(The Oily Press Lipid Library) Kare Larsson, Peter Quinn, Kiyotaka Sato, Fredrik Tiberg - Lipids - Structure, Physical Properties and Functionality-Woodhead Publishing (2006)

Lipids: Structure, physical
properties and functionality

Also in the Oily Press Lipid Library:
Volume 18. Lipid Oxidation
Written by Edwin N. Frankel
Volume 17. Bioactive Lipids
Edited by Anna Nicolaou and George Kokotos
Volume 16. Advances in Lipid Methodology – Five
Edited by Richard O. Adlof
Volume 15. Lipid Analysis (third edition)
Written by William W. Christie
Volume 14. Confectionery Fats Handbook
Written by Ralph E. Timms
Volume 13. Lipids for Functional Foods and Nutraceuticals
Edited by Frank D. Gunstone
Volume 12. Lipid Glossary 2
Written by Frank D. Gunstone and Bengt G. Herslöf
Volume 11. Lipids in Nutrition and Health: A Reappraisal
Written by Michael I. Gurr
Volume 10. Lipid Oxidation
Written by Edwin N. Frankel
Volume 9. Trans Fatty Acids in Human Nutrition
Edited by Jean Louis Sébédio and William W. Christie
Volume 8. Advances in Lipid Methodology – Four
Edited by William W. Christie
Volume 7. Advances in Lipid Methodology – Three
Edited by William W. Christie
Volumes 1– 6. Out of print
Woodhead Publishing in Food Science, Technology and Nutrition
Lipids: Structure, physical

properties and functionality
KÅRE LARSSON
Camurus Lipid Research Foundation, Lund, Sweden
PETER QUINN
King’s College London, London, UK
KIYOTAKA SATO
Hiroshima University, Japan
FREDRIK TIBERG
Camurus AB, Lund, Sweden
Oxford Cambridge Philadelphia New Delhi
Published in association with

Lipid Technology
Published by Woodhead Publishing Limited,
80 High Street, Sawston, Cambridge CB22 3HJ, UK
www.woodheadpublishing.com
www.woodheadpublishingonline.com
Woodhead Publishing, 1518 Walnut Street, Suite 1100, Philadelphia,

PA 19102-3406, USA
Woodhead Publishing India Private Limited, G-2, Vardaan House, 7/28 Ansari Road,
Daryaganj, New Delhi – 110002, India
www.woodheadpublishingindia.com
First published by The Oily Press, 2006

Reprinted by Woodhead Publishing Limited, 2012
© PJ Barnes & Associates, 2006; © Woodhead Publishing Limited, 2012

The authors have asserted their moral rights
This book contains information obtained from authentic and highly regarded sources.
Reprinted material is quoted with permission, and sources are indicated. Reasonable
efforts have been made to publish reliable data and information, but the authors and the
publisher cannot assume responsibility for the validity of all materials. Neither the authors
nor the publisher, nor anyone else associated with this publication, shall be liable for any
loss, damage or liability directly or indirectly caused or alleged to be caused by this book.
Neither this book nor any part may be reproduced or transmitted in any form or by any
means, electronic or mechanical, including photocopying, microfilming and recording, or
by any information storage or retrieval system, without permission in writing from
Woodhead Publishing Limited.
The consent of Woodhead Publishing Limited does not extend to copying for general
distribution, for promotion, for creating new works, or for resale. Specific permission
must be obtained in writing from Woodhead Publishing Limited for such copying.
Trademark notice: Product or corporate names may be trademarks or registered trade-

marks, and are used only for identification and explanation, without intent to infringe.
British Library Cataloguing in Publication Data

A catalogue record for this book is available from the British Library
ISBN 978-0-9531949-9-5 (print)

ISBN 978-0-85709-791-0 (online)
This book is Volume 19 in The Oily Press Lipid Library
Typeset by Ann Buchan (Typesetters), Middlesex, UK

Printed by Lightning Source
Foreword
Over ten years have passed since the publication of Kåre Larsson’s Lipids –
Molecular Organization, Physical Functions and Technical Applications (LMO)
by The Oily Press, then based in Dundee, Scotland, and run by Dr William W.
Christie. The book was soon recognized as a major contribution to the literature
(“This is a book without comparison in the lipid literature...”, Stig E. Friberg in
the Journal of Dispersion Science and Technology, 1995, Vol.16, p.295 and
“...the content is excellent”, Philip W. Wertz in Chemistry and Physics of
Lipids, 1994, Vol.74, p.99). Dr Christie’s excellent choice of Kåre Larsson as
the author was also confirmed (“His expertise in describing the various states
of lipids is second to none”, Edward G. Perkins in INFORM, 1994, Vol.5,
p.1394 and “...written by an acknowledged world expert in his field”, Fred B.
Padley, Lipid Technology, 1994, Vol.6, p.102). Until the publication of LMO
there had been no single, concentrated source of so much information on the
subject: “The strength of this book – and it is enormous – is the fact that the
author has been able to compile in one volume information otherwise found
only in the most widely different kinds of scientific journals” (Friberg).
When LMO was published, Kåre Larsson was a Professor of Food Technol-
ogy in the Chemical Centre at Lund University, Sweden. He is now cofounder
of Camurus AB and Probi AB, and earlier of Biogram AB (which later became
Bioglan AB), and serves as Chairman of the Board of the Camurus Lipid
Research Foundation in Lund. Camurus is a provider of drug delivery systems
and works closely with pharmaceutical manufacturers. Kåre Larsson is a
Fellow of the Royal Swedish Academy of Science and the Academy of
Engineering Science and has authored more than 200 original papers and five
books covering areas of lipid biophysical chemistry, food science and
nutrition, and biomedicine. He is also the named inventor on several patents, of
which four have led to industrial products. In 2001 he won the Rhodia prize of
the European Colloid and Interface Society for his discovery of cubosomes and
hexosomes and explorative work on their applications.
When I asked Kåre Larsson to write a second edition, the comprehensive
coverage of LMO became an obstacle – the subject area had expanded to such
an extent that one author could not cover it alone. But we did not want to resort
to the usual edited book with each chapter written by a different author. The
solution was to invite three other well-known scientists in this field to act as
coauthors – and it is a tribute to Kåre’s reputation that all three agreed.
Peter Quinn, Professor of Biochemistry in the Department of Life Sciences
at King’s College London, UK, is renowned for his work on biological
v
vi FOREWORD
membranes and their constituents, as evidenced by the publication of ten

books, more than 400 research papers, and several patents.
Kiyotaka Sato, Professor in the Graduate School of Biosphere Sciences,
Hiroshima University, Japan, is widely recognized as a leading expert in the
physical chemistry and biophysics of fats and lipids and in the crystallization of
biological soft materials. Among other awards, he was presented with the 2005
AOCS Stephen S. Chang Award for distinguished accomplishments in basic
research that have been used by industry for the improvement or development
of products related to lipids in the area of lipid crystallization and
crystallography.
Fredrik Tiberg is President, Chief Executive Officer and Head of Research
& Development at Camurus AB, and also Adjunct Professor of Surface and
Colloid Chemistry at Lund University. He has published more than 80 original
scientific papers, coauthored several books, and been named as inventor on a
number of patents.
With some major changes in the areas covered, and with four authors instead
of one, we were obliged to define the resulting publication as a new book with
a new ISBN number, rather than as a second edition retaining the old number.
Therefore we took the opportunity of also giving it a new and more appropriate
title: Lipids: Structure, Physical Properties and Functionality. With the know-
ledge and experience of the above-named four authors behind it, I am confident
that this new book will build on the reputation of LMO and be a valued source
of information for many years to come.
Peter J. Barnes
Publisher, The Oily Press
Bridgwater, UK
January 2006
Preface
The ambition behind the new edition of this book is to provide an up-to-date
description of the diversity of lipid molecular arrangements in different
physical states, as a basis for the understanding of lipid functionality in
biological and technical systems. The first edition was published in 1994 with
Kåre Larsson as author, and when he was asked by the publisher to revisit the
text he realized that he could not cover this broad field alone. Three colleagues
joined forces with him, and the present edition has therefore in many aspects
been extended. In some cases the description is deeper with a more narrow
focus. For example, the chapter on the solid state in the earlier edition covered
all lipids, whereas in this new edition there is a very complete demonstration of
the crystal structures and crystallization properties of fatty acids and fats. These
general principles, however, can be applied to all lipids.
In biology, as well as in technical applications such as foods, we are dealing
with soft matter. Lipids form aqueous phases alone or in conjunction with
proteins and polysaccharides. The combination of short-range disorder and
long-range order into liquid-crystalline structures plays a crucial role. A
driving force is the dualistic properties of the molecules in relation to water.
Molecular regions avoiding water contact, in combination with regions
striving towards such contact and interaction, lead to self-assembly, and even
in the liquid state to the formation of organized structures on the colloidal level.
This new edition presents many new results, particularly on the structure and
functions of dispersions of liquid-crystalline phases forming nanostructures
and mesoporous systems.
With regard to the role of lipids in cellular and molecular biology, this book
focuses on biophysical aspects, and discussion of lipid biochemistry is limited
to a chapter on cell membranes.
Kåre Larsson, Lund, Sweden
Peter Quinn, London, UK
Kiyotaka Sato, Hiroshima, Japan
Fredrik Tiberg, Lund, Sweden
January 2006
vii
Contents
Foreword v
Preface vii
1 Basic concepts 1
A. Classification of lipids 1
1. Non-polar lipids
2. Polar lipids
B. Structure, physical properties and functional properties 5
C. Methods for structure characterization 6
References 8
2 Solid-state behaviour of polymorphic fats and fatty acids 9

A. Polymorphism: structures and transformations 11
1. Fatty acids
2. Diacylglycerols
3. Triacylglycerols
4. Molecular interactions and polymorphic structures
B. Phase behaviour of binary mixtures of triacylglycerols 41
1. Binary mixtures of saturated-acid triacylglycerols
2. Binary mixtures of saturated–unsaturated mixed-acid
triacylglycerols
C. Polymorphic crystallization of fats in complex fluid systems 47
1. Polymorphic crystallization from bulk liquid
2. Effects of external factors on polymorphic crystallization
3. Fat crystallization in emulsion droplets
4. Crystallization in a gel phase
D. Conclusions 63
References 63
3 Liquid-crystalline lipid–water phases 73

A. Structures 73
1. Introduction: Lamellar structures and the role of curvature
2. Hexagonal phases
3. Cubic phases
ix
x CONTENTS
4. Membrane protein crystallization in cubic phases

5. Identification of liquid-crystalline lipid structures
B. Phase transitions and phase diagrams 83
1. Introduction
2. Why aqueous lipid phases are formed
3. Relation between molecular geometry and structure of lipid–water
phases
4. The transition between cubic bicontinuous phases proves their
minimal surface structure – The Bonnet relation
5. Other lipid–water phases
C. Phase behaviour of different lipid–water systems 89
1. Monoacylglycerols
2. Phospholipids
3. Sphingolipids
4. Glyceroglucolipids/galactolipids
5. Lipid–glycerol systems
6. Simplified model diagrams of lipid–water systems
D. Main features of phase diagrams involving different types of lipids 99
1. Introduction
2. Ternary and quaternary phase diagrams – some general
features
E. Cholesterol esters 103
References 104
4 The liquid state 107

A. Liquid triacylglycerols 107
B. Simple non-polar lipids in the liquid state 108
C. Polar lipids in the liquid state and the aqueous L2 phase 109
D. Lipid microemulsions 111
E. The L3 phase 111
References 112
5 Lipids at the air–water interface – monolayers and multilayers

in surface films, bubbles and foams 113
A. Monomolecular lipid films at the air–water interface 113
1. Monolayers of polar lipids
2. Domain structure within monolayers
3. Triacylglycerol monolayers and multilayers
4. Competitive spreading of monolayers from fats/oils and proteins
5. Other non-polar lipids
6. Mixed monolayers
7. Lipid monolayers on water and on solids
CONTENTS xi
8. A surface phase formation at an air–water interface of a lung

surfactant extract
B. Bubbles and foams 123
1. Foam structure
2. Foam stability
3. Effects of electrolytes and the hydrophobic force
4. Micellar lifetime and dynamics of bubbles and foams
5. Destabilization of foams
6. Foams stabilized by an interfacial phase
7. Foams based on proteins and emulsified oils/fats
References 127
6 Dispersions of lipid–water phases 129

A. Liposomes, vesicles and the Lα phase 129
1. Formation mechanisms and stability
2. Structure and functionality
B. Non-lamellar colloids of reversed cubic, hexagonal, and ‘sponge’ phase 134
C. Lipid–water dispersions in biological systems 141
1. Lamellar bodies
2. Cubic cell membrane assemblies
References 143
7 Interaction of lipids with proteins and polypeptides 145

A. Introduction 145
B. Surface activity of proteins 146
C. Interaction of soluble proteins with lipid monolayers 148
1. Mechanism of protein penetration into lipid monolayers
2. Protein penetration dependence on initial film pressure
3. Ionic interactions and protein adsorption
4. Effect of subphase pH on protein interaction with lipids
5. Hydrocarbon chain dependence of protein penetration
D. Interaction of lipolytic enzymes with their substrates 159
E. Interaction of proteins and lipids in membranes 162
1. Intrinsic and extrinsic proteins
2. Lipid-anchored membrane proteins
F. Lipoproteins 167
1. Albumin
2. Serum lipoproteins
G. Cubic lipid–water phases 171
References 173
xii CONTENTS
8 Emulsions 175
A. Oil-in-water emulsions 175
1. Emulsification
2. Protein-stabilized emulsions
3. Lipid-stabilized emulsions
B. Water-in oil emulsions 180
C. Demulsification 181
References 181
9 Lipids of biological membranes 183

A. Introduction 183
B. Membrane lipid composition 184
1. Analysis of membrane lipid composition
2. Lipid composition of cell membranes
3. Lipid domains in membranes
C. Membrane function and lipid diversity 194
1. Adaptation to environmental factors
2. Remodelling membrane lipids
3. Control of membrane lipid unsaturation
D. Biochemical mechanisms of lipid homeostasis 201
1. Phospholipases A2 (PLA2)
2. Acyltransferases
3. Acylation/deacylation cycle
4. Phospholipases C
5. Phospholipases D
6. Sphingomyelinases
E. Conclusions 213
References 213
10 Lipid barriers at the environment–body interface 219

A. Introduction 219
B. Skin: The stratum corneum 220
1. Lipid composition of the stratum corneum
2. Structure and organization of the lipids
C. Lipid barriers of the alimentary tract 223
1. Saliva
2. The gastric mucous surface
3. Surface barriers of the lower alimentary tract
4. Intestinal absorption of lipids
D. The peritoneal mesothelium 227
E. Lung surfactant 228
CONTENTS xiii
1. Composition of pulmonary surfactant

2. Formation of pulmonary surfactant
3. Structure and properties of pulmonary surfactant
4. Treatment of pulmonary surfactant deficiencies
References 234
11 Drug delivery 239

A. Colloidal self-assembly structures 240
1. Micelles, microemulsions, and emulsions
2. Lipid crystal dispersions
3. Liposomes
4. Nanoparticles of non-lamellar phases: cubic, hexagonal, and
‘sponge’ phases
B. Liquid crystals 247
C. Summary outlook 248
References 249
12 Foods 251
A. Starch–lipid interaction 251
B. Milk-based lipid products and their analogues 254
1. Milk
2. Whipping of cream, butter and ice cream
3. Drying of milk
4. Hydrogenation of oils – hardened fats
5. Artificial milk/cream, margarine and low-fat spread products
C. Cocoa butter and chocolate 259
D. Vegetable oils, frying oils and emulsion products 260
1. Frying oils
2. Mayonnaise and dressing products
E. Lipids in bread and other baked products 261
1. Interaction of wheat flour with water
2. The baking process
3. Cakes and biscuits/cookies
References 264
Recommended further reading 264
Index 265
CHAPTER 1
Basic concepts
A. Classification of lipids
There is no strict definition of lipids. Extracts in organic solvents from
biological tissues have been termed lipids for a long time. A definition more
consistent with modern views, proposed by Christie (1987), says: “Lipids are
fatty acids and their derivatives, and substances related biosynthetically or
functionally to these compounds”. From a physical point of view it is natural to
divide lipids into two groups:
• polar lipids, which interact with water and form aqueous phases;
• non-polar lipids, which do not form aqueous phases.
These are the definitions we will use in this book. Interfaces, however, are a
special case, where even non-polar lipids can interact with water and form
monomolecular layers.
Simple fatty acids exhibit most of the structural states described in this book.
Oleic acid, for example, behaves like oil in relation to water at room tempera-
ture. If titrated into soap, however, a wide variety of aqueous phases are
obtained. This means that by lowering the pH, dissociated oleic acid – a polar
lipid – becomes undissociated and a non-polar lipid. In this respect fatty acids
represent a special case.
The structures of fatty acids in different states of order represent well the
structures in the solid state, described in Chapter 2, as well as in the disordered
liquid-crystalline phases, described in Chapter 3. The overall organization into
bilayers in the solid state is driven by the carboxyl groups, which associate the
molecules by hydrogen bonds into dimers or polymers. The hydrocarbon
chains are extended to different degrees. In solids, planar zigzag conformations
of the carbon–carbon bonds exist (all-trans), whereas in disordered states,
occurring in liquid crystals and in melts, gauche conformations form dynami-
cally along the chains. Recording of the trans/gauche ratio, for example by
Raman spectroscopy, has demonstrated the conformational changes that occur
during chain-melting transitions.
The raw materials used in lipid technology are fats and oils, with plant
extracts dominating. About 120 million tonnes of vegetable oils were produced
in 2001, the main types being soybean oil (29 million tonnes), palm oil (23
million tonnes) and rapeseed oil (14 million tonnes). The most important polar
lipids, phospholipids, are obtained as by-products during the industrial refining
of vegetable oils into pure triacylglycerols.
1
2 LIPIDS: STRUCTURE, PHYSICAL PROPERTIES AND FUNCTIONALITY
1. Non-polar lipids
General information on the chemistry of lipid molecules can be found in The

Lipid Handbook (Gunstone et al., 1994). Fatty acids can be saturated or
unsaturated, and the predominant configuration of the double bonds is cis. One
industrial approach to modifying the melting points of fats is hydrogenation,
whereby unsaturated chains become saturated, and in parallel some of the
double bonds may also be converted to the trans isomer.
Oils and fats (in what we call fats, at least a part is solid at room temperature)
are triacylglycerols, formerly known as triglycerides. A triacylglycerol
molecule is made up of three fatty acids (acyl groups) attached via ester
linkages to a 3-carbon glycerol ‘backbone’ structure. Because both the types
and the positions of the fatty acid components can vary widely, there are an
enormous number of possible molecular combinations. Natural fats and oils are
complex mixtures of triacylglycerols; butterfat, for example, contains on the
order of 100 000 different molecular species.
Each species is defined by the fatty acids in the 1-, 2- and 3-positions of the
glycerol backbone. For example, soybean oil is dominated by two species:
LiLiLi, in which linoleic acid (Li) occupies all three positions on the glycerol
backbone; and LiLiO, in which oleic acid (O) occupies the 3-position. It should
also be mentioned that some ether analogues of triacylglycerols occur in
nature. Marine lipids, for example, can contain an ether-linked alkyl group in
the 1-position and ester-linked acyl groups in the 2- and 3-positions of the
glycerol backbone. From a physical point of view they still behave like
triacylglycerols.
Diacylglycerols, formerly known as diglycerides, contain two fatty acid
groups, which can be located at different positions on the glycerol backbone.
The same diacylglycerol molecule can exist as two different isomers, depend-
ing on whether the acyl groups are in the 1,2-position (note that this is
equivalent to the 2,3-position) or the 1,3-position. Acyl migration takes place
easily, and during processing an equilibrium mixture consisting of approxi-
mately equal amounts of the two isomers is usually formed.
Sterols occur in free form or as fatty acid esters. The molecules are charac-
terized by a steroid skeleton, which makes most of the molecules rigid. The
most important member is cholesterol – a lipid of great importance as a
constituent of the cell membranes in our body. There are also sterols in plants,
which are closely related to cholesterol, such as stigmasterol. Cholesterol is
transported in the circulation as cholesterol esters within lipoproteins. In the
liver, cholesterol is transformed into bile acids, which are polar lipids that
function like detergents in the intestinal digestion of fats.
Cholesterol has for a long time been regarded as a crucial lipid for the
functioning of mammalian cell membranes. A recent study of knockout mice
lacking the ability to synthesize cholesterol was therefore most surprising, as
BASIC CONCEPTS 3
Figure 1.1 Chemical formulae of phosphatidylcholine (PC), phosphatidylethanolamine (PE) and

phosphatidylinositol (PI). The counter-ion (usually sodium) is denoted by X+.
they were viable even though cholesterol was replaced by another sterol
(Wechsler et al., 2003). An important mechanism induced by cholesterol in the
cell membranes is phase separation of the bilayer into cholesterol-rich domains
(lipid rafts and caveolae; see also Chapters 3 and 9) and cholesterol-poor
regions. Perhaps this segregation was still achieved in the knockout mice by the
alternative sterol.
2. Polar lipids
Monoacylglycerols or monoglycerides are the simplest type of polar lipids, and
are the dominant functional additives used in industrial food processing. They
are produced on a large scale to high purity by molecular distillation. Of the two
isomers, the 1-isomer dominates in the equilibrium mixtures formed by acyl
migration (about 90% 1-isomer and 10% 2-isomer).
Phospholipids or phosphoglycerolipids associate spontaneously into lipid
bilayers in water, which is the basic mechanism behind the formation of
biological membranes. They have acyl groups ester-bound in the 1- and 2-
positions of the glycerol backbone, and a polar group involving phosphate
in the 3-position. The most important members of this lipid class
are phosphatidylcholine (PC), phosphatidylethanolamine (PE) and
phosphatidylinositol (PI), all shown in Figure 1.1. Other phospholipids that

occur commonly are phosphatidylglycerol (PG), phosphatidylserine (PS) and
phosphatidic acid (PA). The aqueous interactions of phospholipids are closely
related to their ionic character. Whereas PC and PE are zwitterionic (containing
one positive group and/or one negative group, depending upon the pH), the
others (i.e. PI, PG, PS, and PA) are all anionic at physiological pH.
The fatty acid in the 1-position of a phospholipid can be split off by the
phospholipase enzyme PLA1, and the fatty acid in the 2-position is released by
PLA2. In this way lysophospholipids are obtained. These are as polar as soaps
and can therefore solubilize the components of cell membranes, causing lysis
of the cells.
Galactolipids or galactosylglycerolipids originate mainly from the thylakoid
membranes of plants, where photosynthesis takes place. Therefore they are
extremely important and probably the most commonly occurring lipids on
earth. The two major classes of galactolipids are monogalactosyl-diacylglycerols
(MGDG) and digalactosyl-diacylglycerols (DGDG).
Sphingolipids contain a long-chain amine termed sphingosine linked by an
amide bond to a fatty acid. Such simple amides are called ceramides, and occur
for example in the skin. Free sphingosine, which is formed enzymatically, is a
cationic detergent-like molecule. The sphingosine molecule contains hydroxyl
groups along the hydrocarbon chain, and one hydroxyl group next to the amine
group can be linked to a polar group. If this group is phosphatidylcholine, the
molecule is rather similar to PC and is called sphingomyelin. Sphingomyelin
associated with cholesterol forms phase-separated domains in mammalian cell
membranes (the lipid rafts and caveolae mentioned above) with specific
biological functions.
Gangliosides are bioactive ceramides occurring mainly in the brain and in
milk. Milk contains two types, called GM3 and GD3, in which either one
(GM3) or two (GD3) sialic acid residues are linked to a galactose unit that,
together with a glucose unit, is bound to the ceramide molecule. Another type
of sphingolipid containing a sugar group, which also occurs in the brain, is the
cerebrosides. A subgroup is the sulfatides, which have a sulfate group attached
to this sugar.
The physical properties of lipids and their relation to molecular shape and
amphiphilic character are central themes in this book. The dualistic properties
of the molecules in relation to water, with flexible hydrocarbon chains avoiding
water contact and a polar head group that tends to orient towards water defines
the amphiphilicity. From a physical point of view, lipid molecules are therefore
often illustrated simply by a line representing the hydrocarbon chain axis,
attached to a circle representing the polar head group. This schematic
representation of structures, not the detailed chemical architecture, will be
used in structural descriptions of the liquid state and the aqueous phases, which
are highly disordered.
BASIC CONCEPTS 5
In industrial applications of lipid functionality (see also Section B below),

the molecules are usually modified by chemical processes. The most important
group of such functional lipids is food emulsifiers, which have many
applications in addition to emulsification. Distilled monoacylglycerols and
phospholipids dominate these emulsifiers. Phospholipids in pharmaceutical
applications are usually prepared from egg yolk, whereas those used in food
applications are by-products from the refining of vegetable oils. After
extraction from soybeans, rapeseed or other oil crops, the free fatty acids are
first removed as soaps by a sodium hydroxide solution. The next step (called
degumming) is removal of the phospholipids by water due to the formation of
an aqueous liquid-crystalline phase. This crude mixture is usually termed
lecithin. Alternatively the word lecithin is used in the case of purified fractions
(and even pure PC is sometimes termed lecithin). Other food emulsifiers are
diacetyl-tartaric esters of monoacylglycerols (under the name DATEM),
sodium and calcium salts of stearoyl-lactylates (termed SSL and CSL, respec-
tively), and polylactoyl esters of stearic acid. Some food emulsifiers are also
used in the formulation of drugs and cosmetics. Important examples include
ethoxylated lipids, such as polyoxyethylene monoacylglycerols and
polyoxyethylene sorbitan monostearate.
Finally, some other types of important lipids should be mentioned. If we
consider fat-soluble vitamins, the significance of the physical state is illustrated
by the fact that their bioavailability depends on their solubilization in
triacylglycerols and phospholipids in the food. We may in this context apply
the simple physical description mentioned above. On this level of description,
vitamin D resembles cholesterol, with a polar head group and a rigid hydro-
carbon chain, whereas each of the others – vitamin A (retinol), vitamin E
(tocopherol), and vitamins K1 and K2 – can be represented by a polar head group
and a flexible hydrocarbon chain. With regard to the complex physiological
functions of vitamins, such a description illustrates the gap between lipid
biochemistry and the physical aspects we deal with in this book. This is
particularly evident when we consider the effects of small chemical modifi-
cations on the diversity of biological signalling by different eicosanoids
(prostaglandins, thromboxanes and leukotrienes). This lipid family is derived
from arachidonic acid, a 20-carbon fatty acid with four double bonds (C20:4).
The physical description of their disordered structures, which represents their
functional states, is at present very primitive.
B. Structure, physical properties and functional properties

Lipids are frequently used in technical applications in order to obtain a specific
function. One example is the emulsification of oil in water. In order to obtain
an emulsion, we can add a polar lipid to an oil–water mixture, and after a
mechanical agitation process, an emulsion is obtained. The polar lipid used is
a functional additive, and it is therefore necessary to discuss the functionality

of the particular lipid used. Other functional properties involve flow behaviour
or rheology. Thus, an oil becomes a plastic fat when we have added a critical
amount of triacylglycerol crystals. Fat crystals as such are responsible for
various functions in food emulsions, for example the ability to whip an
emulsion into cream and ultimately to form butter. Amylose–lipid complex
formation is an important method of modifying starch products, and this
function can best be achieved by polar lipids with one acyl chain only.
Dispersion of sugar crystals in a fat continuum is another example where a
polar lipid is needed for a specific function.
In Chapters 9–12, we will discuss applications of lipid functionality compre-
hensively and try to understand the underlying mechanisms. In order to do so
we must understand how each functional property is related to the physical
properties of our actual system. Emulsification, for example, is related to the
viscosity of the oil and water phases, to the reduction of interfacial tension by
the polar lipid used as emulsifier, to the mechanical properties of the interfacial
film, and finally to the kinetics involved in processing. The best approach to
obtain a detailed knowledge of physical properties is to start from the molecular
arrangements adopted by the lipid components. In that way, forces acting in
different directions can be derived. The combination of long-range order with
short-range disorder is the most significant physical property of lipids when we
consider their biological as well as technical functions. Therefore this unique
structural feature of lipids is a major theme of this book.
Lipid structure should be considered on two levels. Starting with the
molecular level, lipid molecules have a remarkable ability to associate with
other molecules of the same kind (‘self-assembly’) in different geometrical
arrangements. In this way, colloidal aggregates are formed. A soap micelle is
a typical example of a colloidal aggregate formed by polar lipid molecules. A
particle in the size range 10–1000 nm is usually regarded as a colloidal particle.
It is obvious that a colloidal dispersion exhibits a large interfacial area towards
the continuous phase. This is the reason for the close link between surface and
colloid chemistry. Dispersions of oil in water into an emulsion usually give an
oil/water (o/w) interface area of many tenths of m2/g emulsion. The structure at
the interface plays an important role in determining the stability and other
physical properties of the emulsion. Surface and colloid science are important
tools for relating structure to physical properties and functionality.
C. Methods for structure characterization

The polarizing microscope is a valuable tool for characterizing single-crystal
quality and screening the phase properties of lipid–water systems. The first
structure determinations of liquid-crystalline structures by Friedel (1922) and
Sir William Bragg (1934) were based on optical properties. The textures
BASIC CONCEPTS 7
involving optical axis variations were analysed in smectic phases. It was shown
that the so-called focal–conic texture is due to curved arrangements of equidis-
tant and uniform layers, which are flexible (liquid-like). Lack of birefringence
means that a phase is cubic or a true liquid, whereas the birefringence texture
is quite characteristic of other liquid-crystalline phases described later in this
book.
X-ray diffraction is the most powerful method for structure determination in
the solid state. One problem is to prepare single crystals that are perfect enough
to allow resolution of the reflections from the different crystal planes. If this is
achieved, it is possible to determine the exact position in the unit cell of all
atoms other than hydrogen atoms. The unit cell is the smallest repetition unit
within a crystal.
Even when single-crystal data are not available, it is still possible to derive
the main features of the molecular packing from X-ray powder diffraction
curves, which can always be obtained. The reason is the arrangement of the
lipid molecules into layers, where the distance between hydrocarbon chains
can be derived from one region of the diffraction pattern (‘short-spacings’) and
the thickness of the layers can be obtained from another region (‘long-
spacings’). The solid-state behaviour of lipids also involves transitions between
different crystal forms (polymorphs).
The aqueous phases of lipids are liquid-crystalline or liquid, and small-angle
X-ray scattering/diffraction (SAXS) methods are used to characterize the
structures, particularly of the more ordered liquid crystals. As the repetition
distances in the hydrocarbon chain direction are usually very long, small
diffraction angles must be recorded. The detailed diffraction patterns used in
order to identify the structures of different phases is further discussed in
Chapter 3.
Neutron scattering methods are also applied to studies of lipid structures.
The replacement of hydrogen atoms by deuterium in lipids provides the
possibility of locating hydrogen atoms, a feature that is normally not possible
by X-ray diffraction.
The frequency of X-ray (or neutron) radiation is so high in relation to the
movement of the individual atoms that only a static structure is seen by
individual photons. Therefore the diffraction pattern reflects the time average
of the atomic position. In order to obtain information on the dynamic behaviour
of the molecules, various spectroscopic methods are used. The most powerful
technique is NMR (nuclear magnetic resonance) relaxation, and the NMR
method also provides structural information for the identification of liquid and
liquid-crystalline phases. From infrared and Raman spectra we can obtain
information on molecular conformation, for example whether the hydrocarbon
chains of lipid crystals have their zigzag planes parallel or not.
The enormous increase in X-ray intensity achieved by synchrotron sources
has recently made possible time-resolved X-ray diffraction analysis. This has
been particularly important in order to follow the dynamics of formation of

lipid self-assembly in water.
References
Bragg, Sir W. (1934) Nature 133, 445.
Christie, W.W. (1987) High-Performance Liquid Chromatography and Lipids, Pergamon
Books, Oxford, UK.
Friedel, G. (1922) Ann. Phys. 18, 2373.
Gunstone, F.D., Harwood, J. L. & Padley, F.B., eds (1986; second edition 1994) The Lipid
Handbook, Chapman & Hall, London, UK.
Wechsler, A., Braufman, A., Skafir, M., Heverin, M., Gottlieb, H., Damari, G., Gozlan, S.,
Kelner, I., Spivak, O., Moshkin, O., Friedman, E., Becker, Y., Shaliter, R., Einat, P.,
Faerman, A., Björkhem, I. & Feinstein, I. (2003) Science 302, 2087.
CHAPTER 2
Solid-state behaviour of polymorphic fats and
fatty acids
Lipids are a class of compounds that contain long-chain aliphatic hydrocarbons

and their derivatives (Gunstone & Padley, 1997). Examples of lipids include
fatty acids, alcohols, aldehydes, waxes, glycerols, and phospholipids. The solid
phase of lipids reveals a wide variety of significant features, which largely
depend on the physical states of the materials in which the lipid crystals are
included (Larsson, 1994).
Figure 2.1a depicts typical relationships between physical states and func-
tions of fats and lipids. The crystalline state signifies that the major portion of
the material is composed of lipid crystals, as typically represented in confec-
tionery fat (chocolate) (Timms, 2003). Fine particles of sugar, cocoa mass, and
milk powder are suspended in the continuous phase of the cocoa butter crystals,
which comprise about 30% (w/w) of the total mass of chocolate.
The gel state is defined as a two-phase colloidal system consisting of solid
and liquid components, in which the solid behaviour prevails over the sol state
(Clark, 1992). Gel materials have attracted considerable attention in food,
pharmaceutical, and cosmetic applications because of their smooth surface,
viscoelasticity, appearance, ease of handling, and agreeable texture in the
mouth. The morphology, size, density, and crystal network of lipid crystals are
the dominant factors that influence the physical properties of a gel state made
of lipid materials.
An emulsion is defined as a two-phase colloidal system consisting of water
and oil with the help of emulsifiers that reduce the water–oil interfacial energy
(Dickinson & McClements, 1996). There are two types of emulsion, water-in-
oil (w/o) and oil-in-water (o/w). Typical emulsion systems consisting of solid
lipids are butter, margarine, and spread (w/o) and whipped (o/w) systems, in
which the physical properties of the emulsion, such as the spreadability,
whippability and texture, are influenced by the lipid crystals present in the
continuous phase of the w/o emulsion (Rousseau, 2002), in the dispersed phase
of the o/w emulsion (Walstra et al., 2001), or around air bubbles in ice cream
(Goff, 1997; Clarke, 2004).
In each state shown in Figure 2.1a, macroscopic physical properties of fat
crystal networks play dominant roles (Marangoni, 2005). The principles under-
lying the formation processes of lipid structures are common to the above three
states, which include microscopic and macroscopic features as shown in
Figure 2.1b (Marangoni & Hartel, 1998). Polymorphic structures and primary
9
(a)
Physical state Function
crystal shape, melting,
liquid density, rheology
crystal
gel morphology,
liquid rheology
emulsion
stability, barrier,
o/w rheology, carrier
+H 2O &
emulsifier stability, rheology,
w/o texture
(b)
External conditions
Molecules Primary Macroscopic

Flocs Network
(polymorph) particles structures
Figure 2.1 (a) Relationships between the solid-state behaviour of fats and lipids in different physical
states, and lipid function. (b) Relationships between the microscopic and macroscopic features of fat
crystals.
particles of lipid crystals comprise the microscopic features, whereas the

formation of flocs and networks of lipid crystals determines the macroscopic
features. The molecular structures of lipids are revealed in polymorphism and
primary particle formation. Polymorphism remarkably influences the macro-
scopic properties of fat products. For example, there are three polymorphic
forms in triacylglycerol crystals: α, β′ and β. In margarines and shortenings (de
Man & de Man, 2002), the α form is very short-lived and does not exist in the
finished product. The β crystals are initially small, but they tend to grow into
large needle-like agglomerates, resulting in a sensation of sandiness in the
mouth. The β′ crystals are the most desirable, since they are relatively small and
can incorporate a large amount of liquid oil in the crystal network. Although
external thermodynamic and kinetic factors can be more important than
microscopic features, knowledge of the relationships between molecular struc-
tures, particle formation of different dimensions, and their spatial networks
gives us optimal ways of designing materials with desired functionalities.
The polymorphic structures, mixing behaviour of different fats, and
SOLID-STATE BEHAVIOUR OF POLYMORPHIC FATS AND FATTY ACIDS 11
crystallization properties of the principal fats and fatty acids are discussed in
this chapter as they relate to the physical states of bulk, emulsion, and gel. The
fundamental aspects of lipid crystal structures will be discussed first.
A. Polymorphism: structures and transformations

Almost all lipids possess two or more different crystal structures under a given
set of thermodynamic conditions. This multiplicity in crystalline structures of
the same substance is divided into two categories, polymorphism and polytypism
(Sato & Garti, 1988).
The polymorphic behaviour of lipid crystals is basically determined by the
molecular structure, thermodynamic stability, and phase transformation. The
thermodynamic stability of polymorphic forms is illustrated by the relationship
of their Gibbs energy values, G = H – TS, where H is enthalpy, S is entropy,
and T is temperature (Bernstein, 2002). One can usually determine the relation-
ship between G and T by measuring the temperatures and enthalpy values of the
polymorphic transformations and the temperature variation of the solubility
values of the polymorphic forms. Polymorphic forms with greater G values are
less stable than those with lower ones, and have higher solubility values and
lower melting points. The general behaviour of polymorphic phase transforma-
tions of lipids can be discussed using two example polymorphs, A and B, whose
G–T relationships are depicted in Figure 2.2.
In Figure 2.2, enantiotropic and monotropic relations mean the following
properties. In an enantiotropic phase diagram, there is a temperature at which
the two polymorphs can coexist in a stable equilibrium, and one of the
polymorphs will be more stable than the other above or below this temperature.
In a monotropic phase diagram, such a temperature does not exist, and one
polymorph is always less stable than the others below their melting tempera-
tures.
A reversible solid-state transformation may occur under enantiotropic con-
ditions (Figure 2.2 a and b) when the G values of two forms cross at TA–B, if no
steric hindrance prohibits solid-state transformations from B to A or vice versa.
An endothermic enthalpy change is observed at TA–B during a solid-state
transformation when the temperature is increased. In contrast, solid-state
transformations between the two forms are sterically hindered if the activation
free energy for the transformation (ΔG#) exceeds the thermal energy. The
transformation from A to B in this case may occur far above TA–B on heating
(superheating), while the transformation from B to A may occur below TA–B on
cooling (supercooling). Superheating and supercooling phenomena are fre-
quently observed in lipid crystals. Possible steric hindrances may include
rearrangements of the molecular conformation of the bulky aliphatic chains,
carboxylic acid, and glycerol groups, and rearrangement of methyl end groups
at the lamella–lamella interface. The exact relation of the thermodynamic
(a) (c)
L A L
G G
A B
B
TA–B Tm(B) Tm(A) Tm(B)
T T
(b) (d)
ΔG#S ΔG#m
A ΔG#c
G ΔG# G L
meta-
stable
stable
B B
Figure 2.2 Gibbs energy (G) values of two polymorphic forms, A and B, and liquid (L), as a function
of temperature (T). (a) Enantiotropic relation. (b) Activation of Gibbs free energy (ΔG#) for transforma-
tion from metastable to stable forms. (c) Monotropic relation. (d) Solid-state transformation from A to B
(dashed arrow) and melt-mediated transformation from A to B (solid arrows). Tm, melting temperature.
stability cannot be obtained by observing the solid-state transformation in

conditions of superheating or supercooling. In that event, precise measurement
of the solubility of the two polymorphs over a wide temperature range can
reveal the thermodynamic stability.
When steric hindrance prohibits solid-state transformation between the two
forms above and below TA–B and the two forms are present in a nearly saturated
solution, polymorphic transformations from metastable to stable forms occur in
solution, called a solution-mediated or solvent-mediated transformation (Cardew
& Davey, 1985). The driving forces for this transformation are differences in
solubility. The rates of solution-mediated transformation are determined by the
rates of dissolution of the metastable forms and crystallization of the more
stable forms. Usually, the dissolution rate is higher than the crystallization rate,
so the latter process is rate-determining.
A monotropic relationship of the thermodynamic stability is illustrated
in Figure 2.2c, in which each polymorph has its own melting temperature.
The polymorphism of many acylglycerol crystals exhibits this nature. Two
interesting features are revealed in the polymorphic crystallization and trans-

formation processes when transformation from the metastable A form to the
stable B form is hindered in the solid state due to steric interference. The first
feature is that the Ostwald step rule becomes important, predicting that a phase
change may occur step-by-step through increasingly stable phases (Boistelle,
1988; Aquilano & Squaldino, 2001). Metastable form A nucleates first, before
the more stable form B, when nucleation is induced under a substantial driving
force for nucleation such as supercooling or supersaturation. In contrast,
nucleation of the more stable form B prevails when supercooling or super-
saturation is decreased.
The second feature is that two different polymorphic transformation
processes may occur: solid-state or melt-mediated (Sato & Garti, 1988). The
former transformation occurs from A to B during a heating process with an
exothermic enthalpy value below Tm(A). A melt-mediated transformation from
A to B occurs when the temperature increases to above Tm(A) without solid-state
transformation from A to B. Melting of A and crystallization of B in the liquid
that forms by the melting of A occur in this type of transformation. The rate of
melt-mediated crystallization is often significantly higher than that of a solid-
state transformation if the total activation energy values of the melting of A
(ΔG#m) and the crystallization of B (ΔG#c) are lower than that of the solid-state
transformation (ΔG#s) from A to B, as illustrated in Figure 2.2d. Kashchiev &
Sato (1998) discussed the kinetics of overall crystallization of the stable
polymorph when this form is preceded by the formation of a metastable form,
taking into account the transformation processes depicted in Figure 2.2d.
A melt-mediated transformation has technological significance, since the
tempering process performed during fat crystallization in food engineering
corresponds to a melt-mediated transformation. Furthermore, solution-medi-
ated transformation also occurs from A to B below Tm(A), when steric hindrance
prohibits the solid-state transformation and the two forms are present in nearly
saturated solution. This transformation also has practical significance during
processes in which solid fats coexist with liquid oil through which lipid
molecules diffuse preceding the polymorphic transformation.
1. Fatty acids
This section discusses the polymorphic structures of the principal saturated and
unsaturated fatty acids. Fatty acids are the main hydrophobic moieties of lipids
present in biological tissues, and are also employed for multiple purposes in the
food, pharmaceutical, and polymer industries (Chow, 1992). Saturated fatty
acids exhibit a straight chain configuration, allowing for melting points higher
than those of unsaturated fatty acids with the same number of carbon atoms.
The inclusion of a double bond reduces the melting point of an unsaturated fatty
acid, depending on its conformation (cis or trans) and position in the carbon
Table 2.1 Polymorphism in saturated fatty acids
nc Polymorphic forms
Even-numbered
12 A1(t), A-super(t), C(m)
14 A2(t), B(m), C(m)
16 A2(t), B(m), C(m), E(m)
18 A(t), B(m), C(m), E(m)
Odd-numbered
13 A′(t), C′(t)
15 A′(t), B′(t), C′(m)
17 A′(t), B′(t), C′(m)
nc, number of hydrocarbon atoms. m, monoclinic; t, triclinic.
chain. It also introduces multiplicity in the possible molecular conformations

of chain segments separated by the double bond, resulting in complex
polymorphic structures of unsaturated fatty acids.
Saturated fatty acids

Table 2.1 summarizes the number and polymorphic crystalline forms of the
saturated fatty acids with nc values (number of hydrocarbon atoms) of 12
through 18. Triclinic and monoclinic forms commonly occur in both even- and
odd-numbered acids. Crystal structural analyses have been performed for the C
form (Malta et al., 1971), B form (Goto & Asada, 1978), and E form (Kaneko
et al., 1990) of stearic acid (C18:0), as depicted in Figure 2.3a. Forms C and E
contain straight hydrocarbon chains in the unit cells, but the gauche
conformation is found at the end of the chain closest to the COOH group in B.
Single-crystal X-ray diffraction studies showed that, surprisingly, the
molecular volume in the crystal is greatest for C, smallest for B, and between
for E (Goto & Asada, 1978; Kaneko et al., 1990).
Subcell structures are defined as the cross-sectional packing mode of the
zigzag aliphatic chain to characterize the molecular packing of the aliphatic
chains, as illustrated in Figure 2.4 (Shipley, 1986). More than 9 types of subcell
structures have been identified in fatty acids, acylglycerols, and polar lipids
(Pascher et al., 1992); the five subcells depicted in Figure 2.4 are often revealed
in the crystals of fatty acids and triacylglycerols.
All the polymethylene zigzag planes are parallel in T⏐⏐ and this form is
thought to have the densest subcell packing. The O⊥ subcell consists of zigzag
planes that are perpendicular to the planes of its neighbours. Two subcells, O′⏐⏐
and M⏐⏐, contain zigzag aliphatic chains arranged in parallel in orthorhombic
and monoclinic systems, respectively. There is also hexagonal (H) chain
packing in addition to the above four specific subcells, in which the
polymethylene chains do not assume a specific orientation. Instead, they
c/2 c/2
c/2
(a)
a a
a
B form C form E form
(b)
monoclinic orthorhombic
Figure 2.3 (a) Molecular structures of three polymorphs of stearic acid: B, C, and E. In each case, the
projection of half of the unit cell (a, c/2) is shown. (b) Illustration of polytypism of monoclinic type and
orthorhombic type.
H T M
cs
O⊥ O′
bs
as
Figure 2.4 Typical subcell structures present in lipid crystals. Subcell parameters as, bs and cs are given
for O⊥ as an example.
Figure 2.5 Relationship between Gibbs energy (G) and temperature for five crystal forms of stearic
acid.
undergo torsional motion with an aliphatic gauche conformation, making the H

subcell less stable than the other four. The C, B, and E forms of stearic acid
presented in Figure 2.3 have the O⊥ subcell. A triclinic A form with the T⏐⏐
subcell occurs in stearic acid, although it is not included in Figure 2.3.
A higher-order modification called polytypism has been observed in
n-alkanes and saturated and unsaturated fatty acids (Sato & Kobayashi, 1991),
in addition to polymorphism. Polytypism forms due to a stacking mode
difference in the unit cell layers normal to the basal plane while maintaining the
two-dimensional packing arrangements. Monoclinic and orthorhombic
polytypes have been thus far observed, as indicated in Figure 2.3b.
Stearic acid, one of various saturated fatty acids, has been extensively
studied both for polymorphism and polytypism. E and B polymorphs of stearic
acid possess two polytypic structures, Em and Eo (Kaneko et al., 1990) and Bm
and Bo (Kobayashi et al., 1984, 1986a); no polytypic modification was
observed for the C form (denoted Cm). Detailed crystal structures of the
polymorphic forms of stearic acid were described by Kaneko (2001). There-
fore, five polymorphic or polytypic forms with the O⊥ subcell and a triclinic A
form with the T⏐⏐ subcell of stearic acid are present. Thermodynamic stability of
an enantiotropic nature was determined by measuring the solubility and solid-
state transformation behaviour, as illustrated in Figure 2.5, except for the A
form.
The orthorhombic polytype of the B polymorph (Bo) is thermodynamically
more stable than the monoclinic polytype (Bm), as determined by careful
Figure 2.6 Structure relationships of a polytypic–polymorphic transformation of stearic acid.
solubility measurements (Sato et al., 1988). This result is explained by the

contribution of free energy of the lattice vibration longitudinal to the lamella
interface, which lowers the total Gibbs energy of Bo compared to Bm (Kobayashi
et al., 1984). Crystallization from neat liquid only forms Cm, which does not
transform to other forms in the solid state because of steric hindrance. Instead,
crystallization from solution can form Em, Eo, Bm, and Bo.
The four forms, Em, Eo, Bm, and Bo, transform to Cm on heating. However, the
transformation temperatures are higher than the crossing point of the Gibbs
energy values determined by solubility measurements. For example, Bm and Bo
transformed to Cm in the solid state at 40–50ºC, which is greater than the
crossing points of the G values (23ºC for Bm–Cm and 32ºC for Bo–Cm). This
overheating property is explained by two different modes of lattice displace-
ment associated with the polytypic–polymorphic transformation of Bo→Cm
and Bm→Cm of stearic acid (Inaoka et al., 1988). Electron microscopic obser-
vations of the step patterns of cleaved stearic acid crystals before and after the
transformation indicated that an alternative rotation of stearic acid molecules
occurred in the adjacent lamellae around the c-axis of Bo during the Bo→Cm
transformation, keeping the subcell arrangements unchanged. In contrast,
displacements during the Bm→Cm transformation are caused by a collective
inclination of the molecules within the lamellar plane followed by deformation
of the subcell of the aliphatic chain, keeping the symmetry axis (c-axis)
unchanged. We can infer that the lattice displacements illustrated in Figure 2.6
are the primary cause for the overheating of the Bo→Cm and Bm→Cm trans-
formations.
Transformation from Em to Bo did not occur in the solid state. This behaviour
is caused by steric hindrance, which may involve variations in the subcell
arrangements, slip movement of the long-chain molecules, and/or the stacking
sequences at the methyl–methyl end groups. Conversions among the polytypic
Table 2.2 Molecular properties of polymorphism in oleic acid, elaidic acid and erucic
acida
Fatty acid Form Local conformation Subcell

near double bond
Oleic acid γ S-C-S′ O′ ⏐⏐
α S-C-T O′ + O -like
⏐⏐ ⊥
β2 Unclear ⏐⏐-type
β1 A: (174º,cis,173º) (T-C-T) T ⏐⏐
B: (175º,cis,175º) (T-C-T)
Elaidic acid LM (118º,trans,118º) (S-T-S′) O⊥
HM Unclear O + ⏐⏐-type
⊥
Erucic acid γ S-C-S′ O′ ⏐⏐
γ1 S-C-S O′ + O -like
⏐⏐ ⊥
α S-C-T O′ + O -like
⏐⏐ ⊥
α1 S-C-S T ⏐⏐
a
Data taken from Suzuki et al., 1985; Ueno et al., 1994; Kaneko et al., 1997b; Kaneko et al., 1998.
S-C-S′, skew-cis-skew′; S-C-S, skew-cis-skew; T-C-T, trans-cis-trans; S-C-T, skew-cis-trans. A and B
are two asymmetric units in a unit cell. LM, low-melting; HM, high-melting.
forms of stearic acid were also observed in a solution-mediated transformation

in which a metastable Em form transformed to the most stable Bo form through
nucleation and crystal growth of Bo at the expense (dissolution) of Em (Kaneko,
2001).
Unsaturated fatty acids

Unsaturated fatty acids are important lipid molecules that play critical roles in
the functional activities of biological organisms, promoting fluidity and perme-
ability of the membrane through conformational flexibility of their acyl chains
(Gennis, 1989). The number, position, and configuration of the double bonds
are the primary factors that influence the physical and chemical properties of
unsaturated fats and lipids. Hagemann et al. (1975) measured the melting
points of a series of single-acid triacylglycerols made from C18:1 isomers
(monounsaturated fatty acids with 18 carbon atoms) in which the double bond
with cis or trans configuration was placed at various different positions along
the aliphatic chain. It was confirmed that: (a) the melting point was lower for
the cis isomer than for the trans isomer at a given position of the double bond;
(b) the melting points were lower when the double bond was placed at odd-
numbered carbon positions than when it was placed at even-numbered carbon
positions, for both cis and trans configurations; and (c) the melting point was
lower when the double bond was located at a central position along the acyl
chain rather than near either the –CH3 end group or the –COOH group forming
the linkage with the glycerol backbone, irrespective of the cis–trans
skew-cis-skew ′ skew-cis-skew trans-cis-trans

(S-C-S′) (S-C-S) (T-C-T)
Figure 2.7 Olefinic conformations of a cis double bond.
configurations and odd–even positions. The same behaviour was observed in

the thermal properties of lecithins with acyl chain moieties of various C18:1
isomers (Barton & Gunstone, 1975).
As to polymorphic properties, recent studies have clarified that polymor-
phism of the unsaturated fatty acids differs significantly from that of the
saturated fatty acids, in terms of molecular features and transformation behav-
iour (Kaneko et al., 1998). For example, the polymorphic structures of oleic
acid (C18:1, cis-ω9) (Suzuki et al., 1985) and elaidic acid (C18:1, trans-ω9) (Ueno
et al., 1994) are summarized in Table 2.2 as typical examples of cis- and trans-
monounsaturated fatty acids.
The polymorphic structures of unsaturated fatty acids require a discussion of
the cis conformation of a C=C double bond, which is associated with local
conformation of the C–C–C bonds that are adjacent to the double bond, as
indicated in Figure 2.7 (Kaneko et al., 1998). Two conformations of skew (S,
internal rotation angle of 120º and S′, internal rotation angle of –120º) are
considered stable. The trans (T) conformation indicates an almost-straight
zigzag chain.
The subcell structure may be a good reference for a comparison of oleic acid
and elaidic acid, since O⊥ is revealed in elaidic acid, while subcells with parallel
packing, such as O′⏐⏐ and T⏐⏐, are observed in oleic acid. This is evidently a
reasonable consequence since parallel packing is favoured by a chain structure
that is bent at the cis double bond (oleic acid), and the linear chain of the trans
double bond (elaidic acid) is packed with a perpendicular subcell arrangement.
However, this property only applies to oleic and elaidic acids. Diversity in the
subcell structure and local hydrocarbon conformation near the cis double bond
can be observed in different unsaturated fatty acids, as explained in a previous
Figure 2.8 Relationship between Gibbs energy (G) and temperature for the γ, α, β2, and β1 forms of
oleic acid.
review (Kaneko et al., 1998). We discuss polymorphism of oleic acid,

petroselinic acid, linoleic acid, and α-linolenic acid here as representative
examples.
Figure 2.8 illustrates the Gibbs energy–temperature relationship of four
polymorphic forms of oleic acid, which were determined by solid-state trans-
formation behaviour (Suzuki et al., 1985), solubility measurement (Sato &
Suzuki, 1986) and solvent crystallization (Kaneko et al., 1997b). The α and γ
forms exhibit a reversible transformation of an enantiotropic nature. Kobayashi
et al. (1986b) clarified, using vibrational spectroscopic and thermal tech-
niques, that the aliphatic conformation of the ω chain, which is defined as the
chain segment between the double bond and the methyl end, transforms from
ordered (γ) to disordered (α) upon heating, whereas the ordered conformation
of the Δ chain does not change, as illustrated in Figure 2.9. Atomic-level crystal
structure analysis was done for α (Kaneko, 2001) and γ (Kaneko et al., 1996)
polymorphs of erucic acid, whose structural properties are identical to those of
the α and γ forms of oleic acid. The subcell structure of the γ form is O′⏐⏐ for both
the Δ and ω chains. However, the subcell structure of the ω chain changes to
O′⊥-like when the γ form transforms into α, and the local conformation near the
double bond changes to S-C-T, while the subcell structure of O′⏐⏐ of the Δ chain
is unchanged. This property is consistent with the order–disorder transforma-
tion that occurs in the ω chain, which is thought to be due to partial melting of
the ω chain. The same polymorphic properties as those of the α and γ forms of
oleic acid can be observed in other cis-monounsaturated fatty acids, such as
erucic acid (C22:1, cis-ω9) (Kaneko et al., 1993, 1996), asclepic acid (C18:1, cis-
ω7) (Yoshimoto et al., 1991), palmitoleic acid (C16:1, cis-ω7) (Hiramatsu et al.,
1990), and gondoic acid (C20:1, cis-ω9) (Sato et al., 1997), and also in
Figure 2.9 Schematic illustration of a γ–α transformation.
polyunsaturated fatty acids such as linoleic acid (C 18:2, cis-ω6,9) and

α-linolenic acid (C18:3, cis-ω3,6,9) (Ueno et al., 2000). Therefore, we can
conclude that partial melting of the ω chain is a typical structural characteristic
in unsaturated fatty acids caused by introduction of the cis double bond.
Atomic-level structure analyses with X-ray diffraction using single crystals
have been performed on polymorphic forms of the principal cis-
monounsaturated fatty acids, some of which are presented in Figure 2.10. The
common properties revealed in the crystal structures shown in Figure 2.10 can
be summarized as follows.
(1) The COOH groups of neighbouring molecules are located in the same
plane, normal to the chain axes. This arrangement may minimize the
polar intermolecular energy of the crystal.
(2) The cis double bonds of neighbouring molecules are also located in the
same plane. The intermolecular π–π interactions may be stabilized by
this arrangement.
Kaneko et al. (1997b) found a unique polymorphic property of β-type
polymorphs of oleic acid with a T⏐⏐ subcell structure (Table 2.2). Successive
melt-mediated transformations occur from α (Tm=13.3ºC) to β2 and from β2
(Tm=16.0ºC) to β1 (Tm=16.3ºC) on heating, as indicated in Figure 2.8. The β1
form is obtained by this transformation from the melt phase, since its growth
rate is notably lower than that of the other forms by a factor of 10–3 to 10–4. This
uniqueness in the β1 structure is ascribed to its crystal structure, as explained
below.
The β1 form belongs to a triclinic system, containing two independent
molecules in an asymmetric unit, as illustrated in Figure 2.11. An interdigitated
chain-length structure and the local hydrocarbon conformation near the cis
c/2 c c
a
(a)
c/2
a
(b) b a
(c) (d)
Figure 2.10 Crystal structures of polymorphic forms of the principal cis-monounsaturated fatty acids:
(a) γ form of erucic acid; (b) α form of erucic acid; (c) LM (low-melting) form (orthorhombic polytype)
of petroselinic acid; (d) LM form of elaidic acid.
double bond make this polymorph so unique. The olefinic conformation is

trans-cis-trans (T-C-T), and the acyl chains form the T⏐⏐ subcell. A T-C-T-type
local conformation adjacent to the cis double bond was observed for the first
time in oleic acid β1 form; S-C-S, S-C-S′, or S-C-T conformations are usually
observed in monounsaturated fatty acids. The interdigitated structure forms in
such a way that the methyl group of molecule A and the carboxyl group of
molecule B (or vice versa) are located in the same plane. This stacking mode is
unusual, since fatty acids usually form a separated lamellar structure, in which
the methyl terminals are segregated from the carboxyl groups, except for
triclinic A1 and A-super forms of lauric acid (Kaneko, 2001). There are two
aliphatic chains with subcell 1 and subcell 2. Subcell 1 is formed from the chain
Figure 2.11 Crystal structure of the β1 form of oleic acid.
segment consisting of the methyl group of molecule A and the carboxyl group
of molecule B, and subcell 2 is made of the methyl group of molecule B and the
carboxyl group of molecule A. The subcell parameters are summarized in
Table 2.3.
The fact that the interdigitated lamellar structure of β1 was only observed in
oleic acid can be explained as follows. Oleic acid (C18:1) has the same number
of carbon atoms (9) in its ω and Δ chains. For most other cis-monounsaturated
fatty acids, the ω and Δ chains are of different length. Therefore, a void would
form in the interdigitated structure either between the methyl ends or the
carboxyl terminal groups of adjacent molecules. A void between adjacent
carboxyl groups would result in an excessive energy barrier to formation of the
hydrogen bonds that stabilize the structure, and so this type of crystal would not
form.
Table 2.3 Subcell parameters of oleic acid β1 forma
Parameter Subcell 1 Subcell 2
as 4.09 nm 4.18 nm
bs 5.36 nm 5.54 nm
cs 2.54 nm 2.54 nm
α 81.4º 72.2º
β 106.5º 109.7º
γ 120.8º 123.5º
a
Data taken from Kaneko et al., 1997b.
Table 2.4 Crystal structure data of two polymorphs of petroselinic acid
Polymorph Space group Subcell Olefinic conformation

–
HM P1 M⏐⏐ S-C-S
LM Pbca O⊥ S-C-S′
Data taken from Kaneko et al., 1992a,b.
Figure 2.12 Polymorphic behaviour of petroselinic acid.

The polymorphic behaviour varies to a great extent when the position of

the cis double bond in the aliphatic chain is varied, as clearly observed
in petroselinic acid (C18:1, cis-ω12), a positional isomer of oleic acid with
polymorphic structures very different from those of oleic acid but similar in
part to those of stearic acid (Sato et al., 1990; Kaneko et al., 1992a, 1992b,
1997a).
Figure 2.12 depicts the thermodynamic stability relation of two polymorphs
of petroselinic acid, HM (high-melting form, Tm=30.5ºC) and LM (low-
melting form, Tm=28.5ºC). The LM and HM forms did not reveal any
order–disorder transformation, and the crossing point of the Gibbs energy
values of the two forms, 18.7ºC, was determined by measuring the solubility
values of the two forms at different temperatures. No transformation from HM
to LM occurred below 18.7ºC (supercooling), and LM slowly converted to HM
above that temperature (superheating).
An interesting feature was revealed in the crystal structures of the two forms,
–
as summarized in Table 2.4. The space group of HM is triclinic P1, whereas LM
is of orthorhombic space group Pbca with the O⊥ subcell. The crystal shape of
LM is almost the same as that of the B and E forms of stearic acid. Correspond-
ingly, the subcell parameters of O⊥ near the methyl terminals of the LM form of
petroselinic acid are very similar to those of the B and E forms of stearic acid.
Furthermore, the occurrence behaviour of polytypic modifications was almost
identical; e.g. LM has orthorhombic and monoclinic polytypes, similar to B and
E of stearic acid. The stabilizing factor of the O⊥ subcell of the LM form of
petroselinic acid is thought to be stabilization of the methyl-end interface. It is
possible to construct the methyl-end packing in the same manner as that of
even-numbered saturated fatty acids, since petroselinic acid has an even
number of carbon atoms in the ω chain. The absence of any order–disorder
transformation with respect to the ω chain in petroselinic acid may also be
ascribed to this property of methyl-end packing.
The polymorphic behaviour of linoleic acid and α-linolenic acid is summa-
rized in Table 2.5 (Ueno et al., 2000). Thermal analysis uncovered three
polymorphic forms of linoleic acid – LT (low-temperature), MT (medium-
temperature), and HT (high-temperature) – and LT and HT forms of α-linolenic
acid. The X-ray diffraction study indicated that the HT and LT forms of the two
acids exhibited the same diffraction patterns as those of the α and γ forms of
oleic acid. The crystal shapes of the HT forms of linoleic and α-linolenic acids
(Figure 2.13) were also identical to that of the α form of oleic acid. Therefore,
we can conclude that the order–disorder transformation observed in many cis-
monounsaturated fatty acids also occurs in polyunsaturated fatty acids. The
transformation temperatures and the enthalpy and entropy values of the trans-
formation of LT(→MT)→HT, and HT→melt decreased with an increase in the
number of cis double bonds.
Table 2.5 Polymorphic properties of linoleic acid, α-linolenic acid, and oleic acida
Polymorph Ttr (ºC) Δ Htr(kJ/mol) Δ Str(J/mol•K)
Linoleic acid
LT –51.3 (→ MT) 2.6 8.1
MT –35.4 (→ HT) 0.27 0.86
HT –7.2 (→ melt) 33.6 120.9
α-Linolenic acid
LT –60.2 (→ HT) 0.11 0.46
HT –13.0 (→ melt) 27.8 106.1
Oleic acid
γ –2.2 (→ α) 8.76 32.3
α 13.3 (→ melt) 39.6 138.4
a
Data taken from Suzuki et al., 1985; Ueno et al., 2000.
Ttr, transformation temperature. ΔHtr, enthalpy of transformation. ΔStr, entropy of transformation.
2. Diacylglycerols
Monoacylglycerols and diacylglycerols are polar acylglycerols. The crystal-
line and lyotropic liquid crystalline properties of monoacylglycerols were
recently reviewed by Krog (2001). Therefore, this section focuses on
diacylglycerols (DAGs) as a representative group of polar acylglycerols.
Interesting properties were observed in the crystal structures of three types of
saturated–unsaturated mixed-acid DAGs, 1-stearoyl-2-oleoyl-sn-glycerol (sn-
1,2-SODG) (Di & Small, 1993), 1-stearoyl-3-oleoyl-sn-glycerol (sn-1,3-SODG)
(Goto et al., 1995), and 1-stearoyl-2-linoleoyl-sn-glycerol (sn-1,2-SLiDG) (Di
& Small, 1995). Molecular structures of the diacylglycerols composed of one
saturated and one unsaturated fatty acyl chain make up the hydrophobic core of
many biological membranes. Therefore, the interactions of the acyl chains in
the membrane bilayer are of interest because the saturated and unsaturated
chains have marked difficulty in packing together in the crystalline state
(Small, 1984).
The stearoyl and oleoyl chains were stacked in separate leaflets of the double
chain-length structure in the stable form of sn-1,3-SODG (Tm=42.5ºC), in
which both leaflets were packed in the T⏐⏐ subcell (Figure 2.14a). The
molecules formed an extended V-shaped conformation, with the oleoyl and
stearoyl chains coming off the two ends of the glycerol group with an angle of
94º between their planes. The combination of the T⏐⏐ subcell and the trans-skew-
cis-skew-skew-trans (T-S-C-S-S-T) conformation at carbons 7–13 in the oleic
acid leaflet is of particular interest. Hydrogen bonding formed within the
lamellar interface between the oxygen atom of the glycerol sn-2 carbon and the
–C=O group of the oleoyl chain.
There are eight polymorphic forms of sn-1,2-SODG in the dry state: γ2, γ1, α,
HT form of linoleic acid HT form of α-linolenic acid
Figure 2.13 Crystal shapes of the HT forms of linoleic acid and α-linolenic acid.
c/2 c/2 c/2
a
b b
(a) (b) (c)
Figure 2.14 Crystal structures of: (a) sn-1,3-SODG; (b) sn-1,2-DPDG; (c) sn-1,2-SODG.
β4, β3, β2, β1, and β′. Interestingly, the most stable form of sn-1,2-SODG is β′
(Tm=25.7ºC), and the second most stable form is β1 (Tm=23.1ºC). This contrasts
with triacylglycerols (TAGs), whose most stable form is usually β with the T⏐⏐
subcell (to be discussed below). A structure model of the β′ form of sn-1,2-
SODG was constructed as depicted in Figure 2.14c, based on the crystal
structure of 1,2-dipalmitoyl-sn-glycerol (sn-1,2-DPDG) in Figure 2.14b
(Dorset & Pangborn, 1988). A hairpin conformer structure formed in the
monoclinic crystal of sn-1,2-DPDG since two palmitoyl chains were aligned in
the same leaflet with an angle of 92.9º with respect to the lamellar plane.
Hydrogen bonding formed between the oxygen atom of the glycerol sn-3
carbon and the –C=O group of the sn-1 palmitoyl chain. These structures are
basically the same as that of 1,2-dilauroyl-sn-glycerol. An interesting point is
that the palmitoyl chains in sn-1,2-DPDG are packed according to the O⊥
subcell. X-ray powder diffraction data indicate that the β′ form of sn-1,2-
SODG is stacked in a double chain-length structure. However, we cannot
determine whether a hairpin structure like that of sn-1,2-DPDG forms, or a
V-shaped structure like that of sn-1,3-SODG. Figure 2.14c postulates that a
hairpin structure might be formed by decreasing the steric hindrance between
the stearoyl and oleoyl chains that are packed in the same leaflet.
There are four polymorphs of sn-1,2-SLiDG in the dry state: α, sub-α1,
subα2, and β′. The two sub-α forms are metastable low-temperature polymorphs,
and the melting points of the α and β′ forms are 11.6ºC and 16.1ºC,
respectively. Hydrated sn-1,2-SLiDG possesses three forms: αw, sub-αw1, and
sub-αw2. The subcell packing is hexagonal for the α form and O⊥ for the β′ form.
sn-1,2-SLiDG packs much less efficiently than sn-1,2-DPDG, but appears to
pack more efficiently than sn-1,2-SODG.
3. Triacylglycerols
The most naturally abundant non-polar lipids are triacylglycerols (TAGs),
waxes, and cholesterol esters. TAGs are discussed in this section as a repre-
sentative group of non-polar lipids, since TAGs are the predominant constituents
of fats and oils. A TAG is a three-fold ester of glycerol and fatty acids. TAGs
can be divided into three classes with respect to their fatty acid composition.
TAGs with only one type of fatty acid are called monoacid TAGs; those with
two or three types of fatty acids are called diacid and triacid TAGs, both of
which are referred to in this section as mixed-acid TAGs. Diacid TAGs are
further divided into two types, symmetric and asymmetric. Chiral properties
are revealed in the asymmetric diacid TAGs, and also in triacid TAGs.
The physical properties of TAGs are determined by the types of their fatty
acid moieties, e.g. saturated and unsaturated chains, cis and trans double
bonds, short and long chains, chains with even and odd numbers of carbons,
and the positions of the esterified fatty acids on the glycerol carbon atoms.
Knowing the fatty acid compositions of TAGs enables us to modify the
molecular shapes of the fats through hydrogenation, interesterification,
fractionation, or genetic engineering, to obtain desirable physical properties for
fat-based products.
TAG molecules adopt the ideal conformation and arrangement in relation to
their neighbours in the crystalline state to optimize the intra- and intermolecu-
lar interactions and achieve efficient close-packing states. Such states are
characterized by a packing of long-chain molecules within a plane normal to
the chain axis, called lateral packing and defined by subcell, and also by a
stacking of lamellae with a specific lateral packing along the chain axis, called
Figure 2.15 Chain-length structures of triacylglycerol crystals.
chain-length structure. Pioneering work on the polymorphic structures of

various TAG crystals dating back to the 1940s was summarized by the reviews
of Bailey (1950), Lutton (1950) and Malkin (1954), in which, surprisingly in
view of the limited analytical methodology available in those days, basic
concepts of complicated subcell and chain-length structures were already
discussed. Larsson (1966) led to classification of the three fundamental
polymorphs in terms of the subcell structures α (H), β′ (O⊥), and β (T⏐⏐).
In this section, it is worthwhile to consider the chain-length structure in
addition to the subcell and olefinic structures when characterizing the solid-
state behaviour of TAGs at a molecular level. The chain-length structure is
defined as the number of repeating units of the hydrocarbon chains involved in
a unit lamella along the c-axis, as indicated in Figure 2.15. One repeating unit
layer made up of one hydrocarbon chain is called a leaflet. TAGs composed of
the same or very similar fatty acids may form a double chain-length structure,
such as that displayed for the β form of tricaproyl-glycerol (CCC) (Jensen &
Mabis, 1963, 1966). A triple chain-length structure is formed when the
chemical natures of one or two fatty acids differ significantly from the others.
It is notable that the subcell structures of different leaflets in a triple chain-
length structure can transform in different manners. In addition, quatro
chain-length (Sato et al., 2001) and hexa chain-length (Kodali et al., 1989)
structures have been observed in TAGs with complicated fatty acid composi-
tions. We focus on the polymorphic structures of mixed-acid TAGs in this
section, since saturated monoacid TAG structures were summarized in a recent
article (van Langevelde et al., 1999a).
Polymorphism in diacid triacylglycerols

