UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Interactions and permeability properties of vesicles of thylakoid lipids Webb, Murray S 1989

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

Item Metadata


831-UBC_1989_A1 W42_5.pdf [ 8.25MB ]
JSON: 831-1.0098480.json
JSON-LD: 831-1.0098480-ld.json
RDF/XML (Pretty): 831-1.0098480-rdf.xml
RDF/JSON: 831-1.0098480-rdf.json
Turtle: 831-1.0098480-turtle.txt
N-Triples: 831-1.0098480-rdf-ntriples.txt
Original Record: 831-1.0098480-source.json
Full Text

Full Text

INTERACTIONS AND PERMEABILITY PROPERTIES OF VESICLES OF THYLAKOID LIPIDS by MURRAY S. WEBB B.Sc, University of British Columbia, 1980 M.Sc, University of Toronto, 1983 A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE STUDIES Department of Botany We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA 11 May, 1989 ® Murray S. Webb, 1989 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department The University of British Columbia Vancouver, Canada DE-6 (2/88) ABSTRACT The large-scale purification of the major spinach thylakoid lipids by a combination of silica and carboxymethyl-cellulose chromatography is described. Yields of hundreds of milligrams of the lipids, representing 25-40% of the original lipid, have been obtained. In addition, routine purities in excess of 99.7% of the isolated lipids has been demonstrated. The structures of the purified lipids have been confirmed by fatty acid analysis, thin layer chromatography, and "C-NMR. Some minor reassignments to previously published "C-NMR for these compounds are described. In addition, the 'H-NMR spectra for the glycolipids monogalactosyldiacylglycerol (MGDG), digalactosyldiacylglycerol (DGDG), and sulfoquinovosyldiacylglycerol (SQDG), are shown. The resonance assignments for MGDG and SQDG have been obtained by a combination of off-resonance decoupling experiments and by two-dimensional COSY 'H-NMR experiments. Similar experiments with DGDG have failed to resolve the proton assignments due to extensive overlapping of the proton resonances. Interbilayer interactions between large unilamellar vesicles of DGDG in aqueous salt solutions have been examined by light scattering, freeze-fracture electron microscopy, and X-ray diffraction. When suspended in aqueous salt solutions, vesicles of 100 nm diameter were found to aggregate in a rapid and reversible manner to yield aggregates greater than 1000 nm in diameter. Freeze-fracture electron microscopy showed these aggregates to consist of appressed, but not fused, vesicles. Quasi-elastic light scattering and turbidity experiments showed that aggregation was not due to charged impurities of the lipid behaving in accordance with electrostatic double layer theory. Experiments testing the efficacies of various chloride salts indicated a strong correlation existed between ionic radius and ability of the salt to promote aggregatioa Similar experiments examining the effect of sodium salts, glycerol, and pH on vesicle aggregation implicate an interaction between the DGDG head group and structured water as underlying the aggregation process. ii The effect of additions of other lipids on the extent of DGDG aggregation has been examined. Addition of 0.5 to 5.0% of either anionic lipid phosphatidylglycerol (PG) or SQDG inhibited the aggregation of DGDG vesicles, probably by the development of an electrostatic potential. Different effects of PG and SQDG on the concentration of Mg2* required for aggregation indicated that PG may form a bidentate ligand with Mg2* at ^ 5 mol% PG. SQDG did not show this behavior, indicating that its negatively charged sulfonate group is unavailable for cation complex formation. Addition of MGDG to DGDG up to 50 mol% had no effect on the Mg2* requirement for aggregation, but at ^ 25 mol% triggered irreversible vesicle aggregation. This suggests that the MGDG head group is as effective at causing aggregation as the DGDG head group. Further, MGDG probably triggers vesicle fusion at i> 25 mol%. The results suggest that the galactolipids may contribute to the close approach of thylakoids in higher plant chloroplasts. The permeability properties of large unilamellar vesicles of DGDG to , 6Rb*. 36C1~, and JH-glucose have been determined. In addition, the permeabilities of binary, ternary, and quaternary mixtures of thylakoid lipids to , 6Rb* have also been measured. Vesicles of DGDG were found to be 60-130 fold more permeable to Rb* and 46-76 fold less permeable to CI" than phosphatidylcholine vesicles. Vesicles of DGDG and PC were similar in glucose permeability. Electron spin resonance measurement of DGDG bilayer fluidity indicated that fluidity differences could not account for the observed differences in ion permeability. The addition of 50 mol% of MGDG to DGDG vesicles had no effect on Rb* permeability, suggesting that the phase preference of MGDG does not increase bilayer permeability. The addition of SQDG led to a large increase of Rb* permeability. The calculated permeability coefficient to Rb* for a DGDG/MDGD/SQDG/PG (1/2/0.5/0.5) mixture similar to that of thylakoid membranes was 2.0-10"' cm-s1. This value is three orders of magnitude higher than that for phospholipid systems, and ten-fold higher than that for vesicles of pure DGDG. It is concluded that the permeability properties of thylakoid lipids are dominated by oriented surface dipoles and not by bilayer fluidity or acyl chain packing considerations. iii It is also proposed that the high permeability of thylakoid lipids to cations is the main cause of low observed thylakoid membrane electrical potentials, and large proton gradients across thylakoid membranes. It has been proposed previously that the high proportions of saturated phosphatidylglycerols (ie. DPPG) found in chilling-sensitive plants may promote the formation of gel phase lipid, and cause increased metabolite leakage, in the thylakoids of these species at chilling temperatures. The leakage of "Rb* from large unilamellar vesicles of thylakoid lipids containing proportions of disaturated PG (as DPPG) mimicking those of chilling-sensitive and chilling-resistant plants has been measured. This data indicated that no increase in Rb* permeation occurred between any of the tested vesicles systems between 7° and 30° C. Differential scanning calorimetry showed no heat flow indicative of gel to liquid- crystalline phase separation in any of the lipid mixtures, even with DPPG levels as high as 12 mol%. It is concluded that a direct effect of disaturated PG on chilling injury in sensitive plants by an increase of low-temperature thylakoid permeability is unlikely. iv TABLE OF CONTENTS ABSTRACT ii LIST OF FIGURES viii LIST OF TABLES ix LIST OF ABBREVIATIONS x ACKNOWLEDGEMENTS xi 1. GENERAL INTRODUCTION 1 1. Introduction 1 1.1. Light-harvesting and photosynthetic electron transport 1 1.2. Thylakoid Ultrastructure 4 2. Structure and Distribution of Thylakoid Lipids 6 2.1. Chemical Structures 6 2.2. Taxonomic Distributions 10 2.3. Cellular and sub-cellular distribution 11 3. Phase Behavior of Thylakoid Lipids 13 3.1. MGDG 14 3.2. DGDG 15 3.3. PG and SQDG 15 3.4. Binary, and higher, mixtures 16 3.5. The shape concept of lipid phase behavior 17 3.6. Bilayer fluidity 18 4. Lipid-protein interactions in thylakoids 22 4.1. MGDG 22 4.2. DGDG and phospholipids 24 4.3. SQDG 25 4.4. PG 25 5. Inter-bilayer Interactions 28 6. Statement of Objectives 29 2. PURIFICATION AND NMR SPECTROSCOPY OF THYLAKOID GLYCOLIPIDS 31 1. Introduction 31 2. Materials and Methods 33 2.1. Materials 33 2.2. Lipid extraction and purification 33 2.3. Lipid and fatty acid analysis 36 2.4. Preparation of hydrogenated lipids 37 2.5. 1 3 C - N M R spectroscopy 37 2.6. ' H - N M R spectroscopy 38 3. Results and Discussion 39 3.1. Yield and purity of extracted lipids 39 3.2. 1 3 C - N M R of thylakoid glycolipids 43 3.3. l H - N M R of thylakoid glycolipids 49 v 3. SALT-MEDIATED INTERACTIONS BETWEEN VESICLES OF DGDG 62 1. Introduction 62 2. Materials and Methods 63 2.1. Materials 63 2.2. Lipid purification 63 2.3. Vesicle reconstitution 63 2.4. Turbidity measurements 64 2.5. X-ray diffraction 64 2.6. Quasi-elastic light scattering 65 2.7. Freeze-fracture electron microscopy 65 3. Results 66 3.1. Aggregation of vesicles by salts 66 3.2. Effects of different salts on aggregation 72 4. Discussion 81 4. EFFECTS OF LIPID ADDITIONS AND HEAD GROUP MODIFICATIONS ON VESICLE AGGREGATION 87 1. Introduction 87 2. Materials and Methods 88 2.1. Materials 88 2.2. Vesicle reconstitution 88 2.3. Measurement of vesicle sizes and A A 5 0 values 89 2.4. NICOMP 270 standard curve 89 2.5. Enzymatic digestion of DGDG vesicles 89 2.6. Lipid analysis 90 3. Results 91 3.1. Fatty acid and lipid compositions of vesicles 91 3.2. Effects of other lipids on aggregation 97 3.3. Effects of cation mixtures on A A 5 0 values 100 3.4. Effect of head group and fatty acid modifications 100 3.5. Effect of vesicle diameter on aggregation 102 3.6. Reversibility of vesicle aggregation 103 4. Discussion 107 5. PERMEABILITY PROPERTIES OF LARGE UNILAMELLAR VESICLES OF THYLAKOID LIPIDS 113 1. Introduction 113 2. Materials and Methods 114 2.1. Materials 114 2.2. Lipid purification and analysis 114 2.3. Vesicle reconstitution 114 2.4. Rb* flux 115 2.5. CI" flux 116 2.6. Glucose flux 116 2.7. Calculation of permeability coefficients 116 2.8. Election Spin Resonance 117 3. Results 119 3.1. Lipid and fatty acid compositions of vesicles 119 3.2. Efflux of solutes from DGDG and PC vesicles 119 3.3. DGDG vesicle fluidity 125 3.4. Effects of other lipids on Rb*permeability 126 4. Discussion 129 vi 6. EFFECT OF TEMPERATURE ON PHASE BEHAVIOR AND PERMEABILITY OF THYLAKOID LIPIDS 135 1. Introduction 135 2. Materials and Methods 137 2.1. Materials 137 2.2. Lipid purification 137 2.3. Vesicle reconstitution and efflux measurement 137 2.4. Flux calculation 138 2.5. Differential Scanning Calorimetry 138 3. Results 139 3.1. Characteristics of dispersed vesicle 139 3.2. Phase behavior of lipid vesicles 142 3.3. Temperature dependence of vesicle permeability 147 4. Discussion 151 7. GENERAL DISCUSSION 154 1. The importance of well-characterized systems 154 2. Surface properties and vesicle interactions 157 3. A possible mechanism of galactolipid vesicle aggregation 160 4. The role of galactolipids in thylakoid stacking 161 5. Permeability of vesicles and thylakoids 164 6. Chilling stress and membrane permeability 167 8. REFERENCES 171 vii LIST OF FIGURES Figure 1. Schematic representation of higher plant thylakoids 6 Figure 2. Chemical structures of MGDG, DGDG, SQDG, and PG 7 Figure 3. U C - N M R spectrum of DGDG 44 Figure 4. " C - N M R spectrum of MGDG 46 Figure 5. " C - N M R spectrum of SQDG 47 Figure 6. J-resolved 1 } C - N M R spectrum of SQDG 48 Figure 7. ' H - N M R spectrum of MGDG 50 Figure 8. COSY l H - N M R spectrum of MGDG 51 Figure 9. Expanded COSY ^ - N M R spectrum of MGDG 52 Figure 10. ' H - N M R spectrum of SQDG 54 Figure 11. COSY ' H - N M R spectrum of SQDG 55 Figure 12. Expanded COSY J H - N M R spectrum of SQDG 56 Figure 13. ' H - NMR spectrum of DGDG 59 Figure 14. COSY ' H - N M R spectrum of DGDG 60 Figure 15. Expanded COSY 'H-NMR spectrum of DGDG 61 Figure 16. Freeze-fracture electron microscopy of DGDG vesicles 68 Figure 17. Turbidity standard curve 73 Figure 18. Effect of cations on DGDG aggregation 74 Figure 19. Correlation between cation radius and A A 3 0 76 Figure 20. Effect of anions on DGDG aggregation 77 Figure 21. Effect of glycerol on DGDG aggregation 78 Figure 22. Effect of pH on DGDG aggregation 79 Figure 23. Effect of PG, SQDG, and MGDG on DGDG aggregation 94 Figure 24. Effect of lipid additions on DGDG aggregation 95 Figure 25. Aggregation of DGDG, 18:2-DGDG, and DglcDG 101 Figure 26. Standard curve for NICOMP 270 104 Figure 27. Reversibility of aggregation of DGDG/MGDG vesicles 106 Figure 28. Efflux of "Rb* from DGDG and PC vesicles 121 Figure 29. Efflux of "CI- from DGDG and PC vesicles 122 Figure 30. Efflux of 3H-glucose from DGDG and PC vesicles 123 Figure 31. Differential scanning calorimetry of 0% DPPG vesicles 143 Figure 32. Differential scanning calorimetry of 4% DPPG vesicles 144 Figure 33. Differential scanning calorimetry of 8% DPPG vesicles 145 Figure 34. Differential scanning calorimetry of 12% DPPG vesicles 146 Figure 35. Efflux of , 6 Rb + vs. temperature for vesicles 148 Figure 36. Arrhenius plot of , 6Rb* efflux from vesicles 149 viii LIST OF TABLES Table 1. Yield and composition of spinach lipid extracts 40 Table 2. Fatty acid composition of purified thylakoid lipids 41 Table 3. Purity of isolated thylakoid lipids 42 Table 4. QELS measurements of DGDG vesicle diameters in salts 67 Table 5. QELS measurements of vesicle diameters from lipid mixtures 70 Table 6. X-ray diffraction of DGDG lamellar repeat distances 71 Table 7. Fatty acid composition of PG, DGDG, and 18:2-DGDG 92 Table 8. Recovery of lipids from vesicles 93 Table 9. Size and aggregation characteristics of lipid mixtures 96 Table 10. Effect of KC1 on vesicle aggregation in MgCl2 99 Table 11. Recovery of lipids from vesicles 120 Table 12. Fluidity and permeability characteristics of DGDG and PC 124 Table 13. Calculated Rb* permeability coefficients for lipid mixtures 127 Table 14. Recovery of lipids from vesicles 140 Table 15. Size and volume characteristics of vesicles 141 ix LIST OF ABBREVIATIONS PSII Photosystem II PSI Photosystem I LHCII Light Harvesting Complex of PSII LHCII* Oligomer of LHCII LUV Large unilamellar vesicle MLV Multi-lamellar vesicle SUV Small unilamellar vesicle palmitic acid (16:0) palmitoleic acid (16:1) palmitolinolenic acid (16:3) stearic acid (18:0) oleic acid (18:1) linoleic acid (18:2) linolenic acid (18:3) hexadecanoic acid trans-A3-hexadecenoic acid hexadecatrienoic acid (A7, 10, 13) octadecanoic acid octadecenoic acid (A9) octadecadienoic acid (A9, 12) octadecatrienoic acid (A9, 12, 15) M G D G 1,2- diacyl- 3 - / 3 - D - galactopyranosyl- sn- glycerol D G D G 1,2- diacyl- 3 - (a - D - galactopyranosyl- ( 1 - 6 > (S-D- galactopyranosyl)- sn- glycerol SQDG 1,2- diacyl- 3 - (6'- sulpho- a-D- quinovopyranosyl)- sn- glycerol PG 1,2- diacyl- sn- glycero- 3 - phosphoryl-1'- sn- glycerol PC 1,2- diacyl- sn- glycero- 3 - phosphorylcholine PE 1,2- diacyl- sn- glycero- 3 - phosphorylethanolamine PS . l,2-diacyl-sn-glycero-3-phosphorylserine x ACKNOWLEDGEMENTS I wish to begin by thanking my parents for their unqualified support and encouragement through this prolonged foray into graduate studies. I also wish to thank Dr. Beverley R. Green for supervising and supporting this project, as well as for numerous ideas and suggestions regarding its development and presentation. My thanks also to Dr. Edith Camm, Peter Sibbald, Gopal Subramaniam, and Michael White, for lab advice, humour, suggestions, and many interesting discussions. I am also grateful to Dr. P.R. Cullis for making available the extensive lab facilities used in many parts of this project Drs. Mick Hope and Kim Wong were very generous with their time in helping me with freeze-fracture and NMR work. Special thanks to Dr. Colin Tilcock for advice, input, and discussions on many parts of this project and for performing the X-ray diffraction experiments and much of the NMR work. I would also like to thank Dr. Don McRae for help with the ESR measurements and Dr. Dan Lynch for performing the DSC measurements. Finally, I would like to express my appreciation to all of those people who, over the years, have constituted the coffee and beer groups. These people include Rene Belland, Don Champagne, Kali Robson, Lacey Samuels, Frank Shaugnessey, Peter Sibbald, Paul Spencer, Gopal Subramaniam, Randy Thomson, and many others. To Paul Lazarski, my thanks for resurrecting a too-long-dormant fascination with the outdoors world. xi 1 1. GENERAL INTRODUCTION 1. INTRODUCTION Oxygen-evolving photosynthesis is a process common to all algae and higher plants. The importance of photosynthesis as the primary producer in the food chain and as a source of molecular oxygen is obvious. While photosynthesis has central roles in supplying reducing power for a wide variety of cellular metabolic processes, the primary role of photosynthetic electron transport is the generation of ATP and NADPH required for C02- fixation by ribulose 1,5-bisphosphate oxygenase-carboxylase and for N02 and sulfur fixation. Light harvesting and photosynthetic electron transport occur in a group of membranes known as the thylakoids. The thylakoids may be present as unattached membranes in the cytoplasm (in prokaryotic algae), or be localized within a double membrane envelope, the chloroplast, which occurs in algae and in all higher plants. Irrespective of their location, the thylakoids are the predominant membranes in plant cells, representing 60-80% of the total cellular membranes in higher plant mesophyll cells (Forde and Steer, 1976). It follows, then, that the protein and lipid components of the thylakoid are amongst the most abundant organic compounds in the natural world. The general intent of this thesis is to examine the role of the thylakoid lipids in photosynthetic processes. 1.1. Light-harvesting and Photosynthetic Electron Transport Before reviewing previous work on the subject of lipid involvement in photosynthesis, a brief review of the light-reactions of photosynthesis is presented A complete summary of light-harvesting and photosynthetic electron transport is beyond the scope of this Chapter. The 2 interested reader is directed to the comprehensive reviews of Crofts and Wraight (1983), Mathis (1986), Murphy (1986a,b), Ort (1986), Thornber (1986), Whitmarsh (1986), and Green (1988). The light reactions of photosynthesis are currently believed to follow the "Z-scheme" of photosynthetic electron transport The Z-scheme describes the energetics and pathway of electron flow from the primary electron donor, H 20, through photosystem II (PSII) and photosystem I (PSI) to the final electron acceptor NADP\ Electron flow is initiated with photon absorption by the light-harvesting complex (LHQI) attached to PSII. This complex represents approximately one-half of the total thylakoid protein and chlorophyll a, most of the chlorophyll b, and is composed of 2-4 polypeptides of 24-28 kD molecular weight Light energy is absorbed by LHQI, transmuted to exciton and vibrational energy, and rapidly transferred to the PSII core proteins DI and D2. Light absorption may also be accomplished by the so-called internal antenna of PSII: Chlorophyll-protein (CP) 47, CP43, and CP29 (Green, 1988). The reaction center chlorophyll(s) (P680) are probably a chlorophyll a dimer, possibly non-covalently bound between the DI and D2 polypeptides (Tang and Satoh, 1985; Danielus et al., 1987). The viability of the chlorina f2 mutants that lack LHCII suggests that CP47, CP43, and CP29 may also function as primary antenna for PSII. Excitation energy transfer to P680 generates nanosecond electron transfer through the primary electron acceptor, pheophytin, to a quinone acceptor, Q , and through to a second quinone acceptor Q (Diner, A B 1986). The Q acceptor functions as a 2-electron gate (Crofts and Wraight, 1983), it must be B reduced by two electrons and two protons to generate the quinol, Q H, , that passes hydrogen B to plastoquinone (PQ). The "electron hole", P680\ now present at the PSII reaction center is reduced by the splitting of 2H20 to yield 02, 4 protons (released inside the thylakoid lumen), and 4 electrons. This reaction occurs in the oxygen-evolving complex, a peripheral membrane protein 3 complex composed of three polypeptides of 33, 23, and 17 kD molecular weight and containing M n : + that may be involved directly in the H20-splitting process. This complex is located on the inner surface of the thylakoid lumen (Andersson, 1986). Electrons released by water splitting are transferred through intermediate Z to reduce P680\ Lateral electron transport between PSII and the cytochrome b 6 /f complex by PQH 2 may be the rate-limiting step of photosynthetic electron transport (Crofts and Wraight, 1983). Protonation of the quinone head group might lead to decreased head group polarity and confirmational changes that allow the entire molecule to bury in the hydrophobic acyl phase (Katsikas and Quinn, 1981, 1982), then diffuse to the b 6 / f complex. This complex consists of cytochrome b6, cytochrome f, and the Rieske Fe-S protein (Joliot and JolioL 1986). Cytochrome b 6 / f mediates both proton ejection from PQH 2 into the thylakoid lumen, possibly via a true Q-cycle, and the reduction of plastocyanin. This is a low molecular weight Cu-protein bound peripherally to the inside surface of the thylakoid lumen. Plastocyanin transfers electrons to PSI, possibly by binding to a non-core subunit of the complex (Haehnel, 1986). The PSI complex consists of a reaction center composed of two 82-83 kD polypeptides as well as several low molecular weight proteins and an antennal complex, LHCI, containing chlorophylls a and b and several polypeptides in the 22-25 kD range, based on SDS-PAGE (Wollman, 1986). Illumination of the PSI complex leads to excitation of the two reaction center chlorophyll a molecules that constitute P700 to generate P700\ The primary electron acceptors of PSI, F , F , and F , are probably iron-sulfur proteins (Goldbeck, 1988). These A B X pass electrons through ferredoxin to reduce N A D P via a flavoprotein, Fp. 4 Protons accumulated within the thylakoid lumen from water-splitting and electron-transport are used for ATP synthesis via the CFo-CFi ATPase at a HVATP stoichiometry of about 3 (McCarty and Nalin, 1986). The CF 0 -CFi complex is similar to the F 0 - F i complexes of many mitochondrial systems and appears to have an a3^3y8e stoichiometry. 1.2. Thylakoid Ultrastructure Higher plant thylakoid membranes are laterally segregated into two major functional domains; the appressed, or granal, membranes and the non-appressed, stromal, membranes (Staehelin, 1986). A wide range of electron microscope, fractionation, and genetic studies have shown the separation of chlorophyll-protein and electron transport complexes into the two domains (Anderson, 1981, 1982; Gerola, 1981; Staehelin, 1986). The appressed membranes are highly enriched in LHCH, PSII, CP47, CP43, CP29, and the oxygen-evolving complex. In contrast, the non-appressed membranes are enriched in PSI, LHCI, and the CFo-CFi complex. Some complexes appear to be equally distributed between both membrane fractions, particularly the cytochrome b 6 /f complex (Allred and Staehelin, 1986). In addition, there appears to be a reversibly phosphorylated sub-population of LHCII, the so-called mobile-LHC, which in the phosphorylated form is dissociated from PSII and is free to diffuse into the non-appressed membranes (Kyle et al., 1984; Torti et al., 1984) and facilitate energy transfer to PSI. The extent of phosphorylation of LHQI is regulated via the redox state of the PQ pool (Horton et al., 1981), this being a sensitive and direct indicator of the balance of electron flow through PSII and PSI (Bennett, 1983). The dissociation by phosphorylation of some LHQI from PSII cores would appear to generate a sub-population of PSII reaction centers, PSII0, which are also free to diffuse into the non-appressed membranes. 5 The lateral separation of PSII and PSI complexes imposes a requirement for a long-distance electron carrier. Calculations performed by Crofts and Wraight (1983) suggested that PQ diffusion would be too slow to support observed rates of electron transport but were based on an assumed diffusion coefficient for PQ of 10"' cnrV 1 . More recently, ubiquinone diffusion coefficients as high as 10" cm2-s~l have been measured in phospholipid vesicles and mitochondrial membranes (Fato et al.. 1985, 1986). Plastoquinone diffusion coefficients of 1.3-3.5-10"7 cm-s"1 have been reported in phospholipid vesicles (Blackwell et al., 1987). The influence of proteins on PQ diffusion has been estimated to reduce the diffusion coefficient by approximately ten-fold (Blackwell and Whitmarsh, 1989). It is possible that PQ diffusion is rapid enough for its consideration as the primary long-distance electron carrier of photosynthesis. Under lighting conditions favouring PSII activity the reduction of the PQ pool occurs due to faster electron flow into the pool from PSII rather than their faster removal by PSI. This leads to the activation of a membrane-bound kinase that phosphorylates LHCII at one or two surface-exposed sites. Phosphorylation represents charge injection into the appressed membrane regions and promotes thylakoid destacking by electrostatic repulsion, detachment of LHCII from PSII and its diffusion to the non-appressed membranes (Staehelin, 1986). This mobile- LHC is postulated to facilitate energy transfer to PSI, by increasing the cross-sectional area of PSI-associated antenna, and to balance the rates of electron flow through PSI and PSII. In the reverse situation of higher activity of PSI than PSII, a membrane associated phosphatase is expected to remove these phosphate groups and promote mobile-LHC association with PSII. These changes in the distribution of excitation energy are known as the State transitions (reviewed by Staehelin, 1986). The functional significance of lateral segregation is not entirely clear. It has been argued that segregation increases the ability of thylakoids to regulate electron transport under 6 different balances of illuminating wavelenths (Staehelin, 1986). Nevertheless, it has been pointed out (Miller and Lyon, 1985) that algae that do not possess thylakoids differentiated into appressed and non-appressed regions do show the state transitions indicative of LHCII phosphorylation, thylakoid destacking, and increased PSI antennal size. A summary of the above discussion is presented in Figure 1. 2. STRUCTURE AND DISTRIBUTION OF THE THYLAKOID LIPIDS Clues to the function of the thylakoid lipids might be found in their distribution, both taxonomically and cytologically, within photosynthetic organisms. An overview of the chemical structures and distributions of the major lipids is presented. 2.1. Chemical structures Lipids constitute a class of biological macromolecules that are broadly defined as being soluble in non-polar organic solvents. The lipids are divided into two major classes: the complex lipids which possess covalently bound fatty acids and are saponifiable, and the simple lipids which do not possess fatty acids and are not saponifiable. Lipids of biological membranes show strongly amphipathic structures with the hydrophilic and hydrophobic regions bridged by a glycerol "backbone". Long chain fatty acids are esterified at carbons one and two of the sn- glycerol moiety (sn: stereospecific numbering system for glycerol). Esterified at carbon three is the hydrophilic head group consisting of either a phosphate group and various attached residues (the phospholipids) or a sugar moiety (the glycolipids). The predominant lipids of thylakoids (Figure 2) are the uncharged galactolipids monogalactosyldiacylglycerol (MGDG) and digalactosyldiacylglycerol (DGDG). These 7 lumtn L H C * T P Figure 1. Schematic representation of the structure of higher plant thylakoid membranes showing both the path of electron flow through the Z-scheme and lateral differentiation of appressed and non-appressed regions. Abbreviations as in the text Figure modified from Staehelin (1986). 8 CH2OH HO 0 .OH -O CH 2 -OH (a) (c) CH2OH HO>—— O -CH -CH 2 o 0 C-0 C-0 (CH 2 ) 7 JCH 2) 7 CH CH hi 1 CH CH 2 CH 2 CH CH CH | CH 1 CH 2 CH 2 CH CH CH CH CH 2 CH 2 CH 3 CH 3 -CH -CH 2 O O C-0 C-0 (CH 2) 7 JCH 2) 7 CH 2 CH • CH 2 CH | CH 2 CH 2 CH 2 CH CH 2 CH CH 2 CH 2 CH 2 CH CH 3 CH C H 2 CH 3 .OH HO ° CH 2 HO .OH OH (b) O l C H 2 — C H CH 2 9 9 C-0 C-0 (CH 2) 7 JCH 2) 7 CH CH Ii • CH C H 2 CH CH C H 2 CH CH C H 2 C H 3 CH i C H 2 CH CH C H 2 CH CH C H 2 C H 3 H H H - C - C — CH 2 -0-P-0" HO OH 9 C H 2 — C H CH 2 (d) 9 C-0 C H 2 CH HC (CH2)7 C H 2 i C H 2 C H 2 C H 2 C H 3 9 c=o icH2>7 CH CH C H 2 CH • CH C H 2 CH CH CH 2 CH 3 Figure 2. Chemical stmctures of (a) 18:3/18:3-MGDG, (b) 18:3/18:3-DGDG, (c) 18:3/16:0-SQDG, and (d) 18:3/16:1-PG. 9 galactosylglycerols total about 75% of the total thylakoid lipid (50% and 25% respectively). The remainder is comprised mostly of the anionic lipids sulfoquinovosyldiacylglycerol (SQDG) and phosphaudylglycerol (PG) (Allen and Good, 1971). A long-standing debate concerning the presence of phosphatidylcholine (PC) in thylakoids either as an endogenous lipid or as a contaminant present from the extra- chloroplastic membranes is still unresolved. Most workers routinely obtain a few mol% PC in "purifiedB thylakoid preparations. Several other galactolipids have also been detected in plant extracts, including tri- and tetra- galactosylglycerols as well as M G D G and DGDG acylated at the galactose group. These compounds are probably artifacts of tissue homogenization and membrane isolation (Heinz, 1967a,b, 1973; Critchley and Heinz, 1973; Wintermans et al., 1981). Fatty acids are long chain hydrocarbons with a terminal carboxyl group through which the fatty acid is esterified to glycerol to yield the complex lipid. Higher plants and animals commonly contain unbranched chains of 14 to 24 carbons length and may be saturated (no double bonds) or unsaturated (one or more cis double bonds). Fatty acids most commonly esterified to membrane lipids in plants are palmitic acid (16:0; denotes the number of carbon atoms:number of double bonds), stearic acid (18:0), oleic acid (18:1), linoleic acid (18:2), and linolenic acid (18:3). In addition, photosynthetic tissues in plants possess significant quantities of the unusual fatty acid trans-A3-hexadecanoic acid (16:1) esterified specifically to sn-2 of PG. Some species also contain high levels of palmitolenic acid (16:3) located almost exclusively on MGDG. As this discussion implies, fatty acids are specifically distributed between the various lipid classes and, in some cases, bound to specific sites within a given lipid. As a result, the four major classes of lipids (MGDG, DGDG, SQDG, and PG) represent dozens of different molecular species (specific head group and fatty acid combinations) of lipids. Finally, it should be added that the thylakoids also contain a wide variety of lipophilic pigments including the chlorophylls a and b, the xanthophylls, 0 - carotene, and the 10 quinones. Most, or all, of these pigments are likely to be non-covalently bound to the polypeptides of the pigment-protein complexes (Murphy, 1986a,b). The chlorophyll-binding proteins involved in light absorption bind specific types and amounts of the chlorophylls, most likely by hydrophobic interactions. In contrast, the quinones are free to diffuse within the bilayer and are probably only transiently associated with proteins during the electron transport reactions. The exact vertical location of the quinones within the lipid bilayer is a point of some debate (Kingsley and Feigenson, 1981; Stidham et al., 1984). The function of these compounds in photosynthesis has been reviewed (Murphy, 1986b). The other cellular membranes of plants are rich in the phospholipids PC, PG, phosphatidylethanolamine (PE), phosphatidylserine (PS), phosphatidylinositol (PI), and cardiolipid, as well as cerebrosides and sterols (Lynch and Steponkus, 1987). 2.2. Taxonomic distribution The glycolipids of higher plant thylakoids are common to all oxygen-evolving photosynthetic organisms. These lipids have been detected in all higher plants, as well as in the red, green, and brown algae, in the diatoms, in the dinoflagellates and in the cyanobacteria Anabena sp. and Anacystis sp. (reviewed by Douce and Joyard, 1979). The galactolipids have also been found in the presumptive descendent of the original chloroplastic endosymbiont, Prochloron sp., as has a closely related lipid monoglucosyldiacylglycerol (MglcDG) (Murata and Sato, 1983). The lipids of the photosynthetic purple bacteria, Rhodospirillacea, are predominantly those phospholipids common to other bacteria. Clearly these lipids are highly conserved despite significant changes in thylakoid ultrastructure and light absorption strategies utilized by many of these "lower" organisms (Staehelin, 1986). In some of the cyanobacteria, the ability to desaturate MGDG fatty acids beyond 16:1 and 18:1 is not present 11 2.3. Cellular and subcellular distribution It has been implied above that the galactolipids and SQDG occur exclusively in the thylakoid membranes. In fact, significant levels have also been found in the chloroplast envelopes of higher plants (Douce and Joyard, 1979). Consistent with an endosymbiouc origin of chloroplasts, the separation of the inner and outer membranes of the chloroplast envelope (Cline et al., 1981) has shown the inner membrane to be enriched in unsaturated glycolipids. Conversely, the outer membrane was composed predominantly of phospholipids similar to those in the cytoplasmic membranes and possessed lowered glycolipid levels and lower levels of acyl unsaturation. The lateral segregation model of higher plant thylakoid organization suggests that such lateral heterogeneity might also exist for the membrane lipids. Investigations have shown, however, only small enrichments of 4-10 mol% for MGDG in granal membranes compared to stromal membranes or whole thylakoids. Similar, but more variable, results have been reported for the distributions of DGDG, PG, and SQDG (Tuquet et al., 1977; Gounaris et al.. 1983d; Murphy and Woodrow, 1983; Farineau et al., 1984; Tuquet et al., 1986). Of more significance is the lateral heterogeneity in total lipid content of appressed and non-appressed membranes. The former fraction is highly depleted in lipid and enriched in protein, and the reverse true of the latter fraction (Ford et al., 1982; Murphy and Woodrow, 1983). Transmembrane specificity of thylakoid lipid distribution has also been examined by several groups (reviewed by Siegenthaler and Rawyler, 1986; Gounaris et al., 1986). Work of this nature has proceeded by three main techniques. Some workers have employed the external or internal digestion of membranes by lipolytic enzymes and assaying for the preferential destruction of certain lipids (Rawyler and Siegenthaler, 1981, 1985). Others have used the external or internal labelling of lipids, usually the labelling of galactose with 3 H after 12 galactose oxidase digestion (Sundby and Larsson, 1985; Rawyler et al., 1987). Finally, monospecific antibody binding experiments have also been used (Radunz, 1980). All of these methods suffer serious technical difficulties, not the least of which is the lack of uniform accessibility of all lipids in a monolayer to large external enzymes, antibodies, or reagents. In view of the high protein content of this membrane (60% by weight) the basic prerequisite for complete protein or enzyme accessibility seems unlikely. Further, lipolytic studies must be done with great care to avoid extensive membrane reorganization, particularily lipid transbilayer flip-flop, as a result of degradative lipid removal from one monolayer (Op den Kamp, 1979). It is not surprising, then, that there is some disagreement in the literature concerning the transbilayer orientation, if any, of these lipids. MGDG may show a 60/40 outside/inside ratio based on lipolytic and labelling methods (Rawyler and Siegenthaler, 1985; Rawyler et al., 1987) but has also been reported as being mainly inside by antibody binding (Radunz, 1980). Similarly, reported DGDG outside/inside ratios of 15/85 (Rawyler and Siegenthaler, 1985; Rawyler et al., 1987) are supported by monospecific antibody binding (Radunz, 1980) but in disagreement with a study using chemical labelling (Sundby and Larsson, 1985). For SQDG a mainly inside distribution has been inferred based on the high proportion of other lipids supposedly in the outer monolayer (Sundby and Larsson, 1985) but opposite results have been obtained with antibody binding (Radunz, 1980). Workers appear to agree, however, on a 70/30 outside/inside ratio for PG (Radunz, 1980; Rawyler and Siegenthaler, 1980; Siegenthaler and Rawyler. 