particularly notable. First,the anisotropy of T2 relaxation is inconsistent with the theory for surface undulations(T 2 values are shorter for the 0° orientation than for the 90° orientation). Second,compared to the magnitudes of T2 measured for the powder sample 500-700ys),the magnitudes of T2 at particular positions down the chain are significantly longer,ranging from 700ms in the plateau region at 9 = 0 degrees to 7ms for carbon 15at 9 = 70 degrees.It is possible to show that fluctuations affecting the splitting frequency can oc-cur in both 9 and in (see Appendix B). Fluctuations in could arise, forexample, from fluctuations in local membrane thickness, as discussed in Chapter 3.These motions are referred to as fluctuations induced by random diffusion of lipidsthrough regions of different membrane thickness or "thickness fluctuations". Accord-ing to a phenomenological theory describing such thickness fluctuations (Appendix B),the transverse relaxation rate as a function of 9, 1/T2 , should show an angular depen-dence corresponding to [P 2 (cos0)] 2 . As shown in Figure 5.3 (Fl), the T2 data wereadequately approximated by a sum of an orientation independent term, A, and a terminvolving [P 2 (cos0)] 2 in the form 1/T 2 = A + C[P 2 (cos9)] 2 . The data can also be ex-pressed in a more general form that is independent of a specific motional model using asum of terms P o (cos9), P 2 (cos0) and P4 (cos0), where P o(cos0) = 1, P2 (cos0) (3cos 20- 1)/2 and P 4 (cos0) (35cos40 - 30cos 20 + 3)/8. We show in Figure 5.3 (F2), that thedata were also adequately fitted by an expression of the form1 Tan) = A02 ) ;127) P2(CO30)^P4(COS9) (5.2)Reasonable fits, Fl and F2, to the T2 relaxation data were obtained for all carbonpositions, n. A comparison of F1 and F2 for each carbon atom using x 2 methods0.6^0.4cos 002 0010 0.80.6^0.4COS 010 0.8 000.2Chapter 5. Anisotropy of T2 Relaxation in Pure and Mixed Lipid Systems^94C13F2210 01.8 0.6^0.4COS 00.2 0010 0.8 0.6^0.4COS02 00Figure 5.3: T2 1 values (open rectangles) of representative carbon po-sitions of macroscopically oriented POPCd31 (left panel)andPOPC:POPEd31 50:50 (right panel) and the corresponding fitsto the data are presented. The data were fitted (solid curves)by F1) 1/T 2 = A + C[P 2 (cosO)] 2 and F2) 1/T 2 = A oA 2 P2 (cos9) A 4P4 (cos9). All samples were measured at 30°C.Chapter 5. Anisotropy of T2 Relaxation in Pure and Mixed Lipid Systems^95indicated that the fit to equation 5.2 was slightly better in most cases (as can be seenby eye).It is possible to relate the coefficients Atli of equation 5.2, to terms A -10 n) , B (n) andC(n) via equations 5.3 - 5.5,A,(92) = AV + 15—2 B (n) + 5—1 c(n)A (22) = 42 ).f + 21B(n) + c(n)7420 = l-1A (n)f —^-0,LEL „,„(n) + 18 c (n)42^35 35and compare the relative contributions of fluctuations in 0 and thickness to T2relaxation using the ratio A42 ) /A2) . This ratio (calculated for each carbon position) ispresented in Table 5.1. 1 (Values of A2(n2 )1 , B (n) , and C(n) are given in Appendix A). Notethat this is close to or larger than 1 in most cases. For A (4n2) /A n2 ) r-::% 1, the ratio B'(n) 2/On) must be;.:-., 0.175. Thus the contribution from C(n) is larger than from BV(n) for thePOPC d31 system studied. We note that if W (n)(or On)) was zero, the ratio A42 ) /A2 )would be 1.8(or -2.4). Recall that Bqn) and C(n) refer to the relative magnitudes ofcontributions from fluctuations in 0 and respectively. Thus thickness fluctuationsappear to provide the more dominant contribution to T2 relaxation in the POPCd31oriented system.(nIt is of interest to compare the contributions of A02) f , the B(n) and C(n) terms to A0n2 )in equation 5.3 above and their maximum contributions to the overall T2 relaxation.'In order to proceed we assume that 42 )1 = A (42 )1 = 0 (see Appendix B). This is a reasonable firstapproximation since A r'2 ) 'f and A '2 ) 'f contribute to both T 1 and T2 relaxation and it is generally foundthat the orientation dependence of T2 relaxation times is stronger than the orientation dependence ofT 1 . Whereas a variation of 20-30% is observed in T1 measurements of the deuterons on the acyl chain(Morrison and Bloom , unpublished results) roughly a factor of 5 is observed for most carbon positionsin T2 measurements.2 The maximum value which sin 2 Ocos 2 0 can attain is one quarter that of [P2(cosO) 2 . Thus we defineBqn)=B(n)/4.Chapter 5. Anisotropy of T2 Relaxation in Pure and Mixed Lipid Systems^96Table 5.1: Ratios of T2 (F2)fit coefficients for POPCd31 and POPC/POPEd31 bilayersand A02 valuesPOPCd31 POPC/POPEd31 its -1CarbonNumberA42/A22 A02 X10 -4/18 -1 A42/A22 A02 x10-4/28-1C15 0.95 ± 0.20 1.90 + 0.12 0.74 ± 0.30 2.22 + 0.167C14 0.90 + 0.18 2.62 + 0.12 1.19 ± 0.31 2.93 + 0.23C13 1.11 ±0.16 3.03 + 0.12 1.14 + 0.28 3.54 + 0.24C12 1.08 + 0.11 3.30 + 0.09 1.58 + 0.18 3.81 + 0.17C11 1.54 + 0.097 3.78 + 0.10 1.17 + 0.14 4.21 + 0.15C10 0.92 ± 0.41 3.97 ± 0.40 0.92 ± 0.42 5.11 ± 0.47C9 0.91 ± 0.32 4.69 ± 0.29 0.64 ± 0.45 5.46 ± 0.40PI 1.05 + 0.095 5.37 + 0.12 0.98 ± 0.25 6.09 + 0.34For most carbon positions we find that A (07.2L)f (2/15)B(') (1/5)C(n ) . This may beexpected since A02)f is due to angular independent relaxation processes and suggeststhat angular independent relaxation processes make an important contribution to theoverall T2 relaxation in these membrane systems.In order to get a rough estimate of possible correlation times for any surface undu-lations present in these systems we have used values of B(n) determined from equations5.3, 5.4 and 5.5 above. In making these estimates we assume a certain amplitude,<(89)2 >, for surface undulations and obtain correlation times from the measurementof T2. Thus for B a 71,n) (see Appendix B for details), we estimate that such correlationtimes are of the order of 10 -6 - 10 -7 seconds. These are somewhat shorter than in theheadgroup-labelled powder samples.Thickness fluctuations should be further pronounced in the fluid phase of a mixedsystem containing lipids of different intrinsic hydrophobic thicknesses. Since thicknessdifferences had been observed in POPC:POPEd31 dispersions as a function of the ratioChapter 5. Anisotropy of T2 Relaxation in Pure and Mixed Lipid