The polymorphism of diacid-type TAGs differs substantially from that of
monoacid TAGs. Figure 2.16 provides an example of the variation in melting
points and long-spacing values of the most stable forms of diacid TAGs of PPn,
which is a series of TAGs with palmitic acid (C16:0) at the sn-1 and sn-2
positions and a saturated acid with an even number of carbon atoms (n = 0–16)
at the sn-3 position. Kodali et al. (1984) clarified that the most stable form was
β for PP2, PP4, PP10, and PP12, and β′ for PP6, PP8 and PP14, in which the chain-
length structures were double for PP0 (dipalmitoyl-glycerol), PP2, PP4, PP12,
and P16 (PPP), triple for PP6, PP8 and PP10, and quatro for PP14. Correspondingly,
as n increased from 0 to 16, the melting point decreased from 60ºC (for PP0–
PP4), stabilized around 42ºC (for PP4–PP10), and increased from 56ºC (PP12) to
68ºC (PPP). It was clarified that the polymorphic transformation behaviour
differs markedly from one TAG to the next in the diacid TAGs of PPn (Kodali
et al., 1990a,b). Figure 2.17 depicts structural models of PP2, PP6, PP10, and
PP14. The atomic-level crystal structures were subsequently determined for the
β form of PP2 (Goto et al., 1992) and the β′ form of PP14 (Sato et al., 2001).
The polymorphism of symmetric diacid TAGs CnCn+2Cn (n = 10–16) exhib-
ited other unique properties. The most stable form was β′ for C10C12C10,
C12C14C12, C14C16C14, and C16C18C16, and no β form was observed (van Langevelde
et al., 1999b). The long-spacing values and melting points of the β′ forms of the
four TAGs of CnCn+2Cn linearly increased with increasing values of n.
The polymorphic behaviour of three series of TAGs (PPn, SnS, and CnCn+2Cn)
indicated that specific chain–chain interactions of the methyl-end packing,
aliphatic chains and glycerol groups may induce very complicated polymor-
phic behaviour. Interestingly, the β′ form can be more stable than the β form in
some of these diacid TAGs. Therefore, atomic-level structure analyses of the β′
form have recently been conducted using single crystals of PP14 and C10C12C10,
as separately discussed in detail below.
Many researchers have discussed complicated polymorphic structures of
saturated–unsaturated diacid TAGs in terms of subcell packing, chain-length
structure, and aliphatic chain conformation, which may be revealed differently
in saturated and unsaturated moieties (Larsson, 1972; Small, 1986; de Jong et
al., 1991). Figure 2.18a illustrates typical features of polymorphic transforma-
tion of SOS (1,3-distearoyl-sn-2-oleoyl-glycerol) α, γ, β′, β2, and β1 forms. The
chain-length structure converted from double (α) to triple (γ, β′, β2, and β1)
(Sato et al., 1989; Koyano et al., 1989; Arishima & Sato, 1989). The presence
Figure 2.16 Melting points () and long-spacing values () of asymmetric diacid TAGs of the form
PPn (see text for details).
Figure 2.17 Structure models of PP2, PP6, PP10, and PP14.
of two β forms of SOS (Arishima & Sato, 1989) was recently confirmed by a
powder X-ray diffraction study (Peschar et al., 2004). Figure 2.18b depicts the
melting behaviour of the polymorphic forms of 1,3-dipalmitoyl-2-sn-oleoyl-
glycerol (POP), 1,3-palmitoyl-stearoyl-2-oleoyl-rac-glycerol (POS) (Arishima
et al., 1991; Koyano et al., 1991), 1,3-stearoyl-arachidoyl-2-oleoyl-rac-glyc-
erol (AOS) (Arakawa et al., 1998), 1,3-diarachidoyl-2-oleoyl-glycerol (AOA),
and 1,3-dibehenoyl-2-sn-oleoyl-glycerol (BOB) (Wang et al., 1987), which is
basically the same as that of SOS. Their primary polymorphic transformation
properties are summarized below, mostly as referred to SOS, which was
studied with synchrotron radiation X-ray diffraction (Ueno et al., 1997),
vibrational spectroscopy (Yano et al., 1993, 1999) and solid-state nuclear
magnetic resonance (NMR) (Arishima et al., 1996) techniques.
The α forms of all the TAGs shown in Figure 2.18b revealed the same crystal
structure, having the double chain-length arrangement. This indicates that the
stearoyl and oleoyl moieties in the same leaflet are packed in accordance with
Figure 2.18 (a) Molecular model of polymorphism of SOS. (b) Melting points of polymorphic forms
of POP, POS, SOS, AOS, AOA, and BOB (: α, : γ, : β′, Δ: β2, : β1).
the H subcell, and that disordered, non-specific chain packing of the hexagonal
subcell may not cause steric hindrance between saturated and oleic acid
moieties. The α form transforms to the γ form of the triple chain-length
structure, in which the oleoyl and stearoyl leaflets are separated as a result of
chain sorting. The stearoyl leaflet is assumed to be parallel packed, and the
oleoyl leaflets retain a hexagonal subcell structure. The total chain of SOS in
the γ form is arranged normal to the lamella interface. The triple chain-length
structure in the β′ form is maintained in all the above TAGs except for POP, in
which the β′ form has a double chain-length structure. The stearoyl leaflet in the
β′ form is packed according to the O⊥ subcell, while the hexagonal subcell
structure is retained in the oleoyl leaflet. The SOS chains are inclined with
respect to the lamella plane. Finally, the β′ form transforms to two β forms,
which reveal an inclined chain arrangement against the lamellae interface with
long-spacing values of 6.75 nm (β2) and 6.60 nm (β1) for SOS. The subcell
structure is T⏐⏐ for both the stearoyl and the oleoyl leaflets in the β1 form. It is
assumed that the subcell structures of the three leaflets of the β2 form are similar
to T⏐⏐, but the molecular conformation of the oleoyl leaflet may differ from that
of β1.
The differences in the conversion of the subcell structure between the
stearoyl and oleoyl leaflets of SOS were analysed by a polarized FTIR (Fourier
transform infrared) technique. Infrared CH2 scissoring and CH2 rocking re-
gions are good indicators of subcell packing (Yano et al., 1999). The bands of
oleoyl leaflets overlapped those of stearoyl leaflets for the usual hydrogenated
specimen, and thus partial deuteration was attempted so that the stearoyl chains
would be deuterated and the oleoyl chains would be hydrogenated to form
SDOHSD. The FTIR scissoring and rocking spectra of the stearoyl and oleoyl
leaflets of SDOHSD could thus be separated. A single band was observed for both
the SD chain and the OH chain of the α form, corresponding to the hexagonal
subcell. A sharp single band for the SD chain and a broad band for the OH chain
were observed in the γ form, suggesting that the stearoyl groups form parallel
packing and the oleoyl moiety packs in the hexagonal subcell. The scissoring
and rocking spectra of the SD chain in the β′ form exhibited two components,
indicating a O⊥ subcell. Parallel packing was indicated in the oleoyl leaflet in
the β′ form, since no splitting occurred. Both the stearoyl and oleoyl moieties
were packed in the T⏐⏐ subcell in the β1 form. Knowledge of the polymorphic
structures of SOS and other homologues is very important for the confectionery
industry, since the stable polymorphs of cocoa butter, which contains ~80%
POP, POS, and SOS, are thought to be identical to the two β forms of SOS.
Mykhaylyk & Hamley (2004) and Mykhaylyk et al. (2004) recently carried
out an analysis of the electron density profiles of the chain-length structures of
SOS polymorphs, using long-spacing X-ray diffraction patterns. Their findings
supported the structure models depicted in Figure 2.18a. A crystal structure
analysis of the β2 form of SOS performed using powder X-ray diffraction
methods confirmed the existence of two β forms in SOS (Peschar et al., 2004).
However, the kink model at the cis double bond depicted in Figure 2.18a was
not observed. It was revealed instead that the oleoyl and stearoyl chains were
both relatively straight.
The polymorphism of the series of saturated-oleic-saturated mixed-acid
TAGs summarized in Figure 2.18 is notable in cocoa butter (CB), which is the
most popular confectionery fat. CB consists of three major TAGs (POP, POS,
and SOS), plus other minor components (Timms, 2003). The three TAGs
determine the polymorphic nature of CB, which exhibits six polymorphs,
Form I through Form VI (Wille & Lutton, 1966). Form V is important for the
functionality of chocolate, and thus crystallization of CB in Form V and
preservation of this polymorph during long storage are prerequisites for quality
Table 2.6 Thermal data for SSS, SOS, SRS, and SLiS polymorphsa
TAG Polymorph Melting point (ºC) Δ H (kJ/mol) Δ S (J/mol•K)
SSS α 55.0 109.3 333.1

β′ 61.6 142.8 426.5
β 73.0 188.4 544.3
SOS α 23.5 47.7 160.79
γ 35.4 98.5 319.23
β′ 36.5 104.8 338.47
β2 41.0 143.0 455.19
β1 43.0 151.0 477.62
SRS α 25.8 58.1 194.4
γ 40.6 119.6 381.3
β′2 44.3 171.2 539.3
β′1 48.0 184.8 575.3
SLiS α 20.8 40.9 139.2
γ 34.5 137.4 448.7
a
Data taken from Sato et al., 1989; Boubekri et al., 1999; Takeuchi et al., 2000.
control of the end product. The correspondence between the polymorphic

properties of cocoa butter and SOS is thought to be as follows: Forms I and II
of CB correspond to sub-α and α forms of SOS, Forms III and IV of CB
correspond to the β′ form of SOS, and Forms V and VI of CB correspond to the
β2 and β1 forms of SOS. This correspondence was also supported by structural
determination with powder X-ray diffraction methods by Peschar et al. (2004).
The polymorphic structure shown in Figure 2.18a was found to be iso-
morphic to stearic/unsaturated diacid TAGs containing ricinoleic acid (SRS)
(Boubekri et al., 1999) or linoleic acid (SLiS) (Takeuchi et al., 2000) at the sn-
2 position instead of oleic acid, as examined with thermal and X-ray diffraction
studies. Remarkable differences between SOS, SRS, and SLiS were observed
in the presence and absence of the more stable forms, while the common
polymorphic structures were maintained (Table 2.6). Namely, SRS has α, γ,
and β′ forms, but no β form, whereas β′ and β forms are absent in SLiS, which
has only α and γ forms. Thermal data for the polymorphic forms of SRS and
SLiS are provided in Table 2.6 together with those for tristearoyl-glycerol
(SSS) and SOS.
The α and γ forms of SRS and SLiS displayed the same molecular structures
as those in SOS. However, SRS had two β′ forms (β′2 and β′1) as the most stable
forms, and no transformation into β form occurred. The hydrogen bonding in
the ricinoleoyl chains in SRS was so tight that the O⊥ subcell was stabilized
through the carbonyl groups at the acyl chain, probably making β′ the most
stable form (Figure 2.19 a). This property makes the ΔH and ΔS values of
melting of the β′ forms of SRS much greater than those of the β′ forms of SSS
and SOS and even comparable to that of the β form of SSS (Table 2.6).
Figure 2.19 Structure models of: (a) the β′ form of SRS; (b) the α and γ forms of SLiS.
The interactions among the linoleoyl chains at the sn-2 position in SLiS may
stabilize the γ form, prohibiting transformation into the more stable forms β′ or
β. The transformation from γ to β′ or β is associated with the chain inclination
with respect to the lamellar interface, which may be prohibited by the chain–
chain interactions between the linoleoyl moiety with two cis double bonds
(Figure 2.19 b). For this reason, the enthalpy and entropy values for melting of
the γ form of SLiS are much greater than those of SOS and SRS.
Atomic-level structure analyses have not been successful despite the impor-
tance of diacid TAGs that are abundantly present in natural fats, except for the
two cases discussed below. It is worth noting that the structure models cited
above are only hypothetical and are based on powder X-ray diffraction and
infrared absorption data.
Crystal structures of β′ form
Determination of the atomic-level crystal structure using single crystals and
X-ray diffraction can provide the most precise structural information, although
infrared absorption, Raman scattering, solid-state NMR spectroscopy, and
X-ray diffraction studies using polycrystalline samples or oriented samples
also provide valuable information.
Figure 2.20 (a) Structure of the β form of tricaproyl-glycerol (CCC). (b) Structure of the β form of sn-
1,2-dipalmitoyl-3-acetoyl-glycerol (PP2). (c), (d), (e) Possible structure models for the β′ form of
triacylglycerols.
The first atomic-level structure determination was performed for the β forms
of tricaproyl-glycerol (CCC) (Jensen & Mabis, 1963, 1966) and trilauroyl-
glycerol (LLL) (Larsson, 1964), both of which form a double chain-length
structure (Figure 2.20a). Goto et al. (1992) found a new crystal structure with
a triple chain-length structure composed of palmitoyl–acetoyl–palmitoyl leaf-
lets for the β form of the diacid TAG PP2 (sn-1,2-dipalmitoyl-3-acetoyl-glycerol)
(Figure 2.20b). Furthermore, atomic-level analyses of β′ crystals have been
reported for two saturated diacid TAGs, CLC (C10C12C10; 1,3-didecanoyl-
2-dodecanoyl-glycerol) (van Langevelde et al., 2000) and PP14 (1,2-palmitoyl-
3-myristoyl-sn-glycerol) (Sato et al., 2001). Surprisingly, the two β′ crystals
exhibited remarkably different molecular features, probably because CLC is a
symmetric-type diacid TAG and PP14 is an asymmetric-type diacid TAG.
Many researchers have argued about β′ structure, and proposed structure
models are summarized in Figures 2.20c–e (Hernqvist & Larsson, 1982;
Hernqvist, 1988; Birker et al., 1991; Birker & Blonk, 1993; van Langevelde et
al., 1999b; van de Streek et al., 1999). Distinctive differences among the three
models appear in the chain inclination and methyl-end stacking. Figure 2.20c
indicates that all the extended straight chains involving the glycerol groups are
synchronously inclined with respect to the lamella plane. In fact, this structure
was observed in CCC and LLL. In contrast, the aliphatic chains are alternately
inclined against the lamella plane at the methyl end (Figure 2.20d) or at the
glycerol group (Figure 2.20e).
The β forms of CCC and PP2 in Figures 2.20a and 2.20b did not reveal a flat
methyl-end stacking mode arrangement; a zigzag stacking mode comprising a
terrace and step was revealed instead. The angle made by the terrace plane and
the lamella plane, defined as θ, was 12º in CCC and 11º in PP2. The chain
Figure 2.21 Crystal structures of: (a) the β′ form of CLC; (b) the β′2 form of PP14.
inclination structure depicted in Figure 2.20d was observed in the β′2 form of
PP14, whereas the β′ form of CLC displayed the structure depicted in
Figure 2.20e. Interestingly, the formation of each chain inclination is closely
related to the unique pattern of the methyl-end stacking mode.
CLC is a CnCn+2Cn-type TAG in which β′ is the most stable polymorph, as
described above. Melt-grown single crystals were analysed with a synchrotron
radiation X-ray beam to uncover an orthorhombic structure with a space group
of Iba2 with the unit cell lattice parameters and molecular structures depicted
in Figure 2.21a. The CLC molecules were packed in a chair conformation, in
which zigzag planes of the acyl chains were orthogonally packed and displayed
an O⊥ subcell structure.
The crystal structures of the β forms revealed in CCC and LLL and the β′
forms of CLC differ in the following three aspects, in addition to having
different subcell structures.
(1) The TAG molecules in the β form have an asymmetric tuning-fork
conformation, whereas the CLC molecules in the β′ form adopt a chair
conformation.
(2) The TAG molecules in the β form are all straight at the glycerol moieties,
whereas the CLC molecules in the β′ form are bent at the glycerol
moiety.
(3) The mode of methyl-end stacking is straight and flat.
Of the two stable β′ forms of PP14 (β′1 and β′2), a single crystal of β′2 was
crystallized from solution and its atomic-level structure was analysed
(Figure 2.21b). The primary structural properties of the β′2 form of PP14 can be
summarized as follows.
(1) A unit lamella revealed a quatro chain-length structure consisting of two
double-layer leaflets.
(2) The two double-layer leaflets were combined end-by-end in a unit
lamella, and the chain axes were alternately inclined against the lamellar
interface.
(3) The methyl-end stacking mode was largely different, a less stepped
structure at the outer plane (θ1 = 9.6º) and a very stepped structure at the
inner plane (θ2 = 38º).
(4) The two asymmetric units revealed different glycerol conformations,
trans for sn-1(P, palmitic) and sn-2(P) but gauche for sn-3(M, myristic)
in A, and trans for sn-2(P) and sn-3(M) but gauche for sn-1(P) in B
(Figure 2.22a).
(5) The outer methyl end consists of all the palmitic acid chains (–PPP–),
whereas the inner methyl end consists of palmitic–myristic–myristic
chains (–PMM–).
(6) Two asymmetric units, referred to as A and B, formed a hybrid-type
orthorhombic perpendicular subcell (HS3) (Figure 2.22 b), which was
different from the usual O⊥ type depicted in Figure 2.4 that was observed
for the β′ form of CLC.
The β′2 form of PP14 exhibited other unique properties of dynamic molecular
movements, expressed in the thermal movements of fatty acid chains of
different leaflets and the conformational stability of the glycerol group (Sato
et al., 2001), and the growth and equilibrium of the crystal morphology
(Hollander et al., 2003).
We briefly compared the two atomically analysed β′ forms. The CLC β′ form
displayed a bent glycerol conformation and straight methyl-end stacking,
whereas a straight glycerol conformation and stepped methyl-end stacking at
the inner interfaces of the four chain-length structures were revealed in the β′2
form of PP14. Arranging the lauric acid chain at the sn-2 position and the capric
acid chains at the sn-1 and sn-3 positions in the CLC β′ form may produce very
smooth methyl-end stacking. This would occur because the sn-2 chain, which
is longer than the sn-1 and sn-3 chains by two carbon atoms, does not form the
stepped methyl-end stacking mode that was revealed in the β′2 form of PP14,
C1 6 C14
C14
O
O
as
O
O O 3
O 1
2 O 2
3
1 O O
O O
bs
O
C16
C16 HS3
A C16 B
(a) asymmetric unit (b) subcell
Figure 2.22 (a) Asymmetric units, A and B, and (b) hybrid subcell of a PP14 β′2 crystal.
CCC, and PP2. However, the bent conformation at the glycerol group may be
accompanied by some excess lattice energy. We assume that the bent glycerol
conformation must have orthogonal-packed zigzag aliphatic chains, whereas
the parallel-packed chains in the β form do not require a bent glycerol
conformation. The uniqueness of the β′2 form of PP14 is revealed in the presence
of two asymmetric units in a unit cell, the chain-end stacking mode, the lateral
chain packing expressed in the HS3 subcell, the alternate chain inclination
against the lamella interface, and the straight glycerol conformation.
The fact that the β′ form is stabilized when a TAG contains different fatty
acid moieties (diacid TAGs), such as CLC and PP14, may indicate a general
tendency. Milk fat is a natural fat that is fairly stable in the β′ form; its stability
may be attributed to the presence of a high concentration of mixed-acid TAGs.
de Man (1999) indicated that the following factors are prerequisites for
stabilization of the β′ form, based on his observations: (1) fatty acid chain
length diversity, (2) TAG carbon number and diversity, (3) TAG structure, (4)
concentration of liquid oil and (5) temperature fluctuation. The first three
factors are of a molecular nature, which may be partially explained by
stabilization of the methyl-end stacking (as revealed in the β′2 form of PP14)
and/or bending at the glycerol group (as revealed in the β′ form of CLC).
Therefore, the following suggestions may conceivably explain a β′–β transfor-
mation, considering the β′ and β structures presently clarified.
The transformation of CLC from the β′ form causes conformational changes
in the glycerol groups, combined with a rotation of the half-leaflets between the
glycerol and methyl-end groups around the lamella plane normal, and conver-
sion of the subcell structure from O⊥ to T⏐⏐. The transformation of PP14 from the
β′2 form causes rotation of the double-layer leaflets I and II around the lamella
methyl end stacking
aliphatic chain packing
glycerol conformation
Figure 2.23 Molecular interactions operating in TAG crystals.
plane normal and conversion of the subcell structure from hybrid orthorhombic
to T⏐⏐. Therefore, stabilization of the β′ structure may be achieved by increasing
the activation energies necessary for all or some of the molecular movements
associated with the β′–β transformation described above. Clarification of these
processes is very important and further basic research is needed. Furthermore,
diversity in β′ structure should be elucidated; for example, atomic-level
understanding of the β′1 form of PP14 has not been achieved, although a
spectroscopic vibrational study revealed the presence of a O⊥ subcell (Yano et
al., 1997).
4. Molecular interactions and polymorphic structures

We can discuss the following three major molecular interactions that are most
influential in stabilizing the polymorphic structures of TAG crystals by consid-
ering the diversified polymorphic structures of monoacid and diacid-type
TAGs (Figure 2.23). The same arguments can also be applied to the poly-
morphism of fatty acids.
(1) Aliphatic chain packing results from the molecular interactions between
saturated and unsaturated fatty acid chains. It determines the subcell
structures, olefinic conformation, and chain length structure.
(2) Glycerol conformation among the glycerol groups determines the total
configuration (straight or bent) of the TAG molecule. Crystallographic
clarification of membrane lipid structures has revealed diverse varieties
of glycerol conformations of phospholipids and other polar lipids,
which are influenced by glycerol–glycerol interactions (Pascher et al.,
1981; Hauser et al., 1988; Pascher, 1996). This interaction is replaced by
carboxyl groups for fatty acids.
(3) Methyl-end stacking may play important roles in organizing the chain
inclination and chain-length structures. Methyl-end stacking in
combination with the glycerol conformation determines formation of
the triple chain-length structure.
These three factors interrelate in a complicated manner. For example,
optimum stabilization of the methyl-end stacking is related to the quatro chain-
length structure and the formation of HS3 subcell structures in the crystal
structure of the β′2 form of PP14. Stabilization of the stearic and oleic acid chains
causes a conversion from double to triple chain-length structures in SOS. The
same effect forms an interdigitated chain-length structure at the expense of
methyl-end stacking in the β1 form of oleic acid. It is very important to improve
our quantitative understanding of how the polymorphism of TAGs is controlled
by structural stabilization at the molecular level.
B. Phase behaviour of binary mixtures of triacylglycerols

Fats employed in manufacturing foods, cosmetics, and pharmaceuticals are
mixtures of different types of triacylglycerols (TAGs). The complicated behav-
iour of melting, crystallization and transformation, crystal morphology, and
aggregation of a fat is partly due to the physical properties of the crystals of the
TAG components that comprise the fat, and largely to the phase behaviour of
the mixture of different TAGs (Timms, 1984). The mixing phase behaviour of
the component TAGs of natural fats is very significant for texture formation of
the end products (Marangoni, 2005) and the fractionation efficiency of the
functional fat compositions derived from natural oils and fats, in particular for
confectionery fat blending (Timms, 2003). This section discusses the phase
behaviour of binary mixtures of the principal TAGs.
Three types of mixing phases – a solid solution phase, eutectic phase, or
molecular compound formation – can occur when two TAGs are miscible in all
proportions in a liquid phase, as illustrated in Figure 2.24 (Mullin, 2001). The
primary factors that determine the phase behaviour of a TAG mixture are their
molecular shapes and polymorphism, both of which are related in a somewhat
complicated manner. For example, miscible phases form in less stable α and β′
forms, whereas the eutectic phase forms between PPP and SSS (Rossel, 1967),
and between LLL and MMM (Takeuchi et al., 2003). Mixtures of POS–SOS
Figure 2.24 Typical three-phase diagrams of binary mixtures of two components, A and B: (a) solid
solution phase formation; (b) eutectic phase formation; (c) molecular compound (C) formation. S, solid;
L, liquid.
(Wille & Lutton, 1966; Rousset et al., 1998), and SSS–SSE (E; elaidoyl)
(Rossel, 1967; Timms, 1984) revealed eutectic and miscible properties in quite
complicated manners. However, this is not the case between PPP and LLL, in
which eutectic phases are formed in the three polymorphs (Takeuchi et al.,
2003). Mixtures of unsaturated TAGs and saturated TAGs exhibited eutectic
properties, but molecular compound formation occurred only in special combi-
nations of mixtures of saturated–unsaturated mixed-acid TAGs, as reviewed by
Sato et al. (1999). This section discusses new results regarding different types
of binary mixtures of TAGs.
1. Binary mixtures of saturated-acid triacylglycerols

Takeuchi et al. (2003) precisely analysed the phase behaviour of mixtures of
saturated monoacid TAGs in a time-resolved synchrotron radiation X-ray
diffraction (SR-XRD) study on mixtures of PPP–LLL, MMM–LLL, and SSS–
LLL. This work followed a pioneering SR-XRD study of phase behaviour of
the PPP–SSS mixture (Kellens et al., 1991).
Figure 2.25 depicts the binary phase behaviour of MMM–LLL and PPP–
LLL mixtures. Figure 2.25a provides a diagram of the polymorphic occurrence
for MMM–LLL mixtures obtained from DSC (differential scanning calorimetry)
and SR-XRD experiments. This diagram illustrates the following three points:
(1) β′ formed in the mixture system, whereas α transformed directly to β in pure
LLL and MMM; (2) miscible solid-solution phases formed in the metastable α
and β′ forms of the mixtures; and (3) a eutectic phase formed in the most stable
β form. These three results are consistent with the results from the PPP–SSS
system. In contrast, Figure 2.25b illustrates that a diagram of the polymorphic
occurrence for the LLL–PPP mixtures can be subdivided into the following
three regions: (1) the phase behaviour of the LLL–PPP mixtures was mainly
governed by LLL for LLL concentrations above 90%; (2) the β′–β transforma-
tion of the LLL fraction and the α–β transformation of the PPP fraction
Figure 2.25 Binary phase behaviour of (a) MMM–LLL and (b) PPP–LLL mixtures.
occurred separately for LLL concentrations between 50% and 90%, indicating
that phase separation occurred in the three polymorphic forms; (3) the LLL
fraction was dissolved in the PPP fraction for LLL concentrations below 50%.
Hence, the phase behaviours of the LLL–PPP mixtures were mainly governed
by PPP. An SSS–LLL mixture exhibited similar properties.
These results enable us to draw the following conclusions. The metastable
forms have miscible phases when the carbon numbers (Cn) of the fatty acids of
the component TAGs differ by two, but the most stable form has a eutectic
phase. Neither TAG is miscible in any polymorphic form when Cn differs by
four or six.
2. Binary mixtures of saturated–unsaturated mixed-acid triacylglycerols

Neither miscible nor molecular compound-forming systems were observed in
mixtures of saturated monoacid TAGs and unsaturated monoacid TAGs.
However, formation of molecular compound crystals has recently been
observed in specific binary mixture systems of saturated–unsaturated mixed-
acid TAGs such as SOS–SSO (1,2-distearoyl-3-oleoyl-rac-glycerol) (Engstrom,
1992; Takeuchi et al., 2002b), SOS–OSO (1,3-dioleoyl-2-stearoyl-sn-glyc-
erol) (Koyano et al., 1992), POP–PPO (1,2-dipalmitoyl-3-oleoyl-rac-glycerol)
(Minato et al., 1997a), and POP–OPO (1,3-dioleoyl-2-palmitoyl-sn-glycerol)
(Minato et al., 1997b). In a mixture of a saturated monoacid TAG (PPP) and a
saturated–unsaturated mixed-acid TAG (POP), molecular compound was not
formed (Minato et al., 1996). Infrared absorption spectroscopic analysis
indicated that steric hindrance due to repulsive interactions between saturated
and oleic acid moieties operates in the formation of a molecular compound
crystal (Minato et al., 1997c).
Figure 2.26 compares phase diagrams of binary mixtures of PPP–POP and
POP–OPO. Neither miscible nor molecular compound formation phases formed
in the PPP–POP mixture, whereas eutectic mixtures formed in all the poly-
morphic structures of the PPP–OPO mixture. Thermal, X-ray diffraction, and
infrared absorption studies clarified the following properties of the molecular
compound structures revealed in the binary mixtures described above.
(1) Molecular compounds formed at 1:1 concentration ratios in mixtures of

POP–PPO, POP–OPO, SOS–SSO, and SOS–OSO, both in the stable
and metastable states. Immiscible eutectic or monotectic phases formed
in the most stable forms between the component materials and the
molecular compounds.
(2) The molecular compounds were of double chain length, while the stable
forms of the component TAGs were all of triple chain length. This
conversion in the chain-length structure is primarily caused by molecu-
lar interactions through oleic acid chains, which were packed in the
same leaflets in the double layers in a molecular compound crystal, as
depicted in Figure 2.27.
(3) The subcell structures of the molecular compound crystals transformed
from H (αC) to T⏐⏐ (βC) through O⊥ (β′C), in which C refers to the
molecular compound crystal.
SOLID-STATE BEHAVIOUR OF POLYMORPHIC FATS AND FATTY ACIDS
Figure 2.26 Phase diagrams of binary mixtures of (a) PPP–POP and (b) POP–OPO.
45
SOS (β) SSO (β′) OSO (β) SOS–SSO SOS–OSO

compound (β) compound (β)
SOS SOS
SSO OSO
Figure 2.27 Structure models of stable polymorphs of SOS, SSO, and OSO and the molecular
compound crystals of SOS–SSO and SOS–OSO.
(4) Infrared spectroscopic analysis of the molecular conformation of the

oleic acid chains in the βC form indicated that the olefinic group of the βC
form of SOS–SSO and POP–PPO mixtures exhibited neither a S-C-S′
type nor a S-C-S type conformation. However, the βC form of the POP–
OPO mixture was revealed to be a S-C-S type.
Interestingly, miscible mixture phases formed between SOS and SLiS,
whose polymorphic properties are isomorphic. Figure 2.28a depicts the phase
behaviour of α and γ forms observed in SOS–SLiS mixtures in thermal and
synchrotron radiation X-ray diffraction studies (Takeuchi et al., 2002a). This
diagram illustrates the following two points. (1) Miscible phases formed for α
and γ forms of the SOS–SLiS mixtures over the entire concentration range
studied. (2) Although SOS has β′, β2 and β1 as the more stable forms,
transformation into these forms in the binary mixtures was not observed in
simple cooling of the mixture from an elevated temperature. Furthermore, α-
melt-mediation into β′ and β2 occurred through rapid temperature variation,
leading to the immiscible phase for a SOS concentration above 30%.
These results indicate the strong influences of chain–chain interactions
between SOS and SLiS, which are highly polymorph-dependent. Figure 2.28b
provides structural models of the miscible phases of α and γ forms of the SOS–
SLiS mixtures. Oleoyl and linoleoyl chains in the oleic/linoleic acid leaflet may
coexist in the α form of a double chain-length structure and in the γ form of a
triple chain-length structure due to olefinic interactions between the oleoyl and
linoleoyl chains. It is notable that SLiS is most stable in the γ form because of
stabilization of the linoleoyl chain leaflet in the triple chain-length structure.
Therefore, the disordered conformation of the linoleoyl chains of the SLiS
fraction may cause the SOS fraction not to separate from the miscible phase,
and the transformation into β′ or β forms is retarded in a solid state. The phase
separation into β′ and β2 can only occur by melt-mediation.
In summary, the following properties have been observed in mixtures of
saturated–unsaturated mixed-acid TAGs. (1) Eutectic mixtures are formed
(a) (b)
40 α SOS/SLiS
liquid
Temperature (°C)
30
γ
γ SOS/SLiS
20
α
0 20 40 60 80 100
SLiS concentration (%)
Figure 2.28 (a) Phase behaviour of the SOS–SLiS mixture. (b) Structure model of the co-crystal of
SOS and SLiS.
between saturated-acid TAGs and saturated–unsaturated mixed-acid TAGs.

(2) Miscible mixtures form between isomorphic polymorphs. (3) Molecular
compound crystals form between specific TAGs through aliphatic interactions.
These results represent valuable indications for applications, including the
fractionation of saturated and unsaturated TAGs, and the blending of fats to
modify polymorphic transformations and crystallization, as well as to create
new fat crystals using molecular compounds.
C. Polymorphic crystallization of fats in complex fluid systems

This section discusses the polymorphic crystallization behaviour of TAGs in
bulk liquid, in o/w emulsion droplets, and in a gel state. The crystallization
processes in a bulk liquid are somewhat simpler than those that occur in
complex fluid systems such as emulsion and gel states, and are basically
influenced by the polymorphic structures, thermodynamic stability, and kinetic
factors. We first discuss the crystallization processes in bulk liquid with and
without external factors. More complicated crystallization phenomena in the
emulsion and gel states will then be examined.
1. Polymorphic crystallization from bulk liquid

Thermodynamic factors and interfacial energy factors
Polymorphic nucleation is primarily governed by two competitive factors,
interfacial energy and chemical potential differences, both of which depend
significantly on the polymorphic modifications of the TAG under considera-

tion (Aquilano & Squaldino, 2001). A classic equation for the homogeneous
nucleation rate (J) can be written as:
⎡ – fν2γ3 ⎤
⎢ –––––––– ⎥
⎣ kBT(Δμ)2 ⎦
J = Ae (1)
fν2γ3
ΔG# = –––––2 (2)
(Δμ)
where A is a pre-exponential factor involving the frequency of molecules to be

incorporated into growing nuclei, f is the geometrical factor of the critical
nucleus, γ is the interfacial free energy, ν is the volume per molecule, Δμ is the
chemical potential difference, T is the absolute temperature, k B is the Boltzmann
constant, and ΔG# is the activation energy for nucleation. The predominant
factor in the nucleation rate function is in exponential terms, except for
nucleation in high-density liquids, where volume diffusion involved in the A
term becomes more dominant. Δμ is expressed for nucleation from a super-
cooled liquid as follows:
ΔHf ΔT
Δμ = –––––– = ΔSf ΔT (3)
Tf
where ΔHf is the heat of fusion, ΔSf is the entropy of fusion, Tf is the melting
temperature, ΔT is supercooling, defined as Tf – Tc, and Tc is the crystallization
temperature.
All of the exponential factors that appear in Equations 1–3 are polymorph-
dependent except for Tc. Δμ is always higher for the more stable form than the
less stable ones, except for a specific condition in which the G values of the two
forms are the same (see Figure 2.2).
We consider a monotropic case (Figure 2.2c) for simplicity, with two forms,
A (metastable) and B (stable). We assume that pre-exponential factor A is
common to the two polymorphs, and the differences in f and ν are negligible for
A and B. The ΔG# ratio of the two forms then equals:
ΔG#A ⎛ γA ⎞ 3 ⎛ ΔSfBΔTB ⎞ 2
–––– = ⎜ –– ⎟ ⎜ ––––––– ⎟ (4)
ΔG#B ⎝ γB ⎠ ⎝ ΔSfAΔTA ⎠
where the superscripts of γA and γB, etc. indicate the values of interfacial energy
and other factors for the A and B forms. The term [(ΔSfBΔTB)/(ΔSfAΔTA)]2 in
Equation 4 is always greater than 1 since metastable A forms have lower
melting points and smaller ΔSf values than the stable B form. This gives rise to
a significantly higher activation energy for nucleation for A compared with B
Figure 2.29 Schematic illustration of activation energy for nucleation (ΔG#) of two crystal forms
(metastable A and stable B) with varying supercooling (ΔT).
if the interfacial energy values are the same for the two forms. However, the
ratio of the ΔG# values of A and B is determined by conflicting factors of
supercooling and interfacial energy if we reasonably suppose that the γ value of
the low-enthalpy form B is greater than that of the high-enthalpy form A.
Two extreme cases are worthy discussion topics: high supercooling with low
Tc, and low supercooling with high Tc. The ratio of (ΔSfBΔTB)/(ΔSfAΔTA)
approaches 1 in the former and form A is preferably nucleated since its
activation energy for nucleation becomes smaller than that of B. In contrast, the
ratio of (ΔSfBΔTB)/(ΔSfAΔTA) becomes smaller than 1 when supercooling is
decreased, resulting in a smaller ΔG# value and a greater nucleation rate for the
metastable A form than the stable B form. This argument is illustrated in
Figure 2.29.
Figure 2.30 provides an example of polymorph-dependent nucleation of a
TAG. This figure depicts the small-angle X-ray diffraction patterns of the β′
and β forms of LLL, which were monitored by a synchrotron radiation X-ray
beam at isothermal crystallization temperature (Tc) (Ueno et al., 2003b). A
small-angle X-ray diffraction line of 3.35 nm, corresponding to presence of the
β′ form, appeared earlier at Tc = 30ºC than that of the β form (3.25 nm). The
induction time, τ, was obtained by measuring the time interval after the
temperature of LLL reached Tc until the small-angle diffraction lines of the two
forms were detectable. The τ values thus obtained at Tc = 25.0ºC and 30.0ºC are
provided in Table 2.7, together with the values of Tf and ΔSf.
Figure 2.30 Synchrotron radiation small-angle X-ray diffraction (SR-XRD) of polymorphic crystalli-
zation of the β′ and β forms of trilaurin, taken during isothermal crystallization at Tc = 30ºC: (a) without
ultrasound stimulation; (b) with ultrasound stimulation. Details of first-occurring diffraction patterns are
inserted.
The τ value of the β′ form was always smaller than that of the β form. The
results in Table 2.7 indicate that ΔG# was always smaller for the β′ form than for
the β form during crystallization at 25.0ºC and 30.0ºC, since τ is inversely
proportional to the nucleation rate function (J–1). We can use the values of ΔT
and ΔSf in Table 2.7 to calculate that (ΔSfβΔTβ)/(ΔSfβ′ΔTβ′) is 3.2 at Tc = 25.0ºC
and 5.6 at Tc = 30.0ºC. The fact that (ΔSfβΔTβ)/(ΔSfβ′ΔTβ′) is greater than 1 at the
two crystallization temperatures indicates that the thermodynamic driving
force for nucleation is greater for the β form than for the β′ form. This contrasts
with the experimental result and suggests that the smaller interfacial energy for
the β′ form may exceed the thermodynamic driving force for nucleation, as
illustrated in Figure 2.29. A similar observation was made for POP, SOS
(Koyano et al., 1989) and PPP (Sato & Kuroda, 1987).
Table 2.7 Thermal data and crystallization properties of β′ and β polymorphs of trilaurin without and with ultrasound stimulationa
Tc = 30.0ºC Tc = 25.0ºC
Polymorph Tf (ºC) Δ Sf (J/mol•K) Δ T (ºC) t (s) Δ T (ºC) t (s)
β′ 34.0 264.5 4.0 1820 (No ultrasound) 9.0 60 (No ultrasound)

–b (2 s ultrasound) 40 (2 s ultrasound)
β 46.5 357.8 16.5 2820 (No ultrasound) 21.5 300 (No ultrasound)
30 (2 s ultrasound) 100 (2 s ultrasound)
a
Data taken from Ueno et al., 2003b.
b
Crystallization occurred soon after the temperature reached Tc.
SOLID-STATE BEHAVIOUR OF POLYMORPHIC FATS AND FATTY ACIDS
51
2. Effects of external factors on polymorphic crystallization

The above argument addresses the nucleation process of crystals without
external factors. However, various external factors are applied incidentally or
purposely in actual fat crystallization processes, and the polymorphic nuclea-
tion processes are modified by the external factors so that the most favourable
polymorphic forms can be crystallized. The effects of three external factors on
crystallization processes are briefly discussed in this section.
Ultrasound stimulation
Ultrasound at high power and low frequencies can generally assist various
processes in food technology (Povey & Mason, 1998). It has been reported that
high-power ultrasound stimulation remarkably influences both nucleation and
crystal growth by creating additional fresh nucleation sites in the crystalliza-
tion medium (Eskin, 1994). Ultrasound stimulation for a short period at
supercooled conditions accelerated the crystallization of Form V in cocoa
butter (Baxter et al., 1995). We discuss here the effects of ultrasound stimula-
tion on the polymorphic crystallization of fats, which was recently studied for
triacylglycerols (Ueno et al., 2003a,b), cocoa butter (Higaki et al., 2001) and
palm oil (Patrick et al., 2004).
The crystallization behaviour of PPP and LLL in the presence of ultrasound
stimulation was investigated in situ using time-resolved SR-XRD measure-
ments. The results for LLL with ultrasound stimulation at an exposure time of
2 seconds with a power of 100 W are provided in Figure 2.30b and Table 2.7,
together with those without ultrasound stimulation. Figure 2.30b shows a
marked decrease in the induction times (increased nucleation rates) for crystal-
lization of both PPP and LLL, and indicates that crystallization of the most
stable β form was promoted rather than the β′ form as Tc increased (supercool-
ing decreased). However, ultrasound stimulation with a longer exposure time
delayed the crystallization rate due to the rise in sample temperature caused by
the absorption of ultrasound dissipation energy. Similar results were observed
for PPP. This conflicting behaviour indicates that there is an optimal condition
for the acceleration of crystallization by ultrasound stimulation. A pronounced
decrease in the induction time for nucleation results from the melting point shift
due to high-pressure pulses associated with collapsing bubbles. The mecha-
nism that clarifies the preferred nucleation of the β polymorph over the β′ form
may involve conflicting factors, such as the effects of pressure on the melting
temperature of the two polymorphs, which prevailed for the β′ form, and the
effects of rising temperature on stabilization of the crystal nuclei, which
prevailed for nucleation of the β form, but the details are unclear. Thus, details
of the crystallization processes of fats subjected to ultrasound stimulation are
open to interpretation.
Figure 2.31 Time variation of the diffraction intensity of SR-XRD small-angle peaks of polymorphic
forms of cocoa butter. Open symbols, without shear; closed symbols, with shear.
Shear field
Shear is applied incidentally or purposely during the crystallization procedure
in the actual production processes of fat-based products. The effect of applying
a shear field on fat crystallization was first examined in cocoa butter using a
viscometer and DSC under varying shear rates (Feuge et al., 1962). The
polymorphic forms of crystallized cocoa butter (CB) converted from metastable
to more stable forms with an increased shearing rate, and the crystallization rate
was increased (Ziegleder, 1985). Stapley et al. (1999) examined the effect of
shear on the solidification of chocolate. Recent studies using SR-XRD have
demonstrated that transformations from metastable to more stable forms,
particularly to Form V, are accelerated by high shear stress (MacMillan et al.,
2002; Mazzanti et al., 2003, 2004).
Figure 2.31 depicts the time variation of relative intensities of X-ray diffrac-
tion peaks of CB crystals formed after cooling from 50ºC to 18ºC at a rate of
3ºC/minute. Form III appeared first without shear, after which Form IV crys-
tallized at the expense of Form III. In contrast, an accelerated transformation
from Form III to Form V was caused by applying shear stress at 1440 s–1,
without the occurrence of Form IV. The same result was observed with lower
shear rates, and the persistence time of Form III was reduced as the shear rate
was increased. Mazzanti et al. (2003) also observed that the CB crystals were
aligned and that a new phase was created in the shear field. Similar observa-
tions were obtained for milk fat and palm oil.
The effects of the shear field on fat crystallization are considered to involve
changes in the crystallization pathways by means of modification of the mass
transfer conditions to the crystal surface via the boundary layer, segregation of
individual crystallites or some hindering of cluster formation, improved heat
transfer, and increased crystallite collisions. However, further research is
needed to obtain deeper insight into the effects of shear field on fat crystalliza-
tion.
Crystal seeding effect

Adding crystals (‘seeding’) is the optimal technique to control the polymorphic
crystallization of fats. Specific relationships between the crystal seed materials
and the polymorphic forms of certain fats are necessary for the seeding
technique to be functional. These relationships include polymorphic corre-
spondence, aliphatic chain matching, molecular shape, and thermal stability.
Polymorphic correspondence means that the seed material exhibits the same
polymorphic forms as those of the crystallizing materials, mostly β′ or β.
Aliphatic chain matching involves two characteristics, chain length and the
chemical nature of the fatty acid moiety. Differences in the chain length (nc)
between the seed and crystallizing materials are not expected to exceed four.
The molecular shape of the fatty acid moiety relates to its saturation or
unsaturation. For example, seed material without an unsaturated fatty acid
moiety is less effective when saturated–unsaturated mixed-acid TAGs are the
crystallizing material, as observed in the crystallization of CB (Hachiya et al.,
1989a,b). Finally, thermal stability indicates that the seed material will not melt
or dissolve in the liquid phase of the crystallizing materials at the seeding
temperature. Therefore, the melting point of the seed materials must be much
higher than that of the crystallizing material.
A good example of the crystal seeding effects of a high-melting-temperature
fat was observed using BOB β2 crystals to seed crystallization of Form V of
cocoa butter (Hachiya et al., 1989a,b).
3. Fat crystallization in emulsion droplets

As mentioned in Section A.2, the crystallization of fats in o/w emulsion
droplets determines their physical properties, such as stability, whippability,
rheology, and texture. Therefore, controlling fat crystallization is a key process
in various colloid technologies in which an emulsion is subjected to thermal
thaw, causing crystallization and melting of the dispersed oil phase. Crystalli-
zation in an o/w emulsion contributes to the de-emulsification process in the
food industry for whipped cream, the freezing of ice cream, and the coagulation
of oil droplets in chilled mayonnaise.
Numerous studies have been performed to clarify the mechanisms of fat
crystallization in o/w emulsions, as reviewed by Povey (2001) and Coupland
(2002). Many researchers have studied the complexity of fat crystallization in
o/w emulsions, which are affected by many factors including emulsion droplet
(a) Interfacial (b) Volume (c) Inter-droplet

heterogeneous heterogeneous heterogeneous
Figure 2.32 Three types of heterogeneous nucleation occurring in o/w emulsion droplets.
size (Dickinson et al., 1991), the type of emulsifier (McClements et al., 1993;
Palanuwech & Coupland, 2003), droplet–droplet interactions (Hindle et al.,
2000), polymorphism (Ueno et al., 2003a), the effects of additives (Katsuragi
et al., 2001), cooling rate (Lopez et al., 2002), and temperature variation
(Vanapalli et al., 2002). This section briefly discusses the basic properties of
the crystallization processes of fat crystals in o/w emulsion droplets.
Nucleation in emulsion: homogeneous versus heterogeneous

Crystallization in an emulsion occurs by nucleation and growth of crystals that
consist of the materials dissolved in the droplets. Nucleation in pure liquid in
emulsion droplets becomes homogeneous under two conditions: when the
crystalline embryos are entirely immersed in the droplets, and when the small
size of the droplets induces changes in the thermodynamic properties of the
embryos (Mutaftschiev, 2001). Nucleation becomes heterogeneous if these
conditions are not satisfied, and almost all events of fat crystallization in o/w
emulsion droplets are initiated by heterogeneous nucleation. In particular,
Krog & Larsson (1992) discussed how crystallization in an emulsion is
heterogeneously induced by the membrane that forms the emulsion droplets.
First, it is difficult to avoid van der Waals interactions between crystallizing
fat molecules and hydrophobic moieties of emulsifiers or proteins that consti-
tute the interface membrane of the emulsion droplets. Such nucleation may be
called interfacial heterogeneous nucleation. Second, catalytic foreign
materials present in liquid, i.e. when the droplets contain foreign molecules
that behave as nucleation-catalytic centres, lead to heterogeneous nucleation.
The dispersed phase is divided into a number of droplets in an emulsion system.
Nucleation-catalytic materials (hereafter referred to as impurities) are distrib-
uted unequally throughout some of the droplets when the size distribution of
the droplets is wide (poly-dispersed droplets). Larger droplets most likely have
higher impurity concentrations. We can therefore reasonably assume that
nucleation assisted by an impurity is more pronounced in larger droplets than
in smaller ones through volume heterogeneous nucleation.
Third, droplets move rapidly and frequently collide because of their Brownian
(a)
1.2
1.0
0.8
Z V/Z V, m
0.6
0.4
0.2
0
0 0.2 0.4 0.6 0.8 1.0 1.2
r (µm)
(b)
1440
USV (m/sec)
1430
1420
1410
0 400 800 1200
t (min)
Figure 2.33 (a) Size distribution of palm oil-in-water emulsion droplets. Bars indicate the actual size
distribution. ZV and ZV, m are the size distribution function and its average value, respectively, and r is the
diameter of the emulsion droplets. (b) Change in ultrasound velocity (USV) with time during crystal
nucleation and growth. Closed circles are experimental data. Solid and dotted curves are calculated based
on poly-dispersed and mono-dispersed emulsions, respectively.
motion and gravity in the emulsion state. Thermal movement and droplet–
droplet interactions cause various destabilization mechanisms, in which
crystallization due to droplet–droplet interactions plays an important role. The
already-crystallized droplets that collide with other droplets containing super-
cooled liquid cause inter-droplet nucleation, provided that the crystal phase
comes into contact with the supercooled liquid. Figure 2.32 depicts the three
types of heterogeneous nucleation.
Effects of droplet size distribution

Crystallization in emulsion droplets is also influenced by the size distribution
of the droplets. The emulsion is referred to as mono-dispersed if all the droplets
in the emulsion are of the same size. The emulsion is referred to as poly-
dispersed if this is not the case. Most emulsions are poly-dispersed, primarily
because widely employed emulsification techniques form poly-dispersed emul-
sion droplets. Mono-dispersed emulsion droplets are formed with special
techniques using porous membranes or microchannels. It is interesting to
compare the crystallization rate of a poly-dispersed emulsion with that of a
mono-dispersed emulsion.
An analysis of the kinetics of isothermal crystallization of droplets in poly-
dispersed emulsions was performed under conditions in which each emulsion
droplet produced one nucleus only (mono-nucleus model) (Kashchiev et al.,
1998). Expressions for the time dependences of the number of crystallized
droplets and the fraction of crystallized droplet volume were derived and
compared in experiments conducted using an ultrasound velocity (USV)
technique. The USV through o/w emulsions is proportional to the extent of
crystallization of the oil droplets. For example, when an emulsion is kept at a
constant temperature far below the melting point of the droplet material,
progressive crystallization of the droplets in the emulsion brings about a
gradual change of USV in accordance with the nucleation and crystal growth
of the materials in the droplets. We can relate the USV to the percent
of crystallized droplets, based on the USV values for the solid and liquid
of the dispersed phase and for the continuous water phase at different
temperatures.
Figure 2.33a depicts the size distribution of poly-dispersed emulsion drop-
lets of palm oil in the water phase and measurement of USV values at a constant
temperature (10ºC). The emulsification conditions were Tween 20, distilled
water, and an oil:water ratio of 20:80. The melting temperature of palm oil is
around 30ºC; the USV value of the emulsion measured at 10ºC increased with
time due to nucleation and crystal growth of the palm oil, as indicated in
Figure 2.33b. The evolution of the USV value with time best fit the calculation
of poly-dispersed emulsions (solid curve) rather than mono-dispersed emul-
sions (dotted curve). Poly-dispersed n-hexadecane-in-water emulsions were
analysed similarly.
Figure 2.34 (a) Molecular structure of polyglycerol fatty acid esters (PGFEs) with 10 glycerol groups.
R, fatty acid moiety. (b) Temperature variations of the ultrasonic velocity (USV) values of PMF–water
emulsions without (‘pure’) and with the addition of PGFEs with 3, 6, 9 or 12 esterified stearic acid
moieties. The additive concentration was 1% (w/w) with respect to the PMF.
Heterogeneous nucleation affected by additives

Fat crystallization in an emulsion can be modified when certain additives are
included in the oil phase, using the interfacial heterogeneous nucleation that
plays a greater role in o/w emulsion systems than in a bulk system. Additives,
which can promote interfacial heterogeneous nucleation, possess the chemical
and physical properties of emulsifiers with a more hydrophobic nature and
higher melting point, mostly due to the presence of a saturated fatty acid
moiety. The effects of additives on fat crystallization have been observed as an
increase in nucleation rate, the occurrence of new polymorphic structures, and
a change in crystal morphology in the emulsion droplets. An increased nuclea-
tion rate produced by an additive may be interpreted as a reduction in interfacial
energy (see Equation 2) due to the interactions between the additive and fat
molecules. The additives so far examined are sucrose fatty acid oligoesters,
diacylglycerols, and polyglycerol fatty acid esters. This section discusses the
crystallization processes of a palm mid fraction (PMF)-in-water emulsion with

emulsifier additives of polyglycerol fatty acid esters (PGFEs) (Sakamoto et al.,
2004).
PMF is a solid fraction of palm oil with a medium range of melting
temperatures. Although PMF is a multi-component fat, it exhibits polymor-
phism, and the melting points for the different forms are α (13.0ºC), β′
(22.5ºC), and β (35.0ºC). A PMF-in-water emulsion with an average droplet
size of 1.35 μm was prepared using distilled water and Tween 20 emulsifier
mixed with PMF in a water:oil ratio of 80:20. Figure 2.34a depicts the PGFEs
with a large hydrophilic moiety composed of 10 glycerol units; the saturated
fatty acids are esterified with different degrees of esterification. The PGFEs
were added to PMF prior to emulsification. Figure 2.34b illustrates that the
crystallization of PMF markedly increased as the added PGFE became more
hydrophobic, in accordance with an increasing degree of esterification with
stearic acid from 3 to 12, as indicated by the increase in the USV values at all
temperatures. The addition of PGFE at an esterification degree of 12 (indicat-
ing that all 12 hydroxy groups of the 10 polymerized glycerol groups were
esterified with stearic acid) in particular resulted in the greatest USV values.
We also noted that the USV values increased with increasing concentration of
the PGFE additive and with increasing chain length of the saturated fatty acids.
It is essential to observe the effects of hydrophobic PGFE additives on the
nucleation process and on the crystal growth process separately in order to
understand their role in accelerating the crystallization of PMF in an emulsion
system, since additives may interact with the crystallizing materials in different
ways during crystal nucleation and crystal growth. The addition of PGFEs
retarded the crystal growth rate of PMF in the bulk state. This retardation of
crystal growth may also occur during crystallization of the oil phase in an
emulsion system. Therefore, the acceleration of PMF crystallization by PGFEs
in o/w emulsions is attributed to effects on the nucleation process, not on the
crystal growth process.
The effects of additives on the nucleation kinetics in an emulsion can be
considered using interface heterogeneous nucleation and volume heterogene-
ous nucleation, as illustrated in Figure 2.35 (Sakamoto et al., 2004). Interface
heterogeneous nucleation occurs at the oil–water interface, where the additive
molecules are adsorbed and crystallized during cooling because of their high
melting point. Freezing of the fatty acid chains of the adsorbed emulsifier
membranes can act as a template for nucleation of the fat molecules in the oil
phase, which may cooperatively interact with the template films and start to
nucleate. Volume heterogeneous nucleation is nucleation accelerated by tem-
plates that form within the oil phase of the emulsion. The facts that no PGFEs
with short chain-length fatty acid moieties accelerated the nucleation of PMF
crystals, and that fully esterified PGFEs exhibited acceleration effects, indicate
that interfacial heterogeneous nucleation and volume heterogeneous
Water phase
Interface
heterogeneous nucleation
fat
crystals
Volume
Oil phase heterogeneous nucleation
Template of
Tween20 additive
additive
Figure 2.35 Schematic illustration of two types of heterogeneous nucleation processes caused by
PGFE additives.
nucleation may occur in a nonlinear way when the degree of esterification of

decaglycerol fatty acid esters is increased from 3, 6, or 9 (interfacial heteroge-
neous) to 12 (interfacial and volume heterogeneous).
4. Crystallization in a gel phase

We discuss here crystallization in the gel state, in which small amounts of lipid
crystals form a three-dimensional network that entraps liquid fractions of water
or oil phases. Morphology, size, density, and the crystal network of lipid
crystals are the dominating factors that influence the physical properties of the
gel state (Gallegos & Franco, 1999).
Hydroxyl-stearic acid forms a gel state due to strong inter-molecular hydro-
gen binding; however, this gel material cannot be applied to the manufacture of
edible products (Hermansson, 1999). Edible lipid gels are hydrocolloid gels
made of monoacylglycerols and a mixture of high-melting and low-melting fats
(Heertje & Leunis, 1997).
Gel of a polar lipid–water mixture

In a gel phase made of polar lipid emulsifiers and water, a lamellar-type
lyotropic liquid crystal (LC) phase is formed at elevated temperatures. In the
LC phase, the water phase is swelled into a continuous lamellar LC phase.
Figure 2.36 Gelation mechanism of a monoacylglycerol–water mixture.
When this LC phase is cooled to undergo the transformation from LC to

crystalline phases, the lamellar structure involving the swollen water phase is
maintained, forming a highly viscous phase called an α-gel phase Further
cooling forms a rheologically hard phase, which is employed in low-fat table
spreads (Chronakis, 1997).
Monoacylglycerols (MAGs) are polar lipid emulsifiers used in food prod-
ucts. Thermal treatment of a mixture of MAG and water can form a gel state that
is used for spreadable fat products, as illustrated in Figure 2.36. MAGs are first
transformed into a lamellar phase by heating the MAG–water mixture above
the Krafft point (the melting temperature of an amphiphilic compound in
water). Hydration of the MAG is undertaken in this process to form a swollen,
space-filling system of lamellar liquid-crystalline phase. The gel state is
formed upon cooling by a crystal network of plate-like crystals of MAGs that
grow from lamellar structures, which represent the template for crystallization.
The crystallization of a MAG in water involves polymorphic transformation
of the MAG from the lamellar liquid-crystalline state (Lα) through α-gel to the
gel state, in which the MAG molecules form plate-like β-crystals in water. The
following gelation mechanism for the transformation from liquid-crystalline to
the gel state has been proposed: (1) rapid crystallization in a lateral direction
occurs after formation of a nucleus of the MAG crystal, forming the first space-
filling network; (2) this is followed by reinforcement of the network, through
which stacks of crystalline bilayers are formed.
Gel of high-melting fat and low-melting fat mixtures

We have recently reported that a gel state occurs in mixtures of high-melting
(HM) fat and low-melting (LM) fat when special thermal treatments are applied
to the mixture systems, without any other ingredients such as emulsifiers
(Higaki et al., 2003, 2004a, 2004b). The HM fat was fully hydrogenated
Scale bar, 20µm

Temperature
β A: sol
Tf
α
Tc
Time B: gel
Figure 2.37 Thermal treatment to form a β-fat gel.
Figure 2.38 Formation of a β-fat gel from a binary mixture of FHR-B and sal fat olein.
Figure 2.39 Phase behaviour of binary mixtures of SFO and FHR-B (α, β′, and β forms).
rapeseed oil containing a substantial amount of behenic acid (called FHR-B);

the LM fats were sal fat olein (SFO, a low-melting fraction of sal fat) and cocoa
butter (CB). The minimum concentration of the HM fat to produce the gel state
was 1.5% (w/w). Figure 2.37 shows the thermal treatment to form the gel state,
in which a crystal network of β polymorphs of the HM fat formed after melt-
mediated transformation at the final temperature (Tf, around 38ºC) from the α
form of the HM fat, which was formed from the liquid mixture by rapid cooling
at the crystallization temperature (Tc, below 20ºC). The fat mixture became a
sol state without the thermal treatment. This gel is known as β-fat gel.
Optical observations, DSC and X-ray diffraction measurements, and
rheological studies have clarified the following formation mechanism of a β-fat
gel, as illustrated in Figure 2.38.
(1) The β crystals of FHR-B displayed unique microstructures, such as large

numbers, very small size, and uniform distribution. These microstruc-
tures were observed at high cooling rates during the first crystallization,
but not during the slow cooling process, since α crystals of FHR-B were
formed by rapid cooling and transformed to the β crystals during the
thermal treatment. The morphological properties of the FHR-B crystals
in particular play decisive roles in the display of gel-like behaviour;
other fats, such as SSS, BBB, or fully hydrogenated rapeseed oil rich in
stearic acid, did not exhibit this property.
(2) Fine β crystals form the crystal network, incorporating the liquid phase
of LM fats. In contrast, β crystals formed by thermal treatment with slow
cooling, or by rapid cooling to a high Tc, were few in number, much
larger, and were precipitated as aggregates. Therefore, these β crystals
do not form a gel state.
(3) The phase behaviour of binary mixtures of SFO and FHR-B with α, β′,
and β polymorphs indicates that rapid cooling can result in the α form
and that melt mediation into the β form occurs when Tf is set between the
melting points of the α and β forms, as illustrated in Figure 2.39.
D. Conclusions
This chapter discussed the solid-state behaviour of fats and fatty acids, begin-
ning with the fundamental aspects of the structures and transformations of
crystalline materials and ending with crystallization in complex fluid
systems.
Future research areas that may interest scientists and technologists involved
with the main subjects of this chapter include structural analysis of fats with
complicated structures, such as asymmetric mixed-acid TAGs or binary and
ternary fat mixtures. Many naturally occurring and industrially valuable fats
are made of these TAGs and their mixtures, and many of their fine structures
have been unveiled. Clarification of the kinetic processes of transformations
and mixing processes is also very important and warrants further investigation,
possibly using advanced in situ observation techniques such as a synchrotron
radiation X-ray beam. Finally, dispersed systems, such as emulsions and gels,
that exhibit micro-sized and nano-sized structures will become increasingly
significant.
References
Aquilano, D. & Squaldino, G. (2001) Fundamental Aspects of Equilibrium and Crystalliza-
tion Kinetics. In: Crystallization Processes in Fats and Lipid Systems (Garti, N. & Sato,
K., eds), Marcel Dekker, New York, USA, pp.1–51.
Arakawa, H., Kasai, T., Okumura, Y. & Maruzeni, S. (1998) Polymorphism of SOA triglerol
prepared from sal fat. J. Jpn. Oil Chem. Soc. 47, 19–24.
Arishima, T. & Sato, K. (1989) Polymorphism of POP and SOS. III. Solvent crystallization
of β2 and β1 polymorphs. J. Am. Oil Chem. Soc. 66, 1614–1617.
Arishima, T., Sagi, N., Mori, H. & Sato, K. (1991) Polymorphism of POS. I. Occurrence and
polymorphic transformation. J Am. Oil Chem. Soc. 68, 710–715.
Arishima, T., Sugimoto, K., Kiwata, R., Mori, H. & Sato, K. (1996) 13C cross-polarization
and magic-angle spinning nuclear magnetic resonance of polymorphic forms of three
triacylglycerol. J Am. Oil Chem. Soc. 73, 1231–1236.
Bailey, A.E. (1950) Melting and solidification of pure compounds. In: Melting and Solidifi-
cation of Fats, Wiley, New York, USA, pp.117–180.
Barton, P.G. & Gunstone, F.D. (1975) Hydrocarbon chain packing and molecular motion in
phospholipids bilayers formed from unsaturated lecithins. J. Biol. Chem. 250, 4470–
4476.
Baxter, J.F.G., Morris, J. & Gaim-Marsoner, G. (1995) Process for retarding fat bloom in fat-
based confectionery masses. EU Patent Application 95306833 5.
Bernstein, J. (2002) Polymorphism in Molecular Crystals, Oxford University Press, Oxford,
UK.
Birker, P.J.M.W.L. & Blonk, J.C.G. (1993) Alkyl chain packing in a β′ triacylglycerol
measured by atomic force microscopy. J. Am. Oil Chem. Soc. 70, 319–321.
Birker, P.J.M.W.L., de Jong, S., Roijers, E.C. & van Soest, T.C. (1991) Structural investiga-
tion β′ triacylglycerols: an X-ray diffraction and microscopic study of twinned β′ crystals.
J. Am. Oil Chem. Soc. 68, 895–906.
Boistelle, R. (1988) Fundamentals of nucleation and crystal growth. In: Crystallization and
Polymorphism of Fats and Fatty Acids (Garti, N. & Sato, K., eds), Marcel Dekker, New
York, USA, pp.189–226.
Boubekri, K., Yano, J., Ueno, S. & Sato, K. (1999) Polymorphic transformations in SRS (sn-
1,3-distearoyl-2-ricinoleyl glycerol). J. Am. Oil Chem. Soc. 76, 949–955.
Cardew, P.T. & Davey, R.J. (1985) The kinetics of solvent-mediated phase transformations.
Proc. R. Soc. London A398, 415–428.
Chow, C.K. (1992) Fatty Acids in Foods and Their Health Implications, Marcel Dekker, New
York, USA.
Chronakis, I.S. (1997) Structural-functional and water-holding studies of biopolymers in low
fat content spreads. Lebensm. Wissen.-Technol. 30, 36–44.
Clark, A.H. (1992) Gels and gelling. In: Physical Chemistry of Foods (Schwartzberg, H.G.
& Hartel, R.W., eds), Marcel Dekker, New York, USA, pp.263–305.
Clarke, C. (2004) The Science of Ice Cream, The Royal Society of Chemistry, Cambridge, UK.
Coupland, J.N. (2002) Crystallization in emulsions. Curr. Opin. Coll. Interface Sci. 7, 445–450.
de Jong, S, van Soest, T.C. & van Schaick, M.A. (1991) Crystal structures and melting points
of unsaturated triacylglycerols in the β phase. J. Am. Oil Chem. Soc. 68, 371–378.
de Man, J.M. (1999) Relationship among chemical, physical, and textural properties of fats.
In: Physical Properties of Fats, Oils, and Emulsifiers (Widlak, N., ed.), Marcel Dekker,
New York, USA, pp.79–95.
de Man, J.M. & de Man, L. (2002) Texture of fats. In: Physical Properties of Lipids
(Marangoni, A.G. & Narine, S.S., eds), Marcel Dekker, New York, USA, pp.191–217.
Di, L. & Small, D.M. (1993) Physical behavior of the mixed chain diacylglycerol, 1-stearoyl-
2-oleoyl-sn-glycerol: difficulties in chain packing produce marked polymorphism. J.
Lipid Res. 34, 1611–1623.
Di, L. & Small, D.M. (1995) Physical behavior of the hydrophobic core of membranes:
properties of 1-stearoyl-2-linoleoyl-sn-glycerol. Biochemistry 35, 16672–16677.
Dickinson, E. & McClements, D.J. (1996) Advances in Food Colloids, Blackie Academic &
Professional, London, UK.
Dickinson, E., McClements, D.J. & Povey, M.J.W. (1991) Ultrasonic investigation of the
particle size dependence of crystallization in n-hexadecane–water emulsions. J. Coll.
Interface Sci. 142, 103–110.
Dorset, D.L. & Pangborn, W.A. (1988) Polymorphic forms of 1, 2-dipalmitoyl-sn-glycerol:
a combined X-ray and electron diffraction study. Chem. Phys. Lipids 48, 19–28.
Engstrom, L. (1992) Triglyceride systems forming molecular compounds. J. Fat Sci.
Technol. 94, 173–181.
Eskin, G.I. (1994) Influence of cavitation treatment of melts on the processes of nucleation
and growth of crystals during solidification of ingots and castings from light alloys.
Ultrasonic Sonochem. 1, S59–S63.
Feuge, R.O., Landmann, W., Mitcham, D. & Lovegren, N.V. (1962) Tempering triglycerides
by mechanical working. J Am. Oil Chem. Soc. 39, 310–313.
Gallegos, C. & Franco, J.M. (1999) Rheology of food, cosmetics and pharmaceuticals. Curr.
Opin. Coll. Interface Sci. 4, 288–293.
Gennis, R.B. (1989) Biomembranes: Molecular Structure and Function, Springer-Verlag,
New York, USA.
Goff, H. (1997) Ice cream. In: Lipid Technologies and Applications (Gunstone, F.D. &
Padley, F.B., eds), Marcel Dekker, New York, USA, pp.355–368.
Goto, M. & Asada, E. (1978) The crystal structure of the B-form of stearic acid. Bull. Chem.
Soc. Japan 51, 2456–2459.
Goto, M., Kodali, D.R., Small, D.M., Honda, K., Kozawa, K. & Uchida, T. (1992) Single
crystal structure of a mixed-chain triacylglycerol: 1,2-dipalmitoyl-3-acetyl-sn-glycerol.
Proc. Natl Acad. Sci. USA 89, 8083–8086.
Goto, M., Honda, K., Di, L. & Small, D.M. (1995) Crystal structure of a mixed chain
diacylglycerols, 1-stearoyl-3-oleoyl-glycerol. J. Lipid Res. 36, 2185–2190.
Gunstone, F.D. & Padley, F.B., eds (1997) Lipid Technologies and Applications, Marcel
Dekker, New York, USA.
Hachiya, I., Koyano, T. & Sato, K. (1989a) Seeding effects on solidification behavior of
cocoa butter and dark chocolate. I. Kinetics of solidification. J. Am. Oil Chem. Soc. 66,
1757–1762.
Hachiya, I., Koyano, T. & Sato, K. (1989b) Seeding effects on solidification behavior of
cocoa butter and dark chocolate. II. Physical properties of dark chocolate. J. Am. Oil
Chem. Soc. 66, 1763–1770.
Hagemann, J.W., Tallent, W.H., Barve, J.A., Ismail, I.A. & Gunstone, F.D. (1975) Polymor-
phism in single-acid triglycerides of positional and geometric isomers of octadecanoic
acid. J. Am. Oil Chem. Soc. 52, 204–207.
Hauser, H., Pascher, I. & Sundell, S. (1988) Preferred conformation and dynamics of the
glycerol backbone in phospholipids. An NMR and X-ray single-crystal analysis.
Biochemistry 27, 9166–9174.
Heertje, I. & Leunis, M. (1997) Measurement of shape and size of fat crystals by electron
microscopy. Lebensm.-Wissenschaft Technol. 30, 141–146.
Hermansson, M. (1999) The fluidity of hydrocarbon regions in organo-gels, studied by NMR
– Basic translational and rotational diffusion measurements. Coll. Surf. A 154, 303–309.
Hernqvist, L. (1988) On the crystal structure of the β′1-form of triglycerides and mechanism
behind the β′1-β transition of fats. Fat Sci. Technol. 90, 451–454.
Hernqvist, L. & Larsson, K. (1982) On the crystal structure of the β′-forms of triglycerides
and structural changes at the phase transitions liq–α–β′–β. Fette Seifen Anstrichm. 84,
349–354.
Higaki, K., Ueno, S., Koyano, T. & Sato, K. (2001) Effects of ultrasonic irradiation on
crystallization behavior of tripalmitoylglycerol and cocoa butter. J. Am. Oil Chem. Soc.
78, 513–518.
Higaki, K., Sasakura, Y., Koyano, T., Hachiyal, I. & Sato, K. (2003) Physical analyses of gel-
like behavior of binary mixtures of high-melting and low-melting fats. J. Am. Oil Chem.
Soc. 80, 263–270.
Higaki, K., Koyano, T., Hachiya, I. & Sato, K. (2004a) In-situ optical observation of
microstructure of β-fat gel made of binary mixtures of high-melting and low-melting fats.
Food Res. Int. 37, 2–10.
Higaki, K., Koyano, T., Hachiya, I., Sato, K. & Suzuki, K. (2004b) Rheological properties
of β-fat gel made of binary mixtures of high-melting and low-melting fats. Food Res. Int.
37, 799–804.
Hindle, S., Povey, M.J.W. & Smith, K. (2000) Kinetics of crystallization in n-hexadecane and
cocoa butter oil-in-water emulsions accounting for droplet collision-mediated nuclea-
tion. J. Coll. Interface Sci. 232, 370–380.
Hiramatsu, H., Sato, T., Inoue, T., Suzuki, M. & Sato, K. (1990) Pressure effect on
transformation of cis-unsaturated fatty acid polymorphs. 2. Palmitoleic acid (cis-9-
hexadecenoic acid). Chem. Phys. Lipids 56, 59–63.
Hollander, F.F.A., Boerrigter, S.X.M., van de Streek, J., Bennema, P., Meekes, H., Yano, J.
& Sato, K. (2003) Comparing the morphology of β-n.n.n with β′-n.n+2.n and β′-n.n.n-
2 triacylglycerol crystals. J. Phys. Chem. B 107, 5680–5689.
Inaoka, K., Kobayashi, M., Okada, M. & Sato, K. (1988) Stability, occurrence and step
morphology of polymorphs and polytypes of stearic acid, II. Monolamella step morphol-
ogy and composite polymorphic/polytypic transformation. J. Crystal Growth 87, 243–

