New comprehensive biochemistry vol 01 membrane structure

© Elsevier/North- Holland Biomedical Press, 1981
All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or
transmitted, in any form or by any means, electronic, mechanical, photocopying, recording or otherwise
without the prior permission of the copyright owner.

ISBN for the series: 0444 80303 3
ISBN for the volume: 0444 80304 I

Published by:
Elsevier/North-Holland Biomedical Press
335, Jan van Galenstraat, P.O. Box 211
Amsterdam, The Netherlands
Sole distributors for the U.S.A. and Canada:
Elsevier/North-Holland Inc.
52 Vanderbilt Avenue
New York, NY loon

Library of Congress Cataloging in Publication Data

Main entry under title:

New comprehensive biochemistry.
InclUdes bibliographies and indexes.
Contents: v , 1. Membrane structure -- [etc.]
1. Biological chemistry. I. Finean, J. B.
II. Michell, R. H. [DNLM: 1. Membranes--Anatomy and
histology. Wl NE372F v.l / QJ3 532.5 .M3 M534]
Q.D4l5.N48
574.19'2
81-3090
ISBN 0-444-80303-3 (Elsevier/North-Holland : set)

AACR2

Printed in The Netherlands


Membrane structure
Editors

J.B. FINEAN and R.H. MICHELL
Birmingham

1981

ELSEVIER/NORTH-HOLLAND BIOMEDICAL PRESS
AMSTERDAM· NEW YORK· OXFORD


New Comprehensive Biochemistry
Volume I
General Editors

A. NEUBERGER
London

L.L.M. van DEENEN
Utrecht

ELSEVIER/NORTH-HOLLAND BIOMEDICAL PRESS
AMSTERDAM· NEW YORK. . OXFORD

v

Preface
In the former series of Comprehensive Biochemistry the contributions of membranes
to cellular biochemistry were considered in a volume entitled Cytochemistry (1964) in
which the organelles of the cell were considered individually. Since that time the
study of membranes has formed one of the most rapidly expanding fields of biology,
and this volume is devoted to a consideration of only one aspect of this progress,
namely our current understanding of the relationship between membrane structure
and function. Other aspects of membrane biochemistry will be discussed in forthcoming volumes on Phospholipids and on Membrane Transport. One of the outstanding features of recent research on membrane structure has been a transition from the
marked polarisation of views that characterised the 1960s towards a general agreement during the 1970s that all membranes share one basic form of structural
organisation. The aims of this volume are to identify general features of membrane
structure, to discuss in considerable detail some selected aspects that have been
studied intensively in recent years, and to relate some of this molecular information
to individual membrane functions.
We anticipate that most of our readers will already have a general knowledge of
cell structure and of the roles of individual membranes and organelles in particular
cell functions. For those who lack this background information, we would recommend reference to brief monographs such as Membranes and their Cellular Functions
(J.B. Finean, R. Coleman and R.H. Michell, 2nd ed., 1978, Blackwell, Oxford), The
Biochemistry of Cell Organelles, (R.A. Reid and R.M. Leech, 1980, Blackie, Glasgow
and London) and Biological Membranes (R. Harrison and G.G. Lunt, 2nd ed., 1980,
Blackie, Glasgow and London).
J.B. Finean
R.H. Michell
Birmingham, August 1980


CHAPTER I

Isolation, composition and general structure
of membranes
J.B. FINEAN and R.H. MICHELL
Department of Biochemistry, University of Birmingham
P.o. Box 363, Birmingham B15 2TT, us:

1. Historical introduction
Awareness of the existence of a discrete plasma membrane at the surface of cells
gradually emerged as cell biologists of the late nineteenth century observed a variety
of plant cells and single cell organisms and probed their cell boundaries using both
physical and chemical techniques [1-3]. From his studies of the permeability of cells
to a variety of non-electrolytes, Overton [3] was even able to speculate on the lipid
nature of the permeability barrier.
The first significant chemical study of a membrane was not reported until 1925,
when Gorter and Grendel [4] extracted lipid from erythrocytes and spread it as a
monolayer at an air-water interface in order to compare the area that it might
potentially cover with the total surface area of the original erythrocytes. A fortuitous
mutual cancellation of experimental errors allowed the correct conclusion that there
was sufficient lipid to form a lipid bilayet over all or almost all of the surface of the
cell. The probability that the lipid of biological membranes exists predominantly in
bilayer form was later reinforced by physical measurements (optical and electrical)
made both on biological membranes and on isolated lipid (mainly phospholipid)
systems [5]. This has since remained the .dominant theme in considerations of
membrane structure.
The initial suggestion that protein would probably be closely associated with lipid
in plasma membranes (and maybe also in other membranes) was again a speculative
one based on surface tension measurements and on the spontaneous association of
water-soluble proteins with monolayers of lipid spread at an air-water interface.
Although there was no relevant information on the protein components of membranes, Danielli and Davson proposed a general structural scheme [6] for cell
membranes which featured a bilayer of lipid coated at its aqueous interfaces with

layers of protein. Their first suggestion that protein might penetrate into or through
the lipid layer [7] was not based on any direct knowledge of membrane proteins, but
was simply a speculative attempt to account for the occurrence of facilitated
permeation of solutes through plasma membranes.
Early thoughts on membrane structure were confined to the plasma membrane
Finean r Michell [eds.] Membrane structure
© Elsevier/North-Holland Biomedical Press, 1981


2

P.B. Finean and R.B. Michell

Fig. 1. Electron micrographs of liver cells (hepatocytes) isolated by the procedure of Seglen [37]. (A) Lead
citrate-stained section of cells fixed with 1% OS04 and 1% tannic acid. X51000. (B) Freeze-fracture
replica of unfixed cell preparation. X57000.

[8,9]: the more extensive elaboration of membrane-bounded compartments within
the cell was not recognised until the 1950s when improvements in the preparation of
thin sections of tissues for examination by electron microscopy indicated a similar


Isolation, composition and general structure

approximately

3

IO~m

Fig.2. Diagrammatic illustration of variety of organelles in plant and animal cells as revealed by electron
microscopy (from [44]).

general form for both plasma membrane and the membranes of cytoplasmic
organelles (Figs. I and 2). This emphasised the limitations of the Danielli and
Davson membrane model [6] in accounting in structural terms for a much greater
range of functions, and hence inspired the proposal of alternative structural arrange-


4

Fig. 3. Diagram illustrating the chronological order in which the most influential models have been
proposed.


Isolation, composition and general structure

5

ments (Fig.3). In particular, it was realised that: (a) under appropriate conditions
some lipids would adopt configurations other than a bilayer; (b) the fine detail of
membrane structure as seen at high magnification in some electron micrographs
appeared granular; and (c) membranes were dissociated into lipoprotein "particles"
by detergent treatment. This encouraged speculation, especially by biochemists, that
membranes might consist of laterally aggregated arrays of globular lipoprotein
"subunits" (e.g. [10-12,14]).
It has only been as a result of the relatively recent progress in characterisation of
membrane proteins that substantial agreement on a general model of membrane
structure has been reached. In particular, the identification of membrane proteins in
which substantial exposed regions are dominated by non-polar amino acid side

chains led to the realisation that such regions would be likely to associate with
hydrocarbon regions of the membrane lipid phase; parts of these proteins might
therefore be inserted deep into the membrane interior. This, together with an
emphasis on disordered or fluid packing of the lipid hydrocarbon chains and on free
lateral diffusion of membrane components, was then featured in a new membrane
model, the "fluid mosaic" model proposed by Singer and Nicholson [17] in 1972.
This has since been generally accepted as a more realistic expression of the general
characteristics of membranes than any previous model. It may well be the last of the
generalisable membrane models, because experimental work on membranes has now
advanced to the stage at which the structural patterns of individual membranes are
being defined in some detail [18]. As a result, we now know that individual
membranes differ both in the spatial distributions of their molecular components
and in the mobilities of these components.