1986; Unitt and Harwood, 1985). The significance of these distributions, if real, are not known. It is possible that the variable data obtained for the lateral and transmembrane distributions of lipids results, in part, from catalytic degradation of lipids during cell disruption and membrane purification. The very rapid destruction of thylakoid acyl lipids by enzymes released and/or activated during cellular rupture is well known (Gallaird, 1980). Both 13 galactolipid-specific (Sastry and Kates, 1964) and non-specific (Hirayama et al., 1975) activities have been reported in a wide variety of species and tissues. These include potato tubers (Hirayama et al., 1975), leaves of Phaseolus multiflorus (Sastry and Kates, 1964; Burns et al., 1977a,b), and chloroplasts of Phaseolus sp. (Anderson et al., 1974; Shaw et al., 1976) and of Triticum sativum (O'Sullivan et al., 1987). Indeed, rapid degradation of as much as 50% of thylakoid lipids during the isolation of wheat chloroplasts (O'Sullivan et al., 1987) as well as thylakoids, PSII vesicles, and stromal lamellae of barley (Henry et al., 1983) has been reported. Few investigators of lipid composition in sub-cellular and sub-thylakoid membrane preparations report measures taken to avoid lipolytic activity. Further, few workers report even the routine quantification of neutral lipid (free fatty acids and diacylglycerols) fractions in such preparations, possibly because these compounds co-chromatograph with the pigments in many solvent systems. It should be added, though, that catalytic activity is probably species-specific: no degradation of lipid is found in isolated pea thyakoids (Chapman and Barber, 1987). In view of the above discussion, it may be that few of the studies of sub-thylakoid membrane lipid composition are reliable. At the very least, it would be of interest to repeat some of this work with membranes isolated in the presence of lipase inhibitors such as p-chloromercuribenzoate and phenylmercuric acid (Yang et al., 1967; O'Sullivan et al., 1987). 3. PHASE BEHAVIOR OF THYLAKOID LIPIDS The ability of membrane lipids to modulate membrane behavior is determined by their physical characteristics. This section provides a review of the known physical properties of thylakoid lipids as a prelude to a discussion of how these properties might affect photosynthesis. 14 3.1. MGDG The phase behavior of chloroplast galactolipids was first investigated by Rivas and Luzzati (1969). They reported that galactolipid mixtures from maize would adopt a variety of phases, including the hexagonal-II (H^), cubic, and lamellar phases depending on the temperature and water content of the system. The phase is a structure in which the lipids are arranged as long, hexagonally packed, tubes with the head groups near the center of the tubes and the acyl chains splayed outwards. The center of each tube is occupied by an aqueous core. Kxuetz (1970) and Shipley et al. (1973) showed that purified MGDG would adopt the H phase over a wide temperature range of -15° to 80° C at l>50% lipid in water. The extent to which MGDG adopts the H phase is highly dependent on the degree of unsaturation of its acyl chains (Gounaris et al., 1983b). The formation of the H phase by M G D G at 20° C was significantly reduced by dropping the average number of double bonds per MGDG molecule by about one. More complete saturation promoted the formation of a laterally segregated lamellar gel (L/3)-H phase (Mansourian and Quinn, 1986). However, even monounsaturated MGDG can be forced to take up the H phase by higher temperatures (Mannock et al., 1985). Fully saturated M G D G yields a liquid-condensed monolayer (Bishop et al., 1980) indicative of the L0 phase but showing the existence of several metastable gel states with complex phase behavior and thermal history dependence (Sen et al., 1983; Mannock et al., 1985). Reported temperatures for the Lfi-La transition of distearoyl-MGDG are between 45-70° C as measured by differential scanning calorimetry (DSC) (Sen et al., 1983; Gounaris et al., 1983b) and 70°C as measured by X-ray diffraction (Lis and Quinn, 1986). 15 3.2. DGDG The addition of a single galactose to MGDG to yield DGDG results in a lipid that forms a liquid-expanded (Bishop et al., 1980) lamellar (La) phase under all known conditions of temperature and hydration. Polyunsaturated DGDG undergoes the La to L/3 transition upon cooling below -50° C (Shipley et al., 1973). As with MGDG, saturated DGDG shows the gel phase, Lp\ at room temperature. A series of metastable forms of the L/3 phase have been identified (Sen et al., 1983). The distearoyl-DGDG lipid shows the transition from to La at 50-51°C (Sen et al., 1981b; Galanopoulou et al., 1982). 3.3. PG and SQDG The phase behavior of a wide variety of lipids, including PG, has been reviewed recently by Tilcock (1986). For unsaturated PG the La phase is expected under most physiological conditions. Dipalmitoyl-PG (DPPG) shows a LP to La transition at 41° C in H20, but is raised to 54° C in the presence of Mg2*, and to 86° C by Ca2+. This probably results from electrostatic screening of adjacent negatively charged phosphate groups (Tilcock, 1986) and/or ion binding and charge bridging. Thylakoids contain significant quantities of PG with the molecular species 16:0/16:0 (DPPG) and 16:0/16:1 (Murata, 1983). This fact may be relevant to the low-temperature sensitivity of some plant species. It has been proposed (Murata, 1983) that the high levels of high melting-point PG in the thylakoids of chilling-sensitive plants leads to the formation of L/3 phase lipid in the membrane at killing temperatures. This L/3, or gel state, lipid was predicted to have detrimental effects on membrane permeability and protein/enzyme function. The addition of the trans-A3 bond to the 16:0 at sn-2 position drops the temperature of the L/3-La transition by about 10°C below that of DPPG (Bishop and Kenrick, 1987). 16 The phase behavior of thylakoid SQDG has not received as much attention as that of the galactolipids. SQDG purified from an algal source showed the La phase above 20° C, mixed La ' and Lj3 phases from 20° to -10°C, and totally LB at <, -10°C (Shipley et al., 1973). The fatty acid composition of this SQDG was highly saturated compared to that of higher plants. The 18:3 content and molecular species compositon of SQDG has been shown to be highly species-dependent (Murata and Hoshi, 1984). The SQDG purified from spinach showed no evidence of L/3 phase formation above 5°C, as detected by fluorescence polarization of trans-paranaric acid (Murata and Yamaya, 1984). There is no published data concerning the possibility of an SQDG H phase at higher temperatures and high ionic strength. 3.4. Binary, and higher, mixtures The tendency of binary galactolipid mixtures, and of total thylakoid polar lipids, to form the H phase or its inverted micellar intermediate (Sen et al., 1981a; 1982b) is due entirely to unsaturated MGDG (Gounaris et al., 1983b). Replacement of unsaturated M G D G with identical proportions of fully hydrogenated MGDG in thylakoid lipid extracts removed the ability of these preparations to form the H phase. Binary galactolipid mixtures show a complex interplay of La , H , cubic and water phases depending on both water content and temperature (Rivas and Luzzati, 1969; Brentel et al., 1985). Promotion of the La to H transition in binary galactolipid mixtures has been accomplished by both increased temperature and by ethylene glycol addition. This has led Sen et al. (1982a) to conclude that the H transition requires both a large hydrophobic volume (from unsaturated fatty acids and/or thermal motion) and the disruption of lipid head group interaction with hydrating water. 17 The stabilization of the lamellar phase in thylakoid lipid extracts by the presence of the anionic lipids SQDG and PG has been suggested (Gounaris et al., 1983b). It could not be concluded from this work, however, whether stabilization resulted from direct lipid-lipid interactions or, more likely, resulted from electrostatic repulsion between adjacent vesicles. This would lead to the inhibition of vesicle fusion and resultant increase in vesicle size. It could be argued, therefore, that a lamellar phase for MGDG is promoted by smaller vesicle size as a result of more acceptable packing of the cone-shaped MGDG molecule at the inner monolayer of small vesicles. Such a transbilayer asymmetry based on steric considerations has been proposed for MGDG in thylakoid membranes, particularily at the inner surface of the highly curved end margins (Murphy, 1982), but is not supported by empirical evidence (see above). It should be added that the phase(s) adopted by binary galactolipid mixtures is strongly dependent on the reconstitution method used (Sprague and Staehelin, 1984a,b). In addition, differences in vesicle size and internal solute content (Gruner et al., 1985b; Mayer et al., 1985; Perkins «et al., 1988) may arise from different reconstitution schemes and, hence, affect the lipid phase structure. 3.5. The shape concept of lipid phase behavior The phase preferences of membrane lipids may be qualitatively understood by the "shape concept" of lipid polymorphism (Cullis and De Kruijff, 1979; Gruner et al., 1985a). This model is closely related to the quantitative "intrinsic curvature" model (Gruner, 1985) and the "critical packing parameter" of Israelachvili et al. (1977; 1980). These models argue that the phase preference of a membrane lipid originates in the ratio of volumes occupied by hydrophobic acyl regions and hydrophilic head group regions. Factors that increase the ratio of hydrophobic/hydrophilic volumes promote a time-averaged cone, or wedge, shape of the lipid. Dowest energy packing of these lipids would be an "inside-out" phase with clustered interior 18 head groups and external hydrophobic regions - the phase. Such factors include cis-unsaturation of acyl chains, high temperatures, and decreased head group area caused by either decreased hydration or salt-mediated electrostatic screening of charged groups. It is obvious, then, that MGDG adopts the H phase because of both a very high degree of acyl unsaturation and the small swept area of the monogalactosyl head group. Conversely, the DGDG head group is significantly larger than that of MGDG, giving the molecule an overall cylindrical shape that is most easily accomodated by lamellar phase packing. Similarily, PG and SQDG show overall cylindrical shapes due to moderate unsaturation and large head group areas due to inter-lipid electrostatic repulsion from the phosphate and sulfonate groups. Conical and cylindrical shapes for MGDG and DGDG have been obtained by minimum energy modelling of these lipids (Brasseur et al., 1983). This analysis showed the areas occupied per hydrocarbon chain/head group to be 0.96 nmV0.54 nmJ for MGDG and 0.95 nmV0.85 nm2 for DGDG. Inverted and lamellar packing were predicted for these lipids on the basis of these shape considerations. Very similar phase behaviors are observed for the monoglucosyl- and diglucosyl-diacylglycerols isolated from Achdeplasma laidlawii (McElhaney, 1984; Lindblom et al., 1986). 3.6. Bilayer fluidity It is commonly believed that a high degree of thylakoid lipid unsaturation is critical to the low temperature thermal tolerance of photosynthesis. Specifically, low temperature induced reduction of lipid acyl chain motion may be expected to lower the rate of PQ diffusion and, hence, of electron transport (Oquist, 1982). In plants, therefore, a reduction in temperature may be expected to be accompanied by a compensatory increase in acyl 19 unsaturation and decrease in membrane microviscosity (Raison. 1973, 1980; Shinitzky and Henkart, 1979). Increased acyl unsaturation in response to lowered temperature has been observed in wheat root phospholipids (Ashworth et al., 1981), pine chloroplasts (DeYoe and Brown, 1979), microsomes of Dunaliella salina (Lynch and Thompson, 1982), the cyanobacteria Anabena variabilis (Sato et al., 1979; Sato and Murata, 1981); ivy leaves and spruce needles (Senser and Beck, 1984) and in leaves of Vicia jaba (Lem et al., 1980). Other reports show no correlation between fatty acid unsaturation and temperature in Anacystis nidulans (Sato et al., 1979), in chloroplasts (Oquist 1982) and plasma membranes (Hellergren et al., 1984) of Pinus sylvestris, nor in cold-hardened Brassica napus (Stout and LaCroix, 1981). Other temperature reponses of membranes may include changes in phospholipid composition (Horvath et al., 1980, 1981), glycolipid composition (Vigh et al., 1985a), fatty acid positional distribution (Watanabe et al., 1981), and non acyl-lipid membrane components (Lynch and Steponkus, 1987). Several workers have suggested that low temperature stress may be avoided by the seasonal accumulation of the "free-water binding" lipid, DGDG (DeYoe and Brown, 1979; Oquist, 1982), an assumption without empirical support In vivo alteration of thylakoid fluidity has been attempted by several groups. Restall et al., (1979) reported that hydrogenation of 40% of double bonds in spinach thylakoids, primarily in 18:3, had no effect on the rate of photosynthetic electron transport nor on thylakoid ultrastructure. More recent work using a removable water-soluble catalyst has suggested inhibition of electron flow from PSII to PSI after 10% hydrogenation, inhibition through PSII at higher levels of hydrogenation and, insensitivity of PSI electron flow to as much as 50% loss of acyl double bonds (Vigh et al., 1985b; Horvath et al., 1986). These effects were apparently not due to catalyst interactions with antennal or reaction center complexes (Horvath et al., 1986), nor with chlorophylls, carotenoids or PQ (Szalontai et al., 1986). 20 The suggestion by these workers that increased bilayer thickness could be detected, using TEM, after 10% hydrogenation of double bonds (Horvath et al., 1986) is highly questionable. Further, at 10% hydrogenation levels, at which full chain electron transport was inhibited by 30%, electron spin resonance (ESR) measurement of bilayer order (1/fluidity) with the spin probe 12-SASL showed decreased, not increased, acyl order (Vigh et al., 1985b). In addition, even with 40% hydrogenation of double bonds, ESR measurements of rotational correlation times (T 0 ) for the spin probe 16-SASL showed only a minor increase of T 0 from 1.75 ns to 2.0 ns (Horvath et al., 1986). These criticisms are consistent with conclusions reached by Van de Ven et al. (1984) and Koole et al., (1984) that the addition of double bonds to DGDG does not increase hydrocarbon region fluidity. The possibility that hydrogenation affects some other, as yet unexamined, electron transport component remains a significant problem. More convincing results have been obtained by lowering thylakoid fluidity with the addition of cholesterol (Ford and Barber, 1980) or cholesterol hemisuccinate (Yamamoto et al., 1981; Ford and Barber, 1983b) to thylakoids. This work suggested that reduction of bilayer fluidity reduced whole chain electron transport rates, proton uptake, Q oxidation, and B cytochrome f reduction by reducing the rate of PQ diffusion. While these conclusions were similar to those discussed above, the results obtained from the different approaches suggests that acyl unsaturation is not the primary factor influencing overall thylakoid fluidity. This view is more clearly pointed out by Hiller and Raison (1980), Ford and Barber (1983a), and Waggoner et al. (1985) in results showing higher microviscosity, higher degrees of acyl orientational order, and lower fluidity in whole thylakoids than in their purified lipids. A central role of proteins in decreasing overall membrane fluidity is also evidenced by the higher fluidity of stromal membranes, compared to granal membranes. Stromal lamallae show a lower protein/lipid ratio than granal lamallae but nearly identical acyl unsaturation and 21 higher fluidity (Ford et al., 1982; Murphy and Woodrow, 1983). In addition, there is some evidence that plants may adjust their thylakoid fluidity in response to lowered growth temperature by decreasing the protein/lipid ratio of the membrane rather than adjusting acyl unsaturation (Chapman et al., 1983) This result is consistent with a recent ESR investigation showing that increased fatty acid unsaturation did occur in low temperature-grown Synechococcus sp. but was not related to changes in bilayer fluidity nor thermal adaptation (Miller et al., 1988). Finally, a mutant of Arabidopsis Indiana has been obtained that shows greatly reduced levels of 16:3 and 18:3 in all of the chloroplast lipids (Browse et al., 1986). This mutation showed no effect on photosynthetic electron transport rates, C02-fixation, nor membrane fluidity (McCourt et al., 1987). There is evidence that MGDG unsaturation may be involved in high temperature instability of thylakoids. At temperatures of 40-50° C irreversible denaturation of chlorophyll-protein complexes is observed (Cramer et al., 1981) and correlates with loss of oxygen evolution, PSII activity, and photophosphorylation (Thomas et al., 1986b). These temperatures also cause granal destacking, lipid phase separation and extensive H phase •> II formation (Gounaris et al., 1983a). Protection of PSII activity and prevention of H phase formation were both conferred by in vivo hydrogenation of the membrane lipids (Thomas et al., 1986a). In summary, the above discussion suggests that only in a few organisms, and perhaps only in specific membrane systems, can fatty acid unsaturation be considered a primary response to lowered growth temperture. Even when decreased temperature does lead to increased unsaturation, an associated increase in bilayer fluidity does not necessarily occur. The thylakoid lipids may be highly unsaturated in part to facilitate rapid PQ diffusion but primarily to counter the ordering effect obtained from the very high protein content of this membrane. This would serve to ensure that the membrane is in the liquid-crystalline lamellar 22 phase but might confer high temperature instability to the thylakoid. This suggestion is similar to the "homeophasic adaptation" proposal of Silvius and McElhaney (1980). Alternative roles for unsaturated thylakoid lipids are outlined in the next section. 4. LIPID-PROTHN INTERACTIONS IN THYLAKOIDS The heterogeneity of lipid species in the thylakoid, and other membranes has prompted many workers to look for specific interactions between lipids and membrane proteins. These interactions are usually envisaged as involving a so-called "boundary lipid" that may be preferentially associated with a given protein. Such a lipid may be expected to act as a specific activator of protein or enzyme function or act to facilitate protein packing in the membrane (Murphy, 1982, 1986a,b). Work in this area has proceeded by employing several methodologies. Many investigators have looked for the co-purification of a specific lipid during the purification of' a protein, or chlorophyll-protein complex, from the thylakoid. Others have attempted to reconstitute a purified protein or complex into vesicles of known lipid composition. Digestion with specific lipolytic enzymes has also been used. However, higher plant mutants have recently become very useful at clarifying some confusing results in this area. 4.1. MGDG The apparent slight enrichment of MGDG in the appressed membranes discussed in Section 2.3. might result from a reported enrichment of MGDG (up to 75 mol%) in a purified PSII core preparation (Gounaris and Barber, 1985). However, these workers reported that this MGDG was highly saturated, a fact which is in disagreement with the very unsaturated nature of MGDG from most appressed membranes, and from partially purified 23 PSII preparations (Murphy and Woodrow, 1983; Farineau et al., 1984). In addition, this result is not consistent with the finding that unsaturated, but not saturated, MGDG facilitated energy transfer from LHCII to PSII (Siefermann-Harms et al., 1982). MGDG is apparently excluded from purified PSI preparations (Rawyler et al., 1980) and CFo-CFi preparations from spinach and Dunalliela salina (Pick et al., 1985). Despite this, MGDG has been claimed to reconstitute energy transfer from LHCI to PSI in detergent solubilized thylakoids, whereas neither PE nor PC were capable of this (Siefermann-Harms et al., 1987). The exclusion of MGDG from CFo-CFj preparations is difficult to reconcile with reports from the same group showing ATPase activity to be stimulated by unsaturated MGDG (Pick et al., 1984, 1987) but inhibited by saturated MGDG and other lamellar-phase lipids. On the other hand, a requirement for H ^ phase preferring lipids, such as MGDG and PE, over lamellar phase lipids DGDG and PC has also been reported for the Ca2+-ATPase of sarcoplasmic reticulum (Navarro et al., 1984). Galactolipase digestion of thylakoids has shown that removal of galactolipids, particularily MGDG, inhibited whole'chain electron transport and electron flow through PSI (Shaw et al., 1974; Anderson et al., 1976). These effects were shown to be largely due to enzymatic release of free fatty acids, and not to galactolipid removal per se. Such a result is consistent with the conclusions of Rawyler and Siegenthaler (1980) that there was no good correlation between enzymatic lipid removal and photosynthetic electron transport. Galactolipid digestion from thylakoids has been correlated, by other workers, with the destruction of chlorophyll-protein complexes detected by gel electrophoresis (Krupa, 1984) and of PSII reaction centers as judged by freeze-fracture EM (Jacob and Miller, 1986). Neither of these groups controlled for the likely detergent effects of released free fatty acids. 24 A role for mutually compatible packing of M G D G and protein complexes, particularily LHCII, is supported by the finding that all Chlamydomonas mutants lacking CP complexes showed higher DGDG/MGDG ratios (Semenova et al., 1987). 4.2. DGDG and phospholipids Few workers have reported data showing the specific co-isolation of DGDG with any thylakoid-associated protein complex. The probable non- effect of galactolipid removal referred to above applies to both DGDG and MGDG as both of these compounds are digested by galactolipases (Shaw et al., 1974; Rawyler and Siegenthaler, 1980; Krupa, 1984; Jacob and Miller, 1986). Reconstitution into phospholipid vesicles has been accomplished for LHCII. (Mullet and Arntzen, 1980; McDonnel and Staehelin, 1980; Ryrie et al., 1980; Morschel and Staehelin, 1983; Murphy et al., 1984; Sprague et al., 1985), cytochrome b 6 / f (Morschel and Staehelin, 1983), CFo-CFi (Mullet et al., 1981; Morschel and Staehelin, 1983; Pick et al., 1985), for PSII (Murphy et al., 1984); and for PSI (Mullet et al., 1980; Dunahay and Staehelin, 1985; Lopez and Tien, 1984; Orlich and Hauska, 1980). Functional PSI activity has been coupled to CFo-CFi ATP synthetase activity with reconstituted soybean phospholipids (Hauska et al., 1980). A requirement for a neutrally-charged galactolipid lamellar phase (DGDG) over a non-charged phospholipid lamellar phase has been reported for the structural integrity of reconstituted PSII reaction center proteoliposomes (Sprague et al., 1985). Reconstitution into DGDG vesicles has been accomplished for the cytochrome b 6 / f complex (Morschel and Staehelin, 1983), CFo-CF, (Morschel and Staehelin, 1983; Pick et al., 1987), LHCII (Remy et al., 1984; Sprague et al., 1985) and the oxygen-evolving complex of PSII (Gounaris et al., 1983e). No specific requirements for DGDG have been reported by any of these workers. 25 4.3. SQDG More convincing data for lipid-protein interactions has been obtained for the anionic glycolipid SQDG. The earliest was that of the ability of SQDG to protect CFi against thermal inactivation (Livne and Racker, 1969). SQDG has been reported to co-purify strongly with CFo-CF, from both spinach and Dunalliela salina (Pick et al., 1985). Chloroplast CF : binds to vesicles of chloroplast lipids but not to vesicles of PC, even with incorporated CF 0 (Dijkman and Daniel, 1981). In reconstituted systems the ATPase activity of CF 0 -CFi can be stimulated by chloroplast lipids but not by phospholipids (Pick et al., 1984). This stimulation appeared to be due to a CFo-CFi requirement for unsaturated MGDG and at least one other lipid. This "other lipid" may be a mixture of DGDG and SQDG resembling the in vivo proportions (Pick et al., 1987). These workers have suggested that SQDG may affect apparent ATPase activity and ATP-Pi exchange rates by directly affecting membrane proton permeability (Pick et al., 1987). A role for an SQDG interaction with CFo-CFj has been proposed (Sakai et al., 1983; Barber and Gounaris, 1986). Recently Sigrist et al. (1988) have reported the co-purification of SQDG with LHCII of Chlamydomonas reinhardii. This is the first report of this sort of interaction. SQDG and M G D G have both been described as inactive at promoting cytochrome b 6 /f mediated electron transport, compared to the effects of DGDG, PG, and PC. (Chain, 1985). 4.4. PG While there has been a report of PG requirement by PSI (Rawyler and Siegenthaler, 1981a), this is the only report of its kind. This section will concentrate on PG and 16:1 interactions with LHCII, a subject that has been reviewed by Dubacq and Tremolieres (1983). 26 The earliest suggestions of a role of PG and the trans-A3 bond of 16:1 in granal stacking were based on the correlation between 16:1 accumulation in PG and the extent of granal stacking during the greening of etiolated leaves (Tuquet et al., 1977; Lemoine et al., 1982). This correlation extends to the accumulation of chlorophyll b and LHCII in developing leaves and can be obtained under different developmental conditions of temperature, light intensity, and illuminating wavelength and periodicity (Dubacq and Tremolieres (1983). The enrichment of PG and 16:1 has been observed in detergent-solubilized LHCII (Rawyler et al., 1980; Krupa et al., 1987), in the LHCII oligomer LHCII* (Tremolieres et al., 1981; Remy et al., 1982), and in appressed membrane preparations (Murphy and Woodrow, 1983; Gounaris et al, 1983e). Other workers have shown no, or only slight, enrichment of PG in isolated granal membranes (Tuquet et al., 1977; Farineau et al., 1984). Low temperature development of rye leads to the formation of an isolated LHCII with significantly lowered 16:1 content and thylakoids with much lower IJHCIIVLHCII ratio (Krupa et al., 1987; Huner et al., 1987; Krol et al., 1988; Huner et al., 1989) as detected by SDS-PAGE Enzymatic removal of PG from detergent-purified LHCII* resulted in the complete conversion of the oligomer to the monomer (Krupa et al., 1987). Similar results have been obtained by phospholipase A 2 digestion of thylakoids (Remy et al., 1982). Finally, the monomeric LHCII has been reported to strongly reassociate into the LHCII* oligomer in PG-16.-1 liposomes to a much greater extent than in other types of liposomes (Remy et al., 1982, 1984). The involvement of PG and 16:1 in LHCII* formation and, thereby, granal stacking has been seriously challenged by genetic studies. Reduced membrane appression has been found in a barley mutant lacking chlorophyll b (and presumably LHCII) but with normal levels of PG and 16:1. A mutant of Arabidopsis thaliana specifically lacking in 16:1 of PG showed no changes in membrane appression nor energy transfer from LHCII to PSII (Browse 27 et al., 1985; McCourt et al., 1985). In addition, some plant species with no 16:1 present in their PG still show normal ratios of LHCII*/LHCII (Huner et al., 1989). A role for PG and 16:1 in LHCII-mediated excitation transfer and/or granal stacking in vivo must, therefore, be considered unlikely. Developmental results discussed above may be simply due to the co-regulation of genes involved in chlorophyll b synthesis, LHCII polypeptide synthesis and the PG 16:0 desaturase. The co-isolation of PG with LHCII may result from reassociauon of the charged lipid with protein after detergent solubilization of thylakoids. This behavior has been inferred previously (Heinz and Siefermann-Harms, 1981). The preferential association of PG and 16:1 with the purified light-harvesting assemblies of the brown alga Fucus (Caron et al., 1985), despite the differences in LHC structure in brown alga and higher plants (Anderson and Barrett, 1986), supports the suggestion that PG reassocation may be non-specific. Oligomerization of LHCII monomer by PG may also result simply by lipid-induced charge neutralization and/or hydrophobic association. Alternatively, PG may only have an effect on the sensitivity of LHCII* to detergent solubilization (McCourt et al., 1987). A correlation between 16:1 levels and freezing tolerance of some species of cereals has been reported (Huner et al., 1989). A general comment is required at this point regarding the nature of reconstituted lipid- protein systems used in some of the above work. While the ideal reconstituted system (Madden, 1986) may not be necessary for all types of studies, some degree of caution is obligatory. It is not uncommon, for instance, for workers to claim the reconstitution of a given protein with unsaturated MGDG without showing the existence of a lamellar phase instead of the H phase or showing that the protein actually was integrated into a hydrophobic, non-detergent, milieu (Ikegami, 1983; Siefermann-Harms et al., 1982, 1987; Pick et al., 1984, 1987). Furthermore, the problem of the quantity and location of residual detergent is rarely addressed, and some reconstitutions would appear to contain 50% or more 28 of detergent (Siefermann-Harms et al., 1982, 1987). The biological relevance of reconstituted systems of this sort is questionable. 5. INTER-BILAYER INTERACTIONS The physical chemistry of interactions between bilayer membranes is the subject of extensive current research. The principles have been comprehensively reviewed by Rand (1981) and Israelachvili (1985) in relation to phospholipid bilayers and by Barber (1980, 1982) as well as Thome and Duniec (1983) in relation to thylakoids. Bilayer membranes represent dispersed colloidal systems consisting of the continuous (aqueous) phase and the discontinuous (bilayer) phase. Colloids of this sort are characterized by strong interfacial and surface interactions at the boundaries of these phases. Bilayers are subject to long-range attractive van der Waals forces that decay somewhat more slowly than the inverse sixth power of the distance. As a result, van der Waals forces have an effective range of approximately 15 run. Since van der Waals forces arise from attraction between intrinsic fluctuating dipoles, the magnitude of the attractive force is dependent on the polarizability of the lipids and, hence, on the chemical nature of the lipid. In general, the magnitude of the attractive van der Waals force is directly proportional to the Hamaker coefficient at constant water separation. Bilayers are also subject to a long range electrostatic repulsive force, or electrostatic potential, arising from the presence of fixed charges on the membrane surface. Mobile counter ions will be attracted to the fixed surface charges, giving rise to the so-called double-layer. The magnitude and decay length of the electrostatic potential depends in a complex fashion on the concentration and valency of the counterforts, and on the pH of the medium. In general, the electrostatic potential is related to the product of the surface charge densities of 29 the interacting bilayers. The decay length is strongly dependent on the ionic strength of the medium. Surface potentials ^ 20 mV generate an electrostatic potential sufficient to prevent vesicle flocculation. Electrolyte addition, particularity divalent counterions, reduces the electrostatic energy barrier to flocculation and moves the position of the barrier closer to the bilayer surface. Since van der Waals are longer range than repulsive electrostatic forces, vesicles should flocculate at the distance at which attractive forces equal repulsive forces. The above highly simplified version of DLVO (Deryaguin, Lifshitz, Verweey, Overbeek), or double-layer theory has been supplemented with the recent recognition of a rapidly decaying repulsive hydration force acting at very short distances (^2nm) from the bilayer surface (Rand, 1981). The force arises from the energy required to dehydrate hydroxy 1, phosphate, and ammonium groups. Hydration forces show a decay length of approximately 0.2-0.3 nm, and dominate short-range interactions between approaching membranes. The strength of the hydration force is expected to be a function of the solvent, rather than of the solute. Some influence of the phase of a phospholipid solute on the decay length of hydration repulsion has been reported, and suggests that the decay constant is not equal to the diameter of one water molecule (Mcintosh and Simon, 1986). Hydration repulsion is probably the force preventing the aggregation of membranes at direct molecular contact Several additional forces may act between approaching bilayers, including both steric repulsive forces (Mcintosh et al., 1987) and the recently proposed hydration-attraction force (Rand et al., 1988). 6. STATEMENT OF OBJECTIVES The intent of the work described in this thesis was to characterize several biologically-relevant physical properties of well-defined vesicle systems made from thylakoid 30 lipids. In particular, it was intended to examine some of the permeability properties of the thylakoid lipids. Results from a portion of this work are presented in Chapters 5 and 6. In the initial stages of this work, an unreported aggregation behavior of DGDG vesicles was discovered, which led to investigations of the surface interactions of DGDG bilayers. Before doing such work, however, it was necessary to obtain large quantities of thylakoid lipids. Chapter 2 describes the large scale preparation, purification, and composition of glycolipids and PG from spinach. In addition, n C - and 'H-NMR spectra are described for the glycolipids and, where possible, assignments given. Chapters 3 and 4 described some of the interactions occurring between vesicles of highly purified DGDG (Chapter 3) and of mixtures of thylakoid lipids (Chapter 4). The relevance of these data to granal stacking and membrane close approach is discussed. Chapter 5 describes the permeability properties of DGDG vesicles to Rb+, CI", and glucose, as well as the permeability of lipid mixtures to Rb+. In Chapter 6 the relevance of vesicle permeability to low-temperature stress in diilling- sensitive plants is examined. Finally, Chapter 7 presents an overall discussion of the data and draws some general conclusions. 