250.
Jensen, L.H. & Mabis, A.J. (1963) Crystal structure of β-tricaprin. Nature 197, 681–682.
Jensen, L.H. & Mabis, A.J. (1966) Refinement of the structure of β-tricaprin. Acta Crystallogr.
22, 770–781.
Kaneko, F. (2001) Polymorphism and phase transitions of fatty acids and acylglycerols. In:
Crystallization Processes in Fats and Lipid Systems (Garti, N. & Sato, K., eds), Marcel
Dekker, New York, USA, pp.53–97.
Kaneko, F., Kobayashi, M., Kitagawa, Y. & Matsuura, Y. (1990) Structure of stearic acid E
form. Acta Crystallogr. C 46, 1490–1492.
Kaneko, F., Kobayashi, M., Kitagawa, Y., Matsuura, Y., Sato, K. & Suzuki, M. (1992a)
Structure of the high-melting phase of petroselinic acid. Acta Crystallogr. C48, 1057–
1060.
Kaneko, F., Kobayashi, M., Kitagawa, Y., Matsuura, Y., Sato, K. & Suzuki, M. (1992b)
Structure of the low-melting phase of petroselinic acid. Acta Crystallogr. C48, 1054–
1057.
Kaneko, F., Yamazaki, K., Kobayashi, M., Kitagawa, Y., Matsuura, Y., Sato, K. & Suzuki,
M. (1993) Structure of the γ phase of erucic acid. Acta Crystallogr. C 49, 1232–1234.
Kaneko, F., Yamazaki, K., Kobayashi, M., Kitagawa, Y., Matsuura, Y., Sato, K. & Suzuki,
M. (1996) Mechanism of the γ–α and γ1–α1 reversible solid-state phase transitions of
erucic acid. J. Phys. Chem. 100, 9138–9148.
Kaneko, F., Kobayashi, M., Sato, K. & Suzuki, M. (1997a) Martensitic phase transition of
petroselinic acid: Influence of polytypic structure. J. Phys. Chem. B 101, 285–292.
Kaneko, F., Yamazaki, K., Kitagawa, Y., Kikyo, T., Kobayashi, M., Kitagawa, Y., Matsuura,
Y., Sato, K. & Suzuki, M. (1997b) Structure and crystallization behavior of the β phase
of oleic acid. J. Phys. Chem. B 101, 1803–1809.
Kaneko, F., Yano, J. & Sato, K. (1998) Diversity in the fatty-acid conformation and chain
packing of cis-unsaturated lipids. Curr. Opin. Struct. Biol. 8, 417–425.
Kashchiev, D. & Sato, K. (1998) Kinetics of crystallization preceded by metastable-phase
formation. J. Chem. Phys. 109, 8530–8540.
Kashchiev, D., Kaneko, N. & Sato, K. (1998) Kinetics of crystallization in polydisperse
emulsions. J. Coll. Interface Sci. 208, 167–177.
Katsuragi, T., Kaneko, N. & Sato, K. (2001) Effects of addition of hydrophobic sucrose fatty
acid oligoesters on crystallization rates of n-hexadecane in oil-in-water emulsions. Coll.
Surf. B 20, 229–237.
Kellens, M., Meeussen, W., Hammersley, A. & Reynaers, H. (1991) Synchrotron radiation
investigations of the polymorphic transitions of saturated monoacid triglycerides. Part 2:
Polymorphisms study of a 50:50 mixture of tripalmitin and tristearin during crystalliza-
tion and melting. Chem. Phys. Lipids 58, 145–158.
Kobayashi, M., Kobayashi, T., Ito, Y. & Sato, K. (1984) Polytypism in n-fatty acids and low-
frequency Raman spectra: stearic acid B form. J. Chem. Phys. 80, 2897–2903.
Kobayashi, M., Kobayashi, T., Ito, Y. & Sato, K. (1986a) Raman and Brillouin studies on
polytype structures and physical properties of n-alkanes and n-fatty acids. Bull. Mineral
109, 171–184.
Kobayashi, M., Kaneko, F., Sato, K. & Suzuki, M. (1986b) Vibrational spectroscopic study
on polymorphism and order–disorder. J. Phys. Chem. 90, 6371–6378.
Kodali, D.R., Atkinson, D., Redgrave, T.G. & Small, D.M. (1984) Synthesis and polymor-
phism of 1,2-dipalmitoyl-3-acyl-sn-glycerol. J. Am. Oil Chem. Soc. 61, 1078–1084.
Kodali, D.R., Atkinson, D. & Small, D.M. (1989) Molecular packing of 1,2-dipalmitoyl-3-
decanoyl-sn-glycerol (PP10): Bilayer, trilayer and hexalayer structures. J. Phys. Chem.
93, 4683–4691.
Kodali, D.R., Atkinson, D. & Small, D.M. (1990a) Molecular packing in fats, influences of
acyl chain length and unsaturation. J. Disp. Sci. Technol. 10, 393–440.
Kodali, D.R., Atkinson, D. & Small, D.M. (1990b) Polymorphic behavior of 1,2-dipalmitoyl-
3-lauroyl (PP12)- and 3-myristoyl (PP14)-sn-glycerols. J. Lipid Res. 31, 1853–
1864.
Koyano, K., Hachiya, I., Arishima, T., Sato, K. & Sagi, N. (1989) Polymorphism of POP and
SOS. II. Kinetics of melt crystallization. J. Am. Oil Chem. Soc. 66, 675–679.
Koyano, T., Hachiya, I., Arishima, T., Sato, K. & Sagi, N. (1991) Polymorphism of POS. II.
Kinetics of melt crystallization. J. Am. Oil Chem. Soc. 68, 716–718.
Koyano, T., Hachiya, I. & Sato, K. (1992) Phase behavior of mixed systems of SOS and OSO.
J. Phys. Chem. 96, 10514–10520.
Krog, N. (2001) Crystallization properties and lyotropic phase behavior of food emulsifiers.
In: Crystallization Processes in Fats and Lipid Systems (Garti, N. & Sato, K., eds), Marcel
Dekker, New York, USA, pp.505–526.
Krog, N. & Larsson, K. (1992) Crystallization at interfaces in food emulsions. Fat Sci.
Technol. 94, 55–57.
Larsson, K. (1964) The crystal structure of the β-form of trilaurin. Ark. Kemi. 23, 1–15.
Larsson, K. (1966) Classification of glyceride crystal forms. Acta Chem. Scand. 20, 2255–
2260.
Larsson, K. (1972) Molecular arrangement in glycerides. Fette Seifen Anstrichm. 74, 136–
142.
Larsson, K. (1994) Solid state behavior. In: Lipids – Molecular Organization, Physical
Functions and Technical Applications, The Oily Press, Dundee, UK, pp.7–46.
Lopez, C., Bourgaux, C., Lesieur, P., Bernadou, S., Keller, G. & Ollivon, M. (2002) Thermal
properties and structure of milk fat: 3. Influence of cooling rate and droplet size on cream
crystallization. J. Coll. Interface Sci. 254, 64–78.
Lutton, E.S. (1950) Review of the polymorphism of saturated even glycerides. J. Am. Oil
Chem. Soc. 27, 276–281.
MacMillan, S.D., Roberts, K.J., Rossi, A., Wells, M.A., Polgreen, M.C. & Smith, I.H. (2002)
In situ small angle X-ray scattering (SAXS) studies of polymorphism with the associated
crystallization of cocoa butter fat using shearing conditions. Cryst. Growth Des. 2, 221–
226.
Malkin, T. (1954) The polymorphism of glycerides. Progr. Chem. Fats Other Lipids II, 1–
50.
Malta, V., Celotti, G., Zanetti, R. & Martelli, A.F. (1971) Crystal structure of the C form of
stearic acid. J. Chem. B, 548–553.
Marangoni, A.G. (2005) Fat Crystal Networks, Marcel Dekker, New York, USA.
Marangoni, A.G. & Hartel, R.W. (1998) Visualization and structural analysis of fat crystal
networks. Food Technol. 52, 46–51.
Mazzanti, G., Guthrie, S.E., Sirota, E.B., Marangoni, A.G. & Idziak, S.H.J. (2003) Orienta-
tion and phase transitions of fat crystals under shear. Cryst. Growth Des. 3, 721–725.
Mazzanti, G., Guthrie, S.E., Sirota, E.B., Marangoni, A.G. & Idziak, S.H.J. (2004) Novel
shear-induced phases in cocoa butter. Cryst. Growth Des. 4, 409–411.
McClements, D.J., Dickinson, E., Dungan, S.R. Kinsella, J.E., Ma, J.G. & Povey, M.J.W.
(1993) Effect of emulsions type on the crystallization kinetics of oil-in-water emulsions
containing a mixture of solid and liquid droplets. J. Coll. Interface Sci. 160, 293–
297.
Minato, A., Ueno, S., Yano, J., Wang, Z.H., Seto, H., Amemiya, Y. & Sato, K. (1996)
Synchrotron radiation X-ray diffraction study on phase behavior of PPP–POP binary
mixtures. J. Am. Oil Chem. Soc. 73, 1567–1572.
Minato, M., Ueno, S., Smith, K., Amemiya, Y. & Sato, K. (1997a) Thermodynamic and
kinetic study on phase behavior of binary mixtures of POP and PPO forming molecular
compound systems. J. Phys. Chem B 101, 3498–3505.
Minato, A., Ueno, S., Yano, J., Smith, K., Seto, H., Amemiya, Y. & Sato, K. (1997b) Thermal
and structural properties of sn-1,3-dipalmitoyl-2-oleoylglycerol and sn-1,3-oleoyl-2-
palmitoylglycerol binary mixtures examined with synchrotron radiation X-ray diffraction.
J. Am. Oil Chem. Soc. 74, 1213–1220.
Minato, M., Yano, J., Ueno, S., Smith, K. & Sato, K. (1997c) FT-IR study on microscopic
structures and conformations of POP–PPO and POP–OPO molecular compounds. Chem.
Phys. Lipids 88, 63–71.
Mullin, J.W. (2001) Phase equilibria. In: Crystallization (4th edn), Butterworth-Heineman,
Oxford, UK, pp.135–180.
Mutaftschiev, B. (2001) Some specific cases of nucleation. In: The Atomistic Nature of
Crystal Growth, Springer, Berlin, Germany, pp.249–266.
Mykhaylyk, O.O. & Hamley, I.W. (2004) The packing of triacylglycerols from SXAS
measurements: Application to the structure of 1,3-distearoyl-2-oleoyl-sn-glycerol crys-
tal phases. J. Phys. Chem. B 108, 8069–8083.
Mykhaylyk, O.O., Casteletto, V., Hamley, I.W. & Povey, M.J.W. (2004) Structure and
transformation of low-temperature phases of 1,3-distearoyl-2-oleoyl-sn-glycerol. Eur. J.
Lipid Sci. Technol. 106, 319–324.
Palanuwech, J. & Coupland, J.N. (2003) Effect of surfactant type on the stability of oil-in-
water emulsions to dispersed phase crystallization. Coll. Surf. A 223, 251-262.
Pascher, I. (1996) The different conformations of the glycerol region of crystalline acylglycerols.
Curr. Opin. Struct. Biol. 6, 439–448.
Pascher, I., Sundell, S. & Hauser, H. (1981) Glycerol conformation and molecular packing
of membrane lipids: the crystal structure of 2,3-dilauroyl-D-glycerol. J. Mol. Biol. 153,
791–806.
Pascher, I., Lundmark, M., Nyholm, P-G. & Sundell, S. (1992) Crystal structures of
membrane lipids. Biochim. Biophys. Acta 1113, 339–373.
Patrick, M., Blindt, R. & Janssen, J. (2004) The effects of ultrasonic intensity on the crystal
structure of palm oil. Ultrason. Sonochem. 11, 251–255.
Peschar, R., Pop, M.M., de Ridder, D.J.A., van Mechelen, J.B., Driesse, R.A.J. & Schenk,
H. (2004) Crystal structure of 1,3-distearoyl-2-oleoylglycerol and cocoa butter in the
β(V) phase reveal the driving force behind the occurrence of fat bloom on chocolate. J.
Phys. Chem. B 108, 15450–15453.
Povey, M.J.W. (2001) Crystallization of oil-in-water emulsions. In: Crystallization Proc-
esses in Fats and Lipid Systems (Garti, N. & Sato, K., eds,) Marcel Dekker, New York,
USA, pp.251–288.
Povey, M.J.W. & Mason, T.J. (1998) Ultrasound in Food Processing, Blackie Academic &
Professional, London, UK.
Rossel, J.B. (1967) Phase diagram of triglyceride systems. In: Advances in Lipid Research
Vol. 5 (Paolettei, R. & Kritchevsky, D., eds), Adademic Press, New York, USA, pp.353–
408.
Rousseau, D. (2002) Fat crystal behaviour in food emulsions. In: Physical Properties of
Lipids (Marangoni, A. & Narine, S.S., eds) Marcel Dekker, New York, USA, pp.219–
264.
Rousset, P., Rappaz, M. & Minner, E. (1998) Polymorphism and solidification kinetics of the
binary system POS–SOS. J. Am. Oil Chem. Soc. 75, 857–864.
Sakamoto, M., Ohba, A., Kuriyama, J., Maruo, K., Ueno, S. & Sato,K. (2004) Influences of
fatty acid moiety and esterification of polyglycerol fatty acid esters on the crystallization
of palm mid fraction in oil-in-water emulsion. Coll. Surf. B 37, 27–33.
Sato, K. & Garti, N. (1988) Crystallization and polymorphic transformation: an introduction.
In: Crystallization and Polymorphism of Fats and Fatty Acids (Garti, N. & Sato, K., eds),
Marcel Dekker, New York, USA, pp.1–7.
Sato, K. & Kobayashi, M. (1991) Structure, stability and crystal growth of polymorphs and
polytypes of long chain alphatic compounds. In: Crystals – Growth, Properties and
Applications, Vol.13: Organic Crystals I. Characterization (Karl, N., ed.), Springer-
Verlag, Berlin, Germany, pp.65–108.
Sato, K. & Kuroda, T. (1987) Kinetics of melt crystallization and transformation of
tripalmitin polymorphs. J. Am. Oil Chem. Soc. 64, 124–127.
Sato, K. & Suzuki, M. (1986) Solvent crystallization of α, β and γ polymorphs of oleic acid.
J. Am. Oil Chem. Soc. 63,1356–1359.
Sato, K., Morishita, H. & Kobayashi, M. (1988) Stability, occurrence and step morphology
of polymorphs and polytypes of stearic acid, I. Stability and occurrence. J. Cryst. Growth
87, 236–242.
Sato, K., Arishima, T., Wang, Z.H., Ojima, K., Sagi, N. & Mori, H. (1989) Polymorphism
of POP and SOS. I. Occurrence and polymorphic transformation. J. Am. Oil Chem. Soc.
66, 664–674.
Sato, K,.Yoshimoto, N., Suzuki, M., Kobayashi, M. & Kaneko, F. (1990) Structure and
transformation in polymorphism of petroselinic acid (cis-ω-12-octadecenoic acid). J.
Phys. Chem. 94, 3180–3185.
Sato, K., Yano, J., Kawada, I. & Kawano, M. (1997) Polymorphic behavior of gondoic acid
and phase behavior of its binary mixtures. J. Am. Oil Chem. Soc. 74, 1153–1159.
Sato, K., Ueno, S. & Yano, J. (1999) Molecular interactions and kinetic properties of fats.
Prog. Lipid Res. 38, 91–116.
Sato, K., Goto, M., Yano, J., Honda, K., Kodali, D.R. & Small, D.M. (2001) Atomic
resolution structure analysis of β′ polymorph crystal of a triacylglycerol: 1,2-dipalmitoyl-
3-myristoyl-sn-glycerol. J. Lipid Res. 42, 338–345.
Shipley, G.G. (1986) X-ray crystallographic studies of aliphatic lipids. In: The Physical
Chemistry of Lipids, from Alkanes to Phospholipids (Handbook of Lipid Research Vol.
4) (Small, S.M., ed.), Plenum Press, New York, USA, pp.97–147.
Small, D.M. (1984) Lateral chain packing in lipids and membranes. J. Lipid Res. 25, 1490–
1500.
Small, D.M. (1986) Glycerides. In: The Physical Chemistry of Lipids, from Alkanes to
Phospholipids (Handbook of Lipid Research Vol. 4) (Small, S.M., ed.), Plenum Press,
New York, USA, pp.345–394.
Stapley, A.G.F., Tewkesbury, H. & Fryer, P.J. (1999) The effects of shear and temperature
history on the crystallization of chocolate. J. Am. Oil Chem. Soc. 76, 677–685.
Suzuki, M., Ogaki, T. & Sato, K. (1985) Crystallization and transformation mechanisms of
α, β and γ polymorphs of ultra-pure oleic acid. J. Am. Oil Chem. Soc. 62, 1600–
1604.
Takeuchi, M., Ueno, S., Yano, Y., Floter, E. & Sato, K. (2000) Polymorphic transformation
of 1,3-distearoyl-sn-linoleoyl-glycerol. J. Am. Oil Chem. Soc. 77, 1243–1249.
Takeuchi, M., Ueno, S., Floeter, E. & Sato, K. (2002a) Binary phase behavior of 1,3-
distearoyl-2-oleoyl-sn-glycerol (SOS) and 1,3-distearoyl-2-linoleoyl-sn-glycerol (SLS).
J. Am. Oil Chem. Soc. 79, 627–632.
Takeuchi, T., Ueno, S. & Sato, K. (2002b) Crystallization kinetics of polymorphic forms of
a molecular compound constructed by SOS (1,3-distearoyl-2-oleoyl-sn-glycerol) and
SSO (1,2-distearoyl-3-oleoyl-rac-glycerol). Food Res. Int. 35, 919–926.
Takeuchi, T., Ueno, S. & Sato, K. (2003) Synchrotron radiation SAXS/WAXS study of
polymorph-dependent phase behavior of binary mixtures of saturated monoacid
triacylglycerols. Cryst. Growth Des. 3, 369–374.
Timms, R.E. (1984) Phase behaviour of fats and their mixtures. Prog. Lipid Res. 23, 1–38.
Timms, R.E. (2003) Confectionery Fats Handbook, The Oily Press, Bridgwater, UK.
Ueno, U., Suetake, T., Yano, J., Suzuki, M. & Sato, K. (1994) Structure and polymorphic
transformations in elaidic acid (trans-ω9-octadecenoic acid). Chem. Phys. Lipids 72, 27–
34.
Ueno, S., Minato, A., Seto, H., Amemiya, Y. & Sato, K. (1997) Synchrotron radiation X-ray
diffraction study of liquid crystal formation and polymorphic crystallization of SOS
sn-1,3-distearoyl-2-oleoyl glycerol. J. Phys. Chem. B 101, 6847–6854.
Ueno, S., Miyazaki, A., Yano, J., Furukawa, Y., Suzuki, M. & Sato, K. (2000) Polymorphism
of linoleic acid (cis-9, cis-12-octadecadienoic acid) and α-linolenic acid (cis-9, cis-12,
cis-15-octadecatrienoic acid). Chem. Phys. Lipids 107, 169–178.
Ueno, S., Hamada, Y. & Sato, K. (2003a) Controlling polymorphic crystallization of n-alkane
crystals in emulsion droplets through interfacial heterogeneous nucleation. Cryst. Growth
Des. 3, 935–939.
Ueno, U., Ristic, R.I., Higaki, K. & Sato, K. (2003b) In-situ studies of ultrasound stimulated
fat crystallization using synchrotron radiation. J. Phys. Chem. B 107, 4927–4935.
Vanapalli, S., Palanuwech, J. & Coupland, J.N. (2002) Influence of crystallization on the
stability of flocculated emulsions. J. Ag. Food Chem. 50, 5224–5228.
van de Streek, J., Verwer, P., de Gelder, R. & Hollander, F. (1999) Structural analogy between
β′ triacylglycerols and n-alkanes: toward the crystal structure of β′-2 p.p+2.p
triacylglycerols. J. Am. Oil Chem. Soc. 76, 1333–1341.
van Langevelde, A., van Malssen, K., Hollander, F., Pechar, R. & Schenk, H. (1999a)
Structure of mono-acid even-numbered β′-triacylglycerols. Acta Crystallogr. B 55, 114–
122.
van Langevelde, A., van Malssen, K., Sonneveld, E.D., Pechar, R. & Schenk, H. (1999b)
Crystal packing of a homologous series β′-stable triacylglycerols. J. Am. Oil Chem. Soc.
76, 603–609.
van Langevelde, A., van Malssen, K., Dressen, R., Goubits, K., Hollander, F., Peschar, R.,
Zwart, P. & Schenk, H. (2000) Structure of CnCn+2Cn-type (n=even) β′-triacylglycerols.
Acta Crystallogr. B 56, 1103–1111.
Walstra, P., Kloek, W. & van Vliet, T. (2001) Fat crystal networks. In: Crystallization
Processes in Fats and Lipid Systems (Garti, N. & Sato, K., eds), Marcel Dekker, New
York, USA, pp.289–328.
Wang, Z.H., Sato, K., Sagi, N., Izumi, T. & Mori, H. (1987) Polymorphism 1,3-disaturated-
acyl-2-oleoylglycerols: POP, SOS, AOA, BOB. J. Jpn. Oil Chem. Soc. 36, 671–
679.
Wille, R.L. & Lutton, E.S. (1966) Polymorphism of cocoa butter. J. Am. Oil Chem. Soc. 43,
491–496.
Yano, J., Ueno, S., Sato, K., Arishima, T., Sagi, N., Kaneko, F. & Kobayashi, M. (1993) FT-
IR study of polymorphic transformations in SOS, POP and POS. J. Phys. Chem. 97,
12967–12973.
Yano, J., Kaneko, F., Kobayashi, M., Kodali, D.R., Small, D.M. & Sato, K. (1997) Structural
analyses of triacylglycerol polymorphs with FT-IR techniques: II. β′1-form of 1,2-
dipalmitoyl-3-myristoyl-sn-glycerol. J. Phys. Chem. B 101, 8120–8128.
Yano, J., Sato, K., Kaneko, F., Small, D.M. & Kodali, D.R. (1999) Structural analyses of
polymorphic transitions of sn-1,3-distearoyl-2-oleoylglycerol (SOS) and sn-1,3-dioleoyl-
2-stearoylglycerol (OSO): assessment on steric hindrance of unsaturated and saturated
acyl chain interactions. J. Lipid Res. 40, 140–151.
Yoshimoto, N., Suzuki, M. & Sato, K. (1991) Polymorphic transformation in asclepic acid
(cis-ω-octadecenoic acid) Chem. Phys. Lipids 57, 67–73.
Ziegleder, G. (1985) Verbesserte kristallization von kakaobutter unter dem einfluss eines
scherge falles. Int. Z. Lebensm. Technol. Verfallensyechm. 36, 412–418.
CHAPTER 3
Liquid-crystalline lipid–water phases
A. Structures
1. Introduction: Lamellar structures and the role of curvature

The solid-state behaviour of polar lipids, as we learned from the previous
chapter, is characterized by localization of hydrocarbon chains into layers with
weak lateral interaction, whereas there are strong forces between the polar
groups, which form end-group planes. It is therefore natural to expect that such
a structure may ‘melt’ successively on heating: first the hydrocarbon chains
become disordered into a liquid-like structure, with the overlying gross struc-
ture remaining intact, then at a higher temperature complete melting occurs
(Figure 3.1). This actually takes place in lipids with very strong interaction
within the polar sheets. The all-trans conformation of the chains is transformed
into a state of chaotic disorder with high mobility and a large proportion of
gauche conformations along the chains. The combination of disorder on the
atomic scale while the long-range order into layers still persists is the charac-
teristic property of liquid-crystalline phases of lipids, sometimes also termed
mesomorphic phases.
As shown in Figure 3.2, a similar liquid-crystalline phase can also be formed
by penetration of water into a layer between the polar sheets, provided that the
temperature is high enough to allow disorder of the hydrocarbon chains.
The liquid-crystalline phases formed by the addition of water (or another
solvent) are called lyotropic liquid crystals, whereas liquid-crystalline phases
formed by heating (as in Figure 3.1) are called thermotropic liquid crystals.
The liquid-crystalline phases shown in Figures 3.1 and 3.2 are lamellar, and
they are termed Lα phases. The cross-sectional area per hydrocarbon chain is
typically about 30 Å2 and the volume per CH3 group is about 30 Å3, reflecting
the liquid-like disorder. The X-ray short-spacing region shows only a diffuse
halo centred at 4.5 Å, whereas all crystal forms show sharp diffraction lines
around 4 Å due to the close-packing order of the hydrocarbon chains.
A milestone in the understanding of the nature of lipid–water liquid-
crystalline phases was the work by Luzzati and co-workers (1960), which gave
an unambiguous demonstration of the liquid-crystalline character of the hydro-
carbon chains. From this work it became obvious that the hydrocarbon chains
can fill spaces of different geometries, not only lamellar types.
The structures shown in this chapter are of utmost importance in biology as
well as in numerous technical applications. Their relevance in biology was
73
Figure 3.1 Formation of a liquid-crystalline phase just before melting of a crystalline polar lipid. The
hydrocarbon chain axis and the polar head groups are indicated.
Figure 3.2 A liquid-crystalline phase formed from a polar lipid (with two hydrocarbon chains in the
molecule) by the addition of water.
discussed in a commentary by Luzzati (1997) in relation to the “widespread

opinion that lipids are passive components of biological membranes”, provid-
ing “examples that most convincingly attest to the biological significance of
lipid polymorphism, and in particular to the specific role of the cubic phases”.
Luzzati also defined what he called gel phases, which have water layers
alternating with lipid bilayer, like in the lamellar liquid crystalline phase, but
the lipid molecules have crystallized in the gel phase. The gel phase in aqueous
systems of monoacylglycerols was described in Chapter 2. A different gel
phase was described in aqueous systems of saturated phosphatidylcholines by
Janiak et al. (1976): the periodic crystalline bilayer phase Pβ′ (also termed the
ripple phase). The periodicity is due to ripples along the bilayer.
An intermediate type of lamellar phase, with stripes of the bilayer in the
crystalline state alternating with stripes of the bilayer with the liquid-like chain
conformation, occurs in cholesterol–phospholipid systems. The mechanism
behind the formation of such structures is a cholesterol-induced phase
LIQUID-CRYSTALLINE LIPID–WATER PHASES 75
separation within the bilayer. A detailed structure analysis of bilayers consist-

ing of cholesterol and dimyristoylphosphatidylcholine has been reported
(Mortensen et al., 1988), and cholesterol-rich stripes alternate there with
cholesterol-poor stripes. There is an immiscibility gap at between about 8 and
20mol% cholesterol.
If both types of stripes or domains have a liquid-type chain conformation,
there are two similar lamellar liquid-crystalline domains along the bilayers,
which should be regarded as a lamellar liquid-crystalline phase with inner
periodicity. These two bilayer phases are termed Lα and Lα(O), where Lα(O) is
cholesterol-rich and considered to be more ordered. Such structures have also
been described in aqueous phases formed by lung surfactants (Larsson et al.,
2003). Furthermore, these two bilayer structures represent phase separation in
biomembranes, where the cholesterol-rich domains (often containing
sphingolipids) are called lipid rafts (Simons & Ikonen, 2000). The shapes of
each bilayer phase coexisting in bilayers were recently visualized in a well-
defined lipid ternary system consisting of sphingomyelin, dioleoyl-
phosphatidylcholine and cholesterol (Baumgart et al., 2003). Certain proteins
can also induce curvature to these domains, so that flask-like invaginations are
formed, the so-called caveolae. These structural mechanisms are further
discussed in connection with cell membranes.
A kind of hybrid phase, a gel–liquid-crystalline bilayer coexisting in a ripple
phase of dipalmitoylphosphatidylcholine, has been demonstrated by time-
resolved X-ray diffraction studies (Rappolt et al., 2000).
The introduction of curvature in the description of lipid–water phases
represents a new paradigm in the understanding of lipid structures (Hyde et al.,
1997), and this approach will be applied below.
2. Hexagonal phases
The basic structural studies mentioned above were done on soap–water sys-
tems. At that time it was known that soap molecules associate in water into
spherical micelles, which with increased concentration become elongated.
When the concentration of such rod-shaped aggregates is increased, it is natural
to expect that they become oriented and arranged in a hexagonal structure, as
shown in Figure 3.3. There are two simple geometric ways to organize parallel
cylinders or rods, either in a two-dimensional square lattice or in a hexagonal
lattice. The hexagonal lattice has the advantage of a higher packing density per
cross-sectional unit area and a better interaction between the rods. The hexago-
nal structure was also the one found experimentally in the first-studied
soap–water system, and it is now known to be a general structure in amphiphile–
water systems when the amphiphiles are surfactants containing hydrocarbon
chains. One type of surfactant lipid that forms micelles in water is the
lysophospholipids.
Figure 3.3 (Left) A cross-section of the HI structure formed by lipids associated into infinite cylindrical
micelles. (Right) The inverse type of structure (HII), with water cylinders arranged in a continuous
hydrocarbon matrix with interfaces formed by the polar head groups.
Most lipids have hydrocarbon chain regions that require much space. A rod
structure forming a water-continuous phase is therefore not possible, whereas
the inverse type of structure is ideal. Such a structure is also shown in
Figure 3.3. The open and inverse types of hexagonal structures are termed HI
and HII, respectively.
3. Cubic phases
The main features of the bicontinuous cubic lipid–water phases, which are now
generally accepted, are based on minimal surfaces. We will therefore start by
defining some of the concepts behind minimal surfaces.
During the 19th century, a popular topic in mathematics was the analysis of
the minimum surface area defined by borders of various shapes. Soap films
could then be used as experimental models, as a film spanning a particular
frame will minimize its surface area. An example is shown in Figure 3.4.
Figure 3.4 Illustration of a soap film that spans two parallel square-shaped frames.
The curvature at a particular point on a surface is defined by the two principal

radii of curvature, R1 and R2. If we consider the normal vector through this point
and a plane through the normal vector, we can imagine a circle within this
plane, which is a tangent to the surface at the actual point. If further this plane
is allowed to rotate, the circles which form tangents will have different radii,
and the maximal and minimal values are equal to these principal radii R1 and R2
(the sign defines the side of the surface where the circle is located). Two
concepts are now defined; the average curvature, H, and the Gaussian
curvature, G, with:
⎛ 1 ⎞⎛ 1 1 ⎞
H = ⎜ – ⎟ ⎜ –– + –– ⎟
⎝ 2 ⎠ ⎝ R1 R2 ⎠
1
G = ––––
R1R2
The mathematical condition for the minimum area of a surface (such as a soap
film) spanning a certain frame is that H = 0 everywhere along the surface. This
means that the surface is as convex as it is concave at all points within the
surface.
Repetition of a minimal surface unit in three dimensions results in an infinite
three-dimensional structure. Further, if we assume that a surface formed in this
way is space-filling, continuous, and free of self-intersections, it is called a
periodic minimal surface (PMS).
The lipid bilayer of bicontinuous cubic lipid–water phases form PMS.
Earlier theoretical work showed that there are three fundamental PMS struc-
tures. One was called the diamond surface (D-surface), one discovered by
Schwarz more than 100 years ago was called Schwarz’s primitive surface (P-
surface), and one discovered by Schoen (1970) was called the gyroid surface
(G-surface). As will be shown below, these three types of structures exist in
lipids. It had been proposed at an early stage that microemulsions and liquid-
crystalline phases of surfactants might form minimal-surface types of structures
(Scriven, 1976). An experimentally based general introduction to the minimal
surface concept in structure chemistry was reported by Andersson et al. (1988).
We will now describe in detail the step-wise evolution of the cubic PMS
concept of the bicontinuous structures of lipid–water phases. In this way we
can also demonstrate the methods available to determine liquid-crystalline
structures, as well as the limitations of these methods.
Many studies in the area have involved the use of 1-monoolein. However, a
complication of phase studies of monoacylglycerols is isomerization during the
study. Even if the samples initially consist of pure 1-monoolein, acyl migration
during thermal equilibration will result in an equilibrium mixture of about 90%
of the 1-isomer and 10% of the 2-isomer. Even a few hours is sufficient to result
Figure 3.5 Proposed minimal surface structure of a lipid–water phase (Larsson et al., 1980). Two P-
surface structure units along the x-axis are shown. The dotted lines on the unit to the right illustrate a
square frame, which together with frames in the other directions will span a soap film with this P-surface
structure.
in some acyl migration. It might therefore be better to use the name glycerol
monooleate (GMO), without stating the isomer purity.
A detailed structure of the cubic phase of anhydrous sodium myristate was
published by Luzzati & Spegt in 1967. Many observed X-ray diffraction lines
were consistent with space group Ia3d, and a structure consisting of a rod
arrangement of polar groups in a continuum of hydrocarbon chains was
proposed. An X-ray study of a cubic GMO–water phase was first proposed to
indicate space-filling polyhedrons with lipid bilayer faces arranged according
to space group Im3m (Larsson, 1972).
Then an NMR-diffusion study by Lindblom et al. (1979) of a series of cubic
GMO–water phases indicated that the structure was both water and lipid
continuous. As relations in X-ray data with the adjacent Lα phase indicated a
lipid bilayer-based structure unit, opening of the square faces of the earlier
proposed polyhedrons (space group Im3m) was a structure model that seemed
to fulfil both NMR and X-ray data. It was also realized that if the planar faces
were allowed to be curved as minimal surfaces, the proposed structure was
identical to Schwarz’s P-surface (Larsson et al., 1980). The structure is shown
in Figure 3.5. This represented a new structure model of cubic lipid–water
phases different from the prevailing model at that time, with the cubic lipid–
water phases proposed to be arranged as rod systems (Luzzati et al., 1968). The
proposed space group Im3m was not correct, however, as the indexing
involved two phases with related structures, a feature discussed below which
provides strong evidence for the minimal surface structures.
Later, Longley & Mcintosh (1983) analysed a cubic GMO–water phase in
equilibrium with excess water, and from somewhat better diffraction data they
could convincingly conclude that its space group was Pn3m. They also
proposed that the probable structure followed the diamond minimal surface,
the D-surface. This motivated a re-examination of the indexing of the whole

cubic region reported by Lindblom et al. (1979). This showed that it consists of
two cubic phases (Larsson, 1983): first, at low water content, a phase with
space group Ia3d is formed (proposed to be a G-surface structure); and then, at
higher water content, the Pn3m phase reported by Longley & Mcintosh is
obtained. Due to their structural relation, which in fact is the strongest evidence
for the PMS structures, they could be indexed as one single phase with space
group Im3m.
The complete GMO–water phase diagram and the relations between the two
phases involving unit cell size as well as water content were finally reported by
Hyde et al. (1984). The so-called Bonnet relation between these three minimal
surface structures is discussed below, in Section B.4. The minimal surface
structure following the P-surface is formed when amphiphilic polymers are
added to the GMO–water system, as shown below (Landh, 1994).
First the three structures shown in Figures 3.6–3.8 will be described and
compared. We have already seen the P-surface structure unit, which has six
openings to adjacent units. The D-surface shown in Figure 3.6 has four
openings between each unit. Thus the water compartments on each side form
channel systems directed as the carbon bonds in diamond, therefore the name
‘diamond surface’. As in the other minimal surface structures, the two continu-
ous water compartments, separated by one continuous lipid bilayer, are
congruent and have no contact with one another. The G-surface type of bilayer
structure is shown in Figure 3.7. The water channels between the structure
units follow a helical network, with three connections/openings between each
unit.
Figure 3.6 Illustration of the cubic GMO–water structure with the bilayer centred on the D-surface.
The surface was calculated by the nodal surface approximation; see text for details.
Figure 3.7 Schematic illustration of the G-surface structure of cubic lipid–water phases.
Figure 3.8 The structure unit of the P-type of nodal surface illustrated by a net with a proposed standing
wave breathing vibration mode of the bilayer, indicated by transparent layers (with maximal amplitude
at the flat point). (After Andersson et al., 1997).
Why are these different cubic phases formed? A qualitative understanding

may perhaps be based on the connectivity between the structure units. If the P-
surface in Figure 3.5 is considered, it is obvious that the bilayer thickness at the
‘necks’ between the units requires a minimum water content so that there is an
opening between each unit. With six connectivity, it is perhaps not surprising
that the P-surface structure in the GMO system is formed only when the water
compartments are expanded by a polymer, whereas successive reduction of the
aqueous volume gives first the D-surface with four connectivity, and finally the
G-surface structure with three connectivity.
A simple method to approximate the PMS structures was discovered by von
Schnering & Nesper (1991). They found that roots of Fourier series of the first
structure factors give periodic nodal surfaces (PNS) which are very close to the
corresponding PMS (within a few percent). The analytical calculation of the
PMS is very complicated and this approximation has therefore been valuable
and is now routinely used. The physical significance of PNS in relation to PMS
has been considered (Andersson et al., 1997). Thermal undulations in the lipid
bilayer of a cubic phase must form standing waves, as a consequence of the
three-dimensional periodicity, and such motions are centred on nodal surfaces.
It was therefore proposed that the PNS description reflects the true dynamic
structure of the bilayer, whereas the closely related PMS structure represents a
hypothetical static structure. An example is the P-surface, which is obtained as
a PNS from the equation:
cos2πx + cos2πy + cos2πz = 0
The dynamic structure with standing wave vibrations along the bilayer,
which was obtained by the nodal surface description, is shown in Figure 3.8. A
breathing vibration mode, located at regions where the freedom for transverse
motions is expected to be highest, is indicated.
If, instead of zero, a constant different from zero is used in the P-surface
equation given above, the surface obtained is a close approximation of a
surface with constant average curvature. It seems likely that the best descrip-
tion of the bilayer structure is obtained by such surfaces centred on the minimal
surface (or nodal surface), not by parallel surfaces, as the hydrocarbon chain
disorder should be expected to vary along the bilayer and be highest at the
‘necks’ (cf. Figure 3.8).
Minimal surfaces in general possess general physical properties which are
significant features of the bicontinuous cubic phases. The hydrostatic pressure
gradient over an interface is equal to:
⎛ 1 1 ⎞
γ ⎜ –– + –– ⎟
⎝ R1 R2 ⎠
where γ is the interfacial tension.
A minimal surface has an average curvature equal to zero everywhere, as

seen above, which means that the pressure gradient is equal to zero.
Bicontinuous cubic structures have been observed as a specific topological
state of cell membrane assemblies, cf. Hyde et al. (1997) and Landh (1996).
The unit cell length is usually much larger than in the pure lipid systems
discussed above, with unit cell axes of the order of magnitude 10 nm. Such
aggregates of different outside shape and finite inner periodicity are exten-
sively discussed by Andersson et al. (1999). Beautiful electron micrographs of
cubic PMS structures have been described recently in endoplasmic reticulum
(without reference to the cubic structure, however) by Snapp et al. (2003), and
in mitochondria by Deng et al. (1999). Similar particles were also prepared in
simple lipid systems and termed cubosomes (Larsson, 1989). They will be
further discussed in Chapter 6. The relations between different cubic phases in
aqueous systems discussed in Section B.4 below are also relevant to cubic
phases in vivo.
The accumulated evidence for the PMS structures of bicontinuous cubic
phases has recently been reviewed (Larsson & Tiberg, 2005).
Finally, two different types of cubic phases in lipids will be mentioned. The
first type is formed by very polar lipids which are water-soluble as micelles.
Such cubic phases consist of lipid micelles arranged in a cubic lattice (Vargas
et al., 1992). The second type is the inverse type; it contains water aggregates
in a lipid matrix, and is formed for example by unsaturated phosphatidylglycerols
(Seddon, 1990). The same inverse cubic phase is also formed in the ternary
system of diacylglycerols together with phosphatidylcholine and water (Orädd
et al., 1995).
4. Membrane protein crystallization in cubic phases

In the understanding of functions of integral membrane proteins, crystal
structure determination is of utmost importance, and a crucial step is the
preparation of single crystals with good diffraction quality. A breakthrough has
been the crystallization of bacteriorhodopsin via the cubic GMO–water phase,
reported by Landau et al. in 1996. A few other membrane proteins have since
also been crystallized by this method.
Recently the molecular mechanism behind the crystallization of
bacteriorhodopsin has been analysed (Nollert et al., 2001). A model is pro-
posed whereby the bacteriorhodopsin molecules aggregate, induced by the
curvature into planar domains; growth of these domains then leads to crystal
nucleation and growth.
5. Identification of liquid-crystalline lipid structures

As mentioned in Chapter 1, the most powerful method for analysing liquid-
crystalline lipid structures is X-ray diffraction/scattering using the small-angle

region (SAXS). The main features of the diffraction patterns will be summa-
rized here. All liquid-crystalline phases are characterized by a diffuse halo
around 4.5 Å, due to the disordered hydrocarbon chains, and a lack of sharp
diffraction in this wide-angle region.
The Lα phase is periodic in one dimension only, and the diffraction spacings
must therefore obey the ratios 1:1/2:1/3:1/4, etc. Usually only a few lines are
seen.
The HI and HII phases are periodic in two dimensions, and the hexagonal
symmetry gives spacings in the ratios 1:1/√3:1/√4, etc. The different alterna-
tives of relation between associated lipids and water compartments, open and
inverse in relation to water, are usually obvious from the phase diagram. The
open phase (HI) occurs in very polar lipid systems with micellar solubility,
contrary to the inverse type (HII).
The different cubic bicontinuous phases are more difficult to identify. It is
therefore recommended to compare the recorded X-ray pattern with earlier
reported diffraction data of the P-, D-, and G-types of structures. The cubic
phases in general show diffraction spacings in the ratios 1:1/√2:1/√3:1/√4:1/
√5:1/√6:1/√8, etc., and the space group symmetry of the P-, D-, and G-surface
structures each require specific diffraction lines to be absent. It should be
mentioned in this connection that the P-surface forms a lipid structure, which
in spite of its name is not primitive but body-centred (with space group Im3m),
due to the bilayer with mirror symmetry over the minimal surface.
B. Phase transitions and phase diagrams
1. Introduction
The phase rule (also called Gibb’s phase rule) defines the relation at equilib-
rium between the number of phases, P, the number of components, C, and the
number of degrees of freedom, F (e.g. temperature, pressure, composition), by
the relation:
P+F=C+2
Applications of the phase rule and its use in determining phase diagrams of
aqueous systems of lipids will be discussed here. Different liquid-crystalline
phases provide different functions, and the significance of knowing the
composition and temperature range where a particular phase exists is obvious.
The self-assembly of lipid molecules in water due to their dual properties in
relation to water (their amphiphilicity) was mentioned in Chapter 1. There are
a few lipid types which are so polar that they are water-soluble in the form of
micellar aggregates. Above a critical micellar concentration (cmc), the
monomers solved in water associate into aggregates which are usually
spherical with a surface exhibiting polar head groups protecting a hydrophobic

core. Most of the lipids that are discussed below are insoluble in water, and
dispersions of the aqueous phases in excess water are therefore an important
aspect, for example dispersions of liposomes discussed in Chapter 6.
2. Why aqueous lipid phases are formed

In a simplified view, the liquid-crystalline lipid–water phases are formed as a
compromise to satisfy water solubility of one part of the molecule and water
insolubility by another part. From this perspective the affinity of the polar head
groups is the driving force behind formation of the liquid-crystalline phases.
The term hydration force was coined by Rand and Parsegian with co-workers
on the basis of measurements of osmotic stress on the Lα phase of phospholipids
(de Neveu et al., 1976). A repulsion force versus distance was thus recorded.
The curve showed an exponential fall-off with a decay length corresponding to
the thickness of a single layer of water molecules. This force was found to be
the predominant one at distances below 20 Å.
The hydration force has been the subject of controversy for some time.
Recently, Israelachvili & Wennerström (1990) proposed that the force origi-
nated from the movement of the lipid molecules in and out from the bilayer
plane. Finger-like protrusions with a density decreasing exponentially from the
polar head group surface would therefore be the reason for the observed
exponential repulsive force.
3. Relation between molecular geometry and structure of lipid–water phases

As mentioned earlier, it is obvious that lipid molecules with a large polar head
group in relation to the hydrocarbon chain region tend to form HI phases,
whereas molecules with small polar heads and a bulky hydrocarbon chain
region tend to form HII phases. Furthermore, in cases of molecular geometry in
between these extremes, Lα phases are favoured, and on a more detailed level
bicontinuous phases are expected for molecular shapes between those corre-
sponding to the Lα and HII phases.
An approach which can quantitatively relate molecular shape to aqueous
lipid phase structure was introduced by Israelachvili et al. (1977). They defined
a packing parameter equal to V/Al, where V is the molecular volume, l is the
molecular length (in the disordered state) and A is the molecular cross-
sectional area at the water-contact surface. The value of V is experimentally
available from densities (partial specific volumes), and l and A can be calcu-
lated from diffraction data. With a packing parameter of 1, the Lα phase should
be expected, and values >1 indicate the existence of an HI phase, whereas
values <1 indicate the existence of HII phases. Values only slightly smaller than
1 might indicate the existence of a bicontinuous cubic phase.
Figure 3.9 A bilayer description of the HII phase is shown to the left, as an alternative to the usual
monolayer model shown to the right.
Helfrich’s early introduction of the curvature elastic energy (Helfrich, 1973)

has been of great importance in the analysis of bilayer-type structures and their
occurrence in vesicles and biomembranes. The following formula describes the
elastic energy per unit area as a function of curvature:
k1H2
––– + k2G
2
where H and G are the average curvature and Gaussian curvature, respectively
(as defined in Section A.3 above), and k1 and k2 are elastic moduli constants.
The HII phase is usually described as water cylinders covered by lipid
monolayers in a hexagonal packing. Alternatively the HII phase can be regarded
as a bilayer phase with a honeycomb shape, i.e. bilayer units shaped as infinite
ribbons intersecting one another into trigonal corners. These two views of the
HII structures are illustrated in Figure 3.9. Both models might be fruitful to
consider when molecular packing is taken into account. The circular cross-
section of water channels in a hexagonal phase formed by a single lipid means
that there is a regular variation in molecular length in different hexagonal
directions, whereas there can be constant length in the other alternative shown
in Figure 3.9 (except at the intersections of bilayer units). Aqueous systems of
alkanes/phosphatidylcholine have been studied by Sjölund et al. (1988). They
demonstrated how alkanes favour the HII phase and reduce structural frustra-
tions by filling the trigonal void space between phosphatidylcholine monolayers
covering the water cylinders.
The effect of temperature on phase transitions is discussed in Section C
below, and the different phase diagrams shown in Section C will illustrate the
effect of water content.
We will now consider the transitions between the cubic phases. It seems
likely that the cubic transitions discussed below are martensitic in their
character; see Hyde et al. (1997).
4. The transition between cubic bicontinuous phases proves their minimal

surface structure – The Bonnet relation
As mentioned above, the remarkable relation between the diamond (D-surface)
and gyroid (G-surface) types of structures in the GMO–water system became
evident when it was realized that the cubic region consists of two phases
(Larsson, 1983). When the first observed spacing was plotted against the
volume fraction of water, a linear relation was observed. The reason for this
close geometric relation between the minimal surfaces is the so-called Bonnet
relation (see Schoen, 1970).
The Bonnet relation means that the Gaussian curvature distributions of two
surfaces at the transition from one to the other are the same. One surface is thus
an isometric conjugate to the other. Hyde & Fogden (1998) analysed the
variations in Gaussian curvature along the surfaces and introduced the concept
of Gaussian curvature inhomogeneity. Due to the Bonnet relation all three
types of surface (G, D and P) have the same value of calculated inhomogeneity.
These three are the most homogeneous of all known minimal surfaces, and this
may explain why these are the only ones observed in lipid–water systems; they
represent a minimum in bilayer frustration.
In general the water swelling in liquid-crystalline phases, such as the
lamellar phase, is known to increase the disorder (as reflected in the molecular
cross-sectional area per molecule). In any of the three types of bilayer cubic
phases (G, D and P), the disorder will on the contrary decrease with water
swelling (the bilayer will become increasingly planar, as reflected by Gaussian
curvature changes). Obviously water swelling will build up strain, and as
shown below the phase transitions between Bonnet-related phases will reduce
this strain.
Schoen (1970), a pioneer behind minimal surface theory, introduced a
normalized surface-to-volume ratio equal to S/V2/3, with values of 2.4177,
2.4533 and 2.3451 for the D, G and P surfaces, respectively. If swelling of the
G-phase in the GMO–water system builds up bilayer frustration, a transition
with water swelling would ideally result in a higher water content per bilayer
unit volume. This is in fact what happens in the G→D structural transition. The
two-phase region (see below) is quite narrow, and seems to correspond to the
situation where the bilayers on each side of the transition range are equivalent,
i.e. have the same Gaussian curvature distribution. Thus, assuming that the two
cubic phases are Bonnet-related, the water content will increase by about 1.5%
at the transition. This is consistent with the experimental observations (Hyde et
al., 1984).
The G→D transition on swelling corresponding to a Bonnet transformation
requires that the ratio of the unit cell axes is 1.58. The agreement according to
X-ray data of the two cubic phases just on each side of the transition was good,
and this was taken as strong evidence for the proposed minimal-surface types
of structures (Hyde et al., 1984).
The D→P transition in GMO–water systems containing proteins or other

polymers discussed later in this chapter can be understood along the same lines.
Also in this case the strain built up during swelling can be reduced by the
transition, as the same bilayer structure can accommodate more water after the
transition (see the normalized surface-to-area values given above). Other
systems with coexisting cubic phases show similar agreement, as shown by
Hyde & Fogden (1998). Thus the D- and P-phases of the ternary system
didodecylammoniumbromide/cyclohexane/water shows a unit cell axis ratio in
agreement with 1.28, which corresponds to the Bonnet relation in the case of a
D→P type of phase transition. Recently the accumulated evidence from all
reported cubic phase transitions in lipid–water systems was examined and
shown to be fully consistent with the Bonnet requirements (Larsson & Tiberg,
2005).
Although the bilayer mid-surface only is equivalent to the minimal surface,
the Bonnet relations hold even when the whole bilayer is considered, provided
that the bilayer is assumed to have constant thickness (with polar head groups
forming surfaces parallel to the minimal surface).
Cubic phases tend to form single-crystal domains over large volumes – often
throughout the whole bulk of a cubic sample. When there is a phase transition
involving two minimal-surface types of structures, such as the G→D transition
on water swelling of GMO, it can be seen how a sharp front of one phase moves
through the other; see Figure 3.10. The interface is hard to detect as the
refractive indices of the two phases must be very similar.
An interfacial structure between the lipid bilayer of the cubic phase and the
water phase in Figure 3.10, which seems likely, is illustrated in Figure 3.11.
The reason is the favouring of water diffusion by such an interface between
G-phase and water and then between D-phase and water (Figure 3.10). The
bilayer must close towards the water phase in order to avoid water contact with
the bilayer interior. The closing zone can be seen as local formation of an Lα-
type of bilayer.
The sharp front between the two coexisting phases in Figure 3.10 might have
a similar interfacial structure, with an Lα- type of bilayer on each side of an Lα-
type of water film forming the gap between the phases. It should be pointed out
Figure 3.10 Phase transition behaviour when water is added to the GMO–water G-phase at its
maximum swelling. The shaded region illustrates the D-phase successively formed.
Figure 3.11 Proposed interface structure between a G-phase or a D-phase bilayer and a water phase.
The curved double line illustrates the bilayer, and closing of the bilayer structure outwards towards water
by an Lα type of bilayer is indicated by the thick lines at the top.
in this connection that the surface structure depicted in Figure 3.11, with rod-
shaped water channels perpendicularly oriented to the surface, has hexagonal
symmetry when viewed along these channels. Both the G-phase and the
D-phase have closely related geometry and symmetry along this plane, which
is a strong argument for proposing that this plane corresponds to the interface
between the D-phase and the G-phase.
5. Other lipid–water phases

So-called intermediate phases have been found in simple surfactant–water
systems (i.e. binary systems where micellar solutions exist at high water
content), and with increased surfactant concentration the HI phase and the Lα
phase are formed. These intermediate phases exist between the HI and the Lα
phases, and their structures represent successive changes in molecular organi-
zation (Kekicheff & Cabane, 1988).
A different type of aqueous phase is formed by lung surfactants. Recent cryo-
transmission electron microscopy studies of the alveolar lining of mammalian
lungs have shown that the surface layer consists of a coherent phase, which was
proposed to consist of a bilayer with a tetragonal minimal surface structure (M.
Larsson, 2002). A unit of the structure is shown in Figure 3.12. This surface is
the mid-surface of a continuous phospholipid bilayer, free from self-intersec-
tions. The structure corresponds to a texture earlier identified from lung
washings under the name tubular myelin, which according to electron
microscopy studies has been assumed to consist of intersecting bilayers with
square-shaped cross-sections. Beside phospholipids, the lung surface bilayer
consists of a few percent of two hydrophobic proteins with a tendency to
associate apposing bilayers (cf. Chapter 10).
Figure 3.12 The structure unit of the proposed tetragonal minimal surface organization of the lipid
bilayer lining the alveolar surface, reproduced by permission (M. Larsson, 2002).
C. Phase behaviour of different lipid–water systems
1. Monoacylglycerols
We will start with GMO, which has been the subject of numerous recent studies
of the structure of the bicontinuous cubic phases. The first complete phase
diagram is shown in Figure 3.13 (Hyde et al., 1984). Qui & Caffrey (2000) later
also described this phase diagram at temperatures below 20ºC, where there is
complicated behaviour involving metastable phases. Their diagram also shows
some minor deviations at higher temperatures from that shown in Figure 3.13.
The phase transition sequence obtained by heating illustrates the effect of
increased thermal mobility of the hydrocarbon chain; increasing proportions of
gauche conformations correspond to increased chain divergence. The packing
parameter values at the Lα→cubic and cubic→HII phase transitions can be
calculated from available data on the composition and on the corresponding
X-ray data. The effect of water content on the transitions will be discussed in
the last section of this chapter.
A further illustration of the effect of molecular geometry on the phase
behaviour is illustrated in Figure 3.14, where saturated 1-monoacylglycerols
from C8 to C20 are compared. It is obvious that the effect of chain divergence
will increase with chain length; therefore, the hexagonal HII phase grows at the
expense of cubic phases, which in turn grow at the expense of the lamellar Lα
phase.
In many technical applications, industrially distilled monoacylglycerol mix-
tures are used. A common type is obtained from fully hydrogenated lard, which
Figure 3.13 Phase diagram of the glycerol monooleate (GMO)/water system (after Hyde et al., 1984).
Equilibration of the phases was started from the 1-isomer. The two cubic phases are termed the G-phase
and the D-phase, after the corresponding minimal surface structures. The liquid L2 phase is discussed in
the next chapter. Two-phase regions are shaded.
has a fatty acid composition dominated by stearic acid. The phase diagram of
such a mixture behaves like that of a pure component (Krog & Larsson, 1968).
The transition temperatures are very close to those of 1-monopalmitin (Larsson,
1967). The presence of impurities in the form of charged lipid species results in
increased swelling.
2. Phospholipids
Phospholipids, especially phosphatidylcholines, have been the subject of
extensive studies concerning their aqueous interaction and phase properties,
because of their significance in biomembrane physiology. This is illustrated by
several reviews in this field. A special issue of the journal Chemistry and
Physics of Lipids has for example been devoted to phospholipid phase transi-
tions (Volume 57, 1991).
Phosphatidylcholines (PCs)
There is probably no natural lipid that has been studied as thoroughly as egg
yolk PC. The phase diagram is illustrated in Figure 3.15. At a very low water
content, the phase behaviour is very complex, and it is omitted here as it is
hardly relevant in any application. The lamellar Lα phase dominates the phase
diagram.
At low water content, however, the transitions Lα→ cubic→HII are seen,
Figure 3.14 The main features of the aqueous phase diagrams of the saturated monoacylglycerols from
C8 to C20: (a) C8:0–C12:0, (b) C14:0–C18:0, (c) C20:0.
which take place in monoacylglycerols at higher water content. We can still

explain these transitions in the same way: the driving force is the tendency for
increased chain divergence when the thermal mobility increases. Although egg
yolk PC is a mixture of several molecular species, it behaves as one component.
Other PC mixtures of biological origin, such as PC from soybeans, behave in an
almost identical way. Pure synthetic PCs show a similar behaviour.
The phase diagram of dipalmitoyl-PC is shown in Figure 3.16. Apart from
indicating a higher chain-melting temperature, the diagram is quite similar to
that of egg yolk PC, with a dominating Lα phase and the transition sequence
Lα→cubic→HII with increasing temperature at low water content. The transi-
tion region from ordered to disordered hydrocarbon chains can be rather
complicated, as discussed below.
Figure 3.15 Phase diagram of egg yolk PC. C denotes a cubic phase region and S denotes the solid state
(after Small, 1986).
Figure 3.16 Phase diagram of dipalmitoyl-PC (after Cevc, 1991).

Figure 3.17 Main features of the phase diagram of water/C6:0-C16:0-PC. Dotted regions contain two
phases (after Svensson et al., 1993).
When the Lα phase is cooled, it forms a gel phase on crystallization of the

chains. In a small region of the gel phase existence range (close to the Lα
transition and at a water content of 20–30%, w/w) the so-called ripple phase is
obtained, a gel phase with a wave-shaped periodicity along the bilayers (Janiak
et al., 1976). The transition temperature from the Lα phase to the gel phase on
cooling changes from about 41ºC to about 23ºC when the acyl chain length is
reduced from C16 (disaturated) to C14 (disaturated).
The main features of a phase diagram of a PC with one C6:0 chain and one C16:0
chain are shown in Figure 3.17. In this study (Svensson et al., 1993), one chain
was C16:0 and the other chain was successively shortened. The general features
of the phase diagram were similar to that of dipalmitoyl-PC down to C8:0. With
C6:0, however, a quite different diagram was obtained. On swelling of the Lα
phase a hexagonal phase was obtained, and further swelling resulted in a liquid
phase with characteristics of the L3 phase discussed in the next chapter.
Phosphatidylethanolamines (PEs)
Seddon et al. (1984) have reported the phase behaviour of the aqueous systems
of the dialkyl-PE didodecyl-PE, and of the diacyl-PE diarachidonoyl-PE.
Didodecyl-PE in excess water transforms into the Lα phase at about 40ºC, and
at about 100ºC the Lα phase transforms into a cubic phase. At higher tempera-
tures, above 120ºC, the HII phase is formed. The diacyl-PE also forms an Lα
phase in a very narrow temperature interval, and it is transformed directly into
the HII phase on heating. The binary system dipalmitoyl-PE–water is shown in
Figure 3.18. This diagram is rather similar to that of dipalmitoyl-PC–water
(Figure 3.16) discussed above.
There are also reports on the aqueous phase properties of unsaturated PEs
(e.g. see Lindblom & Rilfors, 1989).
When PE–water Lα phases are cooled to temperatures below the chain
crystallization temperature, crystals separate directly in the water phase
Figure 3.18 Phase diagram of the binary system dipalmitoyl-PE/water, drawn by combining data from
Cevc (1991) and Seddon et al. (1984). The unknown region at low water content is shaded.
(‘coagel’) – no gel phase is obtained. This difference compared to the behav-

iour of PC has been related to the cross-sectional area of the molecules, which
reflects the polar head group packing (Hauser et al., 1981). In PCs the cross-
sectional area is 47–54 Å2/molecule, which is so close to that of the Lα form that
a transition to a gel phase is possible without breaking the head-group lattice.
This is not possible in PEs, however, where the cross-sectional area is 35–
42 Å2/molecule.
PEs, like PCs, are zwitterionic, and they therefore become anionic at high pH
and cationic at low pH. In the ionic states these lipids behave like the charged
phospholipids described below.
Phosphatidylinositols (PIs)
The phase diagram of soybean PI–water is shown in Figure 3.19. The acyl
chains are dominated by C 16:0 and C18:2. The Lα phase dominates the
investigated region (room temperature to 70ºC) with a swelling limit of about
77% (w/w) water.
Phosphatidylglycerols (PGs), phosphatidylserines (PSs) and phosphatidic

acids (PAs)
The aqueous phase properties of these lipids are mainly due to the electrostatic
repulsion between the polar head groups, as in the PI–water system; the pH and
counter-ions therefore have strong effects. When the pH is so low that these
Figure 3.19 Phase diagram of the binary system consisting of the sodium salt of soybean PI/water
(after Söderberg, 1990).
lipids become non-ionic, they tend to form HII phases (Hope & Cullis, 1980);
divalent counter-ions like calcium (which contract the polar sheets laterally)
also tend to give HII phases (Harlos & Eibl, 1980; Farren et al., 1983).
In mixed phospholipid systems it is fruitful to consider the average molecu-
lar shape, and whether or not there is an average net charge. An illustrative
example is the mixed system DOPE (dioleoyl-phosphatidylethanolamine) plus
DOPC (dioleoyl-phosphatidylcholine) and milk sphingomyelin, which was
reported recently (Waninge et al., 2003). The phases observed in this system
are the same as those exhibited by each of the components, and their regions in
the phase diagram can be explained by the average molecular shape.
Lysophospholipids
Phospholipases A1 and A2 split off the acyl chains in the 1- and 2-positions
respectively. Phospholipase A2 is particularly important from a physiological
point of view as phospholipids in biological tissues often have arachidonic acid
in the 2-position, and this fatty acid can initiate a cascade of inflammatory
reactions. Lysophospholipids are water soluble in micellar form, and these
micelles can solubilize and destroy cell membranes. Phospholipase degrada-
tion of membrane phospholipids into fatty acids and the lyso compounds is the
reason why snake venom can cause haemolysis.
Arvidsson et al. (1985) have studied lyso-PCs of different chain length
(saturated and unsaturated chains), and found cubic phases in addition to the
expected HI and Lα phases.
Figure 3.20 Molecular arrangement in the different phases of cerebroside from bovine brain and in the
corresponding psychosine formed when the acyl chain is removed (after Abrahamsson et al., 1972).
3. Sphingolipids
Sphingomyelin (SM) is rather similar to PC from a chemical point of view (see
Chapter 1). The SM–water system is also quite similar to the PC–water system
with a corresponding chain melting point (dipalmitoyl-PC, for example, is
similar to milk SM in this respect). The Lα phase dominates, and the zwitterionic
character results in a maximum water layer thickness of about 20 Å. The
saturated and unsaturated members examined behave in a similar way except
for the difference in chain melting temperature (Shipley et al., 1985; Sripada et
al., 1987).
Cerebrosides are sphingolipids with a sugar group attached to the polar head
group. They occur in the brain and their aqueous phases have been studied by
Abrahamsson et al. (1972); see Figure 3. 20. When the acyl group is removed,
the detergent-like molecule that is formed is termed psychosine, the aqueous
phase behaviour of which is also shown in Figure 3.20.
When a hydroxyl group of the sugar in cerebrosides is exchanged for a
sulfate group, sulfatides are obtained. These form micellar solutions in water
(Abrahamsson et al., 1972) and behave in a similar way to another type of lipid
containing a sugar group: the gangliosides, which were described in Chapter 1.
Gangliosides are also soluble in micellar form (Curatolo et al., 1977).
4. Glyceroglucolipids/galactolipids
This type of lipid is particularly abundant in plant cell membranes, and is also
often found in microorganisms. Monogalactosyl-diacylglycerols (MGDG)
form HII phases with excess water, whereas digalactosyl-diacylglycerols
(DGDG) form Lα phases. This general behaviour was observed in MGDG and
DGDG from plant leaves (Shipley et al., 1973), from wheat endosperm
(Larsson & Puang-Ngern, 1979) and from mycoplasma (Wieslander et al.,
1978).
The thylakoid membranes of chloroplasts (where photosynthesis takes place)
contain MGDG and DGDG as major components. It is therefore most interest-
ing to note that in the vegetative state of chloroplasts, the lamellar stacks of flat
membrane sacs have been transformed into a cubic phase (Hyde et al., 1997).
In the mycoplasma membrane, a transition between the Lα phase and the HII
phase appears to occur when approximately equal proportions of MGDG and
DGDG are present (Lindblom & Rilfors, 1989).
5. Lipid–glycerol systems
There are also water-free lipid systems that behave like aqueous systems; these
include polar liquids, which allow hydrogen bonding and dipole–dipole inter-
actions like water. Glycerol is an example considered here. The binary system
GMO–glycerol exhibits the same phases as GMO–water, with similar exist-
ence ranges in the phase diagram (K. Larsson, unpublished data). The transition
temperatures are slightly higher, and the kinetics of the phase changes much
slower. Proteins such as lysozyme can be incorporated and they show similar
denaturation behaviour as in the corresponding aqueous phase. Phases with Lα
structure and L2 phases were prepared from galactolipids and phospholipids
extracted from oats. It was found that oxidation appeared to be almost totally
inhibited in such phases, swollen in glycerol. The reason might be that lipases
in this environment become inactive and lipoxidases require free fatty acids as
substrate. Possibilities for replacing water by an alternative non-toxic sub-
stance, such as glycerol, in technical applications of liquid-crystalline phases
may have been neglected. From a fundamental point of view also it is
remarkable that many different types of lipids, alone or in combination with
other types of biomolecules, show similar behaviour in glycerol as in an
aqueous environment.
6. Simplified model diagrams of lipid–water systems

There are three main types of phase diagrams of binary lipid–water systems.
Strongly polar lipids such as lysophospholipids are water soluble in micellar
forms. When the concentration is increased, the HI phase and then the Lα phase
is formed. Cubic intermediate phases may also exist.
Figure 3.21 Simplified model diagram of one-phase regions of aqueous systems of (left) non-ionic
lipids and (right) ionic or zwitterionic lipids. The limit of swelling is shifted to higher water concentra-
tions in ionic compared to non-ionic lipids, but can be reduced by counter-ions. The hatched regions
represent two phases in equilibrium.
The other two types of phase diagrams are illustrated in Figure 3.21. A
situation was chosen where the three types of phases – cubic, hexagonal and
lamellar – exist in the same system. Only the structures that exist above the
chain melting temperature are considered.
It is interesting to note the opposite tilt direction of the phase boundaries in
the two types of system (the reason why we have classified them in this way).
In non-ionic lipids there is a tendency for the packing parameter to change with
both increased water content and increased temperature, which may induce
Lα→cubic bicontinuous→HII transitions. The increased disorder with increased
water content is considered to be due to increased motion of the polar head
group in and out in relation to the end group surface. Such disorder would also
increase the gauche to trans ratio along the chains. This is confirmed by
diffraction data; the bilayer thickness decreases with increasing water content.
When the polar head groups are held together by ionic forces, the effect of
increased water content in the phase will be quite different. Let us consider a
lipid with one charged group. The lateral electrostatic repulsion between the
polar head groups will be reduced if the distance between them within the polar
sheet is increased. Such a change could take place with an increase in water
content, when water penetrates the polar sheets as well as forming a gap
between the bilayers. This would be equivalent to a change with water
concentration in the sequence HII→cubic→Lα. The effect of increased tem-
perature, however, would be the same as in non-ionic lipids, as discussed
above. This explains the difference in tilt direction of the phase boundaries in
Figure 3.21.
The limit of swelling is similar in zwitterionic and non-ionic lipids, corre-
sponding to a water layer thickness of about 20 Å. In the case of ionic lipids,
where the limit of swelling is increased (related to the electric double-layer),
the presence of a few percent of an ionic lipid in predominantly non-ionic lipid

bilayers is enough to drastically increase the water swelling limit.
D. Main features of phase diagrams involving different types of

lipids
1. Introduction
There are now numerous phase diagrams involving two or more lipids reported
in the literature. Here we will only consider some diagrams of general rel-
evance which can be helpful in order to understand the main features of
multicomponent lipid–water systems.
The binary phase diagrams discussed earlier in this chapter show the effects
of water content as well as temperature. If the number of components is
increased to three, and we want to show the effects of composition, we have to
select a particular temperature and illustrate the composition by a triangle. As
mentioned above, there is a tendency for mixtures within a certain lipid class
(such as soybean triacylglycerols) to behave like one component in relation to
the phase rule. This simplifies the determination of phase diagrams of lipids,
and is most important when attempting to apply basic knowledge from phase
diagrams to discussions of lipid functionality. We will start with examples with
applications in biology (gastrointestinal fat digestion) and technology (peroral
drug delivery).
2. Ternary and quaternary phase diagrams – some general features