Systems^97of the two components (Lafleur et al., 1990), a mixture of these lipids was a firstchoice. Thus T2 measurements were made on oriented samples of POPC: POPE d31(50:50 mol%). For comparison with pure POPC d31 results, a plot of the T2 data forcorresponding CD 2 labelled carbon positions of the palmitoyl chain as a function ofis presented in Figure 5.3 (right panel). The values A (42 ) /A (22) are presented togetherwith those for the POPC d31 system in Table 5.1. As in the POPC d31 oriented samples,the fits, F2, are slightly better than fits, Fl (determined by x 2 ). Compared with thePOPCd31 sample, the ratios A42)/A2) are the same within experimental error indicatingthat contributions to relaxation from fluctuations in 0 and are similar in both thePOPC d31 and POPC:POPE d31 oriented samples. Thus, in the mixed oriented system,the component due to thickness fluctuations appears to make a larger contribution toT2 relaxation than does the component due to fluctuations in 0. However, thicknessfluctuations do not appear to be "enhanced" over those in the POP Cd31 oriented sample.Measurements of ternary lipid bilayers are included since they provide a strikingdemonstration of the influence of hydrophobic mismatch on T2 relaxation. The mem-branes used were POPC d31 based multilamellar vesicles (powder-like) containing eithercholesterol and Leu24 peptide, where the peptide hydrophobic thickness matches thatof its membrane environment, or containing cholesterol and Leu16 peptide (Nezil andBloom, 1991), where the hydrophobic length of the peptide was short causing the av-erage bilayer thickness to decrease by 10%. We can anticipate that POPC d31 moleculesundergoing diffusion will encounter regions of very different thicknesses here. Thuslateral variation in thickness obtained in this manner should be reflected in the T2anisotropy across the spectrum. Accordingly, T2 was measured for three powder sys-tems: 1) POPC d31 multilayers containing 30mol% cholesterol; 2) System 1 containing30% Leu24 peptide by weight; and 3) System 1 containing 30% Leul6 peptide byweight. The relaxation spectra of these are presented in Figure 5.4. The shapes ofChapter 5. Anisotropy of T2 Relaxation in Pure and Mixed Lipid Systems^98the relaxation spectra are very distinctive. It is evident that those for systems 1 and2 have the same general features in that corresponding edges and the shoulders showsimilar relaxation times. We interpret these results to mean that the anisotropy ofT2 relaxation for all labelled CD 2 positions could be fit by a function that includessin2 Ocos 20 terms. However, the relaxation spectrum for system 3 is clearly different.It is suggested, from the shape of this relaxation spectrum, that the mechanism ofrelaxation could be described by a function involving [P 2 (cost9)] 2 terms, as indicatedby the theory for random thickness fluctuations.Finally, some preliminary T2 relaxation studies were performed using A. laidlawiimembranes containing the fatty acid composition, 16:0d31/14:0 to determine if suchrelaxation techniques can be applied to biological membranes. This system was chosenbecause of some results of Chapter 3, where intact membranes containing these fattyacids appeared to be thinner than their derived liposomes. Thus certain features ofthe frequency spectra of B2 and B2 + Leul6 membranes are similar to the derivedliposomes and intact membranes, respectively, of A. laidlawii containing 16:0d31/14:0fatty acids. The relaxation spectrum and Pake-doublets spectrum taken at r = 50fts(for reference) are presented in Figure 5.5. Although noisy, we find that similar fea-tures of the T2 relaxation spectra of B2 and B2 + Leul6 membranes also appear in theT2 relaxation spectrum of this A. laidlawii membrane system. This suggests that sur-face undulations and/or thickness fluctuations may also be important in this biologicalmembrane system.7610 i^I^I^I^1^I^—40 —30 —20 —10 0^10 20Frequency (kHz)130 40Chapter 5. Anisotropy of T2 Relaxation in Pure and Mixed Lipid Systems^99Figure 5.4: Relaxation spectra are shown for the POPC d3i :Cholesterol bi-layer containing 30mol% cholesterol B2 (solid line), with 30%(by weight) Leul6 peptide (double line) or with 30% (byweight) Leu24 peptide added. All samples were measured at30°C. (Plot is courtesy of Dr. F. Nezil.)1^1^i—10 0 10Frequency (kHz)■—20—30 20 30Chapter 5. Anisotropy of T2 Relaxation in Pure and Mixed Lipid Systems^100Figure 5.5: A) Relaxation spectrum and B) Superposition of powder pat-terns for reference are shown for the A. laidlawii intact mem-branes grown using the fatty acids, 16:0d31 and 14:0. Thesample was measured at 37°C.Chapter 5. Anisotropy of T2 Relaxation in Pure and Mixed Lipid Systems^1015.5 DiscussionSome of the earliest observations of surface undulations were obtained from light micro-scopic studies of erythrocytes (Brochard and Lennon, 1975). Scattering of the incidentlight was due to random changes in the surface normal of the erythrocyte membraneproducing an effect which is commonly known as the "flicker effect". More recent ob-servations, also at long wavelength scales, were made using giant vesicles (Evans andRawicz , 1990) where it was possible to observe undulations under the light microscope.From the long wavelength measurements, Evans and Rawicz (1990) found that it waspossible to predict undulations occurring on the mesoscopic (— 200 A) length scale.Only a few investigators have directly studied these and similar types of motions on themesoscopic length scale, which is within the realm of 2 H-NMR transverse relaxationmeasurements (Bloom and Evans, 1991; Stohrer et al., 1991, Watnick et al., 1991).We have studied several different membrane systems both macroscopically orientedmultibilayers and multilamellar vesicles using 2H-NMR T2 relaxation methods to deriveinformation, on the types and modes of motions dominant in lipid membrane systems.It appears that the headgroup deuteriated a-DPPC d2 and chain deuteriated POPCd31samples in the fluid phase show T2 relaxation processes clearly consistent with motionsresulting from fluctuations of the orientation of the local surface normal. Results usingchain deuteriated oriented samples, surprisingly, give