2. Isolation of membranes
Some studies of membrane structure can be made using membranes still organised
into cells; such studies include microscopical examination of membrane organisation
in cells and of membrane-cytoskeletal interactions, X-ray diffraction analysis of cells
which possess ordered membrane arrays, some types of measurement of the mobilities of membrane components, and labelling experiments designed to probe the
asymmetric orientations of surface membrane components. However, purified membrane preparations are needed for most studies of the chemical composition or
spatial organization of membranes.
(a) Criteria for assessing purity

For studies of chemical composition, the chief criterion is that membrane preparations should be pure samples of a single type of membrane [19] but studies of
membrane structure also demand that samples of isolated membranes should
preserve the spatial interrelationships between different molecules that prevail in the
intact, healthy cell. These constraints upon the purity of membrane preparations


6

F.B. Finean and R.B. Michell

used for structural studies are often much more stringent than the requirements to
be met by membrane preparations in which the attribute of interest is some
organelle-specific function (e.g. an enzyme activity) that can be adequately studied
even when the membrane exhibiting it exists only as a component of a membrane
mixture. For membranes which contribute substantially to the total membrane
complement of cells, achievement of homogeneity requires a purification of only a
few-fold, and appropriate techniques may not be unduly complex or difficult to
devise. Many membranes, however, constitute only a very small proportion of the
total mass of the parent cell, and in such cases very substantial purification
(sometimes 50- to 100-fold, or even more) is required to yield a small amount of
homogeneous material for analysis.
The monitoring of membrane purification basically consists of following the
purification of the required membrane by monitoring some membrane-specific
criterion, associated with the simultaneous measurement of a variety of additional
criteria specific for all of the possible contaminant structures. Occasionally the
morphology of a particular membrane structure remains sufficiently distinctive, even
after homogenisation, for electron microscopy and/or phase contrast microscopy to
provide a reliable guide to purification (e.g. mitochondria, rough endoplasmic
reticulum, intestinal epithelial brush borders, secretory vesicles), but much more
often the isolated membrane fragments do not retain a morphology that is sufficiently characteristic for their unequivocal identification (e.g. smooth membrane
fragments may come from, among others, smooth endoplasmic reticulum, plasma
membrane or Golgi complex). In most cases, therefore, the progress of the required
membrane and of contaminants through a fractionation procedure is followed by the
assay of a variety of membrane-specific or "marker" criteria; these are usually
enzyme activities known to be confined to particular membranes in the cell under
study (see, for example, [19] and [20], section 1 of [21], Chapters 1-4 of [22]). A
membrane preparation should only be adjudged "pure"; (a) when the purification

achieved corresponds to that which would be estimated from consideration of the
morphology of the parent cell, and (b) when the concentrations of all known
contaminating membranes, as assessed by the activities of their characteristic marker
enzymes, have been reduced to levelswhere it can confidently be calculated that they
contribute very little of the mass of the isolated membrane preparation.
In going from an homogenate to an isolated subcellular fraction, such enrichment
or depletion in terms of particular membranes is usually expressed in terms of
Relative SpecificActivities (RSAs) of the chosen marker enzymes, these RSAs being
the ratios which compare the specific activities in the final fraction(s) to the specific
activities in the initial homogenates [23]. In interpreting RSAs, it is essential to
remember that the mass contributed by a particular stucture to an isolated fraction
is a function both of the experimentally determined RSA and of the contribution of
the particular organelle to the mass of the parent cell. To illustrate this, consider a
simplified cell with only two membrane systems, a plasma membrane that contains
1% of the cell protein and mitochondria which contain 20%. From this cell one
isolates an SO-fold purified plasma membrane fraction in which the RSA of the


Isolation, composition and general structure

7

mitochondrial marker enzyme remains 1.0, as in the original homogenate; 20% of the
material in this substantially "purified" fraction is contributed by mitochondria. A
second fraction from the same cells has a mitochondrial marker RSA of 4.75, but is
also enriched 5-fold with respect to the plasma membrane marker: reference to the
composition of the original cell shows, however, that 95% of the material in this
"contaminated" sample is derived from mitochondria.
(b) The choice of isolation media and of the starting material