31 2. PURIFICATION AND NMR SPECTROSCOPY OF THYLAKOID GLYCOLIPIDS 1. INTRODUCTION While the thylakoid lipids are dominated by the uncharged glycolipids MGDG and DGDG as well as the anionic lipids SQDG and PG (Harwood, 1980), small amounts of PC, phosphatidylethanolamine (PE), phosphatidylinositol (PI) (Allen and Good, 1971) and other phospholipids (Gounaris et al., 1983e) are also present. Chromatography of plant lipid extracts, and particularily preparative scale chromatography, has been complicated by the presence of a large number of additional compounds including chlorophylls and other pigments (Murphy, 1986a,b), and possible lipid degradation products (Gallaird, 1980) in CHCVCHjOH extracts. Further, the close-migration or co-migration of several thylakoid lipids on silica, gel (particularily DGDG, SQDG, and PE) (Kates, 1972; Allen and Good, 1971) has forced most investigators to resort to several different adsorbents for the purification of the thylakoid lipids. Typically these procedures have involved silica chromatography followed by further clean-up on DEAE-cellulose, DEAE-Sepharose or Florosil. It is not surprising, then, that there exists significant variation in the literature concerning the relative proportions of these lipids, even in the thylakoids of a species as extensively studied as spinach. Reported ratios of MGDG/DGDG vary from 0.9 (34%:40%, Farineau et al., 1984) to 2.0 (50%: 25%, Pick et al., 1985). It may be that these differences, and greater differences in the reported lipid composition of sub-thylakoid particles, arise from chromatographic differences among labs as well as the catalytic degradation of membrane lipids during chloroplast and membrane preparation. Data showing extensive degradation of membrane lipids during photosystem isolation has been published (Henry et al., 1983). In addition, very rapid hydrolysis of membrane lipids by lipases released and/or activated during cellular 32 disruption is well known (Sastry and Kates, 1964; Gallaird, 1980). The recent success of sulfhydryl- enzyme inhibitors in preventing galactolipase activity in wheat chloroplasts (O'Sullivan et al., 1987) should help to clarify these discrepancies. 13C-Nuclear magnetic resonance ( 1 3 C-NMR) spectra and longitudinal relaxation times have been reported for MGDG and DGDG (Johns et al., 1977a) as well as for SQDG (Johns et al., 1978) and PG (Coddington et al., 1981). Relaxation and correlation times for these molecules indicated similar degrees of head group motion for DGDG, SQDG, and PG, but faster motion for MGDG. In addition, higher rates of segmental motion in the fatty acids were measured towards the terminal, methyl, end of the acyl chains. This and other data has been interpreted as indicating a time-averaged cone shape for MGDG and cylindrical shapes for DGDG, SQDG, and PG in bilayers (Murphy, 1986a,b). This "shape concept" (Gruner et al., 1985a) has been used to rationalize the phase behavior of aqueous dispersions of these lipids (Murphy, 1982, 1986a,b) as seen by X-ray diffraction (Shipley et al., 1973) and freeze-fracture electron microscopy (Quinn and Williams, 1983). This Chapter describes a method for the purification of thylakoid glycolipids by a combination of silica and carboxymethyl-cellulose (CM-cellulose) chromatography. Yields of hundreds of milligrams and very high purity have been obtained. High resolution " C - N M R spectra as well as ID and 2D ^ - N M R spectra are described for the glycolipids and, where possible, assignments given. 33 2. MATERIALS AND METHODS 2.1. Materials Materials used here were obtained as follows. All solvents were from BDH and of reagent grade, except for benzene (99%) which was purchased from Fisher. All solvents used for preparative chromatography were re-distilled before use with the exception of benzene which was used without further purification. Other chemicals were: 1,1,3,3-tetramethoxypropane from Aldrich; trichloroacetic acid and BF } /CH 3 OH from BDH; Pt02 from ICN Biochemicals; silica gel 60 and 2.5 x 7.5 x 0.25 cm analytical Silica gel plates from Merck; CD 3OD and CDCl} from MSD Isotopes; ammonium molybdate, ammonium sulphate, butylated hydroxytoluene (BHT), 2-thiobarbituric acid, soy PE, as well as the 16:0, 18:0, 18:1, 18:2, and 18:3 fatty acid standards, from Sigma; pentadecanoic acid (15:0) from Supelco; and CM-cellulose (CM-52, sodium salt) from Whatman. The "C-NMR internal standard tetramethylsilane (TMS, Sigma) was the generous gift of Dr. Pieter Cullis, DepL of Biochemistry, U.B.C. 2.2. Lipid extraction and purification The method of extraction and purification of thylakoid lipids has been described previously in preliminary (Webb and Green, 1987) and partial (Webb et al., 1988) form. Bulk extraction of 2.0 to 2.5 kg of de-veined spinach leaves into 6 1 of cold CHClj/CHjOH (1/2, v/v) after brief homogenization (Polytron) was performed by minor modifications to the method of Allen and Good (1971). Residual fibre and debris were removed by filtration, with suction, over 8 Mm glass wool or cheesecloth. The filter cake was 34 washed several times with CHC1 3 /CH 30H (2/1, v/v). After allowing the phases to separate overnight the CHC13 phase was removed and the upper phase partitioned several times against additional CHC13. The total lipid extract was dried under vacuum and re-dissolved in about 60 ml of CHCIJ/CHJOH/HJO (60/30/3, v/v/v). Lipids were extracted from 2-3 g of spinach leaves as described by Williams and Merrilees (1970) and separated by the ID TLC method of Khan and Williams (1977) on (NH«) 2 S0 4 impregnated 0.25 mm silica gel 60 plates. Preliminary separation of the lipids and removal of bulk chlorophyll was performed by liquid chromatography on a Waters Prep 500 LC system fitted with a Waters Prep Silica 500 column. Separation was performed by pre-equilibration of the column with C H a 3 / C H 3 O H / H 2 0 (60/30/3, v/v/v) then loading the total extract at 50 ml-mhr1 and eluting the lipids at 150 ml-min1 in the same solvent while collecting 200 ml fractions. Deterrriination of the lipid composition of these crude fractions was done by separating aliquots on analytical silica gel plates in acetone:benzene:H20 (91/30/8, v/v/v; Pohl et al., 1970) as below. DGDG purification The DGDG-enriched fraction from above was further purified on a CM-cellulose 25 cm x 1.5 cm (o.d.) column pre-packed in CHC13. Approximately 0.5 g of total lipid in 5-6 ml CHC13 was loaded on the column at 3.5 ml-min"1 with two column volumes of solvent Neutral lipids, pigments, MGDG, and PC were removed by washing with eight column volumes of 7% CH 3 OH in CHC13. Elution of DGDG and PE was accomplished with eight volumes of 12% CH 3 OH in CHC13. Removal of residual PE from the DGDG was performed on a circular, spinning preparative chromatography system (Chromatotron; Harrison Research, Palo Alto, CA) with a 2 mm silica adsorbent About 350 mg of total lipid was applied to the system and DGDG 35 eluted with acetone/benzene/H20 (91/30/8, v/v/v) at 3.0 ml-min-1 and collecting 0.5 or one minute fractions. MGDG purification The MGDG-enriched fractions from the liquid chromatography step were loaded onto the CM-cellulose column with two volumes of CHC13. Neutral lipids and pigments were removed with eight volumes of 0.5% CH 3 OH in CHC13 at 3.5 ml-min1. Recovery of M G D G and PC was accomplished by elution with eight volumes of 5% CH 3 OH as above. Removal of residual PC was performed on the Chromatotron system using a 2 mm silica adsorbent with CHCVacetone (1/1, v/v; Vorbeck and Marinetti, 1965) at 3.0 ml-min1. SQDG and PG purification Later- eluting fractions obtained from the liquid chromatography step routinely contained both SQDG and PG. The SQDG and PG enriched fractions were loaded onto CM-cellulose as described above in CHC13 then all other lipids removed by washing the column with 15% CH 3 OH in CHClj at 3.5 ml-min"1. PG was eluted with 5-8 column volumes of 18% CH 3 OH in CHC13 at the above flow rate. SQDG was subsequently eluted with eight volumes of 25% CH 3 OH in CHC13. Final purification of each lipid was done on the Chromatotron system with a 2 mm silica adsorbent using the acetone/benzene/H20 (91/30/8, v/v/v) system. 36 2.3. Lipid purity and fatty acid analysis Lipids were quantified and their fatty acid composition analyzed by GLC of fatty acid methyl esters. Aliquots of purified lipid were mixed with a known quantity of pentadecanoic acid, dried under N , , then methylated with 14% BF 3 in CH 3 OH at 100° C for 1 h. Methyl esters were extracted twice into hexane then analyzed by GLC on a Hewlett-Packard 5890A gas chromatograph fitted with a 30 m x 0.25 mm (i.&) 25% cyanopropylphenyl capillary column (J & W Scientific) and with a Hewlett-Packard 3390A integrator. Samples were "cold-trapped" by injection of 1 /til aliquots onto a column held at 120° C for two minutes before increasing the temperature at 150C-min-1 to 190° C for 12 minutes for fatty acid methyl ester separation. All samples were checked for longer-chain fatty acids by a further temperature ramp at 15°C-min1 to 220° C for 10 minutes. No lipids contained significant levels of fatty acids longer than 18 carbons. The chromatograph was routinely operated with injector and detector temperatures of 180° C and 250° C respectively, split ratios of 1/50 to 1/100, and a total flow rate of 1 ml-min1. Molar response factors for the fatty acids were calculated from the GLC analysis of known ratios of pentadecanoic acid to fatty acid standards. Yield of purified lipids were checked independently by weighing dried aliquots of the lipid stocks. Malondialdehyde was measured according to Buege and Aust (1978) and the extinction coefficient of 1.56-105 M _ 1-cm _ 1 confirmed after quantitative hydrolysis of authentic 1,1,3,3-tetramethoxypropane in 0.25 M HC1 for 30 min at 21° C (Bond et al., 1980). The presence of chlorophylls and/or carotenoids was assayed in 80% acetone at 645 and 663 nm (Arnon, 1949) as well as 400 and 440 nm respectively. Carotenoid content was calculated using an extinction coefficient of 253 g^l^-cnr 1 (Selstam and Sandelius, 1984). 37 The presence of other lipids was assayed by spotting 0.2-0.3 mg of purified lipids on analytical silica TLC plates (2.5 x 7.5 x 0.25 cm) and developing in either acetqne/benzene/H20 (91/30/8, v/v/v) or CHC1 3 /CH 3 0H/H 2 0 (65/25/4). After development the sides of the plate were spotted with amounts of soy PE corresponding to 0.1 to 5.0 mol% of the applied lipid. Plates were sprayed with the phospholipid reagent of Allen and Good (1971). With mild heating the amount of contaminating phospholipid could be estimated. Subsequent charring by intense heating of the plates allowed assessement of total contaminants by comparison with the applied PE standard. 2.4. Preparation of hydrogenated lipids Saturation of the acyl chains of DGDG and M G D G was done by dissolving 50 mg of lipid in 5 ml benzene and 50 mg of Adam's catalyst (Pt02). Suspensions were bubbled with H 2 at 40° C for 3 h then held under H 2 overnight at 40° C. Bulk catalyst was removed by filtration over glass fibre filters then residual catalyst and any degradation products were removed by preparative chromatography (Chromatotron) on silica using acetone/benzene/H20 (91/30/8, v/v/v) as described above. This procedure was a minor modification of that of Sen et al. (1981b). 2.5. " C - N M R spectroscopy 1 3 C - N M R spectra were obtained on a Bruker WP-200 wide bore spectrometer at 50.3 MHz with ' H broadband decoupling and full Nuclear Overhauser Enhancement Aquisition parameters were: minimum scans = 13,000; interpulse delay (Dl) = 3.8s; pulse width = I2us; spectral width = 11 kHz; and line broadening of 1 Hz. Samples of DGDG (136 mg), SQDG (78 mg), and MGDG (306 mg) were dispersed in CD 3 OD with tetramethylsilane 38 (TMS) as internal standard. 13C-NMR spectra for the hydrogenated samples of DGDG and MGDG were obtained from 30-45 mg lipid in CDjOD/CDCU (4/1 and 1.5/1, v/v, respectively) on a Varian XL-300 spectrometer at the NMR facility, Dept. of Chemistry, U.B.C. as described above but with DI = 0.4s. 2.6. *H-NMR spectroscopy "H-NMR spectrum were recorded on a Bruker WH-400 spectrometer at 400 MHz on ten mg samples of DGDG, MGDG, or SQDG in CD3OD without TMS. A minimum of 256 scans were collected at 18° C using a 6 /xs pulse width and spectral width of 4.8 kHz. Homonuclear ('H) chemical shift correlated spectra (COSY) were obtained for DGDG, SQDG, and MGDG samples on a Bruker WH-400 spectrometer, operating at 400 MHz as above, in the NMR facility, Dept. of Chemistry, U.B.C. This work was performed in collaboration with Drs Colin Tilcock and Kim Wong of the Dept. of Biochemistry, U.B.C. 39 3. RESULTS AND DISCUSSION 3.1. Yield and purity of extracted lipids The yields for the extraction and purification of thylakoid lipids are shown in Table 1. For the minor lipids SQDG and PG yields of approximately 100 mg were routinely obtained and levels of 600-700 mg of MGDG and about 800 mg of DGDG. These values represent 25-40% recovery based on the extraction of small portions of leaf material by a different method (Williams and Merrilees, 1970). A similar recovery of 39% was calculated for total chlorophyll (data not shown). Lipids were eluted at the following concentrations of %CHjOH in CHClj on CM-cellulose: 0% for chlorophyll and neutral lipids, 1% for MGDG, 4% for PC, 9% for PE and DGDG, 16-17% for PG, and 20-25% for SQDG. These results are in close agreement with those of Comfurius and Zwaal (1977) for the neutral lipids, PC, and PE, but this»PG eluted at 3-4% lower CH 3 OH than reported by the above workers. Interestingly, both MGDG and DGDG eluted 4-6% CHjOH earlier than their glucolipid counterparts (Comfurius and Zwaal, 1977). CM-cellulose was particularily good at resolving DGDG from SQDG. This is in contrast to the close migration of these lipids on silica-based systems (Allen and Good, 1971; Rouser et al., 1976; Christie, 1982). Unfortunately, several lipids adsorb strongly to CM-cellulose, requiring extensive washing to facilitate high recoveries of lipid. The recoveries of individual lipids was quite variable. The extract assayed before extensive chromatography showed some enrichment of DGDG and slight decreases of PG and M G D G (Table 1). It is possible, but unlikely, that this represents preferential enzymatic hydrolysis of some lipids during the extraction procedure since there was no increase in the neutral lipid component in the extract when compared to that found in leaves (Table 1). 40 TABLE 1. YIELD AND COMPOSITION OF SPINACH LIPID EXTRACTS. Total lipids were extracted from 2-3 g of spinach leaves by the method of Williams and Merrilees (1970) or from 2-2.5 kg of spinach leaves as described in the Methods. Aliquots of these extracts were separated by the ID TLC method of Khan and Williams (1977), lipids scraped from the plates and quantified by gas-liquid chromatography. Yield of purified lipid '. mol % predicted found Lipid leay£s extract (grams') (grams) (%i PC 4.8 ,7.2 -PE 2.7 2.8 - -SQDG 1.9 1.9 0.175 0.091 52 DGDG 26.7 42.0 2.18 0.850 39 PG 14.3 7.8 0.95 0.080 8 MGDG 41.7 33.3 2.84 0.710 25 NL 1 7.2 4.9 -deludes free fatty acids, diacylglycerols and triacylglycerols 41 TABLE 2. FATTY ACID COMPOSITION OF PURIFIED THYLAKOID LIPIDS. Fatty acid methyl esters of purified lipids were prepared with BF3/methanol and analyzed by gas-liquid chromatography as described in the Methods. Data are means (±S.D.) of 2-5 independent determinations. Lipid mol% fattv acids 16:0 16:1» 16:3 18:0 18:1 18:2 18:3 42.9 0.7 1.0 1.2 0.5 3.7 49.6 (2.8) (1.0) (0.7) (0.2) (0.4) (0.6) (3.3) SQDG DGDG 7.8 - 3.4 0.7 1.2 2.5 84.3 (2.2) (0.8) (0.1) (0.4) (0.5) (2.9) PG 19.6 35.5 - 2.1 1.1 3.8 37.7 (2.0) (1.2) (1.7) (0.4) (0.1) (1.9) MGDG 2.0 0.1 22.7 0.7 0.5 1.5 72.6 (1.5) (0.1) (4.2) (0.2) (0.4) (0.8) (1.5) DGDG 13.8 - - 83.0 3.1 (saL) (1.2) (1.8) (0.6) MGDG 25.5 - - 72.9 1.6 (saL) (0.6) (0.7) (0.2) 'trans-AS-hexadecanoic acid 42 TABLE 3. PURITY OF ISOLATED THYLAKOID LIPIDS. Purity of isolated lipids was determined by analytical TLC as described in the Methods using soy PE as a standard and the phospholipid spray reagent of Allen and Good (1971) for detection. Chlorophyll and carotenoids were quantified in 80% acetone using the extinction coefficients of Arnon (1949) and Selstam and Sandelius (1984) respectively. mol% contaminating major reproducible acyl contaminating mol% mol% purity lipid lipid chlorophyll carotenoids (mn\%) lipid D G D G MGDG ad. ad. none none ad. £0.1 ad. £0.01 99.7 99.7 SQDG PG 1-2 0.2-0.5 PE± DGDG ad. DGDG ad. ad. ad. 98 99.5 ad.: not detectable 43 It is likely that the major decrease in yield, and preferential loss of some lipids, occurred at the extraction and CHClj-phase separation steps as a result of the large scale of the operation. • The fatty acid compositions of the purified lipids (Table 2) were very similar to those reported by other workers for lipids isolated from spinach thylakoids (Allen and Good, 1971; Murphy and Woodrow, 1983; Farineau et al., 1984). Data in Table 2 also indicates the quantitative saturation of fatty acyl residues by catalytic hydrogenation with Pt0 2. Levels of purity of the extracted lipids (Table 3) were very high. Contamination by chlorophyll or other pigments was virtually non-existent Only the MGDG extract displayed a slight yellow color that was identified as a carotenoid contaminant of £0.01 mol%. Other polar lipids were also absent to £0.25 mol% with the exception of SQDG which contained small amounts of DGDG. This resulted from extensive DGDG tailing on CM-cellulose and the lack of separation of these lipids on subsequent silica chromatography. Oxidative degradation of the lipids was monitored by assay for malondialdehyde (MDA), a common product of lipid peroxidation (Buege and Aust, 1978). All lipids showed levels of 1.5 to 3.8 nmoles MDA/Mmole lipid and these values did not change during subsequent lipid storage at -20°C or -80°C under N 2 with 0.05% BHT. Further, these values were similar to control values of 1.7 nmoles M D A / M moles lipid obtained for polar plant lipid dispersions and were far lower than values of 33.1 nmoles M D A / M mole lipid for fully peroxidiied thylakoid lipids (Galanopoulou et al., 1982) 3.2. 1 3 C - N M R of thylakoid glycolipids Unexpected behavior of vesicles of purified DGDG (Chapter 3 and Webb et al., 1988) suggested the occurrence of head group derivatization during the lipid extraction. The identity H U H , , - ^ C H ] O H H 1 0 \ O H J[ H 9 . . . " n H5 HO H4 H3 °^rosr?cj c i c i N — ^ / V H , — C H — C H , " H J O H H I ? ? C - O C - 0 <CH ,h (CH, ) , CM CM CH CH C H , CHj CH CH CH CM C H , C H , O l CH CH CH C H , C H , CHj C H , C - 0 CH M CH solvent C I - C3 H 2 - H 6 H 8 - H 1 2 H7 HI JLA. TMS 200 150 100 ppm 50 Figure 3. Proton-decoupled 1 3 C - N M R spectrum of DGDG in CD 5 OD. 45 and structure of the glycolipids was, therefore, confirmed by 13C-NMR. The broadband proton-decoupled 13C-NMR spectra of DGDG (Figure 3) was identical to that previously reported by Johns et al. (1977a). No peaks were shifted downfield, as would be expected for head group derivatization, nor were any contaminating peaks found. Two small resonances at 23.7 and 33.1 ppm were attributed to A 1 4 and A15 carbons based on the appearance of these peaks in the SQDG spectrum (Figure 5). Palmitic acid is present in small amounts in DGDG (8 mol%) compared to SQDG (43 mol%) (Table 2). The l3C-NMR spectrum of MGDG (Figure 4) was also identical to that reported previously (Johns et al., 1977a) and there was no evidence of derivatization nor contamination of this lipid. Small additional resonances were observed in the acyl region of the spectrum at 28.2 and 131.0 ppm. These peaks likely arise from carbons A7, A10, and A13 of 16:3. These carbons could not be assigned in the original spectrum (Johns et al., 1977a) as the MGDG from the examined species do not contain 16:3. These assignments were completely consistent with the "C-NMR spectra obtained for the hydrogenated lipids DGDG and MGDG (spectra not shown). In both cases the only effect of hydrogenation was to eliminate completely all resonances between 128 and 133 ppm. These peaks are due to the presence of unsaturated bonds in the methylene chain. The "C-NMR spectra of SQDG (Figure 5) was also identical to that published previously (Johns et al., 1978) and shows the diagnostic 54.4 ppm resonance due to the H6 sulfonate group. Qose comparison of this spectrum to that of Johns et al. (1978) revealed the presence of an unassigned resonance at 67.3 ppm and a peak at 72.3 ppm not found in this system. The J-resolved spectrum (Figure 6) was obtained in an attempt to sort out these assignments. The J-resolved spectrum is plotted with quaternary carbons and methylene groups H5 H6 \ CHjOH H O V — — 0 i-O C3 < H4 SOH AcH,—C H3 H 2 0 H ^ CI - C H , HI C2 C H -O O C - 0 C - 0 (CHjh {CHih CH CH CH, CH CH CH CM CH, CH CH i CH, CH CH CHl at CH 9"i CH, O O 1 CH * CH HI SOlvCTl H 3 H5 H4 H6 CI C3 TMS - C H , — CHi — KJ li / 170 130 — i — 9 0 ppm 5 0 -i i— 10 Figure 4. Proton ,-decouplcd " C - N M R spectrum of MGDG in CD,OD. H5 H6 CHjSCV H4 HO TO 9 c C H j -ill C2 CI - C H — -CH 2 0 9 C-O C-0 (CHj)7 JCH2) CH 2 CH CH 2 CH | CH 2 CHj CH 2 CH CHj CH 9"2 9*1 CHj CH CHj CH CH 2 CHj 1 solvent i C1-C3 H2-H5 i CH R CH H I H 6 CHJSOJ-1 8 0 1 4 0 1 0 0 6 0 2 0 p p m Figure 5. Proton-decoupled , J C - N M R spectrum of SQDG in CD,OD. I 1 1 1 1 1 1 1 r 180 140 100 60 20 ppm Figure 6. Proton-decoupled J-resoWed "C-NMR spectrum or SQDG in CDjOD. Spectrum is plotted with quaternary carbons and methylene carbons oriented upwards, methine and methyl' carbons downwards. *. or 49 oriented upwards, and CH and methyl groups downwards. This spectrum clearly indicated that carbons resonating at 64.5 and 67.3 ppm were methylene groups and most likely represent Q and C 3 of glycerol. This change suggested that the 69.9 ppm resonance originally assigned to C 3 Of glycerol probably represents Q . It has been assumed that the remaining head group carbons Hj to H 6 have retained their original assignments. However, in a later correction of the original SQDG spectrum (Johns, 1980) the C2 group was assigned downfield to 71.9 ppm and Hi upfield to the 69.9 ppm peak. This assignment is not inconsistent with the J-resolved spectrum (Figure 6). However, the general approach of constructing a complete spectrum from spectra of fragments of the whole molecule (Johns et al, 1977b) can easily lead to incorrect assignments of resonances in the final structure. The changes discussed above for SQDG suggest that similar assignments might be appropriate for the structures of MGDG and DGDG. Such rearrangements would be entirely speculative and likely be incorrect Further work using 2D NMR spectroscopy is currently in progress to assign unambiguously the " C - N M R resonances in the glycerol and head group regions. 3.3. l H - N M R of thylakoid glycolipids The 400 MHz broadband ' H - N M R spectrum of MGDG is shown in Figure 7. Assignments of fatty acyl protons were routine assignments based on previous work and were consistent with COSY results reported below. All resonances for the galactose and glycerol protons occur between the two solvent peaks at 3.35 and 4.80 ppm. Assignments for these resonances were done by a series of homonuclear single frequency off-resonance decoupling experiments (spectra not shown). In addition, assignments were confirmed by homonuclear OH) chemical shift correlated NMR spectroscopy (COSY) experiments and the contour spectrum for M G D G shown in Figure 8. The spectrum could, for interpretation purposes, be viewed simply as the ID 1 H-NMR spectrum plotted in two dimensions. solvent — C H - C H j - C H — —CHj —CHl,— —CH 2 -CH=CH—CH 2 — ppm Figure 7. 400 MHz 'H-NMR spectrum of MGDG in CD,OD. o 51 ppm L 4 8 L 4 5 L 5 .8 L 5 .5 J_ 6. B • I • i • • I • • • • I • • • - • i • 5 5 5 e 4 . 5 4 . B 3 5 J . B 2 .5 2.8 1.5 I B •$ ppm Figure 8. 400 MHz 'H homonuclear chemical shift correlated NMR spectrum (COSY) of MGDG in CD 3OD. Inset area is shown in Figure 9. 52 Figure 9. 400 MHz ' H homonuclear chemical shift correlated NMR spectrum (COSY) of MGDG in CD 3 OD expanded in the galactose and glycerol regions. 53 As a result, the main diagonal line is an autocorrelation line and peaks off the main diagonal are crosspeaks. Crosspeaks connect protons that resonate at different frequencies (ie. different ID ppm values) but that are bridged by two intervening carbons (Schraml and Bellama, 1988). Therefore, protons of adjacent carbons will share the same crosspeak. An example of assignments by this method for the fatty acyl protons of MGDG is shown in Figure 8. An expanded view of the head group region from 3.35 ppm to 4.5 ppm is shown in Figure 9 with the same assignments as in Figure 7. The broadband 400 MHz l H - N M R spectrum of SQDG is shown in Figure 10 and assignments given. The assignments of fatty acid proton resonances were identical to those described above for MGDG. Assignments of protons from the quinovosyl and glycerol carbons were done by a series of homonuclear single frequency off-resonance decoupling experiments (spectra not shown). In addition, assignments were confirmed by a homonuclear (*H) COSY experiment as described above for M G D G (Figure 11). Assignments in the head group and glycerol region (Figure 12) were complicated by the appearance of more peaks than can be accounted for by the accepted structure of SQDG (Figure 1). That is, there should be 9 groups of proton splittings arising from the 6 quinovosyl carbons and the 3 glycerol carbons (a sharp singlet located downfield at 7.9 ppm arising from the sulfonate proton was not observed on this spectrum) while an examination of Figure 12 should convince the reader of the existence of at least 10 major multiplets. It has been assumed, therefore, that there has been some solvent interactions with SQDG generating additional peaks on the 2D COSY spectrum. The possibility of interaction with the CD 3OD solvent is suggested by the observation that SQDG was not fully soluble in CD 3OD. The NMR tubes contained some precipitated lipid. These additional peaks have been ignored (Figure 12). The assignments in Figure 12 were based on the assignment of the 4.77 ppm doublet to Hi and the 4.5 ppm Figure 10. 400 MHz 'H-NMR spectrum of SQDG in CD,OD. 55 CH • CH i P-0 ~CHj(l*:C) / -CHjdi s)qP CHj —ft / ? s ° C-CH2~ CH?- / - C B i - C H s O j - C H z — 0 O-C - CHj-CH--•CH - C H z - C H — / ID O 88 O * o . 0 8 e .a I ' — i — ' -1—>—•—'—«—r -68 M l.a ppm 2.e l >—i—r I 2 J. i. 3.2 ppm . 3.5 - 4.e Figure 11. 400 MHz »H homonuclear chemical shift correlated NMR spectrum (COSY) of SQDG in CD 3OD. Inset area is shown in Figure 12. 56 Figure 12. 400 MHz ' H homonuclear chemical shift correlated NMR spectrum (COSY) of SQDG in CD,OD expanded in the head group and glycerol regions. 57 quartet to Q . The doublet at 4.77 ppm can be unambiguously assigned to H, as this is the only proton spin-spin coupled to one other proton, and the d assignment is consistent with that in both DGDG and MGDG. The SQDG spectrum shows two separate methyl group triplets centered at 0.97 and 0.90 ppm. Identical integral intensities for these triplets suggested that each triplet originated from a separate fatty acid chain. This is consistent with the near 1:1 mixture of 16:0 and 18:3 in the SQDG fatty acid profile (Table 2). It is likely that the upfield triplet at 0.90 ppm originated from the 16:0 moiety of SQDG since it shows greater line broadening than that at 0.97 ppm, indicative of lower molecular motion. Furthermore, one sharp methyl group triplet was observed from MGDG (Figure 7) and a small additional triplet in the DGDG ' H - N M R spectra (Figure 13). MGDG has only 2.0 mol% 16:0 while DGDG contains 8 mol% 16:0 (Table 2). This prediction was confirmed by the COSY experiment showing the upfield methyl group triplet, but not the downfield triplet, to be connected to the CH 2 group (Figure 11). The 400 MHz broadband l H - N M R spectrum of DGDG is shown in Figure 13 with routine fatty acyl proton assignments based on the COSY spectrum (Figure 14) and in agreement with those for SQDG and MGDG. Expansion of the digalactosyl and glycerol regions of the COSY spectrum (Figure 15) revealed a pattern far more complex than that obtained for MGDG (Figure 9). The indicated assignments for Ci, C 2 , and C 3 of glycerol were identical to those for SQDG and MGDG. However, it is apparent that proton resonances from the additional galactose moiety were not simply superimposed on those of MGDG. This is illustrated by the group at 3.5 ppm which contained C 3 . as in MGDG, and only one other proton as indicated by the presence of one additional crosspeak (Figure 15). In MGDG (Figure 9) the 3.50 ppm group contained C 3 , H 2 , and H 4 . Simple superposition of resonances would have predicted that the 3.50 ppm group of DGDG to contain at least these 58 three resonances from MGDG, plus any others in the same region from the additional galactose. The complexity of the multiplets in the two regions at 3.7 and 3.9 ppm of the DGDG spectrum has made assignments within these groups impossible, even using 2D NMR methodology. - C M - C H j - C H — Figure 13. 400 MHz ' H - N M R spectrum of DGDG in CDjOD. 60 Figure 14. 400 MHz *H homonuclear chemical shift correlated NMR spectrum (COSY) of DGDG in CD 3OD. Inset area is shown in Figure 15. 61 Figure 15. 400 MHz l H homounclear chemical shift correlated NMR spectrum (COSY) of DGDG in CD 3 OD expanded in the digalactosyl and glycerol region. 62 3. SALT MEDIATED INTERACTIONS BETWEEN VESICLES OF DGDG 1.- INTRODUCTION The thylakoid lipids exist in a membrane system which, in vivo, shows large areas of close membrane approach (grana). The interaction of neighbouring thylakoid membranes, referred to as "stacking'', is believed to be mediated by interactions of units of the chlorophyll a/b light-harvesting complex (LHC II) (Staehelin, 1986). It has been proposed that the role of the galactolipids in this interbilayer interaction is merely as a neutral, non-charged, lipid matrix serving to maintain a low surface charge density and, hence, rninimize electrostatic repulsion between adjacent bilayers (Murphy, 1986b). While there have been some suggestions for a role of thylakoid PG and its unique fatty acid trans-A3-hexadecanoic acid in granal stacking (Dubacq and Tremolieres, 1983) this seems unlikely in view of genetic evidence showing normal stacking in a mutant of Arabidopsis thaliana lacking trans-A3-hexadecanoic acid (Browse et al., 1985) and reduced stacking in a barley mutant with normal trans-A3-hexadecanoic acid levels (Bolton et al., 1978). This chapter describes the characterization of well-defined DGDG vesicles and their behavior in dilute salt solutions. Further, evidence is presented suggesting that DGDG may play a role in close membrane approach in thylakoids. 63 2. MATERIALS AND METHODS 2.1. Materials All solvents (BDH reagent grade) were redistilled before use. Other compounds were obtained as follows: galactose and NaCl from BDH; CaCl 2, MgCl 2, K Q and glycerol from Amachem; NaBr from Anachem; NH4C1, NaC104, glucose, and succinate from Fisher; CsCl from Calbiochem; RbCl from ICN; LiCl from Baker-Adamson; NaNO,, NaSCN from MCB; oleic acid, egg phosphatidylethanolamine, egg phosphatidylcholine, dioleylphosphatidic acid, dipalmitoylphosphatidylglycerol, valinomycin, and tricine from Sigma. Atomic absorption analysis of the distilled water from this lab has indicated metal levels <10 ng-ml"1 of Cu, Mn, and Zn. 2.2. Lipid Purification The purification of DGDG and SQDG is described in detail in Chapter 2. 2.3. Vesicle reconstitution Aliquots of lipids in CHC13 were dried under a stream of N 2 and residual solvent removed under reduced pressure overnight at 4°C. Lipids were dispersed at 10 mg-mf1 either by vortexing and sonication (20-30 seconds total in five second bursts followed by cooling on ice, under N 2 ; Branson bath sonicator) or by a reverse-phase evaporation method using Freon-11 (Sprague and Staehelin, 1984a,b). Resultant dispersions were converted to large unilamellar vesicles by repeated extrusion (The Extruder, Lipex Biomembranes) through two stacked 0.1 Mm Nucleopore polycarbonate filters at 2000 kPa N 2 (Hope et al., 1985). When DGDG was dispersed in salt solutions, higher N 2 pressures, up to 4500 kPa, were required 64 for extrusion due to vesicle aggregation. No differences in aggregation properties were observed using the reverse-phase or vortexing/sonication methods of lipid dispersal. For the standard curve of turbidity (A 6 0 0) vs. vesicle size, vesicles in H 2 0 were extruded through 0.4 am filters at 350 kPa and an aliquot removed for A 6 0 0 reading. The remainder was extruded at 0.2 Mm (350 kPa), then 0.1 (2000 kPa), then 0.05 Mm (2500 kPa) and finally 0.01 Mm (4500 kPa) (Mayer et al., 1986) and assayed for A^o. For the valinomycin control, the valinomycin in CHCl, was mixed with DGDG in CHClj at 1 Mg'Mmole1 DGDG (1113:1, mohmol), solvent removed as above and vesicles dispersed in water. 2.4. Turbidity measurements Turbidity readings at 600 nm were obtained on a Varian Cary 210 spectrophotometer using non-aggregated 100 nm vesicles in H 2 0 as blanks. Vesicles were diluted to 1 mg-ml'1 with H 2 0 , then small amounts of concentrated salt solutions added and A«0o recorded immediately. 2.5. X-ray diffraction X-ray powder patterns were recorded using Cu Ka radiation generated on a Rigaku RU-200 microfocus rotating anode X-ray machine coupled, via Franks optics, to the Princeton SIV area detector (Gruner, 1977; Reynolds et al., 1978; Gruner et al., 1982a,b; Tilcock et al., 1984). Radial densitomerization of the images was performed as previously described (Gruner et al., 1982b; Tilcock et al., 1984). Data are presented as the basis vector length of the lamellar lattice versus temperature. Repeat spacings were calibrated against lead stearate and are accurate to ±0.05 nm. 65 DGDG was dispersed at 50% (w/w) in 5 mM EDTA, 100 mM KC1, or water and loaded into acid-cleaned X-ray capillaries which were sealed with an epoxy plug. Samples were ramped in temperature from 0°C to 50° C in 10° C steps with a 10 minute equilibration at each temperature. Diffractions were typically collected over a 60 second exposure. 2.6. Quasi-elastic light scattering (QELS) Vesicle and aggregate diameters were estimated by quasi-elastic light scattering (QELS) using the Nicomp Submicron Particle Sizer 270 at lipid concentrations of 0.1, 1.0, or 10 mg-ml1. Samples were irradiated by a 5 mW Helium-Neon laser at 632.8 nm and the autocorrelation function converted to mean vesicle diameter as described previously (Mayer et al., 1986). 