a different dominant relaxationmechanism. The main mechanism suggested for the oriented samples is supported by aphenomenological theory of thickness fluctuations which describes T2 relaxation due torandom diffusion of lipid molecules through regions of different thickness. These resultsare discussed within the framework of the theory for surface undulations and thicknessfluctuations. It is assumed throughout that the influence of dipolar interactions torelaxation are negligible.Chapter 5. Anisotropy of T2 Relaxation in Pure and Mixed Lipid Systems^102It is of interest to compare the modes of surface undulations in the a-DPP Cd2membranes and the macroscopically oriented POPCd 31 multibilayers using the esti-mated re(n) values and the angular dependence of the relaxation times obtained. Thevalues of Tt(in) can be interpreted as averages over all undulational modes and longervalues correspond to longer wavelengths. Then, for TP ) --:-., 10 -5s, longer wavelengthmodes must be present, on average, than for 7-P ) ,c..--, 10 -6-10 -7s. This suggests thatlonger wavelength modes present in the powder a-DPPCd2 samples are damped outin the oriented multibilayers. Thus the relative contribution of surface undulations torelaxation in the oriented samples could be expected to be small. Furthermore, a smallcontribution to relaxation from thermal fluctuations in the orientation of the surfacenormal in the oriented samples is qualitatively indicated by the lack of similarity ofT2 relaxation times at the edge and shoulder orientations. Quantitively the ratios ofA42) /A22) ;,-, 1 suggest that the contribution from undulational type motions is .c.. --, 1/5that of fluctuations in thickness in both oriented multibilayer systems presented.The above observations are also supported by a comparison of T2 values in orientedand powder POPC d31 samples. The average value of T2 measured for such a powder-system was 1515 ,us. The average value of T2 in the corresponding oriented sample,calculated from values measured at all carbon positions for all angles and weighted bysiner, is1513 ,us, significantly longer than that in the powder. Thus some motionscontibuting to transverse relaxation in the powders must be absent in the orientedsamples.We suggest that the On) term contributing to the relaxation in the oriented sam-ples can also be present in the powder samples. To demonstrate this we examine thePOPC d31 relaxation spectrum (cf. Figure 5.2) and the A02 )-f and On) (see Appendix A)values of the plateau region at 0 = 0. At this angle the intensity of the plateau region'Sin 0 represents the probability density of nuclei with a given value of 0 in a powder sampleChapter 5. Anisotropy of T2 Relaxation in Pure and Mixed Lipid Systems^103is relatively clean (i.e. contains essentially no intensity from other nuclei or angles) andsurface undulations do not contribute to T2 relaxation. The 1/T 2 value of the powderspectrum in this region, Re, 2000 s -1 , is larger than the sum Arniateauf cptateau, ti 12005 -1 suggesting that the CP lateau term may contribute at least as much to T2 relaxationin the powder as in the oriented sample.At the present time one can derive only relative contributions to T2 relaxation fromthickness fluctuations. It is suggested here that these are dominant in the macroscopi-cally oriented multibilayers investigated. It is possible to estimate some plausible upperand lower bounds for correlation times due to random thickness fluctuations assumingthat these arise from random diffusion through regions of different thickness. Usingan approximation for these correlation times (see Appendix B), Te = < A >/D, D =4 x 10' cm2/s (a typical diffusion constant for some fluid membranes (Bloom et al,1978; Cullis, 1976)), and an appropriate range of areas, we can estimate the correla-tion times over which the phenomenological theory describing thickness fluctuations isvalid. Since the average cross-sectional area for a lipid molecule such as POPC is 40-60A2 the motions must be slower than 10 -8s. At the other extreme, the correlation timemust be faster than the magnitude of T2 (< 7 ms for the longest T2 values measuredhere) for the motion to be a mechanism for relaxation. Thus 10's < r < 10's givesa range of allowable correlation times. As an example of a plausible correlation time,a value of 7, of the order of 10,as would allow a lipid molecule to diffuse over an areaof 4000 A2 , Pi 70-100 lipid molecules.The relative contributions to T2 relaxation from fluctuations in and 0 in ori-ented POPC d31 and POPC:POPEd31 are of interest. One would anticipate that themixed system containing lipids of different intrinsic thickness should show a larger con-tribution from fluctuations in than from fluctuations in 0. The similarity in theratios A (4'.2' ) /A ) 1) for all carbon positions of both POPCd 31 and POPC:POPEd31Chapter 5. Anisotropy of T2 Relaxation in Pure and Mixed Lipid Systems^104oriented samples suggests that there is no relative increase in fluctuations in forthe mixed system. Thus if such differences occur, they are not detectable by thesemethods perhaps because, by , the hydrophobic lengths (measured by ) ofPOPCd31 and POPEd31 membranes differ by only a few percent. It could be useful,with similar methods, to study DMPC:DSPC oriented samples for example, wherehydrophobic lengths of the individual components are different by more than 10%.As indicated in the Results section the POPC d31 :Chol system containing either theLeul6 or Leu24 peptide provides an elegant system for the study of the influence ofthickness differences in a membrane. Significant qualitative information can be derivedfrom the data. For the POPC:Chol and POPC:Chol: Leu24 systems, a distinctiveshape of the relaxation spectrum was obtained. The angular dependence of relaxation isclearly consistent with an undulational type of lipid motion mentioned earlier. However,in membranes containing the short peptide, Leu16, the dominant relaxation mechanism,suggested by the shape of relaxation spectrum, appears to be provided by randomthickness fluctuations. These membrane systems containing a specifically labelled lipidcould provide excellent model systems for further