In designing a subcellular fractionation scheme with which to isolate a particular
membrane, there are a number of technical obstacles to be negotiated. The parent
cells must be available in sufficient quantity and adequate purity, a method must be
devised for breaking the cells in an appropriate, usually osmotically protective,
medium, and physical techniques are required by which the desired membrane can
be isolated from the homogenates.
Within cells, membranes normally exist in an aqueous medium rich in small ions
and proteins. However, on dilution in an homogenate this high protein concentration is lost. In addition, few cell fractionations are undertaken in predominantly
ionic media since such media often cause aggregation of organelles and thus impair
the separation. The most common media for subcellular fractionation are physiologically iso-osmotic (approx. 300 mosM) or hyperosmotic solutions of non-permeant
neutral solutes such as sucrose or mannitol. A notable exception to this custom is
provided by skeletal muscle, where the polymerisation of actinomycin in low ionic
strength media means that an ionic medium is sometimes (e.g. [24]), though not
always [25], used for the isolation of Ca2+-pump-rich sarcoplasmic reticulum. In
addition, mammalian erythrocyte surface membranes, the most widely studied of all
membranes, are normally isolated in ionic media (either with or without a divalent
cation chelator such as EDTA or EGTA), but most of these diverse media are of
lower than physiological ionic strength and osmotic activity [26]. Most of the time,
the possible effects on membrane composition and structure of using nonphysiological and non-ionic media for membrane isolation are largely ignored, but
experience with the red cell suggests that such uncritical attitudes may ultimately
have to be abandoned. For example, erythrocyte ghosts made in media of physiological ionic strength and containing small concentrations of divalent cations [27](or
returned rapidly to physiological ionic strength after lysis at lower ionic strength
[26]) may be compared with ghosts isolated in almost ion-free media, often in the
presence of EDTA or EGTA (e.g, [28]). The former, especially after incubation at
37°C to "reseal" them, tend to be resilient spheres or even somewhat biconcave, they
are impermeable to most materials which do not permeate the intact cell, they
generate and sustain ion gradients, and they retain the "cytoskeletal" layer of
spectrin and actin at their inner surface [26,27,29]. The latter, by contrast, adopt
rather irregular shapes, are "floppy" and readily vesiculate, are deficient in spectrin
and actin, are permeable even to macromolecules, and may carry "extra" membraneassociated proteins that have become adsorbed at low ionic strengths (e.g.

8

F.B. Finean and R.B. Michell

haemoglobin and maybe also glyceraldehyde-3-phosphate dehydrogenase; see
Chapter 5) [26,28,30,31]. Such detailed information on the damaging effects of
transferring membranes into environments strikingly different from those prevailing
within cells appears to exist only for the erythrocyte plasma membrane, but it might
be anticipated that other membranes, especially other plasma membranes, might
behave similarly. Renewed attempts to devise effective subcellular fractionation
procedures with which to isolate membranes and organelles in media of physiological ionic strength and composition might well yield remarkably interesting, and
maybe disquieting, insights.
For details of appropriate isolation conditions for individual membranes, it is
usually necessary to consult primary journals: leads into these, and occasionally
technical details, can be found in reviews or compilations such as refs. 20-22, 32, 33,
and Section 1 of [34]. The starting material for subcellular fractionation of animal
cells can be a solid tissue, a population of free-living cells grown in tissue culture, or
a suspension of free-living cells from the body: examples of the latter include various
types of blood cells and various cell-types which either occur naturally or can be
grown in the peritoneal cavity (e.g. macrophages, mast cells, polymorphonuclear
leukocytes or free-living neoplastic cells such as Ehrlich ascites). Body fluids
normally contain mixed cell populations, so a preliminary to subcellular fractionation is usually the isolation of one cell type in homogeneous form: appropriate
techniques include differential and/or density gradient centrifugation, free flow
electrophoresis and differential adsorption onto some surface which differentiates
between cells as a result either of their intrinsic adhesiveness or their ability to bind
to some selective surface-specific ligand (e.g. a lectin or cell-directed immunoglobulin): see, for example, section VB of [35].
Most solid tissues are also heterogeneous, both due to the presence of blood
(which can be removed by perfusion) and to the presence of more than one intrinsic