The phase diagram of a triacylglycerol oil, monoacylglycerols and water is
shown in Figure 3.22. With solubilization of the triacylglycerol molecules, the
transition sequence is:
Lα→cubic bicontinuous→HII→L2.
The L2 phase, a kind of microemulsion, is described in the next chapter.
If we now introduce a fourth component, a saponin, a tetrahedron is needed
in order to define the composition. By selecting a certain triacylglycerol
composition, a ternary phase diagram as shown in Figure 3.23 is obtained.
Figures 3.22 and 3.23 together show the two most important tetrahedron faces,
and provide some insight into the three-dimensional diagram. Saponin forms a
micellar solution in water, and can only solubilize minor amounts of
triacylglycerol oil. Therefore the tetrahedron face with corners representing
saponin, water and triacylglycerol oil shows only two liquid phases. The water-
free fourth face of the tetrahedron shows only an oil phase and a solid phase.
An interesting feature of this system is the narrow existence range of the
lamellar Lα phase (indicated by D on Figure 3.23), with an almost constant ratio
Figure 3.22 Ternary phase diagram at 40ºC for soybean oil triacylglycerols (TG)/water/sunflower oil
monoacylglycerols (MG) (after Lindström et al., 1981). The two-phase region L2+water is dotted, and
the liquid-crystalline (LC) region consists of a lamellar liquid-crystalline phase, a cubic phase, and a
hexagonal HII phase.
of saponin:monoacylglycerol. Ekwall and co-workers introduced the name D

for the lamellar phase in the 1960s, and its use in surfactant literature is
therefore common. The adjacent L2 phase is also remarkable as it is succes-
sively transformed into an L1 phase (ordinary micellar solution).
The ternary phase diagram for ethyl acetate/GMO/water was recently deter-
mined with high accuracy (Imberg, 2003). The regions of L2 and
liquid-crystalline phases are remarkably similar to those of the ternary system
triacylglycerols/GMO/water shown in Figure 3.22, although ethyl acetate is an
organic solvent. A remarkable difference, though, is that no HII phase was
observed with ethyl acetate.
Small amphiphilic molecules are often present in technical applications of
lipids, providing pronounced effects on lipid–water phase equilibria. The
ternary systems of PC and water with ethanol and n-butanol are shown in
Figures 3.24 and 3.25 in order to illustrate such effects.
Only a small proportion of ethanol is needed in order to form an ethanol
continuous phase, and this phase is successively transformed into a water
continuous phase. Another interesting feature is the behaviour of the liposomal
dispersion. The stability of dispersed liquid-crystalline particles in a water
environment is strongly reduced by ethanol.
Among the most efficient solvents for the extraction of lipids from biological
tissues is so-called water saturated n-butanol (WSB). A standard procedure is
to start with 20% (v/v) water in n-butanol, which forms two phases. Then only
trace amounts of lipids obtained from the extraction give a single solvent phase.
Figure 3.23 Phase diagram for saponin/water/monoacylglycerols from sunflower oil (Barla et al.,
1979). Regions with saponin crystals are indicated by black, I shows the cubic region, and D indicates
the lamellar Lα phase region. Regions containing two or three phases are shaded. There is a continuous
transformation of the L2 phase into an L1 phase, and it is therefore not possible to strictly show the
corresponding phase boundaries. Position A in the liquid region represents an L1 phase, which towards
position B becomes an L2 type. The line B–C exhibits a remarkable similarity in extension to that of the
region of the lamellar phase (D), both having stoichiometric monacylglycerol:saponin ratios.
This behaviour can be explained by the phase diagram. There is one single
liquid phase from the water corner to the butanol corner, which might be
regarded as an L1 phase transforming into an L2 phase. Another interesting
feature is the transition from an Lα phase to an HI phase with increasing
solubilization of n-butanol, an effect of changes in the average molecular
shape. The effects of butyric acid were found to be similar to those of butanol.
This might be of physiological interest as this acid is produced by fermentation
in the colon, and by penetration provides nutrition to the epithelial cells.
The relations between liquid-crystalline phases and liquid lipid–water phases,
as reflected in phase diagrams, should be briefly mentioned here. We have seen
above that L2 phases often occur near Lα phases, and that their structures are
also related (with partial ‘melting’ and loss of crystallographic order). In a
Figure 3.24 Phase diagram for soybean phosphatidylcholine (SPC)/ethanol/water at room temperature
(after Söderberg, 1990).
Figure 3.25 Phase diagram for egg yolk PC/n-butanol/water (I. Söderberg, unpublished results). The
large empty region consists of one liquid phase, the dotted region is an L2 phase, and the black region is
an HI phase. Two-phase regions are shaded.
similar way the L3 phases can be regarded as ‘melted’ bicontinuous cubic

phases. An example of such a system is shown in Figure 3.26. The liquid phases
L2 and L3 are further considered in the next chapter.
If the phases towards the water corner in Figure 3.26 are considered, the
cubic phase (Q) can solubilize some propylene glycol (PG), and then an L3
Figure 3.26 The ternary phase diagram for propylene glycol (PG)/GMO/water at room temperature
(after Engström et al., 1998). The cubic phase is termed Q here.
phase is formed which has an almost stoichiometric water:PG ratio. The L3

phase appears almost as a continuation of the cubic phase. A similar phase
diagram was observed for the GMO/water/1-methyl-2-pyrrolidone (MP) sys-
tem (Johansson et al., 2001). At a similar stoichiometric water:MP ratio there
is an L3 phase over a large concentration interval, starting from the cubic phase.
The connection with the cubic phase and the L3 phase is even more striking
here, and may tell us something about the structure of the L3 phase.
E. Cholesterol esters
Fatty acid esters of cholesterol are important from a physiological point of
view; they are found in particles that carry lipids in the circulation associated
into lipoproteins. Phase properties of lipid systems involving cholesterol esters
have been studied by Small & co-workers (e.g. see Small, 1986). The liquid-
crystalline phases are unique in their structure compared to other lipids. Before
melting they can form smectic, cholesteric, and blue phases. The cholesterol
skeletons can pack in layers with a helical twist when going from one layer to
the next. The structural changes are very sensitive to temperature variations,
and this can be observed as colour variations and used as a temperature
indicator in certain esters.
The blue phases are cubic, and they obtain their blue colour due to the size
of the structure repetition unit, with a unit cell length that interferes with the
wavelength of blue light. Three blue phases (BPI, BPII and BPIII) can form
successively on heating. The blue phases BPI and BPII have been shown to
have structures corresponding to the gyroid and diamond minimal surface
structures, respectively, and the lattice parameters are also in agreement with
those required by a Bonnet transformation (Hyde & Fogden, 1998).
References
Abrahamsson, S., Pascher, I., Larsson, K. & Karlsson, K-A. (1972) Chem. Phys. Lipids 8,
152.
Andersson, S., Hyde, S.T., Larsson, K. & Lidin, S. (1988) Chem. Rev. 88, 221.
Andersson, S., Larsson, K. & Jacob, M.Z. (1997) Kristallogr. 212, 5.
Andersson, S., Larsson, K., Larsson M. & Jacob, M. (1999) BIOMATHEMATICS –
Mathematics of Biostructures and Biodynamics, Elsevier, Amsterdam, the Netherlands.
Arvidsson, G., Brentel, I., Kahn, A., Lindblom, G. & Fontell, K. (1985) Eur. J. Biochem. 152,
753.
Barla, P., Larsson, K., Ljusberg-Wahren, H., Norin, T. & Roberts, K. (1979) J. Sci. Food
Agric. 30, 864.
Baumgart, T., Hess, S.T. & Webb, W.W. (2003) Nature 425, 821.
Cevc, G. (1991) Chem. Phys. Lipids 57, 293.
Curatolo, W., Small, D.M. & Shipley, G.G. (1977) Biochim. Biophys. Acta 168, 11.
de Neveu, D.M., Rand, R.P., Ginger, D. & Parsegian, A. (1976) Science 191, 399.
Deng, Y., Marko, M., Buttle, K.F., Leith, A.D., Mieczkowski, M. & Manella, C.A. (1999)
J. Struct. Biol. 127, 231.
Engström, S., Alfons, K., Rasmusson, M. & Ljusberg-Wahren, H. (1998) Progr. Colloid
Polymer Sci. 108, 93.
Farren, S.B., Hope, M.J. & Cullis, P.R. (1983) Biochem. Biophys. Res. Commun. 111, 675.
Harlos, K. & Eibl, H. (1980) Biochemistry 19, 895.
Hauser, H., Pascher, I. Pearson, R.H. & Sundell, S. (1981) Biochim. Biophys. Acta 650, 21.
Helfrich, W. (1973) Z. Naturforsch. A28, 693.
Hope, M.J. & Cullis, P.R. (1980) Biochem. Biophys. Res. Commun. 21, 846.
Hyde, S.T., Ericsson, B., Andersson, S. & Larsson, K. (1984) Z. Kristallogr. 168, 213.
Hyde, S.T., Andersson, S., Larsson, K., Blum, Z., Landh, T., Lidin, S. & Ninham, B. (1997)
The Language of Shape. The Role of Curvature in Condensed Matter Physics, Chemistry
and Biology, Elsevier, Amsterdam, the Netherlands.
Hyde, S.T. & Fogden, A. (1998) Progr. Colloid Polymer Sci. 108, 139.
Imberg, A. (2003) On Phase Behaviours in Lipid/Polymer/Solvent/Water Systems and their
Application for Formation of Lipid/Polymer Composite Particles. Thesis, University of
Uppsala, Sweden.
Israelachvili, J.N., Mitchell, D.J. & Ninham, B.W. (1977) Biochim. Biophys. Acta 470, 341.
Israelachvili, J.N. & Wennerström, H. (1990) Langmuir 6, 873.
Janiak, M.J., Small, D.M. & Shipley, G.G. (1976) Biochemistry 15, 4575.
Johansson, A.K., Linse, P., Picurell, L. & Engström, S. (2001) J. Phys. Chem. B105, 12157.
Kekicheff, P. & Cabane, B. (1988) Acta Crystallogr. B44, 395.
Krog, N. & Larsson, K. (1968) Chem. Phys. Lipids 2, 129.
Landau, E.M. & Rosenbusch, J.P. (1996) Proc. Natl Acad. Sci. USA 93, 14532.
Landh, T. (1996) Cubic Cell Membrane Architecture. Thesis, Lund University, Sweden.
Landh, T. (1994) J. Phys. Chem. 98, 8453.
Larsson, K. (1967) Z. Phys. Chem. (Frankfurt am Main) 56, 173.
Larsson, K. (1972) Chem. Phys. Lipids 9, 181.
Larsson, K. & Puang-Ngern, S. (1979) In: Advances in the Biochemistry and Physiology of
Plant Lipids (Appelquist, L-Å. & Liljenberg, C., eds), Elsevier, Amsterdam, the Nether-
lands, p.27.
Larsson, K., Fontell, K. & Krog, N. (1980) Chem. Phys. Lipids 27, 321.
Larsson, K. (1983) Nature 304, 664.
Larsson, K. (1989) J. Phys. Chem. 93, 7304.
Larsson, M. (2002) A Surface Phase Model of The Alveolar Lining: Ultrastructural Analysis
and In Vivo Applications. Thesis, Lund University, Sweden.
Larsson, M., Larsson, K., Nylander, T. & Wollmer, P. (2003) Eur. J. Biophys. 31, 633.
Larsson, K. & Tiberg, F. (2005) Curr. Opin. Colloid Interface Sci. 9, 365.
Lindblom, G. & Rilfors, L. (1989) Biochim. Biophys. Acta 988, 221.
Lindblom, G., Larsson, K., Johansson, L., Fontell, K. & Forsen, S. (1979) J. Am. Chem. Soc.
101, 5465.
Lindström, M., Ljusberg-Wahren, H., Larsson, K. & Borgström, B. (1981) Lipids 16, 749.
Longley, W. & McIntosh, T.J. (1983) Nature 303, 612.
Luzzati, V., Mustacchi, H., Skoulios, A. & Husson, F. (1960) Acta Crystallogr. 13, 660.
Luzzati, V. & Spegt, P.A. (1967) Nature 215, 710.
Luzzati, V., Tardieu, A., Gulik-Krzywicki, T., Rivas, E. & Reiss-Husson, F. (1968) Nature
220, 485.
Luzzati, V. (1997) Curr. Opin. Struct. Biol. 7, 661.
Mortensen, K., Pfeiffer, W., Sackmann, E. & Knoll, W. (1988) Biochim. Biophys. Acta 945,
221.
Nollert, P., Qui, H., Caffrey, M., Rosenbusch, J.P. & Landau, M. (2001) FEBS Lett. 504, 179.
Orädd, G., Lindblom, G., Fontell, K. & Ljusberg-Wahren, H. (1995) Biophys. J. 68, 1865.
Qui, H. & Caffrey, M. (2000) Biomaterials 21, 223.
Rappolt, M., Pabst, G., Rapp, G., Kriechbaum, M., Amenitsch, H., Krenn, C., Bernstorff, S.
& Laggner, P. (2000) Eur. Biophys. J. 29, 125.
Schoen, A.H. (1970) NASA Technical Report No. 05541.
Scriven, L.E. (1976) Nature 263, 123.
Seddon, J.M. (1990) Biochim. Biophys. Acta 1031, 1.
Seddon, J.M., Cevc, G., Kayl, R.D. & March, D. (1984) Biochemistry 23, 2634.
Shipley, G.G., Green, J.P. & Nichols, B.W. (1973) Biochim. Biophys. Acta 311, 531.
Shipley, G.G., Avecilla, L.S. & Small, D.M. (1985) Biochemistry 24, 2902.
Simons, K. & Ikonen, E. (2000) Science 290, 1721.
Sjölund M., Rilfors, L. & Lindblom, G. (1988) Biochemistry 28, 1323.
Small, D.M. (1986) Handbook of Lipid Research. 4. The Physical Chemistry of Lipids,
Plenum Press, New York, USA.
Snapp, E.L., Hegde, R.S., Francoline, M., Lombardo, F., Colombo, S., Pedrazzini, E.,
Bergese, N. & Lippincott-Schwartz, J. (2003) J. Cell. Biol. 163, 257.
Söderberg, I. (1990) Structural Properties of Monoglycerides, Phospholipids and Fats in
Aqueous Systems. Thesis, University of Lund, Sweden.
Sripada, P.K., Malik, P.R., Hamilton, J.A. & Shipley, G.G. (1987) J. Lipid Res. 28, 710.
Svensson, I., Adlercreutz, P., Mattiasson, B., Miezis, Y. & Larsson, K. (1993) Chem. Phys.
Lipids 66, 195.
Vargas, M., Mariani, P., Gulik, A. & Luzzati, V. (1992) J. Mol. Biol. 225, 137.
von Schnering, H.G. & Nesper, R. (1991) Z. Phys. B Condensed Matter 83, 407.
Waninge, R., Nylander, T., Paulsson, M. & Bergenståhl, B. (2003) Chem. Phys. Lipids 125,
59.
Wieslander, Å., Ulmius, J., Lindblom, G. & Fontell, K. (1978) Biochim. Biophys. Acta 512,
241.
CHAPTER 4
The liquid state
The association of surfactant molecules into micelles in an aqueous solution

was mentioned in Chapter 3. An inverse type of structure, formed by water
aggregates in a polar lipid continuum, was also indicated in the phase diagrams.
These two micellar liquid solutions are termed L1 and L2 respectively. The L1
solution starts from the water corner of the corresponding phase diagram,
whereas the L2 solution starts from a lipid corner. A third liquid phase has also
been found in some phase diagrams, without contact with the water corner or
lipid corner. This phase has been termed L3.
The L1 phase occurs only in aqueous systems of very polar lipids, such as
those involving lysophospholipids. The micelles are formed above the so-
called critical micellar concentration (cmc), and the micelles are usually
spherical with a core of highly disordered hydrocarbon chains. At higher
concentrations these micelles usually fuse into rods.
The liquid state of pure lipids will first be considered, and then the L2 and L3
states.
A. Liquid triacylglycerols
It is known that the melting of a fat into an oil may keep a ‘memory’ of the
crystal structure before melting, and in that way influence the polymorphic
behaviour. On the basis of X-ray scattering data from liquid triacylglycerols,
the persistence of bilayer regions after melting, as shown in Figure 4.1, has
been proposed. Thus a band in the X-ray scattering curve with an integrated
intensity value almost as high as that of the first-order diffraction line has been
observed. At nucleation, the organization within these associated regions
should therefore be expected to influence the structure of the crystallization
nuclei.
The freeze-fracture freeze-etching electron micrograph of a disordered lipid
melt is shown in Figure 4.2, and the corresponding texture of a triacylglycerol
melt is shown in Figure 4.3. Freeze-fracturing is an informative electron
microscopy method for structure studies of lipids. A sample fractured in frozen
condition will open up surfaces along planes where the molecular interaction
forces are weakest, which are the methyl end group planes/regions.
107
Figure 4.1 Proposed bilayer arrangement in the liquid state of triacylglycerols after melting (after
Larsson, 1972).
Figure 4.2 Electron micrograph of freeze-fractured freeze-etched liquid n-dodecanol (after Gulik-
Krzywicki & Larsson, 1984).
B. Simple non-polar lipids in the liquid state

It seems likely that the order of liquid triacylglycerols is related to the need to
accommodate the three hydrocarbon chains in a space-filling arrangement. In
simple lipids such as a fat alcohol, there are no such restrictions. In accordance
with this, no X-ray scattering bands comparable in intensity with those from
triacylglycerols are observed. The freeze-fracture electron micrograph of
n-dodecanol melt shown in Figure 4.2 confirms this disorder.
THE LIQUID STATE 109
Figure 4.3 Electron micrograph of a freeze-fractured sample of milk fat before crystallization, kindly
provided by Wolfgang Buchheim, Kiel. The dimension of the terraces corresponds to a liquid
triacylglycerol bilayer according to X-ray scattering. The terrace height is about 50 Å and the shadow on
the micrograph is of the same order of magnitude.
C. Polar lipids in the liquid state and the aqueous L2 phase

Heating of a liquid-crystalline lipid phase or lipid–water phase will ultimately
result in a liquid phase. If this is an aqueous phase with low water content, we
learned from the phase diagrams of the previous chapter that it is called L2. We
will consider this L2 phase in monoacylglycerol systems, where the water
content can vary from about 30% (w/w) down to zero. At higher water contents
this liquid phase coexists with water. Some X-ray scattering characteristics of
this L2 phase are summarized below:
(1) An Lα phase heated through the transition into an L2 phase shows only
a small shift of the first-order diffraction line and a line broadening at the
transition.
(2) A plot of the scattering band position versus volume fraction of water
gives a linear relation, which is taken as evidence for a lamellar
structure.
(3) Assuming that the scattering band shape is an effect of line broadening
due to reduced size of ordered regions, the size of the scattering units
will be in the range of 300–500 Å.
Figure 4.4 Freeze-fracture electron micrograph of a monoacylglycerol–water (98:2) L2 phase (after

Gulik-Krzywicki & Larsson, 1984). The stacks of bilayer units of finite size appear as differently
oriented structure units.
0.3 μm
Figure 4.5 Freeze-fracture electron micrograph of a monoacylglycerol melt (after Gulik-Krzywicki &
Larsson, 1984).
This indicates that the L2 phase has a structure similar to that of the Lα phase,
with reduction of the infinite periodicity of the liquid-crystalline phase at the
transition into bilayer units, with a size of a few hundred Ångströms along the
bilayer and perpendicular to the bilayer plane. The bilayer thicknesses in both
THE LIQUID STATE 111
structures are about the same. The heating increases the disorder and gives a
tendency to divergence along the hydrocarbon chains. Therefore, after transi-
tion the polar regions with finite water layers are embedded into a continuous
hydrocarbon chain matrix.
The L2 phase is therefore considered to have a structure closely related to
that of the Lα phase: a melted Lα phase with change of long-range order to
‘medium’-range order. Studies by freeze-fracture freeze-etching electron
microscopy of the sunflower oil monoacylglycerols/water system have con-
firmed this structure (Gulik-Krzywicki & Larsson, 1984); see Figure 4.4.
Differently oriented stacks of lamellae can be seen, and their size is also in
agreement with the X-ray line broadening estimation.
When the water composition is reduced to zero, there is no discontinuity in
the scattering curves. Galactolipids in the liquid state show similar X-ray
scattering characteristics. An electron micrograph of a monoacylglycerol melt
is shown in Figure 4.5. This indicates that the structure in the melt of polar
lipids is the same as discussed here for L2 phases: a lamellar bilayer structure
with short-range disorder and medium-range periodicity.
D. Lipid microemulsions
The term microemulsion is used for thermodynamically stable liquid phases
formed by oil, water, a surfactant, and usually also a co-surfactant. The term
was introduced in the description of the transparent liquid phase in systems
based on mineral oils and water (Danielsson & Lindman, 1981). Lipid systems
are different, however, as the oil itself is amphiphilic. Still it seems adequate to
use this term for a liquid transparent phase consisting of water aggregates in a
triacylglycerol oil with a polar lipid acting as surfactant. An example showing
the existence range of such a phase is shown in Figure 3.22, which here is
termed L2. An electron micrograph of a corresponding freeze-fractured phase
is shown in Figure 4.6.
These inverse aqueous phases in lipid systems involving triacylglycerols are
liquid, thermodynamically stable and transparent. It seems logical from a
functional point of view to call them microemulsions, as an alternative to L2
phases. It also seems likely that by increasing the proportion of triacylglycerol
with respect to the polar lipid, the shape of the water aggregates will succes-
sively approach spheres.
E. The L3 phase
A third liquid phase beside the L1 and the L2 phases has been observed in non-
ionic surfactant–water systems (and in some ionic surfactant–water systems
containing salts), and it has been termed L3 (Anderson et al., 1989). It has since
also been observed in lipid systems, where it appears as an extension of a
Figure 4.6 Freeze-fracture electron micrograph of a sunflower oil monoacylglycerol–water–soybean

oil (85:6:9) microemulsion (after Gulik-Krzywicki & Larsson, 1984).
bicontinuous cubic phase towards higher concentrations of water and a non-

aqueous solvent (Engström et al., 1998; Imberg, 2003).
The location in the phase diagrams (see Figure 3.26) indicates that the
structure is a ‘melted’ bicontinuous cubic phase (Engström et al., 1998;
Imberg, 2003). Thus, like the L2 phase description as a ‘melted’ Lα phase, the
L3 phase can be described as a periodic minimal surface structure, where the
long-range order is lost and persists only up to medium ranges – perhaps to a
few hundred nm – in other words, a ‘melted’ bicontinuous cubic phase.
References
Anderson, D., Wennerstöm, H. & Olsson, U.J. (1989) J. Phys. Chem. 93, 4243.
Danielsson, I. & Lindman, B. (1981) Colloids and Surfaces 3, 39.
Engström, S., Alfons, K., Rasmusson, M. & Ljusberg-Wahren, H. (1998) Progr. Colloid
Polym. Sci. 108, 93.
Gulik-Krzywicki, T. & Larsson, K. (1984) Chem. Phys. Lipids 35, 127.
Imberg, A. (2003) On Phase Behaviours in Lipid/Polymer/Solvent/Water Systems and their
Application for Formation of Lipid/Polymer Composite Particles. Thesis, University of
Uppsala, Sweden. [Comprehensive Summaries of Uppsala Dissertations from the
Faculty of Pharmacy, No. 304].
Larsson, K. (1972) Fette-Seifen-Anstrichm. 74, 136.
CHAPTER 5
Lipids at the air–water interface – monolayers
and multilayers in surface films, bubbles and
foams
It is told that Benjamin Franklin described to the Royal Society in 1774 an

experiment that showed how a drop of a vegetable oil spread on the surface of
a water pond, and how the size of the wave damping could be used in order to
estimate the size of the smallest units, the molecules. Most of us are familiar
with such wave damping that can be seen over surface regions on lakes or on the
sea surface (‘sea slicks’), which are caused by lipids originating from the
biomass underneath.
Another milestone was the detailed molecular picture of lipid surface films
on water, presented almost a century ago by Langmuir (1917). Since then the
traditional view of lipids at air–water interfaces involves the formation of
monolayers. The existence of multilayers or even coherent surface phases
identical in structure to three-dimensional phases is, however, a neglected
aspect.
A. Monomolecular lipid films at the air–water interface

Langmuir in his pioneering experiments spread lipids on the surface between
two barriers in a trough, where one barrier was used to compress the film and
the other was used to measure the film pressure. In this way, the film pressure
(Π) versus the molecular area (A), Π–A isotherms, could be recorded.
Monolayers of lipids occur in different two-dimensional states, the gaseous, the
liquid and the solid states. The terminology used is G (gaseous), LE (liquid-
expanded), LC (liquid-condensed) and S (solid). This does not, however, take
into account polymorphism of the solid phases, and is therefore confusing (we
can call the solid phases S1, S2, etc.). It is quite clear now that the LC phase has
a crystalline structure. Its higher compressibility than ‘S phases’ is due to the
tilt of the hydrocarbon chains. It is also clear that the chain conformation in the
LE phase is the same as in the lamellar liquid-crystalline phase, as shown by its
compressibility and molecular cross-sectional area.
When we spread a monolayer with molecules initially far apart, they form a
gaseous state. On compression, first a condensed liquid phase and then solid
phases are formed. Sometimes there is a direct condensation from a gaseous to
a solid state. Two-dimensional pressure–temperature phase diagrams can be
constructed by recording Π–A isotherms at different temperatures.
113
The appearance of different phases in monolayers can be observed by adding

a fluorescent probe, which will partition differently between the phases, a
technique introduced by von Tscharner & McConnel (1981). They have
demonstrated beautifully how two-dimensional crystals of phospholipids are
formed on a water surface, exhibiting shapes reflecting the antipode character
of the molecules.
Another interesting macroscopic feature demonstrated in two-phase regions
of gaseous–liquid phases is the occurrence of a two-dimensional ‘foam’
(Moore et al., 1987). When the film is expanded, the gas ‘bubbles’ grow from
a circular shape and start to pack closely to give polygonal structures with a thin
layer of liquid in between. Such two-dimensional ‘foams’ exhibit an interesting
geometry, which also provides information on foam structure in general
(Stavans & Glazier, 1989).
1. Monolayers of polar lipids

The monolayer behaviour of 1-monomyristin is shown in Figure 5.1, and the
corresponding phase diagram, obtained from Π–A isotherms recorded at
different temperatures, is given in Figure 5.2 (Krog et al., 1985). The first
phase, which is formed during compression from the gaseous state (form I) has
the same structure as the Lα phase from which a monolayer is transferred, so
that a two-dimensional structure is obtained. At about 25 mN/m, there is a
transition into another monolayer phase (form II). The equilibrium between
these two phases, evident as a plateau with constant pressure (at very slow
compression), shows that it is a first-order transition. The cross-sectional area
per molecule at the transition is 28.5 Å2 in form I and 22.5 Å2 in form II (the
linear part of the isotherm extrapolated to the plateau pressure). The cross-
sectional area of form I is consistent with the molecular cross-sectional area of
the Lα phase, whereas the cross-sectional area of form II is in good agreement
with that of the crystalline form of 1-monomyristin.
Another interesting feature of this monolayer phase diagram is the thermal
existence range of the phases. Above 42ºC there is no phase with crystalline
chains. In the three-dimensional aqueous system the gel→Lα phase transition
takes place at the same temperature. 1-Monoelaidin was also studied and the
same close relation between monolayer phases and the three-dimensional
phases was observed. From this it can be concluded:
• If a polar lipid forms an Lα phase, it will give a monolayer phase with the
same molecular conformation in the same temperature interval.
• If a polar lipid forms a gel phase, the monolayer will form a condensed
phase with the same molecular conformation in the same temperature
interval.
Most membrane lipids exhibit these two types of monolayer phases, and the
LIPIDS AT THE AIR–WATER INTERFACE 115
Figure 5.1 A pressure–area (Π–A) isotherm of 1-monomyristin monolayer phases at 25ºC (after Krog
et al., 1985). The proposed structures of the two monolayer phases (form I at low pressure and form II
at high pressure) are also indicated.
Figure 5.2 Phase diagram (phase existence region in relation to temperature and monolayer pressure)
of 1-monomyristin monolayers (after Krog et al., 1985). The shaded area corresponds to collapse of the
monolayer.
relations with the aqueous phases in bulk are important, as so much more
information can be obtained from studies of three-dimensional phases. The
same two phases can also coexist in bilayers, such as cell membranes. If we
consider lipid rafts and caveolae (see Chapter 3), they correspond to two
coexisting monolayer phases, both with the liquid type of molecular conforma-
tion.
2. Domain structure within monolayers

The use of fluorescence microscopy in studies of phase separation in monolayers
was introduced by McConnel, and has revealed phase segregation also in
liquid-type monolayer phases (Okonogi & McConnel, 2004). Extensive stud-
ies of phospholipids and mixed phospholipid–cholesterol monolayers using
fluorescence microscopy have given a detailed description of domain shapes in
cases of coexisting phases, for example where solid and liquid monolayer
phases coexist (Weis, 1991). Domain shapes at equilibrium have been ob-
served by very slow compression/expansion when the Π–A isotherms become
reversible. The two factors determining the shape are the line tension between
the two phases and the electrostatic forces at the interface. When the line
tension dominates and both phases are isotropic (viewed from above), the
shape of the dispersed monolayer is circular. The surface potential is constant
within each phase, and at the phase boundary there is therefore a discontinuity.
Any asymmetry in the molecular orientation in the plane of the domain will also
be reflected in the shape of the domain, as the line tension and also the
electrostatic forces will vary in different directions. Repulsive forces between
different domains may also result in an ordered superstructure.
Dimyristoylphosphatidic acid has been observed to form a hexagonal super-
structure (Lösche et al., 1988).
There is an extensive literature on the monolayer behaviour of phospholipids,
fatty acids and other lipids. We will only discuss triacylglycerols, which
illustrate many general features of lipid monolayers.
3. Triacylglycerol monolayers and multilayers

The interfacial behaviour of triacylglycerol oils is due to the amphiphilic nature
of the molecules, and the spreading of monolayers of oils at interfaces has
important consequences in food technology. In Chapter 2 we learned that
triacylglycerol molecules in bulk distribute their chains in two different
monolayers, whereas in a monolayer on water, all three chains point in the same
direction in relation to the glycerol group.
Monolayers of well-defined triacylglycerols were first described by
Dervichian (1939), who studied the behaviour of mono-acid triacylglycerols.
More details were later revealed in a study by Burch et al. (1968). Tripalmitin
Figure 5.3 Tripalmitin pressure–area (Π–A) isotherm at 10ºC (after Burch et al., 1968). The transition
between the two solid phases, C1 and C2, is indicated by the arrow.
above 44ºC shows no solid phase. At lower temperatures, condensed phases

with crystalline chains are formed, as is evident from shifts in the slope of the
Π–A isotherm. A characteristic isotherm is shown in Figure 5.3. The steep rise
of the curve at low pressure shows that there is a direct transition from a gaseous
film to a monolayer with crystalline chains. From a change in tilt of two linear
regions it can be concluded that there is a transition into another phase with
crystalline chains. Furthermore, the angle of tilt of the acyl chains can be
estimated from the cross-sectional area per molecule. Compression beyond the
collapse point may give multilayers as described in Section A.7 below. This
example also demonstrates how much information there is in a Π–A monolayer
isotherm.
Studies of mixed-chain triacylglycerols have also been reported. Monolayers
of 1,2-dipalmitin-3-acylglycerols with C2–C16 in the 3-position were analysed
(Fahey & Small, 1986, 1988). Short chains, up to C6, exhibit a most remarkable
behaviour, with the short chain immersed into water. The shortest member, C2,
behaves like dipalmitin.
4. Competitive spreading of monolayers from fats/oils and proteins

From monolayer studies we can also determine the equilibrium spreading
pressure (ESP), which is the pressure of a monolayer in contact with a crystal,
with a droplet of a liquid lipid, or with a liquid-crystalline phase. In the case of
a plastic fat (a mixture of fat crystals and oil), the ESP value of the monolayer
formed from an oil is higher and will squeeze out the monolayer formed from
the fat crystals, as indicated in Figure 5.4.
The kinetics of the spreading process can vary a lot. When a triacylglycerol
Figure 5.4 Illustration of competition between a monolayer spreading from a fat crystal and a
monolayer spreading from an oil droplet (where the monolayer conformation is first gaseous).
in the liquid state is dropped on the water surface of a surface balance, the
equilibrium pressure is reached within fractions of a second. If, on the other
hand a triacylglycerol crystal is deposited on the surface, it takes many minutes,
up to an hour, until a constant pressure (the ESP) is obtained.
Proteins are often used in food processing in order to stabilize emulsions and
foams. The ESP value of most proteins is about 20 mN/m, which is higher than
the ESP obtained from common fats and oils. This means that proteins will
squeeze out triacylglycerol molecules from air–water interfaces. The presence
of free fatty acids, however, may change this situation (see also next para-
graph).
5. Other non-polar lipids

Free fatty acids in lipid systems exhibit complex behaviour related to pH. At
high pH they behave as polar lipids, whereas at low pH they can be regarded as
non-polar. The pKa value of a monolayer is different from that of a bulk
solution. The pKa value of fatty acids in aqueous solution is in the range 4.5–
5, whereas in monolayers the corresponding value is 7–8 due to the effect of
lateral electrostatic repulsion between the molecules.
In the undissociated state, long-chain saturated fatty acids form solid
monolayers, as shown in Figure 5.5. There is a direct transition from a gaseous
state of the monolayer into a phase traditionally termed L2 (NB this phase has
no structural relation to the three-dimensional L2 phase). From the cross-
sectional area it can be concluded that the molecules are arranged as in the solid
state (form B or C). At higher pressures, solid forms termed LS and S are
successively obtained. If the compression is stopped at any position along their
existence range, the pressure drops to that of the transition L2→LS. This
pressure is also obtained after collapse of the monolayer and it is also equiva-
lent to the ESP value obtained from crystals of the fatty acids. As we learned
from Chapter 2, all known fatty acid crystal forms have tilted molecules, and
the LS and S forms correspond to vertical molecules. It is therefore not
surprising that the monolayers of these two forms are not stable, exhibiting only
a transitory existence before monolayer collapse.
LIPIDS AT THE AIR–WATER INTERFACE
Figure 5.5 Pressure–area (Π–A) isotherms of (a) palmitic acid and (b) stearic acid on a 0.001 M solution of hydrochloric acid (after Lundquist, 1978).
119
Monolayers of simple esters (Lundquist, 1978) show two remarkable fea-

tures:
• the monolayer polymorphism according to a pressure–area phase
diagram is identical or very similar to that of the three-dimensional
state;
• the esters butyl stearate and amyl stearate form monolayers with
extended molecules, which means that the polar groups are well above
water contact.
Monolayers of 1,2-dipalmitin were studied by Fahey & Small (1986). At
room temperature there was a direct transition from a gaseous state to a
crystalline monolayer, and an ESP value of 31.7 mN/m was observed above the
melting point. Different monolayer behaviour was seen in a diacylglycerol with
very different acyl chains (Larsson, 1973a). First a conformation with the polar
group in contact with water was seen, and on further compression the mol-
ecules became extended with the polar groups far from water contact, as
observed in simple esters.
6. Mixed monolayers
Lipid monolayer phases show similar phase properties to lipid mixtures in
three-dimensional states. The phase rule can be applied (modified, as the
molecules are only free to move in two dimensions). If a binary lipid system is
studied by recording pressure–area isotherms at different proportions of the
two components, it is possible to directly determine if there is full mutual
molecular solubility as, if there is, the isotherms will have an intermediate
appearance to the isotherms of each of the two components studied alone. If the
mixed monolayer is expanded or compressed, an indication of interaction is
revealed by a change in the average molecular area in relation to the molecular
cross-sectional areas of the components at a particular pressure. If there is no
molecular solubility, on the other hand, the isotherm of the mixture will exhibit
the phase properties of each component (taking into consideration the fact that
a phase existing up to a higher pressure will squeeze out a phase with a lower
maximum pressure).
As mentioned above, the pioneering work on phase segregation in fluid
monolayer phases was reported by McConnel and co-workers (Subramaniam
& McConnel, 1987). By using a fluorescent probe they visualized the fluid–
fluid domains in mixtures of cholesterol and dipalmitoyl-phosphatidylcholine.
At a cholesterol concentration of 30mol%, they observed domain shape
fluctuations indicating the existence of a critical point of immiscibility. A
decade later the biological relevance of these observations was realized in
connection with studies of phase segregation in cell membranes (see Chap-
ter 6).
Figure 5.6 Formation of a triple chain-length structure of tripalmitin on a water surface (Larsson,
1973b).
7. Lipid monolayers on water and on solids

When monomolecular films of lipids on a water surface are compressed beyond
the collapse point, multilayers can be successively formed (Larsson, 1973b). It
has thus been demonstrated that fatty acids and their methyl esters form layers
consisting of 3, 5 or 7 monomolecular layers, according to the molecular area
of lipids per unit area at the water surface. This means that after collapse the
monolayer folds into a bilayer unit structure. This is not surprising. Only with
an odd number of monolayers can a hydrophilic surface be exposed towards
water and a hydrophobic surface towards air. In special cases with very weak
polar forces in the end group region, for example in monolayers of ethyl
stearate, the monolayer after collapse can be seen to fold into 2, 3, 4, 5 or 6
monolayers. The crystal structure of ethyl stearate is also consistent with this
behaviour; the structure unit is a monolayer, not a bilayer as in most other
lipids. The possibility to form these regular multilayer structures seems to be
related to crystalline monolayers with a certain degree of disorder of the
hydrocarbon chain packing, corresponding to the hexagonal close-packing
arrangement.
An example of the multilayers formed by triacylglycerols is shown in
Figure 5.6. As we saw in Chapter 2, triacylglycerols in the solid state form unit
layers of double chain length. On a water surface, however, the polar region
must be exposed towards water, forcing the monolayer conformation into a
single chain-length structure. At folding of this monolayer after the collapse
point, re-conformation into the ideal double chain-length unit is adopted.
So-called Langmuir–Blodgett (LB) multilayers can be deposited on a solid
support (Blodgett, 1935). By moving a solid with a hydrophobic surface down
through a monolayer of fatty acids into the water phase, it will become covered
by a monolayer (provided that the conditions for monolayer transfer are
fulfilled). The driving force is reduction of the surface energy. Such transfer
processes are simplest when the lipid molecules are in a solid phase, preferably
in an α-form. When the solid is moved upwards from the water phase through
the monolayer, another monolayer will be transferred, making the surface

hydrophobic in air. By repeating this process, LB layers with desired numbers
of monolayers can be produced. There are interesting technical applications of
LB films in electronics and in optics, some of them based on the ingenious work
by Kuhn & Möbius (1971).
In the following text, we will discuss the monolayers involved in stabilizing
bubbles and foams, but first a special case of air–water surface films will be
considered.
8. A surface phase formation at an air–water interface of a lung

surfactant extract
The alveolar surface lining is an important lipid system, which is discussed
further in Chapter 10. Here we will consider a related system: an extract in
organic solvents of lung surfactant. The interfacial behaviour of such a porcine
lung surfactant extract was recently characterized (Larsson et al., 2002). It
consists of about 1–2% (w/w) of two hydrophobic proteins (SP-B and SP-C),
and the rest is polar lipids, mainly dipalmitoyl-phosphatidylcholine. A lamellar
liquid-crystalline phase is formed in excess water, with a unit layer formed by
two lipid bilayers. When the water phase is exchanged with a physiological salt
solution, a lamellar liquid-crystalline phase is also formed, but now the unit
layer consists of a single lipid bilayer as in an ordinary Lα phase. The two
hydrophobic proteins are cationic and will interact with the anionic
phospholipids in the lipid bilayer in an environment of pure water. In the
electrolyte solution, however, the small ions will shield this interaction. These
proteins are known to link apposing bilayers. If they are distributed differently
in the bilayers, it is not surprising that the structure unit can be either one or two
bilayers.
When a salt-free freeze-dried preparation is swollen in a physiological salt
solution, involving a structural rearrangement from a double-bilayer unit
structure into a single-bilayer unit structure, the remarkable dynamic phenom-
enon shown in Figure 5.7 is observed. Within a few minutes, the air–water
interface of a droplet, placed between a slide and a cover-slip, folds into a
network with an enormous enlargement of the interface towards air. The
interface is strongly birefringent, showing that the enlarged surface consists of
a liquid-crystalline phase. Why is a coherent phase formed at the air–water
interface, not just a monolayer structure? Obviously the air–water interface
provides a driving force for faster growth of the single-bilayer phase obtained
in salt solution, and furthermore lateral growth along the bilayer is much faster
than growth of the thickness of this surface phase. The result will be a folding
process in order to supply the aqueous bulk phase with the desired air–water
interfacial area.
The branched pattern formed at the interface might be compared to the
Figure 5.7 The dynamic swelling behaviour of a lung surfactant extract, as seen under a microscope
(after Larsson et al., 2002). A network of branches at the air–water interface is formed in a time-scale of
minutes. The surface zone of the branches exhibits a uniform birefringence in polarized light.
branched shape of snowflakes, which has recently been discussed by Ben-

Jacob & Levine (2001). Thus competition between diffusion and growth can
influence the outer shape, and diffusion drives the system towards maximal
interfacial tension and irregular shapes.
The formation of a coherent phase at the air–water interface was recently
confirmed by neutron scattering studies (M. Larsson et al., to be published). It
has been described in detail here as it is a remarkable example of the effect of
minor amounts of proteins on lipid bilayer conformation and functionality.
B. Bubbles and foams

Gas bubbles in air (e.g. soap bubbles) or attached to a water surface can be
stabilized by polar lipid or surfactant interfacial films of similar structures to
those described above. The surfactant or lipid films will reduce the interfacial
energy. As the number of bubbles that attach to each other increases, the gas
cells become polyhedral in shape and form a foam. Foam lamellae tend to meet
at angles of 120º, and their characteristic structure is shown in Figure 5.8.
Polar lipids that can stabilize foams effectively should be water-soluble in
order to provide/remove molecules fast enough to form the interfacial
monolayers. Therefore it is only the more polar lipids forming micellar
Figure 5.8 Cross-section through a corner in a foam stabilized by a lipid monolayer.
solutions, like lysophospholipids and soaps, that are efficient as stabilizers of

ordinary foams. The reason is the need for water solubility in order to supply
monomers to the expanding gas–water interface quickly enough.
When gas is blown into a micellar solution in order to form a foam, some gas
cells will leave the water phase and form soap bubbles. The shape is of course
spherical, with both sides of the water film covered by a monolayer.
1. Foam structure
The size of the gas cells and their shape can vary a lot in a given type of foam.
From a statistical study of foams, it has been reported that the average
polyhedral cell has 14 faces (William, 1968). This polyhedron with 6 square
and 8 hexagonal faces had already been described by Lord Kelvin as the
polyhedron with the smallest surface-to-volume ratio.
2. Foam stability
The following factors are the most important ones in stabilizing foams:
• the mechanical properties of the interfacial film;
• the viscosity of the water layer, which determines the drainage of foam
lamellae;
• the repulsive forces between the monolayers on each side of the
lamellae.
Drainage is caused by gravity and also by the pressure at the so-called

Plateau borders, where the lamellae meet. This pressure from the outside gas
phase is smaller than that in the central part of the lamellae, as the surface there
is almost planar. Therefore there is a flow of water towards the Plateau borders,
and then downwards through the connected system of Plateau borders. Repul-
sive forces between the polar head groups, however, will act in the opposite
direction.
Additional factors that influence foam stability are surface film elasticity and
Marangoni effects. Such effects are due to differences in surface tension along
a surface, which for example can be caused by evaporation. In the case of foams
we can consider the situation when one foam lamella starts to burst. This means
that the surface area is increased locally and the surface tension at the hole is
higher than that of the surroundings. The surrounding monolayer with an
attached water layer will therefore move towards the hole and tend to close it.
3. Effects of electrolytes and the hydrophobic force

A remarkable effect of common electrolytes on the stability of gas bubbles in
water has been reported (Craig et al., 1993). Air bubbles introduced into pure
water have lifetimes below a second, whereas the presence of salts increases the
lifetimes to about 10 seconds (which can be compared with surfactants giving
bubbles that last for hours). When the bubble coalescence versus concentration
was followed, there was a steep increase in stability above a critical concentra-
tion. In the case of sodium chloride this concentration was 0.1 M, whereas
divalent ions showed a much lower concentration, 0.04 M in calcium chloride
solutions.
The authors proposed that this phenomenon is related to the long-range
attractive force observed between hydrophobic surfaces at distances of several
hundred Ångströms (Israelachvili & Pashley, 1982). This hydrophobic force
was found to be 10–100 times larger than the van der Waals forces.
The measured hydrophobic force is reduced by the addition of salts, and salts
should therefore be expected to reduce the attractive forces between bubbles
that induce coalescence.
4. Micellar lifetime and dynamics of bubbles and foams

When a foam is prepared, for example by injection of air into a micellar
solution, the surfactant molecules adsorb at the gas–water interface. The
diffusion time of surfactant molecules is much shorter (about 10–4 s) than the
micellar lifetime (about 0.6 s) according to measurements in sodium
dodecylsulfate solution (Oh & Shah, 1991). The lifetime of the micelle,
reflecting monomer formation, is therefore rate-limiting. Furthermore it was
seen that the foam had its minimum stability at the concentration where the
micellar lifetime had its maximum. This concentration agrees with the concen-
tration at which the spherical micelles start to become cylindrical.
If a foam is produced by gas injection from a capillary, the bubble will enter
the solution when the buoyancy force just exceeds the surface tension of the
contact circle between the bubble and the capillary. This demonstrates how the
bubble size decreases with surface tension. Above the critical micellar concen-
tration the monomer concentration is constant, but the bubble size has still been
found to vary (Oh et al., 1992). This could again be related to micellar stability.
The effects of salts on bubbles were discussed above. When a monolayer is
present on the surface, there will be electrostatic effects related to the ionic
character of the surfactant. Furthermore the critical micellar concentration is
reduced.
5. Destabilization of foams
There are many applications of foams, and there are also situations where
foaming must be avoided. Destabilizing mechanisms are therefore of interest.
The addition of salts will reduce the stability of foams based on ionic surfactants,
as explained above. Reduction of the viscosity of the surfactant monolayer is a
mechanism provided by a common foam destabilizer, 2-ethylhexanol. A
protein foam can be destabilized by a liquid fatty acid, such as oleic acid, and
this can be applied in order to reduce overly stable beer foams. The fatty acid
molecules will form separate domains along the surfaces of the foam lamellae,
where rupture will take place.
Another destabilizing mechanism has been described by Blute et al. (1994).
A series of organic electrolytes were studied, which supply counter-ions to a
foam stabilized by an ionic surfactant. Tetramethylammonium bromide and a
number of its homologues drastically reduced the stability compared to the
effect of sodium bromide. It was shown that the effect was related to the
increased area per surfactant molecule caused by the bulky shape of the
counter-ion. It was even seen that non-ionic surfactants were influenced,
indicating that these ions are amphiphilic and adsorb at the interface.
6. Foams stabilized by an interfacial phase

Particularly in foods, where the surfactants are normally insoluble, a foam can
be stabilized by a separate phase which forms the air–water interface. As we
will see in Chapter 8, this is also often the case in food emulsions.
We will consider one particular interfacial phase, which is often used
industrially for foam stabilization. The gel phase of pure monostearin shows
limited swelling in water, with a maximum thickness of the water layers of
about 20 Å. The addition of a few percent (calculated on the basis of total
lipids) of a fatty acid salt causes the water swelling to proceed to a water layer
thickness of several hundred Ångströms (Larsson & Krog, 1973). It is in fact

possible, by tuning the electrostatic repulsion from the anionic acyl groups, to
achieve water layers as thick as visible light wavelengths, and the gel phase can
be seen to change from blue to red at the other end of the visible spectrum. The
chain packing in the bilayer of this gel phase is hexagonal, and when the
bilayers transform into the stable β-form on storage, the whole phase collapses
into crystals plus water.
This kind of gel phase prepared from industrially distilled monoacylglycerol
preparations can be additionally stabilized by the solubilization of guest
molecules, such as polyglycerol esters of fatty acids, into the bilayers to
stabilize foams in various foods, particularly in the baking of cakes. In a recent
thesis (Richardson, 2004), it was shown that the air–water interface of foams
produced by this gel phase in water and in sucrose solution consists of coherent
birefringent layers, forming a matrix of this phase throughout the aqueous
compartments of the foam. On injection of air, the phase will open at surfaces
where it is weakest, which is at the methyl end group gap. The rheological
properties of the foam lamellae of this organized structure contribute a lot to the
foam stability.
7. Foams based on proteins and emulsified oils/fats

Protein molecules tend to adsorb at air–water interfaces due to their amphiphilic
character. They can therefore stabilize foams, and egg white is a well-known
example. An important difference in relation to lipids, however, is that these
foams are usually more stable. The reason is that the re-conformation of the
tertiary structure at the interface driven by the reduction in surface free energy
can be irreversible. We can alternatively consider the dynamics at the interface.
Lipid molecules will adsorb and desorb with a fast exchange between bulk and
interface. Once a protein molecule is adsorbed, on the other hand, it is attached
at many positions along the peptide chain. Desorption then becomes unlikely,
as detachment must take place at all positions simultaneously.
Another type of foam, which we all are familiar with, is whipped cream. The
introduction of air bubbles will remove phospholipids from parts of the surface
of the fat globules. When they become partly ‘naked’ and hydrophobic, they
will adsorb at the interface, surrounded by a phospholipid monolayer. The two-
dimensional aggregation of fat globules at the interface provides mechanical
stabilization of the foam. This type of foam is further discussed in Chapter 12.
References
Ben-Jacob, E. & Levine, H. (2001) Nature 409, 985.
Blodgett, K.B. (1935) J. Am. Chem. Soc. 57, 1007.
Blute, I., Holmberg, K., Jansson, M., Oh, S.G. & Shah, D.O. (1994) J. Am. Oil Chem. Soc.
71, 41.
Burch, T., Larsson, K. & Lundquist, M. (1968) Chem. Phys. Lipids 2, 102.
Craig, V.S.J., Ninham, B.W. & Pashley, R.M. (1993) J. Phys. Chem. 97, 10192.
Dervichian, D.G. (1939) J. Chem. Soc. 7, 931.
Fahey, D.A. & Small, D.M. (1986) Biochemistry 25, 4468.
Fahey, D.A. & Small, D.M. (1988) Langmuir 4, 589.
Israelachvili, J.N. & Pashley, R.M. (1982) Nature 300, 341.
Krog, N., Riisom, T. & Larsson, K. (1985) Encyclopedia of Emulsion Technology, Marcel
Dekker, New York, USA, p.321.
Kuhn, H. & Möbius, D. (1971) Angew. Chem. 83, 672.
Langmuir, I. (1917) J. Am. Chem. Soc. 39, 1848.
Larsson, K. (1973a) Biochim. Biophys. Acta 318, 1.
Larsson, K. (1973b) In: Surface and Colloid Science 6 (E. Matijevic, ed.), J. Wiley & Sons,
New York, USA, p.261.
Larsson, K. & Krog, N. (1973) Chem. Phys. Lipids 10, 177.
Larsson, M., Haitsma, J., Lachmann, B., Larsson, K., Nylander, T. & Wollmer, P. (2002)
Clin. Physiol. Func. Imag. 22, 39.
Lösche, M., Duwwe, H. & Möwald, H. (1988) Colloid Interface Sci. 126, 432.
Lundquist, M. (1978) Progr. Chem. Fats Other Lipids 16, 101.
Moore, B.G., Knobler, C.M., Broseta, D. & Rondelez, J. (1987) Chem. Soc. Faraday Trans.
II 82, 1753.
Oh, S.G. & Shah, D.O. (1991) Langmuir 7, 1316.
Oh, S.G., Klein S.P. & Shah, D.O. (1992) AICAE J. 38, 149.
Okonogi, T.M. & McConnel, H.M. (2004) Biophys. J. 86, 880.
Richardson, G. (2004) Foam Formation and Starch Gelatinization with Alpha-Crystalline
Emulsifiers. Thesis, University of Lund, Sweden.
Stavans, J. & Glazier, J.A. (1989) Phys. Rev. Lett. A 37, 1318.
Subramaniam, S. & McConnel, H.M. (1987) J. Phys. Chem. 91, 1715.
von Tscharner, V. & McConnel, H. (1981) Biophys. J. 36, 409.
Weis, R.M. (1991) Chem. Phys. Lipids 57, 227.
William, R.E. (1968) Science 161, 276.
CHAPTER 6
Dispersions of lipid–water phases
What happens when two or more phases coexist in aqueous lipid systems? Can
the phases be dispersed in one another, and what are the flow, transport, and
stability characteristics of such dispersions? What are the responses to ambient
changes? These are questions of importance in many applications involving
lipid-based systems. Because the continuous phase typically is water we will
here concentrate on the properties of aqueous lipid dispersions of the lyotropic
lipid phases introduced in Chapter 3. Some related structures observed in
living systems are discussed towards the end of the chapter.
A. Liposomes, vesicles and the Lα phase

The realization that phospholipids in excess water can form aggregates consist-
ing of concentric bilayers that entrap part of the solvent into their interior was
a breakthrough discovery (Bangham & Horne, 1964). This discovery has
spurred an enormous interest in lipids and their associated colloids across
academic disciplines. Numerous methods have been developed for preparing
liposomes with different characteristic properties. The formation of liposome
structures involves the fragmentation of lamellar-phase structures that sponta-
neously fuse and form spherical structures to avoid contact at the edges
between water and hydrocarbon moieties. The practical issues of preparing
structurally well-defined liposomes relate to: (1) rapid hydration and swelling
of the ‘water-insoluble’ lipid matrix, (2) fragmentation of the same into
submicron aggregates, and (3) maintaining the colloidal stability. Due to their
particulate and space-dividing nature, liposomes have a number of potential
applications in science and technology, for example as mimics of cell mem-
branes and as drug delivery vehicles.
Every living cell is surrounded by a membrane with the same basic bilayer
structure as liposomes. Because of this, liposome structures have been exten-
sively used as biomimetics in biophysical studies aimed at explaining different
aspects of cell membrane properties and functionality. The lipid bilayer acts as
an encapsulating chemical, mechanical, and electrical barrier separating inter-
nal and external milieus. This is also the fundamental basis behind the use of
liposomes as carriers of internally trapped agents. Besides the ability to
encapsulate water-soluble substances into their interiors, the liposome bilayer(s)
can also be used as ‘amphiphilic’ solvent for lipophilic/amphiphilic sub-
stances. Combined with the often biocompatible (triggering a negligible immune
129
response) and biodegradable properties this makes liposomes attractive

vehicles for the in vivo delivery of diagnostic and therapeutic agents (see
Chapter 11).

The requirement for liposome formation is the occurrence of a two-phase
region where the Lα phase coexists with water (see Figure 6.1). Many natural
lipids, especially membrane lipids, have a packing parameter, V/Al, close to 1
(see Chapter 3, Section B.3), and exhibit an Lα+H2O region at high dilution.
The almost ‘spontaneous’ dispersion formation observed in some systems has
been used as an argument for thermodynamically stable liposomes. The
evidence presented so far is, however, not conclusive, and we continue to view
them as metastable structures that are kinetically locked, due partly to the low
solubility of the lipid building blocks.
The traditional way of preparing liposomes is first to make thin lipid films by
evaporation of solvent, and subsequently to disperse these in water (e.g. see
Szoka & Papahadjopoulous, 1980). This process typically results in
multilamellar vesicles with a broad size distribution. The unilamellar struc-
tures, preferred in most technological applications, require down-sizing by the
use of high-shear methods. This can occur either by a ‘budding-off’ process of
daughter vesicles from preformed bilayers (budding-off or fission model), or
alternatively by self-closing of bilayer fragments (bilayer fragmentation model).
The former represents a continuous process whereby the bilayer topology
changes but integrity is preserved until the point of detachment of the ‘daughter
vesicle’. The latter involves instability at the edges and disruption at an early
stage, followed by self-closing. There is no sharp line between these mecha-
nisms and they are probably only clearly separable in special limiting cases.
The bilayer fragmentation mechanism dominates when liposomes are formed
by high-shear methods including sonication and homogenization/
microfluidization. The budding-off mechanism applies to liposome formation
from larger Lα assemblies due to asymmetric change in the surface area upon
a change in the external milieu (such as ionic strength, pH, temperature,
etc.).
Stability and reproducibility issues have for a long time limited technologi-
cal uses of liposome systems. Vesicles are typically not thermodynamically
stable, but trapped in local free energy minima, and are thus difficult to
manufacture reproducibly. The inherent two-phase nature of liposome systems
is a consequence of the unfavourable curvature and excess bending energy
imposed by the spherical liposome structure, and the ubiquitous van der Waals
attraction causing colloidal instability. The origin of the repulsive interaction
between uncharged liposomes is thermal fluctuations of the membranes and
hydration interactions (Israelachvili, 1992). Kinetic stabilization is facilitated
DISPERSIONS OF LIPID–WATER PHASES 131
Temperature
Dilution +
shear force
Lα
Concentration
Lα
Liposomes
Figure 6.1 Schematic illustration of the Lα phase, phase diagram, and formation of liposomes in the
diluted two-phase region. The cryo-TEM image shows liposomes of dioleoylphosphatidylethanolamine/
Triton X-100. The small arrow indicates an ice crystal, the large arrow an inter-lamellar attachment. Scale
bar: 100 μm (Johnsson & Bergstrand, 2004).
by introducing stabilizing electrostatic and steric repulsion between liposomes.

The former involves mixing charged lipids into vesicle bilayers, but is really
only effective in the case of low ionic strength dispersions, since electrostatic
interactions are effectively screened by the presence of ions. Surface-adsorbed
block copolymers and polymer-conjugated lipids are effective also at high
ionic strengths and are therefore the preferred colloidal stabilizers in many
practical applications. Effective liposome stabilization is achieved by incorpo-
rating small fractions of polyethylene glycol conjugated lipids (PEG-lipids)
into liposome bilayers; these act as a steric stabilizing barrier (see also
Chapter 11). Note that it is important that the solubility of lipids and stabilizing
polymeric molecules are attuned in order to retain colloidal stability over time
under diluted conditions.
The most prominent application of liposome systems is as pharmaceuticals.
The use of liposomes as drug delivery vehicles is today well established. The
demands on pharmaceutical products are very high and commonly require
product shelf lives of at least two years. The requirements of colloidal stability,
an encapsulation efficiency of more than 90%, and low leakage rates over time
periods of several years put very stringent demands on the preparation
processes and material properties of pharmaceutical liposome products.
Achieving sufficient encapsulation efficiency typically requires active load-
ing, which can be achieved in response to ion gradients across the bilayer
membrane surface. Thus for instance the uptake of the hydrophilic anti-
neoplastic agent doxorubicin can be loaded with more than 98% efficiency by
employing a H+ gradient (acidic liposome interior) (Lasic et al., 1995).
Figure 6.2 Schematic illustration of the architecture of different types of liposomes: small and large
unilamellar vesicles (SUV and LUV), multivesicular liposomes (MVL), and multilamellar vesicles
(MLV).

Liposomes are classified according to their number of lamellar enclosures as
multilamellar, oligolamellar or unilamellar vesicles. Unilamellar liposomes are
further classified according to size as small, large or giant unilamellar vesicles
(see Figure 6.2). These structures cover a three-order of magnitude particle size
range, from tens of nanometres to tens of microns. Dispersions of small
unilamellar vesicles are typically translucent, whereas dispersions of larger
structures are milky white due to scattering of visible light.
Many physicochemical and functional properties are affected by liposome
size, including deformability, stability, adhesion, enclosed volume per lipid,
and biological clearance. The liposome size affects inter-bilayer interactions
and can alter basic thermodynamic properties of the bilayer. The gel-to-liquid-
crystal transition temperature (Tc) is, for instance, reported to be lower for small
unilamellar vesicles than for corresponding larger multilamellar structures

(Lichtenberg et al., 1981). Small liposomes have also been shown to be more
fusogenic than larger liposomes (Nir et al., 1983), which might be partly
explained by weakened electrostatic and undulation repulsion and bilayer
defects. Small and defect dipalmitoylphosphatidylcholine (DPPC) liposomes
in the gel state appear particularly prone to fuse with other membrane struc-
tures. Another important size issue relates to interactions in biological systems,
where the size distribution is a key factor determining the circulation time and
distribution in the body (Ishida et al., 2002).
Lamellarity is another obvious property of liposomes. Multilamellar
liposomes are easy to prepare, but their dispersions typically suffer from ill-
defined size distributions and a relatively poor colloidal stability. This is an
inherent problem as additions of steric or electrostatic stabilizers will generally
induce inter-lamellar swelling and thereby promote transitions to unilamellar
structures. Partly due to this compromise relatively few applications of
multilamellar liposomes have been realized, despite the potential of high
loading efficiency and control of release rates by tuning the number of lamellar
vesicular units. Non-lamellar liquid crystalline particle structures are in this
context clearly superior; see the discussion below.
Lipid compositions impact the material properties of vesicular membranes.
The disordering effect of double bonds in acyl chains lowers the gel-to-liquid-
crystal transition temperature (Tc), and thereby the elastic modulus and tensile
strength of phospholipid bilayers. The phase transition behaviour observed for
pure phospholipid systems vanishes for lipid systems with heterogeneous
compositions of hydrocarbon chains, like egg phosphatidylcholine. Increasing
temperature typically results in a strong decrease in surface shear viscosity due
to membrane ‘fluidization’, in analogy with effects of the acyl chain composi-
tion.
Cholesterol stabilizes the liquid crystalline state of phospholipid membranes
over a wide temperature range, yet it increases the membrane elasticity with
a maximal effect at slightly above ~50mol%. Using X-ray diffraction the
solubility limit of cholesterol has been determined to be 66mol% for
phosphatidylcholine and 51mol% for phosphatidylethanolamine (Huang et al.,
1999). Sackman and co-workers first demonstrated the phase segregation
induced by cholesterol solubilized in phosphatidylcholine bilayers with an
immiscibility gap occurring between 8mol% and 20mol% cholesterol (Albrecht
et al., 1978, 1981). This concentration interval is relevant to many biomembranes
and is related to the formation of lipid rafts and caveolae. Cholesterol exerts its
effect via an ordering of lipids locally in the outer regions of acyl chains, while
maintaining a liquid-like interior due to disruption of long-range crystalline
order.
Aside from the traditional structure-based classification of liposomes, a
functional classification has also been introduced based on the most prominent
liposome application as in vivo carriers of therapeutic and diagnostic agents.

Instead of morphological differences this is based on (biological) interaction
characteristics and classified as: (1) nonspecific, (2) inert, (3) specific and (4)
reactive liposomes (Lasic & Papahadjopoulos, 1995). The non-specific cat-
egory comprises conventional liposomes which can be controlled in vitro, but
are relatively unstable and rapidly cleared in vivo. Inert liposomes are those
whose interactions are suppressed by steric stabilizing polymer coatings. Such
liposomes equipped with specific targeting ligands such as surface-tethered
antibodies, lectins or other groups are also capable of interacting specifically.
The fourth group includes very reactive liposomes, such as ambient-sensitive
and fusogenic liposomes.
B. Non-lamellar colloids of reversed cubic, hexagonal, and ‘sponge’

phase
Non-lamellar nanoparticles formed by the dispersion of self-assembled lipid
mesophases have great potential in science and technology due to their special
encapsulating and space-dividing nature – featuring both hydrophilic and
hydrophobic nanodomains whose connectivity can be controlled from discrete
to continuous. Such particles have gained increasing interest due to their
fascinating structural architecture, potential uses, and possible biological
relevance. In contrast to liposomes, which consist of mostly water even for
relatively small liposome radii, the lipid content of these particles is high
(normally in the range of 50–70%, w/w). Because of coexisting hydrophilic
and lipophilic nanodomains and the enormous surface area possessed by these
structures, they have a broad spectrum of applicability comprising lipophilic
and amphiphilic bioactive agents, and peptide and protein drugs.
As already discussed, a particular lipid can form various liquid crystalline
(LC) phases (polymorphism) where small changes in molecular or ambient
properties can cause dramatic morphology and phase changes. With decreasing
spontaneous curvature, and increasing packing parameter, we now know that
these include: L3 (or ‘sponge’), reversed bicontinuous cubic (Q2), reversed
hexagonal (HII), and reversed micellar cubic (I2) phases – all of which feature
different characteristic structural attributes and functional properties.
In analogy with the liposomal dispersions, non-lamellar particles can be
produced in multiphase systems comprising a non-lamellar phase and excess
water. The first versions of a fragmented cubic phase were recognized almost
two decades ago when it was proposed that the glycerol monooleate (GMO)
cubic phase can be dispersed into micron-sized particles by mechanical breakup
in the presence of micellar solutions of bile salts or caseins (Larsson, 1989a,b).
The action of these agents was explained in terms of the formation of a lamellar
envelope on the surface of the cubic phase particle. Later it was discovered that
amphiphilic block copolymers can provide very efficient stabilization of
monoacylglycerol-based cubic phase dispersions (Landh, 1994). A number of

studies clearly demonstrated that polyethyleneoxide-based stabilizers, such as
Pluronic® F127, are suitable candidates for the efficient stabilization of GMO
cubic phase dispersions while simultaneously preserving the inner cubic
structure of the particles (Gustafsson et al., 1996, 1997). The formation of
colloidally stable cubic phase particles was related to the preferential location
of the stabilizing copolymer at the surface of the particles. The presence of
vesicular structures in the dispersions indicated further stabilization of the
cubic phase particles by a coexisting Lα phase.
Scientific and technological progress has, however, been hampered due to
significant problems in preparing kinetically stable, non-lamellar particles with
structurally well-defined properties. Another key issue has been the fact that
only a very limited number of lipid constituents have initially been used for
making such particle structures, most notably unsaturated monoacylglycerols
such as GMO. Unfortunately, due in part to the haemolytic activity, high doses
of monoacylglycerols cannot be used in some in vivo applications, such as
intravenous drug delivery.
Improvements in manufacturing schemes and the use of novel lipid combi-
nations have now finally facilitated the use of non-lamellar phases and
nanoparticles in parenteral drug delivery (Barauskas et al., 2005a; Johnsson et
al., 2005a). Using biocompatible and biodegradable lipid excipients and
reproducible, reliable and scalable manufacturing schemes, the final obstacles
for the use of non-lamellar nanoparticles in pharmaceutical applications have
been overcome.

Reverse phases can be fragmented in excess water to form cubic-, hexagonal-
and ‘sponge’-phase particles. The basic condition is the existence of a multi-
phase region comprising a non-lamellar phase and a dilute aqueous phase.
Fragmentation is facilitated by the coexistence of a lamellar or L3 phase, and
additions of amphiphilic stabilizers. The resulting non-lamellar particles are
generally enclosed by an outer bilayer structure which is rich in the stabilizing
component.
Non-lamellar particle compositions include: (1) structure-forming lipid(s),
(2) amphiphilic and polymeric stabilizers, and (3) water. Most early prepara-
tions feature GMO as the structure-forming lipid, because of its suitable phase
behaviour featuring a bicontinuous cubic phase in equilibrium with water.
When mechanically fragmented in water together with a surface-active stabi-
lizer like Pluronic® F127, cubic phase particles are obtained with a lamellar
surface. In order to obtain small submicron particles the composition must be
exposed to high-energy treatment using for instance a homogenizer/
microfluidizer. However, high energy input results in the formation of a
significant fraction of metastable liposome ‘contaminants’, and dispersions of

limited colloidal stability. It was discovered that these metastable structures
can be converted to equilibrium phase structures by a heat treatment step in
which the dispersion is heated to above 90ºC for a short period of time. The
heating step facilitates the preparation of well-defined and monocrystalline
dispersions with controlled size distribution and high kinetic stability (Barauskas
et al., 2005a). Importantly the method is applicable to a large range of different
non-lamellar particle compositions.
In some cases non-lamellar particle structures can be produced with a
minimum of energy input. One such system comprises diglycerol monooleate
(DGMO) and glycerol dioleate (GDO) (Johnsson et al., 2005b). DGMO is a
lipid that forms an Lα phase in water, whereas GDO only forms an isotropic
reversed micellar solution (L2 phase). When combining these two lipids,
several liquid crystal phases were found, including two cubic phases and the
reversed HII phase. These compositions are readily dispersed by gentle shaking,
offering an alternative to high-energy methods and heat treatment, which can
be detrimental in cases when the non-lamellar composition features sensitive
substances such as therapeutic proteins.
Another process for ‘spontaneously’ forming cubic phase nanoparticles via
dilution of the GMO–Pluronic® F127–ethanol–water system was also intro-
duced (Spicer & Hayden, 2001). It is based on mixing the GMO with ethanol
in miscible proportions and dilution of the system with an aqueous solution
containing a stabilizer, i.e. Pluronic® F127. When the dilution trajectory falls
into a cubic phase region, the cubic phase particles are formed spontaneously
due to molecular diffusion differences at the liquid–liquid interface. However,
this preparation method also resulted in a substantial amount of vesicular
material, and typically the particle size distributions were rather broad. In fact,
judging from the published cryo-transmission electron microscopy (cryo-
TEM) images, the majority of the particles displayed vesicular morphology
(Spicer & Hayden, 2001).
In some applications it is preferable to use liquid or powder precursors. In
such cases reconstitution to dispersion should be easy and reproducible. Spray
drying methods have been used but reconstitution is so far not perfect,
particularly after extended storage. However, new ‘spontaneously’ dispersing
liquid compositions have been elaborated and are currently being tested in
pharmaceutical product development programmes.