testing of such theories.Using the biological membrane A. laidlawii B, as much as could be discerned, fea-tures of the relaxation spectrum resembled both, spectra exhibiting sin 20cos2 9 relax-ation behaviour and [P 2 (cose9)] 2 relaxation behaviour. If membrane topology was suchthat some regions of the membrane were protein deficient and others were proteinenriched (assuming that lipid-protein hydrophobic mismatch was occurring) then thepredominant lipid fluctuations in each of these regions could be from fluctuations in 0and respectively. It is evident that one must initially clarify the relative contri-butions of these types of fluctuations in model and then biological membranes.In general the hydrophobic regions of proteins and lipids in natural membranesare probably relatively well matched under normal conditions. One would expect thatChapter 5. Anisotropy of T2 Relaxation in Pure and Mixed Lipid Systems^105in lipid-protein or lipid-lipid interactions there is a threshold level of mismatch abovewhich segregation of individual species occurs (Mouritsen and Sperotto, 1991). Indeedit has been demonstrated that the presence of > 50% gel state lipid (thick mem-branes), substantially reduces the activity of certain integral membrane proteins ( seeMcElhaney, 1984; Mouritsen and Sperotto, 1991 for reviews); presumably since highprotein activity cannot exist with this level of gel-state lipid in the membrane.In summary, the results presented here indicate that both surface undulations andrandom thickness fluctuations can provide dominant mechanisms for T2 relaxation inmembrane systems. Surface undulations or other types of collective motions have beensuggested for some systems (Bloom and Evans, 1991; Stohrer et al. 1991). Thicknessfluctuations are motions which we suggest to be present in membrane lipid bilayers,particularly pronounced where significant hydrophobic mismatch of membrane compo-nents occurs. The establishment of the presence of such motions in membranes withfurther study may suggest certain consequences for membrane function and dynamics.Chapter 6Future DirectionsIt is conceivable that the range of order consistent with the limits of growth of A.laidlawii as identified in chapters 3 and 4 would be similar in a wide range of fluidplasma membranes. However, for membranes containing high levels of cholesterol, onewould expect a significant increase in the order observed. For instance, the membranelipids of model stratum corneum, (ceramide, cholesterol and free fatty acids 1:1:1)appear to be highly ordered and potentially inhomogeneous mixed in the membrane.An examination of the average order/thickness in other biological membranes wouldbe of interest to determine the distribution of thicknesses present in active membranes.Diffusion rates and mechanisms can provide significant information about the lateraldistribution of membrane components and their miscibility. Such techniques would beuseful in clarifying some questions which have arisen during the course of this thesis. Itis noted that several techniques have been developed to measure diffusion in membranes.In particular the pulsed field gradient NMR methods of Lindblom et al. (1976, 1977)enable one to study not only diffusion rates, but to identify non-micellar isotropic phasestructure and the presence of regions of bounded diffusion in membrane bilayers. Suchmethods could be useful in the study of, for instance, A. laidlawii membranes such asthose containing the fatty acid combination 16:0/14:0 discussed in chapters 3 and 5.Some questions which arise from this work are whether an inhomogeneous distributionof membrane constituents exists in this system and if so, to what extent.Fluorescence recovery after photobleaching (see Knowles, 1992) and 2D 31 P-NMR106Chapter 6. Future Directions^ 107(Fenske and Jarrell, 1991) methods have been identified as useful techniques in mea-suring diffusion constants. We have attempted to directly compare the diffusion coef-ficients, D, in POPC and POPE membranes with these two techniques. The POPCdiffusion coefficients D 3x10 -8 cm2 /s and 5x10 -7 cm2/s from FRAP and 31P NMRrespectively are comparable. However, the POPE diffusion coefficients are different bymore than an order of magnitude D 2.5x10- 8 cm2/s and 5x10 -7 cm 2 /s from FRAPand 31P-NMR measurements respectively. Difficulties with both techniques are identi-fied. For instance, using 2D 31 P NMR techniques an accurate vesicle size is necessaryfor the correct determination of D. This is not always easily achieved with standardtechniques particularly if vesicle aggregation and/or a large distribution in size or shapeare common (as may be the case with POPE vesicles). The requirement of a fluores-cently labelled lipid, accurate definition of the bleach spot size, temperature regulationand vesicle manipulation are often drawbacks using FRAP techniques particularly ifhomogeneity of vesicle constituents is compromised. We note that in particular, im-provements in the NMR methods to better define the distribution of vesicle shapes andsizes is necessary to define an accurate value of D for these measurements.The slow motions present in membranes are of interest. The results identifying thevarious motions, fluctuations in 9 and thickness, in the POPC d31 and POPC:POPEd31systems and the Leu16/Leu24 containing membranes provide qualitative informationon the motions predominant in these systems. Further development of the theorydescribing the T2 relaxation mechanisms will provide more quantitative methods ofidentifying the motions responsible for the relaxation. We hope that the eventualextension of these methods to biological membranes, such as A. laidlawii strain B ofvarious fatty acid compositions, will achieve further understanding of the dynamics ofbiological lipid membranes.Chapter 7BibliographyAbragam A. (1961) Principles of Nuclear Magnetism London:Oxford UniversityPress.Auger, M., Carrier, D., Smith I.C.P. and Jarrell, H.C. (1990) J. Am. Chem. Soc.112, 1373-1381.Auger, M.A., Smith, I.C.P and Jarrell, H.C. (1991) Biophys. J. 59, 31-38.Bhakoo, M., Lewis, R.N.A.H and McElhaney, R.N. (1987) Biochim. Biophys.Acta 922, 344-345.Bhakoo, M., and McElhaney, R.N. (1988) Biochim. Biophys. Acta 945, 307-314.Bligh, E.G. and Dyer, W.J. (1959) Can. J. Biochem. Physiol. 