2.7. Freeze-fracture electron microscopy Vesicles at 10 mg-ml"1 in water were aggregated by the addition of salt to 100 mM KC1, 10 mM MgClj, or left in water, then made to 20 or 25% glycerol (v/v) and frozen in Freon-22 cooled with liquid nitrogen. Samples were fractured in a Balzers 400 instrument at -107°C and <L 8-10"7 torr then coated with 2.0 (±0.1) nm Pt followed by 20 (±4) nm carbon as measured by a quartz crystal thickness monitor. Replicas were cleaned with commercial bleach or 70% chromic acid for several hours, rinsed, and examined in a Zeiss EM-10 electron microscope at 80 kV accelerating voltage. 66 3. RESULTS 3.1. Aggregation of large unilamellar vesicles by salt The addition of salt to hydrated vesicles at lipid concentrations of 10 nig-ml-1 was observed to result in very rapid appearance of a cloudy, readily precipitable, particulate suspension. As shown in Table 4, aggregation could be obtained either by dispersing the lipid directly into salt solutions, or by the addition of salt to vesicles suspended in water. The aggregation process was found to be fully reversible (Table 4). That is, 100 nm DGDG vesicles in water could be aggregated by the addition of KC1 or NaCl to yield particles of more than 1000 nm diameter, and the original 100 nm vesicles could be released by subsequent 1:10 to 1:100 dilution with water. This release is easily observable by light microscopy. Vesicle appression followed by fusion would be expected to yield vesicles with diameters larger than 100 nm, and it would be unlikely for vesicles with the original size distribution to be formed upon dilution. This result strongly suggested that the structure of the aggregate was of appressed, but not fused, vesicles. This was confirmed for DGDG aggregation in 100 mM K.C1 and 10 mM MgCU by freeze-fracture electron microscopy (Figure 16). DGDG vesicles in water (Figure 16a) showed typical vesicle appearance for liquid-crystalline phase lipid dispersions. Fractured lipid surfaces were smooth, and vesicles were never observed to be in direct contact Measurement of vesicle diameters from micrographs gave values in the 100-120 nm range for equatorial fractures. This was verified by a QELS determination of 121 (±30) nm diameter for this preparation. Similarily, the bilayer surface appearance of DGDG vesicles aggregated in 100 mM KC1 (Figure 16b) or 10 mM MgCl 2 (Figure 16c) was identical to that of DGDG in water. Single vesicles in aggregates had equatorial diameters between 80 and 120 nm. No vesicles were 67 TABLE 4. QUASI-ELASTIC LIGHT SCATTERING (QELS) DETERMINATIONS OF THE DIAMETERS OF DGDG VESICLES DISPERSED IN VARIOUS SALT SOLUTIONS. DGDG was dispersed at 10 mg-ml-1 in 0.1 M NaCl or 0.1 M KC1 and extruded to give 100 nm diameter large unilamellar vesicles. Diameters were measured by QELS before and after 1:10 or 1:100 dilution with water. Vesicles dispersed in water were extruded to 100 nm, sized by QELS before and after addition of salt, then diluted with water and diameters measured again by QELS. Data represents mean ±S.D. from representative experiments of 5-8 trials. vesicles made in Diameter (nm) in H,0 outside soln made to after salt addition Diameter fnm^ after 1:10 or 1:100 dilution H 2 0 H 2 0 0.1 M KC1 0.1 M NaCl 121 (±21) 121 (±20) 0.1 M K Q 0.1 M NaCl 0.1 M K Q 0.1 M NaQ >1000 >1000 >1000 >1000 103 (±23) n.d. 104 (±23) 129 (±32) n.a.: not applicable. n.d.: not determined. 68 200 nm 100 nm 100 Figure 16. Election micrographs of freeze-fracture replicas of 100 nm DGDG vesicles at 10 mg-ml1 dispersed in water (16a) or dispersed in water then aggregated by the addition of salt to give 100 mM K.C1 (16b) or 10 mM MgCl 2 (16c). Samples were diluted to 25% (16a,b) or 20% (16c) (v/v) glycerol before freezing from 22° C. Bars represent 200 nm (16a) or 100 nm (16b and 16c). 69 observed to be significantly larger than this range, confirming the lack of vesicle fusion, despite close membrane approach. These preparations showed several striking features. First of these was the close adhesion of the vesicles. While it is beyond the resolution of freeze-fracture to determine if adjacent polar surfaces were in direct contact, it is clear that bilayers are stable within several nanometers of each other. A second interesting aspect was the extensive vesicle flattening. It is possible that this results, in part, from osmotic stress due to external salt addition. A consequence of this flattening is the small radius of curvature at the ends of some vesicles (Figure 16b) that was estimated at 5 nm. Since the salt-induced aggregation of an uncharged lipid was an unexpected finding, it seemed likely that there might be a charged impurity in the DGDG preparation that was behaving in accordance with electrostatic double layer theory. To test this, 2.0 or 2.5 mol% of possible contaminants were added to egg PC, the mixture dispersed at 10 mg-ml"1 in 100 mm KC1, extruded at 100 nm and vesicle or aggregate diameters measured by QELS. As shown in Table 5, egg PC with 2.0 mol% egg PE, oleic acid (18:1), dioleoylphosphatidic acid (DOPA), dipalmitoylphosphatidylglycerol (DPPG) or 2.5 mol% SQDG did not aggregate in 100 mM KC1. Further, pure SQDG dispersed in 100 mM KC1 or 10 mM MgClj did not aggregate. Gas-liquid chromatography analysis of the fatty acid methyl esters of DGDG re-extracted after the experiments indicated that no significant change in the fatty acid profile of the DGDG occurred during the experiments. Thin-layer chromatography of the re-extracted DGDG showed only one compound in these samples. Therefore, it is unlikely that breakdown products of the galactolipid were causing vesicle aggregation. The lamellar repeat distances for DGDG in water, 100 mM KC1, and 5 mM EDTA have been examined by X-ray diffraction. As seen in Table 6, there was no difference in 70 TABLE 5. QUASI-ELASTIC LIGHT SCATTERING (QELS) DETERMINATIONS OF DIAMETERS OF VARIOUS LIPID MIXURES DISPERSED IN SALT SOLUTIONS. Lipid mixtures were dispersed in the solutions indicated and extruded to make 100 nm unilamellar vesicles. Diameters were measured by QELS as in Table 1. In the DGDG and valinomycin experiments the vesicles were dispersed at 10 mg-ml-1 in water, extruded to 100 nm, diluted 1:10 with water and sizes measured by QELS. Concentrated K Q was then added to a final concentration of 100 mM KC1 and sizes measured again by QELS. Data represent means ±S.D. from representative experiments. addition [lipid] lipid (mol%) fmyml-'l salt Diameter (nm) egg PC none 10 100 mM KC1 117 (±24) egg PC 2% egg PE 10 100 mM KC1 121 (±37) egg PC 2% 18:1 10 100 mM K.C1 125 (±35) egg PC 2% DOPA 10 100 mM KC1 124 (±36) egg PC 2% DPPG 10 100 mM KC1 127 (±36) egg PC 2.5% SQDG 10 100 mM KC1 129 (±46) SQDG none 10 H 2 0 119 (±34) SQDG none 10 100 mM KC1 104 (±26) SQDG none 10 10 mM MgCl 2 100 (±32) DGDG DGDG DGDG DGDG none none valinomycin valinomycin HjO 100 mM KC11 H 2 0 100 mM K Q 1 125 (±30) >1000 109 (±28) >1000 1 added outside vesicles. 71 TABLE 6. X - R A Y DIFFRACTION MEASUREMENT OF THE LAMELLAR SPACING OF DGDG IN WATER, 100 mM KC1, OR 5 mM EDTA. DGDG was dispersed at 50% (w/w) in either water, 100 mM KCI, or 5 mM EDTA and X-ray powder patterns recorded from 0° to 50°C. Data represent the lamellar lattice repeat distance and are accurate ± 0.05 nm. lamellar sparing fnm't 100 mM 5 mM Temperature (°C) Water KCI EDTA 0 5.17 5.21 5.22 10 5.20 5.19 5.21 20 5.15 5.17 5.19 30 5.14 5.12 5.15 40 5.13 5.13 5.12 50 5.10 5.07 5.12 72 the DGDG lamellar repeat for any of these solutions between 0°C and 50° C. If a charged contaminant were present, the lamellar spacing should have decreased in the presence of salt due to charge screening. This strongly suggested that aggregation was not due to a charged contaminant. In summary, these and other reasons (see Discussion this Chapter) have led to the conclusion that aggregation was not due to charged impurities of the DGDG. Another possibility was that aggregation might have been due to the generation of a diffusion potential across the bilayers. While aggregation occurred with 100 mM KC1 both inside and outside the vesicles (Table 4) this possibility was further checked by the addition of valinomycin to DGDG vesicles. The presence of valinomycin did not inhibit KC1-mediated aggregation (Table 5). We have also verified that osmotic effects were not causing aggregation by the external addition of 0.2 M glucose or galactose. Neither sugar triggered vesicle precipitation (not shown). 3.2. Effect of different salts on aggregation The efficacy of different salts in causing DGDG vesicle aggregation has been examined. Quasi-elastic light scattering has been shown (Hope et al., 1985) to give very accurate size determinations for vesicles in the 30 to 200 nm range when compared to measurements by freeze-fracture electron microscopy. However, the large, polydisperse, and unstable nature of the aggregates made unambiguous size determination by QELS difficult Instead, turbidity (Atoo) was used as a measure of the degree of aggregation by light scattering. The variation of A«oo with vesicles of defined size is shown in Figure 17. Optical density was seen to vary sigmoidally, rather than linearly, with increasing particle size. 73 Figure 17. Variation of optical density at 600 nm with mean diameters of DGDG vesicles extruded to various sizes in water. Values are representative data from two experiments. 74 Monovalent cation concentration, mM Figure 18. Variation of A«oo (light scattering) during the sequential addition of divalent (18a) or monovalent (18b) chloride salts to 100 nm DGDG vesicles dispersed in water. In Fig 18a the CaJ* (•) and MgJ* (O) salts are shown. In Fig 18b the salts are Cs* (A), Rb* (X), NFL* (o), K* (o), Na* (•), and Li* (•). Data show the means of three or four replicates from a representative experiment. 75 The increase of optical density was nearly linear for vesicles with diameters between 50 and 200 nm, but leveled off with vesicles above 200 nm diameter. Since the starting point was 100 nm vesicles, this indicates that turbidity can be used as a sensitive measure of the stable aggregation of the first few vesicles. Turbidity changes during the titration of DGDG vesicles with various chloride salts are shown in Figure 18a for Ca2* and Mg2* and in Figure 18b for the monovalent salts. For most salts the turbidity increased sharply within a narrow concentration range of 2-3 mM for the divalent salts and 5-20 mM for the monovalent salts. Aggregation was strongly dependent on the ionic species within a valence group. Plotting of the ion concentration required for 50% of total Atoo increase (AA 5 0) against hydrated ionic radii (Figure 19) shows that cation efficacy was strongly correlated to ionic radius. The most effective cations were those with small hydrated radii or large crystal radii (Cs* and Rb*), while the least effectiye were those with large hydrated radii or small crystal radii (Na* and Li*). This relationship holds for Ca2* and Mg2* chloride salts (not shown). It should be added that the standard deviations of the turbidity readings, omitted for clarity, were usually 10-15%, yielding standard deviations of AAJO of about 1 mM. The possibility that anions could be involved in aggregation was also examined. Turbidity changes during the titration of DGDG vesicles with various sodium salts are shown in Figure 20. Of those examined, CI" was the strongest aggregation-promoting anion. Replacement of CI" with NO,", B r , C104", or SCN" led to higher salt concentrations being required for aggregation. Aggregation was never observed with the SCN" ion up to 100 mM under the conditions described here. An analysis similar to that in Figure 19 for the anions was not possible due to the lack of available ionic radii for polyatomic anions. 76 6-4-2-0 ' i i i i i i 0.32 0.33 0.34 0.35 0.36 , 0.37 0.38 0.39 Hydrated ionic radius, nm Figure 19. Relationship between ion concentration required for 50% of maximal A«oo increase (AAJO) and hydrated ionic radii of various monovalent chloride salts. The A A J 0 values were interpolated from Fig. 18b, the radii data from (Conway, 1981). Symbols are the same as given in Fig. 18b. 77 Monovalent anion concentration, mM Figure 20. Variation of A«oo during sequential addition of monovalent sodium salts CI" (•, data from Fig. 18b for comparison), NCy (O). Br (•), CICv (•). and SCN" (A) to 100 nm DGDG vesicles dispersed in water. Data represent means of three or four replicates from a representative experiment 78 6 0 -< 4 0 -2 0 -8 —r— 10 T-12 Addition, (v/v) —!— U —r-16 1B 20 Figure 21. Variation of A«oo (light scattering) during the sequential addition of water (O) or glycerol (©) to 100 nm DGDG vesicles at 1 mg-ml1 in water aggregated by the addition externally of 10 mM MgClj. Data shows the means of three or four replicates from a representative experiment 79 MgCla concentration, mM Figure 22. Effect of pH on the concentration of MgClj required to aggregate DGDG vesicles. Lipid was dispersed at 10 mg-ml1 in water then diluted with succinate-HQ (pH 3.5) (•), succinate-HQ (pH 5.0) (O), or or Tricine (pH 7.5) (•) to final buffer concentrations of 1 mM and lipid concentrations of 1 mg-ml"1. Data shows means of three or four replicates from a representative experiment 80 During the preparation of samples for freeze-fracture electron microscopy it was noticed that the addition of glycerol as a cryoprotectant caused a clearing of the normally opaque nature of aggregated vesicles. The possibility that glycerol might inhibit or reduce aggregation was, therefore, examined. The addition of glycerol resulted in disruption of MgCl2-induced aggregates (Figure 21). A similar effect was observed with 100 mM KC1-induced aggregates (not shown). Addition of identical volumes of water showed that observed A«oo decreases in glycerol' were not due to dilution of scattering particles. This effect of glycerol explains the apparent discrepancy between QELS determination of aggregate diameters as >1000 nm for DGDG in KC1 or MgCl 2 (Table 4) and the freeze-fracture micrographs indicating aggregate diameters in the 200-500 nm diameter range (Figure 16b and 16c) in the presence of 20-25% glycerol. The effect of changing the bulk solution pH on MgCl 2 concentrations required for vesicle aggregation is shown in Figure 22. There was no aggregation in the absence of Mg J + at any pH. Decreasing the pH from 7.5 to 5.0 decreased the A A J 0 for MgCl 2 from 6 mM to 2 mM. Further pH decrease to 3.5 correlated with a decrease of A A 5 0 to 1 mM. It is clear that MgCl 2 - induced aggregation occurred in DGDG vesicles over the pH range experienced by thylakoids in viva The small decrease of A A J 0 between pH 5.0 and pH 3.5, a range that brackets the pKa of thylakoid membranes (Barber, 1980), suggests that charge neutralization of surface ionizable groups in this pKa range was not responsible for aggregation. 81 4. DISCUSSION This Chapter has presented evidence showing that large unilamellar vesicles of the neutrally charged thylakoid galactolipid DGDG aggregate in the presence of a variety of salts. Since this has not been reported previously, and the result was somewhat surprising, it was important to eliminate charged impurities as the causal agent in aggregation. Several lines of evidence support this conclusion. Firstly, no phospholipid contaminants could be detected in the DGDG preparation at greater than 0.25 mol% by phospholipid assay (Chapter 2). No other compounds could be detected at about 2 mol% by UC-NMR and thin-layer chromatography (Chapter 2). Secondly, the addition of plausible contaminants to egg PC bilayers did not trigger aggregation (Table 5), nor did pure SQDG vesicles aggregate under conditions used here. Thirdly, the fact that vesicle aggregation at pH 3.5 still required the addition of salts (Figure 22) suggests that charge neutralization or charge screening by protons was not a prerequisite for aggregation. Charged contaminants with pKa's above 3.5, once neutralized by protons, would be expected to aggregate without salt addition in this pH range. Fourthly, electrostatic double layer theory holds that charge screening is valence-dependent, but ionic species-independent within a valence group (Barber, 1980, 1982). A strong dependence on the ion species within both monovalent and divalent cation groups (Figure 19) and the monovalent anions (Figure 20) has been shown. Furthermore, the efficacies of the monovalent cations shown here does not follow that known for the association of these ions with charged phospholipid membranes (Hauser et al., 1976). Thus, these vesicles were not behaving as if a charged impurity was being screened as predicted by electrostatic double layer theory. Fifthly, and in a related vein, the activity of both anions (Figure 20) and cations argues against electrostatic effects or specific ion adsorption to the bilayer surface (Rand, 1981). Finally, the lack of a change in the lamellar repeat distance of DGDG between water, 100 mM KCI, or 5 mM EDTA strongly suggests that aggregation was not due to a charged contaminant 82 (Table 6). If a charged contaminarit had been present in these bilayers, then screening of the charged groups by the addition of either K.C1 or EDTA would be expected to reduce the magnitude- of electrostatic repulsion and allow closer bilayer approach under the influence of attractive" Van der Waals forces. Such a change would have been observed as a decrease in the lamellar repeat distance in KC1 or EDTA (Loosely- Millman et al., 1982), but was not seen in this system (Table 6). Given the above considerations, the mechanism involved in salt-induced DGDG vesicle aggregation is not clear. The strong correlation between cation efficacy and effective ionic radius (Figure 19) suggests that a water interaction may be involved. The observed sequence: Cs* = Rb+ = N H / = K + >Na + >Li + correlates well with the ability of these cations to break the structure of water as indicated by infrared spectroscopy (Verrall, 1973) and arguments based on the extent of hydration of the ions (Franks, 1984). For the cations, the structure-breaking ions were were the most effective at promoting DGDG vesicle aggregation. On the other hand, while the efficacy of anions at promoting aggregation agrees well with the lyotropic series for the salting-out of proteins (Record et al., 1978): Q->N0 3 >Br =C104->SCN the orientation is opposite to the cations, with the structure-breaking anions (SCN", C104") being the least effective at causing bilayer aggregation (Figure 20). Similar sequences have been observed for other hydrophobic colloidal systems (Eagland, 1975) in which oxygen atoms are an important component Eagland (1975) has suggested that anions probably interact preferentially with the hydration of hydrophilic groups while the cations will interact more strongly with the hydration of hydrophobic methylene groups. As a result of these tendencies, 83 the effect of structure-breaking anions was expected to be the reverse of structure-breaking cations, as observed in our data. Furthermore, these interactions were expected to be additive (Eagland, 1975). Interactions between approaching membranes in aqueous solutions are dominated by attractive Van der Waals forces, repulsive electrostatic forces, and repulsive hydrostatic forces (Israelachvili, 1985). Given the lack of electrostatic repulsive forces in uncharged DGDG bilayers, the major repulsive force preventing DGDG aggregation is the hydration force. If DGDG were to show a specific interaction with structured water, and that interaction could be disrupted by dissolved ions, then it is possible that the steeply rising hydrostatic repulsive force observed between phospholipid bilayers (Rand, 1981) could be reduced between approaching DGDG bilayers. Supporting evidence for such a view has been obtained by Johnston et al. (1985) who observed glycolipid-ion interactions in cerebroside monolayers and concluded that ion-induced changes in water structure could explain observed effects of salts on gluco- and galacto-cerebroside monolayer expansion. Wieslander et al. (1978) have reported effects of CaCl2 and MgCl 2 on the degree of hydration of Achdeplasma laidlawii diglucosyldiacylglycerol as detected by 2 H-NMR. Tomoaia-Cotisel et al. (1983) have observed effects of salts on the monolayer properties of distearoyl derivatives of MGDG and DGDG. These authors interpreted their data as indicating that hydrated ions were penetrating into the region between adjacent headgroups and causing monolayer expansion. Since these investigators reported no difference between the efficacies of Na+ and Mg2* and observed that CI" was the least effective of the anions, it is likely that they were investigating a different phenomenon from that reported here. A direct analogy between these results and those presented here is, or course, difficult because of the differences in structures of the glycolipids. Clearly, however, glycolipid-ion interactions are extensive. 84 On the other hand, an effect of DGDG on attractive forces between bilayers is also possible. It is known that the magnitude of Van der Waals forces acting between bilayers, as reflected by the Hamaker coefficient, may vary as much as 10-fold depending on the lipid composition of the interacting bilayers (Rand, 1981). Direct measurements of the forces acting between approaching DGDG bilayers in water (Marra, 1985, 1986) has shown the Hamaker coefficient for DGDG to be six-fold higher than that obtained for dipalmitoylphosphatidylcholine. At present it is not feasible to explicitly describe a mechanism for salt-induced DGDG aggregation. The possibility exists that the DGDG head group has a weak electrostatic charge or dipole moment that is being screened by salt addition. However, the effect of glycerol (Figure 21) on aggregation and the correlation between ion efficacy in breaking water structure and in causing vesicle aggregation strongly implies head group-water interactions as underlying the aggregation process. Interacting particles may aggregate in either a primary energy minimum with a small interparticle separation, or in a secondary energy minimum existing at larger interparticle distances (Israelachvili, 1985). While most aggregation processes are considered to be due to interaction in the primary minimum, aggregation of phosphatidylserine vesicles in the secondary energy minimum, has been reported (Nir et al., 1981). Marra (1985) has identified an energy minimum between approaching DGDG bilayers in water at 1.3 nm separation; however this was not identified as a primary or secondary minimum nor was the effect of salts on the position and depth of the minimum examined. Although appressed DGDG vesicles appear to be in very close contact (Figure 16b and 16c), the limits of resolution of freeze-fracture electron niicroscopy do not allow the measurement of the intervesicle separation in these aggregates. These structures may represent loose aggregates of vesicles condensed in a secondary energy minimum. 85 Calculations have been performed that indicate that the lipid concentration of the chloroplast stroma is in the 2-4 mg-ml-1 range. These calculations were based on: Barber's (1980) estimate of 200 m1 of thylakoid surface area per m2 of leaf surface area; an area per DGDG molecule of 0.7 nm2; assuming approximately 50% of exposed thylakoid surface area is due to lipid; and average leaf thickness of about 1 mm; approximately 50% of total leaf volume being occupied by photosynthetically active cells, and; 10% of mesophyll cell volume is occupied by chloroplasts. While it is obvious that many of these values are both approximate and species-dependent, the calculations nonetheless indicate that the lipid concentrations used in this study are close to those observed in the chloroplast Similarily, the ion concentrations used here are in tine with the concentrations found in isolated chloroplasts (Gross and Hess, 1974; Nakatani et al., 1979; Demmig and Gimmler, 1983) and those used for reversible experimental unstacking and restacking of isolated thylakoids (Staehelin, 1986). It appears, then, that DGDG vesicles reversibly aggregate in lipid and salt concentrations relevant to those found in vivo. This type of aggregation has not been specifically reported in the plant lipid literature. This may, in part, be due to the disruptive effect of glycerol on aggregation (Figure 21). The routine addition of glycerol to vesicle suspensions as a cryoprotectant would prevent such structures from being visible by freeze-fracture electron microscopy. The effect of salts on the interactions of vesicles made from total chloroplast lipids has been examined by Gounaris et al. (1983c). These authors observed increased turbidity and vesicle size upon incubation of total lipid dispersions in salts. These authors interpreted their results as indicating the screening of the surface ionizable groups of PG and SQDG by cations and protons in agreement with classic electrostatic double layer theory. However, Gounaris et al. (1983c) obtained ion-specific differences in effectiveness of Ca2* and Mg 2 + similar to those reported here, but not expected from electrostatic double-layer theory. Since neither the biphasic nor pH effects seen by those 86 authors were observed in purified DGDG vesicles, it seems likely that the results of Gounaris et al. (1983c) were a consequence of using a mixture of thylakoid lipids showing different tendencies to aggregate at specific ion concentrations. This would include the pH-dependent but ionic species-independent aggregation due to screening of the anionic lipids PG and SQDG (Gounaris et al., 1983c), as well as the mostly pH-independent (Figure 22) but ionic-species dependent (Figure 18-20) aggregation reported here. In summary, evidence has been presented indicating that vesicles of the thylakoid galactolipid digalactosyldiacylglycerol aggregate strongly in the presence of physiologically relevant levels of aqueous salt solutions. Further, these data suggests that the mechanism by which vesicle aggregation occurs is probably related to the degree of hydration of the bilayer surface. 87 4 . EFFECTS OF LIPID ADDITIONS AND HEAD GROUP MODIFICATION ON VESICLE AGGREGATION 1. INTRODUCTION In the previous Chapter (and Webb et al., 1988) the reversible aggregation of vesicles composed of pure DGDG was demonstrated. In vivo, however, DGDG proportions range from only 20-30 mol% of the total thylakoid lipid (Murphy, 1986a,b). Therefore, the effect of the other thylakoid lipids MGDG, SQDG, and PG, on the ability of DGDG vesicles to aggregate was of considerable interest Inhibition of bilayer close approach by the addition of either of the anionic lipids PG or SQDG to DGDG would be predicted simply due to electrostatic repulsion. On the other hand, if DGDG aggregation is indeed due to galactose-water interactions, then the M G D G head group may also facilitate bilayer appression by a similar mechanism. The experiments described in this Chapter were designed to test the effect of physiologically relevant lipid and salt mixtures on galactolipid vesicle aggregation. This work is similar in approach to previous work by Gounaris et al. (1983c) who reported the results of interactions between polydisperse vesicles composed only of a total lipid mixture extracted from spinach chloroplasts. 88 2. MATERIALS AND METHODS 2.1.Materials All materials used were identical to those described in Chapter 3 with the following exceptions. Egg PG (sodium salt, derived from egg PC) was obtained from Sigma (P-5531) and used without further purification. Diglucosyldiacylglycerol (DglcDG) was the generous gift of Dr. R.N. McElhaney, Dept. of Biochemistry, University of Alberta, Edmonton, Alberta, Canada, and was also used without further purificatioa DGDG enriched in 18:2 (18:2-DGDG) was purchased from Serdary Research Laboratories (London, Ontario), and was purified as described for DGDG in Chapter 2 on both CM-cellulose and silica chromatography. Carboxylate beads were obtained from Polysriences (Warrington, PA), trypsin (from Bovine pancreas, EC from Sigma (T-1005, Type XI), and galactose oxidase (purified from Dactylium dendroides, EC also from Sigma (Type V). 2.2 Vesicle reconstitution Pure lipids, and lipid mixtures, were dried and dispersed into H 2 0, salt, or buffer solutions at 10 mg lipid-mf1 as described in Chapter 3. For most experiments vesicle dispersions were extruded by ten passes through two 0.1 Mm Nuclepore filters at 2000 kPa N 2 . Vesicles of DglcDG were passed through these filters three times at that pressure. To test the effect of vesicle size on aggregation, DGDG dispersions were passed through 0.4, 0.2, 0.1, 0.05, and 0.01 Mm filters as described in Chapter 3. 89 2.3. Measurement of vesicle sizes and AA50 values Vesicles dispersed as outlined above were diluted to 1 mg lipid-ml"1 and sizes estimated by QELS. AA50 values for MgCl2 were measured by turbidity as described in Chapter 3. Reversibility of aggregation was measured by dilution of the aggregated vesicles to 0.1 or 0.2 mg lipid-ml-1 and to a final MgCU concentration well below the AA50 value determined for that sample, then sizes measured by QELS. 2.4. NICOMP 270 standard curve Carboxylate beads showing a narrow size distribution were diluted to 0.05% solids in H20 and sizes measured by QELS. The reported size was that obtained if further dilution of the beads did not affect the mean particle size. Results reported include those values obtained from both the NICOMP (no assumptions) and Gaussian fits to the scattered light intensity. The NICOMP values reported are those obtained assuming the beads were vesicles, rather than solid particles. 2.5. Enzymatic digestion of DGDG vesicles DGDG vesicles were dispersed at 10 mg lipid-ml"1 as described above. For galactose oxidase digestion the concentrated enzyme was added to the vesicles to yield a final buffer containing 6.5 mM Tris-HCl (pH 6.8 or 8.0), 1 unit enzyme-ml1, 5.9 Mg protein-ml1, and 10 mg lipid-ml"1. The mixture was incubated at 30° C and aliquots removed at various times for 10-fold dilution and measurement of AA50 for MgCl2. For trypsin digestion, the enzyme (in 1 mM HC1) was added to DGDG vesicles to a final trypsin concentration of 20 ug-val'1. Vesicle incubation and AA50 measurements were done as above. Lipid extraction and TLC after the digestion experiments showed no evidence for lipid breakdown nor change in the migratory properties of DGDG. Trypsin activity was confirmed by digestion of the artificial 90 substrate benzoyl arginine p-nitroanaline to the colored product p-nitroanaline. 2.6. Lipid analysis In vesicles composed of mixtures of lipids the preparations were checked for complete incorporation of all of the added lipids by re-extracting the vesicles into C H C l } / C H 3 O H (1/1, v/v) at the end of the experiment Aliquots were spotted onto analytical silica gel plates (0.25 mm x 2.5 cm x 7.5 cm, Merck) and developed in acetone:benzene:water (91/30/8, v/v/v) (Pohl et al., 1970). Plates were sprayed with the phospholipid reagent of Allen and Good (1971) and chaired with a heat gun. After allowing the background blue coloi to fade the plates were photographed on Polaroid P/N 665 film and the negatives scanned by densitometry (Helena Quick Scan). The recovery of each lipid in the dispersion was estimated by weighing the areas under the lipid peaks. Fatty acid profiles were determined by G L C of fatty acid methyl esters as in Chapter 2. 91 3. RESULTS 3.1. Fatty acid and lipid compositions The fatty acid compositions of lipids used specifically in this Chapter are given in Table 7. Fatty acid profiles for the other lipids are described in Chapter 5. Egg PG showed high levels of 16:0 (44 mol%) and 18:1 (30 mol%) and an overall fatty acid profile very similar to that of egg PC (Foley et al., 1988). Diglucosyldiacyglycerol (DglcDG) was extracted from the unsaturated fatty acid-auxotroph Achdeplasma laidlami B grown in a medium supplemented with palmitic and oleic acids. This addition is seen in the 18:1-enriched fatty acid profile of DglcDG (Table 7) and was similar to those reported by McElhaney et al. (1973). The fatty acid compositon of DGDG obtained from Serdary Research Lab showed very high levels of 18:2 (73 mol%). By comparison, the DGDG purified from spinach leaves, that used routinely here, was highly enriched in 18:3 (85 mol%, Chapter 2). The fatty acid composition of Serdary DGDG was very similar to that reported for DGDG purified from wheat flour (Foley et al., 1988). All of these lipids were expected to adopt the liquid-crystalline phase when dispersed in excess water at room temperature (Chapter 1). The lipid composition of vesicles re-extracted into CHCl 3 /CH;OH (1/1) after aggregation experiments is given in Table 8. This data shows that the final lipid compositon of the vesicles was very similar to that of the dispersed lipid mixture. This result was obtained even when high levels (50 mol%) of MGDG were dispersed into the vesicles. Similar results were obtained in other mixtures (Chapters 5 and 6). This data is in contrast to that of Sprague and Staehelin (1984a,b) who reported that MGDG could not be quantitatively transferred into the aqueous phase by conventional hydration methods. The reason for this 92 TABLE 7 THE FATTY ACID COMPOSTION OF LIPIDS USED. Fatty acid methyl esters were prepared with BF 3 /methanol and analyzed by gas-liquid chromatography. Data presented are the means (± S.D.) from 4 determinations. Lipid mol% fattv acids Mil. 18:0 18:1 1L2 egg PG 43.6 (3.3) 11.7 (0.5) 30.2 (1.6) 14.4 (1.2) DglcDG 37.3 - - 5.9 56.5 (2.4) (1.0) 1.7 18:2-DGDG 15.4 - - 1.3 6.4 72.9 3.9 (1.3) (0.1) (0.1) (1.4) (0.1) 'trans-AS-hexadecanoic acid. 93 TABLE 8. RECOVERY OF LIPIDS FROM VESICLES. Vesicles were re-extracted into CHCl 3 /CH } OH (1/1, v/v) after aggregation experiments. Aliquots were separated on silica TLC in acetone:benzene:water (91/30/8, v/v/v) and charred with the phospholipid spray of Allen and Good (1971). Plates were photographed then negatives scanned on a Helena Quick Scan gel scanner and recoveries estimated by weighing the areas under the lipid peaks. Data are the means ( ± standard deviation) of at least three separate preparations. mixture dispersed % of recovered lipid DGDG MGDG SQDG ±G_ DGDG 100 DGDG/PC 3/1 69.5 (6.8) 30.4 (6.7) DGDG/PC 1/3 32.2 (6.4) 67.7 (6.4) DGDG/MGDG/PC 1/2/1 29.5 (1.4) 46.0 (6.2) 24.5 (4.8) DGDG/MGDG/SQDG 1/2/1 33.2 (6.5) 46.1 (4.4) 20.7 (2.0) DGDG/MGDG/PG 1/2/1 33.4 (3.3) 45.0 (0.1) 22.9 (1.3) DGDG/MGDG/SQDG/PG 30.9 (1.8) 44.0 (2.7) 13.0 (3.5) 12.0 (1.0) 9 4 1 i i 2 3 4 Addition, mole% Figure 23. Effect of the addition of 0-5 mol% egg PG (•). SQDG (•), or MGDG (A) to DGDG on the AAj0 values (mM MgCl5) required for vesicle aggregation. Data are the means of three determinations from a representative experiment Standard deviations are about 1 mM. 95 MgClj concentration, mM Figure 24. Plot of turbidity (% total A 6 0 0 increase) against added MgCl, concentration for the indicated lipid mixtures. Data are the means of three determinations from a representative experiment 96 TABLE 9. SIZE AND AGGREGATION CHARACTERISTICS OF LIPID MIXTURES. Lipid mixtures were dispersed in H 2 0 and extruded to make 100 nm unilamellar vesicles. Diameters were measured by QELS as described in Chapter 3 at 1 mg lipid-ml"1 in H 20. Vesicles were aggregated by the sequential addition of MgCl2 to determine the AA 5 0 value (mM MgCl2) and sizes measured again by QELS. Finally, vesicles were diluted to 0.1 or 0.2 mg lipid-ml"1 and to the final MgCl2 concentration (mM) shown and sizes measured again by QELS. Data represents mean ± standard deviation from representative experiments of 2-4 separate trials. lipid size (nm) aggregated vesicles diluted vesicles mixture in H iQ size (nm) AA. . size (nml fMpCUl DGDG 124 (46) £1000 5.4 143 (37) 1.2 DGDG/PC 121 (22) £1000 4.0 126 (29) 1.2 3/1 DGDG/PC 115 (30) 133 (46) £200 147 (60) 40 1/3 DGDG/MGDG/PC 121 (22) £1000 4.1 £1000 1.2 1/2/1 DGDG/MGDG/SQDG 137 (53) £1000 115 £1000 20 1/2/1 DGDG/MGDG/PG 134 (45) £1000 42 £1000 12 1/2/1 DGDG/MGDG/SQDG/PG 132 (45) £1000 65 £1000 28 1/2/0.5/0.5 97 difference is not clear, however, it is possible that the slow removal of bulk organic solvent from the lipid stock solutions may lead to MGDG phase separation prior to vesicle dispersion. Considering the magnitude of experimental error in these measurements, the differences beween vesicles before and after the experiments (Table 8) were small and were probably due to the preferential charring of unsaturated fatty acids by this spray reagent (Chapter 6). The extent of fatty acid unsaturation is different for each of these lipids (Table 7 and Chapters 2, 5, and 6). 