Whereas liposomes are classified according to lamellarity and size, non-
lamellar phase particles are classified according to phase structure as
bicontinuous cubic (Q2), reversed hexagonal (HII), ‘sponge’ (L3), or reversed
20
10
0
0.01 0.1 1 10
Particle size (µm)
Figure 6.3 Particle size distributions of a homogenized and subsequently heat treated (125ºC) GMO–
Pluronic® F127 dispersion at a lipid:polymer ratio of 9:1. The size distribution increases with the total
amphiphile concentration (1%, 2%, 3%, 4% and 5%, w/w) used in the heat treatment step, and remains
unaltered by subsequent changes in concentration caused by, for instance, dilution (Barauskas et al.,
2005a).
micellar cubic (I2) particles. Cubosome® and Hexosome® are frequently used
trade names of cubic and hexagonal phase dispersions.
The size of non-lamellar particles can be controlled in the size range from
about 100 nm to several μm without changes in the inner morphology. One
example of size distributions obtainable for the GMO–Pluronic® F127 system
is shown in Figure 6.3.
By changing the phase structure the connectivity of water and lipid domains
can also be varied. This makes non-lamellar particles attractive for loading,
encapsulation and controlled release purposes in pharmaceutical and diagnos-
tic applications, as the phase structure can easily be adjusted to address a
specific technological need and fit a certain compound. By using two structur-
ing lipids with different spontaneous curvature, it is possible to change between
different particle phase structures simply by changing the ratios of the two
components. An example of such a system is GDO–DGMO, where GDO forms
a L2 phase and DGMO a lamellar phase. Combined in different ratios they form
a range of nanostructured phases including two cubic phases and the reversed
HII phase (Johnsson et al., 2005b). Some key properties of the different
reversed-phase dispersions are given below.
Figure 6.4 Cryo-TEM images of bicontinuous cubic phase (Q2) dispersions viewed in the [0,0,1]
direction.
Figure 6.5 Cryo-TEM images of reverse hexagonal phase (HII) dispersions showing a ‘top view’ of
well-ordered nanoparticles with reverse hexagonal ‘honeycomb’ structure.
Cubic phase particles – Cubosomes®

Figure 6.4 shows an image of bicontinuous cubic phase particles with an
internal monocrystalline cubic phase enclosed by a bilayer surface structure. A
characteristic feature of structurally well-ordered cubic phase particles is their
cubic geometry. The particle size can be varied in the interval from ~100 nm to
several μm. However, for long-term kinetic stability (months to years) the size
should be <1 μm. Fragmentation of the reversed micellar cubic phase (I2) is
also possible, yielding nanoparticles with discrete water domains.
Hexagonal phase particles – Hexosomes®

Monocrystalline hexagonal phase particles are shown in Figure 6.5. The
particles always appear projected along the [1,0] direction in the cryo-TEM
images; implying that they have a high aspect ratio. To prove this hypothesis,
angle-resolved cryo-TEM measurements have been performed to visualize
other facets of the particles. These studies confirmed that well-ordered hexago-
nal phase particles are shaped as hexagonal prisms with a high width-to-height
aspect ratio of ~50 (Barauskas et al., 2005b). Less well-ordered Hexosome®
particles adopt more spherical shapes.
Nanostructured liquid particles – sponge (L3) particles and reversed micellar

(L2) particles
Thermodynamically stable sponge (L3) and reversed micellar (L2) phases can
also be fragmented into colloidal dispersions; see example in Figure 6.6.
Sponge phase particles reveal an inner network of interconnected lamellae,
sometimes referred to as molten bicontinuous cubic phase. This structure
terminates in an external bilayer defining the particle surface. Compared to
liquid crystalline particles, L3 particles incorporate much more water, as L3
phases occur in the water-rich region of the phase diagram. The shape of the
disordered L3 phase particles is typically spherical as exemplified in Fig-
ure 6.6. It is noteworthy that the particles might comprise regions of different
phase structure. Recent results show nanoparticles with L2 and L3 phase
structures in coexistence.
While L3 particles contain substantial amounts of water, L2 phases contain
relatively small amounts (~10%). Dispersion of an L2 phase in water requires
the presence of stabilizer and might result in a change in the structure of the
dispersed phase, including the appearance of new structural elements.
Using the drug delivery application as an example, it is easy to understand
that the encapsulation and release properties of, for instance, a hydrophilic
substance will be different in a bicontinuous cubic phase particle, compared to
a reversed micellar (L2) or micellar cubic (I2) phase particle, which has discrete
water domains. It is also obvious that a large water-soluble protein is more
easily incorporated into ‘sponge’ phase particles with a relatively high degree
Figure 6.6 Cryo-TEM images at different magnifications of particles with sponge phase characteris-
tics.
of water swelling compared to hexagonal phase particles with small water

channels.
To facilitate loading and encapsulation it is possible to modify electrostatic
interactions by including small amounts of anionic or cationic components in
the structure-forming matrix.
The presence of ionic lipids at the particle surface will increase the colloidal
stability through electrostatic repulsion between particles. As discussed in the
case of liposomes, the repulsion is effectively weakened by the presence of
electrolyte as predicted by classical Derjaguin–Landau–Verwey–Overbeek
(DLVO) interaction potential theory.
Effective stabilization is achieved by the surface-tethered polymer seg-
ments, for instance polyethylene glycol (PEG) chains. Surface-active molecules
with large PEG segments having a molecular mass of several thousand Daltons,
such as block copolymers and some PEG-lipids, are largely excluded from the
nanostructured particle interior and therefore accumulate at particle surfaces.
They provide effective stabilization without significantly affecting the phase
structure. Smaller amphiphiles can partition more evenly and thereby affect the
structure of the interior mesophase. Ethoxylated surfactants typically swell the
aqueous domains and shift the phase structure towards less negatively curved
structures. Such stabilizing agents will typically facilitate fragmentation, in-

crease the degree of swelling, and stabilize the particle structure.
An important feature of any colloidal system used in technological applica-
tions is that it should exhibit physical and chemical stability for extended
periods of time. Physical stability concerns particle size as well as encapsula-
tion retention, whereas chemical stability refers to lipid oxidation and hydrolysis
as well as to any solubilized/encapsulated compounds. Using appropriate
compositions, the stability of non-lamellar particle structures can be very good,
showing stability for months to years at room and refrigerated temperatures.
Note that critical phase transitions in the storage/application temperature
interval should preferably be avoided.
As the main application of non-lamellar phase particles relates to the
encapsulation of biologically active molecules, it is worth mentioning some
key beneficial features of these structures:
• high solubilizing/loading capacity for a broad spectrum of substances;
• protective of sensitive drug substances;
• effective encapsulation is possible, giving controlled release;
• effective in vivo transporters;
• enhanced circulation.
The last point is related to the optional PEG coating and small particle size,
which protects the particles from recognition by the immune system and
engulfment by the macrophages; see also Chapter 11.
C. Lipid–water dispersions in biological systems

The only dispersed particles of simple lipid–water phases that can be clearly
identified in biological tissues are aggregates of the lamellar liquid-crystalline
phase and of cubic phases.
1. Lamellar bodies
The special organelles for lipid storage and secretion have been called lamellar
bodies, and they can vary in size from 100 nm to 2400 nm (Schmitz & Muller,
1991). They occur for example in the gastrointestinal tract, in the peritoneum,
and in the mesodermal cell layer of joints. The lamellar bodies of two other
tissues will be considered here.
Assemblies of concentric lipid bilayers – like multilamellar liposomes – are
found in the lamellar bodies secreted from type II epithelial cells in the lung.
There are indications from cryo-TEM studies that the unit layer consists of two
lipid bilayers (M. Larsson, 2002), but still the main arrangement is similar to
that of liposomes. Epidermal lamellar bodies are involved in regeneration of
the stratum corneum layer of the skin. These organelles are different, however,
as the equidistant lipid bilayers are not concentrically arranged, but approxi-
mately planar. The stacks of membrane discs which form rods and cones in the
retina of the eye are another example of a parallel bilayer organization. These
two types of organization represent extremes of the Lα phase conformations in
closed particles/organelles.
2. Cubic cell membrane assemblies

In connection with the first experimental demonstration of cubic lipid–water
particles (Larsson, 1989), some examples of cell membrane systems were
shown which were proposed to be cubic, due to the resemblance between their
electron microscopy periodicity and that of known cubic lipid–water phases.
Cubic cell membranes were later the subject of a thesis by Landh (1996). He
was able to identify hundreds of examples of cubic membrane arrangements by
a convincing reconstruction analysis. Thus in relation to an electron micro-
graph, a particular plane in the unit cell of the corresponding cubic minimal
surface structure was identified. The thickness of a corresponding section was
then varied to reach agreement with the original electron microscopy texture.
Minimal surface structures in cell membranes were also reviewed in a book by
Hyde et al. (1997). An illustrative example of the formation of a cubic
membrane system is shown in the cover figure of the October 2003 issue of The
Cell. It shows how an endoplasmic reticulum texture, with a liposomal mem-
brane in the centre, transforms into a cubic membrane.
The length scale in biological cubic phases is larger than that of correspond-
ing phases comprising essentially only lipids or surfactants, by about one order
of magnitude. The same three types of structures – the G, D, and P types – and
no others have been observed in biological samples. Cubic membranes were
first associated with pathological conditions, based on numerous observations
(Landh, 1996). The dense t-tubuli membrane system, for example, penetrates
the muscle cell in the transverse direction and is involved in the delivery of
calcium ions to the actin–myosin complex. In disease states, such as muscle
dystrophy, they appeared to be transformed into cubic membranes. Later
observations indicated that cubic membranes can be formed during apoptosis
(programmed cell death). Furthermore, they may represent a vegetative state of
a cell.
A minimal surface bilayer conformation is possible only when the normal
asymmetry of the cell membrane is lost, so that the compartments on each side
can become equal (or chirality related). Phosphatidylserine (PS) is normally
located in the inside monolayer of the membrane. Exposition of PS on the
outside of the membrane is a signal for apoptosis, and results in elimination of
the cell by phagocytosis.
Deng & co-workers (1999) found that the mitochondria in a certain amoeba
will, on starvation, undergo a transformation into a cubic minimal surface
Figure 6.7 Amoeba mitochondria showing transition into a cubic membrane conformation on starva-
tion (from Deng et al., 1999, with permission).
membrane system. A remarkable feature is that feeding after starvation showed

that this transition is reversible; see Figure 6.7.
The formation mechanism of a cubic conformation in biomembrane systems
is likely to sometimes involve transitions from the ‘normal’ closed vesicle
conformation. As an example of such hypothetical transformations, we may
consider an assembly of vesicles located at the carbon atom position of the
diamond lattice. If the lattice is compressed, the vesicles will successively
come closer to one another. Then, at a certain intervesicular distance, formation
of a catenoid opening between adjacent vesicles will reduce the total surface
area. A phase transition should therefore be expected, resulting in fusion of the
vesicles into one closed compartment. This means that the average curvature
has to be constant, so that the pressure over the membrane is the same
everywhere. A process that might be based on such a mechanism is the mass-
cooperative fusion of vesicles in the synaptic cleft: the prerequisite for nerve
signal conduction.
References
Albrecht, O., Gruler, H. & Sackmann, E. (1978) J. Phys. 39, 301–303.
Albrecht, O., Gruler, H. & Sackmann, E. (1981) J. Colloid Interface Sci. 79, 319–338.
Bangham, A.D. & Horne, R.W. (1964) J. Mol. Biol. 8, 660–668.
Barauskas, J., Johnsson, M., Joabsson, F. & Tiberg, F. (2005a) Langmuir 21, 2569–2577.
Barauskas, J., Johnsson, M. & Tiberg, F. (2005b) Nano Lett. 5, 1615–1619.
Deng, Y.R. et al. (1999) J. Struct. Biol. 127, 231–239.

Gustafsson, J., Ljusberg-Wahren, H., Almgren, M. & Larsson, K. (1996) Langmuir 12, 4611–
4613.
Gustafsson, J., Ljusberg-Wahren, H., Almgren, M. & Larsson, K. (1997) Langmuir 13, 6964–
6971.
Huang, J., Bubolz, J.T. & Feigenson, G.W. (1999) Biochim. Biophys. Acta 1417, 89–100.
Hyde, S.T., Andersson, S., Larsson, K., Blum, Z., Landh, S., Lidin, S. & Ninham, B. (1997)
The Language of Shape, Elsevier, Amsterdam, the Netherlands.
Ishida, T., Harashima, H. & Kiwada, H. (2002) Biosci. Rep. 22, 197–224.
Israelachvili, J. (1992) Intermolecular and Surface Forces, Academic Press, London, UK.
Johnsson, M. & Bergstrand, N. (2004) Colloids Surfaces B Biointerfaces 34, 69–76.
Johnsson, M., Barauskas, J. & Tiberg, F. (2005a) J. Am. Chem. Soc. 127, 1076–1077.
Johnsson, M., Lam, Y., Barauskas, J. & Tiberg, F. (2005b) Langmuir, 21, 5159–5165.
Landh, T. (1994) J. Phys. Chem. 98, 7304–7314.
Landh, T. (1996) Cubic Membrane Architectures. Thesis, Lund University, Sweden.
Lasic, D.D., Ceh, B., Stuart, M.C.A., Guo, L., Fredrik, P.M. & Bahrenholz, Y. (1995)
Biochim. Biophys. Acta 1239, 145–156.
Larsson, K. (1989a) J. Phys. Chem. 93, 7304–7314.
Larsson, K. (1989b) J. Disp. Sci. Technol. 10, 351.
Larsson, M., Terasaki, O. & Larsson, K. (2003) Solid State Sci. 5, 109.
Lasic, D.D. & Papahadjopoulos, D. (1995) Science 217, 1245–1246.
Lichtenberg, D. et al. (1981) Biochemistry 20, 3462–3467.
Nir, S., Bentz, J., Wilshut, J. & Duzgunes, N. (1983) Progr. Surf. Sci. 13, 1–124.
Schmitz, G. & Muller, G. (1991) J. Lipid Res. 10, 1539–1570.
Spicer, P.T. & Hayden, K.L. (2001) Langmuir 17, 5748–5756.
Szoka, F.C. & Papahadjopoulous, D. (1980) Ann. Rev. Biophys. Bioeng. 9, 467–508.
CHAPTER 7
Interaction of lipids with proteins and
polypeptides
A. Introduction
Proteins are linear polymers of amino acids linked by peptide bonds between
the α-carboxyl group of one amino acid and the α-amino group of the next.
There are 20 protein amino acids and each protein is distinguished by the
primary sequence of these amino acids in the polymer. The factor that charac-
terizes the different amino acids is the side chain attached to the α-carbon atom.
Amino acids are categorized into several groups with common chemical
features. Some amino acids have charged functional groups such as carboxyl or
amino groups, while others are hydrophobic and contain branched aliphatic
groups or aromatic residues. The physical properties of the protein depend on
the proportion of each type of amino acid in the protein and the positions of the
amino acids in the polypeptide chain. These, in turn, dictate the higher-order
folding arrangements referred to as the three-dimensional secondary and
tertiary structure of the native protein.
The manner in which the protein folds depends largely on the local forces
experienced by the different side chains and the need for the polymer chain to
adopt a conformation with relatively high entropy. The forces at play are those
due to the solvent environment as well as the proximity of other amino acid
residues of the protein or of proteins with which it interacts. In the case of
certain membrane proteins the ‘solvent’ may include the lipid matrix of the
membrane, and interactions of this type may be required to fold the protein into
its native configuration. On the other hand there is evidence that interaction of
the proteins with membrane lipids is required to impose a bilayer conformation
on the surrounding membrane lipids, and is therefore an essential factor in
preserving the structure and properties of the membrane itself.
Non-ionic surfactants interact only weakly if at all with water-soluble
proteins but can solubilize intrinsic membrane proteins because of their
hydrophobic domains, which otherwise render them insoluble in aqueous
media. There is, however, strong interaction between ionic surfactants and all
proteins. One of the most studied of this type of interaction is between sodium
dodecyl sulphate and proteins. The non-specific cooperative binding of sodium
dodecyl sulphate to soluble proteins results in unfolding of the polypeptide
chain. After reduction of any disulphide bridges in proteins, sodium dodecyl
sulphate, above the critical micellar concentration, interacts with the polypep-
145
tide with a stoichiometry of 1.4 g detergent per gram of protein. This inter-
action imposes a regular helical structure on the polypeptide chain, which
becomes bent in the shape of a hairpin. The length of the resulting complex is
a function of the length of the polypeptide chain and, because of the predictable
conformation combined with a constant charge to mass ratio, differences in
hydrodynamic and electrophoresis properties can be exploited in separation
strategies for complex mixtures. The most notable system is the separation of
proteins by polyacrylamide gel electrophoresis in the presence of sodium
dodecyl sulphate (SDS-PAGE), which separates polypeptides on the basis of
size.
Where proteins interact with biological membranes at the aqueous–lipid
interface the charges of the acidic membrane lipids provide a particular
environment capable of interacting with basic amino acids of the protein. In the
case of proteins that are interpolated into the hydrophobic interior of the
membrane the environment is more conducive to the location of amino acids
with non-polar side chains. Clearly, both proteins and membrane lipids have
hydrophilic and hydrophobic groups, which interact to determine the structure
and conformation of the complex. This chapter will explore the nature of these
interactions and identify examples that illustrate the principles of how proteins
interact with lipids to modulate lipid structure and protein function.
B. Surface activity of proteins

All proteins are hydrated in their native state and this interaction with water is
essential to fold the protein into its biologically active conformation. Likewise,
virtually all proteins contain amino acid residues that are comfortable in a
hydrophobic environment or can only be inserted into an aqueous environment
at the expense of entropic free energy. This amphipathic character can easily be
demonstrated by the surface activity of proteins. For example, the creation of a
two-phase oil–water system when performing a lipid extraction of biological
samples leaves a layer of protein at the interface which is soluble in neither the
organic nor the aqueous phase. Moreover, most solutions of proteins will form
a monolayer at an air–water interface or can be spread on an air–water interface
to form a stable interfacial film.
Serum albumin is a typical soluble protein that exists in its native conforma-
tion under physiological conditions of pH, electrolyte concentration and
temperature. The protein can be spread at an air–water interface as a monolayer
of molecules in one of two well-defined configurations. This is illustrated by
the experiment shown in Figure 7.1, in which [14C]-labelled bovine serum
albumin is spread onto an aqueous surface at a surface density of about 50 nm2/
molecule (Quinn & Dawson, 1970). A surface pressure–area isotherm (solid
symbols) shows that the protein monolayer can be compressed up to a collapse
surface pressure of about 16 mN•m–1. If the monolayer is then expanded and the
INTERACTION OF LIPIDS WITH PROTEINS AND POLYPEPTIDES 147
Figure 7.1 Surface pressure–area isotherms of serum albumin monolayers at an air–water interface.
Monolayers were spread initially at high molecular density () and were subsequently expanded and
equilibrated at low surface density before recompression (). The protein molecules undergo a
conformational change to occupy a greater molecular area when unconstrained by pressure from
surrounding molecules in dense monolayers.
film is allowed to equilibrate at low surface density the isotherm changes so that
the molecules occupy almost twice the surface area per molecule at equivalent
surface pressures (open symbols). The measurement of surface radioactivity
confirms that there is the same number of protein molecules in the monolayer
so that the observed effect must be due to a change in conformation of the
protein molecules in the film. One explanation is that when the protein is spread
at high density the compact structure assumed by the protein in solution is
preserved by the tight packing of the molecules at the air–water interface.
Expansion of the surface area reduces these constraints, allowing the protein to
adopt a more expanded configuration consistent with the prevailing interfacial
forces associated with the surface tension of water. This illustrates the point
that protein folding (or unfolding) is dictated by the forces exerted by the
surrounding solvent molecules.
The structure of monomolecular films of rhodopsin purified from bovine
retinal rod outer segment disk membranes as a complex with octyl glucoside at
the air–water interface has been examined by infrared absorption and X-ray
reflectivity measurements (Lavoie et al., 2002). It was found that when the
protein was spread as a monolayer at a surface pressure of 5 or 10 mN•m–1 it

retained a secondary structure and thickness consistent with that of the native
rhodopsin. Monolayers spread at 0 mN•m–1 and allowed to equilibrate for
30 min before compression showed an increase in β-sheet conformation at the
expense of α-helical structure and a decrease in thickness consistent with an
unfolding of the protein to occupy a greater surface area per molecule. The
surface pressure–area isotherms were of the same type as that observed for
bovine serum albumin, indicating that membrane proteins like rhodopsin
behave in a similar manner to typically water-soluble proteins in films at the
air–water interface.
C. Interaction of soluble proteins with lipid monolayers

One of the most convenient systems for investigating the interaction between
proteins and polypeptides with lipids is the use of Langmuir film techniques
(Brezesinski & Mohwald, 2003). The principle of the method relies on the
insolubility of lipids in water and their concentration-dependent ability to
decrease the surface tension of water at an oil or air interface. The magnitude
of the decrease in the surface tension of water is a measure of the surface
pressure (π) of the monolayer, i.e. π = 72 – x, where the surface tension of
water is 72 mN•m–1 and x is the surface tension when a monolayer of surfactant
molecules is present. The presence of a monolayer of lipid can be verified by
creation of a surface pressure–area isotherm in which there is an increase in
surface pressure on reduction in the surface area per molecule of lipid. At the
point where a reduction in surface area per lipid molecule does not result in an
increase in surface pressure, more molecules are present than are able to form
a monolayer and the film is said to have collapsed. The interaction between the
insoluble lipid monolayer and soluble proteins present in the aqueous subphase
can be monitored by changes in surface pressure of the monolayer. Quantitative
changes in the composition of the monolayer can be determined by surface
radioactivity measurements. The principle of this measurement is that when
lipids or proteins are labelled with radionuclides that decay by weak β-
emission, such as 14C, 32P, 35S, etc., only labelled molecules present in the
surface layer can be detected as those present in the subphase are effectively
shielded by water. Other methods of measuring changes in the structure or
composition of the film include interfacial junction potential and surface
reflectivity or absorbance.
1. Mechanism of protein penetration into lipid monolayers

The polar lipids of cell membranes and other weak detergents form stable
monomolecular films at an air–water or oil–water interface. The properties of
such monolayers have been described in Chapter 5. Lipid monolayers have
proved to be invaluable for studies of the interaction between lipids and soluble
proteins; the lipid molecules at the interface are oriented and arranged in a well
understood way and their concentration can be readily altered by careful
expansion and compression of the film. A number of surface parameters can be
used to ascertain the extent and nature of the interaction with a soluble protein
in the bulk phase, including the consequences of interaction on the structure
and biological activity of the protein. Thus an increase in the surface pressure
of a film on adding protein is generally assumed to represent the penetration of
part at least of the space-occupying ‘volume’ of the protein into the film, albeit
this may not reach to the hydrophobic fatty acyl chains. Because of the various
factors that contribute to the surface pressure exerted by a monolayer (the
equation of state has kinetic, cohesive and electrostatic terms) it is clear that the
‘volume’ of protein penetrating into a particular film may not always be
proportional to the surface pressure increment. The ‘volume’ of protein pen-
etrating into the monolayer is related to the surface pressure increment by a
factor that depends on the force–area curve of the lipid film and any specific
interactions between the components. Measurement of the interfacial potential
of a monolayer also provides evidence of protein–lipid interactions, although
the interpretation of any changes is often problematic. This is because surface
potential is the sum of the vertical components of the intrinsic dipoles of the
lipid molecules, including the ionic dipoles of the head groups as well as the
oriented dipoles of water molecules hydrating the film. Any perturbation of the
aligned lipid molecules or ionic interactions or displacement of lipid molecules
by the adsorbing protein can affect these dipoles. Finally, with the use of
radioactively labelled proteins, direct measurement of the surface radioactivity
of the adsorbed protein is possible. Thus the total amount of protein adsorbed
per unit area of the surface phase can be derived but not the precise location or
orientation.
The monolayer method does have one possible disadvantage: unless the
penetrating parts of the proteins are exactly the same size and shape as the film
molecules, holes will tend to form in the unimolecular layer. To maintain the
planar structure the hole will fill with air unless some compensating collapse or
realignment of the fatty acyl chains can take place. Unless this realignment
occurs the adsorption energy may be changed somewhat by the energy required
to create the hole.
Penetration of protein into lipid monolayers can be observed when the initial
surface pressure of the monolayer is low, and increasing amounts of protein are
added to the aqueous subphase. This can be seen in Figure 7.2. [14C]-labelled
serum albumin was injected under a monolayer of egg phosphatidylethanolamine
spread at an initial surface pressure of 2 mN•m–1. The change in interfacial
junction potential shows a linear relationship but this parameter is the sum of
many vertically oriented dipoles and is not subject to a straightforward inter-
pretation. With increasing amounts of serum albumin the surface pressure
Figure 7.2 Changes in surface pressure (), radioactivity () and interfacial junction
potential () on injecting [14C]-labelled serum albumin underneath monolayers of
phosphatidylethanolamine initially spread at 2 mN•m–1.
increments at equilibrium exhibit a relationship to protein concentration that is

characteristically biphasic, shown by arrows indicating the first and second
inflection points on the surface pressure curve in Figure 7.2. The surface
radioactivity curve is also biphasic but it is noteworthy that, whereas the
pressure increment reaches a maximum, the radioactivity continues to increase
despite the presumption that additional penetration of protein does not take
place. This result clearly demonstrates that the protein does not react quantita-
tively with the monolayer. Using the surface radioactivity measurements it is
possible to calculate the percentage of added protein adsorbed to the surface;
values for albumin and cytochrome c adsorbed to different lipid monolayers
are presented in Table 7.1. It can be seen that the minimum concentration of
protein required to produce the maximum surface pressure increment occurs
when a relatively small proportion of the added protein is adsorbed on the
monolayer. Measurements of radioactive protein in the aqueous subphase
confirm that protein does not adsorb to the trough. Furthermore, removal of the
monolayer and its replacement by a fresh lipid film resulted in adsorption of
protein from the aqueous subphase to form a new film at equilibrium with
almost identical composition to that of the initial film. Other evidence also
indicates that once the protein is adsorbed to the lipid monolayer it does not
desorb even if the aqueous subphase is replaced by fresh medium or medium
containing unlabelled protein.
With knowledge of the surface concentration and area of both protein and
lipid at any surface pressure from surface radioactivity and surface pressure–
area isotherms, respectively, it is possible to infer how the protein penetrates
into the lipid monolayer. Thus, the area made available by the increase in
Table 7.1 Percentage of protein added under lipid monolayers of 2 mN•m–1 adsorbed to
the film
Added protein (%)
Protein Lipid First inflection Second inflection
Albumin Phosphatidylcholine 20 12
Phosphatidylethanolamine 14 7
Stearic acid 21 15
Cytochrome c Phosphatidylcholine 10 2
Phosphatidylethanolamine 26 19
Stearic acid 74 26
Table 7.2 Relationship between the space created in lipid monolayers and the area
occupied by protein at the interface at the first inflection point
Pressure Space in lipid film Area of protein

Protein Lipid (mN•m–1) (mm2/m2) (mm2/m2)
Albumin Phosphatidylcholine 8.1 1.65 1.86

Phosphatidylethanolamine 8.5 1.78 1.64
Stearic acid 4.1 0.43 0.51
Cytochrome c Phosphatidylcholine 10.8 1.94 2.54
Phosphatidylethanolamine 10.8 2.20 2.31
Stearic acid 6.4 0.60 0.78
surface pressure on the lipid molecules can be compared with the area occupied
by the penetrating protein assuming this is fully unfolded in the film. The
calculated area occupied by the protein at the first inflection point in the film
pressure showed reasonably close agreement with the area made ‘available’ by
compression of the lipid molecules (Table 7.2). This is consistent with the
penetration of folded protein molecules in their entirety into gaps in the
expanded films at low pressure and occupying spaces equivalent to that of the
unfolded protein at the air–water interface. At surface pressures greater than
the first inflection point the relationship completely breaks down and clearly
more protein is interacting with the film than could be accounted for by the
unfolded protein molecules entering the available space in the monolayer. This
means that only part of the protein enters the film, possibly the more hydropho-
bic domains of the polypeptide chain. It is likely that at low film pressures the
energy required for the protein molecules to enter the film and unfold at the
interface is comparatively low because of the greatly expanded nature of the
monolayer, but that at higher pressures the adsorption energy required for this
type of penetration is not available in the system. It is noteworthy that the
pressure of the first inflection point is less than the collapse pressure of the
protein monolayer at the air–water interface, whereas the second inflection
point is usually above this pressure.
The penetration of proteins containing the so-called START domain, a
conserved protein fold of a group of lipid-transfer and membrane-targeting
proteins, into lipid monolayers has been investigated (Angeletti et al., 2004).
The proteins themselves, such as StarD7, form stable monolayers by simple
adsorption from solution at the air–water interface. The maximum surface
pressure was found to be 18 mN•m–1 due to adsorption as a Gibbs monolayer,
from which a value of 28 kJ•mol–1 for the free energy of protein adsorption
could be derived. This value is similar to those of a variety of other amphipathic
proteins, consistent with the high tendency of this group of proteins to adsorb
at interfaces. Interestingly, no desorption of protein was observed after succes-
sive expansion and recompression cycles of the film up to the collapse pressure
of about 36 mN•m–1. However, a reversible rearrangement was observed in the
interfacial organization of the protein, which was evidenced by an abrupt
change in molecular packing area. The protein was found to penetrate into a
variety of phospholipid monolayers, with a preference for phosphatidylserine
over phosphatidylglycerol.
2. Protein penetration dependence on initial film pressure

The extent of protein adsorption to phospholipid monolayers depends among
other factors on the initial surface pressure of the film. The simplest type of
reaction is where the lipids of the monolayer have head groups which are
uncharged or zwitterionic at the pH of the subphase. This is exemplified by the
adsorption of [14C]-labelled cytochrome c to monolayers of phosphatidylcholine,
as shown in Figure 7.3. This shows that the change in surface pressure due
to interaction with the protein is inversely proportional to the initial surface
pressure of the monolayer. Furthermore, the change in surface pressure
was directly related to the number of protein molecules adsorbed to the
interface, with no interaction if the initial pressure of the film exceeded about
20 mN•m–1. When films to which protein had adsorbed at initial pressures of
less than 20 mN•m–1 were compressed, almost complete desorption of the
protein was observed. Because the effect was independent of the concentration
of salt in the subphase it may be concluded that the interaction was largely via
non-ionic forces.
The value of the initial surface pressure at which there is no increase in
pressure upon injection of protein into the subphase depends on the lipid
composition of the monolayer and the particular protein. This pressure has
been determined for amyloid precursor protein for different lipid monolayers in
Figure 7.3 Changes in surface parameters on adding [14C]-labelled cytochrome c to the subphase
beneath phosphatidylcholine monolayers at various initial surface pressures.
order to investigate how the protein may be metabolized in Alzheimer’s disease

(Lahdo & De La Fourniere-Bessueille, 2004). The cut-off pressures were
in the order cholesterol (32 mN •m –1 ), sphingomyelin (28 mN • m –1 ),
phosphatidylcholine (23 mN•m–1) and phosphatidylserine (11 mN•m–1).
3. Ionic interactions and protein adsorption

Where the net charges on the protein and the lipid are opposite in sign the
attraction between the protein and the lipid monolayer results in enhanced
penetration of the protein into the monolayer as well as adsorption by electro-
static interactions. This is illustrated in Figure 7.4, which shows the adsorption
of [14C]-labelled cytochrome c to monolayers of phosphatidic acid spread on
subphases of 10 mM or 1 M NaCl at different initial pressures. While the
inverse relationship between the initial film pressure and change in pressure
resulting from penetration of protein into the monolayer is observed on both
subphases, the magnitude of the change in pressure is considerably greater
when the electrostatic charges on the reactants are not screened by the high salt
concentration. Thus, the maximum initial pressure at which protein is pre-
vented from entering the monolayer is almost the collapse pressure of the
phospholipid monolayer when the charges are fully expressed. Furthermore,
the amount of protein adsorbing to the film is much greater in the latter case and
adsorption of protein is clearly taking place without any hydrophobic inter-
actions between the components.
The nature of the adsorbed protein has been investigated by comparing the
amount of [14C]-labelled cytochrome c that could be desorbed from mono-
molecular films of phosphatidylethanolamine spread at an initial surface
pressure that resisted penetration, by the addition of concentrated saline to the
Figure 7.4 Changes in surface parameters on adding [14C]-labelled cytochrome c to the subphase
beneath phosphatidic acid monolayers spread at various initial surface pressures on 10 mM NaCl (closed
symbols) and 1 M NaCl (open symbols).
subphase at intervals after injection of the protein into the subphase. The
adsorption of protein to the film was almost completely prevented when the
initial subphase was 1 M NaCl, consistent with the case presented in Fig-
ure 7.4. However, if adsorption was allowed to take place in 10 mM salt
solution and then solid NaCl was added to increase the concentration in the
subphase to 1 M, then only a proportion of the adsorbed protein was displaced.
The amount removed depended on the time of the interaction with the film,
reducing to a constant proportion of about 20% after equilibration had been
achieved. This result suggests that once electrostatic interaction has taken
place the adsorption is subsequently stabilized by non-ionic bonding. This
could be hydrogen bonding between the peptide chains and the polar head
groups of the phospholipid or, less likely, van der Waals interactions between
hydrophobic side chains of amino acids and the methylene groups of the
ethanolamine oriented in the interfacial region.
4. Effect of subphase pH on protein interaction with lipids

Cytochrome c is a very basic protein and charged membrane lipids are acidic so
it may be expected that there will be a strong mutual attraction between the
soluble protein and the negatively charged interface. The relationship between
the extent of protein interaction with monolayers and subphase pH is
Figure 7.5 The effect of electrostatic charge of monolayers of phosphatidylethanolamine and [14C]-
labelled cytochrome c on (a) adsorption to films at surface pressures of 26 mN•m–1 and (b) penetration
into films spread on subphases of 1 M NaCl at a surface pressure of 10 mN•m–1.
complicated and not clearly understood. The problem is illustrated by the

interaction of cytochrome c with monolayers of phosphatidylethanolamine
spread on subphases of different pH, as shown in Figure 7.5. At film pressures
greater than 26 mN•m–1 there is no penetration of cytochrome c into the
phospholipid monolayer but the amount of protein adsorbed to the film is
directly related to the charge difference between the reactants (Figure 7.5a).
The maximum absorption is observed when the charge on the lipid interface is
approximately equal to the opposite charge on the protein. The effect of pH on
the penetration of cytochrome c was examined on monolayers spread at an
initial pressure of 10 mN•m–1 on a subphase containing 1 M NaCl to prevent
protein adsorption to the film. It can be seen from Figure 7.5b that very little
penetration (or adsorption) of protein takes place when both protein and lipid
are positively charged (pH 3). Maximum penetration of the monolayer is
observed when the net positive charge on the protein is greatest and this
progressively decreases as the net charge decreases. This is also consistent with
the rate of penetration of the protein into the film measured before equilibrium
has been established. When the subphase pH is increased from pH 7 to 9 there
is a dramatic decrease in the amount of protein penetrating and a corresponding
increase in the pressure of the film. An explanation for this could be that the
protein undergoes a change in conformation as the isoelectric point is ap-
proached and this change leads to an occupation of larger domains of the
protein within the monomolecular film. This form of the protein apparently
creates a more stable monolayer, evidenced by the increased surface pressure.
Interaction between the protein and the lipid is again reduced as both carry
negative charges at pH greater than 10.5.
The dependence of mutual charges on the strength of the interaction between
soluble proteins and monomolecular films has also been examined in the
albumin–phosphatidycholine system by altering the surface charge of the film
and the net charge of the protein. The isoelectric point of serum albumin is 4.7
and the net charge can be made positive at pH 4.5 and negative at pH 5.5. The
interfacial potential of a zwitterionic phosphatidylcholine monolayer can be
made negative by inclusion of 10mol% phosphatidic acid or dicetyl phosphate
and positive by inclusion of 10mol% stearylamine, and the magnitude of the
potential is virtually unaffected between pH 4.5 and pH 5.5. The results
obtained from measurements of the penetration of albumin into the various
monolayers maintained at different initial pressures were unambiguous; pen-
etration into phosphatidylcholine monolayers was observed up to an initial
pressure of about 20 mN•m–1 irrespective of the subphase pH. Addition of
negatively charged phosphatidic acid or dicetyl phosphate caused a marked
increase in the penetration of the protein but only when the protein was
positively charged. By contrast, the inclusion of stearylamine in the film had
little effect at either pH or indeed if the net negative charge on the protein was
increased by adjusting the subphase pH to 7. This indicates that the electro-
statically facilitated penetration shows some specificity for the sign of the net
charges on the protein and lipid interfaces. The explanation for this differential
effect of the sign of the charge on the interface on penetration is obscure. It may
be expected that not all proteins would necessarily show the same charge-
assisted penetration characteristics, although the trend observed in the serum
albumin system has been observed in a wide variety of proteins. Furthermore,
it has been reported that a number of proteins bind only weakly to oil emulsions
if the stabilizer is a positively charged detergent, but relatively strongly if
negatively charged detergents are used.
5. Hydrocarbon chain dependence of protein penetration

The maximum pressure of monolayers of egg phosphatidylcholine and
phosphatidylethanolamine allowing penetration of cytochrome c is not
dependent on whether the hydrocarbon chains are unsaturated or fully satu-
rated. The pressure increment – initial pressure curves have slightly different
slopes but the initial pressure at which protein penetration is prevented is
invariably about 20 mN•m–1. It is well known that hydrogenation of unsaturated
lipids causes a pronounced reduction in the area occupied by each phospholipid
at the interface. Thus hydrogenation of egg phosphatidylcholine and
phosphatidylethanolamine reduced the cutoff pressure of monolayers pen-
etrated by cytochrome c from 74 to 50 mN•m–1 and 79 to 47 mN•m–1, respectively.
Clearly the limiting pressure allowing penetration does not depend on the space
available between the lipid head groups. Presumably for penetration to take
place the forces assisting adsorption must be greater than the work required to
push the occupying portion of the protein into the film, thereby increasing the
surface pressure. However, at lower pressures it is likely that for a specified
change in the pressure increment on adding protein, a much smaller volume of
protein would enter the film of hydrogenated lipid due to the steeper slope of its
surface area–pressure isotherm.
Although unsaturation of fatty acids appears to have little effect on the
pressure allowing penetration, it is found that penetration is enhanced by an
increase in length of the fatty acyl chains. This is exemplified by the ability of
cytochrome c to penetrate monomolecular films of phosphatidylcholine of
molecular species C16:0/C16:0, C16:0/C18:0 and C16:0/C18:1, the results of which are
presented in Figure 7.6. Protein penetration into 1,2-dipalmitoyl-
phosphatidylcholine (C16:0/C16:0 PC) had a cutoff initial pressure of 14 mN•m–1,
compared with 20 mN•m–1 for 1-palmitoyl-2-stearoyl-phosphatidylcholine (C16:0/
C18:0 PC). The chain length difference is only 2 carbons in the fatty acyl chain
in the sn-2 position of the glycerol and this has a negligible effect on the area
occupied by the molecules in the film. The cutoff point is the same for
phosphatidylcholine molecules having stearate (C16:0/C18:0 PC) or oleate (C16:0/
C18:1 PC) attached to the sn-2 position, so the effect is chain-length dependent
Figure 7.6 The effect of hydrocarbon chain substituents of phosphatidylcholine on penetration of

cytochrome c into monolayers spread at different initial surface pressures.
and not dependent on the presence of unsaturated double bonds in the chain.
The molecular species containing the unsaturated acyl chain, however, does
occupy a larger surface area at equivalent surface pressure and this explains the
difference observed in the slopes of the initial pressure penetration curves. The
penetration of albumin into saturated and polyunsaturated molecular species of
phosphatidylcholines is found to produce similar results.
A systematic study of the effect of the chain length of phosphatidylcholines
on the penetration of serum albumin showed that the cutoff point for dimyristoyl
and dipalmitoyl molecular species was 20 mN•m–1 but increased to 35 and
41 mN•m–1 for the distearoyl and dibehenoyl molecular species, respectively.
These differences are likely to be due to the marked differences in the
properties of the respective monolayers. Thus distearoyl and dibehenoyl
monolayers are condensed at 20ºC, while dipalmitoyl- and dimyristoyl-
phosphatidylcholine monolayers are liquid-expanded. This again means that
the pressure limit of protein penetration cannot be related to the gaps between
the phospholipid molecules at the interface, since dimyristoyl-
phosphatidylcholine, which has the most space ‘available’, is not penetrated at
pressures which allow penetration into distearoyl- or dibehenoyl-
phosphatidylcholines.
It is not clear what additional forces assist protein penetration into the
longer-chain molecular species of phosphatidylcholines at higher pressures.
Direct hydrophobic interactions would involve extensive penetration of un-
folded domains of the protein because the hydrophobic side chains of individual
amino acid residues of the polypeptide chain are not long enough to penetrate
between the acyl chains of the phospholipid without some degree of unfolding.
D. Interaction of lipolytic enzymes with their substrates

The characteristics of the lipid–water interface are able to affect the interaction
of specific proteins and to modulate their activity. Many of these proteins are
enzymes that act on the lipids themselves, such as triacylglycerol lipases and
phospholipases as well as apolipoproteins (Zidovetzki & Lester, 1992; Soulages
et al., 1996). Many lipolytic enzymes share a common feature in that their
hydrolytic activity against their normal substrates is greatly influenced by the
presence of lipids that are not substrates of the reaction or known to be bound
to the enzyme. For hydrolysis to take place the water-soluble enzyme must be
attracted to the substrate located at the lipid–water interface. As has been seen
from the factors that govern the interaction of proteins with monomolecular
films of lipid, the charges carried by the reactants need to be favourable, and the
magnitude of the attractive force increases with the size of the opposite charges
carried by the protein and the potential at the substrate interface. In addition, the
enzyme must orient about the interface in such a way that the substrate is
presented in a favourable manner for the hydrolytic reaction to proceed.
Finally, the products of the reaction must diffuse away from the reaction site to
be replaced by a fresh substrate molecule. This process is often complicated by
the fact that reaction products are often hydrophobic in character and remain
concentrated in the substrate, thereby influencing the rate of hydrolysis simply
by diluting the substrate and altering the manner in which the enzyme interacts
with the substrate–water interface.
One of the characteristic features of the hydrolysis of lipid substrates by their
respective enzymes is the existence of a time delay or lag period before
activation of the enzyme is observed. The lag period in activation of
phospholipase c from Bacillus cereus has been investigated by assay of activity
of the enzyme against large unilamellar vesicles composed of different substrate
mixtures (Ruiz-Arguello et al., 1998). A lag period in activation of the enzyme
was noted in substrate mixtures consisting of phosphatidylcholine (the normal
substrate for the enzyme), phosphatidylethanolamine (hydrolysed to a lesser
extent), sphingomyelin and cholesterol (neither are substrates for the enzyme).
The duration of the lag period was found to vary between 8 s and more than
30 min depending on the proportion of the different lipids in the substrate
mixture. Furthermore, the maximum rate of hydrolysis varied from 0.4 min–1 to
more than 55 min–1, again depending on the particular substrate mixture. The
presence in the substrate of lipids that tended to destabilize bilayers and favour
hexagonal-II structure, such as phosphatidylethanolamine and cholesterol, was
found to enhance activity and shorten the lag phase. Conversely, lipids that are
known to stabilize bilayers, such as sphingomyelin, had the opposite effect on
enzyme activity. This suggests that while the substrate is presented in the form
of a bilayer the enzyme is sensitive to the local instability that assists orienta-
tion of the substrate about the active site of the enzyme. The interesting feature
of all reaction mixtures, however, was that the proportion of the substrate
hydrolysed during the lag period was invariably 0.10% of the total substrate
present. The explanation for this observation is that the creation of local rafts of
bilayer enriched in the diacylglycerol product of the reaction results in aggre-
gation of the vesicular substrate which, in turn, is responsible for the acceleration
in the rate of hydrolysis.
An indication of the effect of reaction products on the physical properties of
the substrate has been obtained from studies of the effect of phosphatidic acid
in monolayers of phosphatidylcholine subjected to hydrolysis by
phospholipase D from Streptomyces chromofuscus (El Kirat et al., 2002). The
enzyme is a member of a superfamily that includes endonucleases, helicases,
lipid synthetases and enzymes able to catalyse the formation or hydrolysis of
phosphodiester bonds. In its reaction against a substrate of phosphatidylcholine
a phosphatidyl enzyme intermediate is formed which is subjected to a nucle-
ophilic substitution by a molecule of water to release phosphatidic acid. The
effect of phosphatidic acid in monolayers of substrate at the air–water interface
on the surface elasticity modulus (Ks) and the lag period before accelerated
hydrolysis is observed is shown in Figure 7.7. The surface elasticity modulus is
derived from the surface pressure–area isotherm from the relationship:
⎛ ∂π ⎞
Ks = – A ⎜ –– ⎟
⎝ ∂A⎠
where A is the molecular area at the corresponding surface pressure, π. The
greater the value of Ks for a monolayer the less it is subject to deformation. The
correlation between Ks and lag time shown in Figure 7.7 suggests that deforma-
tion of the substrate–water interface is an essential step in orienting the enzyme
about the phospholipid substrate, and that the product is instrumental in
modulating this process.
A number of studies have examined the effects of substrate presentation and
the molecular species preferences of secretory phospholipase A2. This enzyme
hydrolyses the fatty acid esterified to the sn-2 position of the glycerol backbone
of diacylglycerophosphatides. The activities of the different subclasses of this
enzyme are known to be modulated by proteins and peptides such as melittin
and phospholipase A2-activating protein that are believed to act by modifying
the manner of presentation of the substrate (Koumanov et al., 2003). The
effects of such proteins in activating the enzymes differ depending on whether
the substrate is in the form of a dispersion of pure lipid or is present in a
biological membrane. These differences highlight the role of regulatory peptides
in the biological function of these phospholipases. The presence of non
Figure 7.7 The value of the surface elasticity modulus (Ks) at 30 mN•m–1 and the enzyme activity lag
time (determined as the time preceding a 5% decrease in monolayer area) plotted against the proportion
of phosphatidic acid mixed with the phosphatidylcholine substrate. Data taken from El Kirat et al., 2002.
substrate lipids is also found to influence catalytic activity. Type-II secretory

phospholipase A2 can be ‘activated’ by displacement from an interface com-
prised of susceptible diacylglycerophospholipid substrate by binding to an
interface of non-susceptible phospholipid such as sphingomyelin. The sphin-
gomyelin may be presented in the form of a separate dispersion added to the
assay mixture or as a phase-separated domain within the bilayer of substrate
molecules. The affinity of the enzyme for the sphingomyelin interface may, in
turn, be reduced by association of the sphingomyelin with cholesterol to form
a liquid-ordered phase. This effect is observed as ‘activation’ of the enzyme by
cholesterol.
The action of one phospholipase can trigger the activity of another phos-
pholipase with different substrate specificity. Such actions may represent
manifestations of the highly complex biochemical homeostatic mechanisms
that are present in living cells and are responsible for maintaining the molecular
species composition of cell membranes within relatively narrow limits. One
example is the activation of secretory phospholipase A2 by ceramide, the
product of the action of sphingomyelinase on its substrate, sphingomyelin
(Koumanov et al., 2002). The activation of secretory phospholipase A2 by
ceramide is additional to ‘activation’ by release of the enzyme bound to
sphingomyelin domains. The activation is apparently due to the creation of
phase-separated domains of bilayer and hexagonal-II phase in the substrate
dispersion. The presentation of diacylglycerophospholipid substrate at the

interface between the two phase structures would appear to be the most
plausible explanation for the enhanced activity of the enzyme. This may be
related to the observation that substrate molecular species with a polyunsatu-
rated fatty acid, such as arachidonic acid, in the sn-2 position of the glycerol
were much preferred as substrates compared with molecular species with
relatively saturated fatty acids acylated to the sn-2 position.
The substrates for triacylglycerol lipases are usually presented in the form of
an emulsion such as dietary fat droplets or lipoproteins. These particles consist
of substrate presented in a bulk phase which is stabilized at the aqueous
interface by a monolayer surface phase composed of phospholipid and other
proteins. The enzyme catalysis takes place in the surface phase, so both the
substrate and the enzyme must partition from their respective bulk phases into
the interfacial region for reaction to occur. In many cases stabilization of lipid
emulsions by phospholipids impedes access of the enzyme to its substrate and
this is evidenced by a significant lag period that may last for hours before lipase
activity accelerates towards equilibrium. Such a lag period can often be
reduced or even eliminated by the introduction of reaction products, such as
diacylglycerol or fatty acids; some lipases require cofactor proteins to augment
their catalytic function (Brockman, 2000).
E. Interaction of proteins and lipids in membranes

The association of proteins with lipids represents a major structural feature of
all living cells, particularly the membranes. The other form of lipid–protein
complex is the lipoproteins; the properties of these structures are examined in
the following section.
The contemporary model of membrane structure derives from the fluid-
mosaic model proposed by Singer & Nicolson (1972). The amphipathic lipids
of the membrane are arranged in a bilayer configuration which serves as a fluid
matrix for association of the membrane proteins. A cartoon of the fluid-mosaic
model of membrane structure is shown in Figure 7.8. The individual compo-
nents distribute randomly within the plane of the structure except where,
through the sum of short-range interactions, the assembly of specialized
structures such as membrane junctions takes place. While the lateral diffusion
of both proteins and lipids may occur at a relatively fast rate in the plane of the
membrane and about an axis perpendicular to it, the reorientation of molecules
within the membrane plane is a rare or prohibited event.
1. Intrinsic and extrinsic proteins

The membrane proteins were originally divided into two categories: intrinsic
or integral proteins, and extrinsic or peripheral proteins. The former were
Figure 7.8 A cartoon depiction of the fluid-mosaic model (after Singer & Nicolson, 1972).
characterized by interaction with the hydrophobic interior of the lipid bilayer

matrix while the latter were confined to association with the lipid at the aqueous
interface. The distinction between these two groups of proteins was obtained
operationally in experiments of the type reported by Tanner & Boxer (1972).
Their experiments were performed on human erythrocyte membranes, which
were shown by SDS-PAGE to contain 6 major proteins comprising approxi-
mately two-thirds of the total membrane protein. When membrane suspensions
prepared in isotonic salt solutions were exposed to low salt concentrations in
the presence of a chelating agent, two of the major proteins, one identified as
spectrin, were removed from the membrane. Examination of the structure of the
extracted membranes by freeze-fracture electron microscopy showed no change
in the size or distribution of the membrane-associated particles in the hydro-
phobic interior, suggesting that the extracted proteins did not associate with the
acyl chains of the lipids. It was concluded that these proteins were attached to
the membrane predominantly by electrostatic interactions. Subsequent expo-
sure of the membranes extracted with low salt concentration to a 1 M salt
solution released almost quantitatively another protein, identified as the glyco-
lytic enzyme D-glyceraldehyde 3-phosphate dehydrogenase. As expected, all
the enzyme activity originally associated with the membrane was recovered in
the salt extract. High salt concentrations were also found to cause disruption of
the membrane structure with the formation of small membrane vesicles in
which the remaining components were likely to have undergone considerable
rearrangement. This was consistent with the disappearance of membrane-
associated particles from the freeze-fracture plane of the membrane interior. At
this stage about half of the original protein had been extracted from the
erythrocyte membrane. Further extraction of the membrane vesicles with
aqueous pyridine removed another major protein, identified as the protein
carrying the MN-antigenic site. The solvent treatment resulted in disruption of
the bilayer integrity, consistent with the removal of proteins that interact with
the hydrocarbon chains of the lipids. The remaining two major proteins of the
erythrocyte membrane could only be extracted by replacement of the
membrane lipids with detergent. The intrinsic proteins could therefore be

distinguished from the extrinsic proteins as they required disruption of the
bilayer arrangement of the membrane lipids to remove them from the
structure.
The conformation of extrinsic proteins is determined by forces that operate
in the folding of soluble proteins in an aqueous environment, and the secondary
and tertiary structure resembles that of typical soluble proteins. The structure of
intrinsic proteins, however, is different because they are adapted to interact
with the hydrocarbon domain of the lipid bilayer. The most common adaptation
is observed in the primary structure, in which segments of the polypeptide
chain numbering about 20–25 residues are comprised predominantly of amino
acids with hydrophobic side chains. Such segments are often arranged in a
helical structure and span the lipid bilayer from one side to the other. It has been
suggested that the precise length of the membrane-spanning segment is matched
to the thickness of the lipid bilayer matrix in the sorting of intrinsic membrane
proteins by lipid domain formation. Some intrinsic membrane proteins, such as
glycophorin, have only one membrane-spanning hydrophobic segment. In the
case of glycophorin this segment tends to associate with another glycophorin
molecule to form a dimer, and this represents the functional configuration of
the protein. Another large group of integral membrane proteins, numbering
more than 150 well characterized examples, have a motif consisting of seven
transmembrane helices bundled together to form a channel. Well represented
examples in this group are the G-protein coupled receptors responsible for a
variety of cell signalling processes, and bacteriorhodopsin of the purple
membrane of Halobacterium halobium. Ion channels are an example where
monomers of transmembrane proteins associate into oligomers such as tetramers
in which each of the four subunits has a pair of membrane-spanning helices that
bundle together to form a water-filled channel. An exception to the transmem-
brane helical motif is represented by the bacterial porins, a class distinguished
by the creation of transmembrane pores by the formation of tetramers with each
subunit consisting of β-strands that create a wall surrounding a central channel.
A band of aliphatic side chains and a border of aromatic amino acid side chains
serve to position the protein in the membrane.
A method used to determine the topography of the protein with respect to the
lipid bilayer is to construct a hydropathy plot, which relates the relative polarity
of the amino acid side chains with their position in the primary structure of the
protein. Hydropathy plots of many intrinsic proteins show that the polypeptide
chain may pass many times across the membrane. Some have C- and N-termini
on the same side of the membrane, while in others they are on opposite sides.
Intrinsic proteins also differ in the extent to which they extend into the
protoplasmic or cytoplasmic surfaces of the structure.
Figure 7.9 Glycerylphosphorylinositol components of GPI-anchored proteins.
2. Lipid-anchored membrane proteins

Treatment of cell membranes with phospholipases releases some proteins
which otherwise could not be removed by manipulation of the electrostatic
interactions between the lipids and proteins. This observation led to the
recognition of a third class of membrane proteins that are tethered to the lipid
bilayer by covalent attachment to a lipid. Such proteins displayed on the outer
surface of the plasma membrane are anchored by derivatives of
glycerylphosphorylinositol, the so-called GPI-anchored proteins. Two types of
GPI lipid anchors are shown in Figure 7.9. One consists of diacyl-
glycerophosphorylinositol, in which the inositol sugar of the phospholipid is
attached via an extended glycosyl chain (GP) to the protein. The other is
exemplified by acetylcholine esterase (AChE), in which sn-1 alkyl, sn-2 acyl
glycerophosphorylinositol is also esterified to a fatty acid via the inositol ring.
The proteins that are typically anchored by GPI tethers are involved in
transmembrane signalling or recognition processes. The lipid anchor itself
provides a ligand allowing the protein to be selectively filtered during
endocytotic or exocytotic pathways. Two mechanisms of sorting are possible
for cycling GPI-anchored proteins from the cell surface through the endosome
compartment and back to the cell surface (Overath & Engstler, 2004). One is
positive sorting, whereby the GPI-anchored protein is concentrated within the
endosome membrane and buds off in vesicles enriched in the protein. Alterna-
tively, negative sorting involves the gradual enrichment of the endosome itself
by the removal of membrane depleted in the GPI-anchored protein. There is
Figure 7.10 Configurations of (a) fatty acid and (b) isoprenyl anchors of membrane proteins (P).
also a distinction between GPI-anchored proteins that are cycled through the
endosome compartment via clathrin-coated vesicles (such as prion protein) and
those that are not (including thy-1). There is accumulating evidence that the
lipid composition of membranes and the creation of lipid microdomains plays
a key role in the sorting of different GPI-anchored proteins during both
endocytosis from the cell surface and repackaging in the secretory process
(Mayor & Riezman, 2004).
The lipid anchors tethering proteins to the cytoplasmic surfaces of cell
membranes are either long-chain fatty acids or prenyl lipids. Cytosolic proteins
like v-Src associate with the plasma membrane via a single fatty acid, typically
myristate or palmitate, attached to an N-terminal glycine residue of the protein.
Other cytosolic proteins are anchored to the plasma membrane by prenylation
of one or two cysteine residues at or adjacent to the C-terminus of the protein;
examples of this group include Ras and Rab proteins. The structures of these
lipid anchors comprised of farnesyl (C15) and geranylgeranyl (C20) groups are
shown in Figure 7.10. The possession of a lipid anchor is not in itself sufficient
to permanently anchor the protein to the membrane; additional sites of interac-
tion with the lipids are also present. These additional attachment sites may be
a polybasic patch, exemplified by the six consecutive lysine residues in Ras that
bind to the head groups of acidic phospholipids. Alternatively, additional
palmitoylation via labile thioester linkages may strengthen the association of
the protein with the cytoplasmic membrane surface.
The lipidation of proteins appears to be a dynamic process, and palmitoylation
is not invariably accomplished with palmitic acid. Thus other fatty acids such
as myristic, stearic, oleic, linoleic and arachidonic acids have also been
identified at additional lipidation sites on lipid-anchored proteins. Moreover,

some cytoplasmic and even transmembrane proteins are known to be
palmitoylated, usually at two or more sites, without prior myristoylation or
palmitoylation. The case of the trimeric G-proteins is particularly interesting.
The α-subunit is myristoylated, palmitoylated or both while the γ-subunit is
prenylated. In its inactive resting state the trimeric αβγ oligomer is known to be
excluded from membrane lipid rafts, but upon activation the α-subunit is
targeted to raft domains while the βγ subunit complex remains in the fluid
membrane matrix (Moffett et al., 2000). Similarly, phosphorylation of the
β2-adrenergic receptor upon agonist binding and subsequent dephosphoryla-
tion during inactivation has been found to correspond with palmitoylation of
the receptor and its subsequent deacylation (Moffett et al., 2001).
F. Lipoproteins
Proteins can interact with lipids to form water-soluble complexes. Such
complexes are exploited in the mobilization of complex lipids within living
organisms. There are two main types of complex: complexes formed between
monomeric proteins and lipids, and large lipoprotein complexes.
1. Albumin
Free fatty acids bind to a variety of proteins during mobilization in the body.
They are transported in the bloodstream as a complex with plasma albumin.
Plasma albumin is a conformerically flexible protein that can adopt multiple
conformations of approximately equal energy to accommodate the binding of
ligands. The protein has at least five fatty acid binding sites, three of which are
of significantly higher affinity than the remainder. The mechanism of binding
of the fatty acids and other lipophilic drugs to the protein has been investigated
by nuclear magnetic resonance (NMR) methods (Lucas et al., 2004). The
average residence lifetime of a long-chain fatty acid bound to a high-affinity
site was found to be greater than 66 ms, whereas short-chain fatty acids like
octanoic acid have residence times of only a few milliseconds. The lifetimes are
relatively short because the fatty acids exchange readily between different
binding sites on the protein. Dissociation of the fatty acid from the protein, on
the other hand, takes place on a timescale of seconds.
The dissociation of fatty acids from plasma albumin is the rate-limiting step
in the delivery of fatty acids to target cells. Their dissociation from the complex
at the site of entry into cells is assisted by the presence in the plasma membrane
of proteins with a high affinity for fatty acids (McArthur et al., 1999). One such
protein, membrane fatty acid-binding protein (FABPpm), binds tightly to free
fatty acids and prevents destabilization of the membrane due to the presence of
the free fatty acid in the structure. The entry of the free fatty acid into the outer
leaflet of the membrane lipid bilayer matrix establishes a transmembrane

gradient which can be dissipated by flip-flop of the uncharged fatty acid from
the outer to the inner membrane leaflet. This process can take place spontane-
ously or may be assisted by another fatty acid-binding protein of the membrane,
fatty acid translocase (FAT/CD 36). Another soluble fatty acid-binding protein
present in the cytoplasm, FABPc, serves to remove free fatty acids present on
the cytoplasmic leaflet and to transport them through the cytoplasm. There is
also evidence that fatty acids also bind to caveolin-1. This protein is a
component of caveolae, which are thought to deliver lipid to the different
subcellular organelles. It remains unclear whether or not intracellular traffick-
ing of fatty acids mediated by FABPc and by the vesicular caveolae mechanism
act in concert with one another.
Apart from transport to the cell, metabolic mobilization relies on activation
of long-chain acyl-coenzyme A (acyl-CoA). This process is mediated by a
family of fatty acid-transport proteins referred to as FATP 1–5/6, which are
known to possess acyl-CoA synthetase activity. There is evidence from the
variable nature of the N-terminus of these fatty acid-binding proteins that they
may be able to deliver fatty acids to particular membrane sites within the cell.
The activation of fatty acids takes place at the highly conserved AMP-binding
site located at the cytosolic domain of FATP whence they are primed for
metabolism at the appropriate organelle.
2. Serum lipoproteins
Lipoprotein complexes circulate in the mammalian bloodstream to distribute a
cargo of lipids from their site of synthesis, usually the liver, to the peripheral
tissues. There has been an extensive research effort to characterize lipoproteins
because of their association with heart disease and atherosclerosis. The lipids,
mainly triacylglycerols, cholesterol and cholesterol esters, occupy a central
core surrounded by a shell of polar lipids and proteins. The proteins act to
stabilize the lipid droplet and provide recognition sites for targeting the
complex to the appropriate site of delivery. The particles are usually spherical
in shape and are classified according to their buoyant density. The characteris-
tics of the different classes of human lipoproteins are presented in Table 7.3. As
expected, the density increases with increasing protein to lipid ratio. The
protein components are synthesized in the liver and small intestine. While
many of the ten major apoproteins found in lipoproteins are common to more
than one class of lipoprotein, their combination is distinct in each of them.
Chylomicrons are the least dense of the lipoproteins and are responsible for
packaging fats, cholesterol and other lipids taken up from the diet in the
bloodstream and conducting them about the body. Because they consist almost
entirely of triacylglycerols they have a buoyant density of <0.95 g•cm–3. The
major protein is apolipoprotein B-48 (apoB-48), which has a molecular mass of
Table 7.3 The size and composition of mammalian lipoproteins
Lipoprotein Density Diameter Triacylglycerol: Apoprotein

(g•cm–3) (nm) cholesterol ratio
Chylomicrons <0.95 1000 10 B-48, C, E

Very-low-density (VLDL) 0.95–1.006 50 2.3 B-100, C, E
Intermediate-density (IDL) 1.006–1.019 30 1.0 B-100, E
Low-density (LDL) 1.019–1.063 20 0.2 B-100
High-density (HDL) 1.063–1.210 10 0.03 A
240 kDa and forms an amphipathic shell around the spherical fat globule, in
which the interior surface of the protein is hydrophobic and the exterior is
hydrophilic. Triacylglycerols and cholesterol are also synthesized in the liver,
and amounts in excess of requirements are packaged into very-low-density
lipoproteins and exported to the peripheral tissues. The proteins stabilizing
very-low-density lipoproteins include apoB-100 and apoE. ApoB-100 is an
extremely large protein comprised of more than 4500 amino acid residues and
has a molecular mass of 513 kDa. ApoB-48 is formed from the first 48% of
apoB-100 and arises from the post-transcriptional editing of apoB-100 mRNA
in the intestine. The relationship between apoB-100 and apoB-48 has been the
subject of considerable interest (Brodsky et al., 2004).
The intermediate-density lipoproteins result from depletion of the
triacylglycerol content of very-low-density lipoproteins by the action of lipases
associated with capillary surfaces and their consequent enrichment in choles-
terol esters. These intermediate-density lipoproteins may be taken up by the
liver and further processed or converted into low-density lipoproteins by
hydrolysis of more triacylglycerol. Low-density lipoprotein is the major carrier
of cholesterol and consists of a core of about 1500 cholesterol molecules
esterified mainly to linoleate. The non-polar lipid is stabilized by a monolayer
of phospholipid and apoB-100. High-density lipoprotein is also involved in
cholesterol transport but the source of cholesterol is that scavenged from
apoptosing and dying cells and membranes undergoing metabolic turnover.
The cholesterol is esterified to a long-chain fatty acid in a reaction catalysed by
an acyltransferase intrinsic to the high-density lipoprotein. The cholesterol
esters are rapidly transferred to very-low-density or low-density lipoproteins
by a specific transfer protein or targeted to the liver in their high-density
lipoprotein vector.
ApoA, apoC and apoE are referred to as exchangeable apolipoproteins and
they are responsible for regulating the traffic of lipids into and out of cells by
acting as cofactors for plasma enzymes and ligands for cell-surface receptors.
The exchangeable apolipoproteins share the same genomic heritage and there-
fore possess structural similarities (Saito et al., 2004). One particular feature is
nonpolar surface
positively-charged
residues Class A
negatively-charged
residues
Class Y
Class G
Figure 7.11 Helix wheel plots of the three classes of α-helical segments found in apoA-1, apoA-IV and
apoE apolipoproteins (adapted from Segrest et al., 1992).
a primary sequence arranged in α-helical motif in which the basic residues are
located near the hydrophilic–hydrophobic interface and acidic residues are
clustered at the centre of the polar face. The helical segments are often
interrupted by proline residues. This so-called Class A motif, which is exempli-
fied by apoA-1 lipoprotein, is illustrated together with Class Y and Class G
motifs in Figure 7.11. ApoE isoforms have a predominance of Class G helical
motifs and the Class Y helical motif dominates the secondary structure of
apoA-IV. The N-terminal amphipathic helices of apoE are bundled into 4
antiparallel strands forming an elongated globular structure with the hydropho-
bic faces oriented into the interior. The C-terminal adopts a coiled-coil helix
structure that is more exposed to the aqueous phase. In the absence of lipid,
apoA-1 adopts a two-domain configuration similar to apoE: an N-terminal
helical bundle extending into the central part of the primary structure and a
C-terminal helical domain that is less well organized. Both apoA-1 and apoE
aggregated into oligomers in the absence of lipid via hydrophobic interactions
between residues located at the C-termini of the respective molecules.
The self-association in aqueous media via the C-terminal domains of
apoA-1 and apoE is consistent with the role of the C-termini in the interaction
of these proteins with lipids. When apoE binds to lipid it has been suggested
that the initial binding takes place at the C-terminal, which then induces the
4-helix bundle of the N-terminus to reorganize so that the hydrophilic faces of
the helices open out and become available for binding to the lipid. Similar
reorganizations are believed to occur when apoA-1 interacts with lipid. Thus,
an initial binding takes place at the C-terminal domain, which is arranged in an
elongated hairpin structure. Following this there is a conformational change
involving residues 1–43 which serves to unmask a hydrophobic domain in
residues 44–65 of the protein.
The driving force for the formation of complexes between the apolipoproteins
and lipids appears to be a favourable change in enthalpy. A conformational
transition from random coil to α-helix on binding of apoA-1 to lipid, for
example, is associated with an enthalpy change of the order of –5 kJ•mol–1
per α-helical segment. This represents a total enthalpy change of about
–11 kJ•mol–1 and is additional to the enthalpy change accompanying the
interaction of apoA-1 with lipid, which is about –40 kJ•mol–1. The change in
conformation of the protein therefore contributes significantly to the binding
affinity between the protein and the lipid.
Two conformers of high-density lipoprotein have been characterized, one
discoid in shape and the other spherical. The discoid particles are comprised of
a phospholipid bilayer disk with two molecules of apoA-1 encircling the edges
where the acyl chains are exposed. The size is limited by the length of the
apoA-1 molecules, which are arranged in a belt of α-helices stacked one on the
other in an antiparallel orientation. A similar discoid particle has also been
described for apoE but in this structure 4 molecules of the protein are arranged
at the periphery of the disk. The spherical form of high-density lipoprotein
varies in diameter and has more neutral lipid than the discoid form. The
amphipathic helices of the apolipoproteins are believed to be interpolated
between the phospholipid molecules rather than at the periphery. The extent of
interaction of the protein with the lipid is greater as the proportion of protein in
the particle decreases.
G. Cubic lipid–water phases