37, 911-917.Bloch, K. (1976) in Reflections in Biochemistry (Kornber, A., Horecker, B.L.,Cornudella, Z., and Oro, J., Eds.) Pergamon Press Oxford.Bloch, K. (1983) CRC Crit. Rev. Biochem. 14, 47.Bloom, M., Burnell, E.E., Roeder, S.B.W. and Valic, M.I. (1977) J. Chem Phys.66, 3012-3021.Bloom, M. (1979) Can. J. Phys. 57, 2227-2230.Bloom, M., Davis, J.H., and Valic, M.I. (1980) Can. J. Phys. 58, 1510-1517.108Bibliography^ 109Bloom, M. and Smith, I.C.P. (1985) Progress in Protein-Lipid Interactions, Else-vier Science Publishers B.V., Amsterdam, The Netherlands.Bloom, M. and Sternin, E. (1987) Biochemistry 26, 2101.Bloom, M. and Evans E. (1991) Observations of surface undulations on the meso-scopic length scale by NMR In: Biologically Inspired Physics, pp 137-147 (ed. L.Peliti) Plenum Press, New York.Bloom, M., Evans, E. and Mouritsen, 0.G., (1991). Quart. Rev. Biophys.,24,293-397.Bloom, M., Burnell, E.E., MacKay, A.L., Nichol, C.P., Valic M.I. and Weeks G.(1978) Biochemistry, 26, 5750-5762.Boggs, J.M. (1987) Biochim. et Biophys. Acta 906, 353-404.Bonmatin, J.-M., Smith, I.C.P., Jarrell, H.C. and Siminovitch, D. J. (1990) J.Am. Chem. Soc. 112, 1697-1704.Brochard, F. and Lennon, J.F. (1975) J. de Physique 36, 1035-1047.Brown, M.F. and Seelig J. (1978) J. Am. Chem. Soc. 17, 381-384.Butler, K.W., Johnson, K.G. and Smith I.C.P. (1978) Arch. Biochem. Biophys.191, 289-297.Cain, J., Santillan, G. and Blasie, J.K. (1972), in Membrane Research, Fox, F.,Ed., New York, N.Y., Academic Press.Cullis, P.R. (1976) FEBS Lett. 70, 223-228.Cullis, P.R. and de Kruijff, B. (1979) Biochim. et Biophys. Acta. 559, 399-420.Bibliography^ 110Cullis, P.R. and Hope, M.J. (1984) Physical properties and Functional Roles ofLipids and Membranes. In Biochemistry of Lipids and Membranes, eds. Vance,D.E. and Vance, J.E., 25-72. The Benjamin/Cummings Publishing Company,Inc.Cullis, P.R., Hope, M.J. and Tilcock, C.P.S. (1986) Chem. Phys. Lipids 40,112-144.Danielli, J.F. and Dayson, H. (1934) J. Cell. Comp. Physiol. 5, 495-508.Davis, J.H., Jeffrey, K.R., Bloom, M., Valk, M.I., and Higgs, T.P. (1976) Chem.Phys. Lett. 42, 390-394.Davis, J.H., Bloom, M., Butler, K.W. and Smith I.C.P. (1980) Biochim. Biophys.Acta. 597, 477-491.Davis, J.H. (1983) Biochim. Biophys. Acta. 737, 117-171.Davis, P.J., Efrati, H., Razin, S. and Rottem, S. (1984) FEBS Lett. 175, 51-54.deKruijff, B., Demel, R.A. and van Deenen, L.L.M.(1972) Biochim. Biophys.Acta. 255, 331-347.deKruijff, B., van Dijck, P.W.M., Goldback, R.W., Demel, R.A. and van Deenen,L.L.M. (1973) Biochim. Biophys. Acta. 330, 269-282.deKruijff, B., Cullis, P.R., Radda, G.K. and Richards, R.E. (1976) Biochim.Biophys. Acta. 419, 411-424.DeYoung, L. and Dill, K.A. (1988) Biochemistry 27, 5281-5289.Dufourc, E.J. (1983) Ph. D. Thesis, University of Ottawa, Ottawa, Canada.Bibliography^ 111Dufourc, E.J. and Smith I.C.P. (1986) Chem. Phys. Lipids 41, 123-135.Epand, R.M. (1990) Biochem. Cell. Biol. 68, 17-23.Eriksson, P.-D., Rilfors, L., Wieslander, A., Lundberg, A. and Lindblom, G.(1991) Biochemistry 30, 4916-4924.Evans, E. and Needham, D. (1987) J. Phys. Chem. 91 4219-4228.Evans, E. and Rawicz, W. (1990) Phys. Rev. Lett. 64 2094-2097.Fenske, D.B. and Jarrell, H.C. (1991) Biophys. J. 59, 55-69.Gally, H.U., Pluschke, G., Overath, P. and Seelig, J. (1979) J. Am. Chem. Soc.18, 5605-5610.Gally, H.U., Pluschke, G., Overath, P. and Seelig, J. (1981) Biochemistry 20,1826-1831.Gorter, E. and Grendel F. (1925) J. Exptl. Med. 41, 439-443.Gruner, S.M., Cullis, P.R., Hope, M.J. and Tilcock, C.P.S. (1985) A. Rev. Bio-phys. Biophys. Chem. 14, 211-238.Gurr, M.I. and Harwood, J.L. (eds.) (1991) Lipid Biochemistry fourth edition,Chapman and Hall, London.Helfrich, W. and Servuss, R.-M. (1984) Nuovo Cim. D 3, 137-151.Hsiao, C.Y.Y., Ottaway, C.A. and Wetlaufer, D.B. (1974) Lipids 9, 913-915.Huang, T.-H., DeSiervo, A.J., Homola, A.D. and Yang, Q.-Y. (1991) Biophys. J.59, 691-702.Bibliography^ 112Huschilt J.C., Hodges, R.S., and Davis, J.H. (1985) Biochemistry 24, 1377,1386.Ipsen, J.H., Mouritsen, O.G. and Bloom, M. (1990) Biophys. J. 57, 407-412.Israelachvili, J.N., Marcelja, S. and Horn, R.G. (1980) Quarterly Reviews of Bio-physics 13, 121-200.Jarrell, H.C., Butler, K.W., Byrd, R.A., Deslauriers, R., Ekiel, I. and Smith,I.C.P. (1982) Biochim. Biophys. Acta. 688, 191-200.Jarrell, H.C., Tulloch, A.P. and Smith, I.C.P. (1990) Biochemistry 22, 5611-5619.Jarrell, H.C., Jovall, P.A., Giziewicz, J.B., Turner, L.A. and Smith, I.C.P. (1987)Biochemistry, 26, 1805-1811.Johnson, E.J., Mahlberg, F.H., Rothblat, G.H. and Phillips, M.C. (1991) Biochim.Biophys. Acta. 1085, 273-298.Kang, S.Y., Kinsey, R.A., Rajan, S., Gutowsky, H.S., Gabridge, M.G. and Old-field, E. (1981) J. Biol. Chem. 255, 1155.Kelusky, E.C., Dufourc, E.J. and Smith, I.C.P. (1983) Biochim. Biophys. Acta.735, 302-304.Keough, K.M.W. (1984) Biomembranes 12, 55-88.Koblin, D.D. and Wang, H.H. (1981) Biochim. Biophys. Acta. 649, 717-725.Koenig, S.H., Ahkong, Q.F. Brown, R.D. III, Lafleur M., Spiller, M. , Unger, E.and Tilcock, C. (1992) Magnetic Resonance in Medicine 23, 275-286.Knowles, D. (1992) Ph.D. Thesis, University of British Columbia, Vancouver,B.C.Bibliography^ 113Lafleur, M., Fine, B., Sternin, E., Cullis, P.R. and Bloom, M. (1989) Biophys. J.,56, 1037-1041.Lafleur, M., Cullis, P.R. and Bloom, M. (1990a) Bur. Biophys. J., 19, 55-62.Lafleur, M., Bloom, M. and Cullis, P.R. (1990b) Biochem. Cell Biol. 68, 1-8.Lafleur, M., Cullis, P.R., Fine, B. and Bloom, M. (1990c) Biochemistry 29, 8325-8333.Lindblom, G., Persson, N.-O. and Arvidson, G. (1976) Adv. Chem. Ser. 152,121-141.Lindblom, G. and Wennerstrom, H. (1977) Biophys. Chem. 6, 167-171.MacDonald, P.M., Sykes, B.D. and McElhaney, R.N. (1984) Can. J. Biochem.Cell Biol. 62, 1134-1150.MacDonald, P.M., Sykes, B.D., McElhaney, R.N. and Gunstone, F.D. (1985a)Biochemistry 24, 177-184.MacDonald, P.M., Sykes, B.D. and McElhaney, R.N. (1985b) Biochemistry 24,2237-2245.Marcelja, S. (1974) Biochim. Biophys. Acta. 367, 165-176.Marsh, Derek (ed.) (1990) Handbook of Lipid Bilayers, CRC Press Inc., BocaRaton, Florida.McElhaney, R.N. and Tourtellotte, M.E. (1970) Biochim. Biophys. Acta. 202,120-178.McElhaney, R.N. (1974) PAABS REVISTA 3, 753-767.Bibliography^ 114McElhaney, R.N. (1982) Chem. Phys. Lipids 30, 229-259.McElhaney, R.N. (1984) Biochim. Biophys. Acta. 779, 1-42.McElhaney, R.N. (1986) Biochem. Cell. Biol. 64, 58-65.McElhaney, R.N. (1989) CRC Crit. Rev. Microbiol. 