3.2. Effects of other lipids on DGDG aggregation In order to assess the impact of other lipids on the extent of salt-induced DGDG vesicle aggregation, vesicles of specific lipid mixtures have been measured for their A A 5 0 values in response to MgCl2. The effect of the addition of 0.5-5.0 mol% of egg PG, SQDG, or MGDG to DGDG vesicles is shown in Figure 23. Addition of either anionic lipid significantly increased the AA 5 0 value required for aggregation. The presence of 1 mol% egg PG or SQDG raised the AA 5 0 from 5.5 mM for pure DGDG to 12 and 20 mM, respectively. Further lipid addition to 5 mol% increased the A A 5 0 value linearly for SQDG and in a saturation curve for PG (Figure 23). In contrast to the anionic lipids, the addition of MGDG in the 0.5-5.0 mol% range had no effect on DGDG aggregation (Figure 23). Further addition of MGDG up to 50 mol% also had no significant effect on the AA 5 0 values for the mixtures (data not shown). The possibility that MGDG is also active at promoting vesicle appression in salt was investigated. The aggregation of vesicles composed of different lipid mixtures was followed during the sequential addition of MgCl2 (Figure 24). The A A 5 0 values are summarized in Table 9. Pure DGDG vesicles showed a A A j 0 of 5.5 mM. Reduction of the DGDG content 98 to 75 mol% by PC addition (DGDG/PC, 3/1) had no significant effect on the AA 3 0 value (4.0 mM). However, further reduction of DGDG to 25 mol% (DGDG/PC, 1/3), a proportion similar to- that in thylakoids, yielded vesicles that did not aggregate at 5*200 mM MgCl2. Replacement of some of this PC with MGDG to give a DGDG/MGDG/PC (1/2/1) mixture resulted in vesicles with a A A 5 0 of 4.1 mM. This value was identical to that obtained from DGDG/PC (3/1) vesicles, both samples composed of 75 mol% galactolipid. This result strongly suggests that MGDG and DGDG are equally active at promoting vesicle aggregation in the presence of salts. If MGDG were non-participatory, or even inhibitory, in membrane close approach, then A A J 0 from the DGDG/MGDG/PC (1/2/1) mixtures would be expected to be higher than those obtained from the DGDG/PC (3/1) mixture. The above results suggest that the zwitterionic PC head group has no inhibitory effect on galactolipid vesicle aggregation. This is in contrast to the observation that the replacement of PC in the DGDG/MGDG/PC (1/2/1) mixtures with either spinach PG (DGDG/MGDG/PG, 1/2/1) or SQDG (DGDG/MGDG/SQDG, 1/2/1) increased the A A J 0 values for these mixtures to 42 and 115 mM, respectively. A mixture with equal proportions of PG and SQDG (DGDG/MGDG/SQDG/PG, 1/2/0.5/0.5) yielded an intermediate A A J 0 of 65 mM MgCl2. Therefore, both PG and SQDG actively inhibit vesicle close approach, probably by electrostatic repulsion. These data also shows different degrees of inhibition of aggregation for the two anionic lipids; the SQDG mixture required higher MgCl2 concentrations for complete charge screening than did the PG mixture. This is also shown by Figure 23 in which the 5 mol% PG mixture peaked at a AA J 0 value of 43 mM while the 5 mol% SQDG value was 51 mM and linearly increasing at this concentration. These results strongly suggest different orientations of the negatively charged groups of these two lipids, resulting in different interactions of these groups with ions in the aqueous phase. 99 TABLE 10. EFFECT OF KC1 ON VESICLE AGGREGATION IN MgCl2. Mixtures of DGDG/MGDG/SQDG/PG (1/2/0.5/0.5) were dispersed in H 2 0, 100, or 200 mM KC1 and extruded to make 100 nm unilamellar vesicles. Diameters were measured by QELS as described in Chapter 3 at 1 mg lipid-ml'1 in H 2 0. Vesicles were aggregated by the sequential addition of MgCl2 to determine the AA 5 0 value (mM MgCl2) and sizes measured again by QELS. Finally, vesicles were diluted to 0.1 or 0.2 mg lipid-ml-1 and to the final MgCl2 concentration (mM) shown and sizes measured again by QELS. Data represents mean ± standard deviation from representative experiments of 2-4 separate trials. fKCn (mM) size (nm) in KC1 aggregated vesicles diluted vesicles size (nm) AA™ (,mM) size (nm) fMgCM (mM) 0 132 (45) £1000 65 £1000 28 100 114 (27) £1000 57 £1000 20 200 113 (22) £1000 49 £1000 20 100 3.3. Effect of cation mixtures on A A 3 0 values An attempt was made to determine the effect of cation mixtures on galactolipid vesicle aggregation. Recently, Schroppel-Meier and Kaiser (1988) measured K* and MgJ* concentrations in spinach chloroplasts at 180 and 18 mM respectively. DGDG/MGDG/SQDG/PG (1/2/0.5/0.5) vesicles were dispersed in 0, 100, and 200 mM KC1 and the A A J 0 values for MgCl2 measured (Table 10). These data show an 8 mM decrease in MgCl2 required for aggregation with each 100 mM increase in KC1 concentration. This is strong evidence that electrostatic repulsion is the primary force preventing close approach in these vesicles. The 10:1 efficacy ratio for divalent: monovalent cations in screening surface charges (Barber, 1980), arising from the ion valency raised to exponential terms in the Boltzmann distribution, is similar to the 100:8 ratio reported here. Slight differences in these ratios probably originate in some degree of ion binding to the bilayer surface. 3.4. Effect of head group and fatty acid modification In view of the above effects, it was of interest to determine if specific aspects of galactolipid structure are required for vesicle aggregation. The turbidity of 18:2-enriched DGDG vesicles during the addition of MgCl2 is compared to that of 18:3-DGDG in Figure 25. A slightly increased AA J 0 of 8.6 mM was determined for the 18:2-DGDG vesicles. Given the extreme sensitivity of the A A J 0 to low level charged contaminants (Figure 23) it was not possible to attribute this difference specifically to the change of fatty acid unsaturation. Interpolation of Figure 23 indicates that charged contaminants in the 0.25-0.5 mol% range would be sufficient to raise the AA J 0 to 9 mM. Aggregation of DglcDG vesicles by the addition of MgCl2 is also shown in Figure 25. 101 MgClj concentration, mM Figure 25. Plot of turbidity (% total A«oo increase) against added MgCl2 concentrations for vesicles of DGDG (•). 18:2-DGDG (•), and DglcDG (A). 102 Under all circumstances, the AAj 0 values for DglcDG were far higher than for DGDG. In initial experiments DglcDG gave AA 5 ( ) of about 37 mM. After re-extraction of the lipid and re-dispersal for replicate experiments values of 60-80 mM were obtained. Reversibility of aggregation was confirmed by QELS as described for DGDG (data not shown). There is, however, some question of the purity of the DglcDG preparation since TLC and GLC analysis of this lipid after all of these experiments showed 3-4 major bands on TLC and changes in the fatty acid compositon of the lipid (data not shown). Attempts to modify DGDG structure by digestion with galactose oxidase and trypsin were not successful. Galactose oxidase digestion was intended to induce a net negative surface charge density by oxidation of the H 1 2 alcohol. Trypsin digestion was attempted because of the extensive use of this enzyme to investigate the role of proteins in mediating granal stacking and because of the known activity of trypsin against other substrates including fatty acids. Neither treatment changed the A A } 0 of DGDG vesicles even after extended digestion for 6 hours at 30° C. Failure to add negative charge density to DGDG vesicles by galactose oxidase digestion was probably due to the galactolipid being a poor substrate for the enzyme, as suggested by other workers (Sundby and Larrson, 1985). In contrast, trypsin shows some hydrolytic activity against non- protein substrates, including fatty acid esters (Hofstee, 1957). However, no effect on DGDG vesicles was noted, nor were lipid degradation products visible by TLC after the experiment (data not shown). 3.5. Effect of vesicle diameter on aggregation Israelachvili (1985) has shown for the aggregation of two identical vesicles of radius R that the adhesion energy at equilibrium is proportional to RJ and that the total adhesion force is equal to 2rrRcj0 (where <J0 is the adhesion energy per unit area of approaching 103 surfaces). This predicts, therefore, that larger vesicles will have both larger adhesion energies and greater force of adhesion. As a result, small DGDG vesicles would be expected to require higher MgCl2 concentrations to aggregate. It was of interest, therefore, to test if DGDG vesicles of various sizes would all aggregate in MgCl2. A standard curve for the NICOMP 270 particle sizer (Figure 26) indicated a good correlation between measured and actual particle sizes for monodisperse beads of less than 1 nm diameter. DGDG vesicles extruded to a variety of sizes between 58 and 360 nm diameter all aggregated in both 10 mM MgCl2 and 100 mM KCI at lipid concentrations of 1 mg-ml1. It was not possible to measure different efficacies of MgCl 2- induced aggregation for vesicles of different sizes due to the lower reliability of QELS for polydisperse preparations and due to the high turbidity of non-aggregated 360 nm vesicles. 3.6. Reversibility of vesicle aggregation The addition of small amounts (<5%) of the lipids PG, SQDG, and MGDG to DGDG vesicles (Figure 23) had no effect on the reversibility of vesicle aggregation (data not shown). In all cases, vesicles released from the aggregates during dilution retained their original diameters of about 120 nm. However, in vesicle systems containing 50 mol% MGDG, aggregation was not reversible despite the dilution of MgCl2 to levels well below the AA J 0 value (Tables 9 and 10). Irreversibility was observed even with the presence of 25 mol% anionic lipids PG and SQDG (Table 9). These data suggested that the presence of MGDG triggered vesicle fusion subsequent to MgCl2-induced close approach of bilayers. Larger vesicles produced by vesicle fusion were observed by QELS as being £1000 nm in diameter (Tables 9 and 10). These observations are consistent with the proposed role of inverted micellar 104 Figure 26. Standard curve for the NICOMP 270 particle sizer. Plot of measured sizes using NICOMP (•) or Gausssian (•) distributions against actual bead size. Data represent the mean ( ± S.D.) of three independent determinations. 105 structures, adopted by hexagonal-II phase lipids such as MGDG, as intermediates in bilayer fusion (Cullis and Hope, 1978, 1988). It is not known, however, how much H phase lipid is needed to trigger bilayer fusion. Experiments were performed in which binary mixtures of DGDG and MGDG were dispersed into H 20, aggregated by MgCl2 addition, then re-diluted with H 2 0 to examine the reversibility of aggregation (Figure 27). Addition of MGDG to DGDG up to 50 mol% had no effect on the size of extruded vesicles in H 20, nor on the rmnimum size of aggregated vesicles (Figure 27) as measured by QELS. Irreversible galactolipid vesicle aggregation was only observed in vesicles containing £ 25 mol% MGDG. A significant increase in the size of release vesicles occurred at 25-30 mol%, below this value the original 120 nm vesicles were re-released from the aggregates. Formation of large vesicles, £1000 nm in diameter, required 40 mol% MGDG in DGDG. 106 Figure 27. Effect of MGDG on reversibility of vesicle aggregation. Plot of size (measured by QELS) of vesicles dispersed in H,0 ( A ) , after aggregation in MgCl, (•). and of re-diluted vesicles (n). Representative data from three experiments. 107 4. DISCUSSION The experiments described in this Chapter represent an attempt to determine if aggregation observed for pure DGDG vesicles occurs in physiologically relevant lipid and salt mixtures. Data presented in Figures 23 and 24 and in Table 9 indicate that the tendency of galactolipid vesicles to aggregate in MgCl2 is strongly inhibited by the presence of the anionic lipids PG and SQDG. It is likely that a significant amount of added salt is required to suppress electrostatic repulsion in these mixtures. These results may explain the failure of previous workers to observe DGDG vesicle aggregation in salt solutions. The presence of only 1 mol% charged lipid contaminant would retain sufficient electrostatic potential to prevent vesicle close approach and aggregation in 10 mM MgCl2 (Gounaris et al., 1983c) or 10 mM CaCl2 (Sen et al., 1982b). The addition of either neutrally charged lipid MGDG or PC did not inhibit aggregation (Figures 23 and 24) as expected from DLVO theory. The different curves obtained for PG and SQDG (Figure 23) shows that these lipids interact to different extents with the Mg 2 + ion. A saturation curve was observed for PG. At proportions of 5 mol% a A A 5 0 value of 43 mM was obtained and was very similar to that obtained for the DGDG/MGDG/PG (1/2/1) mixture (42 mM) where PG was 25 mol% (Table 10). Since PG was more effectively screened by Mg 2 t than was SQDG at higher charged lipid proportions, and was saturable, the formation of Mg 2*-PG complexes in which a single Mg 2 + screened two PG molecules seemed likely. This interpretation is consistent with neutron diffraction results indicating that the negatively charged phosphorous oxygen of E. colt PG is readily accessible to ions in the aqueous phase (Mischel et al., 1987) and with the association of Ca 2 + with the phosphate group of PC (Altenbach and Seelig, 1984) and PS (Dluhy et al., 1983), possibly as a bidentate ligand. Association of Mg 2 + with PG may be expected based on the known effects of both Ca 2 + and Mg 2 + on the phase transition 108 temperature of DPPG (Borle and Seelig, 1985). In contrast, the SQDG molecule also undergoes electrostatic screening but requires higher Mg 2 + levels to achieve similar suppression of the electrostatic potential. For example, 25 mol% SQDG in DGDG/MGDG mixtures required 115 mM MgCl, for aggregation compared to 42 mM for 25 mol% PG in the same mixtures (Table 10). The linear nature of Figure 23 for SQDG argues that the negatively-charged sulphonate group of this lipid is unavailable for Mg 2*-SQDG complex formation. It would appear, then, that PG charges are suppressed at least in part by a charge neutralization mechanism involving ion adsorption. In contrast, SQDG charges are suppressed by a delocalized charge screening mechanism in which the counterion forms an "atmosphere" of opposite charge around the lipid head group. That the electrostatic repulsion is the primary force preventing close approach in these mixtures is shown by the near 10:1 efficacy ratio of divalent: monovalent cations (Barber, 1980). These results are consistent with theoretical and empirical data showing the decrease of adhesion energy between vesicles of DGDG with only 2-4% added DOPG (Evans and Needham, 1987, 1988). The similar efficacy of MGDG and DGDG in promoting vesicle aggregation in salt solutions (Figure 24, Table 9) suggests that the MGDG galactose shows the same interactions with water as proposed for DGDG (Chapter 3). Little is known about the specifics of galactolipid head group orientation except that the digalactosyl group of DGDG appears to lie parallel to the plane of the bilayer (McDaniel, 1988). Low hydration of the DGDG head group has been reported (Rand and Parsegian, 1988), and other workers have indicated that the hydration of MGDG is similar to that of DGDG based on X-ray diffraction results (Brentel et al., 1985). Poor hydration of the bilayer surface coupled with effective salt competition for hydrating water may lead to vesicle aggregation by extensive interlamellar H-bonding and/or salting-out No effect on aggregation after changing the DGDG fatty acid unsaturation by an average of two double bonds per molecule (Table 7 and Figure 25) and 109 the possibility that the diglucosyl head group is fax less capable of promoting vesicle aggregation are consistent with a specific role of galactolipids in bilayer close approach. It should be noted, however, that a direct comparison between DglcDG and DGDG is difficult because of the additional structural differences between these molecules: 1) DglcDG is enriched in 18:1 (Table 7) whereas DGDG is high in either 18:2 or 18:3 (Table 7 and Chapter 2) despite the apparent lack of influence of fatty acid unsaturation on aggregation; 2) both sugars are a-linked in DglcDG (McElhaney, 1984) rather than a- (outer galactose) and 0- linked (inner galactose) in DGDG; 3) the outer glucose of DglcDG is a-linked to the C-2 of the inner glucose (McElhaney, 1984) compared to the 0- linkage to C-6 in DGDG. Nevertheless, the DglcDG head group has been reported to be poorly hydrated, and that the degree of DglcDG hydration was altered by MgCl2 addition (Wieslander et al., 1978). Similarily, the DGDG head group is as poorly hydrated as PE (Shipley et al., 1973; Rand and Parsegian, 1988) with an estimated 14-15 H 2 0 bound per lipid (Brentel et al., 1985). In the presence of 200 mM KCI, vesicles composed of mixtures of lipids similar to those found in thylakoids aggregated around 45-50 mM MgCl2, a value far higher than the 18 mM stromal concentration recently reported for spinach chloroplasts (Schroppel-Meier and Kaiser, 1988). Several considerations, however, indicate that forces acting on these vesicles may be relevant to thylakoids in vivo. First, aggregation was tested here at 1 mg lipid-ml"1 while the concentration of lipid in the chloroplast may be significantly higher (Chapter 3). Secondly, the highly constrained nature. of granal stack ultrastructure (Staehelin, 1986) implies a locally high "membrane concentration". Third, due to the negative surface charge density of thylakoids and the exponential variation of valency in the Boltzmann distribution, divalent cations, and Mg 2 + in particular, are expected to be highly concentrated above bulk levels at the thylakoid surface (Barber, 1980). A specific concentration value for ions in the chloroplast has little obvious meaning since ions may be concentrated at membrane surfaces by double-layer 110 interactions and their activities in the highly viscous aqueous stromal phase are not known. Finally, the concentrations of PG and SQDG in these vesicles (12 mol% each) axe similar, but not identical, to those in spinach thylakoids. This mixture probably over-estimates the proportion of SQDG in thylakoids, which is usually 5-8 mol% (Murphy, 1986a,b; Gounaris et al., 1983e). It is concluded, therefore, that the forces promoting aggregation of physiologically relevant lipid mixtures in vesiculo may significantly contribute to granal stacking in vivo. Data presented here suggests a role of MGDG in promoting vesicle fusion. The presence of MGDG at 50 mol% in ternary and quaternary mixtures (Figure 24 and Table 9) as well as at £ 2 5 mol% in binary mixtures with DGDG (Figure 27) led to irreversible vesicle aggregation. While this is not an unambiguous assay for vesicle fusion, a similar method using the irreversibliity of turbidity changes in phosphatidylserine vesicles has been reported (Ohki, 1988). This method was reported to give results very similar to those obtained by a more rigorous assay using the fluorescent measurement of the mixing of vesicle aqueous contents. Since the fusion of two distinct bilayers requires the temporary disruption of normal bilayer structure, it is generally agreed that a non-bilayer intermediate must occur during fusion. Several possible intermediates have been proposed (see Wilschut and Hoekstra, 1984, 1986) including the inverted micelle (Cullis and Hope, 1978, 1988), point defects in bilayer structure formed by ion-induced lateral phase separation, defects due to osmotic pressure gradients (Lucy and Ahkong, 1986), and bilayer surface tension variations (Ohki, 1988). However, the inverted micelle remains the most attractive of the intermediates because it allows a structurally reasonable method for mediating non-leaky bilayer fusion. Other intermediates show no compelling way of disrupting bilayer structure without leakage of the vesicular contents. Ill While these data are consistent with the model of Cullis and Hope (1978, 1988), they are not inconsistent with other mechanisms of bilayer fusion. The detection of inverted micelles in these preparations by conventional freeze-fracture electron microscopy would not demonstrate a role of H phase lipids and lipidic particles in bilayer fusion. This is because inverted micelles observed in many fusing phospholipid systems may result from long-term (2 hour) storage of the vesicles with Ca J + and/or cryoprotectant glycerol (Bearer et al., 1982) and may not be due to fusion per se. Inverted micelles that are legitimate fusion intermediates probably possess lifetimes too short to be observed by freeze-fracture. Other observed inverted micelles are probably stable, long-lifetime forms, not acting as fusion intermediates, but could be formed after the fusion event It is possible, then, that other non-bilayer structures with short lifetimes are the real fusion intermediates and observed inverted micelles simply a consequence of membrane-membrane interaction (Papahadjopoulos et al., 1988; Wilschut and Hoekstra, 1984, 1986; Bearer et al., 1982). Therefore, MGDG may simply destabilize DGDG bilayers to the extent that fusion occurs as mediated by some other structural intermediate. The results presented here are consistent with those reported by Gounaris et al. (1983c) who found the increase of spinach total lipid vesicle size could be triggered by either salt addition or decreased pH. These results were interpreted in terms of classical DLVO double-layer theory in which surface charge screening by protons and/or salts promoted vesicle flocculation by Van der Waals attraction. The data in this Chapter suggest that: 1) both galactolipids DGDG and MGDG promote vesicle aggregation in MgCl2 to similar extents; 2) the galactose moeity may be specifically required for this interaction with hydrating water; no effect of limited changes in fatty acid unsaturation was found; 3) both PG and SQDG inhibit vesicle close approach by electrostatic repulsion, however, their different efficacies at inhibiting aggregation in salt solution implies very different orientations of the negatively charged group towards aqueous phase ions; 4) it is argued that these forces may contribute to close 112 approach observed in granal thylakoids but may represent a small contribution to the LHCII-mediated stacking; 5) vesicles with £ 2 5 mol% MGDG showed irreversible aggregation, implying an ability of MGDG in triggering bilayer fusion, a property which may be relevant to etioplast development 113 5. PERMEABILITY PROPERTIES OF LARGE UNILAMELLAR VESICLES OF THYLAKOID LIPIDS 1. INTRODUCTION A major function of the thylakoid is the development and maintenance of a transmembrane proton gradient of approximately three pH units. It is clear that, in order to avoid short-circuiting the electrical portion of the protonmotive gradient, the thylakoid must be an effective permeability barrier to ions. Analysis of the permeability properties of phospholipid vesicles has shown that the phospholipid bilayer is the primary barrier to transmembrane movement of charged solutes and is sufficiently impermeable to potassium and sodium to maintain concentration gradients for time periods up to days (Deamer and Bramhall, 1986). Typical permeability coefficients for these ions are in the 10"12 to 10"13 cm-sec"1 range (Deamer and Bramhall, 1986). Corresponding values have not been reported for the plant thylakoid lipids; although a recent report by Foley et al. (1988) suggested that the permeability of DGDG and DGDG/MGDG mixtures to KCI was similar to that of PC. The permeabilities of a uniform size population of large unilamellar vesicles of the galactolipid DGDG to "Rb\ 36C1", and 3H-glucose, and of mixtures of thylakoid lipids to , 6Rb*. have been assessed. Data are presented showing that the permeability characteristics of DGDG are significantly different from those of PC. Further, the effect of adding other thylakoid lipids on permeability of DGDG vesicles is shown and the importance of these results to the function of the thylakoid is discussed. 114 2. MATERIALS AND METHODS 2.1. Materials Materials used in this work were obtained as follows: 5-doxyl-stearate and 12-doxyl-stearate form Aldrich; KC1 from Amachem; , 6RbCl and Na"Cl from Amersham; . NaCl, sodium acetate and all solvents (reagent grade and re-distilled before use) from BDH; D-glucose from Fisher; mannitol from Mallinckrodt; Freon-11 from Matheson; RbCl from Merck; 3 H(2)-D-glucose and U C - cholesterol from New England Nuclear; Sephadex G-50 (coarse) from Pharmacia, and all others from Sigma. Soybean PC (Type HI—S, Sigma) was judged pure by analytical TLC and used without further purification. 2.2. Lipid purification and analysis The extraction, purification, and anlysis of lipids is described in Chapter 2. 2.3. Vesicle reconstitution Lipids in CHC1 3/CH 30H were mixed, dried rapidly under a stream of N 2 at 40° C then any residual solvent was removed under vacuum overnight If included, valinomycin (in CHC13), nystatin (CH3OH), or gramicidin D (CH3OH) were added at an ionophore/lipid ratio of 1/500 (mole/mole). Lipids were dispersed at 10 mg-ml-1 into 1 ml of salt solution or buffer (see below) by vortexing and sonication. These multilamellar dispersions were converted to large unilamellar vesicles of approximately 100 nm diameter by repeated extrusion through 0.1 Mm Nuclepore filters (The Extruder, Lipex Biomembranes, Vancouver, B.C., Canada) at 2000 kPa N 2 . Vesicles were periodically checked for size by light scattering (Nicomp 270 Submicron Particle Sizer) 115 and were always at 120 (±30) nm diameter. Increased dispersion turbidity associated with vesicle fusion or aggregation (Chapter 3) was not observed. 2.4. Rb+ flux Lipid or lipid mixtures, containing nystatin, were dispersed as described above into 1 ml H,0 containing 1-5 uCi "RbCl, usually at 110 M M RbCl. External , 6Rb* was removed by passing the vesicles over a 10 ml (wet bed) column of Sephadex G-50 (packed into a 10 ml disposable syringe) pre-equilibrated in equal concentration of NaCl and pre-spun at 100 g for 30 seconds. Vesicles were eluted from this column by spinning at 100 g for 30 s and the timing of the experiment was started at the end of this spin. At various times 100 nl aliquots of the vesicles were loaded onto a 1 ml Sephadex G-50 column, equilibrated and pre-spun as above, but packed into Bio-Rad Econo columns. External tracer was removed by another spin at 100 g for 30 seconds. Label remaining in the vesicles was assayed by LSC. All efflux experiments were performed at 30° C. Measurement of the vesicle recovery from the 1 ml columns was checked using DGDG vesicles labelled with 14C-cholesterol. Results indicated routine recovery of 90% of the vesicles with 5*1.5% carry-over of external label. In initial experiments it was ensured that "Rb* co-eluting with the vesicles was trapped within the vesicles, rather than surface bound, by the addition of valinomycin (1 M1 in CHC13) to 100-150 /xl of vesicles to a final valinomycin/lipid ratio of about 1/250. These vesicles were incubated for a further 15 min at 30°C then co-eluting "Rb* measured as above. This treatment invariably reduced the co-eluting label to background levels. An identical 1 M1 volume of CHC13 did not cause leakage of Rb* from these vesicles under the same conditions. This indicated that vesicle structure remained intact during valinomycin addition and, therefore, that co-eluting "Rb* was trapped within the vesicle rather than surface bound. 116 2.5. CI- flux Efflux of 36C1" was measured as described for "Rb* except that vesicles, containing gramicidin D in place of nystatin, were dispersed into 5 nd of Na 3 6Cl at 9.30 mM. The external solution was changed to 9.19 mM NaCl, 0.11 mM sodium acetate and sampled as described above. 2.6. Glucose flux The efflux of D-glucose was measured as above except that lipid, without added ionophore, was dispersed into 1 ml of 50 mM D-glucose, 5 mM Tricine-NaOH (pH 7.5) containing 5 ud 3 H - D - glucose. The external solution was changed to 50 mM mannitol, 5 mM Tricine-NaOH (pH 7.5), and sampled as described above. 2.7. Calculation of permeability coefficients Under conditions of the flux of an electroneutral solute, or the non-electrogenic flux of ionic solutes by movement of ion pairs, the permeability coefficient (P) may be simply calculated from: P=J/AAC where J is the flux in Mmoles-s"1, AC is the concentration gradient experienced by the solute of interest, and A is the total surface area of the vesicle preparation (Deamer and Bramhall, 1986). With the specific activity of the internal contents of the vesicles, the initial slope of the efflux curve (y moles remaining vs time) gives J. Lipid measurement by assay for galactose and/or phospholipid gives the area term by assuming 0.7 nmJ per lipid molecule (Bishop et al., 1980). 117 Alternatively, the permeability coefficient was calculated using: P=(kV/A)-(C/AC) where k is the first order rate constant of solute efflux in s"1, V is the trapped volume of the vesicles determined from the zero time sample, A is the total membrane surface area determined from the lipid assay, and C was the internal concentration of the solute of interest The rate constant, k, was determined from the slope of a least squares linear regression of ln(%0/%) vs time (seconds) as described previously (Brunner et al., 1980), where %0 and % represent the % solute remaining trapped within vesicles at time zero and time t respectively. For chloride efflux the initial external Cl~ concentration 0, therefore the rate constant, k, was determined by a modified ln(%0/%) expression (Brunner et al., 1980). For most of these regressions the r2 value was £0.97, indicating first order kinetics for the efflux. For both methods of calculation an area value equal to one-half the total area determined from the lipid assay has been used in order to account for the single monolayer of lipid "seen" by the diffusing ion. Statistical significance refers to a two-tailed "Student's" test at a a =0.05 significance level. 2.8. Electron Spin Resonance Vesicles were made from 2.7 mg of DGDG, combined with spin labels 5-doxyl-stearic acid (I[12,3]) or 12-doxyl-stearic acid (I[l,14]) at molar probe/lipid ratio of 1/200, by the reverse phase evaporation method (Sprague and Stahelin, 1984a,b) and dispersed into 0.2 mis of 150 mM KC1 and 0.5 mM Tris-HCl (pH 8.0). Spectra were recorded on a Varian Model E9 EPR spectrometer at 21° C with a microwave power of 10 mW and time constant of one second. For the slow anisotropic 118 motion of the I[12.3] probe the order parameter, S, was calculated from (Gaffney, 1976): S = { ( A | | - A _ L - C ) / ( A | | -2Aj_-2C)}-1.723 where: C=1.4G-0.053(A||-Aj_ ) and A || and Aj_ equal one-half of the separation, in gauss (G), of the outer and inner extrema, respectively, of the I[12,3] spectrum. For the relatively rapid anisotropic motion of the spin label I[l,14] the order parameter was not calculable, however, the rotational correlation time, T 0 , was calculated from: r„ = 6.510 1 0tj 0[( v/ho/h. l)-l] seconds where C J 0 is one-half the width of the midfield peak and h0 and h,i represent the heights of the midfield and highfield peaks, respectively, of the I[l,14] spectrum. 119 3. RESULTS 3.1. Lipid and fatty acid composition of vesicles As reported in Chapter 4, the final lipid compositions of dispersed vesicles was very similar to the composition of lipid mixtures (Table 11). 