Cubic lipid–water phases are known to form ternary complexes with proteins,
and such complexes have been exploited in the crystallization of proteins for X-
ray diffraction examination. A partial phase diagram of the ternary system
lysozyme–monoolein–water has been constructed (Ericsson et al., 1983) and is
Figure 7.12 Ternary phase diagram of lysozyme–monoolein–water system.
presented in Figure 7.12. The most remarkable feature of the phase diagram is
that a relatively large proportion of the water-soluble protein can be incorpo-
rated to form a lipid–protein–water phase without any ionic interactions of the
type illustrated in monomolecular films described above. It was also found that
the protein occupied the aqueous phase of the complex in its native configura-
tion. As discussed below, this discovery was to have considerable implications
for the crystallization of both soluble and membrane proteins. The incorp-
oration of protein results in an expansion of the cubic phase lattice formed by
monoolein–water. The cubic domain of the phase diagram shown in
Figure 7.12 contains all three fundamental cubic structures observed in lipid–
water systems of the type found in monoolein–water, namely the gyroid surface
CG, the diamond surface CD, and the primitive surface CP, in order of increasing
protein:water ratio.
Complex cubic phases are formed with ternary lipid–water systems like
monoolein mixed in proportions of two parts protein solution or dispersion
with three parts of lipid. When such mixtures are treated with precipitants such
as non-ionic detergents or salts the protein begins to crystallize within hours of
incubation at 20ºC. The method can be used to grow crystals of soluble as well
as membrane proteins and other organic and inorganic molecules.
The precise process of crystallization from these tertiary lipid phases has
been examined in some detail (Misquitta et al., 2004). Precipitants like Na+/K+
phosphate salts, for example, provoke a reduction in water activity which
favours protein–protein interactions. Three-dimensional structures are created
when protein–protein contacts are established between successive layers. The

key to successful crystallization of proteins is the action of the precipitant to
destabilize the cubic phase of monoolein, which is the principal host in the
lipid–protein complex, so that it tends to form a liquid-crystalline lamellar
phase. A group of agents used to induce the transition are the alkyl glycosides,
in which hexyl, octyl, nonyl or decyl hydrocarbon chains are combined with
sugar residues of glucose or maltose.
References
Angeletti, S., Maggio, B. & Genti-Raimondi, S. (2004) Biochem. Biophys. Res. Commun.
314, 181–185.
Brezesinski, G. & Mohwald, H. (2003) Adv. Colloid Interface Sci. 100–102, 563–584.
Brockman, H.L. (2000) Biochimie 82, 987–995.
Brodsky, J.L., Gusarova, V. & Fisher, E.A. (2004) Trends Cardiovasc. Med. 14, 127–132.
El Kirat, K., Besson, F., Prigent, A.F., Chauvet, J.P. & Roux, B. (2002) J. Biol. Chem. 277,
21231–21236.
Ericsson, B., Larsson, K. & Fontell, K. (1983) Biochim. Biophys. Acta 729, 23–27.
Koumanov, K.S., Momchilova, A.B., Quinn, P.J. & Wolf, C. (2002) Biochem. J. 363, 45–51.
Koumanov, K., Momchilova, A. & Wolf, C. (2003) Cell Biol. Int. 27, 871–877.
Lahdo, R. & De La Fourniere-Bessueille, L. (2004) Biochem. J. 382, 987–994.
Lavoie, H., Desbat, B., Vaknin, D. & Salesse, C. (2002) Biochemistry 41, 13424–13434.
Lucas, L.H., Price, K.E. & Larive, C.K. (2004) J. Am. Chem. Soc. 126, 14258–14266.
Mayor, S. & Riezman, H. (2004) Nat. Rev. Mol. Cell Biol. 5, 110–120.
McArthur, M.J., Atshaves, B.P., Frolov, A., Foxworth, W.D., Kier, A.B. & Schroeder, F.
(1999) J. Lipid Res. 40, 1371–1383.
Misquitta, Y., Cherezov, V., Havas, F., Patterson, S., Mohan, J.M., Wells, A.J., Hart, D.J. &
Caffrey, M. (2004) J. Struct. Biol. 148, 169–175.
Moffett, S., Brown, D.A. & Linder, M.E. (2000) J. Biol. Chem. 275, 2191–2198.
Moffett, S., Rousseau, G., Lagace, M. & Bouvier, M. (2001) J. Neurochem. 76, 269–279.
Overath, P. & Engstler, M. (2004) Mol. Microbiol. 53, 735–744.
Quinn, P.J. & Dawson, R.M. (1970) Biochem. J. 116, 671–680.
Ruiz-Arguello, M.B., Goni, F.M. & Alonso, A. (1998) Biochemistry 37, 11621–11628.
Saito, H., Lund-Katz, S. & Phillips, M.C. (2004) Prog. Lipid Res. 43, 350–380.
Segrest, J.P., Jones, M.K., De Loof, H., Brouillette, C.G., Venkatachalapathi, Y.V. &
Anantharamaiah, G.M. (1992) J. Lipid Res. 33, 141–166.
Singer, S.J. & Nicolson, G.L. (1972) Science 175, 720–731.
Soulages, J.L., van Antwerpen, R. & Wells, M.A. (1996) Biochemistry 35, 5191–5198.
Tanner, M.J. & Boxer, D.H. (1972) Biochem. J. 129, 333–347.
Zidovetzki, R. & Lester, D.S. (1992) Biochim. Biophys. Acta 1134, 261–272.
CHAPTER 8
Emulsions
Emulsions are dispersions of one liquid phase in a continuous phase, which

also is liquid. In Chapter 6, some of the dispersed lipid–water phases discussed
were liquid, and they might therefore also be described as emulsions. Here,
however, we will only discuss emulsions consisting of water and fats/oils, i.e.
those in which the oil phase consists of triacylglycerols. If the oil phase is
emulsified in water, the emulsion is an oil-in-water (o/w) emulsion. The inverse
emulsion, with water droplets in an oil continuum, is called a water-in-oil
(w/o) emulsion. Important applications of lipid emulsions are found in foods,
and certain food aspects of emulsions are covered in Chapter 12.
Among the first things we learn in chemistry and physics is that oil and water
do not mix. Recently, however, this wisdom has been challenged (Pashley,
2002). Pashley found after removing the dissolved gases from an oil–water
mixture that a cloudy emulsion formed spontaneously. There have also been
speculations on technical applications of this phenomenon (e.g. see New
Scientist, 22 February 2003, p.17). Here, however, we will only consider
conventional emulsions produced by traditional technology.
A. Oil-in-water emulsions
1. Emulsification
Dispersing one liquid phase in another requires energy input, which is directly
proportional to the increase in interfacial area and to the interfacial tension.
This energy can be provided by a stirrer, for example an ‘Ultra-Turrax’. The
higher the shear rate, the smaller are the droplets formed. Various mechanisms
behind droplet formation have been described by Walstra (1973). A droplet in
laminar flow will burst when the velocity gradient exceeds a critical value. The
viscosity of the dispersed phase in relation to the continuous phase is therefore
important. Turbulent flow is more efficient when the continuous phase has low
viscosity, for example.
If different equipment is compared, the particle size is reduced when an
Ultra-Turrax process is replaced by ultrasonication, and further reduced when
a valve homogenizer is used (the mixture is forced through a valve, resulting in
a pressure gradient). A log–log plot of the droplet diameter versus energy input
shows a linear relation. Pressure fluctuations result in cavity formation, which
is a major mechanism in the valve homogenizer and during ultrasonication. At
very high viscosity a colloid mill is the best equipment for emulsification.
175
Figure 8.1 Schematic illustration of a common globular protein, like β-lactoglobulin (left), and a
protein with a flexible peptide chain, like β-casein (right) at the oil–water interface.
2. Protein-stabilized emulsions
This is a common emulsion type in foods. The protein used will orient and
change its conformation at the oil–water interface in order to reduce the
interfacial energy. In the early literature it was usually assumed that the protein
molecules were unfolded. Studies by ellipsometry and atomic-force microscopy
have since modified this picture (e.g. see Wahlgren & Arnebrant, 1991). Thus
the molecules often tend to be globular and compact. There are two major
arrangements, as indicated in Figure 8.1.
Proteins with globular tertiary structure, like whey proteins, adsorb with a
conformation close to the native state. Proteins lacking a well-defined tertiary
structure, on the other hand, such as casein molecules, are more unfolded, with
the hydrophilic segments extending out into the water phase. The driving force
is the charges along the peptide chain, usually with a dominance of negative
groups. This provides an efficient steric stabilization, and the caseins are very
powerful emulsifiers. It takes several minutes for a protein to reach a plateau
value of interfacial tension, which means that after emulsification, various
relaxation phenomena take place. Ageing phenomena in interfacial protein
films have been followed by Dickinson (1992). In interfacial films of
β-lactoglobulin, an increase in film viscosity was seen corresponding to
disulphide bridge formation.
The main factor behind the high stability of emulsions based on proteins is
the irreversible character of the adsorption of the protein molecules at the
interface. Thus the peptide chain is attached at several positions and desorption
would require that all these positions are detached at the same time, which is
very unlikely.
EMULSIONS 177
Figure 8.2 Structure of an emulsion stabilized by a PC monolayer and a bilayer phase at the oil–water
interface.
3. Lipid-stabilized emulsions
In most cases when polar lipids are used to stabilize emulsions, they form a
separate phase at the oil–water interface. We might say that they form
‘interphases’. If a liposomal dispersion is formed by phosphatidylcholine (PC)
in water, the liposomes can encapsulate an introduced oil phase by mechanical
opening of the bilayer structure. The first emulsion used for parenteral nutri-
tion, ‘Intralipid’, prepared from egg yolk PC and soybean oil, consists of such
a lamellar liquid-crystalline phase at the oil–water interface, as shown in
Figure 8.2 (see also Chapter 11).
Friberg and co-workers introduced phase equilibria into discussions of lipid
emulsion stability (Friberg et al., 1969), which has been a fruitful approach.
When there is equilibrium between oil, water and the Lα phase, a stable
emulsion with a structure like that shown in Figure 8.2 can be obtained.
Such emulsions can be further stabilized if the Lα phase is transformed into
a gel phase (Larsson, 1978). Such an interface can be prepared by cooling
during homogenization through the temperature range at which the bilayers
will crystallize (Figure 8.3).
Technical emulsification processes used for foods and for pharmaceutical
products often involve homogenization under cooling. This results in forma-
tion of either a gel phase or surface-active crystals, which form a solid film at
the interface (Krog & Larsson, 1992). These leaf-shaped crystals expose a
polar head surface towards water and a methyl end group surface towards oil.
This phenomenon is neglected in the literature and will therefore be described
in detail here.
The interfacial tension at the sunflower oil–water interface at different
Figure 8.3 Effect of temperature on the oil–water interfacial tension with monostearin at different
concentrations solved in the oil (after Krog & Larsson, 1992).
Figure 8.4 Critical temperatures (Tγ) at which the interfacial tension starts to decline rapidly versus the
concentration of monomyristin (glycerol monomyristate; GMM), monostearin (GMS), and monobehenin
(GMB) (after Krog & Larsson, 1992).
EMULSIONS 179
Figure 8.5 Binary phase diagram for monomyristin–sunflower oil (after Krog & Larsson, 1992).
concentrations of monostearin in the oil is shown in Figure 8.3. At a certain

critical temperature, Tγ, the interfacial tension during cooling starts to decline
strongly. The variation of this temperature with the concentration of
monoacylglycerols of different chain length is shown in Figure 8.4. The shapes
of these curves follow the phase separation of monoacylglycerol crystals from
the oil phase. A corresponding phase diagram is shown in Figure 8.5.
Many technical emulsions are produced under these conditions, and it is
clear that the interface consists of crystals of the polar lipid. The interfacial
layer has also been separated and analysed by X-ray diffraction, showing that
the film consists of monoacylglycerol crystals.
Another practical demonstration of the significance of this phenomenon is
the destabilization of protein-based emulsion utilized in ice cream, where the
emulsion should undergo coalescence at melting of the ice crystals (see
Figure 8.6 Interfacial tension versus temperature of: (A) 0.2% (w/w) glycerol monostearate
(monostearin) in sunflower oil towards distilled water; (B) pure sunflower oil towards 0.01% milk
proteins in water; (C) 0.2% glycerol monostearate in sunflower oil towards 0.01% milk proteins in water.
The situation in (C) is desired in ice cream (after Krog & Larsson, 1992).
Chapter 12). Milk proteins can give overly stable emulsions, but the addition of
some monoacylglycerols will solve the problem by squeezing out the protein
molecules from the interface, with the effect shown in Figure 8.6.
B. Water-in oil emulsions

This type of emulsion is not very common. They are not as easy to stabilize as
o/w emulsions; electrostatic repulsion, for example, cannot be used. Butter and
margarine are often described as w/o emulsions, but this is not really true. The
water droplets are immobilized in the continuous fat phase by a crystal
network, and when the crystals melt the emulsion separates.
Aluminium and calcium soaps have been reported to be efficient stabilizers
of w/o emulsions (Busch & Neuwald, 1973). Wool wax is the lipid fraction
extracted from raw wool, and it has been used for more then a century to
stabilize w/o emulsions in skin care products. It consists of fatty acid esters of
sterols and fat alcohols, and an even better emulsifier is obtained from the
separated alcohols, dominated by cholesterol. The emulsification mechanism
is again crystallization of surface-active crystals at the oil–water interface, as
seen by X-ray diffraction during cooling (Hoppe & Larsson, 1981).
Formation of w/o emulsions is a problem in crude oil reservoirs. The oil
EMULSIONS 181
consists of hydrocarbons with small amounts of polar components, which act as

emulsifiers (McMahon, 1992).
C. Demulsification
Flocculation and coalescence limit the lifetime of emulsions. Sometimes it is
important to break down an emulsion, and the mechanisms involved are
therefore important to understand. Flocculation is reversible, but if the inter-
facial film becomes thinner, film rupture and coalescence take place.
Addition of electrolytes will reduce the electrostatic repulsion and therefore
favour demulsification. Increasing the temperature also favours demulsification,
by decreasing the viscosity of the film between the particles. Physical
demulsification involves separation methods like separation, filtration and
flotation. More efficient is chemical demulsification, for example the use of
organic solvents or surfactants that destroy the interfacial film structure.
References
Busch, G. & Neuwald, F. (1973) In: Proceedings of the VII Congress IFSCC, Gesellschaft
Deutsches Kosmetik-Chemiker, Hamburg, Germany, p.171.
Dickinson, E.I. (1992) In: Emulsions – A Fundamental and Practical Approach (Sjöblom, J.,
ed.), Kluwer Academic Publishers, London, UK, p.23.
Friberg, S.E., Mandell, L. & Larsson, M. (1969) J. Colloid Interface Sci. 29, 155.
Hoppe, U. & Larsson, K. (1981) J. Dispersion Sci. Technol. 2, 433.
Krog, N. & Larsson, K. (1992) Fat Sci. Technol. 94, 55.
Larsson, K. (1978) Prog. Chem. Fats Other Lipids 16, 163.
McMahon, A.J.M. In: Emulsions – A Fundamental and Practical Approach (Sjöblom, J.,
ed.), Kluwer Academic Publishers, London, UK, p.135.
Pashley, R.M. (2002) J. Phys. Chem. B 107, 2724.
Wahlgren, M. & Arnebrant, T. (1991) Trends Biochem. 9, 209.
Walstra, P. (1973) Chem. Eng. Sci. 48, 333.
CHAPTER 9
Lipids of biological membranes
A. Introduction
All biological membranes contain a highly complex assortment of polar lipids.
While there are relatively few major lipid classes there is a whole spectrum of
molecular species within each class, which differ in the type, length and
number of unsaturated residues representing their hydrophobic component.
There is a general consensus, based on observations that have been obtained by
a variety of biophysical methods, that the lipid constituents of biological
membranes are arranged in a bilayer configuration in which the polar groups
are located in contact with the aqueous medium on the outside and the
hydrocarbon substituents are oriented towards the interior to form a hydro-
phobic domain that excludes water. It has been argued on the basis of the
hydrophilic to hydrophobic balance within the molecules that membrane lipids
have a relatively low critical micellar concentration, and that a discrete
distribution of domains within the molecule is responsible for creating a stable
bilayer structure.
Although the bilayer arrangement appears to be the dominant phase of many
lipids in biological membranes (Sachs et al., 2003) this is not necessarily the
phase preferred by individual molecular species of lipid isolated from particu-
lar biological membranes. If polar lipids, for example, extracted from biological
membranes are dispersed in aqueous media a mixture of phases is seen at
temperatures relevant to the growth of the organism from which the lipids were
extracted. If the constituent lipids are separated into molecular species and then
dispersed in aqueous media they are found to form one of a number of well-
characterized phases. The structures include bilayer phases in which the
hydrocarbon components are arranged in a crystal lattice, a gel phase or a liquid
crystal configuration. In addition, a number of non-bilayer phases, such as
hexagonal-II, cubic phases, etc. (Yang et al., 2003), are also found at tempera-
tures approximating the growth temperature of the organisms from which the
lipid was extracted. It is argued that lipids which tend to form non-bilayer
phases are constrained into a bilayer arrangement by their interaction with
other components of the membrane.
The composition of the molecular species, even of membranes that perform
very simple functions, is highly complex, and individual molecular species of
lipid may number more than 100 distinct chemical entities. In general, indi-
vidual molecular species of lipids that will form any one of these particular
phases are likely to be constituents of most biological membranes. Of particular
183
interest, therefore, is the phase behaviour of complex lipid mixtures of the

type encountered in biological membranes, since this will provide information
which will help in understanding the factors that govern the structure and
stability of the lipid matrix of membranes. It is largely these factors
that determine the limits within which organisms will grow and ultimately
survive.
The lipid bilayer of biological membranes was originally regarded as a
simple fluid matrix with which the different intrinsic and extrinsic membrane
proteins interact. Indeed, reconstitution experiments have demonstrated that
certain functions of membrane proteins often require only incorporation into a
lipid bilayer, and few specific requirements for particular lipids have been
identified. The concept that the lipid matrix acts simply as a barrier to the
movement of solutes, however, was abandoned when it became clear that the
distribution of lipids on either side of the membrane was different and this
asymmetry was maintained by an active process which, if disturbed, resulted in
lethal consequences (Quinn, 2002). More recently, it has become clear that the
lateral distribution of lipid molecular species is also asymmetric, leading to the
formation of domains in the plane of the membrane (Edidin, 2001; London,
2002). The creation of these domains is believed to be the result of specific
interactions between particular membrane lipids located in one or the other
leaflet of the membrane. As a consequence of domain formation the distribu-
tion of membrane proteins is altered, and this alteration is associated with a
potentiation of signalling functions (Anderson & Jacobson, 2002; Kenworthy,
2002).
This chapter will review evidence of the complexity of membrane lipid
composition, how the lipid composition is established in cell membranes, the
biochemical processes that are in place to regulate membrane lipid composi-
tion, and how interactions between particular membrane lipids serve to create
domains within the membrane. It is only with an understanding of these
processes that the various functions that membranes perform, such as signal
transduction, fusion, etc., can be described at the molecular level.
B. Membrane lipid composition

The task of characterizing the lipid composition of cell membranes was, until
recently, a relatively tedious and inaccurate procedure. The application of a
variety of sophisticated analytical techniques to reveal the detailed molecular
species composition of a variety of cell membranes has led to a greater
appreciation of the roles of lipids in membrane function. The following section
presents some of the approaches currently used to analyse the molecular
species composition of biological membranes.
LIPIDS OF BIOLOGICAL MEMBRANES 185
1. Analysis of membrane lipid composition
The conventional methods of membrane lipid analysis involve an initial

separation of the lipid classes according to their polarity by thin-layer chroma-
tography (TLC) on a polar silica gel or by high-performance liquid
chromatography (HPLC) on an apolar grafted silica gel. Once the lipid class is
isolated, subsequent separation of the molecular species of each lipid class can
be undertaken. The final step requires a high degree of purity of each of the lipid
classes to obtain a reliable pattern of fatty acyl residues located at the sn-1 and
sn-2 positions of glycerolipids. The profile of fatty acid methyl esters is
obtained by gas–liquid chromatography (GLC) after an additional step consist-
ing of position-selective enzymatic release by purified phospholipases A1 or
A2. The direct assessment of molecular species of diacylglycerols (Gaskell &
Brooks, 1977) and ceramides (Vieu et al., 2002) can also be obtained by a one-
step HPLC and GLC procedure. For diacylglycerols, derivatization of the
primary hydroxyl group is required following enzymatic cleavage of the polar
head group by phospholipase C from Clostridium welchii or Bacillus cereus
(which also has sphingomyelinase activity). For phosphatidylinositols, the use
of a specific phospholipase C from Staphylococcus aureus is recommended.
The pre-purification of ether ethanolamine phosphatides is required to prevent
overlapping of the chromatographic patterns of diacyl, alkyl and alkenyl
compounds. Purification of the derivatized products (either acetylated or
trimethylsilylated) is also preferred before further analysis. Intramolecular acyl
migration from position sn-2 to sn-3 is common, and although it does not
change the fatty acid composition the positional distribution is altered. A
partial acyl migration which is dependent on pH results in the appearance of
1,3-diacylglycerol derivatives.
Recently a single-step analysis of the molecular species of endogenous
ceramides and of the ceramide moiety of sphingomyelins in biological sam-
ples, using GLC, has been introduced (Vieu et al., 2002). Silylated
sphingomyelins were quantitatively converted to monosilylated ceramide upon
injection into the chromatography system, whereas the free ceramides were
disilylated on the primary and secondary alcohol functions, as confirmed by
mass spectrometry (MS). The reproducible shift of the retention times between
the mono- and disilylated derivatives enables simultaneous quantification of
the different sphingomyelin and ceramide molecular species. Overlapping
diacylglycerols were first removed by mild saponification of the lipid extract.
Three major free ceramides (NC16:0, NC24:0, and NC24:1) were quantified in
HEPG2 and Chinese hamster ovary (CHO) cells (Lichtenbergova et al., 1998).
Upon induction of apoptosis in CHO cells by C6-ceramide, the disappearance
of the C6-ceramide, its partial conversion to C6-sphingomyelin, and the promi-
nent increase of NC16:0 ceramide could be detected. The method represents a
unique procedure for the simultaneous analysis of sphingomyelin and ceramide
molecular species at a level of sensitivity appropriate to monitor the variation

of the different pools in biological samples.
The straightforward procedure described above gives only the overall fatty
acid composition of diacylglycerols or ceramides as total carbon number and
number of double bonds. Therefore C36-Δ4 may hold for palmitoyl-arachidonyl
(C16:0-C20:4) as well as for dilinoleoyl species (C18:2-C18:2) or any other combina-
tion. A rapid MS procedure has been developed in which the direct introduction
(DCI) of a TLC-purified phospholipid class gives the pattern of the molecular
species in a single step. The procedure has been applied to screen the molecular
species of sphingomyelin of red blood cell membranes (Katsikas & Wolf,
1995). The conditions gave spectra with three discrete areas, corresponding to
ions originating from sphingoid bases, fatty acids and ceramides. Eight sphingoid
bases were identified and sphingosine was the major contributor. The rest of the
sphingoid bases have 16–19 carbon atoms and 0–2 double bonds. Fourteen
fatty acids have been detected with 16–26 carbon atoms and 0–2 double bonds.
Palmitic acid was the most abundant fatty acid. Thirty-eight ceramides were
identified. Differences between some sphingoid bases, fatty acids and ceramides
were detected which have been attributed to both diet and other environmental
and genetic factors.
Fast atom bombardment methods have also been applied to analyse the lipid
composition forming the annulus around intrinsic membrane proteins and
protein–lipid anchors. An example of complete resolution of the structure is the
mass spectrometric analysis of human acetylcholinesterase of the red blood cell
membrane (Roberts et al., 1988). The glycoinositol phospholipid membrane
anchor was found to consist of a glycan linked through a glucosamine residue
to an inositol phospholipid. Analysis by fast atom bombardment mass
spectrometry with negative ion monitoring and by the complementary tech-
nique of collision-induced dissociation revealed molecular and daughter ions
that indicated a plasmanylinositol with a palmitoyl group on an inositol
hydroxyl.
The introduction of tandem MS–MS has also considerably alleviated these
analytical tasks. The procedure does not require the complete separation of
every lipid class forming the membrane lipid matrix. The first stage in the mass
spectrometer (quadrupole filter, Q1) is used to trace the parent ion after a
characteristic neutral loss detected in the second stage of the equipment
(usually another quadrupole filter, Q2). For example, the polar head group of
phosphatidylcholine (PC) can be monitored by the loss of the 183 Da
phosphorylcholine group by scanning each of the parent ions (mass-to-charge
ratio, m/z = M) in Q1 which have generated an ion m/z ratio of M–183 in Q2.
The fragmentation between Q1 and Q2 in a gas collision chamber serves to
produce the product ions identified in the second mass analyser. A scan of the
product ions allows a direct assignment of the fatty acids associated with a
selected lipid. The whole procedure can be applied in the ‘selected reaction
monitoring’ (SRM) mode with a maximum selectivity and sensitivity for each
of the lipid classes. Due to the experimental simplicity of the method, with no
pre-purification of the lipid extract required, the major advantages are (1) that
a definitive assessment of the fatty acid composition, which hitherto could not
be obtained from the overall carbon number and double bond content, can be
obtained, (2) minor molecular species among a few dominant species can be
identified, and (3) accurate quantification can be achieved, which is not
possible if the sample has to be pre-purified through multiple separation steps
with varying yield.
Nano-electrospray ionization tandem mass spectrometry (ESI–MS) brings
analysis of membrane lipids down to the picomole level. The soft ionization
procedure applied in ESI–MS is appropriate for small volumes of
chloroform:methanol extracts from less than 106 CHO cells (in practice down
to 103). The resolution of the molecular species of the distinct lipid classes and
for the plasmalogen analogues is obtained with negative ion mode [M–H]– for
anionic lipids and positive mode [M+H]+ for choline-containing phospholipids.
Ethanolamine phosphatides are responsive in both modes. When operated in
the single-stage MS, ESI mass spectra of lipid extracts show almost exclusively
molecular ion species without fragmentation. The assignments of the molecu-
lar ions can be obtained by recording the product ion spectra in Q2 after
selection in Q1 and activation in the collision cell. The polar head group
detected in Q2 can be lost as a charged or a neutral fragment. The proportion of
ion product (head group lost as a charged product) increases with the collision
energy, whereas neutral loss is favoured at voltages less than 30 V. In a
precursor ion scan the second mass analyser is set to transmit only ions from the
parent ion selected by Q1 (for instance Q2 is set to detect species with
m/z = 184 Da, corresponding to H + plus the phosphorylcholine of
phosphatidylcholine and sphingomyelin), whereas in the neutral loss scan the
analyser Q2 is scanned in a synchronized fashion with Q1 but with an offset to
lower m/z values equivalent to the neutral loss of interest (for instance, –183 Da
to monitor phosphatidylcholine).
Using the higher mass resolution of tandem Fourier transform ion cyclotron
resonance mass spectrometry (Fridriksson et al., 1999), over 90 different
phospholipid species were resolved in detergent-resistant membranes (DRM)
prepared from mast cells. Coupling between the high-affinity receptor for
immunoglobulin E (IgE), located on the external surface of the cell, and
tyrosine kinase, Lyn of the Src family, found exclusively on the cytoplasmic
side of the membrane, associated with the DRM, is accompanied by a shift to
polyunsaturated molecular species of choline- and amino-phosphatides found
in this sub-fraction. Taking into account the critical role of fatty acids in the
association of phospholipids with cholesterol, a transition of the liquid-ordered
to liquid-disordered fluid phase was inferred from this compositional shift (Ge
et al., 1999).
Application of these powerful analytical methods is essential if the signifi-

cance of changes in minor species is to be fully appreciated in a particular cell
type or during cell activation. The dominant molecular species may be consid-
ered to form the basic building blocks of the matrix. In practice, the high
sensitivity of the method allows analysis of small samples, which greatly
enhances the utility of the method. Thus analysis of membrane fractions from
cultured cells can be accomplished with a relatively high degree of repro-
ducibility (Brugger et al., 1997).
In addition to monitoring changes in molecular species by direct analysis,
methods have also been developed to monitor changes indirectly. One such
method is to biosynthetically incorporate probes into membrane lipids in living
cells and to follow changes in the probes in situ. An example of this technique
is the feeding of paranaric acid to cells in tissue culture; this then becomes
incorporated into the constituent membrane lipids as a substitute for normal
fatty acids (Ritov et al., 1996; Tyurina et al., 1997). Separation and analysis by
conventional chromatography is simplified by following fluorescent ana-
logues. Changes in membrane phospholipid composition resulting from
enhanced turnover, oxidation, etc., can be detected at levels not accessible
easily by direct chemical analysis (Shvedova et al., 2001, 2002).
2. Lipid composition of cell membranes

The composition of the lipid component of biological membranes must be
considered in the context of the particular molecular species of lipid that are
present and the relative proportions in which they occur. The lipids are
characterized by classes with different basic structures, such as glycerolipids
and sphingolipids, and the substituents (both polar and non-polar) they possess
– for example, the length, unsaturation and presence of branched groups on
associated hydrocarbon chains, and the position and mode of attachment to the
glycerol backbone. Sterols are an interesting case in that the type of sterol
observed in membranes is characteristic for animals (cholesterol in higher
organisms, ergosterol and 7-dehydrocholesterol in yeasts) or plants
(camposterol, sitosterol, stigmasterol). The sterols have structures that are
relatively similar, indicating that the functions performed by sterols in biologi-
cal membranes may be highly conserved. What these functions are remains
unclear but a role in lateral domain creation is strongly inferred by recent
evidence. Specific functions identified with other membrane lipids are few.
Specific glycolipids are known to be associated with the ABO blood group
substances, and phosphorylated molecular species of glycerophosphoinositides
are involved in transmembrane signalling cascades. Finally, the
glycerophospholipids act as a reservoir of the metabolic precursor of eicosanoid
biosynthesis, arachidonic acid.
Each morphologically distinct membrane in cells possesses a characteristic
lipid composition. The constituent lipids of each membrane appear to be

preserved within relatively narrow limits by biochemical mechanisms that are
as yet not completely understood. The processes of membrane lipid turnover
and homeostasis are discussed below but some basic questions relating to the
purpose of membrane lipid diversity arise. These are:
(1) Why do (most) cells in higher organisms maintain such a complex lipid
matrix despite being protected from variations in their environment by
the homeostatic regulation of surrounding fluids (e.g. composition of
the extravascular fluid and of serum)?
(2) Are the narrow ranges (homeostatic regulation) of any of the membrane
lipids critical to the cell? Are there cell functions specifically dependent
on the presence of a particular native lipid which, if substituted, results
in perturbation of that function?
(3) Is lipid complexity predicated by the complex assortment of membrane
proteins, or do lipids have their own rules for assembling which govern,
in turn, the way proteins are ‘passively’ inserted into the matrix?
Lipid homeostasis for a complex mixture represents a cost in terms of
metabolic expense and gene diversity, which encodes all enzymes responsible
for catalysing the biosynthetic and degradation pathways. Multiple cross-
regulation would be expected to achieve the characteristic composition of
membranes. The maintenance of a complex lipid composition requires regu-
lated pathways to repair the alterations (literally ‘homeostasis’) induced by the
‘activated enzymes’ perturbing the cell membranes.
As we have seen, methods of lipid analysis have currently achieved a level
of sophistication such that a large number of molecular species can be identi-
fied accurately and quantitatively in a short period of time. Methods of
establishing the physical consequences of the diversity of lipid composition
have not kept pace with these developments. The reason for this is mainly that
methods employed to characterize the conformation and structure of the
membrane lipid matrix are averaging techniques, such as spectroscopy, that are
unable to distinguish subtle differences in local environment created by
domains dependent on lipid complexity. As a result much information is
indirect and based on the construction and examination of models to simulate
the lipid matrix of biological membranes. Frequently such models diverge
considerably from the real world because the low energy barriers that separate
conformational states in the complex mixtures which represent biological
membranes are less likely to occur in defined lipid mixtures. Furthermore,
some molecular species may be present in relatively minor proportions and
would not be expected to greatly influence the phase structure of the membrane
lipid matrix. Others, for example cholesterol and sphingomyelin in the plasma
membrane, may dominate the lipid composition of the membrane and exert a
major impact on the structure and properties of the membrane. Clearly,
characterizing the physical consequences and influence of individual molecu-

lar species of membrane lipid on the structure and properties of the bilayer
matrix remains a considerable challenge.
3. Lipid domains in membranes

The segregation of the lipids of cell membranes into separate domains is now
known to underlie membrane functions like signal transduction, fusion, etc.
This realization has come about by the characterization of so-called liquid-
ordered phases formed between choline phosphatides and cholesterol. These
phases are created by specific interactions between molecules which segregate
from domains of fluid lipids to form a ‘platform’ or ‘raft’ which attracts lipid-
anchored membrane proteins. The segregation of these proteins appears to be
required for their function.
Cell membranes can be fractionated according to their solubility in mild
detergents. Kirkpatrick and colleagues (Kirkpatrick et al., 1974) were among
the first to report selective solubilization of membrane lipids by treatment with
Triton X-100. They showed that membrane proteins and lipids were solubilized
in human erythrocyte membranes with increasing concentrations of detergent
up to 5 mM. Under these conditions the insoluble fractions were found to
preferentially retain sphingomyelin and cholesterol. Since then many studies
have shown that treatment of a range of cell membranes with mild detergents
at low temperatures (0–4ºC) results in selective solubilization of the structure
(Ahmed et al., 1997; Janes et al., 2000; Li et al., 2001; Schroeder et al., 1998;
Wang & Silvius, 2001; Xu & London, 2000; Xu et al., 2001). The insoluble
fraction floats at low density upon gradient centrifugation and the fraction can
be easily harvested (Brown & Rose, 1992; Parton et al., 2002). Residual
detergent is then removed and the resulting membrane can be further purified
or subfractionated.
The lipid composition of detergent-resistant membrane fractions isolated
from a particular membrane, especially the proportion of cholesterol, depends
to a certain extent on the type of detergent used. For example, membrane
fractions that are insoluble in 1% (w/v) Lubrol WX are found to be soluble in
Triton X-100 at the same concentration (Roper et al., 2000). Analysis of the
lipid composition of detergent-resistant membrane fractions from human
macrophages and fibroblasts showed that Lubrol WX-insoluble membranes
had more unsaturated phosphatidylcholine, and a lower ratio of cholesterol to
phosphatidylcholine, than fractions insoluble in Triton X-100 (Drobnik et al.,
2002).
Some concern has been expressed as to whether the low-temperature condi-
tions (0–4ºC) used to prepare DRMs by solubility in non-ionic detergents
converts fluid lipid domains into liquid-ordered phases that resist solubiliza-
tion by detergent. This question has been addressed by preparing DRM
Table 9.1 Molar proportions of the major lipids of human, sheep and goat red blood cell
ghost membranes and detergent-resistant membranes (DRM) derived from them
Molar proportion (%)
Human Human Sheep Sheep Goat Goat

Lipid ghosts DRM ghosts DRM ghosts DRM
SM 12.1 44.5 41.6 43.6 25.9 32.3

PC 16.7 4.6 0.1 0.3 2.1 4.5
PS/PI 6.4 8.7 5.1 6.5 14.0 15.6
PE 10.8 2.2 14.0 4.2 9.7 16.6
Cholesterol 54.0 39.9 39.2 45.4 48.4 31.1
SM, sphingomyelin; PC, phosphatidylcholine; PS, phosphatidylserine; PI, phosphatidylinositol; PE,

phosphatidylethanolamine.
fractions at different temperatures and using different types of detergent. A

recent study has been reported on preparations of detergent-resistant
membrane fractions from microvilli membranes from pig small intestine
(Braccia et al., 2003). The lipids present in DRMs prepared using Triton X-100
at low temperatures were compared with those in DRMs isolated from the same
membrane using Brij 98 at physiological temperatures. The lipid profiles of the
DRMs were closely alike suggesting that the overall solubilizing properties of
the two detergents were similar but the insoluble fraction obtained with Brij 98
was more enriched with phospholipids. These results indicate that rafts are not
an artefact arising from thermotropic phase behaviour of saturated molecular
species of lipid in the temperature range 0–37ºC, and are consistent with the
results of other studies of DRM fractions isolated with Brij 98 (Drobnik et al.,
2002). This infers that insolubility in non-ionic detergent depends on a phase
structure formed by the association of particular lipid constituents that is stable
over the temperature range 0–37ºC, rather than thermotropic phases that are
generated within this temperature range.
Analysis of the lipid composition of detergent-resistant membrane fractions
prepared by treatment of human, sheep and goat (ruminant) erythrocyte
membranes with Triton X-100 at low temperature has been used to determine
the types of lipids that are solubilized and those that remain in the insoluble
membrane fraction (Koumanov et al., 2005). The major lipid classes in each of
the DRM fractions and the parent membranes from which they were derived are
presented in Table 9.1.
The relative proportion of major lipids did not differ greatly from one
Table 9.2 Composition of fatty acids amide linked to the sphingosine of sphingomyelin
isolated from human and goat erythrocyte ghosts and corresponding detergent-resistant
membrane (DRM) fractions
Composition (% of total fatty acids)a
Fatty acid Human ghosts Human DRM Goat ghosts Goat DRM
C16:0 41.45 ± 5.68 8.63 ± 0.74 27.94 ± 1.35 44.26 ± 5.21

C18:0 6.25 ± 0.74 6.36 ± 0.66 3.27 ± 0.42 3.44 ± 0.29
C18:1 1.15 ± 0.21 1.49 ± 0.21 7.80 ± 0.62 0.86 ± 0.09
C18:2 0.10 ± 0.01 0.18 ± 0.03 0.79 ± 0.09 0.15 ± 0.01
C20:0 0.70 ± 0.00 2.22 ± 0.03 0.07 ± 0.01 0.06 ± 0.00
C22:0 5.78 ± 0.48 14.72 ± 2.02 0.76 ± 0.09 0.65 ± 0.05
C24:0 20.02 ± 2.35 62.10 ± 5.69 1.59 ± 0.10 1.92 ± 0.18
C24:1 24.02 ± 1.87 1.47 ± 0.19 55.58 ± 4.85 47.07 ± 6.25
C26:0 0.33 ± 0.04 2.78 ± 0.21 0.11 ± 0.02 0.09 ± 0.01
C26:1 0.19 ± 0.01 0.05 ± 0.00 2.10 ± 0.18 1.52 ± 0.28
Total saturated 74.54 96.81 33.74 50.41
Total
monounsaturated 25.36 3.01 65.48 49.44
Total
polyunsaturated 0.10 0.18 0.79 0.15
a
Mean ± standard deviation of the mean of 3 replicates.
preparation to another, suggesting that detergent treatment leads to reproduc-

ible DRM fractions, at least in respect of their lipid composition. The most
notable features of the lipid compositions are firstly, the relatively high
proportion of sphingomyelin in erythrocyte ghosts from goats (or other
ruminants), which results from a replacement of phosphatidylcholine by
sphingomyelin. Secondly, sphingomyelin appeared to be the most resistant of
the polar lipids to solubilization by the detergent and the most abundant lipid
found in the DRM fractions. The fact that the sphingomyelin content of human
erythrocyte ghosts was low relative to that of goat erythrocyte ghosts was
reflected in the proportion of the ghost membrane insoluble in Triton X-100:
the human DRM fraction represented only 5% of the ghost membrane, whereas
27% of the lipid of the goat erythrocyte membrane was insoluble in Triton
X-100. The ratio of sphingomyelin to cholesterol is of particular interest.
Cholesterol represents approximately half the total lipid of erythrocytes of
both species, but the proportion of cholesterol recovered in the DRM
fractions is significantly less than this. Indeed, cholesterol is present in
about the same proportion as sphingomyelin in each case. This suggests
that the amount of cholesterol that resists solubilization by detergent is
determined by the extent to which sphingomyelin remains insoluble in Triton
X-100. The ratio of sphingomyelin to cholesterol in the goat erythrocyte

membrane compared to that in the DRM fraction suggests that in this
membrane excess cholesterol is excluded from the domain that remains
insoluble in Triton X-100. The result implies that it is the presence of saturated
molecular species of sphingomyelin that determines the detergent-resistant
domain, and not the cholesterol.
To investigate the factors that govern the partitioning of sphingomyelin into
the DRM fraction of human and goat erythrocyte membranes, the molecular
species of sphingomyelin recovered in the DRM fractions were compared with
those in the membrane ghosts from which they were derived. The fatty acids in
amide ester linkage to the sphingosine isolated from human and goat erythro-
cyte ghost membranes and the corresponding DRM fractions were analysed
and the results are shown in Table 9.2. In both human and goat the saturated
molecular species of sphingomyelin dominate the DRM fractions at the ex-
pense of monoenoic fatty acids. This is achieved in the human membranes by
an approximately three-fold enrichment of C22:0 and C24:0 molecular species of
sphingomyelin and the almost complete exclusion of molecular species of
sphingomyelin associated with C24:1 fatty acid. Similarly, in the goat mem-
branes the DRM fraction is enriched in the C16:0 molecular species and contains
relatively less C18:1 and C24:1 sphingomyelin. The overall effect is that the more
saturated molecular species of sphingomyelin dominate the DRM fractions,
and these molecular species appear to preferentially interact with cholesterol in
a stoichiometry of 1:1.
While sphingomyelin and cholesterol have tended to achieve prominence
in raft lipid composition more recent studies have indicated that
glycerophospholipids and diacylglycerols are also constituents of raft frac-
tions. One method that has been used to identify such components is the
detection of lipids from radioactive precursors that are isolated in raft fractions
(Rouquette-Jazdanian et al., 2002). The method involves feeding cells with
radiolabelled glycerol, fatty acids or water-soluble polar groups and identify-
ing the complex lipids into which they are biosynthetically incorporated. Using
this approach it was shown that DRM fractions derived from the human
leukaemic T-cell line Jurkat had a considerably higher cholesterol content than
the parent membrane, and that polar lipids incorporating [3H]-labelled glycerol
were present. These glycerophospholipids included choline, ethanolamine,
serine and inositol phosphoacylglycerols with a preponderance of
phosphatidylcholine and phosphatidylserine. Incorporation of radiolabelled
fatty acid precursors into the phospholipids showed preferential labelling of
DRM lipids with saturated fatty acids such as palmitic and stearic acids rather
than oleic, linoleic and arachidonic acids. These results are consistent with
those from other studies, indicating that the DRM fractions contain predomi-
nantly saturated molecular species of membrane lipids (Bunnell et al., 2000;
Dietrich et al., 2001).
Analysis of the proteins in such fractions shows that they are predominantly
attached to the membrane by lipid anchors. Such proteins include
glycosylphosphatidylinositol (GPI)-anchored proteins originating from the
outer surface of the plasma membrane and diacylated cytoplasmic proteins
such as certain Src family kinases originating from the inner surface of the
membrane (Bagnat et al., 2000; Moffett et al., 2000). The lipid composition of
rafts, like the protein constituents, differs from the composition of the parent
membrane, providing evidence that fractionation depends on the solubility of
the lipid in detergent solution.
Differential solubility in non-ionic detergents such as Triton X-100, CHAPS
and Lubrol has been exploited to define trafficking pathways of membrane
proteins in cells (Mairhofer et al., 2002; Sakyo & Kitagawa, 2002; Slimane et
al., 2003). The implication of such studies is that membrane lipid domains are
primarily responsible for directing the traffic.
Preparations of DRM fractions insoluble in 1% Nonidet P-40 have been used
to examine signal transduction pathways in lymphocytes (Zubiaur et al., 2002).
CD38 signal transduction in T-lymphocytes was said to be due to formation
of supramolecular signalling complexes of CD3 molecules bearing immuno-
receptor tyrosine-based activation motifs together with protein tyrosine kinases.
In other studies insoluble fractions in 1% Brij 58 at 4ºC were used to demon-
strate that T-cell receptors are associated into cell-surface domains upon
receptor stimulation, providing a mechanism for coupling receptor location
with downstream signalling cascades (Montixi et al., 1998).
C. Membrane function and lipid diversity

Because the lipid composition of each membrane is maintained relatively
constant under given growth conditions it may be inferred that this is required
for the membrane to perform its particular functions. Since membrane
functions vary widely, it is perhaps not surprising that this is reflected in the
diversity of lipid molecular species present. This section examines how
membrane lipids are thought to mediate membrane functions.
1. Adaptation to environmental factors

Membrane complexity allows flexibility in physiological functions in the face
of varying environmental situations. While the detailed molecular species of
membrane lipids are complex their physical properties are invariably those of
weak surfactants with critical micelle concentrations in the nanomolar range.
Most molecular species of lipids form either a fluid bilayer or hexagonal-II
phase when dispersed in dilute salt solutions at physiological temperatures.
The exceptions are the sterols and heavily glycosylated lipids, which are
sparingly soluble and micelle-forming, respectively. The membrane lipid
A B
Figure 9.1 Freeze-fracture electron micrographs of: (A) chloroplast thylakoid membrane; (B) total
polar lipid extract of chloroplast thylakoid membrane dispersed in the same medium as in (A).
matrix is believed to be exclusively bilayer in arrangement, so that the

interaction of the non-bilayer-forming with other membrane constituents must
constrain them into a bilayer configuration. This is consistent with structures
observed when total polar lipid extracts of biological membranes are dispersed
in aqueous systems. The dispersion consists of phase-separated bilayer and
non-bilayer structures. This is illustrated in Figure 9.1, which shows freeze-
fracture electron micrographs of chloroplast thylakoid membrane and a total
polar lipid extract of the membrane dispersed in the same medium. Changes in
temperature, salinity, etc. alter the stability of the matrix and this may be
compensated to some extent by relocation of lipid and protein constituents into
domains where stability may be altered in local regions of the membranes.
Some indication of the advantages of complexity in membrane molecular
species can be appreciated when cells adapt to different growth conditions.
Adaptation of poikilothermic organisms to changes in environmental condi-
tions involves changes in the lipid molecular species of the membrane matrix.
The consequences of changes in membrane lipid molecular species are such
that the balance of lipids tending to form non-bilayer phase and those forming
a bilayer under any particular environmental condition are preserved. Since
this condition can be achieved by redistribution of existing fatty acids within
the lipid classes present, changes in the length or unsaturation of fatty acid
substituents, or a change in the spectrum of lipid classes, it is not unexpected
that all of these strategies are employed by different organisms. Furthermore,
by examining the responses of poikilothermic organisms to changes in environ-
mental factors it is possible to gain some insight into the mechanisms of lipid
homeostasis in homeothermic organisms.
2. Remodelling membrane lipids
Membrane lipid remodelling of molecular species can be achieved by a

redistribution of the fatty acyl esters within a particular class of lipid or between
different lipid classes. For example, mixtures of phospholipids with saturated
fatty acyl chains with molecular species acylated to unsaturated fatty acids
results in a phase separation of gel and fluid phases. A redistribution of the fatty
acids between the two types of lipid results in a homogeneous fluid bilayer at
physiological temperatures. Remodelling without de novo fatty acid biosynthe-
sis has been observed in the adaptation of microorganisms such as protozoa
(Dickens et al., 1982) and algae (Sato & Murata, 1980). Another strategy is to
change the length of the acyl chains, as the amphipathic balance is dependent
on the hydrophobic–hydrophilic affinities within the complex lipid molecule.
Such adaptations have been reported in yeast (Okuyama et al., 1979), where a
shift in growth temperature from 35ºC to 10ºC resulted in a change in the ratio
of C16 to C18 fatty acids from 1.4 to 4.7 on a timescale suggesting direct effects
on the synthase rather than a control exerted at the level of genetic transcrip-
tion. Branched-chain fatty acids also have a lower melting point compared to
straight-chain fatty acids and many microorganisms manipulate the branched-
chain composition of their constituent membrane lipids in response to changed
environmental conditions (Klein et al., 1999). The most common method of
adaptation, however, is an alteration in the extent of unsaturation of the fatty
acyl residues. The insertion of a single cis double bond into a stearic acid
molecule, for instance, reduces the melting point from 70ºC to 13ºC due to the
fact that the chains are unable to pack into close alignment. Changes in
desaturase activity can be bought about by direct modulation of the desaturase
enzyme pathways (Pugh & Kates, 1984), at the level of gene transcription
(Gombos et al., 1996; Wada et al., 1990), and by altered availability of
substrates (Rutter et al., 2002; Thomas et al., 1998).
While the lipid homeostatic process in homeothermic organisms is not
required to cope with the changes that poikilothermic organisms are subjected
to, it is likely that the strategies employed to preserve the structure and stability
of the matrix are similar. For example, membrane phospholipid remodelling
has been found in tumour cells to be associated with increased levels of fatty
acid synthase. The products of the synthase are long-chain saturated fatty acids
which incorporate into membrane phospholipids. The fate of these saturated
molecular species has been traced to their incorporation into detergent-resist-
ant membrane fractions, leading to the suggestion that fatty acid synthase
indirectly modulates transmembrane signal processes mediated by membrane
rafts (Swinnen et al., 2003).
To obtain an understanding of the principles underlying the regulation of
membrane composition the response of simple organisms to changes in
environmental factors has been examined in some detail. One of the favoured
organisms for such studies is the unicellular organism Acholeplasma laidlawii,

which has only one cell membrane. Studies of this organism have shown that
the response to changes in environmental conditions, including low tempera-
ture, is to tailor the lipid composition so as to preserve a constant surface charge
density, a similar phase equilibrium, and a spontaneous curvature parameter
determined by the polar lipid mixture within relatively narrow limits (Karlsson
et al., 1996, 1997; Lindblom et al., 1986). The suggestion that spontaneous
curvature of lipid structures is a functionally important membrane parameter
subject to regulation by cells and is one of the constraints controlling lipid
composition of the membrane lipid matrix was first proposed by Gruner
(1989). As already noted, interaction of non-bilayer-forming lipids with charged
lipids and intrinsic membrane proteins tends to impose a bilayer configuration
on lipids that would, when dispersed alone in aqueous systems, form a non-
bilayer structure. The proportion of non-bilayer lipid is therefore likely to be
influenced by the ratio of lipid to protein as well as to other bilayer-forming
lipids.
The acclimation of rye and oat cells to low temperatures results in changes in
plasma membrane lipid composition which correlate with ultrastructural changes
induced by freezing. The increased proportion of phospholipids relative to
cerebrosides in the plasma membranes of acclimated rye cells was rationalized
to increase tolerance to freezing by retaining more interfacial water at the
membrane surface, thereby preventing freeze-induced dehydration and the
induction of non-lamellar phase transitions (Webb & Steponkus, 1993). Uemura
& Steponkus (1997) have recently extended their studies to determine the
effect of acclimation of rye leaves on the lipid composition of the inner and
outer membranes of the chloroplast envelope. Contrary to earlier observations
on Acholeplasma laidlawii, temperature acclimation of rye plants results in an
increase in the proportion of bilayer at the expense of non-bilayer-forming lipid
in the envelope membranes of the chloroplasts. It was suggested that the
increased stability of the lamellar configuration at low temperature was a major
factor contributing to improved freezing tolerance. Nevertheless, some caution
may be warranted in reaching such a conclusion because other factors such as
the extent of unsaturation of the lipids, lipid:protein ratios, etc. also contribute
to the overall phase stability of the membrane. This is of particular relevance in
light of a report of cold-regulated genes (COR) in Arabidopsis which code for
polypeptides targeted to the chloroplasts and enhance the freezing tolerance of
protoplasts by lowering the temperature of the lamellar to non-lamellar phase
transition of membrane lipids (Artus et al., 1996).
Examination of the factors that influence the activity of enzymes responsible
for the synthesis of bilayer- and non-bilayer-forming lipids has indicated that
the level of activity necessary to preserve the phase equilibrium within optimal
limits is controlled at the level of the membrane. The enzymes thus appear to be
responsive to the physical state of the membrane matrix in which they operate.
Thus the maintenance of membrane composition within relatively narrow

limits and its adaptation to varying environmental conditions exploits bio-
chemical processes that are able to sense the phase state of the lipid bilayer
matrix and perform the necessary homeostatic adjustments. This implies that
the genetic control of the relative proportions of the different lipid classes is
rather indirect and therefore there is probably little scope for direct genetic
manipulation of the relative proportions of lipid classes to enhance the toler-
ance of organisms to low temperature.
3. Control of membrane lipid unsaturation

Poikilothermic organisms grown at low temperatures show an increased level
of unsaturation of their membrane lipids compared to those grown at higher
temperatures (Harwood et al., 1994). The effect of this adjustment on lipid
phase equilibrium, however, is often complicated by an associated change in
the ratio of lipid to protein in the membranes (Palta et al., 1993). The
connection between growth temperature and alteration of lipid unsaturation
has been examined in unicellular photosynthetic organisms by observing the
response of the organism to saturation of membrane lipids at optimal growth
temperatures using membrane-impermeable homogeneous catalysts (Quinn,
1998). In studies of Dunaliella salina, Vigh et al. (1988) hydrogenated cells
briefly so that only lipids of the plasma membrane were saturated and observed
the effects on growth rates. They found that the cells stopped growing immedi-
ately but growth resumed after about 12 hours, by which time the unsaturation
of the membrane lipids was restored to the original levels. More recently, Lehel
et al. (1993) applied the method to Synechococcus PCC6803 and showed that
hydrogenation of the plasma membrane lipids of the intact cyanobacterium
stimulated expression of the desA gene coding for proteins involved in lipid
unsaturation.
Nine mutants of the overwintering annual Arabidopsis thaliana defective in
membrane lipid unsaturation have been isolated (Browse & Somerville, 1994).
The effects of mutations of two of these genes affecting lipid unsaturation in the
chloroplast on chloroplast ultrastructure and chilling sensitivity have been
reported. The fad5 mutation involves a single nuclear mutation causing a defect
in the activity of chloroplast Δ9-desaturase responsible for insertion of a double
bond into C16:0 fatty acyl residues of monogalactosyldiacylglycerol (Kunst et
al., 1989a,b). The chloroplast membranes of this mutant are characterized by
an increased proportion of palmitic acid and a corresponding decrease in
unsaturated C16:0 fatty acids in the lipids. Another mutation, fad6, results in a
defect in the activity of Δ12-desaturase responsible for introducing double
bonds into 16-carbon and 18-carbon fatty acids of monogalactosyl- and
digalactosyldiacylglycerols (Browse et al., 1989; Hughly et al., 1989). The
chloroplast membranes of these mutants contain relatively high proportions of
Figure 9.2 Electron micrographs of freeze-fracture replicas of chloroplasts showing the effect of
temperature on particle distribution in thylakoid membranes of Arabidopsis mutants defective in the
synthesis of unsaturated membrane lipids. (A) Chloroplasts from wild-type Arabidopsis and (B)
chloroplasts from Arabidopsis with the fad6 mutation, both thermally quenched from 20ºC. (C,D)
chloroplasts from wild-type Arabidopsis equilibrated and thermally quenched from 4ºC. The scale bar
corresponds to 100 nm.
monoenoic fatty acids, as further desaturation of these molecular species is

prevented by the defective allele. Both these mutants become sensitive to
chilling and are no longer able to acclimate for frost tolerance (Hughly &
Somerville, 1992). Another mutation, fad2, causing a defect in Δ12-desaturase
affecting the insertion of a double bond into oleoyl residues of extra-chloroplastic
membranes, also produces plants with a phenotype that does not allow them to
survive long exposure to low, non-freezing, temperatures (Miguel et al., 1993).
Examination of the ultrastructure of chloroplast membranes of the fad5 and
fad6 mutants showed that the protein complexes of the photosynthetic
membrane tended to dissociate and become aligned into characteristic arrays

within the membrane (Tsvetkova et al., 1994, 1995). The conditions of array
formation are illustrated in Figure 9.2. Wild-type Arabidopsis shows a random
orientation of membrane-associated particles in exoplasmic and protoplasmic
fracture planes in both appressed and non-appressed regions of the thylakoid
membranes when thermally quenched from around the growth temperature of
20ºC (Figure 9.3A). Particles form into characteristic arrays in the appressed
region of the membrane when the chloroplasts are cooled to 4ºC (Figures
9.3C,D). This effect is also seen at 20ºC in the mutant strains (Figure 9.3B),
indicating that the unsaturated lipid is responsible for preventing alignment of
particles at the growth temperature. The alignment of particles is likely to be
associated with low-temperature photoinhibition typical of photosynthetic
membranes containing more saturated lipids (Tasaka et al., 1996). Constraint
of photosystem-II complexes in these arrays may inhibit processing of the D1
protein (Oquist, 1995). These observations indicate that the molecular species
of lipid plays a role in the organization of intrinsic membrane proteins into
functional oligomeric complexes.
Several studies from Murata’s laboratory (Ariizumi et al., 2002; Murata &
Wada, 1995; Murata et al., 1992; Nishida et al., 1996) reported the production
of transgenic plants with altered levels of polyunsaturated membrane lipids and
sensitivity to exposure to chilling temperatures. The results of these experi-
ments are illustrated in Figure 9.3. Tobacco plants, which are considered to
have a chilling sensitivity intermediate between those of squash (chilling
sensitive) and Arabidopsis (chilling resistant), when transformed with acyl-
ACP:glycerol-3-phosphate acyltransferase from either squash or Arabidopsis
became more chilling sensitive or resistant, respectively. This was correlated in
each case with decreased or increased levels of polyunsaturated membrane
lipids. Cold tolerance in tobacco is also enhanced by expression of a gene for
chloroplast ω3-desaturase from Arabidopsis (Kodama et al., 1994, 1995) or Δ9-
desaturase from Anacystis nidulans (a chilling-sensitive organism)
(Ishizaki-Nishizawa et al., 1996). Transformation of Arabidopsis with
Escherichia coli acyl-ACP:glycerol-3-phosphate acyltransferase, which is
unable to discriminate between C16:0-ACP and C18:0-ACP, has been reported by
Wolter et al. (1992) to increase the proportion of saturated molecular species of
phosphatidylglycerol and confer chilling sensitivity on the transgenic
plants.
The molecular species of the two acidic lipids of chloroplast thylakoid
membranes, phosphatidylglycerol and sulphoquinovosyldiacylglycerol, tend
to be more saturated than other lipid classes. In the case of
sulphoquinovosyldiacylglycerol no correlation was found to exist between the
extent of unsaturation of this lipid and the chilling sensitivity of the plant from
which it originated. By contrast, several studies reported a correlation between
the contents of the C16:0/C16:0 plus C16:0/C16:1(Δ3-trans) molecular species of
Figure 9.3 Genetic manipulation of membrane lipid unsaturation of Nicotiana tabacum and Arabidopsis
and the consequence on exposure of the transgenic plants to chilling conditions.
phosphatidylglycerols and chilling sensitivity (Roughan, 1985). The liquid-

crystalline to gel-phase transition temperature of these molecular species is
above chilling temperature and the rationale explaining the lesion was that the
presence of gel-phase lipid resulted in an increased solute permeability across
the membrane. Subsequently, Wu & Browse (1995) examined the chilling
sensitivity of the fab1 mutant of Arabidopsis, which is defective in acyl-
ACP:glycerol-3-phosphate acyltransferase activity, and found that although
the level of saturated molecular species of phosphatidylglycerols represented
43% of the total phospholipid and exceeded that found in many chilling-
sensitive plants the mutants were not chilling sensitive. However, the long-term
survival of the mutant strain at 2ºC was affected, suggesting that the level of
unsaturation of the phosphatidylglycerols may only be one of a number of
factors controlling chilling sensitivity in this plant.
D. Biochemical mechanisms of lipid homeostasis

Complex lipids found in cell membranes are frequently assembled with ester
bonds linking the individual components. All these bonds are subject to
enzymic hydrolysis and the regulation of these enzymes underlies the
homeostatic regulation of membrane lipid composition. Furthermore, the
specific activation of particular hydrolytic enzymes is a common strategy for
the conduct of a number of membrane functions, including signal transduction.
This section will review the properties of some of these hydrolases and factors
that are responsible for their regulation.
1. Phospholipases A2 (PLA2)
Hydrolases that attack the acyl ester bonds linking fatty acids to the sn-1 and sn-
2 positions of the glycerol backbone of phospholipid molecules are categorized
by their positional specificity as A1 and A2, respectively. A number of early
studies showed evidence of phospholipase A activity hydrolysing fatty acyl
residues at the C-2 position of the glycerol moiety of phosphoacylglycerols
(PLA2) by enzymes that were bound to cell membranes and acting on endog-
enous substrates. The enzyme activity was shown to change the level of cAMP
in fat cells (de Cingolani et al., 1972) and to be influenced by the insulin:glucagon
ratio (Polonovski et al., 1974; Wolf et al., 1977). The possibility that the
enzyme retro-regulates membrane fluidity was also implied in these early
reports (Petkova et al., 1987). The notion that these enzymes are involved in
signal transduction by mediation of metabolic turnover of membrane
phospholipids has been less favoured by the discovery of cytosolic species of
PLA2 (cPLA2), whose action appears more appropriate to this function (Clark
et al., 1991). It was also realized that phospholipids of the membrane matrix
were relatively resistant to enzyme hydrolysis. A number of mechanisms are
responsible for the protection of phospholipids against PLA2 attack. These
include:
(1) Limitation of the penetration of the enzyme into the substrate due to the
tight packing of the lipid molecules.
(2) Restricted access of the enzyme to its preferred substrate. For instance,
exogenous secretory type II PLA2 outside cells cannot gain access to
phosphatidylethanolamine substrate located in the inner leaflet of plasma
membranes.
(3) Dilution of susceptible substrates within non-substrate membrane lipids
such as sphingolipids and cholesterol.
These limitations serve to restrict enzyme activity to a relatively small propor-
tion of the lipids forming the membrane lipid matrix, leading to the conclusion
that the hydrolysis of such a small number of molecules would be unlikely to
result in any physiological consequences.
Detection of phospholipid turnover resulting from the action of endogenous
PLA2 acting on the membrane lipid matrix requires pre-labelling of
phospholipids by biosynthetic incorporation of radiolabelled precursors, either
32
P- or 14C-labelled fatty acids, administered beforehand to the organism.
Furthermore, the release of lysoderivatives and fatty acids can only be related
to PLA2 under particular experimental conditions in which subsequent enzyme
action on the products (lysophospholipases, transacylase, oxygenation of
PUFA, etc.) are negligible. These conditions are not easily achieved and assays
of enzyme activity against exogenous substrate, i.e. dispersions of radiolabelled
or fluorescent analogues of phospholipids, are assumed to reflect the mem-
brane-bound enzyme activities. Despite such problems it has been shown that
activities against exogenous and endogenous substrates vary according to the
fluidity of the membrane matrix (Momchilova et al., 1985, 1986).
The demonstration that cPLA2 is the particular phospholipase involved in the
release of arachidonic acid from the sn-2 position of phosphatidylcholine under
conditions where the cascade of reactions leading to eicosanoid biosynthesis is
triggered has been fundamental to understanding the role of PLA2 in this
process (Clark et al., 1991). In this work it was shown that cloning and
expression of a cDNA encoding a high molecular mass (85.2 kDa) cPLA2,
assigned to type IV PLA2, has no detectable sequence homology with the non-
pancreatic secreted forms of PLA2 (type II) described previously. Whereas
type II PLA2 represented non-specific activities observed earlier, cPLA2 selec-
tively cleaved arachidonic acid from microsomes of intact biological
membranes. It was demonstrated that cPLA2 translocated to membrane vesicles
in response to physiologically relevant changes in free calcium concentration.
By contrast, secretory type II PLA2 is known to be calcium-dependent with an
optimum concentration greater than 1 mM consistent with its preferential
extracellular activity. Moreover, an amino-terminal 140 amino acid fragment
was identified in cPLA2 which translocated to natural membrane vesicles in a
Ca2+-dependent fashion. Maximal activation required phosphorylation of Ser-
505 via the MAP kinase (mitogen-activated protein kinase) pathway (Lin et al.,
1993). Treatment of cells with agents that stimulated the release of arachidonic
acid caused increased serine phosphorylation and activation. The site of cPLA2
phosphorylation by MAP kinase, Ser-505, was identical to the major site of
cPLA2 phosphorylation observed in phorbol ester-treated cells. Replacement
of Ser-505 with Ala resulted in a mutant cPLA2 that caused little or no enhanced
agonist-stimulated arachidonate release from intact cells.
The delineation of the two functionally distinct domains of cPLA 2 was
approached experimentally by Nalefski et al. (1994). Isolation and sequence
analysis of cPLA2 cDNA clones from four different species revealed several
highly conserved regions. The N-terminal conserved region is homologous to
that of a number of other Ca2+-dependent lipid-binding proteins. The first 178
residues of cPLA2, containing the homologous Ca2+-dependent lipid-binding
(CaLB) motif, and another recombinant protein containing the cPLA2(1–178)
fragment placed at the C- terminus of the maltose-binding protein (MBP-
CaLB) associated with membranes in a Ca2+-dependent manner. Both cPLA2
and MBP-CaLB also bind to synthetic liposomes at physiological Ca2+ concen-
trations, demonstrating that accessory proteins are not required for this process.
In contrast, ΔC2, a truncated cPLA2 lacking the CaLB domain, failed to
associate with membranes and failed to hydrolyse liposomal substrates but
hydrolysed monomeric 1-palmitoyl-2-lysophosphatidylcholine at an identical

rate to that of cPLA2 in a Ca2+-independent fashion. On the basis of the
requirement for calcium, the Ca2+-independent catalytic domain and the regu-
latory CaLB domain were clearly distinguished. In Chinese hamster ovary
(CHO) cells the binding of cPLA2 to membrane can be modulated by the ratio
of sphingomyelin to cholesterol (Klapisz et al., 2000), which in turn modulates
membrane susceptibility. Cholesterol depletion of CHO-2B cells by treatment
with methyl-β-cyclodextrin resulted in inhibition of the release of arachidonic
acid whereas the restoration of cholesterol by incubation with cholesterol-
loaded cyclodextrin relieved the inhibition. Conversion of CHO-2B cellular
sphingomyelin to ceramides by Staphylococcus aureus sphingomyelinase
enhanced endogenous cPLA2 activation by calcium ionophore and epinephrine
as well as uptake by cells of C2-ceramide and C6-ceramide analogues. These
results were confirmed in vitro with purified human recombinant cPLA2 acting
on a model phospholipid substrate. The enzyme activity was inhibited by
sphingomyelin but reactivated by ceramides as well as by cholesterol added to
glycerophospholipid liposomal substrates containing sphingomyelin.
These results regarding cPLA2 are in agreement with earlier observations of
secretory type II PLA 2 (Koumanov et al., 1998). Secretory type II
phospholipase A2 (sPLA2) is inhibited by sphingomyelin. Cholesterol, either
mixed with the model glycerophospholipid substrate or added to the assay
medium in separate liposomes, effectively counteracts this inhibition. The
inhibition of fatty acid release, assayed by quantitative gas chromatography–
MS, is observed when sphingomyelin is added to erythrocyte membranes as
substrate. Hydrolysis of sphingomyelin by bacterial sphingomyelinase sup-
presses its inhibitory potency. The addition of cholesterol to sphingomyelin
liposomes with a 1:1 stoichiometry completely relieves the inhibition of sPLA2.
An enzyme binding assay showed that sPLA2 binds with relatively high affinity
to the sphingomyelin, after either phase segregation at the assay temperature or
on the pure sphingomyelin liposomes added to the incubation medium. Choles-
terol was shown to suppress the binding of the enzyme to the sphingomyelin
interface. Interestingly, it turns out that the specificity for the release of
polyunsaturated fatty acids, mostly C20:4n-6 for cPLA2, is a property that can
also be acquired by other types of PLA2 when the ratio of sphingomyelin to
cholesterol is manipulated. When the ratio is decreased from 10 to 1 in the lipid
mixture serving as the substrate the release by sPLA2 of C20:4n-6 relative to
C18:1n-9 increases from 1.5 to 2.06. Such evidence serves to exemplify how the
manner of presentation of substrate to the enzymes is able to modulate their
hydrolytic activity.
2. Acyltransferases
The activity of acyltransferase able to reverse the action of PLA2 was initially
described in membranes of the endoplasmic reticulum (Magargal et al., 1978).