17 1-32.McElhaney, R.N., DeGier, J. and van Deenen, L.L.M. (1970) Biochim. Biophys.Acta. 219, 245-247.McElhaney, R.N., DeGier, J. and Van Der Neut-Kok, E.C.M. (1973) Biochim.Biophys. Acta. 298, 500-512.Monck, M.A., Bloom, M., Lafleur, M., Lewis, R.N.A.H., McElhaney, R. N. andCullis, P.R. (1992) Biochemistry 31, 10037-10043.Monck, M.A., Bloom, M., Lafleur, M., Lewis, R.N.A.H., McElhaney, R. N. andCullis, P.R. (1993) Biochemistry 32, 3081-3088.Morrison, C. and Bloom, M. (1993) J. Magn. Res. (in press).Mouritsen, O.G. and Bloom, M. (1984) Biophys. J. 46 141-153.Mouritsen, O.G. and Sperotto, M. (1992) Thermodynamics of Lipid-Protein In-teractions in Lipid Membranes: The Hydrophobic Matching Condition, In: Ther-modynamics of Surface Cell Receptors (Jackson, M., ed.) pp 127-181, CRC PressInc., Boca Raton, FloridaNes, W.R. and McKean, M.L. (1977) Biochemistry of Steroids and Other Isopen-tenoids, University Park Press, Baltimore, Maryland.Nes, W.R. and Nes, W.D. (1980) Lipids in Evolution, Plenum Press, New York.Bibliography^ 115Needham, D. and Evans, E. (1988) Biochemistry, 27, 8261-8269.Nezil, F.A. and Bloom, M. (1992) Biophys. J. 61:1176-1183.Oldfield, E., Chapman, D. and Derbyshire W. (1971) FEBS Lett. 16, 102-104.Oldfield, E., Chapman, D. and Derbyshire W. (1972) Chem. Phys. Lipids 9,69-81.Oldfield, E., Meadows, M. and Glaser, M. (1976) J. Biol. Chem. 251, 6147-6149.Pauls, K.P., MacKay, A.L. and Bloom, M. (1983) Biochemistry 22, 6101-6109.Perly, B., Smith, I.C.P. and Jarrell, H.C. (1978) Biochemistry 17, 2727-2740.Philips, M.C., Johnson, W.J. and Rothblat, G.H. (1987) Biochim. Biophys.Acta., 906, 223-276.Rance, M. (1980) Ph.D. Thesis, University of Guelph, Guelph, Ontario, Canada.Rance, M., Jeffrey, K.R., Tulloch, A.P., Butler, K.W. and Smith, I. C.P. (1982)Biochim. Biophys. Acta. 688, 191-200.Rance, M. and Byrd, R.A., (1983) Obtaining high-fidelity spin-1/2 powder spec-tra in anisotropic media: phase-cycled Hahn echo spectroscopy. J. Magn. Reson.,52 221-240.Razin, S. (1982) in Organization of Prokaryotic Cell Membranes (Ghosh, B. K.,Ed.) CRC Press, Boca Raton, Fl.Read, B.D. and McElhaney, R.N. (1975) J. Bacteriol. 123, 47-55.Reif, F. (1965) Fundamentals of Statistical and Thermal Physics, McGraw-Hill,New York.Bibliography^ 116Rilfors, L., Lindblom, G., Wieslander, A. and Christiansson, A. (1984) LipidBilayer Stability in Biological Membranes In Membrane Fluidity, eds. MorrisKates and Lionel Manson, 205-245, Plenum Publishing Corporation.Rilfors, L., Wikander, G. and Wieslander, A. (1987) J. Bacteriol. 169, 830-838.Rohmer, M., Bouvier, P. and Ourisson, G. (1979) Proc. Natl. Acad. Sci. U.S.A.76, 847-851.Rommel, E., Noack, F., Meier, P. and Kothe, G. (1988) J. Phys. Chem., 92,2981-2987.Rose, M.E. (1957), Elementary Theory of Angular Momentum Wiley and SonsInc., New York.Rottem, S. (1979) in The Mycoplasmas (Barile, M.F. and Razin, S., Eds.) Vol.1, Academic Press, New York.Saito, Y., Silvius, J.R. and McElhaney, R.N. (1977) Arch. Biochem. Biophys.182, 443-454.Seelig, J. and Seelig, A. (1974) Biochemistry 13, 4839-4845.Seelig, J. and Seelig, A. (1977) Biochemistry 16, 45-50.Seelig, J. (1977) Q. Rev. Biophys., 10, 353-418.Seelig, J. and Seelig, A. (1980) Q. Rev. Biophys. 13, 19-61.Seelig, J. and Waespe-Sarcevic, N. (1978) Biochemistry 17, 3310-3315.Siegel, D.P., Burns, J.L., Chestnut, M.H. and Talmon, Y. (1989) Biophys. J. 56,161-169.Bibliography^ 117Silvius, J.R. and McElhaney, R.N. (1978) Can. J. Biochem. 56, 462-469.Silvius, J.R. (1979) Ph.D. Thesis, University of Alberta, Edmonton Alberta,Canada.Silvius, J.R., Mak, Nanette and McElhaney, R.N. (1980) Biochim. Biophys. Acta.597, 199-215.Singer, S.J. and Nicolson, G.L. (1972) The fluid mosaic model of cell membranes.Science, Wash., 175, 720-731.Sternin, E. (1985) Data acquisition and processing: a systems approach Rev. Sci.Instrum., 56, 274-282.Sternin, E., MacKay, A.L. and Bloom, M. (1983) J. Magn. Reson. 55, 274-282.Sternin, E., Fine, B., Bloom, M., Tilcock, C.P.S., Wong, K.F. and Cullis, P.R.(1988) Biophys. J. 54, 689-694.Stockton, G.W., Johnson, K.G., Butler, K.W., Polnaszek, D.F., Cyr, R. andSmith, I.C.P. (1975) Biochim et Biophys. Acta. 401, 535-539.Stockton, G.W., Johnson, K.G., Butler, K.W., Tulloch, A.P., Boulanger, Y.,Smith, I.C.P., Davis, J.H. and Bloom, M. (1977) Nature 269, 267-268.Stockton, G.W. and Smith, I.C.P. (1976) Chem. Phys. Lipids 12, 251-263.Stohrer, J., Graner, G., Reimer, D., Weisz, K., Mayer, C., and Kothe, G. (1991)Collective lipid motions in bilayer membranes studied by transverse deuteron spinrelaxation. J. Chem. Phys. 95, 672-678.Thewalt, J., Kitson, N., Araujo, C., MacKay, A.L. and Bloom, M. (1992) Biochem.Biophys. Res. Commun. 188, 1247-1252.Bibliography^ 118Thurmond, R.L., Dodd, S.W. and Brown, M.F. (1991) Biophys. J. 59, 108-113.Valic, M.I., Gorrisen, H., Cushley, R.J. and Bloom, M. (1979) Biochemistry 18,854-859.Vance, D.E. and Vance, J.E. (eds.) (1984) Biochemistry of Lipids and Mem-branes, The Benjamin/Cummings Publishing Company, Inc., Menlo Park, Cali-fornia. 25-72.van der Neut-kok, E.C.M., De Gier, J., Middelbeek, E.J. and van Deenen, L.L.M.