3.2. Efflux of solutes from DGDG and PC vesicles The efflux of , 6Rb* from large unilamellar vesicles of pure DGDG and pure soy PC is shown in Figure 28. Tracer flux occurred at a significantly greater rate from DGDG vesicles than from PC. In order to specifically measure Rb+ flux rates it was necessary to ensure that the counter-flow of Na* into the vesicles, and/or the co-transport of CT out of the vesicles occurred at rates greater than or equal to that of Rb\ The development of a Rb* electrical potential was avoided by the addition of the anion ionophore nystatin to these vesicles. Under conditions where Rb+ efflux is not matched by Na* influx, chloride will freely exchange through the nystatin complex to reach a new chemical potential, neutralizing a Rb+ electrical potential. However, the efflux of , 6 Rb + from DGDG vesicles was similar with or without (not shown) the inclusion of nystatin. This result argues that in DGDG vesicles without nystatin either the development of a Rb+ electrical potential is too small to influence further Rb* efflux and/or that DGDG permeability to CI" is equal to, or greater than, that of Rb\ The permeability of DGDG and soy PC to 56C1" is compared in Figure 29. Phosphatidylcholine was significandy more permeable to CI" than was DGDG. In these experiments the development of a CI" diffusion potential was short-circuited by the addition of gramidicin D. Both DGDG and PC vesicles were relatively permeable to glucose, leaking 120 TABLE 11. RECOVERY OF LIPIDS FROM VESICLES. Vesicles were re-extracted into CHClj/CH 3OH (1/1, v/v) after efflux experiments. Aliquots were separated on silica TLC in acetone:benzene:water (91/30/8, v/v/v) and charred with the phospholipid spray of Allen and Good (Allen and Good, 1971). Plates were photographed then negatives scanned on a Helena Quick Scan gel scanner and recoveries estimated by weighing the areas under the lipid peaks. Data are the means ( ± standard deviation) of at least three separate preparations. mixture dispersed DGDG DQDQ 100 % of recovered lipid MGDG SQDG EG DGDG/MGDG 3/1 73.7 (3.1) 26.3 (3.1) DGDG/MGDG 1/1 54.3 (3.9) 45.6 (3.9) MGDG/PC 1/1 44.9 (3.1) 55.1 (3.1) DGDG/PC 1/1 45.9 (3.8) 54.1 (3.8) DGDG/MGDG/PG 1/2/1 33.6 (4.2) 48.0 (3.5) 18.3 (4.9) DGDG/MGDG/SQDG 1/2/1 30.5 (2.9) 51.4 (4.9) 18.0 (3.8) DGDG/MGDG/SQDG/PG 29.8 (0.3) 49.1 (3.5) 8.9 (1.5) 12.2 (2.6) 1/2/0.5/0.5 DGDG/MGDG/PC 1/2/1 29.6 (2.5) 49.2 (7.0) 21.1 (9.4) soy PC 100 121 20-w-r— — — I ; — i 1 I i 1—' 0 10 20 30 40 50 60 Time, min Figure 28. Plot of % Rb* remaining in large unilamellar vesicles of DGDG (•) and soy PC (e) vs. time at 30°C. Vesicles were made with 0.110 mM RbCl inside and containing 1-5 dCi "Rb\ External solution was changed to an equal concentration of NaCl by gel filtration on Sephadex G-50. Aliquots of vesicles were removed at variuos times and external tracer removed by another passage over Sephadex G-50. Labelled "Rb4 co-eluting with the vesicles was assayed by LSC. 122 0 \ i i I I I I 1 0 10 20 30 40 50 60 Time, min Figure 29. Plot of % CI" remaining in large unilamellar vesicles of DGDG (•) and soy PC (•) vs. time at 30°C. Vesicles were made in 9.30 mM NaCl containing 5 uCi "CI". External solution was changed to 9.19 mM NaCl, 0.11 mM sodium acetate by gel filtration on Sephadex G-50. Aliquots of vesicles were removed at various times and external tracer removed by another passage over Sephadex G-50. Labelled "CI" co-eluting with the vesicles was assayed by LSC. 123 o f i i I I I I O 10 20 30 40 50 60 Time, min Figure 30. Plot of % glucose remaining in large unilamellar vesicles of DGDG (•) and soy PC (•) vs. time at 30° C. Vesicles were made in 50 mM D-glucose, 5 mM Tricine (pH 7.5) containing 5 vCi 3H(2)-D-glucose. External solution was changed to 50 mM mannitol, 5 mM Tricine (pH 7.5) by gel filtration on Sephadex G-50. Aliquots of vesicles were removed at various times and external tracer removed by another passage over Sephadex G-50. Labelled 3H-glucose co-eluung with the vesicles was assayed by LSC. TABLE 12. SUMMARY OF PERMEABILITY AND FLUIDITY CHARACTERISTICS OF DGDG AND SOYBEAN PC VESICLES. The double bond index (DBI) was calculated from the fatty acid composition of the lipids (Table 2 and not shown), assuming no positional specificity, and indicates the average number of double bonds per lipid molecule. Electron spin resonance (ESR) spectra yielded the rotational correlation times, r 0 , and order parameters. S. respectively of I[1.14] and I[12,3] spin probes. Permeability coefficients were calculated from P=J/AAC or using P=(kV/A)-(C/AC) as described in the Methods and represent the means (±S.D.) of £three independant preparations. The DGDG/PC row represents the ratio of data presented in the above two rows. Lipid J2BL ESR r. (ns) S P (ems') P=J/A-AC_ P=(kV/AHC/AQ_ -Rhl glucose Cl/Rh' -Rbl. glucose CI /Rb' DGDG 5.40 1.32 0.670 3.0-10"" 10-ltr" 3.2-Hr" 3.41 (0.410-") (8.610-,e) (0.2-10-") 3.4-10"" 9.310" 1.110" 2.7 soy PC 3.43 0.668* 5.0-10-" 7.8-10-' 3.410 1 0 1.510* (3.8-10-") (3.6-10-') (1.410") 2.6-10-" 4.2-10-' 2.410" 1.6-10* DGDG/PC 1.57 1.00 60.0 0.013 0.90' 130 0.022 0.46' 'not significantly different. 'calculated from Hubbell and McConnell (1969) for soy PC dispersions. 125 70-80% of trapped glucose within one hour (Figure 30). Permeability coefficients were calculated using two different flux equations as descibed in the Methods (Table 12). These data indicate that DGDG was 60-130 fold more permeable to Rb* than was PC (permeability coefficients for these lipids were 3-10"10 cm-s"1 for DGDG and about 4-10"12 cms"1 for PC). This relationship did not hold for CI" with DGDG showing 76-46 fold lower permeability to CI" than PC (M0"9 cm-s"1 and about 7.8-10"1 cm-s1). In contrast, both kinds of vesicles had the same permeability to D-glucose (3-10"10 cm-s1). For the flux of Rb* and glucose from both DGDG and PC vesicles the calculated permeability coefficients were very similar using either P=J/AAC or P=(kV/A)-(C/AC). Identical permeability coefficients were also obtained using either equation for Rb* flux from vesicles composed of mixtures of thylakoid lipids (see below). 3.3. Measurement of DGDG vesicle fluidity It has been reported (De Gier et al., 1968) that differences in acyl chain unsaturation, and presumably in bilayer fluidity, can influence passive membrane permeability to solutes. It was of interest, therefore, to determine if DGDG vesicles were more fluid than those of PC and if such fluidity properties could account for the differences in the permeability properties between DGDG and PC liposomes. The electron spin resonance spectrum (not shown) of DGDG vesicles labelled with the spin probe I[12,3], which samples the hydrocarbon region near the polar head group, yielded an order parameter, S, of 0.670 (Table 12). A very similar value of 0.668 has been calculated from spectra obtained from PC vesicles probed in this region of the hydrocarbon tails (Table 12). When the spin probe I[l,14] was' used the order parameter was not calculable as a result of the spectrum change to more rapid anisotropic motion of the doxyl group. Alternatively, the rotational correlation time, r 0 , the time required for the spin probe to rotate through an arc of one radian, has been 126 determined. Vesicles or DGDG yielded a T 0 of 1.32 ns while a similar value of 2.21 ns was calculated for the same probe in oriented multilayers of egg PC (Griffith and Jost, 1976). Calculation of the double bond index (DBI). the average number of acyl double bonds per lipid molecule, shows that DGDG was 1.5 fold higher in acyl unsaturation than soy PC (Table 12). These results clearly indicate that the fluidity of pure DGDG vesicles was similar to that of PC bilayers in both the upper and central areas of the hydrocarbon region. The higher double bond index (DBI, Table 12) for DGDG compared to PC may explain the slightly faster rotational mobility of the I[1.14] probe deep in the hydrocarbon region (1.32 ns vs. 2.21 ns) but is clearly not directly related to acyl chain order near the polar head groups. Similar r 0 values of 1.2 to 1.3 ns have been reported by Hiller and Raison (1980) for vesicles of total lipid extracts from barley and spinach chloroplasts, and 1.65 ns in vesicles of total lipid extracts of maize chloroplasts (Nolan and Bishop, 1978). These investigators also obtained order parameters close to those reported here. Hiller and Raison (1980) measured S values of 0.66 to 0.67 in phospholipid and galactolipid mixtures from barley chloroplasts with the spin probe 6-doxyl-stearic acid. Nolan and Bishop (1978) reported a 2Aj| value of 57.1 gauss for total chloroplast lipid liposomes measured with I[ 12,3] compared to the value of 53.5 gauss determined here for DGDG vesicles with the same probe. All these data suggest that differences in fluidity cannot account for the large differences in permeability characteristics between DGDG and soy PC vesicles. 3.4. Effects of other lipids on DGDG permeability Large unilamellar vesicles were made from mixtures of other purified thylakoid lipids with DGDG to assess their affect on vesicle permeability to Rb* (Table 13). Calculated permeability coefficients were very similar using either P=J/AAC or P=(kV/A)-(C/AC); values 127 TABLE 13. CALCULATED Rb+ PERMEABILITY COEFFICIENTS FOR LIPID MIXTURES. Permeability coefficients were calculated using P=J/A-AC or using P=(kV/A)-(C/AC) as described in the Methods. Values are the mean (± standard deviation) of at least three separate vesicle preparations. Lipid P = . T / A - A C P (cm-s1) P = ( k V / A H C / A C l DGDG/MGDG 3/1 3.0-10"10 ( ± 2.0-10-10) 3.2-10" DGDG/MGDG 1/1 3.4-10-10 ( ± 1.6-10"10) 2.6-10"10 MGDG/PC 1/1 2.0-10"11 (±1.0-10-") 1.6-10" DGDG/PC 1/1 2.8-10-" (±0.8-10- n ) 2.3-10-11 DGDG/MGDG/PC 1/2/1 4.4-10"11 ( ± 2.0-10-") 4.0-10"11 DGDG/MGDG/PG 1/2/1 2.2-10"10 ( ± 0.2-10-10) 2.9-10" DGDG/MGDG/SQDG 1/2/1 3.4-10-' (±2.0-10-') 3.2-10-' DGDG/MGDG/SQDG/PG 1/2/0.5/0.5 1.4-10-' ( ± 1.4-10"10) 2.5-10"' 128 discussed below are those obtained using P=J/A-AC. The addition of MGDG to DGDG at 25 mol% and 50 mol% neither increased nor decreased the permeability of vesicles to Rb\ the permeability coefficients remaining at about 3.2-10"10 cm-s"1 in spite of the fact that MGDG is a hexagonal phase-forming species. The equimolar addition of MGDG to PC caused a four-fold increase in Rb* permeability. Similarly, the addition of equimolar DGDG to PC resulted in a 5-6 fold increase in Rb* permeability above values for pure PC. These vesicles were an order of magnitude less permeable to Rb* than pure DGDG vesicles (Table 12). When PC was lowered to 25 mol% in the mixture DGDG/MGDG/PC (1/2/1), the permeability was even higher. However, these mixed lipid vesicles were an order of magnitude less permeable to Rb* than pure DGDG vesicles, i.e. they displayed behavior intermediate between pure DGDG and pure PC vesicles. Other ternary and quaternary mixtures of thylakoid lipids were also tested for permeability to Rb* (Table 13). The permeability of DGDG/MGDG/PG (1/2/1) to Rb* was similar to that of DGDG/MGDG (1/1), both near 2.5-10"10 cm-s1. The replacement of PG with another acidic lipid, SQDG, to give DGDG/MGDG/SQDG (1/2/1) caused a dramatic ten-fold increase in vesicle permeability up to the highest value measured here of 3.4-10"' cm-s"1. Replacement of half of this SQDG with PG to give DGDG/MGDG/SQDG/PG (1/2/0.5/0.5) gave a permeability coefficient midway between DGDG/MGDG/PG (1/2/1) and DGDG/MGDG/SQDG (1/2/1). This mixture is of particular interest because it approximates the lipid composition of the thylakoid membrane. This mixture was an order of magnitude more permeable to Rb* than pure DGDG, and three orders of magnitude more permeable than pure soy PC. All ternary and quaternary mixtures of thylakoid lipids were significantly more permeable to Rb* than DGDG/MGDG/PC (1/2/1). 129 4. DISCUSSION Model membrane studies, primarily using phospholipids derived from animal systems, have indicated that the lipid bilayer is the main barrier to transmembrane solute flux and is sufficiently impermeable to maintain observed solute concentration gradients (Deamer and Bramhall, 1986). This chapter reports results concerning the permeability of large unilamellar vesicles composed of lipids extracted from the thylakoids of spinach chloroplasts, to Rb\ G , and glucose. Large unilamellar vesicles of defined size and lipid composition, including vesicles with the same proportion of MGDG as in native thylakoid membranes (Table 12) have been produced. The high MGDG content is in contrast with the results of Sprague and Staehelin (1984a,b) as discussed in Chapter 4. These dispersions showed no sign of increased turbidity that would be indicative of vesicle aggregation (Chapter 3) or aggregation followed by fusion. It was important, therefore, to keep the salt concentration lower than that which causes vesicle aggregation (Chapter 3). It is concluded that vesicles used here were discrete, unfused vesicles of about 100 nm size. The likely presence of inverted micelles in the MGDG-containing preparations (Sprague and Stahelin, 1984a,b) did not affect the permeability of these vesicles to Rb*. This argues that the presence of MGDG at 50 mol% in the thylakoid membranes does not necessarily increase thylakoid membrane permeability to cations. The permeability properties of control soybean PC vesicles measured here were in good agreement with those reported by other workers. Reported literature values for PC are 10"12 to 10"15 cnvs"1 for Na+ and K + (Deamer and Bramhall, 1986) and 3.3-10"12 for Rb* (Mimms et al., 1981); and reported CI" values are several orders of magnitude higher than those of Na* (Deamer and Bramhall, 1986). Glucose permeability to 16:0/18:1-phosphatidylcholine vesicles of 100 nm diameter gave P equal to 1.5-10"10 (Wong, K., and Cullis, P.R., personal 130 communication). In addition, permeability coefficients of 0.3 and 1.1-10"10 cm-s1 have been reported for planar bilayers and small unilamellar vesicles, respectively, of egg PC at 25° C (Brunner et al., 1980). It is concluded, therefore, that the methods and calculations employed in this study were measuring permeability in a manner consistent with previous workers. The permeability of large unilamellar vesicles of DGDG to various ionic solutes was significantly different from soybean PC vesicles. DGDG liposomes were 60-130 fold more permeable to Rb* than PC vesicles, but 76-46 fold less permeable to CI". However, DGDG vesicles were statistically identical to PC vesicles in terms of glucose permeability. Interestingly, calculated permeability coefficients for DGDG to both charged and uncharged solutes were all very similar, in the 10"10 range (Table 12). These results are not in agreement with those reported recently by Foley et al. (1988) who reported that multilamellar dispersions of DGDG and binary mixtures of DGDG and MGDG form bilayers as impermeable to KC1 as PC. This discrepancy is probably due, in part, to their use of turbidity changes as a measure of osmotic swelling and, hence, permeability of vesicles (Bangham, 1968) in the presence of concentrations of cations that have been shown to cause reversible vesicle aggregation (Chapter 3). This aggregation was measured by turbidity changes upon salt addition in the 1-20 mM concentration range, and was not related to osmotic swelling (Chapter 3). The possibility that higher fluidity of DGDG vesicles compared to PC vesicles may explain the difference in permeability of these lipids was examined. It is clear from a comparison of r 0 and S values in Table 12 that the fluidity of DGDG vesicles was not significantly higher than that of soy PC vesicles. Van de Ven and Levine (1984) have reported similar motional properties of diphenylhexatriene in DGDG and soy PC bilayers. Related work with cholestane spin labels in DGDG bilayers led Koole et al. (1984) to the conclusion that the introduction of additional double bonds does not increase the rate of hydrocarbon chain motion in DGDG. Data summarized in Table 12 demonstrate that fluidity 131 differences in bilayers of DGDG cannot explain the differences observed between PC and DGDG in Rb* and CT permeability. Furthermore, even if higher fluidity were able to explain higher Rb* permeability, similarly increased permeation did not occur for CI" and glucose as would be expected for a simple fluidity-based explanation to permeation. The membrane potential profile model of Flewelling and Hubbell (1986) explains the large differences in cation and anion permeabilities observed in phospholipid bilayers as a result of oriented dipole moments on the lipid molecule. In particular the model minimizes the contributions of oriented water dipoles and the phosphate dipole and emphasizes a large contribution of ester dipoles to a resultant membrane electrical potential of several hundred millivolts with the membrane interior positive. Such a potential is expected to promote anion permeation and develop an energy barrier opposing cation flux. This would explain the several orders of magnitude higher permeability of anions compared to cations in phospholipids (Table 12, ClVRb*). The data reported here are not consistent with the model of Flewelling and Hubbell (1986). The DGDG molecule should contain both ester dipole moment vectors present on diacylphospholipids but have no phosphate dipole contribution. However, the galactose moeities of DGDG may have the same orientation as the galactose on galactocerebrosides (McDaniel, 1988), generating a small dipole potential, positive outside (Maggio et al., 1978), similar in direction to that expected for the phosphate dipole (Flewelling and Hubbel, 1986). If ester dipoles are the major contributors to permeability discrimination against cations, then DGDG vesicles should show the same difference in cation and anion permeation as observed in PC vesicles. It was observed, however, that DGDG vesicles show a small, statistically insignificant, higher permeability to CT than Rb* (Table 12). It seems likely, therefore, that the contribution of head group and water dipoles to ion permeation through phospholipid bilayers 132 has been underestimated and/or the contribution due to ester dipoles overestimated (Flewelling and Hubbel, 1986). Consistent with this interpretation is the observation of large dipole potentials • in monolayers of ceramide (Maggio et al., 1978) with unesterified fatty acids. For DGDG vesicles the similar permeability coefficients for CI" and Rb+ (Table 12), ions of very similar radii (Jones, 1975) and identical charge quantity, suggests ion permeation c)ccurring primarily under the influence of Born energy considerations with smaller contributions from charged groups and dipole potentials. The addition of other lipids to DGDG vesicles had very different effects on vesicle permeability to Rb+. The equimolar addition of MGDG (Table 13) did not significantly increase the permeability. However, the same addition of PC to DGDG lowered the permeability by ten-fold. This result suggests that MGDG has the same, or similar, permeability properties discussed above with reference to DGDG. The further addition of PG to give DGDG/MGDG/PG (1/2/1) caused a small, but not statistically significant, decrease in Rb* leakage. The addition of the other anionic lipid, SQDG, has the opposite effect of causing an order of magnitude increase in Rb* permeability. The mixture resembling the in vivo mixtures of thylakoid lipids, DGDG/MGDG/SQDG/PG (1/2/0.5/0.5) had a Rb+ permeability coefficient of 2.0-10"' cm-s-1. These results are consistent with the well-known high permeability of thylakoids to cations and anions. Rottenberg et al. (1972) have reported rapid, and similar, rates of passive efflux for "Rb\ 36C1", and 14C-sorbitol from lettuce chloroplasts and report a low activation energy for these fluxes. Barber (1972) has calculated a permeability coefficient for CI" in spinach thylakoids of 2.1-10"10 cm-s*1. This value is close to that observed here for DGDG vesicles of 9-10"10 (Table 12). Furthermore, a similar permeability coefficient of 5.1-10"10 cm-s"1 can be calculated from reported (Ball et al., 1985) NaCl and KC1 fluxes of 2.3-10"13 mol-cm"J-s"1 into Beta vulgaris thylakoids exposed to 450 mM solute concentration gradients. 133 These results argue that the apparent leakiness of thylakoids is an intrinsic property of the thylakoids, in particular of the thylakoid lipids, and not a result of the high surface/volume ratio of thylakoids (Barber, 1972; Ball et al., 1985; Ort, 1986). A number of workers (Hind et al., 1974; Deamer and Packer, 1969; Chow et al., 1976) have observed the rapid light-induced, redistribution of Mg + J, K \ and CI' in chloroplasts after an initial membrane potential of £100 mV is established (Avron, 1977). Under the influence of these ion fluxes, the membrane potential rapidly decays to steady-state levels of about 10 mV. It is proposed that the high permeability of thylakoid lipids to many solutes is the main cause of high thylakoid membrane permeability and of low observed thylakoid membrane electrical potentials. Upon light-induced increase in the protonmotive force, rapid redistribution of charged ions, particularily Mg*J and CI" occurs as energetically allowed by high lipid permeability and as driven by the electrical portion of the protonmotive gradient This flux dissipates the electrical potential of the thylakoid. This may also explain the large trans-thylakoid proton gradients of 3-4 pH units under steady state illumination. In mitochondria the low phospholipid bilayer permeability to cations leads to the development of a significant membrane potential of £100 mV with an accompanying pH gradient of 1 to 1.5 pH units (Ferguson and Sorgato, 1982). In chloroplasts the lack of a contributory membrane potential necessitates a larger proton concentration gradient of 3-4 pH units. The 1 to 1.5 ApH units of mitochondria is insufficient to support ATP synthesis in chloroplasts (Avron, 1977) . In summary it has been demonstrated that large unilamellar vesicles of the thylakoid lipid digalactosyldiacylglycerol, as well as binary, ternary, and quaternary mixture of thylakoid lipids show permeability properties significantly different from those of phosphatidylcholine. Furthermore, it is argued that these vesicles serve as good models for studying the permeability of thylakoids to ionic and non-ionic solutes. It seems reasonable to propose that 134 these permeability properties, primarily those of the galactolipids, may have a significant effect on function of the intact thylakoid and account for the low membrane potentials and high pH gradients observed in higher plant chloroplasts. 135 6. THE EFFECTS OF TEMPERATURE ON PHASE BEHAVIOR AND PERMEABILrTY OF THYLAKOID LIPID VESICLES 1. INTRODUCTION Many tropical and subtropical species of higher plants show a wide variety of stress symptoms, and frequently plant death, when exposed to low, non-freezing, temperatures in the 10° to 12° C range. Physiological changes associated with chilling injury (reviewed by Lyons, 1973 and Graham and Patterson, 1982) include immediate cessation of protoplasmic streaming, changes in respiratory rates, and extensive changes in general metabolite levels. Early attempts to link this range of physiological responses with a single "trigger" or site of damage were those of Lyons (1973) and Raison (1973). Both workers suggested that the primary plant response to cold was a lipid phase transition of cellular membranes from the liquid-crystalline phase to the gel -phase. This theory (particularily that of Lyons, 1973) predicted that significantly increased cellular permeability to ions and solutes would result from the formation of "cracks and channels" in the solid state lipid. A recent version of Lyon's original hypothesis specific to chloroplast membranes has been proposed by Murata and co-workers (Murata et al., 1982; Murata, 1983; Murata and Hoshi, 1984; Murata and Yamaya, 1984). The hypothesis suggests that elevated proportions of high melting-point molecular species of phosphatidylglycerol (PG) in thylakoids promotes gel phase formation at chilling temperatures in sensitive plants. The hypothesis has been supported by the detection of phase transitions in PG purified from chilling-sensitive plants, but not in PG from chilling-resistant plants, nor in other thylakoid lipids (Murata and Yamaya, 1984). In addition, phase transitions have been detected by differential scanning calorimetry (DSC) in total polar lipids from chilling-sensitive plants (Raison and Wright, 1983; Raison and Orr, 136 1986a) as well as by fluorescence methods in thylakoid, mitochondrial, and plasma membrane preparations and their lipid extracts (Bishop and Kenrick, 1987; Raison and Orr, 1986b; Pike, 1982; Fork et al., 1981). Other workers have found positive correlations between chilling sensitivity and proportions of the high melting-point fatty acids (16:0, 16:1, and 18:0) and/or with the high melting-point molecular species of PG (Murata et al., 1982; Murata, 1983; Murata and Hoshi, 1984; Li et al., 1987). Other workers have found no correlation between the content of high melting-point fatty acids and chilling sensitivity (Roughan, 1985; Kenrick and Bishop, 1986) and have challenged the generality of the hypothesis. It has apparently escaped attention that there is a total lack of data concerning the variation of permeability with temperature in thylakoid membranes, or their lipid vesicles, from chilling-sensitive and chilling-resistant species. In the previous chapter the permeability properties of vesicles made from in vivo mixtures of chloroplast lipids at 30° C were described. It was of interest, therefore, to test the effect of lowered temperature on Rb* permeability in vesicles mimicking those of thylakoids from chilling-sensitive and chilling-resistant plants. These results were related to differential scanning calorimetry (DSC) detection of phase behavior of the lipids, done in collaboration with Dr. Dan Lynch, Dept. of Agronomy, Cornell University. 137 2. MATERIALS AND METHODS 2.1. Materials All materials used were identical to those described in Chapter 5 except that dipalmitoylphosphatidylglycerol (DPPG), sodium salt, was obtained from Sigma. 2.2. Lipid Purification Isolation, purification, and quantification of thylakoid lipids was performed as described in Chapter 3. 2.3. Vesicle reconstitution and efflux measurement The reconstitution of lipid vesicles was performed as in Chapter 5 with several modifications. Ten mg of total lipid (containing nystatin at an ionophore/lipid molar ratio of 1/500) were dispersed into 1.0 ml of 5 mM MgCl2, 0.2 mM RbCl, and 3-5 uCi "RbCl. Lipid mixtures were of the general form DGDG/MGDG/SQDG/total PG (1/2/0.5/0.5, mol/mol/mol/mol) and varied from containing no DPPG (0% DPPG sample) to 12 mol% DPPG. In all cases the total PG content, DPPG plus spinach PG, was maintained at 12 mol%. Vesicles were extruded at 21° C by three passes through a single 0.1 um Nuclepore filter at 2000 kPa for 0% DPPG, 3500 kPa for 4% DPPG, and 6500 kPa for 8% and 12% DPPG samples. Vesicles were pre-equilibrated for £ 3 0 minutes at the temperature of the efflux experiment Leakage was measured as described in Chapter 5. Care was taken to pre-equilibrate all equipment and Sephadex at the measurement temperature. 138 2.4. Flux calculations Calculation of the permeability coefficient has been done using: P=k(V/A) where the symbols have the meanings defined in Chapter 5. Lipid was measured by the method of Roughan and Batt (1968). Vesicles sizes were estimated by quasi-elastic light scattering (QELS) as outlined in Chapter 3, but with vesicles at 1 mg-ml-1 in 5 mM MgCl2, 0.2 mM NaCl. Thin layer chromatography and measurement of vesicle lipid composition was accomplished as in Chapter 5. As before, statistical tests refer to < the Student's test at a significance level a =0.05. 2.5. Differential Scanning Calorimetry (DSC) Vesicle samples were prepared as above in 5 mM MgCl2 and 0.2 mM NaCl then were sealed under N 2 and maintained for about 24h at 4 ° C during shipment to Cornell for analysis by DSC. Vesicles were concentrated by centrifugation then the samples were assayed on the Perkin-Elmer Series 7 Thermal Analysis System between - 5 ° and 60° to 70° C at 10 K-min-1. All samples were taken through an initial heating run and subsequent cooling and re-heating runs if more than one endotherm was observed. 139 3. RESULTS 3.1. Characteristics of dispersed vesicles The lipid composition of the dispersed vesicles is given in Table 14. Since there was no effect of the temperature of the efflux experiment on the lipid composition of the vesicles (not shown); results from all temperatures have been combined. As reported in Chapter 5, it has been possible to obtain vesicles containing 50% MGDG and proportions of the other lipids similar to that in the dispersed mixture. Apparently lowered levels of PG in those preparations containing DPPG (Table 14) probably result from the inability of the spray reagent to detect this disaturated lipid. The presence of DPPG in these preparations was confirmed by positive reaction of this lipid with the molybdenum reagent A comparison of vesicle size and volume characteristics is presented in Table 15. Quasi-elastic light scattering indicated that 0% DPPG vesicles were approximately 180 nm in diameter, with a relatively narrow size distribution. All preparations containing DPPG showed larger average sizes of about 500 nm and a very heterogenous size distribution. While these values are probably not very accurate, the larger size of the DPPG-containing vesicles was confirmed by calculation of the trapped volume (Table 15). The 0% DPPG sample showed a trapped voume of 3.4 ul-umoV1 compared to 15.7 ulumol1 for all DPPG-containing samples. The trapped volume of 3.4 ulumoY1 corresponds well to that expected for vesicles of about 150 nm diameter (Mimms et al., 1981) assuming that: vesicles are spherical and unilamellar; an area/lipid molecule of 0.7 nm2, and; a bilayer thickness of 4 nm. Similarly, trapped volumes of 15.7 /iHimol"1 correspond to unilamellar vesicles of 450-500 nm diameter. This value was very similar to that indicated by QELS (Table 15). 140 TABLE 14. RECOVERY OF LIPIDS IN VESICLES. Vesicles were re-extracted into CHC1 3/CH 30H (1/1, v/v) after efflux experiments. Aliquots were separated on silica analytical TLC plates in acetone:benzene:water (91/30/8, v/v/v) and charred with the phospholipid spraty reagent of Allen and Good (1971). Plates were photographed then negatives scanned on a Helena Quick Scap gel scanner and recoveries estimated by weighing the areas under the lipid peaks. Data are means ( ± standard deviation) of 10 separate preparations. % of recovered lipid % DPPG DQDQ MGDG SQDG PG 24.8 52.8 12.2 10.1 (±3.6) (±2.8) (±1.4) (±2.5) 26.7 52.2 12.2 8.8 (±2.6) (±7.8) (±2.7) (±3.7) 29.4 49.6 14.5 6.4 (±4.4) (±6.3) (±3.0) (±3.1) 12 29.7 55.8 13.9 (±4.2) (±5.9) (±2.4) 141 TABLE 15. SIZE AND VOLUME CHARACTERISTICS OF VESICLES. Vesicles were measured for size by quasi-elastic light scattering in vesicles diluted to about 1 mg-ml1 in 5 mM MgCl2, 0.2 mM NaCl. Trapped volume and V/A ratios were calculated as described in the methods. Data are the means ( ± standard deviation) of 10 separate preparations. diameter trapped volume V/A % DPPG (nm) ; cul-umol lipid1) fcnvlfr*) 0 169 (±45) 3.40 (±0.43) 1.58 (±0.27) 4 475 (±164) 15.9 (±5.90) 7.57 (±2.80) 8 536 (±150) 15.7 9±5.0) 7.03 (±1.78) 12 518 (±161) 15.7 (±5.4) 7.45 (±2.51) 142 This result strongly suggests that the physical properties of these bilayers were markedly altered by the presence of DPPG, even at levels as low as 4 mol%. The exact nature of this change in unknown, but these results suggest that the deformability, or elasticity, of DPPG-containing bilayers is significantly lower than those without DPPG. These results show that vesicles used here were probably unilamellar in nature and, therefore, that the V/A ratio may be used directly in calculation of permeability coefficients without modification of the area value (A) for the presence of several bilayers. It is likely that the higher values of V / A obtained for all DPPG samples (Table 15) resulted from the larger size of the DPPG-containing vesicles. 3.2. Phase behavior of thylakoid lipid vesicles The phase behavior of thylakoid lipid vesicles was examined by DSC. For all vesicles stored at 4 ° C before DSC analysis, no endo- or exo-therms were observable between - 5 ° and 70° C (Figures 31 to 34). All samples were subjected to both heating and cooling runs to look for thermal history-dependant phase behavior. However, no evidence for phase changes were observed in any samples. An interesting complication was observed in initial experiments in which the vesicles were stored on dry ice before DSC. Analysis of these vesicles (not shown) revealed that for the 0% DPPG, a lipid mixture very similar to that of chilling-resistant species, no endotherm was observed in the - 5 ° to 70° C range. The scans obtained for frozen 4%, 8%, and 12% DPPG samples showed small, broad, endotherms occurring between 14° and 56° C. The enthalpy and peak temperatures of these endotherms did not correlate well with the DPPG content of the vesicles. The appearance of frozen vesicles was markedly different from the 4 ° C stored vesicles. Frozen vesicles were very turbid, giving the appearance of aggregated, and 143 Figure 31. Differential scanning calorimetry scans of 0% DPPG vesicles in 5 mM MgCl ;, 0.2 mM NaCl. Vesicles representing about ten mg of lipid were scanned between - 5 ° and 70° C at 10 K-min1. Solid line represents the heating scan, the dashed line the cooling scan. 144 10.0 20.0 30.0 40.0 50.0 60.0 Tamparature ( * O Figure 32. Differential scanning calorimetry scans of 4% DPPG vesicles in 5 mM MgCl2 and 0.2 mM NaCl. Vesicles representing about ten mg of lipid were scanned between 0 and 70° C at 10 K-mhr1. The solid line represents the heating scan, the cooling scan (not shown) also showed no detectable heat flow. 145 o o a 4.0-r 3.5 3.0 as ao-i r l . 0 20.0 30.0 40.0 sao eao Tomperaturo (* D Figure 33. Differential scanning calorimetry heating scans of 8% DPPG vesicles in 5 mM MgCl2, 0.2 mM NaCl. Vesicles representing about ten mg of lipid were scanned between 0C and 70° C at 10 K-mhr1. The solid line represents the heating scan, the dashed line the cooling scan. 146 Figure 34. Differential scanning calorimetry scans of 12% DPPG vesicles in 5 mM MgCl2, q0.2 mM NaCl. Vesicles representing about ten mg of lipid were scanned between - 5 ° and 60° C at 10 K-min 1. The solid line represents the heating scan, the dashed line the cooling scan. 147 possibly fused, vesicles. It is likely that changes in interbilayer distances, and vesicle sizes during freezing lead to lipid demixing and phase behavior not indicative of the vesicles with which the permeability experiments were performed. These data show that heat flow indicative of phase separation did not occur in any vesicles containing DPPG, even with as much as 12 mol% added disaturated PG. The extremely broad endotherms observed for frozen 4%, 8%, and 12% DPPG samples probably indicates that DPPG gel phases resulted from long term (24h) storage of the vesicles on dry ice, and/or from vesicle fusion and subsequent lipid demixing. 3.3. Temperature dependence of vesicle permeability Permeability coefficients calculated for the efflux of Rb+ from mixed lipid vesicles between 7° and 30° C are plotted in Figure 35. No significant differences in Rb+ permeability were found between any of the vesicle systems in the range from 7° to 16° C. This zone brackets that at which chilling- sensitive plants are most susceptible to low temperature damage. The 4% and 8% DPPG samples showed 2-fold higher permeabilities at 23° and 30° C than shown by the 0% and 12% samples. These differences were not statistically significant, however, and probably not indicative of systematic differences between these vesicle systems. These results are not consistent with the higher permeabilities expected from DPPG-containing vesicles at lower temperatures, particularily those between 10° and 15° C. For reactions obeying Arrhenius' theory the slope of a In k vs. 1/T plot is equal to -Ea/R where Ea is the reaction activation energy and R is the gas constant. Least squares regressions of In k vs. 1/T for all vesicles between 7° and 30° C (Figure 36) yielded correlation coefficients (rJ) of 0.90, 0.94, and 0.91 for 0%, 4%, and 8% DPPG samples respectively. Correlation coefficients of these magnitudes strongly indicate the absence of 148 x "I 15 20 Temperature, (C) Figure 35. Plot of permeability coefficients for Rb+ efflux from vesicles vs. temperature for vesicles containing various proportions of DPPG. Vesicles were dispersed into 5 mM MgCl2, 0.2 mM RbCl containing 3-5 nCi of "RbCl, then exchanged into 5 mM MgCl2, 0.2 mM NaCl. Data represents the means (±S.D.) of 2-4 independant determinations for 0% (•), 4% (•), 8% (O), and 12% (•), DPPG-containing vesicles. Figure 36. Arrhenius plot of In k vs. 1/T for the efflux of Rb+ from vesicles containing various proportions of DPPG. Lines represent linear regressions through the points indicated. Symbols and replication as indicated in Figure 35. 