The subcellular distribution of oleoyl-CoA:1-acyl-sn-glycero-3-phosphocholine
acyltransferase (EC 2.3.1.23) has been examined in cultured swine aorta
endothelial cells and smooth muscle cells. The distribution of acyltransferase
activity was found to be similar to that of enzymes of the endoplasmic
reticulum. Treatment of microsomal membranes with digitonin caused a
specific increase in the density of plasma membranes, allowing their conven-
ient separation from endoplasmic reticulum vesicles. The plasma membranes
were shown to be free of acyltransferase activity. In contrast to the activity in
prokaryotes such as Escherichia coli, where modulation occurs by the lipid
environment of the substrate (Scheideler & Bell, 1986), the only phospholipid-
regulating activity in eukaryote cells was sphingomyelin (Momchilova et al.,
1999).
The influence of membrane lipids on arachidonoyl-CoA:lyso-
phosphatidylcholine acyltransferase has been investigated in microsomal
membranes from control and ras-transformed NIH 3T3 fibroblasts. Of all the
phospholipids examined only sphingomyelin induced activation of
acyltransferase in membranes from ras-transformed cells. No effect on the
acyltransferase was observed in microsomal membranes from control
fibroblasts. Diacylglycerol was found to inhibit acyltransferase in both cell
lines, whereas ceramide accumulation induced inhibition only in membranes
from the transformed cells.
These results appeared to contradict previous experiments where the influ-
ence of different phospholipases C on acyltransferase activity of membranes of
the endoplasmic reticulum was tested (Sonoki & Ikezawa, 1976). Three kinds
of phospholipase C (EC 3.1.4.3) were used to selectively hydrolyse
phospholipids in rat liver microsomes, and their effects on the acyl-
CoA:glycerophosphate and acyl-CoA:lysophospholipids acyltransferase
systems were examined. The glycerophosphate acyltransferase (EC 2.3.1.15)
system was rapidly inactivated by treatment with phospholipase C of Pseu-
domonas aureofaciens or Bacillus cereus and the loss of activity paralleled the
degradation of phosphatidylcholine and phosphatidylethanolamine. The 1-
acylglycerylphosphorylcholine acyltransferase (EC 2.3.1.23) system was only
partially inactivated under the same conditions, whereas the 1-
acylglycerophosphate acyltransferase (EC 2.3.1.51) system retained most of its
activity even when more than 95% of phosphatidylcholine and
phosphatidylethanolamine had been hydrolysed. The results demonstrate the
heterogeneity of acyltransferase systems with respect to their dependence on
the intact membrane phospholipids. Hydrolysis of more than 80% of
phosphatidylinositol by phosphoinositidase of B. cereus did not significantly
affect these acyltransferase systems. These results suggest that 1-
acylglycerophosphate and 1-acylglycerylphosphorylcholine acyltransferase
systems do not require specific phospholipids such as phosphatidylcholine,
phosphatidylethanolamine or phosphatidylinositol for their catalytic activities,

but that the integrity of these phospholipids is necessary for the proper
functioning and stability of the enzymes.
A purification scheme has now been proposed for the enzyme (Kerkhoff et
al., 2000). The defined experimental conditions that allowed the extraction of
the integral membrane protein lysophospholipid:acyl-CoA acyltransferase
[EC 2.3.1.23] from membranes while maintaining the full enzyme activity,
using the non-ionic detergent n-octyl glucopyranoside (OGP) and solutions of
high ionic strength, have been reported. Detailed lipid analysis of the different
protein–lipid–detergent mixed micelles showed that the protein–lipid–OGP
mixed micelles were relatively enriched with sphingomyelin compared to
protein–lipid–CHAPS mixed micelles, indicating that the differences in the
solubilization efficiency may be due to the ability of OGP to extract more
sphingomyelin from membranes.
The activity of acyltransferase in the plasma membrane of liver cells has
been suggested to play a major role in the secretion of bile lecithins (Colard et
al., 1980). Phospholipid acyltransferase activities of plasma membranes have
been investigated with various acyl-CoA thioesters (palmitoyl, stearoyl, oleoyl,
linoleoyl and arachidonoyl) with and without added lysoderivatives. Different
patterns of incorporation were observed for each acyl-CoA into endogenous
phosphatidylcholine and phosphatidylethanolamine. The turnover rates calcu-
lated with tracer amounts (10 μM) of acyl-CoA thioesters were five times faster
for the polyunsaturated than for the saturated acyl molecular species of
phosphatidylethanolamine and phosphatidylcholine. Arachidonoyl-CoA was
the most efficient acyl donor at low concentrations and maximal turnover rate
was observed at about 25 μM. No saturation was observed at concentrations up
to 10 μM linoleoyl-CoA. Linoleoyl-CoA transacylase acylated the lyso com-
pounds in the following order: lysophosphatidylcholine > lysophosphatidyl-
serine = lysophosphatidylinositol. Lysophosphatidylethanolamine was
found to inhibit linoleate incorporation into the phosphatidylethanolamine.
No satisfactory explanation was given for this effect so it remains uncertain
as to whether it is connected with an interference of translocators for
lysoderivatives and phosphatidylethanolamine. Linoleoyl-CoA transacylation
was not affected by the fatty acyl moiety at the 1-position of the lysophospha-
tidylcholine.
3. Acylation/deacylation cycle
Few details have been published on the relative activities of PLA2 and
acyltransferase in relation to the homeostatic regulation of membrane lipids.
Regulation of phosphorylation/dephosphorylation was suggested to be a key
factor from studies of the incorporation of 14C-labelled palmitoyl CoA into
membrane phospholipids via the deacylation/acylation cycle conducted in rat
liver microsomes (Hutson & Higgins, 1987). This activity was reversibly
inactivated by treatment of the microsomes with 105 000 g supernatant in
either the presence or absence of ATP and MgCl2. These observations suggest
that the acylation cycle is controlled by a mechanism involving phosphoryla-
tion/dephosphorylation. Because the pool of lysolecithin in the membranes
was not altered by conditions that increase incorporation of palmitoyl CoA into
phospholipid, it is likely that the site of regulation of deacylation/acylation is
the acyltransferase reaction rather than the phospholipase. This conclusion was
reached on the basis of studies of resting tissue; however, results obtained on
activation of granulocytes provided a different perspective (Tou, 1981). The
effect of phorbol myristate acetate, known to reproduce the stimulated oxidative
activities characteristic of phagocytosis, was examined on the metabolism of
the fatty acyl groups of granulocyte phospholipids and compared with that of
phagocytic stimuli. Phorbol myristate acetate was found to stimulate the
labelling of phosphatidylethanolamine, phosphatidylcholine and
phosphatidylinositol by 1-[14C]-palmitic acid but not by U-[14C]-glycerol.
Challenge of the cells with starch granules, by contrast, selectively increased
the labelling of phosphatidylinositol by both radioactive tracers. Labelled
palmitic acid was found at both the sn-1 and sn-2 positions of phospholipids;
more radioactivity was recovered from the 2-position. The radioactivity at both
positions was enhanced in stimulated cells. These data suggest that phorbol
myristate acetate increased palmitic acid incorporation into glycero-
phospholipids by increasing the acylation of the lysoderivatives, and that starch
granules enhanced the formation of phosphatidylinositol via both de novo
synthesis and acylation of the lysoderivative. Both phorbol myristate acetate
and starch granules selectively augmented the incorporation of 1-[14C]-
arachidonic acid into phosphatidylinositol, which exhibited the highest
specific radioactivity among the phospholipids in control and in stimulated
cells. The significance of the increased incorporation of arachidonic acid into
phosphatidylinositol is thought to reflect the degradation that produces the
eicosanoid precursor.
The high specificity of the recipient lysoderivative has been examined in
platelets, another activable circulating cell (McKean & Silver, 1985). The
transfer of unsaturated fatty acyl groups to 1-alkyl-sn-glycero-3-phosphocholine
is many-fold slower than to 1-acyl-sn-glycero-3-phosphocholine. The CoA
esters of linoleate and arachidonate, two unsaturated fatty acyl groups com-
monly found in platelet phospholipids, are the preferred fatty acyl group
donors. In macrophages, three mechanisms of reacylation participating in the
remodelling of membrane phospholipids were examined. Intact alveolar
macrophages were found to acylate alkyl- and acyl-lysophospholipids with a
high selectivity for arachidonate. A specific mechanism appears responsible
for the incorporation of arachidonate into lysophospholipids in intact cells,
since the kinetic pattern for the formation of the C20:4 species was different from
that of all other species. This specificity was investigated in more detail by
examining the enzymatic acylation of 1-alkyl-2-lyso-sn-glycero-3-
phosphocholine by membrane fractions; in the absence of CoA, ATP, and
Mg2+, this lysophospholipid was still re-acylated with a high preference for
arachidonate that was independent of added free fatty acids. The addition of
CoA alone increased the rate of acylation of 1-alkyl-2-lyso-sn-glycero-3-
phosphocholine, mainly due to an increase in the formation of species other
than those containing arachidonate. When CoA, ATP, and Mg2+ were present,
the macrophage membranes catalysed the acylation of 1-alkyl-2-lyso-sn-
glycero-3-phosphocholine without preference for arachidonate. The acylation
of alkyl- and acyl-lysophospholipids by rabbit alveolar macrophages appears
to take place by three distinct mechanisms: a CoA-independent transacylation,
a CoA-dependent transacylation (reverse reaction catalysed by acyl-CoA
acyltransferase), and an acyl-CoA-dependent acylation. The CoA-independent
transacylation reaction is unique in that it is specific for arachidonate and
accounts for the selective acylation of alkyl- and acyl-lysophospholipids by
arachidonate in membrane preparations of alveolar macrophages. This reaction
appears to be extremely important in the remodelling of phospholipid molecu-
lar species and the mobilization of arachidonate into ether-linked lipids. The
transfer of arachidonate to 1-alkyl-2-lyso-sn-glycero-3-phosphocholine is also
of importance in the termination step for platelet activating factor (1-alkyl-2-
acetyl-sn-glycero-3-phosphocholine; PAF) activity, whereby 1-alkyl-2-
arachidonoyl-sn-glycerol-3-phosphocholine (a stored precursor of both PAF
and arachidonic acid metabolites) is restored.
In rabbit alveolar macrophages the transacylation system was shown to
exhibit a complex selectivity according to distinct donor and acceptor and CoA
dependency (Sugiura et al., 1987). Microsomes were found to acylate 1-[3H]-
alkyl-glycero-3-phosphocholine (1-alkyl-GPC; lyso-PAF) in the absence of
any cofactors, indicating the presence of transacylation activity. The
transacylation activity was comparable to the activity of acyl-CoA:1-alkyl-
GPC acyltransferase. The fatty acyl moieties introduced into 1-alkyl-GPC from
membrane lipids by microsomes were mainly C20:4n-6. [14C]-labelled C20:4n-6,
C20:5n-3, C22:4n-6, and C22:6n-3 were transferred efficiently from diacyl-GPC to
1-alkyl-GPC in a CoA-independent manner. The transfer rates for C16:0, C18:0,
and C18:1 from diacyl-GPC to 1-alkyl-GPC were relatively low in the presence
and absence of CoA. On the other hand, the transfer of C20:4 from diacyl-
glycero-3-phosphoethanolamine (GPE) or diacyl-glycero-3-phosphoinositol
(GPI) to 1-alkyl-GPC or 1-acyl-GPC was markedly increased by the addition of
CoA. This observation confirmed that acylation can be a specific and active
pathway for polyunsaturated fatty acids cleaved from the sn-2 position of
phospholipids serving as substrate during cell activation. In the activated
human neutrophil the circulation of arachidonate between alkyl and alkenyl
derivatives participates in the generation of the lysoderivative precursor of
PAF (Sugiura et al., 1987). These studies indicate that lyso-PAF is formed by
the transfer of arachidonate from 1-O-alkyl-2-arachidonoyl-GPC to the alkenyl-
lyso-GPE by a CoA-independent transacylase reaction.
Mass measurements revealed a rapid loss of arachidonate from 1-radyl-2-
acyl-GPE and a concomitant increase in alkenyl-lyso-GPE upon stimulation of
the neutrophils by addition of ionophore A23187. The circulation of polyun-
saturated fatty acids between alkyl-PC and lyso-PAF via alkenyl-PE was
demonstrated in a variety of tissues (Blank et al., 1995; Nixon et al., 1996).
Microsomal membranes from six different rat tissues (spleen, lung, kidney,
brain, testis, and liver) were found to possess CoA-independent transacylase
activity that could both acylate lyso-PAF (1-hexadecyl-2-lyso-sn-glycero-3-
phosphocholine) and then deacylate the 1-hexadecyl-2-acyl-sn-glycero-
3-phosphocholine product via the transacylation of added exogenous 1-alk-1′-
enyl-2-lyso-sn-glycero-3-phosphoethanolamine. Analysis of molecular species
of 1-hexadecyl-2-acyl-sn-glycero-3-phosphocholine before and after addition
of 1-alk-1′-enyl-2-lyso-sn-glycero-3-phosphoethanolamine as the acyl
acceptor demonstrated a high selectivity for polyunsaturated fatty acids (>3
double bonds/acyl group) in both the acylation and deacylation processes that
occurred in testicular microsomal membranes. The transfer of acyl groups by
the transacylase appeared to be equally effective for either arachidonic or
docosapentaenoic (n-6) fatty acids, whereas linoleic and oleic acids were not
transferred. The results indicate the PAF-related transacylase is widely
distributed among tissues and, although highly selective for polyunsaturated
acyl groups, does not discriminate selectively among the polyunsaturates.
A very particular role was ascribed to the deacylation/reacylation
of lysoderivatives on the inner monolayer of red blood cell membrane. This
was the maintenance of the highly asymmetric distribution of
phosphatidylethanolamine (PE) in ruminant membrane and the relatively low
content of phosphatidylcholine (PC) (Florin-Christensen et al., 2001). Rumi-
nant erythrocytes are remarkable for their choline-phospholipid anomalies,
namely low or absent PC along with high sphingomyelin levels. Another
anomaly in bovine erythrocytes affects aminophospholipids: PE shows almost
absolute asymmetry, with only 2% of the total present in the outer leaflet.
Furthermore, PLA2, an enzyme located on the external surface of the erythro-
cytes, shows 3-fold greater activity against PC than against PE. Because
acylation of PE is by far the most important biosynthetic event in this cell
following deacylation by PLA2 to generate lyso compounds, the selective
reacylation of lyso-PE on the inner side can account for the asymmetry of PE
distribution, and the departure of lyso-PC extracted by serum albumin for the
low content of PC in bovine erythrocyte membranes.
4. Phospholipases C
Phospholipases C (PLC) are enzymes that hydrolyse diacylglycerophospholipids

to diacylglycerols and a water-soluble phosphorylated product. Both lipophilic
and water-soluble products are involved in the transduction of membrane
signals and act as modulators of enzymes or ion channels. There are four main
families of mammalian PLC: PLCβ, PLCγ, PLCδ and PLCε, which are charac-
terized by their structural organization and mechanism of regulation (Rebecchi
& Pentyala, 2000; Rhee, 2001). All families are regulated by numerous cell
regulators but some are specific to particular phospholipases. The PLCβ family
are uniquely regulated by heterotrimeric G-proteins; PLCγ enzymes are regu-
lated by both receptor and non-receptor tyrosine kinases; PLCδ are modulated
by agents which include RhoGAP and αH; PLCε enzymes contain a GTP
exchange factor and are regulated by RAS that interacts via two RAS-binding
domains (Kelley et al., 2001; Lopez et al., 2001).
The PLCβ family are enzymes that specifically hydrolyse phosphatidylinositol
4,5-bisphosphate (PIP2) to form diacylglycerol and inositol-1,4,5-trisphosphate.
The enzymes are activated to differing extents by G-protein αq subunits and by
G-protein βγ dimers (Rhee, 2001). It is known that the C-terminal region of
PLCβ is essential for stimulation by αq but the phospholipase can still be
activated by Rho GTPases and G-protein βγ subunits that bind to different
regions of the enzyme (Illenberger et al., 2003a,b). Thus the catalytic subdomains
of PLCβ2 are all that is required for efficient stimulation by βγ dimers, whereas
additionally the putative pleckstrin homology domain is required for stimula-
tion by Rho GTPases. Amongst the Rho GTPases, Rac 1 and Rac 2 were found
to be more important stimulators than Cdc42 and all are implicated in receptor-
mediated stimulation of PLCβ2 activity. The molecular mechanisms of
stimulation by either heterotrimeric G-proteins or Rho GTPases is presently
unknown, but targeting of the enzyme to the substrate and allosteric regulation
of the enzyme are both factors (McLaughlin et al., 2002). Pleckstrin homology
domains serve to modulate the activity of their catalytic sites upon interaction
with either the substrate or G-protein activators (Philip et al., 2002).
PLCγ, like PLCβ, acts on PIP2. Two isoforms have been identified: PLCγ1,
which is widely distributed in mammalian tissues; and PLCγ2, the expression
of which is restricted mainly to haematopoieic cells (Wilde & Watson, 2001).
PLCγ1 has a function essential to growth and development, and cells deficient
in the enzyme fail to mobilize Ca2+ in response to epidermal growth factor but
proliferate normally. Animals defective in PLCγ2 survive but affected mice
show decreased B-cell number, failure of platelets to aggregate in response to
challenge with collagen, and a failure to respond to Fc receptor-mediated
processes (Wang et al., 2000). The molecular features that distinguish PLCγ
from other PLC isotypes are: (1) the presence of two Src homology domains,
located within a pleckstrin homology domain, which are responsible for
localization at the membrane substrate; and (2) an activation by phosphoryla-

tion of multiple tyrosyl residues on the enzyme, mediated by tyrosine kinases
at sites that are system-dependent.
The regulatory process controlling activation of PLCε is mediated by two
Ras/Rap-1 associating domains located at the C-terminus of the molecule and
a CDC25 homology domain near the N-terminus. These sites suggest that
modulation is achieved by binding of Ras family GTPases to these sites (Kelley
et al., 2001). The molecular basis of activation of PLCε by Ras and Rap-1
provoked by platelet-derived growth factor stimulation of cells has been
investigated by Song et al. (2002). It was found that binding platelet-derived
growth factor to its receptor site on the cell surface induces persistent activation
of PLCε. An initial rapid phase of activation of the enzyme is mediated by Ras
and the prolonged activation is sustained by Rap-1. The CDC25 homology
domain, which displays a guanine nucleotide exchange factor activity for Rap-
1, but not Ras, is a requirement for the prolonged activation of the enzyme.
5. Phospholipases D
Phospholipase D (PLD) hydrolyses phospholipids, yielding phosphatidic acid
and water-soluble bases such as choline (Singer et al., 1997). The role of
phosphatidic acid generation in membranes and its role in signalling remains
conjectural. The membrane-bound product does not occur in appreciable
proportions although locally it may be generated in amounts that may have
consequential effects on membrane structure and stability. Efforts to implicate
phospholipase D in signalling have therefore concentrated on how and where
the enzyme is activated on a subcellular level.
Two genes coding for phospholipase D activity have been identified in
mammals: PLD1 and PLD2. Both genes have been cloned and overexpressed
in different cell lines (Frohman et al., 1999). The enzyme PLD1 is located in
association with intracellular membranes and is known to be active in living
cells. PLD2, by contrast, is associated with plasma membranes but has low
resting activity that can be activated by a variety of factors including protein
kinase C, family members of ADP-ribosylation factor and Rho, and the lipid
PIP2 (Singer et al., 1997; Exton, 2002).
Targeting of PLD1 to membranes may involve palmitoylation of cysteine
residues at a domain near the C-terminus of the protein (Sugars et al., 2002).
The importance of the domain in the vicinity of the palmitoylated cysteinyl
residues suggested that not only did palmitoylation target the enzyme to the
membrane but there was specificity in the membrane site of interaction.
How phosphatidic acid may act as a signalling device at the sites of enzyme
activity has been inferred from evidence that the phospholipid binds to specific
proteins. Candidates for phosphatidic acid binding include Raf-1, a serine/
threonine kinase (Rizzo et al., 2000), cAMP-specific phosphodiesterase (Grange
et al., 2000), the mammalian target of rapamycin, mTOR (Fang et al., 2001),
protein phosphatase-1 (Jones & Hannun, 2002) and Src homology 2-domain
containing protein tyrosine phosphatase (Frank et al., 1999). Recently, a solid-
phase adsorption system has been described to identify traffic-related
phosphatidic acid binding proteins and proteins that may be implicated in PLD-
dependent pathways, and a number of specific proteins have been characterized
(Ktistakis et al., 2003).
6. Sphingomyelinases
Sphingomyelinases are enzymes that cleave the phosphorylcholine moiety
from sphingomyelin to yield lipophilic ceramide. The significance of
sphingomyelinases has been recognized by the discovery that sphingomyelin is
a substrate progenitor for initiation of signal transduction pathways (Kolesnick,
1991) in which ceramide acts as a second messenger in functions like cell
differentiation, proliferation and apoptosis (Venkataraman & Futerman, 2000;
Ohanian & Ohanian, 2001; Hannun & Obeid, 2002). Five different categories
of sphingomyelinase have been characterized on the basis of their optimal
conditions for catalytic activity (Samet & Barenholz, 1999), and a further
sphingomyelinase has been identified in bacteria (Goni & Alonso, 2002).
Evidence from model membrane studies has indicated that the action of
sphingomyelinase in generating ceramide from sphingomyelin can induce
lamellar to non-lamellar phase transitions leading to membrane fusion
(Kinnunen & Holopainen, 2000) and to lateral phase separations of ceramide-
enriched domains which exist in gel phase in fluid bilayers of phospholipid
(Huang et al., 1996). Ceramide, but not dihydroceramide, is also able to induce
the formation of pores in phospholipid bilayers, a property that has been
attributed to the extensive hydrogen-bonding capacity of ceramide (Siskind et
al., 2002).
There is evidence that the action of neutral sphingomyelinase causes domain
formation of phases enriched in ceramide and a consequent clustering of cell-
surface receptors. One example is the clustering of L-selectin in lymphocytes
(Junge et al., 1999). The action of a Zn-dependent acid sphingomyelinase in
response to interleukin-1β treatment of human fibroblasts has been found to be
associated with depletion of sphingomyelin and a corresponding increase of
ceramide in the caveolae compartment (Liu & Anderson, 1995). The interleukin-
1β treatment was also shown to be associated with decreased platelet-derived
growth factor-induced thymidine uptake by these cells, suggesting that pertur-
bation of transmembrane signalling processes had occurred. Similar studies of
the action of nerve growth factor signalling have emphasized the importance of
intact caveolae for sphingomyelinase action (Dobrowsky, 2000).
Ceramide generation and the creation of rafts has been shown to be essential
for optimal Fas signalling and induction of apoptosis in both B- and T-
lymphocytes (Kirschnek et al., 2000). On the basis of these studies a model has
been proposed for the action of ceramide in signalling processes associated
with apoptosis (Cremesti et al., 2002). Essentially, the engagement of Fas
triggers translocation of acid sphingomyelinase to the plasma membrane,
where it acts on its substrate segregated into sphingomyelin–cholesterol rafts.
The formation of ceramide induces coalescence of the rafts into large domains
in which oligomerization of downstream effectors such as FADD/MORT-1
and pro-caspase-8 can take place, leading to the Fas death signal.
E. Conclusions
The lipid matrix of biological membranes consists of a complex assortment of
polar lipids arranged in a bilayer configuration. The matrix usually consists of
a mixture of many different molecular species within each lipid class present.
The arrangement of lipids in the bilayer is asymmetric. Lateral asymmetry and
the creation of domains with defined physical properties is produced by
specific interactions between certain lipids such as the complex formed be-
tween cholesterol and the more saturated molecular species of
choline-containing phospholipids. Transbilayer asymmetry is generated by
ATP-dependent aminophospholipid translocases that sequester the
aminophospholipids on the cytoplasmic membrane leaflet.
The precise molecular species of lipid and the proportions in which they are
present differ from one membrane to another but the identity is preserved
within relatively narrow limits. The biochemical mechanisms responsible for
maintaining membrane homeostasis are not fully understood. Hydrolytic en-
zymes specific for particular ester bonds in complex lipids are reasonably well
characterized and appear to be involved in regulating the metabolic turnover of
the lipids. The regulation of such enzymes has been shown to represent the
molecular mechanism of a range of membrane functions, including transmem-
brane signal transduction.
References
Ahmed, S.N., Brown, D.A. & London, E. (1997) Biochemistry 36, 10944.
Anderson, R.G. & Jacobson, K. (2002) Science 296, 1821.
Ariizumi, T., Kishitani, S., Inatsugi, R., Nishida, I., Murata, N. & Toriyama, K. (2002) Plant
Cell Physiol. 43, 751.
Artus, N.N., Uemura, M., Steponkus, P.L., Gilmour, S.J., Lin, C. & Thomashow, M.F. (1996)
Proc. Natl Acad. Sci. USA 93, 13404.
Bagnat, M., Keranen, S., Shevchenko, A. & Simons, K. (2000) Proc. Natl Acad. Sci. USA 97,
3254.
Blank, M.L., Smith, Z.L., Fitzgerald, V. & Snyder, F. (1995) Biochim. Biophys. Acta 1254,
295.
Braccia, A., Villani, M., Immerdal, L., Niels-Christiansen, L.L., Nystrom, B.T., Hansen,
G.H. & Danielsen, E.M. (2003) J. Biol. Chem. 278, 15679.
Brown, D.A. & Rose, J.K. (1992) Cell 68, 533.

Browse, J., Kunst, L., Anderson, S., Hughly, S. & Somerville, C. (1989) Plant Physiol. 90,
522.
Browse, J. & Somerville, C. (1994) In: Arabidopsis (Meyerowitz, E. & Somerville, C, eds),
Cold Spring Harbor Press, New York, USA, p. 881.
Brugger, B., Erben, G., Sandhoff, R., Wieland, F.T. & Lehmann, W.D. (1997) Proc. Natl
Acad. Sci. USA 94, 2339.
Bunnell, S.C., Diehn, M., Yaffe, M.B., Findell, P.R., Cantley, L.C. & Berg, L.J. (2000) J.
Biol. Chem. 275, 2219.
Clark, J.D., Lin, L.L., Kriz, R.W., Ramesha, C.S., Sultzman, L.A., Lin, A.Y., Milona, N. &
Knopf, J.L. (1991) Cell 65, 1043.
Colard, O., Bard, D., Bereziat, G. & Polonovski, J. (1980) Biochim. Biophys. Acta 618, 88.
Cremesti, A.E., Goni, F.M. & Kolesnick, R. (2002) FEBS Lett. 531, 47.
de Cingolani, G.E., van den Bosch, H. & van Deenen, L.L. (1972) Biochim. Biophys. Acta
260, 387.
Dickens, B.F., Ramesha, C.S. & Thompson, G.A., Jr (1982) Anal. Biochem. 127, 37.
Dietrich, C., Bagatolli, L.A., Volovyk, Z.N., Thompson, N.L., Levi, M., Jacobson, K. &
Gratton, E. (2001) Biophys. J. 80, 1417.
Dobrowsky, R.T. (2000) Cell Signal 12, 81.
Drobnik, W., Borsukova, H., Bottcher, A., Pfeiffer, A., Liebisch, G., Schutz, G.J., Schindler,
H. & Schmitz, G. (2002) Traffic 3, 268.
Edidin, M. (2001) Trends Cell Biol. 11, 492.
Exton, J.H. (2002) FEBS Lett. 531, 58.
Fang, Y., Vilella-Bach, M., Bachmann, R., Flanigan, A. & Chen, J. (2001) Science 294, 1942.
Florin-Christensen, J., Suarez, C.E., Florin-Christensen, M., Wainszelbaum, M., Brown,
W.C., McElwain, T.F. & Palmer, G.H. (2001) Proc. Natl Acad. Sci. USA 98, 7736.
Frank, C., Keilhack, H., Opitz, F., Zschornig, O. & Bohmer, F.D. (1999) Biochemistry 38,
11993.
Fridriksson, E.K., Shipkova, P.A., Sheets, E.D., Holowka, D., Baird, B. & McLafferty, F.W.
(1999) Biochemistry 38, 8056.
Frohman, M.A., Sung, T.C. & Morris, A.J. (1999) Biochim. Biophys. Acta 1439, 175.
Gaskell, S.J. & Brooks, C.J. (1977) J. Chromatogr. 142, 469.
Ge, M., Field, K.A., Aneja, R., Holowka, D., Baird, B. & Freed, J.H. (1999) Biophys. J. 77,
925.
Gombos, Z., Wada, H., Varkonyi, Z., Los, D.A. & Murata, N. (1996) Biochim. Biophys. Acta
1299, 117.
Goni, F.M. & Alonso, A. (2002) FEBS Lett. 531, 38.
Grange, M., Sette, C., Cuomo, M., Conti, M., Lagarde, M., Prigent, A.F. & Nemoz, G. (2000)
J. Biol. Chem. 275, 33379.
Gruner, S.M. (1989) Biophys. J. 56, 1045.
Hannun, Y.A. & Obeid, L.M. (2002) J. Biol. Chem. 277, 25847.
Harwood, J.L., Jones, A.L., Perry, H.J., Rutter, A.J., Smith, K.L. & Williams, M. (1994) In:
Temperature Adaptation of Biological Membranes (Cossins, A.R., ed.), Portland Press,
London, UK, p.107.
Huang, H.W., Goldberg, E.M. & Zidovetzki, R. (1996) Biochem. Biophys. Res. Commun.
220, 834.
Hughly, S., Kunst, L., Browse, J. & Somerville, C. (1989) Plant Physiol. 90, 1134.
Hughly, S. & Somerville, C. (1992) Plant Physiol. 99, 197.
Hutson, J.L. & Higgins, J.A. (1987) Biosci. Rep. 7, 73.
Illenberger, D., Walliser, C., Nurnberg, B., Diaz Lorente, M. & Gierschik, P. (2003a) J. Biol.
Chem. 278, 3006.
Illenberger, D., Walliser, C., Strobel, J., Gutman, O., Niv, H., Gaidzik, V., Kloog, Y.,
Gierschik, P. & Henis, Y.I. (2003b) J. Biol. Chem. 278, 8645.
Ishizaki-Nishizawa, O., Fujii, T., Azuma, M., Sekiguchi, K., Murata, N., Ohtani, T. & Toguri,
T. (1996) Nature Biotechnol. 14, 1003.
Janes, P.W., Ley, S.C., Magee, A.I. & Kabouridis, P.S. (2000) Semin. Immunol. 12, 23.
Jones, J.A. & Hannun, Y.A. (2002) J. Biol. Chem. 277, 15530.
Junge, S., Brenner, B., Lepple-Wienhues, A., Nilius, B., Lang, F., Linderkamp, O. & Gulbins,
E. (1999) Cell Signal. 11, 301.
Karlsson, O.P., Rytomaa, M., Dahlqvist, A., Kinnunen, P.K. & Wieslander, A. (1996)
Biochemistry 35, 10094.
Karlsson, O.P., Dahlqvist, A., Vikstrom, S. & Wieslander, A. (1997) J. Biol. Chem. 272, 929.
Katsikas, H. & Wolf, C. (1995) Biochim. Biophys. Acta 1258, 95.
Kelley, G.G., Reks, S.E., Ondrako, J.M. & Smrcka, A.V. (2001) EMBO J. 20, 743.
Kenworthy, A. (2002) Trends Biochem. Sci. 27, 435.
Kerkhoff, C., Trumbach, B., Gehring, L., Habben, K., Schmitz, G. & Kaever, V. (2000) Eur.
J. Biochem. 267, 6339.
Kinnunen, P.K. & Holopainen, J.M. (2000) Biosci. Rep. 20, 465.
Kirkpatrick, F.H., Gordesky, S.E. & Marinetti, G.V. (1974) Biochim. Biophys. Acta 345, 154.
Kirschnek, S., Paris, F., Weller, M., Grassme, H., Ferlinz, K., Riehle, A., Fuks, Z., Kolesnick,
R. & Gulbins, E. (2000) J. Biol. Chem. 275, 27316.
Klapisz, E., Masliah, J., Bereziat, G., Wolf, C. & Koumanov, K.S. (2000) J. Lipid Res. 41,
1680.
Klein, W., Weber, M.H. & Marahiel, M.A. (1999) J. Bacteriol. 181, 5341.
Kodama, H., Hamada, T., Horiguchi, G., Nishimura, M. & Iba, K. (1994) Plant Physiol. 105,
601.
Kodama, H., Horiguchi, G., Nishiuchi, T., Nishimura, M. & Iba, K. (1995) Plant Physiol. 107,
1177.
Kolesnick, R.N. (1991) Prog. Lipid Res. 30, 1.
Koumanov, K.S., Quinn, P.J., Bereziat, G. & Wolf, C. (1998) Biochem. J. 336, 625.
Koumanov, K.S., Tessier, C., Momchilova, A.B., Rainteau, D., Wolf, C. & Quinn, P.J. (2005)
Archiv. Biochem. Biophys. 432, 150.
Ktistakis, N.T., Delon, C., Manifava, M., Wood, E., Ganley, I. & Sugars, J.M. (2003)
Biochem. Soc. Trans. 31, 94.
Kunst, L., Browse, J. & Somerville, C. (1989a) Plant Physiol. 91, 401.
Kunst, L., Browse, J. & Somerville, C. (1989b) Plant Physiol. 90, 943.
Lehel, C., Los, D., Wada, H., Gyorgyei, J., Horvath, I., Kovacs, E., Murata, N. & Vigh, L.
(1993) J. Biol. Chem. 268, 1799.
Li, X.M., Momsen, M.M., Smaby, J.M., Brockman, H.L. & Brown, R.E. (2001) Biochemistry
40, 5954.
Lichtenbergova, L., Yoon, E.T. & Cho, W. (1998) Biochemistry 37, 14128.
Lin, L.L., Wartmann, M., Lin, A.Y., Knopf, J.L., Seth, A. & Davis, R.J. (1993) Cell 72, 269.
Lindblom, G., Brentel, I., Sjolund, M., Wikander, G. & Wieslander, A. (1986) Biochemistry
25, 7502.
Liu, P. & Anderson, R.G. (1995) J. Biol. Chem. 270, 27179.
London, E. (2002) Curr. Opin. Struct. Biol. 12, 480.
Lopez, I., Mak, E.C., Ding, J., Hamm, H.E. & Lomasney, J.W. (2001) J. Biol. Chem. 276,
2758.
Magargal, W.W., Dickinson, E.S. & Slakey, L.L. (1978) J. Biol. Chem. 253, 8311.
Mairhofer, M., Steiner, M., Mosgoeller, W., Prohaska, R. & Salzer, U. (2002) Blood 100, 897.
McKean, M.L. & Silver, M.J. (1985) Biochem. J. 225, 723.
McLaughlin, J.N., Thulin, C.D., Bray, S.M., Martin, M.M., Elton, T.S. & Willardson, B.M.
(2002) J. Biol. Chem. 277, 34885.
Miguel, M., Jas, D., Dooner, H. & Browse, J. (1993) Proc. Natl Acad. Sci. USA 90, 6208.
Moffett, S., Brown, D.A. & Linder, M.E. (2000) J. Biol. Chem. 275, 2191.
Momchilova, A., Petkova, D., Mechev, I., Dimitrov, G. & Koumanov, K. (1985) Int. J.
Biochem. 17, 787.
Momchilova, A., Petkova, D. & Koumanov, K. (1986) Int. J. Biochem. 18, 659.
Momchilova, A., Markovska, T. & Pankov, R. (1999) Biochem. Mol. Biol. Int. 47, 555.
Montixi, C., Langlet, C., Bernard, A.M., Thimonier, J., Dubois, C., Wurbel, M.A., Chauvin,
J.P., Pierres, M. & He, H.T. (1998) EMBO J. 17, 5334.
Murata, N. & Wada, H. (1995) Biochem. J. 308(1), 1.
Murata, N., Ishizaka-Nishizawa, O., Higashi, S., Hayashi, H., Tasaka, Y. & Nishida, I. (1992)
Nature 356, 710.
Nalefski, E.A., Sultzman, L.A., Martin, D.M., Kriz, R.W., Towler, P.S., Knopf, J.L. & Clark,
J.D. (1994) J. Biol. Chem. 269, 18239.
Nishida, I., Swinhoe, R., Slabas, A.R. & Murata, N. (1996) Plant Mol. Biol. 31, 205.
Nixon, A.B., Greene, D.G. & Wykle, R.L. (1996) Biochim. Biophys. Acta 1300, 187.
Ohanian, J. & Ohanian, V. (2001) Cell Mol. Life Sci. 58, 2053.
Okuyama, H., Saito, M., Joshi, V.C., Gunsberg, S. & Wakil, S.J. (1979) J. Biol. Chem. 254,
12281.
Oquist, G., Campbell, D., Clarke, A.K and Gustafsson, P. (1995) Photosynthesis Res. 46, 151.
Palta, J.P., Whitaker, B.D. & Weiss, L.S. (1993) Plant Physiol. 103, 793.
Parton, R.G., Molero, J.C., Floetenmeyer, M., Green, K.M. & James, D.E. (2002) J. Biol.
Chem. 277, 46769.
Petkova, D.H., Momchilova-Pankova, A.B. & Koumanov, K.S. (1987) Biochimie 69, 1251.
Philip, F., Guo, Y. & Scarlata, S. (2002) FEBS Lett. 531, 28.
Polonovski, J., Bard, D., Colard, O. & Bereziat, G. (1974) Comptes Rendues Séances Soc.
Biol. Fil. 168, 1211.
Pugh, E.L. & Kates, M. (1984) Lipids 19, 48.
Quinn, P.J. (1998) In: Aqueous Phase Organometallic Catalysis: Concepts and Applications
(Cornils, B.A.H. and Herrmann, W.A., eds), Wiley-VCH, Weinheim, Germany, p. 418.
Quinn, P.J. (2002) Subcell. Biochem. 36, 39.
Rebecchi, M.J. & Pentyala, S.N. (2000) Physiol. Rev. 80, 1291.
Rhee, S.G. (2001) Annu. Rev. Biochem. 70, 281.
Ritov, V.B., Banni, S., Yalowich, J.C., Day, B.W., Claycamp, H.G., Corongiu, F.P. & Kagan,
V.E. (1996) Biochim. Biophys. Acta 1283, 127.
Rizzo, M.A., Shome, K., Watkins, S.C. & Romero, G. (2000) J. Biol. Chem. 275, 23911.
Roberts, W.L., Santikarn, S., Reinhold, V.N. & Rosenberry, T.L. (1988) J. Biol. Chem. 263,
18776.
Roper, K., Corbeil, D. & Huttner, W.B. (2000) Nature Cell Biol. 2, 582.
Roughan, P.G. (1985) Plant Physiol. 77, 740.
Rouquette-Jazdanian, A.K., Pelassy, C., Breittmayer, J.P., Cousin, J.L. & Aussel, C. (2002)
Biochem. J. 363, 645.
Rutter, A.J., Thomas, K.L., Herbert, D., Henderson, R.J., Lloyd, D. & Harwood, J.L. (2002)
Biochem. J. 368, 57.
Sachs, J.N., Petrache, H.I. & Woolf, T.B. (2003) Chem. Phys. Lipids 126, 211.
Sakyo, T. & Kitagawa, T. (2002) Biochim. Biophys. Acta 1567, 165.
Samet, D. & Barenholz, Y. (1999) Chem. Phys. Lipids 102, 65.
Sato, N. & Murata, N. (1980) Biochim. Biophys. Acta 619, 353.
Scheideler, M.A. & Bell, R.M. (1986) J. Biol. Chem. 261, 10990.
Schroeder, R.J., Ahmed, S.N., Zhu, Y., London, E. & Brown, D.A. (1998) J. Biol. Chem. 273,
1150.
Shvedova, A.A., Tyurina, Y.Y., Tyurin, V.A., Kikuchi, Y., Kagan, V.E. & Quinn, P.J. (2001)
Biosci. Rep. 21, 33.
Shvedova, A.A., Tyurina, J.Y., Kawai, K., Tyurin, V.A., Kommineni, C., Castranova, V.,
Fabisiak, J.P. & Kagan, V.E. (2002) J. Invest. Dermatol. 118, 1008.
Singer, W.D., Brown, H.A. & Sternweis, P.C. (1997) Annu. Rev. Biochem. 66, 475.
Siskind, L.J., Kolesnick, R.N. & Colombini, M. (2002) J. Biol. Chem. 277, 26796.
Slimane, T.A., Trugnan, G., Van, I.S.C. & Hoekstra, D. (2003) Mol. Biol. Cell 14, 611.
Song, C., Satoh, T., Edamatsu, H., Wu, D., Tadano, M., Gao, X. & Kataoka, T. (2002)
Oncogene 21, 8105.
Sonoki, S. & Ikezawa, H. (1976) J. Biochem. (Tokyo) 80, 1233.
Sugars, J.M., Cellek, S., Manifava, M., Coadwell, J. & Ktistakis, N.T. (2002) J. Biol. Chem.
277, 29152.
Sugiura, T., Masuzawa, Y., Nakagawa, Y. & Waku, K. (1987) J. Biol. Chem. 262, 1199.
Swinnen, J.V., Van Veldhoven, P.P., Timmermans, L., De Schrijver, E., Brusselmans, K.,
Vanderhoydonc, F., Van de Sande, T., Heemers, H., Heyns, W. & Verhoeven, G. (2003)
Biochem. Biophys. Res. Commun. 302, 898.
Tasaka, Y., Gombos, Z., Nishiyama, Y., Mohanty, P., Ohba, T., Ohki, K. & Murata, N. (1996)
EMBO J. 15, 6416.
Thomas, K., Rutter, A., Suller, M., Harwood, J. & Lloyd, D. (1998) FEBS Lett. 425, 171.
Tou, J.S. (1981) Biochim. Biophys. Acta 665, 491.
Tsvetkova, N.M., Brain, A.P. & Quinn, P.J. (1994) Biochim. Biophys. Acta 1192, 263.
Tsvetkova, N.M., Apsotovova, E.L., Brain, A.P.R., Williams, W.P. & Quinn, P.J. (1995)
Biochim. Biophys. Acta 1228, 201.
Tyurina, Y.Y., Tyurin, V.A., Carta, G., Quinn, P.J., Schor, N.F. & Kagan, V.E. (1997) Arch.
Biochem. Biophys. 344, 413.
Uemura, M. & Steponkus, P.L. (1997) Plant Physiol. 114, 1493.
Venkataraman, K. & Futerman, A.H. (2000) Trends Cell Biol. 10, 408.
Vieu, C., Terce, F., Chevy, F., Rolland, C., Barbaras, R., Chap, H., Wolf, C., Perret, B. &
Collet, X. (2002) J. Lipid Res. 43, 510.
Vigh, L., Horvath, I. & Thompson, G.A., Jr (1988) Biochim. Biophys. Acta 937, 42.
Wada, H., Gombos, Z. & Murata, N. (1990) Nature 347, 200.
Wang, D., Feng, J., Wen, R., Marine, J.C., Sangster, M.Y., Parganas, E., Hoffmeyer, A.,
Jackson, C.W., Cleveland, J.L., Murray, P.J. & Ihle, J.N. (2000) Immunity 13, 25.
Wang, T.Y. & Silvius, J.R. (2001) Biophys. J. 81, 2762.
Webb, M.S. & Steponkus, P.L. (1993) Plant Physiol. 101, 955.
Wilde, J.I. & Watson, S.P. (2001) Cell Signal. 13, 691.
Wolf, C., Colard-Torquebiau, O., Bereziat, G. & Polonovski, J. (1997) Biochimie 59, 115.
Wolter, F.P., Schmidt, R. & Heinz, E. (1992) EMBO J. 11, 4685.
Wu, J. & Browse, J. (1995) Plant Cell 7, 17.
Xu, X. & London, E. (2000) Biochemistry 39, 843.
Xu, X., Bittman, R., Duportail, G., Heissler, D., Vilcheze, C. & London, E. (2001) J. Biol.
Chem. 276, 33540.
Yang, L., Ding, L. & Huang, H.W. (2003) Biochemistry 42, 6631.
Zubiaur, M., Fernandez, O., Ferrero, E., Salmeron, J., Malissen, B., Malavasi, F. & Sancho,
J. (2002) J. Biol. Chem. 277, 13.
CHAPTER 10
Lipid barriers at the environment–body
interface
A. Introduction
Living organisms exist in a polar air/aqueous environment and the interior of
cells is dominated by a hydrophilic milieu. The interface between the environ-
ment and the interior of cells is created by the plasma membrane. The barrier to
the permeation of solutes in and out of the cell is represented by the hydropho-
bic domain of the lipid bilayer matrix of this membrane. Multicellular organisms
have evolved specialized barrier structures to supplement the function of the
plasma membrane, protect the underlying tissues and organs, and assist them in
their functions. Like the membranes of individual cells, the barrier to the
passage of water and solutes is a hydrophobic domain created by polar lipids
together with specialized proteins which serve to assist in the organization and
maintenance of the lipid structures. While the outer surface skin is dry the inner
surfaces of the body are wet. The internal surfaces consist of hydrophilic
proteoglycans and mucins which form an aqueous, gel-like phase, the mucus.
A surface film of polar lipids is often a component of the mucous zone, with a
significant influence on the properties of the surface. The flow and regenera-
tion of the serbium, representing the hydrophobic surface of the skin, and the
hydrophilic mucus, covering the inner surfaces, both serve to clean and protect
the underlying tissues from harmful microorganisms.
Below the mucous layer lies the endothelial plasma membrane, which is
replenished from within the cell. The transport of polar lipids into the lumen via
the mucous layer is responsible for conferring an amphiphilic character to
mucous surfaces, the turnover of which has often been neglected. The presence
of polar lipids in surface layers produces physical properties aiding tissue
function. The condensed monolayer lining the alveoli of the lung is essential to
prevent collapse of the lung. By contrast, saliva and the oral mucosa do not
have a lipid monolayer covering the surface, and the physical properties
reflected in the surface tension differ markedly from those of lung surfactant
(Sefton et al., 1992). The reduction in surface tension at the air–water interface
by components in the saliva is about 30 mN/m, which is significantly lower
than would be expected if the polar lipids were oriented at the interface.
This chapter will examine the composition and structure of barriers created
in the gastrointestinal tract, in skin, and in the surfactant lining the alveoli of the
lung. The mechanism of regulation and secretion of lipid barriers will also be
considered.
219
B. Skin: The stratum corneum

The mammalian stratum corneum represents the boundary separating the
internal organs and tissues from a variable and often hostile environment. The
basic structure of the stratum corneum consists of keratin-filled corneocytes
surrounded by multilamellar arrangements of lipids (Elias, 1983). It is the
properties of this layer that are responsible for the particular functions it is
required to perform.
One of the principal functions of the stratum corneum is to prevent loss of
water from the body, but prevention of water entry is also a factor of some
importance. Trans-epidermal water loss in humans under normal conditions is
of the order of 500 ml per day but may be much greater if the layer is damaged
or breached. Another function is to prevent penetration of foreign agents and
xenobiotics and, as a therapeutic route, the stratum corneum is a major
challenge in the administration of pharmaceutical agents.
As well as impeding xenobiotic reagents the stratum corneum must protect
the body from invasion by a variety of microorganisms. One strategy by which
this is achieved is desquamation, in which the major cells of skin tissue, the
corneocytes, move out from their site of differentiation and growth and slough
off from the surface. This constant renewal prevents microbes from success-
fully colonizing the skin. In addition, chemical defences in the form of
antimicrobial peptides and toll receptors mediating defence responses within
the body complete the armoury of the stratum corneum against microbial
infection.
Another potential source of damage to the stratum corneum is irradiation by
ultraviolet light. A strategy evolved to deal with this is the production of
pigments like melanin by the epidermal melanocytes; these pigments are then
transferred to keratinocytes and ultimately corneocytes where they act to
absorb the harmful radiation.
1. Lipid composition of the stratum corneum

The major components of the human stratum corneum are polar lipids. The
factor which distinguishes the lipid composition of the stratum corneum from
that of the lipid matrix of cell membranes is the virtual absence of phospholipids
in the stratum corneum. The lipid composition is instead characterized by the
presence of three types of polar lipids: ceramides, long-chain free fatty acids,
and cholesterol.
A number of ceramide molecular species have been identified and character-
ized in the stratum corneum (Ponec et al., 2003; Robson et al., 1994; Stewart
& Downing, 1999; Wertz et al., 1985). The ceramides consist of a sphingosine,
phytosphingosine or 6-hydroxysphingosine base to which is attached either a
long-chain fatty acid or an α-hydroxy fatty acid. The structures of ceramides
identified in the stratum corneum are shown in Figure 10.1.
LIPID BARRIERS AT THE ENVIRONMENT–BODY INTERFACE 221
Figure 10.1 Ceramides of human stratum corneum. The particular features of CER1–CER9 are
indicated by the following keys: A, α-hydroxy fatty acid; EO, linoleic acid bound to ω-hydroxy fatty
acid; H, 6-hydroxysphingosine; N, non-hydroxy fatty acid; P, phytosphingosine; S, sphingosine (adapted
from de Jager et al., 2004).
Epidermal sphingomyelins of different molecular species are found to be

precursors of 2 of the 7 naturally occurring ceramides of the stratum corneum,
ceramides 2 and 5. Ceramide 2 is the most abundant ceramide in the epidermis
and ceramide 5 is believed to be derived exclusively from hydrolysis of
sphingomyelin (Uchida et al., 2000). The decrease in activity of
sphingomyelinase has been shown to be associated with defects in the barrier
properties of the stratum corneum (Jensen et al., 2004). The pathway of
biosynthesis of some acyl ceramides in the stratum corneum does not appear to
involve a substrate precursor of ceramide generated by the action of
sphingomyelinase on sphingomyelin. Studies using inhibitors of
glucosyltransferase and β-glucocerbrosidase have shown that acyl ceramides
are synthesized in a pathway involving the glucosylation of ceramide by
glucosyltransferase, the ω-hydroxylation by hydroxylase to form ω-
hydroxyglucosylceramide, and finally deglucosylation of acylglucosylceramide
by β-glucocerebrosidase (Takagi et al., 2004).
2. Structure and organization of the lipids
The structure, phase behaviour and organization of lipids of the stratum

corneum have been examined in detail using model systems. The most common
mixtures consist of bovine brain ceramide type III or type IV with cholesterol
and/or palmitic acid (Fenske et al., 1994; Kitson et al., 1994; Moore et al.,
1997; Neubert et al., 1997; Percot & Lafleur, 2001; Wegener et al., 1997).
Rather than using naturally occurring ceramides that are comprised of a range
of different molecular species (Ponec et al., 2003) other studies have used
synthetic ceramides that contain a single molecular species (Chen et al., 2000;
Moore & Rerek, 2000; Ohta & Hatta, 2002; Raudenkolb et al., 2003).
X-ray diffraction methods have shown that the lipids in these mixtures form
a structure consisting of two coexisting crystalline lamellar phases with repeat
spacings of about 6 nm and 13 nm, respectively (Bouwstra et al., 1991, 1995;
McIntosh et al., 1996). Comparison of these mixtures containing synthetic
ceramides with mixtures of cholesterol, fatty acids and ceramides extracted
from human and porcine stratum corneum have indicated similar structure and
phase behaviour (Bouwstra et al., 1998, 2002). The findings indicate that the
structure with a periodicity of about 13 nm depends on the particular ceramide
and its proportion with respect to cholesterol in the mixture. The packing of the
lipids into the lamellar phases has been found to be mainly orthorhombic and
is poised with hexagonal packing in the temperature range 35–37ºC. This
configuration of the hydrocarbon chains depends on the presence of the fatty
acid (Bouwstra et al., 2002; de Jager et al., 2004). Raman spectroscopic studies
have shown that the orthorhombic crystalline packing of the hydrocarbon
chains of the stratum corneum was disrupted in skin disorders like psoriasis
(Osada et al., 2004).
Electron density calculations of the unit cell representing the lamellar phase
of about 13 nm periodicity have been reported (McIntosh, 2003). On the basis
of these calculations it has been postulated that the unit cell comprises two
bilayers, each of which has an asymmetric distribution of constituent lipids, in
particular the cholesterol which is said to be located predominantly in the outer
monolayer of each bilayer. Studies of the effects of pH on the electron density
profile indicated that the water layers on the outer edges of the profiles
increased in thickness with increasing pH without significant increase in the
distance separating the bilayers at the centre of the profile. This was interpreted
as co-localization of the ionizable palmitic acid with cholesterol in the outer
monolayers of the bilayer structures. The model for the structure of the stratum
corneum proposed on the basis of this evidence consists of a repeating structure
of two bilayers with a largely symmetric arrangement of most of the ceramides
present (ceramides 2–5). Ceramide 1 is asymmetrically arranged because of its
distinctive hydrocarbon substituent which, unlike the other saturated molecu-
lar species of ceramide, contains an unsaturated ester-linked linoleic acid
chain. Cholesterol does not interact readily with unsaturated hydrocarbon

chains and prefers the proximity of saturated hydrocarbons; it therefore locates
on the same side of the bilayer as the palmitic acid and opposite to that of the
monolayer in which the linoleoyl chain resides (McIntosh, 2004). The func-
tional consequences of this arrangement of lipids are that the tight packing
density between the cholesterol and free fatty acids impedes the permeability of
water across the stratum corneum, and the narrow hydration channels separat-
ing the bilayers under acid pH conditions limit lateral diffusion of water
transversely along the bilayer structures.
C. Lipid barriers of the alimentary tract

The alimentary tract extends from the mouth to the anus. The environments
encountered in the different regions of the tract vary widely. The surface
coatings therefore differ to cater for the particular functions performed during
the digestive process.
1. Saliva
Normal saliva has relatively low amounts of polar lipids, which are present in
a concentration of about 10 mg/100 ml. The classes of lipids that are found
include sphingolipids, glyceroglucolipids and glycerophospholipids. The ma-
jor secretions from the parotid and submaxilliary glands are, however, rich in
proteins which have strongly surface-active properties and help to form a
protective interfacial film over the entire oral cavity.
Lipids of saliva are an important secretion from the salivary glands and they
are required to maintain the integrity of both soft and hard oral tissues. Elevated
levels of lipids in saliva are associated with a high incidence of dental caries
and the development of plaque, calculus and resulting periodontal disease. The
lipids of saliva have been shown to affect the penetration of hydrophobic
substances across the oral mucosa, alter the interaction of salivary proteins with
calcium, and influence glycosyltransferase activity associated with cariogenic
bacteria. Furthermore the phospholipids of saliva interact heterotypically with
proteins and glycoproteins to form a dynamic continuum that dictates many
physiochemical and biological properties of saliva (Slomiany et al., 1988).
These properties include viscoelasticity, lubricative properties, proteolytic
susceptibility and the formation of protective coatings over the oral mucosa and
tooth enamel (Slomiany et al., 1989).
The factors that regulate the secretion of lipids by the salivary glands are
unclear, although secretion in general from the glands is effected mainly
through neural, paracrine and endocrine systems. Stimulation via the
parasympathetic nerve results mainly in an increase in salivary fluid secretion,
while stimulation via sympathetic nerves affects the secretion of macromolecular
constituents of saliva (Cook et al., 1994). Thus, studies of acinar cells of

salivary glands have shown that β-adrenergic agonists mediating cAMP
production stimulate protein and mucin secretion (Slomiany & Slomiany,
2004). Furthermore, isoproterenol stimulation of sublingual gland cells in
tissue culture leads to an increased release of phospholipids (Slomiany et al.,
1991).
2. The gastric mucous surface

The gastric wall contains surface mucus cells that secrete mucus and
phospholipids. These secretions line the stomach wall and protect it from the
action of secretions required for the digestive process. The gastric mucus forms
a gel layer which acts as a lubricant for ingested food and protects against injury
from acid pH, protease enzymes, alcohol, hypertonic and hypotonic foods,
spices, etc.
The mucous lining of the gut is normally in a dynamic balance in which
material eroded from the luminal surface by proteolytic action or mechanical
abrasion is replenished by synthesis and secretion from goblet cells which are
a component of the underlying epithelium. Protection against proteolysis is
afforded by mucopolysaccharides which display their carbohydrate moieties
on the luminal surface, rendering the polypeptide component inaccessible to
hydrolysis. It should be noted that this fact makes mucin poorly digestible and
represents a nutritional deficit to the animal.
Mucins fall into two categories: those that are secreted and those which
remain as components of the plasma membranes of the epithelium. Secreted
mucins are characterized by their relatively large size (about 2 MDa) and a
carbohydrate complement that represents 50–80% of the dry weight of the
mucin. The carbohydrates are linked predominantly to the polypeptide chain
by O-glycocydic linkages. The high carbohydrate content allows secreted
mucins to form viscoelastic gels. Mucins bound to the membranes share
common structural features with secreted mucins, except that they contain
polypeptide sequences that anchor them to the membrane and they cannot form
three-dimensional gels. Instead they are displayed on the surface of the cell
where they form the so-called glycocalyx. The mucins can also be subdivided
into two types depending on whether the monosaccharides are uncharged
(neutral mucins), or carry acidic groups such as sulphates (sulphomucins) or
neuraminic acid (sialomucins). The distribution of neutral, sulpho- and
sialomucins varies along the alimentary tract and can be influenced by diet
(Montagne et al., 2004).
The mucin layer is also a repository of a variety of microorganisms that
change with diet and are involved in the normal digestive processes. The gut
ecosystem has been defined in terms of three components: the diet, the
commensal flora, and the gut mucosa. The last component includes the
digestive epithelium, the gut-associated lymphoid tissue and the overlying

mucus. The complex carbohydrates of mucin act as specific adherence sites for
both commensal and pathogenic bacteria. The success with which the mucin
layer can protect the epithelium from infection depends on its thickness and its
ability to fix bacteria (Araneo et al., 1996). Furthermore, the carbohydrate may
act as a substrate for fermentation by microorganisms and the products may
alter the properties of the mucous layer (Corfield et al., 1992).
There is evidence that, in addition to mucin, phospholipids are secreted from
mucus cells of the epithelium. The phospholipids are likely to be positioned
within the mucous layer at an interface determined by the concentration of
proteoglycans in the aqueous phase. The concentration of proteoglycans is
greatest at the epithelial layer where they are produced and decreases towards
the luminal surface where they exist in a water-rich phase. Because the
phospholipids do not partition into the aqueous mucin phases, the phospholipid
layer retains a separate identity where it can act as a hydrophobic barrier to the
penetration of solutes. An oligolamellar lining comprised of phospholipids has
been observed at the gastric wall and it has even been proposed to constitute the
physical identity of the gastric barrier (Hills, 1990). Studies based on contact
angle measurements have also indicated the possibility of a hydrophobic
barrier at the gastric surface, and correlations with the reduction in
hydrophobicity of this barrier and peptic ulcer disease have been established
(Goggin et al., 1992). There is, however, an apparent misunderstanding of the
nature of the hydrophobic barrier. The surface is considered to be hydrophobic
under physiological conditions because of the orientation of the phospholipid
molecules, but in fact contact angle measurements simply reflect the amphiphilic
properties due only to the presence of phospholipid molecules at an interface
between the mucous phases described above. The orientation of the
phospholipids with their hydrocarbon chains at the air interface is created by
the method used to perform the contact angle measurement.
One of the potential hazards of the presence of phospholipids in the gastric
mucosa is that under the highly acid conditions there is a measurable amount of
acid hydrolysis of the fatty acyl ester groups. A product of such hydrolysis is
lysophospholipids. These lysophospholipids are known to damage the gastric
mucosa in dogs (Duane et al., 1986) and represent a potential risk factor for
gastric ulcers in humans (Bukholm et al., 1997). Such effects can be prevented
by supplementation with polyunsaturated molecular species of
phosphatidylcholine (Demirbilek et al., 2004). Gastric lipases, some of which
originate from salivary secretions, are known to be responsible for the diges-
tion of dietary fats in the stomach (Moreau et al., 1990). These enzymes have
higher activity against acylglycerols and do not hydrolyse phospholipids
appreciably. A surface film of phosphatidylcholine molecules covering the
mucous phase should not therefore be influenced by the presence of such
lipases.
3. Surface barriers of the lower alimentary tract
Mucus forms a lining along the entire intestinal wall. The small intestine is
particularly efficient at digesting food and absorbing the nutrients thereby
released. Water absorption is a major function of the colon. Common features
of this region of the mucous layer are the presence of microorganisms and the
location of important components of the immune system. T-cell lymphocytes in
the epithelium cells control reactions against microorganisms and food compo-
nents, and this region acts as the mammalian equivalent of the bursa in birds,
where B-cells are differentiated.
The low pH of the stomach provides an effective protection against poten-
tially pathogenic microorganisms. When food enters the duodenum the pH is
normalized and antimicrobial protection is afforded by the resident microbial
flora which inhabits the mucous layer (Strompfova et al., 2004). Lactobacilli
are known to inhibit the growth of potentially pathogenic bacteria (Johansson
et al., 1993) and there is evidence that they may stimulate an immune response
(Fuller, 1991). For this reason they are sometimes described as probiotics.
Their packing density increases progressively down the alimentary tract from
about 106 to 108 colony-forming units per millilitre of mucus from one end to
the other. Colonization with different Lactobacillus strains has been followed
in vivo and a corresponding decrease of Gram-negative bacteria was observed
(Johansson et al., 1993). Helicobacter pylori are among the bacteria that are
associated with gastrointestinal inflammatory diseases and ulcer formation
(Suerbaum & Michetti, 2002). Their mode of action is to neutralize the low pH
of gastric secretions by extracellular urease activity. The bacteria move through
the mucous layer and adhere to glycerolipid receptors in the gastric wall with
the aid of flagella (Lingwood et al., 1989). It is interesting to note that
colonization of the intestinal wall by lactobacilli appears to be host specific, so
the particular strain that colonizes the mucosa of one animal species will not
normally be found in the epithelium of another.
The surface area of the intestine is manyfold greater than the surface area of
the skin. To put the metabolic demands on the maintenance of the intestinal
epithelium into some perspective, the cell membranes are renewed every two
days. In consequence the transfer of the lipids of the cell membranes into the
mucous layer is considerable and the presence of these lipids imparts a non-
wettable character to the surface. It is possible that this hydrophobic character
is due to accumulation of phospholipid within a phase-separated domain in the
mucous layer, as suggested above for the gastric mucosa.
The lipid spectrum in the intestinal mucosa is highly complex. In addition to
the lipids of the constituent membranes of the epithelial cells, bile containing
phospholipids, cholesterol and bile acids, as well as food and the degradation
products produced by pancreatic enzymes, are present. The surface of the brush
border of the small intestine is covered by a diffuse layer of carbohydrate
representing the oligosaccharide components of the glycoproteins and

glycolipids present in the plasma membrane of the epithelial cells. This
glycocalyx can be regarded as a separate phase in relation to the mucus. Outside
this layer there is a film of mucus which flows constantly from the mucosal
cells. One physical property of the glycolipids is their relatively high critical
micelle concentration due to the preponderance of hydrated sugar residues.
This affects their affinity for the membrane where they are anchored by their
hydrocarbon chains. Loss of glycolipids into mucus, because of solubility, may
be an important mechanism underlying renewal of the intestinal membrane.
A special group of lipids play an important role in the mucous layer of the
colon. These are short-chain fatty acids which are the fermentation products of
dietary fibre produced by bacteria colonizing the colon. These short-chain fatty
acids represent one of the primary metabolic substrates for endothelial cells of
the colon and are a factor in determining susceptibility to carcinoma of the
colon (Lupton, 2004). The exogenous administration of phospholipids has
been reported to prevent acetic acid-induced colitis, and phosphatidylinositol
appeared to be more effective than phosphatidylcholine (Fabia et al., 1992).
4. Intestinal absorption of lipids

Triacylglycerols are digested by pancreatic lipase/colipase and carboxylic
ester lipase. There is interplay between these two enzymes in the absorption of
the nutritionally important C20:4, C20:5 and C22:6 fatty acids (Chen et al., 1989).
Other fats can be digested into fatty acids and 2-monoacylglycerols by lipase/
colipase only. The partial digestion of dietary lipids by lingual and gastric
lipases can be neglected in this connection. The phase behaviour of the
lipolysis products is a significant factor in the absorption process. Bile acids
contribute to micellar solubilization of monoacylglycerols and fatty acids next
to the brush border membranes, and the monomers are taken up by specific
proteins in the membranes of those cells involved in lipid absorption, the
enterocytes (Cao et al., 2004).
Phospholipids in the diet, from bile, or released from the brush border
membranes are digested by phospholipase A2, and the lysophosphatides parti-
tion into the enterocyte membrane. The lysophosphatides serve as a substrate
for esterification and eventual conversion into triacylglycerols via the
monoacylglycerol pathway, or into phospholipids via the phosphatidic acid
pathway. Pre-chylomicron particles are then formed and exported into the
lymphatic system. In the case of fatty acids with medium chain lengths (C8–
C10), the free fatty acids are exported directly into the portal blood.
D. The peritoneal mesothelium

Peritoneal fluid is known to contain phospholipids and it has been proposed
that one function that these phospholipids may perform is to repel water and
lubricate the tissue (Ziegler et al., 1989). Prolonged dialysis has been shown to
result in a decrease in the phospholipid content, and the high risk of adhesions
following dialysis has been attributed to this decrease, which is believed to be
related to loss of phospholipids from the surface of mesothelial cells (Ar’Rajab
et al., 1991). This is consistent with experimental evidence showing that
intraperitoneal administration of phosphatidylcholine could prevent tissue
adhesions in rats.
E. Lung surfactant
The alveoli of the lung are lined with a surfactant that is essential for normal
pulmonary function. The surfactant consists of a few specialized proteins and
polar lipids which are synthesized and secreted by type II pneumocytes and
which then adsorb to the air–water interface in the alveoli (Piknova et al.,
2002). The process of synthesis and secretion of pulmonary surfactant onto the
alveolar surface is illustrated schematically in Figure 10.2.
The surfactant is required to function immediately at birth, when the air–
water interface expands dramatically as the amniotic fluid is expelled, and to
remain in place thereafter during cycles of compression and expansion in the
respiratory cycle. The remarkable feature of the surfactant is that it is able to
form a film that reduces surface tension at the alveolar air–liquid interface
(Goerke, 1998). In this way it stabilizes the small air spaces of the lung during
exhalation when compression of the structure reaches surface pressures well
above that where films would be expected to collapse. This property of
pulmonary surfactant is believed to reside in the lipid constituents rather than
the proteins that are found associated with the surfactant.
In addition to its primary role in preserving the structure of the functioning
lung, pulmonary surfactant represents the interface between the environment
and the underlying body tissues. The surfactant is known to contain agents that
participate in the innate immune response to microbial infection and to act in
other aspects of immune and inflammatory processes within the lung (Crouch
& Wright, 2001; Vayrynen et al., 2004).
1. Composition of pulmonary surfactant

Pulmonary surfactant can be recovered simply by bronchoalveolar lavage and
separated by differential centrifugation into large and small fractions. The
composition of pulmonary surfactant is remarkably similar for most mamma-
lian species. The predominant component is polar lipid of which 90% is
phospholipid and the remainder is neutral lipid, mainly cholesterol. The
surfactant contains about 5–10% surfactant-associated proteins. The proteins
are dominated by two large hydrophilic proteins, designated SP-A and SP-D,
and two hydrophobic proteins, SP-B and SP-C.
Type I
pneumocyte
Air space
Air-water
interface
Adsorption Type II
pneumocyte
Air Lamellar
body
Fluid
Figure 10.2 Schematic of pulmonary surfactant in an alveolus of the lung. Type II pneumocytes
synthesize the surfactant constituents and package them into the concentric bilayers of lamellar bodies.
After secretion, the bilayers unravel and adsorb to the air–water interface (see inset). During exhalation,
the decreasing alveolar surface area compresses the interfacial film. The arrangement of the surfactant
lipid underlying the surface monolayer has not yet been clearly established (adapted from Piknova et al.,
2002).
The hydrophobic proteins are believed to be responsible for the transport of

surfactant, delivered in the form of vesicles into a thin film lining the surface of
the alveoli. The current model to explain this function is that SP-B and SP-C act
as catalysts to reduce the potential energy of structural intermediates between
bilayers and surfactant monolayer (Schram & Hall, 2004; Walters et al., 2000).
The hydrophilic surfactant-associated proteins, SP-A and SP-D, are mem-
bers of a family of glycoproteins referred to as collectins. This group of proteins
is characterized by a structure consisting of a short, cysteine-rich N-terminal
cross-linking domain followed by a collagen domain connected to an α-helical
segment which terminates in a C-type lectin domain (Kuroki & Voelker, 1994).
The function of the most abundant of the surfactant-associated proteins, SP-A,
deduced from in vitro studies, is to augment the surface tension-lowering
properties of the phospholipids (Rodriguez-Capote et al., 2001; Venkitaraman

et al., 1990) in order to prevent protein inactivation of surfactant (Cockshutt et
al., 1990; Holm et al., 1990) and loss of surfactant properties associated with
oxidative stress (Rodriguez Capote et al., 2003). The primary function of SP-
D is to bind to glycoconjugates displayed on the surface of a variety of
pathogens, thereby enhancing phagocytosis by macrophages and clearance
from the lung (Holmskov et al., 2003). Binding of SP-D to bacteria also inhibits
their growth and induces leakage of cell contents (Wu et al., 2003).
The lipid composition of bovine pulmonary surfactant is shown in
Table 10.1. It can be seen that choline phosphatides dominate the phospholipid
composition and represent about 83% of the total lipid. The acidic phospholipids
comprise about 15%, with phosphatidylglycerol being the most prominent. The
neutral lipid, cholesterol, makes up the remainder. While the lipid composition
of bovine pulmonary surfactant is typical of mammalian species some differ-
ences in the proportions of acidic lipids have been noted, although the relative
proportion of these lipids remains fairly constant. Some changes in phospho-
lipid composition have been detected during development (Rau et al., 2004)
and cholesterol levels are known to vary between species and can alter with
different physiological states (Orgeig & Daniels, 2001). It has been suggested
that phase separation of cholesterol into enriched and depleted domains in the
phospholipid bilayer is responsible for the functional properties of the sur-
factant (Larsson et al., 2003a). With the advent of more sensitive methods of
analysing surfactant preparations the lipidomics of pulmonary surfactant
associated with its function in development and disease states will no doubt be
revealed in greater detail (Bernhard et al., 2004).
2. Formation of pulmonary surfactant

The components of the surfactant are synthesized, assembled and secreted from
alveolar type II pneumocytes (see Figure 10.2). Apart from the distinctive
surfactant-associated proteins the cells must also synthesize phospholipids
which have a preponderance of disaturated fatty acyl chains. The pathway of
phospholipid biosynthesis in type II cells is indistinguishable from that ob-
served in other cells and the molecular species composition is believed to be
modified after synthesis of the phospholipid. The most likely manner in which
this remodelling occurs is via the Land’s deacylation/reacylation cycle, in
which phospholipase A2 catalyses the hydrolysis of the fatty acid in the C-2
position of the glycerol to produce the lysophospholipid, which is subsequently
reacylated with palmitoyl-CoA (Batenburg & Haagsman, 1998).
The usual site of phospholipid biosynthesis is the endoplasmic reticulum,
from which the lipids are distributed to other organelles. One of the conspicu-
ous features of Type II cells in the late-term foetus is the presence of large
granules of glycogen which dominate the cytoplasm of the cell. A recent
Table 10.1 Lipid composition of bovine pulmonary surfactant
Lipid % by weight
Dipalmitoylphosphatidylcholine 34.3
Unsaturated phosphatidylcholinesa 33.6
1-palmitoyl-2-myristoylphosphatidylcholine 7.1
Phosphatidylglycerol 9.6
Choline plasmalogen 6.8
Phosphatidylethanolamine 2.4
Phosphatidylinositol 1.0
Lyso-bis-phosphatidic acid 0.9
Sphingomyelin 1.6
Cholesterol 2.6
a
Mainly 1-palmitoyl-2-palmitoleoyl and 1,2-palmitoleoyl molecular species (adapted from Veldhuizen,
2004).
ultrastructural study to identify the location of fatty acid synthase and CTP-
phosphocholine cytidylyltransferase within type II cells showed that the
glycogen granules were populated by lipid-synthesizing enzymes as well as
surfactant proteins and surfactant storage organelles (Ridsdale & Post, 2004).
It was suggested that this location would ensure that the supply of substrate for
phospholipid biosynthesis would not be rate limiting at a time of maximum
surfactant production.
The surfactant produced within type II pneumocytes is accumulated in the
cytoplasm in the form of lamellar bodies. These structures are lysosome-
derived organelles and are characterized by a low luminal pH and a bounding
membrane containing lysosome-associated membrane proteins. Ultra-
structural studies have shown that multivesicular structures containing sur-
factant components fuse to form so-called composite bodies and ultimately
mature into lamellar bodies (Stahlman et al., 2000). During this process the
complete complement of surfactant components is assembled from their re-
spective sites of synthesis or post-translational modification in the Golgi
apparatus. The formation of lamellar bodies is dependent on the expression of
the surfactant-associated protein SP-B (Foster et al., 2003).
Lamellar bodies are secreted via the apical membrane of type II cells. The
factors that control secretion are conjectural because conflicting evidence has
been obtained from studies performed in vivo and those carried out in vitro.
Nevertheless, the process of secretion of lamellar bodies from type II
pneumocytes appears to be similar to that of other cell types in that it is
regulated by secretagogues (Rooney, 2001) and targeted to the plasma mem-
brane by syntaxin 2 and SNAP-23 (Abonyo et al., 2004).
Figure 10.3 An electron micrograph of tubular myelin structures from large aggregates of lung
surfactant isolated from rat lung lavage. Scale bar: 500 nm.
3. Structure and properties of pulmonary surfactant

Once secreted into the air spaces on the surface of the alveoli, the lamellar
bodies are transformed into a more complex structure referred to as tubular
myelin. The structure of tubular myelin is unique and is illustrated in Fig-
ure 10.3. Ultrastructural examination of large aggregates isolated from lung
lavage preparations indicate that lamellar bodies are interconnected with
tubular myelin structures characterized by fused tubes on a square pattern of
about 60 nm dimension. Extraction of the phospholipids from such structures
leaves the framework of SP-A protein, which is believed to interact with the
phospholipids via their carbohydrate recognition domains (Nag et al., 1999).
Furthermore, the surfactant-associated proteins, particularly SP-A, are known
to be required to induce tubular myelin formation (Larsson et al., 2003b),
although the structure of tubular myelin formed in model systems can be
modified by the hydrophobic surfactant-associated protein SP-B (Morrow et
al., 2004). Indeed there is evidence that SP-B alone is required for adsorption
of the surfactant to the water–air interface by reducing the bending energy of an
intermediate structure with a low radius of curvature (Schram et al., 2003).
When examined by small-angle X-ray diffraction methods the surfactant
forms remarkably regular repeating structures. This is shown in Figure 10.4.
(A)
Figure 10.4 (A) Synchrotron X-ray diffraction pattern recorded from large aggregates of rat lung
surfactant isolated from lavage preparations. (B) Relationship between the d-spacing (S) and scattering
intensity of the sample shown in (A). The inset shows that the diffraction maxima fit a straight-line linear
regression giving the expression y = 33.5x – 0.0926.
The synchrotron X-ray diffraction pattern in Figure 10.4A can be interpreted as

a multilamellar structure from the relationship between scattering intensity and
d-spacing (Figure 10.4B), which shows 7 orders of reflection of a repeat
spacing of 33.5 nm. The structure, as judged by X-ray diffraction spacings, is
stable on heating up to 50ºC.
The process of conversion of bilayer structures into the ultimate surfactant–
air interface is presently unknown. The conventional model has been a
monolayer of surfactant at an air–water interface, and the biophysical proper-
ties of lung surfactant and its constituents have been characterized in great
detail from monomolecular film studies. This model has been challenged
recently on the basis of ultramicroscopic studies with a model in which a
surfactant bilayer in continuum with a monolayer oriented at the air interface
forms the functional unit (Larsson et al., 1999).
4. Treatment of pulmonary surfactant deficiencies