(1974) Biochim. Biophys. Acta. 332, 97-103.Vist, M.R. and Davis, J.H. (1990) Biochemistry 29, 451- 464.Watnick, P.A., Dea, P. and Chan, S.J. (1990) P.N.A.S., 87, 2082-2086.Watson, D. (1960) Clin. Chim. Acta. 5, 637-643.Wieslander, A. and Selstam, E. (1987) Biochim. Biophys. Acta. 901, 250-254.Wieslander, A., Christiansonn, A., Rilfors, L. and Lindblom, G. (1980) Biochem-istry 19, 3650-3655.Appendix ATable A.1: Values of A,(L)f , 13 (n ) and C(n ) for POPC d31 and POPC/POPEd31 fitsPOPCd31 s -1 POPC/POPEd31 s -1Carbon A fl2)f B (n) C(n) A(j2)f B(n) B (n)Number (n)C15 61.6 324.3 427.3 71.4 455.1 450.4± 25.6 + 122.6 + 31.6 + 36.5 + 174.5 + 44.6C14 120.1 376.3 461.3 155.5 262.8 514.5± 25.0 + 119.7 + 30.8 + 51.2 ± 24.4 + 62.4C13 160.5 301.6 513.7 195.6 323.9 575.6±25.2 ± 120.7 ± 31.1 ± 51.9 ± 247.8 ± 63.3C12 164.1 364.4 588.0 206.5 304.2 670.0± 19.8 + 95.0 ± 24.4 ± 38.1 +18.2 ± 46.5C11 230.6 127.5 651.1 224.8 383.6 723.3+ 20.7 + 99.0 ± 25.5 + 32.4 ± 154.6 ± 39.5C10 184.3 550.8 698.5 259.3 653.0 824.5+ 85.4 + 409.2 ± 105.3 ± 102.6 + 489.7 ± 125.2C9 269.1 522.5 652.7 272.2 885.1 778.2± 60.6 ± 290.7 + 74.7 ± 87.7 ± 418.8 + 107.0P1 289.4 565.9 863.2 318.2 710.3 978.6± 25.5 ± 122.0 + 31.4 ± 73.7 ± 351.6 ± 89.9119Appendix BPhenomenological Theory of Transverse Relaxation in MembranesEffect of molecular motion on 2H NMR properties 1The NMR spectra of fluid bilayer membranes exhibit axial symmetry with respectto the bilayer normal 71. We mean by this that molecular motions are axially symmetricwith respect to ii so that, for example, in the presence of an external magnetic fieldH., the averaging of the quadrupolar interaction of a spin-1 nucleus by the molecularmotions that are fast on the "NMR time scale" gives rise to a quadrupolar splitting 2wwhich depends only on the angle 0 between H. and 77. More precisely, for a deuteron( 2H) replacing a H on a C-H bond and approximating the electric field gradient (efg) asan axially symmetric tensor about the C- 2 H bond characterized by an angular frequencywQ P-- 27r x 1.25 x 10 5 s -1 (Davis, 1983; Seelig, 1977) , one can express the splittingparameter in the formco = wQP2(cos0)Scp (B.1)where P2 (,u) = (3u 2 — 1)/2 is a Legendre polynomial, and the orientational order pa-rameter Sap = S20 = < P2(cos/3) 5- fastmotions is a measure of the averaging of thequadrupolar interaction due to modulation of the angle 0 between the C- 2H bonddirection and it' by fast motions.The motions that give rise to this motional averaging are also responsible for lon-gitudinal (spin-lattice) and transverse relaxation, commonly characterized by the time1 This appendix was a contribution from Dr. Myer Bloom120Appendix B. Phenomenological Theory of Transverse Relaxation in Membranes 121constants T 1 and T2 respectively. The relaxation rates may be conveniently expressedin terms of the spectral densities of the correlations functions of the fluctuating spin-dependent interactions. The correlation functions are characterized in turn by a spec-trum of correlation times 7-. (Davis, 1983; Abragam, 1961). The relevant spectraldensities contributing to T i-1 are evaluated at w o and 2wo , where w o is the angularLarmor frequency in the field H., while T2 1 has contributions from spectral densitiesevaluated at zero frequency, in addition to those evaluated at w o and 2wo . From adefinitive study of the dependence of T i-1 on wo in DMPC extending to very low mag-netic fields (Rommel et al., 1988), it appears that the correlation times fall into twoclasses, those associated with relatively fast motions having values Tc < TIP-s.', 10 -8s thatcontribute to the field dependence of T1 1 only at high fields, i.e . corresponding to wo> 27r x 10 7s -1 , and those associated with relatively slow motions having values of Tc >Tcs ''.'"', 10 -6s that contribute appreciably to T i-1 only at very low magnetic fields. Theslow motions would be expected to contribute to T2 1 , but not to T i-1 under the con-ditions of the experiments reported in this paper . We shall make use of this empiricaldemonstration of the clumping of the correlation times into two categories, long andshort, to develop a tentative model for transverse relaxation in terms of two distinctcontributions as follows:—1 = Rs + R2 (B.2)T2 2where 11,12 may be inferred from T 1 measurements, and especially from the depen-dence of T 1 on wo . It is assumed to have its origin in motions for which Tc < 7-1. In thesame spirit a theory for 14 may be constructed using approximations appropriate forrelatively slow motions. Such a theory is presented in the following paragraphs wherewe characterize .IT; as the adiabatic contribution to the transverse relaxation rate.Appendix B. Phenomenological Theory of Transverse Relaxation in Membranes 122One of the features of transverse relaxation of special interest is its orientation de-pendence. A general form for the orientation dependence of spin-lattice and transverserelaxation rates for spin-1 systems, valid within the Redfield approximation, has beenderived recently (Morrison and Bloom, 1993). For axial symmetry, this takes the form1^A^A Di^n\^A ni^n\y. -,. = 11-2i/-2 COSu) ii4iFACOSu) (B.3)for Ti = T1z , T i , or T2, where the coefficients A 01, A2, and A4, are dependent on co., cor-relation times, the orientational order parameters S20 and S40 and the thermodynamicvariables required to characterize the system.Adiabatic Contributions to Transverse RelaxationWe suppose that the fast molecular motions discussed above establish a quadrupolarsplitting 2w given by equation B.1 in terms of Sc.!) and 0. The slower motions modulatew adiabatically, where the word adiabatic is used in the same sense as in pages 427ff ofthe "bible" of Abragam(1961). We assume that the adiabatic motions may be dividedinto two classes, those that modulate 0 and SCD independently, with effective correlationtimes Teo and TcA, respectively. We associate these motions with:(i) fluctuations in local curvature that give rise to changes in 9 without any appre-ciable changes in bilayer thickness, and, consequently, without changes in SCD.