150 discontinuities, or break-points, in this plot A lower correlation (r2=0.39) for the 12% DPPG sample resulted from an outlying point at 353-10"5 K (10° C). Removal of this point from the regression gave an r2 of 0.95. This procedure may be justified when it is considered that the higher rate constant generating this In k value does not indicate higher permeability of the 12% DPPG sample at 10° C (Figure 35). This happened because an unusually low V/A value for 12% DPPG at 10° C resulted in a calculated permeability coefficient identical to the other lipid mixtures at 10° C (Figure 35). Calculated activation energies for Rb+ efflux from vesicles using the slope of In k vs. 1/T were 4.73, 11.3, 11.6, and 7.35 kcal-mol-1 for the 0%, 4%, 8%, and 12% DPPG samples respectively. The Ea for the 12% DPPG sample increases to a more consistent value of 11.4 kcal-mol"1 if the 10° C sample is removed from the slope calculation. Similar Ea values were obtained by the regression of the In P vs. 1/T plot (not shown). Values of 15-17 kcal-mol"1 have been reported for K* permeation of a variety of phospholipid bilayers and 5-8 kcal-mol1 for CI" permeability through PC and phosphatidic acid vesicles (Papahadjopoulos and Watkins, 1965). 151 4. DISCUSSION It is a widely held view that there is a direct effect of low temperature on the function and stability of plant membranes (Graham and Patterson, 1982). Low temperature is seen as acting primarily on the activation energies of membrane-bound enzymes and on the permeability properties of cellular membranes. Both effects could be manifested through changes in membrane lipid fluidity and/or partial phase separation of the membrane lipid (Lyons, 1973). Much support for a role of disaturated PG in chilling sensitivity has been obtained by correlations between the content of high melting-point molecular species of PG (or of high melting-point fatty acids) and the degree of chilling sensitivity. If it is assumed that the total PG content of thylakoids is approximately 10 mol% (Nishihara et al., 1980) in most species, then total levels of disaturated PG in chilling sensitive plants may be calculated to be in the ranges: 0.8-5.6 mol% (Murata et al., 1982); 2.6-6.5 mol% (Murata, 1983); 0.6-4.4 mol% (Kenrick and Bishop, 1986); and fal mol% (Roughan, 1985) in some species. Comparable values for chilling-resistant plants are: 0-0.7 mol% (Murata et al., 1982); 0.3-1.9 mol% (Murata, 1983) and; 0-4.2 mol % (Kenrick and Bishop, 1986). This indicates that the levels of disaturated PG in chilling-sensitive and chilling-resistant plants overlap strongly. These considerations have led several investigators (Kenrick and Bishop, 1986; Roughan, 1985) to conclude that levels of disaturated PG more strongly correlate with the genetic origin of the plant rather than with its degree of chilling sensitivity. Thermotropic phase behavior of various lipid mixtures (Figures 31-34) mimicking those of chilling-sensitive and chilling-resistant plants shows that phase separation could not be obtained with as much as 12 mol% added DPPG. This result is in disagreement with those of Raison and Wright (1983) who triggered the onset of phase separation at 10° C by the 152 addition of 1 mol% DPPG to wheat thylakoid (chilling-resistant) polar lipids. Unlike these workers (Raison and Wright, 1983) it has not been possible here to show a strong relationship between amount of added DPPG and the onset of phase separation. Both 4% and 8% DPPG samples showed very broad, non-descript, endotherms extending over several tens of degrees. Only in the 12% DPPG vesicles subjected to freezing was a distinct separation, centered at 24° C, observable. These results with freeze-thaw vesicles, however, were not reproducible as other dry ice-frozen vesicles failed to show transitions in the -5 to 70° C range (not shown). These results are in agreement with those of Low et al. (1984) who found no correlation between the degree of chilling-sensitivity of COa-fixation and the presence of DSC-detected thylakoid membrane endotherms. It should be added that Low et al. (1984) did observe endotherms in both the thylakoids, and their polar lipids, from the chilling-resistant spinach. A similar endotherm was not detected in this study (Figure 31). This difference may illustrate the importance of consistent dispersion methods and the purity of polar lipid extracts used for calorimetric studies. Epand (1985) has reported significant changes to the lamellar/hexagonal-II phase transition temperature of phosphatidylethanolamines by the addition of very low levels of added diacylglycerols or lyso-PC. The possibility of enzymatic (O'Sullivan et al., 1987) and/or oxidative (Gallaird, 1980) breakdown of lipids during the isolation of thylakoids is high, but is routinely ignored by most investigators. Consistent with this interpretation, the published literature on the occurance of phase separations in thylakoids of chilling-sensitive (and chilling-resitant) plants is in extensive disagreement (see General Discussion). A causal relationship between gel/liquid-crystal phase separation and increased solute permeability is crucial to the chilling sensitivity hypothesis. Increased permeation near the transition temperatures may be expected to arise due to either: a) packing defects, or instability, at the boundary of the solid-fluid phase interface due to different lipid packing 153 arrangements, and/or; b) increase lateral density fluctuations of the lipids, arising from greater lateral compressibility of lipids near their phase transition temperature, may increase the frequency of appearance of "statistical pores" through which solutes might pass (reviewed recently by Deamer and Bramhall, 1986). It has been demonstrated that there is a lipid dependence to increased permeation near phase transition temperatures. Saturated PC's, but not saturated phosphatidylethanolamines (PE's), are very permeable to Na+ and glucose (Singer, 1981) as well as K* (Noordam et al., 1982) at their respective transition temperatures. Further, equimolar mixtures of saturated PC's and PE's were not more permeable to K* at their transition temperatures than in the liquid-crystal phase despite the presence of a well defined phase separation (Noordam et al., 1982) detected by DSC. In summary, the calorimetric and permeability properties of vesicles composed of lipid mixtures mimicking those of chilling-sensitive and chilling-resistant plants have been examined. DSC curves indicated that no gel/liquid- crystal phase separation occurred for any unfrozen vesicle sample containing 4%, 8%, or 12% DPPG. These vesicles did not, in addition, show increased permeability to Rb+ in the same temperature range. It is concluded, therefore, that a direct effect of disaturated PG on chilling injury by increasing low-temperature permeability is unlikely. 154 7. GENERAL DISCUSSION 1. The importance of well-characterized systems An important aspect of the work reported in this thesis is the use of well-defined lipid and vesicle systems. This includes the characterization of the component lipids for structure, purity, and fatty acid compositon; the reconstitution of vesicles of defined, and confirmed, lipid composition as well as defined size and number of bilayers; and the reconstitution of vesicles experiencing a known electrochemical gradient Each of these topics will be discussed in turn. The importance of using lipids of defined purity is highlighted by results presented in Chapter 4. The ability to observe DGDG vesicle aggregation is strongly dependent on the proportions of charged lipids present in the galactolipid (Figure 23). The presence of a 1 mol% contaminant would eliminate vesicle aggregation in 10 mM MgCl2 (Gounaris et al., 1983c), and probably in 10 mM CaCl2 (Sen et al., 1982b) as well as 0.1 M NaCl (Sen et al., 1981b, 1982b), by the development of an electrostatic repulsive force. This may explain the failure of previous workers to observe galactolipid aggregation in salt solutions. Purities of £ 99.5 mol% have been obtained here (Table 3), whereas it is not uncommon for workers to use galactolipid preparations of approximately 95 mol% purity (cf. Waggoner et al., 1985). Until now, no clear criteria for the minimum purity of galactolipids have been available. In addition, it should be added that the visualization of trace contaminants in lipid preparations is not simple. The ability to detect a contaminant by TLC is dependent on the adsorbent and solvent system used, on the degree of loading of the sample, on the degree of separation of the major compound and the contaminant as well as on the detection reagent The visualization of contaminants present at 2-3 mol% is very difficult using non-destructive 155 reagents such as 2',7,-dichlorofluorescene or Rhodamine 6G even with heavy sample loading. As a result, the commonly-occurring phrase "purity checked by TLC" has little meaning without specification of the separation system, sample loading, and detection methods. As a further example of the importance of lipid purity, Epand (1985) and Epand et al. (1988) have reported that trace amounts of lyso-phospholipids and diacylglycerols have a significant impact on the L0~H.jj transition temperature of unsaturated PE's. These additions did not affect the temperatures of the L/3-La transition (Epand, 1985). However, the addition of lipids in larger proportions is known to broaden the temperature range, and mminish the enthalpy, of L/3-La transitions of a wide range of phospholipid and phospholipid-sterol mixtures. These considerations may explain why no endotherms were detected in DPPG-containing vesicles (Figures 31 to 34) while such transitions have been detected by other workers in this temperature range. Many investigators have used relatively uncharacterized "total polar lipid" dispersions (Fork et al., 1981; Pike, 1982; Raison and Wright, 1983; Raison and Orr, 1986a) while other workers have examined whole thylakoid preparations, also of unexamined lipid composition (Fork et al., 1981). The enrichment of high-melting point lipids in these preparations is a possibility. In addition, if catalytic degradation were to occur during the isolation of these preparations (Sastry and Kates, 1964; Gallaird, 1980; O'Sullivan et al., 1987), the release of high-melting point products may generate the species undergoing L/3 phase formation. The temperature of the L/3 - L a phase transition for phosphatidic acid, a possible degradation product, is higher than that for similar molecular species of PE (Tilcock, 1986). It has been possible here to produce vesicles of well-defined lipid composition, including high proportions of MGDG (Tables 8, 11 and 14). These results are contrary to those of Sprague and Staehelin (1984a,b) who reported that quantitative transfer of > 30 mol% MGDG to the aqueous phase could not be accomplished by conventional hydration 156 methods. The reason for the disparity between their results and those reported here is not clear. It may be that preferential solubility of certain lipids over others in organic solvents leads to lipid phase separation during the solvent removal prior to vesicle dispersal. Such separation would be assisted by slow solvent removal. This would, of course, promote the existence of phase separated regions in the dispersed lipid mixture. While the routine characterization of vesicular lipid composition after experiments is easy and rapid, it has not been often reported in liposome experiments. The use of vesicles of defined size distribution has been crucial to the results reported here. In general, large unilamellar vesicles show several advantages over MLV and SUV preparations. These advantages include 1) differential surface tensions on inner and outer monolayers of small vesicles is avoided, 2) transbilayer asymmetry of lipid distribution induced by surface tension and head group packing constraints are avoided, 3) high trap volumes can be obtained, and 4) in permeability experiments, unilamellar vesicles avoid interpretational difficulties arising from multiple barriers to solute movement in MLV systems. The use of LUV's has allowed the unambiguous conclusion that DGDG aggregation is reversible (Table 4) and that £ 25 mol% MGDG triggers irreversible aggregation of binary galactolipid vesicles (Figure 27). Such demonstrations would be extremely difficult with the polydisperse preparations usually obtained by conventional lipid dispersal methods. In addition, effects of size have been reported in the coupling ratio of reconstituted cytochrome c oxidase (Madden et al., 1984) and in the permeability properties of phospholipid vesicles (Deamer and Bramhall, 1986). • Finally, flux measurements have been performed under conditions in which the membrane potential, A«P, should be zero. This has been accomplished by the inclusion of the ionophore carrying the counterion of the ion of interest The ionophore should drop the 157 energy barrier to counterion diffusion to minimal levels and the development of a A * inhibitory to efflux of the ion should not occur. If this were not done then the flux of the measured ion would become coupled, via the A*, to the flux of the slower diffusing counterion (Deamer and Bramhall, 1986). This may explain why the permeability coefficient reported here for PC is somewhat higher than reported by other workers (Toyoshima and Thompson, 1975). 2. Surface properties and vesicle aggregation A large part of the results in this thesis point to the importance of the surface properties of the thylakoid galactolipids as determinants of bilayer behavior. In particular, these results emphasize the effects of the degree of hydration of the bilayer surface and the orientation of the hydrating water molecules. Evidence that aggregation does not occur via double-layer effects includes: l)the lack of any detectable charged contaminant at very stringent levels (Chapter 2), 2) the lack of a decrease in the DGDG lamellar repeat in the presence of salt (Table 6), 3) and the effect of glycerol as an inhibitor of aggregation (Figure 21). While the differential efficacies of the cations (Figure 18) suggest that classical DLVO theory cannot explain vesicle aggregation, some specificity of ion association to charged phospholipid bilayers is known (Hauser et al., 1976), and might also occur with galactolipid dipoles. If galactose residues do indeed possess a small, positive outside, dipole potential similar to the cerebrosides (Maggio et al., 1978) then aggregation could be due to some degree of double-layer screening by added anions. If so, then the different efficacies of the cations (Figure 18), the X-ray diffraction data (Table 6), and the glycerol inhibition of aggregation (Figure 21) would remain unexplained. 158 Data presented in Figure 23 and Table 9 show strong but different effects of added anionic lipids on the degree of aggregation of galactolipid vesicles. These results show that the effects of the two lipids PG and SQDG cannot be lumped together simply because they possess the same charge. The contribution of each to the bilayer surface charge density is an interaction of both lipid concentration in the bilayer and the divalent ion concentration. PG may have rninimal impact on increasing the surface charge density of both vesicles and thylakoids if, as is argued, Mg 2 + forms a bidentate ligand with PG, at proportions higher than 5 mol%, and neutralizes the phosphate negative charges. Barber and Gounaris (1986) have argued that most or all of the SQDG present in the thylakoid is associated with, and charge neutralized by, integral membrane proteins. This position was based on the observation that sufficient SQDG is present in the thylakoid to account for the entire surface charge density of these membranes. Barber and Gounaris (1986) pointed out that it is known that most or all of this surface charge is derived from proteins, and therefore, the SQDG charges must be neutralized in vivo. Similar screening considerations could well apply to PG as to SQDG since PG levels in thylakoids are similar to, or exceed, those of SQDG. The aggregation of DGDG and DGDG/MGDG vesicles is most likely due to an effect of added salts on hydrating water layers. It is interesting to note that DGDG saturated by an average of about 2 double bonds per molecule (Table 7) aggregated at the same concentration of MgCl2 as DGDG purified from spinach (Figure 25). An effect of acyl saturation on the aggregation of DGDG vesicles may not be expected if such interaction is dominated by the head groups. However, it is known that saturated lipids are less hydrated than unsaturated lipids (Sen and Hui, 1988), probably as a result of reduced area per molecule and reduced spacing between adjacent head groups. It would be interesting to test for the aggregation of fully saturated DGDG, since the "18:2-DGDG" used here is still polyunsaturated and retains a high area per molecule. Such an experiment would have to be 159 done at high temperatures ( £ 50° C), those above the L0-La transition, so as to avoid aggregation due to the gel state (Wong and Thompson, 1982). The participation of both galactolipids MGDG and DGDG (Figure 24), but not the glucolipid DglcDG from A. laidlami, suggests a specific role for the galactose residue in membrane close approach. As outlined in Chapter 4, however, there are some technical problems with the determination of AAJO for DglcDG. As a result these data must be taken, for the time being, as preliminary. Nonetheless, interactions of ions with cerebroside monolayers by changes in water structure (Johnston et al., 1985) and with the hydration of DglcDG (Wieslander et al., 1978) have been reported. In addition, an effect on the ion-binding of a disaturated PG by dimannosyldiacylglycerol from Micrococcus luteus, possibly mediated by hydration effects, has been reported (Lakhdar-Ghazal and Tocanne, 1988). The degree of hydration of galactolipids has been measured by several groups. Most workers agree that the degree of hydration of DGDG is very low (McDaniel, 1988; Rand and Parsegian, 1988) and probably similar to, or less than, the hydration of PE at 7-12 H20/lipid (Sen and Hui, 1988), and DglcDG at 8 H20/lipid (Wieslander et al., 1978). A value of 14-15 H20/lipid has been reported for DGDG (Brentel et al., 1985). The single dissenting value is a recent report of 50 H 2 0/DGDG measured in micelles of DGDG in hexane (Sen and Hui, 1988). Very similar degrees of hydration have been reported for MGDG and DGDG at 12-13 and 14-15 H20/lipid respectively (Brentel et al., 1985) and are not inconsistent with somewhat lower numbers of 5 H20/lipid reported for both MGDG (Sen and Hui, 1988) and distearoyl-MGDG (Sen et al., 1983). 160 3. A possible mechanism for galactolipid vesicle aggregation It was suggested in Chapter 3 that an explicit mechanism for salt-induced aggregation could not be described. However, a rather more speculative model is presented here. It is suggested that both galactolipids hydrate strongly to structured water. This water may exist as a large array, or wave, of liquid water (Watterson, 1987a,b,c, 1988). This unit of liquid water was estimated to be composed of a cubic unit 3.4 nm on a side and to give rise to the strong hydration repulsion force arising between most macromolecules (Watterson, 1987c, 1988). Hydrogen-bonding of the galactose residues to this "structured" water is proposed to be disrupted by the addition of salts, particularly chaotropic salts such as Cs* and Rb*. Removal of the ability of the galactolipids to hydrogen-bond to water results in an energetically unfavourable higher entropy state of reordering of water and/or galactose orientations to accomodate each other. This situation may be resolved by either: 1) direct hydrogen-bonding between galactose residues of facing bilayers, or: 2) hydrogen bonding between galactoses of facing bilayers through 1-2 "free" H 2 0 molecules. The former, and less likely, case would represent aggregation in the primary energy minimum of direct molecular contact An arrangement similar to the latter case has been proposed recently as the new "hydration-attraction" force postulated to exist between bilayers of PE (Rand et al., 1988). This suggestion is consistent with the previous discussion outlining the case against aggregation as a consequence of electrostatic screening and Van der Waals attractioa Further, it explains the effects of glycerol as follows. If glycerol were to replace water as the solvent hydrogen-bonding to galactose residues then salt addition would be unable to disrupt the solvent-solute interaction and, hence, the solubility of the bilayer lipids. As a result glycerol would be observed as an inhibitor of salt-mediated aggregation (Figure 21). This proposal is also consistent with the finding of free energies of adhesion for DGDG vesicles twice those 161 of PE vesicles and 20-fold times those of PC vesicles (Evans and Needham, 1987, 1988). In addition, DGDG has been reported to possess a Hamaker coefficient of 3-1011 J (Evans and Needham, 1988) to 7.5-10"21 J (Marra, 1985, 1986), values 2.5 to 6 fold higher than those found for PC and PE. It should be added that in DGDG monolayers mechanically forced together (Marra, 1986) in 0.8 M NaCl, no salt remained between the monolayers at the "adhesive contact" separation of 0.6 nm. This fact also suggests the lack of a direct interaction between ions and galactolipids of the type that might be expected for electrostatic screening. 4. Role of the galactolipids in thylakoid stacking It seems pertinent at this point to consider the role of galactolipids in granal stacking in vivo. It has been proposed (Murphy, 1986b) that the function of galactolipids in thylakoid stacking is the promotion of a low thylakoid surface charge density that might otherwise inhibit membrane close approach. It has also been suggested (see Introduction Section 4.4.) that PG and 16:1 may have a role in stacking by causing the oligomerization of LHCII to LHCII*, commonly considered the active component of granal stacking. Analysis of the contributions of electrostatic repulsive and van der Waals attractive forces to granal stacking by classical DLVO theory has been performed (Sculley et al., 1980; Barber, 1980; Thome and Duniec, 1983). Making some assumptions about the surface charge density, Hamakers coefficients, protein/lipid ratios, and dielectric constants, Sculley et al. (1980) showed that under conditions where granal stacking is stable, attractive van der Waals forces do not exceed, or at best equal the repulsive electrostatic forces and stacking cannot, theoretically, occur. The analyses by Barber (1980) and Thome and Duciec (1983) agreed with these workers. The problem may be avoided in vivo by the reduction of surface charge 162 density by cation binding by proteins and/or the lateral segregation of charged protein components (Sculley et al., 1980, Barber, 1980). The reduction of hydration repulsion due to the replacement of adsorbed cations with smaller hydrated protons has also been proposed as a mechanism for allowing attractive forces to overwhelm repulsive forces in thylakoids (Thome and Duniec, 1983). It is proposed here that the attractive forces that promote the aggregation of galactolipid vesicles in salts represent the "missing force" that unambiguously promotes granal stacking. It has been argued in Chapters 3 and 4 that vesicle aggregation occurs in physiologically relevant lipid mixtures and at lipid and salt concentrations similar to those in chloroplasts. The reduction of aggregation by added anionic lipids PG and SQDG (Figures 23 and 24 and Table 9) may not be inhibitory in vivo for the following reasons. The proportions of SQDG in the bilayer have probably been overestimated by the use of 12 mol% SQDG. Further, Barber and Gounaris (1986) have argued that most or all of the charge due to SQDG must be charge neutralized by SQDG association with protein. A similar argument may also apply for PG charges by neutralization with either protein or M g 2 \ as a bidentate ligand (Figure 23). Finally, the use of bulk Mg 2 + concentrations to evaluate aggregation may underestimate the ion concentration present at the bilayer surface due to double-layer effects and unknown activity coefficients (Barber, 1980). Given these considerations, the A A 5 0 value of 49 mM MgCl2 in 0.2 M KCI for DGDG/MGDG/SQDG/PG (1/2/0.5/0.5) vesicles (Table 10) is of the same order of magnitude as the 18 mM bulk Mg 2 + concentraton recentiy measured for spinach chloroplasts (Schroppel-Meier and Kaiser, 1988). This proposal is consistent with the observation that the galactolipids occur throughout the plant kingdom in all species that show appressed thylakoid membranes (Introduction Section 2.2.). This is also consistent with the observation that stacking occurs at normal or slightly reduced levels in mutants lacking chlorophyll b, and presumably the LHCII, but 163 retaining normal galactolipid levels (Bolton et al., 1978). This proposal may also explain, in part, the observations that trypsin treated thylakoids can be induced to stack at higher salt concentrations (Carter and Staehelin, 1980; Jennings et al, 1981) and that thylakoid stacking has been observed in membranes suspended in salt-free buffer (Gross and Prascher, 1974). It should be made clear that vesicle adhesion is not being proposed as the primary cause of granal stacking. The lack of an effect of trypsin digestion on DGDG vesicle aggregation (Chapter 4) and its rapid inhibition of thylakoid stacking (Gerola, 1981) indicates that DGDG does not maintain nor regulate the degree of membrane appression. It is suggested only that inter- galactolipid forces significantly reduce the energy barrier of initial membrane close approach and, hence, push the balance of forces unambiguously in favour of membrane appression. An important consequence of membrane appression in galactolipid vesicles containing MGDG is irreversible aggregation (Figure 27). These data are interpreted to indicate the occurrence of MGDG-triggered vesicle fusion. However, the assays employed here are not diagnostic of fusion and this conclusion, while the most likely, is tentative. The prevention of extensive bilayer fusion in thylakoids may be due to a stabilizing influence of proteins on MGDG phase behavior. Alternatively, the 2-4 nm gap between appressed granal membranes (Heslop-Harrison, 1963; Nicolson, 1971; Nir and Pease, 1973) may be too large to permit the bilayer de-stabilization that precedes fusion. In this sense, MGDG is distinct from the other HJJ phase forming species, such as PE, in other membranes that have been proposed to promote bilayer fusion (Cullis and Hope, 1978, 1988). In order to maintain the structural and functional capacity of the thylakoid, the tendency of MGDG to induce membrane fusion must be prevented. A stabilization of MGDG by its involvement in the packing of proteins (Murphy, 1982, 1986a,b) is possible. 164 5. Permeability of vesicles and thylakoids Data presented in Chapter 5 argue that the differences in permeabilities of DGDG and PC vesicles cannot be ascribed to differences in fluidity of the bilayer. Scarpa and De Gier (1971) have shown small increases in K* permeability of PC vesicles with increasing unsaturation, but not to the extent observed here. In addition, it is clear that fluidity increases cannot explain the data since DGDG permeability to Rb* was higher than that of PC, but the reverse situation is true for CY permeability. These considerations point to a role of electrical properties in regulating the degree of permeation of ionic solutes such as K* and CI -. This conclusion was reached in some of the earliest studies on ion permeation of phospholipid bilayers (Papahadjopoulos and Watkins, 1967) and is still under debate (Flewelling and Hubbell, 1986). The recent quantitative treatment (Flewelling and Hubbell, 1986) of dipole contribution to the total energy barrier to ion permeation has stressed the dominance of ester dipoles as inhibitors of cation permeation and promoters of anion permeation. This treatment ignores the contribution of dipoles associated with the head groups and bound water molecules, and, therefore is inconsistent with data presented here. It is likely that oriented dipole vectors must be invoked to explain the CI/K* specificity of phospholipid bilayers. However, these dipoles cannot be attributed predominantly to ester dipoles (Flewelling and Hubbell, 1986) as DGDG and PC would show similar C1VK+ permeability ratios. Further, different Cl"/K* permeability ratios are observed among a variety of phospholipids (Papahadjopoulos and Watkins, 1967). Similarity, the difference cannot arise from dipoles originating in the head group structure as both digalactosyl and phosphorlycholine residues are expected to have positive-outside dipoles (Maggio et al., 1978; Flewelling and Hubbell, 1986). The only possible remaining dipole contributor is from oriented water molecules at the bilayer surface. Both the degree of 165 hydration of the head group and the polarization of water may affect the contribution of a water dipole to a total bilayer dipole potential profile. It is interesting to note, in this vein, that the identically charged lipids PG and SQDG appear to have opposite effects on Rb* permeability of mixed lipid vesicles (Table 13). Furthermore, it is interesting that PC is more hydrated than PS (Sen and Hui, 1988) and DGDG (Brentel et al., 1985) and has the highest CT/K* tested (Papahadjopoulos and Watkins, 1967; Table 12). This interpretation of the importance of head group hydration to ion permeability is consistent with the observation (Table 13) that addition of MGDG up to 50 mol% did not significantly affect Rb* permeability of these vesicles. That is, at MGDG proportions in which the bilayer is susceptible to destabilization and fusion (Figure 27), permeabiity was not affected (Table 13). This indicates: 1) that neither inverted micelles nor H ^ phase were present in these bilayers (in which case MGDG permeability is identical to DGDG permeability) or; 2) the presence of inverted micelles in these bilayers did not increase Rb* permeability, as expected if interfacial regions dominate permeability properties. A causal relationship between head group hydration and bilayer permeability would be of great interest A link between membrane fluidity and glucose permeability is implied by the data presented in Table 12. Increased acyl unsaturation has been associated with increased permeability of non-electrolyte solutes to lipid vesicles and A. laidlawii cells (Romijn et al., 1972; McElhaney et al., 1973; Demel et al., 1968; De Gier et al., 1968, 1971). In DGDG vesicles the degree of order in the upper regions of the acyl chains was very similar to that found in PC systems (Table 12). Near the methyl terminal end of the chain the rotational correlation times were approximately 50% of those for PC vesicles (Table 12) indicating molecular rotation at about twice the rate of that in PC. Similarily, the permeability coefficient for glucose efflux from DGDG vesicles was similar to that for PC vesicles. 166 It is likely that different solutes permeate bilayers by a range of different mechanisms. This is indicated by known effects of bilayer thickness and the number of double bonds on solute permeation through membranes (Deamer and Bramhall, 1986). Alternatively, some solutes may penetrate by passing through so-called "statistical pores" (Trauble, 1971). The importance of different permeation mechanisms is most clearly demonstrated by the extremely high permeability of bilayers to H*. Evidence has been reviewed (Deamer and Bramhall, 1986) suggesting that H* flux may be mediated via a water-wire, or row, oriented within the acyl region of the bilayer. The results with galactolipid and mixed lipid vesicles (Table 13) clearly indicate that the high ionic permeability of thylakoid membranes (Avron, 1977) is caused by the permeability properties of their constituent lipids. While the addition of proteins to pure lipid vesicles results in increased passive bilayer permeability (Van Hoogevest et al., 1983; Romans et al., 1981; Van der Steen et al., 1982), vesicles mimicking lipid compositions found in vivo, are already three orders of magnitude more leaky than phospholipid-based vesicles. The permeability coefficients obtained for vesicles are very similar to those for NaCl, KC1, and CI" from spinach and Beta vulgaris thylakoids (Barber, 1972; Ball et al., 1985). This suggests that the thylakoid lipids determine the permeability properties of the intact thylakoid membrane. This conclusion is in contrast to the conclusions of Barber (1972), Ball et al., (1985) and Ort (1986) that the "apparent" leakiness of the thylakoid arises from the exceptionally high surface/volume ratio of thylakoid membranes. It is proposed, therefore, that galactolipid permeability properties are the cause of the low A * and high ApH found in thylakoids. It could be argued that such a proposal is dependent on the assumption of normal galactolipid permeability to protons. That this is a valid assumption is supported by the following. Firstly, it is clear that thylakoids are not extremely leaky to protons as they maintain a steady-state transmembrane gradient of 3-4 pH 167 units (Avron, 1977). Secondly, the high permeability of galactolipid vesicles to Rb* does not imply high permeability of protons if, as mentioned above, protons permeate by a mechanism distinctly different from those of the other ions (Deamer and Bramhall, 1986). Thirdly, measurements of proton flux in DGDG MLV's has suggested that the galactolipid shows identical proton permeability to PC MLV's (Foley et al., 1988). On the other hand. Pick et al., (1984, 1987) have claimed that the proton permeability of chloroplast lipids is far higher than that of phospholipids. These workers, however, never measured lipid proton permeability, only the permeability of cholate/CF0-CF!/lipid proteoliposomes. Finally, in thylakoids, proton flux through the membrane occurs via: 1) flux through C F 0 - C F ! coupled to ATP synthesis; 2) flux through CF 0 -CFi not coupled to ATP sysnthesis, and; 3) non-specific leakage of the thylakoid itself. Measurements of non-specific proton flux in dark thylakoids indicates that this leakage contributes only 1% of the total thylakoid proton flux and has a permeability coefficient of 2.0 to 3.0-10"5 cm-s1 (Schonfeld and Schickler, 1984; Schonfeld and Kopeliovitch, 1985). This value is very similar to values of 10"3 to 10"5 cm-s-1 reported for a variety of lipid vesicle and membrane systems (Deamer and Bramhall, 1986). 