Failure of lung function may result from acute injury to the lung or from
premature development of the foetus leading to neonatal respiratory distress
syndrome. The conditions have been recognized for more than 300 years but
only over the past 50 years has there been progress in treatment. Respiratory
distress syndrome is usually observed in premature infants born after less than
32 weeks’ gestation. The lung surfactant composition of infants at term is
indistinguishable from that of adults, but in premature infants the surfactant is
deficient in surfactant phospholipids as well as having markedly reduced levels
of SP-A and SP-B (Ballard et al., 2003). The most promising line of treatment
for these conditions is surfactant replacement therapy, and such formulations
are presently undergoing phase III trials in human patients (Haitsma et al.,
2004; Schultz & Kesecioglu, 2004). The strategies employed in producing
these formulations is to restore the normal composition of the pulmonary
surfactant as well as to tackle the processes that may inactivate the existing
surfactant.
References
Abonyo, B.O., Gou, D., Wang, P., Narasaraju, T., Wang, Z. & Liu, L. (2004) Biochemistry
43, 3499.
Araneo, B.A., Cebra, J.J., Beuth, J., Fuller, R., Heidt, P.J., Midvedt, T., Nord, C.E.,
Nieuwenhuis, P., Manson, W.L., Pulverer, G., Rusch, V.C., Tanaka, R., van der Waaij,
D., Walker, R.I. & Wells, C.L. (1996) Zentralbl. Bakteriol. 283, 431.
Ar’Rajab, A., Ahren, B., Rozga, J. & Bengmark, S. (1991) J. Surg. Res. 50, 212.
Ballard, P.L., Merrill, J.D., Godinez, R.I., Godinez, M.H., Truog, W.E. & Ballard, R.A.,
(2003) Am. J. Respir. Crit. Care Med. 168, 1123.
Batenburg, J.J. & Haagsman, H.P. (1998) Prog. Lipid Res. 37, 235.
Bernhard, W., Pynn, C.J., Jaworski, A., Rau, G.A., Hohlfeld, J.M., Freihorst, J., Poets, C.F.,
Stoll, D. & Postle, A.D. (2004) Am. J. Respir. Crit. Care Med. 170, 54.
Bouwstra, J.A., Gooris, G.S., van der Spek, J.A. & Bras, W. (1991) J. Invest. Dermatol. 97,
1005.
Bouwstra, J.A., Gooris, G.S., Bras, W. & Downing, D.T. (1995) J. Lipid Res. 36, 685.
Bouwstra, J.A., Gooris, G.S., Dubbelaar, F.E., Weerheim, A.M. & Ponec, M. (1998) J.
Investig. Dermatol. Symp. Proc. 3, 69.
Bouwstra, J.A., Gooris, G.S., Dubbelaar, F.E. & Ponec, M. (2002) J. Invest. Dermatol. 118,
606.
Bukholm, G., Tannaes, T., Nedenskov, P., Esbensen, Y., Grav, H.J., Hovig, T., Ariansen, S.
& Guldvog, I. (1997) Scand. J. Gastroenterol. 32, 445.
Cao, J., Hawkins, E., Brozinick, J., Liu, X., Zhang, H., Burn, P. & Shi, Y. (2004) J. Biol.
Chem. 279, 18878.
Chen, H., Mendelsohn, R., Rerek, M.E. & Moore, D.J. (2000) Biochim. Biophys. Acta 1468,
293.
Chen, Q., Sternby, B. & Nilsson, A. (1989) Biochim. Biophys. Acta 1004, 372.
Cockshutt, A.M., Weitz, J. & Possmayer, F. (1990) Biochemistry 29, 8424.
Cook, D.I., Van Lennep, E.W., Roberts, M.L. & Young, J.A. (1994) In: Physiology of the
Gastrointestinal Tract (Johnson, L.R., ed.), Raven Press, New York, USA, p.1061.
Corfield, A.P., Wagner, S.A., Clamp, J.R., Kriaris, M.S. & Hoskins, L.C. (1992) Infect.
Immun. 60, 3971.
Crouch, E. & Wright, J.R. (2001) Annu. Rev. Physiol. 63, 521.
de Jager, M.W., Gooris, G.S., Dolbnya, I.P., Bras, W., Ponec, M. & Bouwstra, J.A. (2004)
J. Lipid Res. 45, 923.
Demirbilek, S., Gurses, I., Sezgin, N., Karaman, A. & Gurbuz, N. (2004) J. Pediatr. Surg. 39,
57.
Duane, W.C., McHale, A.P. & Sievert, C.E. (1986) Am. J. Physiol. 250, G275.
Elias, P.M. (1983) J. Invest. Dermatol. 80(Suppl.), 44S.
Fabia, R., Ar’Rajab, A., Willen, R., Andersson, R., Ahren, B., Larsson, K. & Bengmark, S.
(1992) Digestion 53, 35.
Fenske, D.B., Thewalt, J.L., Bloom, M. & Kitson, N. (1994) Biophys. J. 67, 1562.
Foster, C.D., Zhang, P.X., Gonzales, L.W. & Guttentag, S.H. (2003) Am. J. Respir. Cell Mol.
Biol. 29, 259.
Fuller, R. (1991) Gut 32, 439.
Goerke, J. (1998) Biochim. Biophys. Acta 1408, 79.
Goggin, P.M., Marrero, J.M., Spychal, R.T., Jackson, P.A., Corbishley, C.M. & Northfield,
T.C. (1992) Gastroenterology 103, 1486.
Haitsma, J.J., Papadakos, P.J. & Lachmann, B. (2004) Curr. Opin. Crit. Care 10, 18.
Hills, B.A. (1990) Med. J. Aust. 153, 76.
Holm, B.A., Venkitaraman, A.R., Enhorning, G. & Notter, R.H. (1990) Chem. Phys. Lipids
52, 243.
Holmskov, U., Thiel, S. & Jensenius, J.C. (2003) Annu. Rev. Immunol. 21, 547.
Jensen, J.M., Folster-Holst, R., Baranowsky, A., Schunck, M., Winoto-Morbach, S., Neumann,
C., Schutze, S. & Proksch, E. (2004) J. Invest. Dermatol. 122, 1423.
Johansson, M.L., Molin, G., Jeppsson, B., Nobaek, S., Ahrne, S. & Bengmark, S. (1993) Appl.
Environ. Microbiol. 59, 15.
Kitson, N., Thewalt, J., Lafleur, M. & Bloom, M. (1994) Biochemistry 33, 6707.
Kuroki, Y. & Voelker, D.R. (1994) J. Biol. Chem. 269, 25943.
Larsson, M., Larsson, K., Andersson, S., Kakhar, J., Nylander, T., Ninham, B. & Wollmer,
P. (1999) J. Dispers. Sci. Technol. 20, 1.
Larsson, M., Larsson, K., Nylander, T. & Wollmer, P. (2003a) Eur. Biophys. J. 31, 633.
Larsson, M., van Iwaarden, J.F., Haitsma, J.J., Lachmann, B. & Wollmer, P. (2003b) Clin.
Physiol. Funct. Imaging 23, 199.
Lingwood, C.A., Law, H., Pellizzari, A., Sherman, P. & Drumm, B. (1989) Lancet 2, 238.
Lupton, J.R. (2004) J. Nutr. 134, 479.
McIntosh, T.J., Stewart, M.E. & Downing, D.T. (1996) Biochemistry 35, 3649.
McIntosh, T.J. (2003) Biophys. J. 85, 1675.
McIntosh, T.J. (2004) Chem. Phys. Lipids 130, 83.
Montagne, L., Piel, C. & Lalles, J.P. (2004) Nutr. Rev. 62, 105.
Moore, D.J., Rerek, M.E. & Mendelsohn, R. (1997) Biochem. Biophys. Res. Commun. 231,
797.
Moore, D.J. & Rerek, M.E. (2000) Acta Derm. Venereol. Suppl. (Stockh.) 208, 16.
Moreau, H., Bernadac, A., Tretout, N., Gargouri, Y., Ferrato, F. & Verger, R. (1990) Eur. J.
Cell Biol. 51, 165.
Morrow, M.R., Stewart, J., Taneva, S., Dico, A. & Keough, K.M. (2004) Eur. Biophys. J. 33,
285.
Nag, K., Munro, J.G., Hearn, S.A., Rasmusson, J., Petersen, N.O. & Possmayer, F. (1999)
J. Struct. Biol. 126, 1.
Neubert, R., Rettig, W., Wartewig, S., Wegener, M. & Wienhold, A. (1997) Chem. Phys.
Lipids 89, 3.
Ohta, N. & Hatta, I. (2002) Chem. Phys. Lipids 115, 93.
Orgeig, S. & Daniels, C.B. (2001) Comp. Biochem. Physiol. A Mol. Integr. Physiol. 129, 75.
Osada, M., Gniadecka, M. & Wulf, H.C. (2004) Exp. Dermatol. 13, 391.
Percot, A. & Lafleur, M. (2001) Biophys. J. 81, 2144.
Piknova, B., Schram, V. & Hall, S.B. (2002) Curr. Opin. Struct. Biol. 12, 487.
Ponec, M., Weerheim, A., Lankhorst, P. & Wertz, P. (2003) J. Invest. Dermatol. 120, 581.
Rau, G.A., Vieten, G., Haitsma, J.J., Freihorst, J., Poets, C., Ure, B.M. & Bernhard, W. (2004)
Am. J. Respir. Cell Mol. Biol. 30, 694.
Raudenkolb, S., Hubner, W., Rettig, W., Wartewig, S. & Neubert, R.H. (2003) Chem. Phys.
Lipids 123, 9.
Ridsdale, R. & Post, M. (2004) Am. J. Physiol. Lung Cell Mol. Physiol. 287, L743.
Robson, K.J., Stewart, M.E., Michelsen, S., Lazo, N.D. & Downing, D.T. (1994) J. Lipid Res.
35, 2060.
Rodriguez Capote, K., McCormack, F.X. & Possmayer, F. (2003) J. Biol. Chem. 278, 20461.
Rodriguez-Capote, K., Nag, K., Schurch, S. & Possmayer, F. (2001) Am. J. Physiol. Lung
Cell Mol. Physiol. 281, L231.
Rooney, S.A. (2001) Comp. Biochem. Physiol. A Mol. Integr. Physiol. 129, 233.
Schram, V., Anyan, W.R. & Hall, S.B. (2003) Biochim. Biophys. Acta 1616, 165.
Schram, V. & Hall, S.B. (2004) Biophys. J. 86, 3734.
Schultz, M.J. & Kesecioglu, J. (2004) Curr. Drug Targets 5, 445.
Sefton, J., Arnebrant, T. & Glantz, P.O. (1992) Acta Odontol. Scand. 50, 221.
Slomiany, B.L., Murty, V.L., Sarosiek, J., Piotrowski, J. & Slomiany, A. (1988) Biochem.
Biophys. Res. Commun. 151, 1046.
Slomiany, B.L., Murty, V.L., Mandel, I.D., Zalesna, G. & Slomiany, A. (1989) Arch. Oral
Biol. 34, 229.
Slomiany, B.L., Sengupta, S., Fekete, Z., Murty, V.L. & Slomiany, A. (1991) Gen.
Pharmacol. 22, 969.
Slomiany, B.L. & Slomiany, A. (2004) Biochem. Biophys. Res. Commun. 318, 247.
Stahlman, M.T., Gray, M.P., Falconieri, M.W., Whitsett, J.A. & Weaver, T.E. (2000) Lab.
Invest. 80, 395.
Stewart, M.E. & Downing, D.T. (1999) J. Lipid Res. 40, 1434.
Strompfova, V., Laukova, A. & Ouwehand, A.C. (2004) Folia Microbiol. (Praha) 49, 203.
Suerbaum, S. & Michetti, P. (2002) N. Engl. J. Med. 347, 1175.
Takagi, Y., Nakagawa, H., Matsuo, N., Nomura, T., Takizawa, M. & Imokawa, G. (2004)
J. Invest. Dermatol. 122, 722.
Uchida, Y., Hara, M., Nishio, H., Sidransky, E., Inoue, S., Otsuka, F., Suzuki, A., Elias, P.M.,
Holleran, W.M. & Hamanaka, S. (2000) J. Lipid Res. 41, 2071.
Vayrynen, O., Glumoff, V. & Hallman, M. (2004) Pediatr. Res. 55, 55.
Veldhuizen, R. & Possmayer, F. (2004) In: Subcellular Biochemistry. Membrane Dynamics
and Domains (Quinn, P.J., ed.), Springer, London, UK, p.359.
Venkitaraman, A.R., Hall, S.B. & Notter, R.H. (1990) Chem. Phys. Lipids 53, 157.
Walters, R.W., Jenq, R.R. & Hall, S.B. (2000) Biophys. J. 78, 257.
Wegener, M., Neubert, R., Rettig, W. & Wartewig, S. (1997) Chem. Phys. Lipids 88, 73.
Wertz, P.W., Miethke, M.C., Long, S.A., Strauss, J.S. & Downing, D.T. (1985) J. Invest.
Dermatol. 84, 410.
Wu, H., Kuzmenko, A., Wan, S., Schaffer, L., Weiss, A., Fisher, J.H., Kim, K.S. &
McCormack, F.X. (2003) J. Clin. Invest. 111, 1589.
Ziegler, C., Torchia, M., Grahame, G.R. & Ferguson, I.A. (1989) Perit. Dial. Int. 9, 47.
CHAPTER 11
Drug delivery
Most drugs are solids, which can be taken orally. When drugs are formulated
into pills from powders, lipids are used as solid lubricants. Lipids can also play
a direct role in improving drug efficacy, safety, and patient convenience,
thereby making it possible to introduce new and improved therapies. This has
stimulated the creative design and study of lipid self-assembly systems for
delivery through different portals of the body.
Drug substances are frequently amphiphilic and sometimes they have low
aqueous solubility. Lipid self-assembly structures present a wide spectrum of
different drug delivery systems in which amphiphilic and lipophilic drugs can
be solubilized. Lipid–water phases also have the obvious advantage of being
able to encapsulate water-soluble drugs by solubilization in aqueous compart-
ments. In this way sustained release can be achieved, as well as protection
against digestive enzymes. A further important aspect where lipid systems can
have a direct influence is permeability through biomembranes, such as the
surface lining of the gastrointestinal tract. Many lipid systems are colloidal in
nature, for which there are well-known pathways in the form of lymphoid
follicles, Peyer’s patches of the GALT (gut-associated lymphoid tissue), and
normal enterocytes. Our natural ‘delivery system’ operates partly by lipids in
synergistic interplay with various proteins and enzymes.
A wide variety of lipid self-assembly structures with specific structural and
functional attributes can be prepared by exploiting the self-assembly properties
unique to different lipids. These structures may therefore vary with regards to
both lipid composition and morphology, and include well-known colloidal
structures like micelles, emulsions, different types of liposomes, and more
sophisticated self-assembly structures such as the reverse liquid crystals and
their colloidal dispersions. The word lipid is generously used in this text and
also includes other surface-active agents (surfactants).
The spectrum of properties and functions represented by lipid self-assembly
objects appears almost unlimited, but is in practice restricted by cost, safety and
regulatory demands. In reality there is only a limited number of approved lipids
and surfactants with prior use in pharmaceutical products. Therefore, introduc-
tions of new excipients should greatly stimulate the design and use of novel
drug delivery systems based on self-assembled lipids, and lead to new and
improved therapies in the future.
Many drugs on the market and in different development stages have
suboptimal properties which potentially could be improved by the use of a
suitable delivery system; see Table 11.1.
239
Table 11.1 Issues addressed by lipid-based drug delivery systems
• Low aqueous solubility

• Storage and in vivo stability
• Local irritancy
• Poor absorption
• Rapid clearance
• Dose-limiting toxicity
• Poor distribution to target organs
There are also a large number of promising candidates which are currently
unavailable because a suitable means of delivery is lacking. As more water-
insoluble drugs are discovered through the use of modern screening techniques,
the need for effective and flexible drug delivery systems will increase in both
preclinical and clinical development stages. The large number of emerging
biotech drugs also face delivery problems that potentially can be solved by
using lipid-based drug delivery systems. This text focuses on colloidal self-
assembly structures, as they are particularly useful in many practical
applications. For controlled release purposes, however, bulk-phase structures
are often preferred due to their improved retention ability.
A. Colloidal self-assembly structures

Colloidal sized lipid particle dispersions and their pre-concentrated dosage
forms have found important applications in a range of drug products and in all
important routes of administration. The properties of such structures can be
advantageously utilized to facilitate solubilization, stabilization and delivery
of different drug substances. Fat-soluble vitamins (A, D, and E) are usually
delivered as emulsions (including when delivered in their natural forms as
integral components of foods, which are often emulsions). Well-know self-
assembly structures used in pharmaceutical products include Taxol® (paclitaxel
micelles; Bristol-Myers Squibb), Sandimmune Neoral ® (cyclosporin
microemulsion; Novartis), Diprivan® (propofol emulsion; AstraZeneca) and
Doxil® (doxorubicin ‘Stealth’ liposome; ALZA).
More recently, nanoparticles of reverse lipid phases of cubic, hexagonal, and
disordered ‘sponge’ phases (Cubosome®, Hexosome®, and Flexosome®), have
emerged as attractive drug delivery vehicles in several applications. A broad
solubilization spectrum, high drug payloads, protection of sensitive substances
like peptides and proteins, and the ability to facilitate permeation make these
systems very interesting alternatives to the more commonly used micelle,
microemulsion, emulsion, and liposome systems. Similar to microemulsions,
emulsions and liposomes, non-lamellar phase particles can be designed to self-
disperse into colloidal dispersions, and can therefore be administered in the
DRUG DELIVERY 241
form of pre-concentrates. This property is essential in many applications where

water-free liquid or powdered pre-concentrates are the desired dosage forms.
The properties and functions of some key lipid-based systems are summarized
below.
1. Micelles, microemulsions, and emulsions

These are the simplest of the self-assembly structures formed by lipids and
surfactants, and are typically used in pharmaceutical products as solubilizers
and carriers of sparingly soluble compounds. Micelles and microemulsions
are thermodynamically stable with large interfacial area, whereas emulsions
and liposomes generally are kinetically stabilized by surfactants and
polymers.
Low solubility together with sluggish dissolution severely reduces the
bioavailability of many water-insoluble drugs. Ideally, an oral formulation
should enable uptake by keeping the drug in the dissolved state throughout its
transit through the gastrointestinal tract. Successful oral lipid formulations on
the market include Avodart® (dutasteride; Glaxo SmithKline), Kaletra®
(lopinavir/ritonavir; Abbott) and Fortovase® (saquinavir; Roche); see also
Strickley (2004).
These are examples of self-emulsifying drug delivery systems (SEDDS),
whose efficiency partly relates to their rapid in vivo fragmentation and distribu-
tion when contacted with intestinal fluids. A hallmark of a SEDDS is
self-emulsification to submicron particles when released into the lumen –
despite very gentle agitation. From a physicochemical point of view the process
is a phase inversion phenomenon and the composition of a SEDDS must
mediate very low interfacial tension between the oil and aqueous phases. The
product Sandimmune Neoral® (cyclosporin A; Novartis) uses this approach,
resulting in improved dose linearity, reduced variability, and improved absorp-
tion when taken in the absence of food (Kovarik et al., 1994; Mueller et al.,
1994a,b). In this respect it should be emphasized that SEDDS typically cannot
be utilized for parenteral drug delivery because of safety issues regarding the
excipients. This is partly due to the comparatively high activity of the excipients
used in SEDDS formulations. Micelles and o/w emulsions have been used
extensively in parenteral formulations and are in such cases composed of
surfactants or lipids with relatively low activity such as Polysorbate 80,
Cremophor EL, egg lecithin, and soybean oil. Examples of drugs intended for
the intravenous route utilizing micelles as solubilization aids include Taxol®
(paclitaxel; Bristol-Myers Squibb) and Sandimmune® (cyclosporin; Novartis),
which both take advantage of the excellent solubilizing ability of Cremophor EL
micelles (provided as a non-aqueous pre-concentrate containing ethyl alcohol).
Polysorbate 80 is a micelle-forming surfactant commonly used in protein
parenteral formulations to minimize interfacial denaturation of proteins, but is
also used for solubility management in, for instance, Taxotere® (docetaxel;
Aventis).
As early as 1869 it was reported that fat could be given subcutaneously to
malnourished patients (Menzel & Perco, 1869). Stabilized emulsion droplets
have been used as vehicles for parenteral nutrition for decades. The emulsion
is injected intravenously, and the particles must be smaller than ~1 μm in size
to avoid emboli (blockage) of fine capillaries. In the circulation these particles
adsorb lipoproteins and become similar to chylomicrons (Zilversmit, 1971).
Lipophilic drugs such as diazepam can be solubilized in such oil droplets and
delivered this way (von Dardel et al., 1976). The most well-known parenteral
emulsion product on the market is probably Diprivan® (propofol; AstraZeneca),
which is presented as a ready-made o/w emulsion based on soybean oil and egg
lecithin as stabilizer. Water-in-oil emulsions have been used for intramuscular
injection in order to form depots for the controlled release of anti-cancer drugs
(Hashida et al., 1977). As early as 30 years ago, multiple emulsions (w/o/w)
were described for the delivery of hydrophilic anti-cancer drugs such as
methotrexate (Elson et al., 1970).
Drugs for transdermal administration can be dissolved in the oil or water
phases of an emulsion. The o/w emulsions have better cosmetic properties,
whereas the w/o emulsions often provide better uptake. The reason is the so-
called occlusion effect: as the stratum corneum layer becomes more hydrated,
its barrier function is reduced. Emulsions are the most commonly used carriers
for topical drug delivery, and are used in numerous marketed products.
2. Lipid crystal dispersions

An alternative type of topical formulation providing a mechanically very
resistant surface layer is described as lipid crystal dispersions.
A dispersion of lipid crystals prepared by controlled crystallization from
monoacylglycerol liposomes results in leaf-shaped crystals with hydrophilic
surfaces (Larsson, 1968). Such crystalline dispersions consisting of a mixture
of monolaurin and monomyristin can immobilize 60–80% (w/w) water into a
homogeneous semi-solid mass. It gives a coherent film when spread onto the
skin, with a structure like that of a shingle roof. An electron micrograph is
shown in Figure 11.1. Spreading the dispersion onto the skin opens the crystals
so that the hydrophobic surfaces are exposed outwards. A first application was
the use of this dispersion as an ‘invisible glove’ for protection against mineral
oil exposure for metal industry workers (Larsson, 1975), with a significant
reduction in the incidence of oil-induced acne. Similar protection against
monomer penetration during work with acrylate polymers has also been
documented (Glantz et al., 1976).
This kind of lipid crystal dispersion can be used as a general ointment/cream
base. Dithranol is an efficient agent against psoriasis, but therapy has been
DRUG DELIVERY 243
Figure 11.1 Electron micrograph of hydrophilic monoacylglycerol crystals prepared as an aqueous

lipid crystal dispersion. The smallest fracture steps correspond to the lipid bilayer thickness (kindly
supplied by Professor A-M. Hermansson).
complicated by oxidative degradation and colouring of the skin. This hydro-

phobic substance can be incorporated into the lipid crystals (between the
non-polar sheets), and has resulted in the product Micanol™ in which these
problems are eliminated (Glantz et al., 1976) Another product based on this
concept is the antimicrobial Microcid™ (Glantz & Larsson, 1982). Its anti-
microbial effect is due to monolaurin itself and to hydrogen peroxide, which is
stabilized on the hydrophilic surfaces (Glantz & Larsson, 1982). It has now
become a standard treatment against impetigo in Sweden. The main reason is
that this therapy will not induce the development of antibiotic-resistant strains
of skin microorganisms, unlike therapies used earlier.
3. Liposomes
Drugs encapsulated in liposomes may circulate in the bloodstream for extended
periods compared to the same drugs in a non-liposomal form, thereby extend-
ing the treatment effect and simplifying the dosing regimen for physicians and
patients. In some cases, liposomal drugs have shown improved accumulation at
the tumour or infection site, thus delivering higher concentrations of the drug
to the disease target. The most important application of liposomes is in
infectious disease and cancer therapy (Lasic, 1993).
Liposomes represent dispersed, closed, bilayer assemblies of the lamellar

phase in excess water. Biologically active molecules, such as proteins, can be
incorporated into bilayers or the aqueous interior of the liposomes, thereby
creating primitive mimetics of biological cells. More important in this context
is that loading and encapsulation of drug substances in the interior aqueous
core or in the bilayer domain is possible. Liposome characteristics depend on
manufacturing protocol and excipient composition. They can be made in the
size range from unilamellar vesicles as small as 20 nm to multilamellar
structures 10 μm in diameter.
Among common lipid drug carrier structures, liposomes are particularly
adapted for the delivery of hydrophilic drugs. This is due to the presence of an
internal aqueous milieu encapsulated by an outer ‘impermeable’ bilayer struc-
ture, which acts as a chemical, mechanical, and electrical barrier against the
ambient medium. Due to the low permeability of protons and other cations they
can be prepared with acidic internal pH, which facilitates the loading of water-
soluble drugs (weak bases) and can be used to enhance drug stability. Indeed,
it is very difficult to achieve sufficient encapsulation without using active
loading principles, as this seldom leads to encapsulating efficiencies above
50% and requirements are often much higher (>90%). The other key issue in
drug development which often determines the success of the liposome formu-
lation approach is stability – both chemical and physical.
Liposomes have been studied as drug carriers since the 1970s, but it took
more than two decades to develop the first successful product applications.
Current liposomal drugs on the market and in the late development phase are
typically intravenous products featuring water-soluble and/or highly toxic or
tissue-irritant active substances, like lipid-soluble amphotericin B (AmBisome®;
Gilead Sciences Inc.) and water-soluble doxorubicin (Doxil®; ALZA). The
liposome carrier can reduce the harmful effects of such substances on healthy
tissues, thereby offering the potential for an improved safety profile. Inside the
liposome the substance is inactive. Liposomes alter the biodistribution of
amphotericin B so that nephrotoxicity is practically eliminated because of
avoidance of accumulation in the kidney, which can only filter in molecules
smaller than about 70 kDa. In addition to improved solubilization and targeting
of the reticuloendothelial system (RES) this is an example of the evasion mode
of liposome action.
Avoidance of extravasation decreases the distribution volume and can
increase the circulation half-life. The plasma half-life can be greatly enhanced
by tethering stabilizing polymer chains at the liposome surface, using e.g.
PEG-lipids whereby so-called Stealth® liposomes are formed (Allen &
Hansen, 1991; Klibanov et al., 1990; Lasic et al., 1991). Indirectly this leads to
preferential distribution towards tissues with enhanced vascular permeability
and thereby to passive targeting of the leaky vasculature associated
with tumours and diseased tissues, thereby increasing drug concentration at
DRUG DELIVERY 245
these sites (Ishida et al., 2002). The first drug using the PEG-stabilized
liposome approach is Doxil®, an anti-cancer drug for the treatment of refractory
ovarian cancer and AIDS-related Kaposi’s sarcoma. In retrospect it is some-
what surprising that the well-known steric stabilization approach was not
explored at an earlier stage in the pharmaceutical development of liposome
products.
Aside from passive targeting, liposome structures can be equipped with
‘homing devices’ such as surface-tethered antibodies. However, such struc-
tures become increasingly complex. It is noteworthy that even ‘simple’ things
are difficult in pharmaceutical development, and that such complex systems
will require substantial potential rewards in terms of therapeutic index and
market potential to even be considered for drug development.
Utilization of liposomes in administration routes other than intravenous is so
far limited because of the complexity of liposome formulations and the
existence of better and simpler alternatives. However, there are a number of
liposomal drug delivery systems on the market intended for the subcutaneous
or intramuscular routes. One example of a depot technology based on liposomes
is DepoFoam® from SkyePharma (Mantripragada, 2002). This technology
exploits foam structures based on liposome aggregates that are 10–30 μm in
size, and has been shown to provide a sustained release of some drugs for up to
1 month.
The technology is used in the marketed product Depocyt® (cytarabine;
SkyePharma) for the treatment of neoplastic meningitis by intrathecal delivery,
and is amenable to clinical product development. The formulation, which is
administered once every 2 weeks, has overcome the limitations of current
therapies for the disease; these require 2–3 doses per week by means of lumbar
puncture, and the rapid clearance of the conventional formulations results in
inadequate distribution. The controlled-release formulation has significantly
improved the pharmacokinetic profile of the active drug substance.
Another current interesting clinical development concerns transdermal
liposome technology (Cevc & Blume, 1992). Such systems are in clinical
development, e.g. Transfersomes® with ketoprofen developed by IDEA, claim-
ing a significantly improved bioavailability.
4. Nanoparticles of non-lamellar phases: cubic, hexagonal, and ‘sponge’

phases
As has been discussed earlier, lipids can self-assemble into various elaborate
non-lamellar geometries. From a drug delivery perspective these have attrac-
tive structural attributes and functional properties. Examples of lipid-based
liquid crystalline nanoparticles (LCNP) are presented in Chapter 6.
Because of the coexisting hydrophilic and lipophilic domains and the very
large surface area possessed by these structures, they have a broad spectrum of
applicability to different drug substances, encompassing small molecules as

well as peptides and proteins. The primary issues addressed by the non-lamellar
liquid crystalline phases and their corresponding nanoparticles are solubiliza-
tion, encapsulation, and protection of sensitive drugs in vivo against rapid
degradation by endogenous enzymes. Especially noteworthy is the
nanostructured interior of the particles, featuring both hydrophilic (aqueous)
and lipophilic (lipid) domains. In contrast to liposomes, which consist mostly
of water even for relatively small liposome radii, the lipid content of typical
LCNP is high (normally in the range of 50–70%).
In this context it can be pointed out that normal o/w emulsions are best
adapted for strongly lipophilic drugs that require an oil medium for optimal
solubilization, whereas amphiphilic drugs – and in particular peptides and
proteins – are not usually well formulated in emulsions. Also, drug payloads
can be increased considerably using a LCNP drug delivery system rather than
liposomes.
Important recent developments allow for sub-micron sized non-lamellar
dispersions to be formed spontaneously by self-dispersion or self-
emulsification from a water-free formulation pre-concentrate similar to the
SEDDS microemulsions. Liquid crystalline nanoparticle dispersions can be
produced simply by gentle magnetic stirring instead of requiring high-energy
dispersion techniques. High drug payload characteristics for a broad spectrum
of drugs, together with in situ emulsification, imply that such systems are
optimal for oral applications where the self-emulsifying property is desired
for convenience and efficient uptake. Such systems have shown promise in
the oral delivery of peptides and proteins, such as insulin (Chung et al.,
2002).
Early non-lamellar particle preparations featured lipids that were not well-
suited for parenteral drug delivery, as exemplified by the unsaturated
monoacylglycerol preparation GMO. Poorly reliable and poorly scalable manu-
facturing schemes further limited the scope of application of the early
non-lamellar particle preparations. Later improvements concerning both manu-
facturing schemes and the excipient base have now finally facilitated the use of
non-lamellar phases and nanoparticles in parenteral drug delivery. Using
biodegradable and well-tolerated lipid excipients and reproducible, reliable
and scalable manufacturing schemes, the final obstacles for the use of non-
lamellar drug delivery systems in parenteral applications have now been
overcome. Non-lamellar particles are typically covered by surface-tethered
polyethlylene glycol chains, and are thus partly protected from discovery by the
RES.
Other promising delivery routes for non-lamellar dispersions are the nasal
and transdermal routes. The Eurocine™ L3 adjuvant is a sprayable nasal
vaccine based on a low-viscosity sponge phase. Composed of endogenous and
pharmaceutically accepted lipids it is approved for human use, and a successful
DRUG DELIVERY 247
phase I trial of its use for nasal immunization with diphtheria vaccine has
recently been performed (Haile et al., 2004).
An example of a colloidal self-assembly structure not belonging to the above
mentioned classes of phase structures is the ‘ribbon-like’ amphotericin B
Abelcet™ formulation. This is another example that clearly illustrates the kind
of benefits that can be achieved by incorporating drugs into novel lipid self-
assembly objects.
B. Liquid crystals
Drugs often have an optimum therapeutic concentration range, above which
they are toxic and below which they are essentially ineffective. In formulations
where there is no controlled release function, the in vivo concentration of the
drug can frequently build to a level which is supra-optimum and then rapidly
decrease to a sub-optimum level. The objective of a sustained release formula-
tion is to deliver the drug within the therapeutic concentration range for a
prolonged period of time. One way to obtain sustained release is to incorporate
the drug in a porous matrix, such as those characteristic of liquid crystalline
materials. It has been recognized for several decades that lyotropic liquid
crystalline phases formed by lipids and surfactants can provide suitable matri-
ces for the sustained release of different kinds of drugs (Engström et al., 1992).
Developments regarding the use of liquid crystalline phases in drug delivery
applications are summarized in two recent reviews (Drummond & Fong, 1999;
Shah et al., 2001).
The release profile of a drug from the matrix is partly determined by drug-related
factors such as the diffusion constant, solubility, and partition coefficient, and
partly by factors related to the delivery system, such as porosity, tortuosity, and
pore connectivity. For liquid crystalline systems the release rate can be regulated
by tuning of the phase structure. It is for instance intuitive that the release of an
aqueous-soluble drug will be much slower in a reverse micellar cubic phase (I2),
where aqueous domains are separated by lipid bilayers, than in a bicontinuous
cubic phase (Q2), in which water channels form a continuous network.
In cases where the encapsulation is effective, drug release becomes deter-
mined by the rate of biodegradation of the liquid crystalline matrix, which in
the case of lipids often involves lipases.
The first marketed product using the special features of liquid crystals for
controlled release purposes is Elyzol® dental gel (Colgate Oral Pharmaceuti-
cals). The formulation is based on a mixture of monoacylglycerols and
triacylglycerols presented in the form of a water-free suspension that trans-
forms to a controlled release matrix of reversed hexagonal phase when injected
into the periodontal pocket (Norling et al., 1992). The phase transition is
triggered by the absorption of water, which also results in the bioadhesive
properties of the lipid matrix.
This is one example illustrating the potential of using liquid crystalline

phases as depots in pharmaceutical applications. Several preclinical and clini-
cal investigations are currently ongoing in which liquid crystal systems that
form in situ are exploited for the controlled release of various drug substances,
including small molecules and peptides. These feature both injectable and
topical routes of administration.
The manufacture of liquid crystalline depot systems involves just a few
standard pharmaceutical processing steps. Products can be provided in con-
venient ready-to-inject non-aqueous precursor form, without the need for
complicated reconstitution and mixing steps. Further advantages of such a
system include the relatively low viscosity of the injectable formulations,
allowing the use of thin needles, and a generally high solubilizing capacity
(with drug payloads of up to 30%), allowing smaller injection volumes.
Proteins and peptides are stabilized in the nanostructured lipid membrane
environment and can be protected from degradation by endogenous enzymes,
thus providing improved bioavailability.
Liquid crystalline formulations can provide in vivo sustained release for
small molecules and peptide drugs over long periods of time. Results from in
vivo preclinical pharmacokinetic studies show that the drug peptide octreotide
(a somatostatin analogue) is released at a relatively constant rate in vivo from
a liquid crystal matrix during at least a one-month period without signs of a
rapid initial release ‘burst’ subsequent to injection (Thuresson et al., 2005).
C. Summary outlook
The past decades have witnessed increased momentum in the research and
development of lipid self-assembly objects as delivery vehicles for various
types of therapeutic agents. These can facilitate the solubilization, stabilization,
controlled release, and transport of a wide range of drug substances that would
otherwise pose problems. The majority of lipid-based drug delivery products
are micellar, microemulsion, and emulsion pre-concentrates or stable disper-
sions, used primarily to address solubility issues. Several liposomal products
have emerged on the market in recent years which typically improve the
therapeutic index (decrease toxicity) by decreasing local exposure to free drug
and facilitating more favourable biodistribution profiles. Non-lamellar particle
structures appear from many aspects ideal for a range of drug delivery applica-
tions. As key challenges related to manufacturing, stability, and reproducibility
have now finally been overcome, these systems provide new opportunities for
optimizing drug product performance. Non-lamellar lipid-based liquid crystal-
line nanoparticles (LCNP) are currently being implemented as a promising
approach in a number of products under development. The need for simple
controlled release technologies enabling, for instance, less frequent injections
is another issue where lipid-based systems can add substantial value, in
DRUG DELIVERY 249
particular liquid crystalline systems in which the nanostructure can be tuned

according to particular drug properties and applications.
References
Allen, T.M. & Hansen, C. (1991) Biochim. Biophys. Acta 1068, 133–141.
Cevc, G. & Blume, G. (1992) Biochim. Biophys. Acta 1104, 226–232.
Chung, H., Kim, J., Um, J.Y., Kwon, I.C. & Jeong, S.Y. (2002) Diabetologia 45, 448–451.
Drummond, C.J. & Fong, C. (1999) Curr. Opin. Colloid Interface Sci. 4, 449–456.
Elson, L.A., Mitchley, B.C.V., Collings, A.J. & Schneider, R. (1970) Eur. J. Clin. Biochem.
15, 87.
Engström, S., Larsson, K. & Lindman, B. (1992) US Patent US 5 753 259.
Glantz, P. & Larsson, K. (1982) J. Dispersion Sci. Technol. 3, 573.
Glantz, P., Larsson, K. & Nyquist, G. (1976) Odont. Rev. 27, 265.
Haile, M. et al. (2004) Vaccine 22, 1498–1508.
Hashida, M., Takahashi, Y., Muranishi, S. & Sezaki, H. (1977) J. Pharmacokin. Biopharm.
5, 225.
Ishida, T., Harashima, H. & Kiwada, H. (2002) Biosci. Rep. 22, 197–224.
Klibanov, A.L., Maruyama, K., Torchilin, V.P. & Huang, L. (1990) FEBS Lett. 268, 235–237.
Kovarik, J.M., Mueller, E.A., van Bree, J.B., Tetzloff, W. & Kutz, K. (1994) J. Pharm. Sci.
83(3), 444–446.
Larsson, K. (1968) In: Surface Active Lipids in Foods, Society of Chemical Industry, London,
UK, p.8.
Larsson, K., Björnberg, A. & Hellgren, L. (1975) Opusc. Med. 4, 162.
Lasic, D.D. (1993) Liposomes: From Physics to Applications, Elsevier Science Publishers,
New York, USA.
Lasic, D.D., Martin, F.J., Gabizon, A., Huang, S.K. & Papahadjopoulos, D. (1991) Biochim.
Biophys. Acta 1070, 187–192.
Mantripragada, S. (2002) Prog. Lipid Res. 41, 392–406.
Menzel, A. & Perco, H. (1869) Wien. Med. Wschr. 19, 517.
Mueller, E.A. et al. (1994a) Pharmaceutical 11, 151–155.
Mueller, E.A. et al. (1994b) Pharmaceutical 11, 301–304.
Norling, T. et al. (1992) J. Clin. Periodontol. 19, 687–692.
Shah, J.C., Sadhale, Y. & Chilukuri, D.M. (2001) Adv. Drug Delivery Rev. 47, 229–250.
Strickley, R.G. (2004) Pharm. Res. 21, 201–230.
Thuresson, K., Tiberg, F., Johansson, M., Harwigsson, I., Joabsson, F. & Johnsson, J. (2005)
Patent Application PCT GB05/002217.
von Dardel, O., Mebius, C. & Mossberg, T. (1976) Acta Anaesthesiologica Scand. 20(3),
221–224.
Zilversmit, D.B. (1971) J. Biol. Chem. 249, 2645.
CHAPTER 12
Foods
We will start by considering a typical food product: sausage. The structure and
biophysical properties of the muscle cell and the fat cell, the biological tissues
from which sausage is prepared, have been characterized in detail. In the
transformation of these raw materials into a food, the biological structure is
destroyed, membrane-bound enzymes are released, and small molecules (re-
lated to flavour) are redistributed between different compartments. We are,
then, far from a molecular understanding of the roles of different components
in the product. Only on the colloidal level can the new structure be described.
It is an emulsion of fat and a dispersion of fat cells and connective tissue in a
continuous aqueous gel phase, which is formed by the homogenized muscle
proteins.
This example is given as an illustration of the knowledge gap between
biology and degraded biological tissues, which are recombined into food
products. It also demonstrates that the fundamental description of foods on the
colloidal level requires knowledge of the structure of interfaces and the forces
between colloidal structural elements. Lipids are usually the most surface-
active components of a food system, and they are therefore of great importance
in food processing.
As most of the applications of lipid functionality discussed below are closely
linked to the fundamental properties described in Chapters 1–8 with references,
only a list of recommended reading will be given at the end of this chapter.
A. Starch–lipid interaction
Starch consists of two glucose polymers: amylose, which is linear with α-(1,4)
links; and amylopectin, which contains the same linear segments in a branched
structure. The branches are attached by an α-(1,6) glucoside bond, and the
structure is indicated in Figure 12.1. Amylopectin is considered to be one of the
largest molecules in nature. A common ratio of amylose to amylopectin is 1:3.
Cereals, tubers, and many other plants store energy as starch granules. These
are concentric structures with alternating semi-crystalline layers and amor-
phous layers – a remarkable structure, particularly as amylopectin molecules
mainly form the ordered layers, whereas most of the amylose is localized in the
disordered layers.
One of the most important reactions in food processing is starch gelatiniza-
tion. Less than 1% of potato starch in water, for example, is enough to give a
251
Figure 12.1 Illustration of the amylopectin branch structure and the location of the three types of
chains (A, B, and C) (after Eliasson et al., 1987).
strong gel structure by heating through the gelatinization temperature, about

60ºC. This involves the following steps:
• amylose molecules from the amorphous regions start to leak out;
• water penetrates the surface layer and swells the granules;
• the crystallinity of the amylopectin layers is destroyed by water
solubilization;
• swollen granules attach to one another by entanglements involving
mainly the outer A-chains (see Figure 12.1) of amylopectin at the
surface.
The process is cooperative with overlap between these steps, due to variations
in granule crystallinity.
If amylose in water solution is exposed to a monoacyl lipid such as a
monoacylglycerol, an inclusion complex with crystalline structure separates.
Its structure is shown in Figure 12.2. By the helical twist of the amylose
molecule, a hydrophobic core is formed which fits the diameter of the hydro-
FOODS 253
Figure 12.2 Proposed structure of the amylose–monoacylglycerol complex. The amylose backbone is
shown with the monoacylglycerol acyl chain indicated by the hydrogen atoms (after Carlsson et al.,
1979).
carbon chain in an extended conformation. The polar head group is too large to
be included.
Amylopectin molecules do not interact as strongly with lipids as amylose.
The outer A-chains (see Figure 12.1), however, can also form a similar
inclusion complex. The effects of lipids on starch in foods can be explained by
these two mechanisms.
The presence of monoacylglycerol molecules at the surface of starch gran-
ules, when gelatinization starts, will result in precipitation of amylose–lipid
complex on the surface. This will hinder the diffusion of amylose molecules out
from the granule. The gel volume will be smaller as water transport inwards is
also reduced. Furthermore the gelatinization temperature may shift upwards
due to the barrier effect of the complex on the surface.
One application of amylose–lipid complex formation is to reduce the sticki-
ness of industrial potato powder products and pasta products. Amylose
molecules in the water phase outside the swollen granules will result in a sticky
consistency.
An important application of the amylopectin complexing reaction with lipids
is to increase the shelf-life of bread due to so-called antistaling. Bread becomes

hard during storage not due to drying but due to crystallization of starch –
staling. Again the most commonly used lipid type is monoacylglycerols. Their
presence at starch gelatinization during heating in the oven will result in
complex formation with the outer A-chains of the amylopectin molecule. This
in turn will hinder the close packing required at recrystallization within the
starch granules, which is the staling mechanism. In order to supply lipid
monomers efficiently, the Lα phase of the monoacylglycerol has proved to be
best.
Starch is not fully gelatinized in bread due to the water-poor environment. If
starch is recrystallized during storage of a food with fully gelatinized starch,
water separation can take place. This recrystallization (retrogradation) process
can be delayed by the amylopectin–lipid complexing reaction. Pre-manufac-
tured desserts are examples of such products.
It can finally be mentioned that very polar lipids with one hydrocarbon chain,
which are water-soluble in micellar form, also form inclusion complexes with
amylose and amylopectin. An example is lysophospholipids. The correspond-
ing amylose complex does not precipitate as a solid, however, and the functional
properties are different.
B. Milk-based lipid products and their analogues

By tradition the dairy field has been the most advanced branch of food
processing and has pioneered food research in general.
1. Milk
Milk secreted from the mammary glands is a complex biological fluid contain-
ing growth factors, antibodies and enzymes with significance far beyond
nutrition. Here we will only focus on bovine milk as a colloidal system,
involving the transformations into various foods. Enzymes are important in
these changes. The proteolytic enzymes, for example, are responsible for the
degradation of caseins to form the cheese matrix, and lipase activity contributes
to the development of flavour.
The milk fat globules, which amount to about 4% (w/w) of milk, have a broad
size distribution with a predominating diameter range of 1–10 μm. The milk fat
globule membrane is illustrated in Figure 12.3.
When the fat globule is secreted, it brings the cell membrane from the gland
as a coat with an important stabilizing function. A protein-rich layer dominated
by xanthine oxidase is located between the globule surface and the membrane
bilayer. The triacylglycerols consist of ‘ordinary’ fatty acids (mainly oleic acid
and palmitic acid) provided via the blood, and short-chain fatty acids (such as
butyric acid) formed in the mammary gland. The short acyl chains are located
FOODS 255
Figure 12.3 Schematic illustration of the milk fat globule membrane. Phospholipid (with polar head
groups denoted by open circles) and cholesterol molecules (with polar groups shown by filled circles) at
the globule surface and in the outside bilayer are indicated. The aqueous protein layer is indicated with
PL.
in the sn-3 position, and the corresponding crystals tend to form triple chain-
length structures. Crystallization of milk fat is extremely complicated, and
even after prolonged storage all three types of crystal forms (α, β′ and β) can
coexist (see Chapter 2).
Milk proteins are dominated by caseins, consisting of α-, β- and κ-casein
molecules. These molecules are extremely amphiphilic, and are associated into
micelles in the milk. When milk is homogenized in order to obtain a uniform
globule size and avoid creaming, with a resultant tenfold increase in the surface
area between globules and milk serum, caseins will act as emulsifier and fill
empty space at the globule surface.
Consumer milk is standardized with regard to fat content, and is therefore
reconstituted after separation of the cream from the milk. It then has to be
pasteurized or sterilized, at about 80ºC or 125ºC, respectively.
2. Whipping of cream, butter and ice cream
Whipped cream
A typical composition after centrifugation into whipping cream is a fat content
of 30% (w/w). Whipping of cream involves the introduction of so many air
cells that we approach a foam structure. The air cells are stabilized by a two-
dimensional matrix of aggregated fat globules at the air–water interface.
Therefore a critical amount of fat globules is needed. Furthermore, fat crystals
must be present in the fat globules in order to hinder coalescence.
The probable structures at the three phase interfaces are illustrated in
Figure 12.4. When air cells are introduced, the fat globules will attach to their
Figure 12.4 Illustration of a probable interfacial structure at the initial stage of the whipping of cream.
surface in order to reduce the surface energy. Polar lipids from the surface will
spread along the interface, and the resulting partly ‘naked’ fat globules will
tend to adsorb at the interface with the exposed hydrophobic regions turned
towards air. Aggregation of globules at the interface will also be driven by the
exposed triacylglycerol core.
Butter
If the whipping is continued after the foam-like structure has been formed, the
new air cells will remove more of the globule coating, and aggregation of the
globules in three dimensions will start. This will result in a phase inversion into
a w/o emulsion. The emulsion oil continuum coexisting with an outside
buttermilk phase will spontaneously take up about 20% (w/w) of water. The
proposed interfacial extraction of the milk fat globule membrane on whipping
is confirmed by the presence of the membrane lipids in buttermilk.
The fat crystal network is an important quality factor of the butter, its
spreading properties, etc. Therefore, certain tempering regimes have been
developed for optimal crystal growth during the whipping process, termed
churning. Usually the temperature of the cream is first lowered to about 8ºC to
give numerous crystal nuclei. It is then raised to about 19ºC in order to favour
growth of the stable polymorphic forms over that of the unstable ones. The
temperature is then lowered to about 10ºC in order to obtain more crystals, and
churning is started.
An alternative to churning is to concentrate the cream by centrifugation to
the fat composition of butter. The concentration of fat globules is then above
that corresponding to close-packed spheres, which means that the globules
have to adopt a polyhedral shape. This emulsion is therefore very fragile, and
a minor mechanical disturbance is enough to cause phase inversion.
The advantage with this method is that the phase inversion can be initiated
when the desired composition of the butter is reached. Legislation in most
countries requires an exact composition of butter, which by traditional churn-
ing is not straightforward to achieve. Furthermore, the production of buttermilk
as a by-product is avoided. On the other hand, the presence of fat globule
membranes in the fat phase complicates the formation of a crystal network.
FOODS 257
Ice cream
Traditional ice cream is prepared from whipped cream, and represents an
interesting system with features in common with those of butter. About 10%
(w/w) of sugar and flavourings, like vanilla, are dissolved in the cream before
whipping. Whipping (or the actual agitation) has to be performed under
cooling. The crucial step is that the water phase must be transformed into ice
just before phase inversion of the fat globules takes place. The structure of the
fat globules on the verge of coalescence is then immobilized by the ice
crystals.
The typical sensory characteristics of ice cream are partly caused by the
cooling sensation that accompanies the melting process and partly by the
mouth-feel involved in coalescence of the fat globules during melting of the fat
and the release of flavour compounds.
Ice cream should have air cells to about half its volume. The ice crystals must
be less than about 30 μm in diameter in order to avoid a ‘sandy’ mouth-feel.
Hydrocolloids such as cellulose derivatives are often added in order to reduce
crystal growth during storage. Even at the low temperature at which ice cream
is stored, all the water is not frozen but can contribute to crystal growth.
Ice cream can also be prepared from a vegetable oil mix, which is first
emulsified into an artificial cream. Then the same processing conditions as
described above have to be fulfilled, including the presence of fat crystals. Milk
proteins are usually used for emulsification. A complication is that caseins are
such efficient emulsifiers that coalescence of the fat globules in the mouth is
inhibited. This problem can be solved by the addition of a small amount of
monoacylglycerols to the oil. These will displace enough protein molecules for
the coalescence process to take place. The mechanism behind this was de-
scribed in Chapter 8.
3. Drying of milk
It is often practical to store milk as a dry powder. When the surrounding water
is replaced by air, there is a tendency for the globule surface layer to rupture.
The addition of polar lipids and carbohydrates has proven to be an efficient way
of stabilizing the surface structure during drying so that the emulsion is
reconstituted on subsequent exposure to water. The polar lipid should supply
an Lα phase which can emulsify any oil phase that separates during the drying
process. The carbohydrate will accumulate at the globule surface during
drying. The polar surface is therefore stabilized and can persist even when
water is removed. Disaccharides and maltodextrins are usually used.
The same mechanism is used by certain microorganisms as protection
against drying damage. Trehalose is known to be secreted during the drying of
fungi, for example, and coats their surface.
4. Hydrogenation of oils – hardened fats

The melting point of a fat is related to the average lengths of its hydrocarbon
chains and their degree of unsaturation. Tristearin, for example, melts at about
70ºC, and when one double bond is introduced to give triolein, melting takes
place below room temperature. If, therefore, some double bonds in the fat are
eliminated by hydrogenation, the melting point can be adjusted upwards to a
desired value. In order to illustrate this, we can consider the production of a
rapeseed oil margarine. The proportion of crystalline fat is adjusted by blend-
ing the rapeseed oil with the same oil hydrogenated to a melting point of usually
41ºC. The final melting point will be lowered due to the eutectic effect of the
non-hydrogenated oil.
The hydrogenation (‘hardening’) reaction, like all chemical reactions, pro-
ceeds in both directions. Therefore in the reverse reaction, when a cis double
bond is saturated, an equilibrium of cis and trans double bonds is produced
until this double bond is fully saturated. The trans double bond has only a minor
effect on the melting point compared to the saturated bond.
There has been concern about physiological effects of trans double bonds in
cell membranes. In order to avoid their production, the oil can first be fully
hydrogenated so that there are no double bonds. It can then be interesterified
with an oil with cis double bonds, so that the melting point is reduced to the
desired value.
5. Artificial milk/cream, margarine and low-fat spread products

In tropical regions with a lack of milk, artificial milk is produced from
vegetable fats. If whippability is desired, the fat must be partly crystalline at
room temperature. As an emulsifier, skim milk powder is often used.
Stearoyl-lactyl-lactate has also proven to be a good emulsifier, probably
because it is ionic and provides electrostatic repulsion forces between the fat
globules.
Napoleon III is considered to have initiated the preparation of a butter
substitute, which was called margarine after one of the inventors. Margarine is
produced by feeding a mixture of an oil phase and a water phase, usually
acidified skim milk, into a scraped-surface heat exchanger. Crystallization of
part of the oil/fat phase is the stabilizing mechanism of this w/o emulsion; thus
the water droplets are immobilized by the plastic fat. An emulsifier is also
added (at a fraction of a percent) to the oil phase in order to obtain water
droplets of uniform size. Usually ‘lecithin’ obtained from soybean oil refine-
ment is used. Traditionally the water phase amounts to 20% (w/w), to give the
same composition as in butter.
The ideal polymorphic form for providing a crystal network with good
rheological properties is the β′ form. This poses a dilemma, as transformation
FOODS 259
to the stable β form may take place during storage. A fat composition favouring
stability of the β′ form is therefore required. Retardation of the phase transition
is favoured by:
• crystals containing fatty acids with varying chain lengths;
• co-crystallization of a lipid with a strong tendency to form stable crystals
with orthorhombic chain packing.
When margarines are used for frying, it is important to avoid spattering due
to the presence of large water droplets. Special emulsifiers with anti-spattering
effects are therefore used.
Margarines for use as spreads on bread are usually prepared from mixtures
of butter fat and vegetable fat in order to provide the taste of butter and good
spreading properties at refrigerator temperature. A typical spread has 75% milk
fat and 25% soybean oil or rapeseed oil as the fat phase, with a water phase
similar to that of butter. Margarines with a lower fat content have increased on
the market. Such spread products are also a convenient form for the incorpora-
tion of special nutrition-enhanced lipids, as so-called functional foods. One
example is a spread product containing plant sterols able to reduce blood
cholesterol levels.
In order to incorporate 40–60% (w/w) of a water phase, special emulsifiers
have to be used. The water droplets in an ordinary margarine are about 5 μm in
diameter. Larger water droplets are obtained when the water content is in-
creased, and then there is a risk of microorganism growth. Distilled
monoacylglycerol preparations are therefore used to reduce water droplet size.
Furthermore there is a need to increase the viscosity of the water phase; gelatin
is often used for this purpose.
C. Cocoa butter and chocolate

The cucumber-like fruits of the cacao tree were first used in Central America
and Mexico. Because of their high value, they were even used for payment, as
reflected by their original Latin name Amygdala pecuniara. The tree was later
named Theobroma cacao by Linnaeus.
Cocoa is the name used for the fermented beans. After roasting, the cocoa
butter is separated from the cocoa powder. This powder is dominated by
proteins and contains a residual cocoa butter content of about 10–20% (w/w).
Chocolate consists of cocoa butter as a continuous phase with dispersed
particles of sugar and cocoa powder. The fat-free part of the cocoa powder
gives the characteristic flavour. Phospholipids are used as a dispersion agent.
The sharp melting behaviour is a unique property of cocoa butter compared to
other known natural fats. It is almost completely solid at room temperature and
fully melted at the temperature in the mouth. The characteristic brittle consist-
ency and glossy appearance of cocoa butter are also required in chocolate. The
fat crystallization behaviour is therefore of utmost importance. This was partly

considered in Chapter 2.
Cocoa fat is an extremely homogeneous triacylglycerol mixture, with oleic
acid in the 2-position and stearic or palmitic acid in the 1- and 3-positions. This
regularity induces a tendency to form triple chain-length crystal structures as
well as the less stable double chain-length structures. The polymorphic behav-
iour is therefore complicated, with the existence of six crystal forms. The most
stable is a β form.
At crystallization it is important to obtain the most stable crystal form in the
product. Otherwise a phase transition may take place, resulting in a greyish and
rough surface – so-called blooming. It is therefore necessary to apply a
tempering process in order to first obtain nuclei of the most stable crystal form,
and then let these grow at the expense of less stable polymorphs.
In the production of dark chocolate or milk chocolate, the components are
mixed and the particle sizes reduced to less than about 35 μm in diameter. The
fluid chocolate mass is then ‘conched’, a process named after the machines that
are used. Conching involves exposure of the mass to air by agitation at about
60ºC for several hours, in order to reduce off-flavours.
The function of the phospholipids, usually soybean lecithin at a level of
about 0.5%, is to stabilize the dispersion of solids with hydrophilic surfaces by
the formation of monolayers of lecithin at the cocoa powder and sugar surfaces.
As cocoa butter is expensive, substitutes have been developed, and these are
classified as:
• cocoa butter equivalents;
• cocoa butter replacers;
• cocoa butter substitutes.
The triacylglycerol preparations classified as cocoa butter equivalents consist
of similar triacylglycerols to those found in cocoa butter, but produced by
enzymatic synthesis. The lower-quality cocoa butter replacers and cocoa butter
substitutes consist of lauric acid-rich triacylglycerol mixtures like coconut oil,
or hardened fats from, for example, rapeseed.
D. Vegetable oils, frying oils and emulsion products
1. Frying oils
Foods can be cooked by heating in water or in oil, and the difference in surface
temperature and chemical environment results in quite different food charac-
teristics, as we all know. A problem with the repeated re-use of frying oils is the
oxidative changes that occur, and the risk of formation of toxic degradation
products. Saturated oils will of course have better stability, but they have
obvious disadvantages from a nutritional point of view.
FOODS 261
Quality control of frying oils is based on measurements of the free fatty acids
content, the foaming properties of the oil, its viscosity, or its colour. An
interesting aspect of frying oil functionality is the presence of polar lipids,
which will influence heat diffusion at the surface and thus the heat transfer
efficiency. It has been shown that the amount of frying oil needed can be
reduced by half if phospholipids are added to the oil. A disadvantage, however,
is the risk of dissociation of the polar group, with the release of amines with
undesirable aroma effects.
2. Mayonnaise and dressing products

Mayonnaise is an o/w emulsion with egg yolk used as emulsifier. Vinegar
provides antimicrobial protection by lowering the pH to about 4. A remarkable
feature of this emulsion is its stability at such a high volume fraction of oil.
According to a US standard, for example, the volume fraction of oil must be at
least 65%. Closest-packing arrangements of spheres (cubic or hexagonal close-
packing) have a volume fraction of about 74%.
The reason for the high stability of mayonnaise is the emulsion mechanism
we learned about in Chapter 8. The egg yolk phospholipids provide an Lα type
of lipid bilayer protection to the emulsified droplets. It is easy to see the
corresponding birefringent shell around the droplets under a polarizing micro-
scope.
Dressing products are also often based on egg yolk as emulsifier and can then
be regarded as diluted mayonnaise. The same oil droplet size persists (about
5 μm diameter) and the water content is usually about 70%. Egg yolk as
emulsifier is sometimes exchanged for lipids; one reason is the risk of patho-
genic microorganisms (Salmonella in particular) in eggs. Even if a mild heat
treatment is used to pasteurize the egg while preserving its functional proper-
ties, some uncertainty may remain with regard to the possible presence of
temperature-resistant strains. Plant phospholipids are an alternative emulsifier
to egg yolk; caseins and caseinates are also used.
E. Lipids in bread and other baked products
1. Interaction of wheat flour with water

Wheat flour is the ground wheat kernel endosperm. It consists of about 65%
(w/w) starch, 12% gluten proteins, 14% water, 2% lipids, and the rest is
minerals and other proteins and carbohydrates. We have already discussed
starch–water interactions. It is well known that the unique baking performance
of wheat compared to other cereals is due to the gluten proteins. In contact with
excess moisture they take up a certain amount of water and form the viscoelastic
gluten gel. The main proteins in this gel (gliadins and glutenins) are extremely
Figure 12.5 Phase diagram of wheat lipids plus water (after Eliasson & Larsson, 1993). The different
phases are here termed: LC-L, denoting the lamellar liquid-crystalline phase; LC-H, the inverse
hexagonal phase; and L, a liquid phase without contact with the water or lipid corners, and therefore
probably an L3 phase.
hydrophobic, and their amphiphilic properties are the basis for their behaviour
in bread-making.
The lipid fraction also interacts with water, and is in fact a part of the gluten
gel (about 10%). The functionality of the lipids in baking will be summarized
here. In order to understand the effects of lipids it is convenient to start from a
phase diagram, as shown in Figure 12.5. It was constructed as a ternary system,
although there are thousands of different lipid species present. On chromato-
graphic fractionation of the total lipids, there is one type of fraction that does
not form aqueous phases. These non-polar lipids form one natural corner of the
phase diagram. The remaining fractions, dominated by galactolipids and
phospholipids, interact with water. They are therefore, according to our defini-
tion in Chapter 1, polar lipids, and together they are used here as the second
lipid corner of the phase diagram.
The dotted line on Figure 12.5 corresponds to the composition of the lipids
in wheat flour. If we follow this line, we will reach the water content where the
Lα phase, non-polar oil and a liquid phase (probably L3) are in equilibrium. The
partition of water between the gluten gel and the embedded lipid phases is not
known, but there is evidence indicating that the phases in this region contribute
significantly to the bread-making process.
FOODS 263
Figure 12.6 Illustration of the lamellae between gas pores in bread dough during expansion in the oven
(after Eliasson & Larsson, 1993). Lipid monolayers spread and stabilize the surface on drying and
contraction of the gluten gel phase.
2. The baking process

Ordinary white bread is prepared by mixing wheat flour, water and yeast in the
proportions 100:65:2 (some salt is usually also added). The mixing process
results in incorporation of air cells, which act as nuclei for the carbon dioxide
produced when yeast ferments starch. The gas cells are stabilized by glutenin
molecules.
When the dough is heated in the oven, there is a fast gas expansion into a pore
structure, and then the wheat lipids play an important role. The structure at the
gas–pore interface is indicated in Figure 12.6.
On heating, the gluten gel loses its cohesiveness and contracts. The dispersed
lipids, considered ideally to exist as an Lα phase, then take over the interface
(part of the evidence for this is that wheat with poor baking performance can be
improved by the addition of lipids to the dough, which gives an Lα phase).
Fixation of the bread structure takes place by gelatinization of starch. The
importance of lipids as antistaling additives in bread was discussed in Section A
above.
It should finally be mentioned that new continuous processes for industrial
dough mixing have been developed, which require fat crystals to be present.
Fully hardened fats used as the fat crystals should survive the gelatinization
temperature of starch. These crystals must be less than about 30 μm in diameter

in order not to be detected in the mouth. It seems as if the fat crystals are needed
in this rapid process in order to stabilize the foam lamellae during oven spring.
3. Cakes and biscuits/cookies

Cakes are baked from batters containing wheat flour, egg, sugar, fat and a gas-
forming agent. The mechanism behind the structural transformation in the
oven, however, is quite different to that described above for bread-making. Air
cells are introduced when the sugar is mixed into the flour–fat–egg paste (some
of the sugar is solved and some flour is dispersed into the water phase from the
egg). The first changes on heating are induced by melting of the fat (usually
butter or margarine). The air cells then move over from the fat phase to the
water phase. During expansion of the gas cells by the production of carbon
dioxide on further heating, a phase inversion into an o/w emulsion takes place.
The gas cells fuse into a pore system, and structure fixation then takes place due
to gelatinization of the starch. When the final temperature is reached, Maillard
reactions and caramelization at the surface will produce aroma compounds,
which are partly solved in the fat film covering the pore system.
Cookies (biscuits) are baked from high-fat batters; typical weight propor-
tions are 100:150:50 of wheat flour, butter (or margarine), and sugar. Similar
phase transformations take place on heating as described above when gas cells
move over to the aqueous phase, although the water phase here has a smaller
volume than in cakes.
References
Carlsson, T., Larsson, K., Dinh-Nguyen, N. & Krog, N. (1979) Starch 31, 222.
Eliasson, A-C., Larsson, K., Andersson, S., Hyde, S.T., Nesper, R. & Von Schnering, H-G.
(1987) Starch 39, 147.
Recommended further reading

Eliasson, A.-C. & Larsson, K. (1993) Cereals in Breadmaking. A Molecular/Colloidal
Approach, Marcel Dekker, New York, USA.
Sjöblom, J., Friberg, S., & Larsson, K., eds (2003) Food Emulsions (4th edn), Marcel Dekker,
New York, USA.
Walstra, P. & Jenners, R. (1984) Dairy Chemistry and Physics, J. Wiley & Sons, New York,
USA.
Index
Air–water interface 113 fatty acids 13

alpha (α) crystals 18 in foods 251
alveolar surface 89 seeding 54
amylopectin 252 shear fields 53
amylose–monoacylglycerol complex 253 triacylglycerols 28
aqueous lipid dispersions 129 ultrasound stimulation 52
artificial milk 258 cubic cell membranes 142
average curvature 77 cubic lipid nanoparticles 135, 138
curvature 77
Beta (β) crystals 18 curvature homogeneity 86
beta prime (β′) crystals 18
bicontinuous structures 77 Dental gel 247
biological membranes 183 digestion of fats 99
biscuits 264 domain structures 116
blooming of fat 260 dressings 261
blue phases 103 drug delivery 239
Bonnet transformation 86 dynamic swelling 123
bread
baking process 263 Elastic energy 85
gas pores 263 egg yolk phosphatidylcholine (PC) 102
wheat 261 emulsion crystallization 54
wheat lipids 261 emulsions
bubbles 123 coalescence 181
butanol–lipid–water system 102 crystalline-interface-stabilized 178
butter 255 flocculation 181
buttermilk 256 lipid-stabilized 177
oil-in-water (o/w) 175
Cakes 264 protein-stabilized 176
chain packing subcells 15 water-in-oil (w/o) 175
chocolate 259 environment–body interface 219
cholesterol 2 equilibrium spreading pressure (ESP) 117
cholesterol esters 103 ethanol–lipid–water system 102
churning 256
cis and trans bonds 258 Fat blooming 260
cocoa butter 259 fat crystallization 47, 54
colloidal self-assembly 240 fat digestion 99
competitive monolayer spreading 117 fat gels 62
consumer milk 255 foam interfacial phase 126
cookies 264 foam stability 125
cream 255 foams 123
cream base 242 food crystals 257
critical micellar concentration 83 frying oils 260
crystal size 257
crystallization Galactolipids 4
diacylglycerols 26 gas pore stabilization 263
emulsions 54 gastric mucous surface 224
fats 47, 54 gastrointestinal lactobacilli 226
265
gauche conformation 73 low-fat spreads 258

Gaussian curvature 77 lung lamellar bodies 231
gel phase 61, 74 lung surfactant 228
glove, invisible 242 lung surfactant extract 122
Hardened fats 258 Maltodextrins 257

heterogeneous nucleation 55 Marangoni effect 125
hexagonal lipid nanoparticles 138 mayonnaise 261
homogeneous nucleation 55 membrane lipids and proteins 162, 183
honeycomb shape 85 milk
hydration force 84 artificial 258
hydrogenation 258 buttermilk 256
hydrophobic force 125 consumer 255
hydrostatic pressure 81
Nucleation
Ice cream 257 heterogeneous 55
intermediate phases 87 homogeneous 55
intestinal mucosa 226
invisible glove 242 Oleic acid 26
L2 phase 109 Phosphatidylcholine (PC)

L3 nanoparticles 139 egg yolk 102
L3 phase 111 soybean 102
lamellar bodies 141 phospholipids 4
Langmuir–Blodgett multilayers 121 plant sterols 259
linoleic acid 26 plastic fats 258
linolenic acid 26 Plateau border 125
lipid absorption 227 polymorphism
lipid(s) diacylglycerols 26
classification 1 fats 47
definition 1 fatty acids 13
galactolipids 4 triacylglycerols 28
non-polar 2 polypeptide–lipid interaction 145
phospholipids 4 pressure–area isotherms 114
polar 3 propylene glycol–lipid–water system 103
raw materials 1 protein foam 127
sphingolipids 5 protein–lipid interaction 145
lipid-anchored proteins 165 protein monolayers 147
lipid crystal dispersions 242 protein penetration 148
lipids and proteins in membranes 162, 183 protein structure 145
lipid–water systems
monoacylglycerols 89 Respiratory distress syndrome 234
phospholipids 90 ripple phase 74
ternary systems 99
lipolytic enzymes 159 Saliva 223
lipoproteins 167 saponin 99
liposomal drugs 244 saturated fatty acids 14
liposome encapsulation 243 sausage 251
liposomes 129, 132 seeding effect in crystallization 53
liquid-crystalline drug carriers 247 self-dispersion 246
liquid crystals serum proteins 168
cubic 76 shear field crystallization 53
hexagonal 75 skin
lamellar 73 lipid composition 220
thermotropic 73 stratum corneum 220
liquid state 107 solid-state transformation
INDEX 267
enantiotropic 12 trehalose 257

monotropic 12 triacylglycerols 100
soybean phosphatidylcholine (PC) 102 tubular myelin 232
sphingolipids 96
sponge nanoparticles 140 Ultrasound-stimulated crystallization 52
starch gelatinization 252 unsaturated fatty acids 18
starch–lipid interaction 251
‘stealth’ nanoparticles 246 Vesicles 129, 132
sterols 2
stigmasterol 2 Whipped cream 255
surfactant lipids 229
surfactant proteins 229 X-ray diffraction
long-spacings 7
Ternary phase diagrams 99 short-spacings 7
thermotropic liquid crystals 73 small-angle (SAXS) 83
trans/gauche ratio 1, 73 subcell parameters (from WAX) 15

(The Oily Press Lipid Library) Kare Larsson, Peter Quinn, Kiyotaka Sato, Fredrik Tiberg - Lipids - Structure, Physical Properties and Functionality-Woodhead Publishing (2006)

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

(The Oily Press Lipid Library) Kare Larsson, Peter Quinn, Kiyotaka Sato, Fredrik Tiberg - Lipids - Structure, Physical Properties and Functionality-Woodhead Publishing (2006)

Uploaded by

Copyright:

Available Formats

Lipids: Structure, physical

properties and functionality

Lipids: Structure, physical

Oxford Cambridge Philadelphia New Delhi

Published in association with

Woodhead Publishing, 1518 Walnut Street, Suite 1100, Philadelphia,

First published by The Oily Press, 2006

© PJ Barnes & Associates, 2006; © Woodhead Publishing Limited, 2012

Trademark notice: Product or corporate names may be trademarks or registered trade-

British Library Cataloguing in Publication Data

ISBN 978-0-9531949-9-5 (print)

This book is Volume 19 in The Oily Press Lipid Library

Typeset by Ann Buchan (Typesetters), Middlesex, UK

membranes and their constituents, as evidenced by the publication of ten

2 Solid-state behaviour of polymorphic fats and fatty acids 9

3 Liquid-crystalline lipid–water phases 73

4. Membrane protein crystallization in cubic phases

4 The liquid state 107

5 Lipids at the air–water interface – monolayers and multilayers

8. A surface phase formation at an air–water interface of a lung

6 Dispersions of lipid–water phases 129

7 Interaction of lipids with proteins and polypeptides 145

9 Lipids of biological membranes 183

10 Lipid barriers at the environment–body interface 219

1. Composition of pulmonary surfactant

11 Drug delivery 239

General information on the chemistry of lipid molecules can be found in The

Figure 1.1 Chemical formulae of phosphatidylcholine (PC), phosphatidylethanolamine (PE) and

phosphatidylinositol (PI), all shown in Figure 1.1. Other phospholipids that

In industrial applications of lipid functionality (see also Section B below),

B. Structure, physical properties and functional properties

a functional additive, and it is therefore necessary to discuss the functionality

C. Methods for structure characterization

been particularly important in order to follow the dynamics of formation of

Lipids are a class of compounds that contain long-chain aliphatic hydrocarbons

Molecules Primary Macroscopic

particles of lipid crystals comprise the microscopic features, whereas the

A. Polymorphism: structures and transformations

stability cannot be obtained by observing the solid-state transformation in

interesting features are revealed in the polymorphic crystallization and trans-

Table 2.1 Polymorphism in saturated fatty acids

nc, number of hydrocarbon atoms. m, monoclinic; t, triclinic.

chain. It also introduces multiplicity in the possible molecular conformations

Saturated fatty acids

undergo torsional motion with an aliphatic gauche conformation, making the H

Figure 2.6 Structure relationships of a polytypic–polymorphic transformation of stearic acid.

solubility measurements (Sato et al., 1988). This result is explained by the

Fatty acid Form Local conformation Subcell

Oleic acid γ S-C-S′ O′ ⏐⏐

Erucic acid γ S-C-S′ O′ ⏐⏐

forms of stearic acid were also observed in a solution-mediated transformation

Unsaturated fatty acids

skew-cis-skew ′ skew-cis-skew trans-cis-trans

Figure 2.7 Olefinic conformations of a cis double bond.

configurations and odd–even positions. The same behaviour was observed in

review (Kaneko et al., 1998). We discuss polymorphism of oleic acid,

Figure 2.9 Schematic illustration of a γ–α transformation.

polyunsaturated fatty acids such as linoleic acid (C 18:2, cis-ω6,9) and

double bond make this polymorph so unique. The olefinic conformation is

Figure 2.11 Crystal structure of the β1 form of oleic acid.

Table 2.3 Subcell parameters of oleic acid β1 forma

Parameter Subcell 1 Subcell 2

Table 2.4 Crystal structure data of two polymorphs of petroselinic acid