(ii) fluctuations in bilayer thickness x or equivalently the membrane area A, thatgive rise to fluctuations in SCD without any appreciable changes in 0.It is emphasized that the assumption of independence of these types of motionscannot be justified on theoretical grounds at the present time. It is likely that the ac-tual modes responsible for the adiabatic motions involve a coupling between membranecurvature and thickness fluctuations. The assumption of codependence should be con-sidered as a first approximation in the establishment of a rigorous theory of adiabaticAppendix B. Phenomenological Theory of Transverse Relaxation in Membranes 123motion.Explicit connections between x, A and Sap for acyl chains have been discussed byseveral authors (see, e.g. Ipsen et al., 1990). Thermodynamic considerations allowone to relate the mean squared fluctuations < (60) 2 > and < (SA) 2 > to measurableproperties of membranes. These two types of fluctuations are governed, under someconditions, by the so-called curvature energy K, (Bloom and Evans, 1991) and theisothermal area compressibility (1/K a ) (Bloom et al., 1991; Lipowsky, 1991), respec-tively, as follows:< (80)2 >,_< 92 > < >2_ kBT Am4rn,< (86) 2 >,-‹ A2 > < A > 2 = BTKa(B.4)(B.5)where Am and A, in B.4 are the maximum and minimum wavelengths for curvaturefluctuation modes, respectively, and K a is defined in terms of the fractional variationof area with surface tension u as 1/Ka = (1/A)(OA/Oa)T. Expressions analogous toequation B.5 can be found in many texts (see, eg. Reif, 1965; page 300) for volumefluctuations in three dimensional systems.It is easy to show that the thermal fluctuations in 0 and A, given by equationsB.4 and B.5 respectively, give rise to renormalizations of the quadrupole splitting ofonly a few percent, which may be ignored for present purposes. For example curvaturefluctuations lead to a reduction in Scp by < P2 (cas60) > 1-3/2< (60) 2 >. From thevalues used in the paper by Bloom and Evans (1991) of n a P.,- 5 x 10'ergs and A m /Am20, equation B.4 gives < (60) 2 > 1.8 x 10 -2 at room temperature. Similarly, thetypical value of K a 200 dynes/cm for fluid membranes (Needham and Evans, 1988),gives kBT/Ka 2A 2 << < A > 2, also leading to small changes in the quadrupolesplittings for practical cases.Appendix B. Phenomenological Theory of Transverse Relaxation in Membranes 124If these motions give rise to uncorrelated fluctuations in the quadrupolar splittingthey result in adiabatic relaxation given by the classic formula (Abragam, 1961) forthe transverse relaxation rate arising from a fluctuating interaction characterized by asecond moment AM 2i and correlation time Tci satisfying the condition AM2i7- << 1,R;^E R2 i = E Am20-ci^ (B.6)(8466) A2 < (89)2 > Tc9 + (sco/aA),2 < (SA)2 > TcA^ (B.7)9^•= c,), 2 [-iSLjszn2 Ocos 2 < (69) 2 > 7,9 + [P2(cos9)] 2 [(8ScD/SA)B] 2 < (811) 2 > TcAlBsin2 8cos 20 C[P2 (cos0)] 2^(B.8)where the coefficients B and C are defined by Eqs. B.5 and B.8. Returning toEq. B.2, identifying R2 with i =2 and the coefficients A02 , A22 , A42 in Eq. B.3, andexpressing sin20cos 2 0 and [P 2 (cos9)] 2 in terms of the Legendre polynomials, we canwritewhere1^A— /102 A22P2(cosO) A42P4(cosO)T2A02 = + 1E2 B CA22 =^,7+ B + 2 CA42 = 4,2 — LB+ 18T5c(B.9) Theoretical Predictions for the B and C coefficients in adiabatic Transverse Relaxation: Appendix B. Phenomenological Theory of Transverse Relaxation in Membranes 125A theory for contributions of thermally induced curvature fluctuations to the trans-verse relaxation rate (i.e. the B coefficient) has been published (Bloom and Evans,1991). In this theory, the curvature fluctuations are expressed in terms of a spectrumof curvature modes of wavelengths A q having correlation times Tq and B is expressedas a product of terms of the form < (8190 2 > Tcq . Furthermore Tcq may be expressed interms of < (890 2 > Tcq with the help of Eqs. B.4 and B.8. The major uncertainty isin the damping mechanism. While < (8 9 q )2 > can be obtained reliably from thermo-dynamic considerations and the equipartition theorem, Tcq should depend sensitivelyon factors such as the degree of hydration, tension and the multi-layer character of themembranes, that, in turn, depend on the method of preparation and thermal history ofthe membranes. Within the range of uncertainty of these factors, it was concluded thatthe predicted values of T2 due to this mechanism could easily fall in the observed rangeof approximately 100 to 1000 ,us typically found for 2 11 NMR in membranes (Bloomand Evans, 1991).We should further emphasize that a second mechanism, associated with order-director fluctuations (Stohrer et al., 1991), has been proposed for a sin 29cos 2 0 contribu-tion to R. Experiment must ultimately distinguish between these different possibilities.It may well be that both contribute to R,,s2 , with curvature fluctuations possibly being ofgreater importance for spins located in the head-group and order-director fluctuationsfor spins on the chains.We are unaware of any published theory for the contributions of area (or equivalentlythickness) fluctuations to 112. Again one can construct a theory based on sinusoidalmodes of area fluctuations which take into account an empirical or theoretical con-nection between Sap and membrane area A. In order to get a feeling for the order ofmagnitude anticipated for the C term in N, we look at the predicted value of C fordeuterons on acyl chains for which we have found ( to our surprise) that an area modeAppendix B. Phenomenological Theory of Transverse Relaxation in Membranes 126characterized by < A > gives a contribution to R; that is independent of < A >.In order to calculate C in Eq B.8, we first model the hydrophobic region of afluid membrane as an incompressible fluid and make use of the empirical relationshipbetween bilayer thickness x, its maximum value in the fluid phase xm (or, alternatively,the corresponding membrane area A =< A > with its minimum value A m ) and theaverage chain order parameter S (Ipsen et al., 1990)x= ^aS -I- 13, with 0.5a + /3 =1A xm(B.13)We then note that the order parameter profile corresponding to the variation ofScp with chain position has been found to have a universal form for a given value ofS (Lafleur et al., 1990) so that one can express aScp/aA in Eq. B.8, for a given S, interms of the experimentally measurable coefficients ECD = aSaDiaS as follows:aScD Ecn as^r, AmaA = '-ECD^ (B.14)From the definition of C in Eq. B.8, making use of Eq. B.5 and assuming thatfluctuations in area are relaxed by spatial diffusion so that TcA pA/D, where D isthe translational diffusion coefficient for diffusive motions parallel to the plane of themembrane and p is a dimensionless coefficient of order unity, we writekB TC = pELL0,2Q(S a)2 DICa(B.15)For the typical values of WQ = 2ir x 1.25 x 105 s -1 , T = 300K ,Dc-se, 4 x 10 -8cm 2s -1and K a 200 dynes cm -1 , we obtain C ti 3000pE6(S+ /3/a) 2 indicating that C is ofthe same order of magnitude as B; thus both the area and curvature fluctuations arecapable of producing values of T2 as short as about 100 its.