6. Chilling Stress and Membrane Permeability If, as suggested in the preceeding discussion, the permeablity of bilayers to ionic solutes is dominated by electrical considerations, then moderate changes in acyl packing should not be expected to significantly alter bilayer ionic permeability. That is the lateral phase separation of some lipid to yield "patches" of L0 phase in an otherwise La bilayer should not significantly alter the bulk dipole moment that a diffusing ion would experience. Indeed, Lj3 phase formation results in greatly increased ion and sugar permeability in saturated PC's but not in saturated PE's (Singer, 1981; Noordam et al., 1982). This was interpreted by Noordam et al. (1982) as indicating that hydrogen-bonding between adjacent PE head groups 168 was too strong to allow the lateral lipid density fluctuations presumed to be the cause of increased ion permeation at the transition temperature for PC's. It is clear that the formation of a mixed L/3-La does not necessarily result in increased bilayer permeability to salts. The suggestion (Lyons, 1973; Murata, 1983) that the formation of L/3 phase lipid in thylakoids of higher plants, from DPPG and 16:0/16:1-PG, would increase thylakoid permeability has little empirical support Therefore, even if phase separation had been detected in lipid mixtures containing DPPG (Figures 31 to 34), the lack of an effect on Rb+ (Figures 35 and 36) would not be inconsistent This is further shown by the finding in Ca 2*-PG systems that DCS-detected endotherms do not necessarily represent L/3-La phase separation, but possibly other gel-state structural rearrangements (Borle and Seelig, 1985). If electrical properties do indeed dominate ion permeation, then it would be of some interest to repeat the flux experiments in Chapter 6 with a non-charged solute such as glucose. Fluorescence changes consistent with phase separation have been reported in semi-purified polar lipid extracts (Fork et al., 1981; Pike, 1982) and chloroplasts (Fork et al., 1981) from chilling sensitive species. In addition, DSC-detected endotherms have been detected at 10-12°C in total thylakoid polar lipid extracts of sensitive plants (Raison and Wright 1983; Raison and On, 1986a). However, contrary results have been reported by Low et al. (1984) who observed no correlation between chilling-stress and endotherms in either thylakoids or their polar lipids from various plant species. In addition, Low et al. (1984) observed endotherms in the membrane lipids of spinach, a species generally considered to be chilling-resistant Raison and Wright (1983) have reported that the addition of small quantities (<^ 5 mol%) of DPPG or DPPC to thylakoid polar lipids triggered the appearance of DSC endotherms in the 10° C range. In addition, enthalpy calculations suggested that other lipids in addition to the disaturated phospholipid were undergoing the transition to L/3 phase. 169 Results reported here are not in agreement with those of Raison and Wright (1983) and of other workers showing phase separation in thylakoids of chilling-sensitive plants. It has been stated previously that the phase behavior of galactolipid mixtures is markedly dependent on the dispersion method (Sprague and Staehelin, 1984a,b, and this work). One of the dispersion methods used by Raison and Wright (1983) was a poorly characterized passive overnight hydration technique. The other method used freeze-thaw vesicles which, in this work, caused the aggregation of vesicles, lipid demixing and the appearance of endotherms in the 16-30°C range that were not visible in unfrozen vesicles. The interpretation here that DPPG at concentrations up to 12 mol% does not phase separate from the galactolipids is consistent with the report that up to 50% DPPG in bilayers of dimannosyldiacylglycerol from Micrococcus luteus does not phase separate at 20° C even in the presence of Na* or Mg 1 + (Lakhdar-Ghazal and Tocanne, 1981). These authors suggested that the formation of extensive non-ionic hydrogen-bonding between DPPG and the glycolipid caused the high miscibility of DPPG with the glycolipid, a possibility that might explain the lack of phase separation reported here. Results presented here suggest that such phase changes may not even occur in cold-stressed chilling-sensitive plants. The influence of proteins on the phase behavior of these lipids is also unknown. Calculations by Murphy and Woodrow (1983) suggest that there is barely sufficient lipid present in the appressed membranes to form a monolayer of solvating lipid around the proteins. If true, then the existence of bulk phase lipid available to undergo phase transitions is questionable. As a final comment on the chilling-sensitivity hypothesis Unking gel phase lipid to low temperature damage, it should be noted that another proposal of the theory is that gel phase lipid may inhibit or destroy the activity of membrane bound enzymes (Raison, 1973; Lyons, 170 1973). Many investigators have looked for a correlation between the temperature of the presumptive lipid phase separation and the temperature of "break points" in the Arrhenius plots of enzyme activity against 1/T. Nonetheless, these sorts of studies have suffered extensive theoretical criticism (Silvius and McElhaney, 1978; Silvius et al., 1981; Kubo, 1985). In support of such criticisms empirical evidence has shown that break points found in the Arrhenius plots of plasma membrane ATPase activity could not be detected in fluidity measurements of plasma membranes nor of their purified lipids (Ishakawa and Yoshida, 1985; Yoshida et al., 1986). To summarize, data presented in this thesis have been interpreted to suggest that galactolipid hydration is very important in determining the physical characteristics of galactolipid vesicles, and of vesicles made of mixtures of thylakoid lipids. The degree of head group hydration, and the influence of dissolved solutes on hydration, may have marked effect on the interactions between galactolipid vesicles. The possible physiological relevance of this behavior is unclear at this time. An apparent consequence of the galactosyl head group structure is the unusual permeabiity properties of vesicles of these lipids compared with those of the phospholipids. The relevance of these results to thylakoid energetics has been discussed. Detailed knowledge of the orientation and structure of the galactosyl head groups would be helpful in interpreting some of this data. Future work could also be directed to the accurate measurement of head group hydration in a variety of aqueous solutions and the direct measurement of the forces between galactolipid vesicles. 171 8. REFERENCES 1. Allen, C.R. and Good, P. 1971. Meth. Enzymol. 23: 523-547 2. Allred, D.R. and Staehelin, L.A. 1986. J. Bioenerg. Biomemb. 18: 419-436 3. Altenbach, C. and Sellig, J. 1984. Biochem. 23: 3913-3920 4. Anderson, J.M. 1981. F.E.B.S. Lett. 124: 1-10 5. Anderson, J.M. 1982. Photobiochem. Photobiophys. 3: 225-241 6. Anderson, J.M. and Barrett, J. 1986. In: Encyclopedia of Plant Physiology, Vol. 19, pp. 269-285 7. Anderson, M.M., McCarty, R.E, and Zimmer, EA. 1974. Plant Physiol. 53: 699-704 8. Andersson, B. 1986. In: Encyclopedia of Plant Physiology, Vol. 19, pp. 447-456 9. Amon, D.I. 1949. Plant Physiol. 24: 1-15 10. Ashworth, EN., Christiansen, M.N., St. John, J.B., and Patterson, G.W. 1981. Plant Physiol. 67: 711-715 11. Avron, M. 1977 Annu. Rev. Biochem. 46: 143-155 12. Ball, M.C., Mehlhom, R.J., Terry, N., and Packer, L. 1985. Plant Physiol. 78: 1-3 13. Bangham, A D . 1968. Prog. Biophys. Mol. Biol. 18: 31-95 14. Barber, J. 1972. Biochim. Biophys. Acta 275: 105-116 15. Barber, J. 1980. Biochim. Biophys. Acta 594: 253-308 16. Barber, J. 1982. Annu. Rev. Plant Physiol. 33: 261-295 17. Barber, J. and Gounaris, K. 1986. Photosyn. Res. 9: 239-249 18. Bearer, E E , Duzgunes, N., Friend, D.S., and Papahadjopoulos, D. 1982. Biochim. Biophys. Acta 693: 93-98 19. Bennett, J. 1983. Biochem. J. 212: 1-13 20. Bishop, D.G. and Kenrick, J.R 1987. Phytochem. 26: 3065-3067 21. Bishop, D.G., Kenrick, J.R., Bayston, J.H., Macpherson, AS., and Johns. S.R. 1980. Biochim. Biophys. Acta 602: 248- 259 22. Blackwell, M.F. and Whitmarsh, J. 1989. Manuscript submitted 172 23. Blackwell, M.F., Gounaris, K., Zara, S.J., and Barber, J. 1987. Biophys. J. 51: 735-744 24. Bolton, P.. Wharfe, J. and Harwood, J.L. 1978. Biochem. J. 174: 67-72 25. Bond, A.M., Deprez, P.P., Jones, R.D., Wallace, G.G., and Briggs, M.H. 1980. Anal. Chem. 52: 2211-2213 26. Borle, F. and Seelig, J. 1985. Chem. Phys. Lipids 36: 263-283 27. Brasseur, R., De Meutter, J., Goormaghtigh, E., and Ruysschaert, J.M. 1983. Biochem. Biophys. Res. Commun. 115: 666-672 28. Brentel, I., Selstam, E , and Lindblom, G. 1985. Biochim. Biophys. Acta 812: 816-826 29. Browse, J., McCourt P., and Somerville, CR. 1985. Sci. 227: 763-765 30. Browse, J., McCourt P., and Somerville, C. 1986. Plant Physiol. 81: 859-864 31. Brunner, J.. Graham, D.E, Hauser, H., and Semenza, G. 1980. J. Memb. Biol. 57: 133-141 32. Buege, J.A., and Aust S.D. 1978. Meth. Enzymol. 52: 302-310 33. Burns, D.D., Gallaird, T., and Harwood, J.L, 1977a. Biochem. Soc. Trans. 5: 1302-1305 34. Burns, D.D., Gallaird, T., and Harwood, J.L. 1977b. Phytochem. 16: 651-654 35. Caron, L , Dubacq, J.P., BerkalorT, C , and Jupin, H. 1985. Plant & Cell Physiol. 26: 131-139 36. Carter, DP. and Staehelin, LA. 1980. Arch. Biochem. Biophys. 200: 374-386 37. Chain, R.K. 1985. F.E.B.S. Lett 180: 321-325 38. Chapman, D.J., and Barber, J.. 1987. Meth. Enzymol. 148: 294-319 39. Chapman, D.J., De-Felice, J. and Barber, J. 1983. Plant Physiol. 72: 225-228 40. Chow, W.S., Wagner, W., and Hope, A.B. 1976. Aust J. Plant Physiol. 3: 853-861 41. Christie, W.W. 1982. "Lipid Analysis", 2nd ed., Pergamon Press, Oxford. 42. Coddington, J.M., Johns, S.R., Leslie, D.R., Willing, R.I., and Bishop, D.G. 1981. Aust J. Chem. 34: 357-363 43. Comfurius, P. and Zwaal, R.F.A. 1977. Biochim. Biophys. Acta 488: 36-42 44. Conway, B.E 1981. "Ionic Hydration in Chemistry and Biophysics", Elsevier, Amsterdam 45. Cramer, W.A., Whitmarsh, J., and Low, P.S. 1981. Biochem. 20: 157-162 173 46. Critchley, C. and Heinz, E 1973. Biochim. Biophys. Acta 326: 184-193 47. Crofts, A.R. and Wraight, CA. 1983. Biochim. Biophys. Acta 726: 149-185 48. Cullis, P.R. and De Kruijff, B. 1979. Biochim. Biophys. Acta 559: 399-420 49. Cullis, P.R. and Hope, M.J. 1978. Nature (London) 271: 672-675 50. Cullis, P.R. and Hope, M.J. 1988. In: Molecular Mechanisms of Membrane Fuson, (Ohki, S., Doyle, D., Flanagan, T.D., Hui, S.W., and Mayhew, E , eds.) Plenum Press, New York, pp. 37-51 51. Danielius, R.V., Satoh, K., van Kan, P.J.M., Plijter, J.J., Nuijs, A M . and van Gorkom, H.J. 1987. F.EB.S. Lett 213: 241-244 52. Deamer, D.W. and Bramhall, J. 1986. Chem. Phys. Lipids 40: 167-188 53. Deamer, D. and Packer, L 1969. Biochim. Biophys. Acta 172, 539-545 54. De Gier, J., Mandersloot, J.G., and Van Deenen, L L M . 1968. Biochim. Biophys. Acta 150: 666-675 55. De Gier, J., Mandersloot J.G., Hupkes, J.V., McElhaney, R.N., and van Beek, W.P. 1971. Biochim. Biophys. Acta 233: 610-618 56. Demel, R.A, Kinsky, S.C., Kinsky, C.B., and van Deenen, L L M . 1968. Biochim. Biophys. Acta 150: 655-665 57. Demmig, B. and Gimmler, H. 1983. Plant Physiol. 73: 169-174 58. DeYoe, D.R. and Brown, G.N. 1979. Plant Physiol. 64: 924-929 59. Dijkmans, H. and Daniel, A . - M . 1981. Can. J. Biochem. 59: 16-21 60. Diner, B.A 1986. In: Encyclopedia of Plant Physiology, Vol. 19, pp. 422-436 61. Dluhy, R A , Cameron, D.G., Mantsch, H.H., and Mendelsohn, R. 1983. Biochem. 22: 6318-6325 62. Douce, R and Joyard, J. 1979. Adv. Bot Res. 7: 1-116 63. Dubacq, J.-P., and Tremolieres, A 1983. Physiol. Veg. 21: 293-312 64. Dunahay, T.G., and Staehelin, L A 1985. Plant Physiol. 78: 606-613 65. Eagland, D. 1975. In: "Water, A Comprehensive Treatise", (Franks, R, ed.) vol. 5, pp. 1-74, Plenum Press, New York 66. Epand, R.M. 1985. Biochem. 24: 7092-7095 67. Epand, R.M., Epand, R.F., and Lancaster, C.R.D. 1988. Biochim Biophys. Acta 945: 174 161-166 68. Evans, E and Needham, D. 1987. J. Phys. Chem. 91: 4219-4228 69. Evans, E. and Needham, D. 1988. In: "Molecular Mechanisms of Membrane Fusion", (Ohki, S., Doyle, D., Flanagan, T.D., Hui, S.W., and Mayhw, E , eds.). Plenum Press, New York, pp. 83-99. 70. Farineau, N., Guillot-Salomon, T., Tuquet, C. and Farineau J. 1984. Photochem. Photobiol. 40: 387-390 71. Fato. R., Battino, M., Castelli, G.P., and Lenaz, G. 1985. F.EB.S. Lett. 179: 238-242 72. Fato, R, Battino, M., Degli Esposti, M., Parenti Castelli, G., and Lenaz, G. 1986. Biochem 25: 3378-3390 73. Ferguson, S.J., and Sorgato, M.C. 1982 Annu. Rev. Biochem. 51: 185-217 74. Flewelling, RF. and Hubbell, W.L 1986. Biophys. J. 49: 541-552 75. Foley, A A , Brain, A.P.R, Quinn, P.J., and Harwood, J.L. 1988. Biochim Biophys. Acta 939: 430-440 76. Ford, R. and Barber, J. 1980. Photobiochem. Photobiophys. 1: 263-270 77. Ford, R. and Barber, J. 1983a. Biochim. Biophys. Acta 722: 341-348 78. Ford, R.C. and Barber. J. 1983b. Planta 158: 35-41 79. Ford, R.C., Chapman, D.J., Barber, J., Pedersen, J.Z., and Cox, R.P. 1982. Biochim. Biophys. Acta 681: 145-151 80. Forde, J., and Steer, M.W. 1976. J. Exp. BoL 27: 1137-1141 81. Fork, D.C.. Van Ginkel, G.. and Harvey, G. 1981. Plant & Cell Physiol. 22: 1035-1042 82. Franks, F. 1984. "Water", pp. 65, Royal Soc. Chem, London 83. Gaffney, B.J. 1976. in Spin Labelling: Theory and Applications (Berliner, L.J., ed.), pp. 567-571, Academic Press, New York 84. Galanopoulou, D., Williams, W.P., and Quinn, P.J. 1982. Biochim. Biophys. Acta 713: 315-322 85. Gallaird, T. 1980. In "The Biochemistry of Plants: a comprehensive treatise", Vol. 4, (Stumpf, P.K., ed.), pp. 85-116 86. Gerola, P.D. 1981. Physiol. Veg. 19: 565-580 87. Goldbeck, J.H. 1988. Biochim. Biophys. Acta 895: 167-204 175 88. Gounaris. K.. and Barber, J. 1985. F.EB.S. Lett. 188: 68-72 89. Gounaris, K., Brain, A.P.R., Quinn, P.J., and Williams. W.P. 1983a. F.EB.S. Lett 153: 47-52 90. Gounaris, K., Mannock, D.A., Sen, A., Brain, A.P.R., Williams, W.P., and Quinn, P.J. 1983b. Biochim. Biophys. Acta 732: 229-242 91. Gounaris, K... Sen, A., Brain, A.P.R., Quinn, P.J. and Williams, W.P. 1983c. Biochim. Biophys. Acta 728: 129-139 92. Gounaris, K., Sundby, C , Andersson, B. and Barber, J. 1983d. F.EB.S. Lett 156: 170-174 93. Gounaris, K., Whitford, D., and Barber, J.. 1983e. F.E.B.S. Lett 163: 230-234 94. Gounaris, K.. Barber, J., and Harwood, J.L 1986. Biochem. J. 237: 313-326 95. Graham, D., and Patterson, B.D. 1982. Ann. Rev. Plant Physiol. 33: 347-372 96. Green, B.R. 1988. Photosyn. Res. 15: 3-32 97. Griffith, O.H. and Jost P.C. 1976. in Spin Labelling, Theory and Applications, pp. 453-523, Academic Press, New York 98. Gross, E L and Hess, S.G. 1974. Biochim. Biophys. Acta 339: 334-346 99. Gross, E L and Prasher, S.H. 1974. Arch. Biochem. Biophys. 164: 460-468 100. Gruner, S.M. 1977. Ph.D. Thesis, Princeton University, Princeton, NJ 101. Gruner, S.M. 1985. Proc. Natl. Acad. Sci. 82: 3665-3669 102. Gruner, S.M., Milch, J.R., and Reynolds, G.T. 1982a. Rev. Sci. Iristrum. 53: 1770-1778 103. Gruner, S.M., Rothschild, K.J., and Clark, N.A. 1982b. Biophys. J. 39: 241-251 104. Gruner, S.M., Cullis, P.R., Hope, M.J., and Tilcock, C.P.S. 1985a. Ann. Rev. Biophys. Biophys. Chem. 14: 211-238 105. Gruner, S.M., Lenk, R.P., Janoff, A.S. and Ostro, M.J. 1985b. Biochem. 24: 2833-2842 106. Haehnel, W. 1986. In: Encyclopedia of Plant Physiology, Vol. 19. pp. 547-559 107. Harwood, J.L 1980. In: "The Biochemistry of Plants. A Comprehensive Treatise". Vol 4. (Stumpf, P.K., ed.) Academic Press, N.Y., pp 1-55 108. Hauser, H., Levine, B.A., and Williams, R.J.P. 1976. Trends Biochem. Sci. 1: 276-281 109. Hauska, G. 1986. In: Encyclopedia of Plant Physiology, Vol. 19, pp. 1-84 176 110. Hauska, G.. Samoray, D., Orlich, G., and Nelson, N. 1980. Eur. J. Biochem. I l l : 535-543 111. Heinz, E. 1967a. Biochim. Biophys. Acta 144: 321-332 112. Heinz, E. 1967b. Biochim. Biophys. Acta 144: 333-343 113. Heinz, E 1973. Z. Pflanzenphysiol. 69: 359-376 114. Heinz, E , and Siefermann-Harms, D. 1981. F.EB.S. Lett. 124: 105-111 115. Hellergren, J., Lundborg, T., and Widell, S. 1984. Physiol. Plant 62: 162-166 116. Henry, L E A . , Mikkelsen, J.D. and Moller, B.L 1983. Carlsberg Res. Commun. 48: 131-148 117. Heslop-Harrison, J. 1963. Planta 60: 243-260 118. Hiller, R.G. and Raison, J.K. 1980. Biochim. Biophys. Acta 599: 63-72 119. Hind, G., Nakatani, H.Y., and Izawa, S. 1974. Proc. Natl. Acad. Sci. 71: 1484-1488 120. Hirayama, O., Matsuda, H., Takeda, H , Maenaka, K., and Takatsuka, H. 1975. Biochim Biophys. Acta 384: 127-137 121. Hofstee, B.H.J. 1957. Biochim Biophys. Acta 24: 211-213 122. Hope, M.J., Bally, M.B., Webb, G., and Cullis, P.R. 1985. Biochim Biophys. Acta 812: 55-65 123. Horton, P., Allen, J.F., Black, M.T. and Bennett, J. 1981. F.EB.S. Lett 125: 193-196 124. Horvath, I., Vigh, L , Belea, A , and Farkas, T. 1980. Physiol. Plant 49: 117-120 125. Horvath, I., Vigh, L , and Farkas, T. 1981. Planta 151: 103-108 126. Horvath, G., Droppa, M., Szito, T., Mustardy, L A , Horvath. LI. and Vigh, L 1986. Biochim. Biophys. Acta 849: 325-336 127. Hubbell, W.L and McConnell, H.M. 1969. Proc. Natl. Acad. Sci. 64: 20-27 128. Huner, N.P.A., Krol, M., Williams, J.P., Maissan, E , Low, P.S., Roberts, D., and Thompson, J.E 1987. Plant Physiol. 84: 12-18 129. Huner, N.P.A., Williams, J.P., Maissan, E E , Myscich, E G . , Krol, M., Laroche, A., and Singh, J. 1989. Plant Physiol. 89: 144-152 130. Ikegami, I. 1983. Biochim. Biophys. Acta 722: 492-497 131. Ishikawa, M. and Yoshida, S. 1985. Plant & Cell Physiol. 26: 1331-1334 177 132. Israelachvili, J. 1985. In: "Physics of Amphiphiles: Micelles, Vesicles and Microemulsions", (Degiorgio, V., and Corti, M., eds.) North-Holland, Amsterdam, pp. 24-58 133. Israelchvili, J.N., Mitchell, D.J., and Ninham, B.W. 1977. Biochim. Biophys. Acta 470: 185-201 134. Israelachvili, J.N., Marcelja, S., and Horn, R.G. 1980. Quar. Rev. Biophys. 13: 121-200 135. Jacob, J.S. and Miller, K.R. 1986. J. Cell Biol. 103: 1337-1347 136. Jennings, R.C., Gerola, P.D., Garlaschi, F.M., and Ford, G. 1981. Plant Physiol. 67: 212-215 137. Johns, S.R. 1980. J. Proc. Roy. Soc. New South Wales 113: 69-80 138. Johns, S.R., Leslie, DR., Willing, R.I., and Bishop, D.G. 1977a. Aust. J. Chem. 30: 813-822 139. Johns, S.R., Leslie, DR., Willing, RJ., and Bishop, D.G. 1977b. Aust. J. Chem. 30: 823-834 140. Johns, S.R., Leslie, D.R., Willing, RJ.. and Bishop, D.G. 1978. Aust. J. Chem. 65-72 141. Johnston, D.S., Coppard, E , and Chapman, D. 1985. Biochim. Biophys. Acta 815: 325-333 142. Joliot, P. arid Joliot A. 1986. In: Encyclopedia of Plant Physiology, Vol. 19, pp. 528-538 143. Jones, M.N. 1975. "Biological Interfaces", Elsevier, Amsterdam 144. Kates, M. 1972. "Techniques of Lipidology. Isolation, analysis and identification of lipids". From: Laboratory Techniques in Biochemistry and Molecular Biology, Vol. 3 (Work, T.S. and Work, E , eds), North-Holland, Amsterdam, pp. 267-610 145. Katsikas, H. and Quinn, P.J. 1981. F.EB.S. Lett. 133: 230-234 146. Katsikas, H. and Quinn, P.J. 1982. Biochim. Biophys. Acta 689: 363-369 147. Kenrick, J.R., and Bishop, D.G. 1986. Plant Physiol. 81: 946-949 148. Khan, M.U. and Williams, J.P. 1977. J. Chromatogr. 140: 179-185 149. Kingsley, P.B. and Feigenson, G.W. 1981. Biochim. Biophys. Acta 635: 602-618 150. Koole, P., Dammers, A.J., van Ginkel, G., and Levine, Y.K. 1984. Biochim. Biophys. Acta 777: 297-305 151. Kxeutz, W. 1970. Adv. Bot Res. 3: 53-169 152. Krol, M., Huner, N.P.A., Williams, J.P., and Maissan, E 1988. Photosyn. Res. 15: 115-132 178 153. Krapa, Z. 1984. Photosyn. Res. 5: 177-184 154. Krupa, Z. ( Huner, N.P.A., Williams, J.P., Maissan, E., and James, D.R. 1987. Plant Physiol. 84: 19-24 155. Kubo, K. 1985. J. Theor. Biol. 115: 551-569 156. Kyle, D.J., Kuang, T.-Y., Watson, J.L, and Arntzen, CJ. 1984. Biochim. Biophys. Acta 765: 89-96 157. Lakhdar-Ghazal, F. and Tocanne, J.F. 1981. Biochim. Biophys. Acta 644: 284-294 158. Lakhdar-Ghazal, F. and Tocanne, J.F. 1988. Biochim. Biophys. Acta 943: 19-27 159. Lem, N.W., Khan, M., Watson, G.R, and Williams, J.P. 1980. J. Exp. Bot 31: 289-298 160. Lemoine. Y., Dubacq, J.-P., and Zabulon, G. 1982. Physiol. Veg. 20: 487-503 161. Li, T., Lynch, D.V., and Steponkus, P.L 1987. Cryo. Lett. 8: 314-321 162. Lindblom, G., Brentel I., Sjolund, M., Wikander, G., and Wieslander A 1986. Biochem. 25: 7502-7510 163. Lis, L.J., and Quinn, P.J. 1986. Biochim. Biophys. Acta 862: 81-86 164. Iivne, A. and Racker, E. 1969. J. Biol. Chem. 244: 1332-1338 165. Loosely-Millman, M.R, Rand, R.P., and Parsegian, V.A 1982. Biophys. J. 40: 221-232 166. Lopez, J.R. and Tien, H.T. 1984. Photobiochem. Photobiophys. 7: 25-39 167. Low, P.S., Ort, D.R, Cramer, W.A, Whitmarsh, J., and Martin, B. 1984. Arch. Biochem. Biophys. 231: 336-344 168. Lucy, J.A and Ahkong, Q.F. 1986. F.RB.S. Lett. 199: 1-11 169. Lynch, D.V. and Steponkus, P.L 1987. Plant Physiol. 83: 761-767 170. Lynch, D.V. and Thompson Jr., G.R 1982. Plant Physiol. 69: 1369-1365 171. Lyons, J.M. 1973. Ann. Rev. Plant Physiol. 24: 445-466 172. Madden, T.D. 1986. Chem. Phys. Lipids 40: 207-222 173. Madden, T.D., Hope, M.J., and Cullis, P.R. 1984. Biochem. 23: 1413-1418 174. Maggio, B., Cumar, R A , and Caputto, R. 1978. Biochem. J. 171: 559-565 175. Mannock, D.A., Brain, AP.R., and Williams, W.P. 1985. Biochim. Biophys. Acta 817: 289-298 179 176. Mannsourian, A.R. and Quinn, RJ. 1986. Biochim. Biophys. Acta 855: 169-178 177. Marra, J. 1985. J. Colloid Interface Sci. 107: 446-458 178. Marra, J. 1986. J. Colloid Interface Sci. 109: 11-20 179. Mathis, P. 1986. Photosyn. Res. 8: 97-111 180. Mayer, LD. , Hope, M.J., Cullis, P.R, and Janoff, AS. 1985. Biochim. Biophys. Acta 817: 193-196 181. Mayer, LD. , Hope, M.J., and Cullis, P.R. 1986. Biochim. Biophys. Acta 858: 161-168 182. McCarty, R.E and Nalin C M . 1986. In: Encyclopedia of Plant Physiology, Vol. 19, pp. 576-583 183. McCourt, P., Browse, J., Watson, J., Arntzen, CJ, and Somerville, C.R. 1985. Plant Physiol. 78: 853-858 184. McCourt, P., Kunst, L , Browse, J., and Somerville, C.R 1987. Plant Physiol. 84: 353-360 185. McDaniel, RV. 1988. Biochim Biophys. Acta 940: 158-164 186. McDonnel, A. and Staehelin, LA. 1980. J. Cell Biol. 84: 40-56 187. McElhaney, RN. 1984. Biochim Biophys. Acta 779: 1-42 188. McElhaney, R.N., De Gier. J., and Van der Neut-Kok, E G M . 1973. Biochim. Biophys. Acta 298: 500-512 189. Mcintosh, TJ. and Simon, S.A. 1986. Biochem 25: 4058-4066 190. Mcintosh, T.J., Magid, AD. , and Simon, S.A 1987. Biochem 26: 7325-7332 191. Miller, K.R and Lyon, M.K.. 1985. Trends Biochem Sci. 10: 219-222 192. Miller, M., Pedersen, J.Z., and Cox, RP. 1988. Biochim. Biophys. Acta 943: 501-510 193. Mimms, LT. , Zampighi, G., Nozaki, Y., Tanford, C , and Reynolds, J.A. 1981. Biochem. 20: 833-840 194. Mischel, M., Seelig, J., Braganza, LF. , and Buldt, G. 1987. Chem. Phys. Lipids 43: 237-246 195. Morschel, E , and Staehelin, L A . 1983. J. Cell Biol. 97 : 301-310 196. Mullet, J.E and Arntzen, CJ . 1980. Biochim. Biophys. Acta 589: 100-117 197. Mullet, J.E, Burke, J.J., and Arntzen, CJ. 1980. Plant Physiol. 65: 814-822 180 198. Mullet, J.E. Pick, U., and Arntzen, C.J. 1981. Biochim. Biophys. Acta 642: 149-157 199. Murata, N. 1983. Plant & Cell Physiol. 24: 81-86 200. Murata, N., and Hoshi, H. 1984. Plant & Cell Physiol. 25: 1241-1245 201. Murata, N. and Sato, N. 1983. Plant & Cell Physiol. 24: 133-138 202. Murata, N., and Yamaya, J. 1984. Plant Physiol. 74: 1016-1024 203. Murata, N., Sato, N., Takahashi, N., and Hamazake T. 1982. Plant & Cell Physiol. 23: 1071-1079 204. Murphy, D.J. 1982. F.EB.S. Lett 150: 19-26 205. Murphy, D.J. 1986a. In: Encylcopedia of Plant Physiology, Vol. 19 pp. 713-725 206. Murphy, D.J. 1986b. Biochim. Biophys. Acta 864: 33-94 207. Murphy. D.J. and Woodrow. I.E 1983. Biochim. Biophys. Acta 725: 104-112 208. Murphy, D.J, Crowther, D., and Woodrow, I.E 1984. F.EB.S. Lett 165: 515-155 209. Nakatani, H.Y., Barber, J., and Minski, M.J. 1979. Biochim. Biophys. Acta 545: 24-35 210. Navarro, J., Toivio-Kinnucan, M., and Racker, E. 1984. Biochem. 23: 130-135 211. Nicolson, G .L 1971. J. Cell Biol. 50: 258-263 212. Nir, I. and Pease, D.C. 1973. J. Ultrastructure Res. 42: 534-550 213. Nir, S., Bentz, J., and Dugunes, N. 1981. J. Colloid Interface Sci. 84: 266-269 214. Nishihara, M., Yokota, K., and Kito, M. 1980. Biochim. Biophys. Acta 617: 12-19 215. Nolan, W.G. and Bishop, D.G. 1978. Arch. Biochem Biophys. 190: 473-482. 216. Noordam, P.C., Killian, A., Oude Elferink, R.F.M., and De Gier, J. 1982. Chem. Phys. Lipids 31: 191-204 217. O'Sullivan, J.N., Warwick, N.W.M., and Dalling, M.J. 1987. J. Plant Physiol. 131: 393-404 218. Ohki, S. 1988. In: Molecular Mechanisms of Membrane Fuson, (Ohki, S., Doyle, D., Flanagan, T.D., Hui, S.W., and Mayhew, E , eds.) Plenum Press, New York, pp. 123-138 219. Op den Kamp, J.A.F. 1979. Annu. Rev. Biochem 48 : 47-71 220. Oquist G. 1982. Plant Physiol. 69: 869-875 221. Orlich. G. and Hauska, G. 1980. Eur. J. Biochem. Il l: 525-533 181 222. Ort, D.R. 1986. In: Encyclopedia of Plant Physiology, Vol. 19, pp. 143-196 223. Papahadjopoulos, D. and Watkins, J.C. 1967. Biochim. Biophys. Acta 135: 639-652 224. Papahadjopoulos, D., Meers, P.R., Hong, K... Ernst, J.D., Goldstein, I.M., and Duzheunes, N. 1988. In: Molecular Mechanisms of Membrane Fuson, (Ohki, S., Doyle, D., Flanagan, T.D., Hui, S.W., and Mayhew, E., eds.) Plenum Press, N.Y., pp. 1-16 225. Perkins, W.R., Minchey, S.R., Ostro, M.J, Taraschi, T.F., and JanofT, AS. 1988. Biochim. Biophys. Acta 943: 103-107 226. Pick, U., Gounaris, K., Admon, A., and Barber, J. 1984. Biochim Biophys. Acta 765: 12-20 227. Pick, U., Gounaris, K., Weiss, M. and Barber, J. 1985. Biochim. Biophys. Acta 808: 415-420 228. Pick, U., Weiss, M., Gounaris, K., and Barber, J. 1987. Biochim. Biophys. Acta 891: 28-39 229. Pike, C.S. 1982. Plant Physiol. 70: 1764-1766 230. Pohl, P., Glast, H., and Wagner, H. 1970. J. Chromatogr. 49: 488-492 231. Quinn, P.J. and Williams, W.P. 1983. Biochim Biophys. Acta 737: 223-266 232. Radunz, A 1980. Z. Natforsch. C. 35: 1024-1031 233. Raison, J.K. 1973. J. Bioenerg. 4: 285-309 234. Raison, J.K. 1980. In: "The Biochemistry of Plants: A Comprehensive Treatise:, Vol. 4., (Stumpf, P.K., ed.), pp. 7-83. 235. Raison, J.K., and On, G.R 1986a. Plant Physiol. 80: 638-645 236. Raison, J.K., and Orr, G.R 1986b. Plant Physiol. 81: 807-811 237. Raison, J.K., and Wright, L.C. 1983. Biochim Biophys. Acta 731: 69-78 238. Rand, RP. 1981. Annu. Rev. Biophys. Bioeng. 10: 277-314 239. Rand, RP. and Parsegian, V.A. 1988. In: Molecular Mechanisms of Membrane Fuson, (Ohki, S., Doyle, D , Flanagan, T.D., Hui, S.W., and Mayhew, E., eds.) Plenum Press, New York, pp. 73-81 240. Rand, R.P., Fuller, N., Parsegian, V.A, and Rau, D C . 1988. Biochem 27: 7711-7722 241. Rawyler, A and Siegenthaler, P.A. 1980. Eur. J. Biochem. 110: 179-187 242. Rawyler, A. and Siegenthaler, P.A. 1981a. Biochim. Biophys. Acta 635: 348-358 182 243. Rawyler, A. and Siegenthaler, P.A. 1981b. Biochim. Biophys. Acta 638: 30-39 244. Rawyler, A. and Siegenthaler, P.A. 1985. Biochim. Biophys. Acta 815: 287-298 245. Rawyler, A., Henry, L.E.A., and Siegenthaler, P.A. 1980. Carlsberg Res. Commun. 45: 443-451 246. Rawyler, A., Unitt, M.D., Giroud, C , Davies, H , Mayor, J.-P., Harwood, J.L, and Siegenthaler, P.A. 1987. Photosyn. Res. 11: 3-13 247. Record, M.T., Anderson, C.F., and Lohman, T.M. 1978. Quar. Rev. Biophys. 11: 103-178 248. Remy, R., Tremolieres, A., Duval, J.C., Ambard-Bretteville, F., and Dubacq, J.P. 1982. F.EB.S. Lett 137: 271-275 249. Remy, R., Tremolieres, A., and Ambard-Bretteville, F. 1984. Photobiochem. Photobiophys. 7: 267-276 250. Restall, C.J., Williams, W.P., Percival, M.P., Quinn, P.J. and Chapman, D. 1979. Biochim. Biophys. Acta 555: 119-130 251. Reynolds, G.T., Milch, J.R., and Gruner, S.M. 1978. Rev. Sci. Instrum. 49: 1241-1249 252. Rivas, E and Luzzati, V. 1969. J. Mol. Biol. 41: 261-275. 253. Romans, A.Y., Allen, T.M., Meckes, W., Chiovetti, Jr., R., Sheng, L , Kercret H.. and Segrest J.P." 1981. Biochim. Biophys. Acta 642: 135-148 254. Romijn, J.C., van Golde, LM.G. , McElhaney, R.N., and Van Deenen, L L M . 1972. Biochim. Biophys. Acta 280: 22-32 255. Rottenberg, H., Grunwald, T., and Avron, M. 1972. Eur. J. Biochem. 25: 54-63 256. Rouser G., Kritchevsky, G., and Yamamoto, A. 1976. In: "Lipid Chromatographic Analysis", Vol. 3 (Marinetti, G.V., ed.), Marcel Dekker Inc., New York, 2nd ed., pp 713-776 257. Roughan, P.G. 1985. Plant Physiol. 77: 740-746 258. Roughan, P.G. and Batt, R.D. 1968. Anal. Biochem. 22: 74-88 259. Ryrie, I.J., Anderson, J.M., and Goodchild, DJ. 1980. Eur. J. Biochem. 107: 345-354 260. Sakai. W.S., Yamamoto, H.Y., Miyazaki, T., and Ross, J.W. 1983. F.EB.S. Lett 158: 203-207 261. Sastry, P.S. and Kates, M. 1964. Biochem. 3: 1280-1287 262. Sato. N.. and Murata, N. 1981. Plant & Cell Physiol. 22: 1043-1050 263. Sato, N., Murata, N., Miura, Y., and Veta, N. 1979. Biochim. Biophys. Acta 572: 19-28 183 264. Scarpa, A. and De Gier, J. 1971. Biochim. Biophys. Acta 241: 789-797 265. Schonfeld, M. and Kopeliovitch, B.S. 1985. F.EB.S. Lett 193: 79-82 266. Schonfeld, M. and Schickler. H. 1984. F.EB.S. Lett. 167: 231-234 267. Schraml, J., and Bellama, J.M. 1988. "Two-dimensional NMR spectroscopy". From: "Chemical Analysis, A Series of Monographs" Vol. 97 (Winefordner, J.D., ed), John Wiley and Sons, New York. 268. Schroppel-Meier, G. and Kaiser, W.M. 1988. Plant Physiol. 87: 822-287 269. Sculley, M.J., Duniec, J.T., Thome. S.W., Chow, W.S., and Boardman, N.K. 1980. Arch. Biochem. Biophys. 201: 339-346 270. Selstam, E , and Sandelius, A.S. 1984. Plant Physiol. 76: 1036-1040 271. Semenova, G.A, Severina, EP., and Ladygin, V.G. 1987. Fiziol. Rast 34: 74-79 272. Sen, A and Hui, S.-W. 1988. Chem. Phys. Lipids 49: 179-184 273. Sen, A , Williams, W.P., Brain, AP.R., Dickens, M.J, and Quinn, P.J. 1981a. Nature 293: 488-490 274. Sen, A , Williams, W.P., and Quinn, P.J. 1981b. Biochim Biophys. Acta 663: 380-389 275. Sen, A , Brain, A.P.R., Quinn, P.J., and Williams, W.P. 1982a. Biochim. Biophys. Acta 686: 215-224 276. Sen, A , Williams, W.P., Brain, A.P.R., and Quinn, P.J. 1982b. Biochim Biophys. Acta 685: 297-306. 277. Sen, A , Mannock, D.A, Collins, D.J., Quinn, P.J., and Williams, W.P. 1983. Proc. Roy. Soc. Lond. B 218: 349-364 278. Senser, M. and Beck, E 1984. J. Plant Physiol. 117: 41-55 279. Shaw, A.B., Anderson, M.M., and McCarty, R.E 1976. Plant Physiol. 57: 724-729 280. Shinitzky, M. and Henkart, P. 1979. Int Rev. Cytol. 60: 121-147 281. Shipley, G.G., Green, J.P., and Nichols, B.W. 1973. Biochim Biophys. Acta 311: 531-544 282. Siefermann-Harms, D., Ross, J.W., Kaneshiro, K.H., and Yamamoto, H.Y. 1982. F.EB.S. Lett 149: 191-196 283. Siefermann-Harms, D , Ninnermann, H., and Yamamoto, H.Y. 1987. Biochim. Biophys. Acta 892: 303-313 284. Siegenthaler, P. and Rawyler. A. 1986. In: Encyclopedia of Plant Physiology, Vol. 19, pp. 693-705 184 285. Sigrist, M., Zwillenberg, C , Giroud, C , Eichenberger, W., and Boschetti. A. 1988. Plant Sci. Lett. 58: 15-23 286. Silvius, J.R. and McElhaney, R.N. 1980. Proc. Natl. Acad. Sci. 77: 1255-1259 287. Silvius, J.R. and McElhaney, R.N. 1981. J. Theor. Biol. 88: 135-152 288. Silvius, J.R., Read, B.D., and McElhaney, R.N. 1978. Sci. 199: 902-904 289. Singer, M. 1981. Chem. Phys. Lipids 28: 253-267 290. Sprague, S.G. and Staehelin, L A . 1984a. Biochim. Biophys. Acta 777: 306-322 291. Sprague, S.G. and Staehelin, L A 1984b. Plant Physiol. 75: 502-504 292. Sprague, S.G., Camm, E L , Green, B.R., and Staehelin, L A . 1985. J. Cell Biol. 100: 552-557 293. Staehelin, LA. 1986. In: Encyclopedia of Plant Physiology, Vol. 19, pp. 1-84 294. Stewart, J.C.M. 1980. Anal. Biochem. 104: 10-14 295. Stidham, M.A, Mcintosh, T.J., and Siedow, J.N. 1984. Biochim Biophys. Acta 767: 423-431 296. Stout, D.G., and Lacroix, L.J. 1981. Z. Pflanzenphysiol. 101: 355-360 297. Sundby, C. and Larsson, C. 1985. Biochim. Biophys. Acta 813: 61-67 298. Szalontai, B., Droppa, M., Vigh, L , Joo, F., and Horvath, G. 1986. Photobiochem. Photobiophys. 10: 233-240 299. Tang, X.-S. and Satoh, K. 1985. F.EB.S. Lett. 179: 60-64 300. Thomas, P.G., Dominy, P.J., Vigh, L , Mansourian, A.R., Quinn, P.J., and Williams, W.P. 1986a. Biochim Biophys. Acta 849: 313-140 301. Thomas, P.G., Quinn, P.J., and Williams, W.P. 1986b. Planta 167: 133-139 302. Thornber, J.P. 1986. In: Encyclopedia of Plant Physiology, Vol. 19, pp. 98-142 303. Thome, S.W. and Duniec, J.T. 1983. Quar. Rev. Biophys. 16: 197-278 304. Tilcock, C.P.S. 1986. Chem. Phys. Lipids 40: 109-125 305. Tilcock, C.P.S., Bally, M.B., Farren, S.B.. Cullis, P.R, and Gruner, S.M. 1984. Biochem 23: 2696-2703 306. Tomoaia-Cotisel, M., Zsako, J., Chifu, E , and Quinn, P.J. 1983. Chem. Phys. Lipids 34: 55-64 185 307. Torti, R, Gerola, P.D., and Jennings, R.C. 1984. Biochim. Biophys. Acta 767: 321-325 308. Toyoshima, Y. and Thompson, T.E. 1975. Biochem. 14: 1525-1531 309. Trauble, J. 1971. J. Memb. Biol. 4: 193-209 310. Tremolieres, A., Dubacq, J.-P., Ambard-Bretteville, F., and Remy, R. 1981. F.EB.S. Lett 130: 27-31 311. Tuquet C , Guillot-Salomon, T., De Lubac, M., and Signol, M. 1977. Plant Sci. Lett 8: 59-64 312. Tuquet C , Guillot-Salomon, T., and Farineau, N. 1986. Physiol. Plant 67: 620-625 313. Unitt, M.D. and Harwood, J.L 1985. Biochem. J. 228: 707-711 314. Van de Ven, M.J.M. and Levine. Y.K. 1984. Biochim. Biophys. Acta 777: 283-296 315. Van der Steen, A.T.M., De Kruijff, B., and De Gier, J. 1982. Biochim. Biophys. Acta 691: 13-23 316. Van Hoogevest P., Du Maine, A.P.M., and De Kruijff, B. 1983. F.EB.S. Lett 157: 41-44 317. Verrall, R.E 1973. in Water, A Comprehensive Treatise (Franks, F., ed.), Vol. 3, pp. 211-264, Plenum Press, New York 318. Vigh, L , Horvath, I., van Hasselt P.R., Kuiper, P.J.C. 1985a. Plant Physiol. 79: 756-759 319. Vigh, L , Joo, F., Droppa, M., Horvath, L , and Horvath, G. 1985b. Eur. J. Biochem. 147: 477-481 320. Vorbeck, M.L and Marinetti, G.V. 1965. J. Lipid Res. 6: 3-6 321. Waggoner, C , Aronson, H., More, J., and Berg, S.P. 1985. Photosyn. Res. 6: 57-71 322. Watanabe, T., Fukushima, H., Kasai, R., and Nozawa, Y. 1981. Biochim. Biophys. Acta 665: 666-73 323. Watterson, J.G. 1987a. Phys. Chem. Liq. 16: 313-316 324. Watterson, J.G. 1987b. Phys. Chem. Liq. 16: 317-320 325. Watterson, J.G. 1987c. Biochem. J. 248: 615-617 326. Watterson, J.G. 1988. Mol. Cell Biochem. 79: 101-105 327. Webb, M.S. and Green, B.R. 1987. In: "Progress in Photosynthesis Research", (Biggins J., ed.), Nijhoff, Dordrecht pp. 197-200 328. Webb, M.S., Tilcock, C.P.S., and Green, B.R. 1988. Biochim. Biophys. Acta 938: 323-333 186 329. Whitmarsh, J. 1986. In: Encyclopedia of Plant Physiology, Vol. 19, pp. 508-527 330. Wieslander, A., Ulmius, J., Lindblom, G., and Fontell, K. 1978. Biochim. Biophys. Acta 512: 241-253 331. Williams, J.P. and Merrilees, P.A. 1970. Lipids 5: 367-370 332. Wilschut, J. and Hoekstra, D. 1984. Trends Biochem. Sci. November, 1984: 479-483 333. Wilschut, J. and Hoekstra, D. 1986. Chem. Phys. Lipids 40: 145-166 334. Wintermans, J.F.G.M., van Besouw, A and Bogemann, G. 1981. Biochim. Biophys. Acta 663: 99-107 335. Wollman,. F .A 1986. In: Encyclopedia of Plant Physiology, Vol. 19. pp. 487-495 336. Wong, M. and Thompson, T.E. 1982. Biochem. 21: 4133-4139 337. Yamamoto, Y., Ford, R.C., and Barber, J. 1981. Plant Physiol. 67: 1069-1072 338. Yang, S.F., Freer, S., and Benson, A A 1967. J. Biol. Chem. 242: 477-484 339. Yoshida, S., Kawata, T., Uemura, M., and Niki, T. 1986. Plant Physiol. 80: 152-160 


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items