cell population. Although this heterogeneity is often ignored, there has been a
marked tendency in recent years for individual cell populations to be isolated from
tissues before functional studies are undertaken. This has allowed, for example, the
properties of hepatoeytes [36,37] and of Kupffer cells [38] from mammalian liver to
be studied separately. Although potential disadvantages of such techniques include
the smaller amounts of starting material that are usually available and the possibility
that the tissue dissociating techniques may damage molecular components exposed
on the surface of the cells, it is to be hoped that this approach may soon be more
widely adopted when isolating membranes for structural studies. There are various
techniques for weakening the forces or structures (e.g. collagen fibrils) that hold cells
together prior to dissociation of tissues to form cell suspensions, of which the most
useful are treatments with either chelators such as EDTA (e.g. [39]) or collagenase
(e.g. [36,37]). With some tissues, it is already customary for isolation of a pure cell
suspension to precede subcellular fractionation (e.g. fractionation of adipocytes,
rather than heterogeneous adipose tissue [40].
When a cell to be fractionated possesses a substantial cell wall (e.g. bacteria, fungi
or higher plants) which may both impede its homogenisation and render purification


Isolation, composition and general structure

9

of plasma membranes remarkably difficult, then these walls can be either removed
or substantially weakened by prior digestion with enzymes.
(c) Separation of subcellular components

In general, the bulk separation of organelles and membrane fragments from cell
homogenates may exploit any physical differences between the various particles in
the homogenate. In practice, however, the great majority of separations are either by

differential rate centrifugation, distinguishing particles of different sizes, or by
isopycnic density-gradient centrifugation, with separation the result of differences in
particle densities. In recent years two other techniques have been developed that are
potentially of general applicability. The first is free-flow electrophoresis ([20], pp.
78-86) in which a suspension of mixed membranes is carried slowly and continuously down a vertical curtain of flowing buffer across which an elecrical field is
applied. At the bottom of the buffer curtain the various separated streams of
particles with differing charge characteristics flow into a row of tubes. Modem
equipment for free-flow electrophoresis can, in a few hours, separate either cells or
membranes of different charges in substantial quantities. The second relatively novel
technique, based on membrane surface characteristics, is the phase partition of a
membrane mixture in an aqueous two-phase polymer system such as 5.3% wIw
Dextran (M, 500000)/4.1 % polyethylene glycol (M, 6000)/°.1 M phosphate buffer,
pH 6.5 ([20], pp. 71-75). Both phase partition and free-flow electrophoresis have
already found applications in separating certain types of plasma membrane from
other structures [20]. Finally, there ate a number of specifically designed techniques
which take advantage of the singular characteristics of particular membranes: for
example, isolation of plasma membranes of phagocytic cells by retrieving the
phagocytised particles and their associated phagocytic vacuole membrane (e.g. [41]
and [20], pp. 88,89), binding of carbohydrate-rich surface membranes to bead-bound
lectins [42] or binding of antigen-bearing surface membranes to immobilised anti-cell
-surface immunoglobulins [43].
As noted above, membranes for structural studies often need to be freer of
contaminants than for many functional studies. For most membranes this calls for
the use of some hybrid fractionation procedure which adopts, in sequence, more
than one of the above techniques: since it readily accommodates the largest
quantities of material, differential rate centrifugation is almost invariably the first of
these sequential steps.
Even when an organelle population (e.g. mitochondria, chloroplasts or secretory
vesicles) has been purified to "homogeneity" by these techniques it often retains
more than one membrane system or else both membrane and some enclosed

solute(s). In such cases, a second equally rigorous round of particle disruption and
fractionation has to be undertaken if an homogeneous membrane preparation (e.g.
of mitochondrial inner membranes or lysosome membranes) is to be obtained.


10

F.B. Finean and R.B. Michell

3. Membrane proteins and glycoproteins
Membranes are selectively permeable barriers which compartmentalise, and thereby
exert considerable control over, cellular metabolism. They provide the support and
working environment for a great variety of enzymes, receptors and antigens, each of
which interacts with soluble material in the aqueous milieu either at one or both
surfaces of the membrane. Sites may also be provided through which the membrane
can interact with cytoskeletal elements, as in cell movement or during secretion, or
with extraneous surfaces (for example, in cell-cell interactions or in the interactions
between cells and other solid supports) [44,45]. All of these functions are achieved by
a hydrated structure that is essentially constructed of a bilayer of lipid molecule with
which various types of proteins and glycoproteins are associated: some penetrate
through the lipid bilayer, some are inserted into it only from one side, whilst others
are associated with the membrane in a more superficial manner which does not
involve direct interaction with the lipid bilayer [17,46,47]. Most membrane functions
are functions of the membrane proteins and glycoproteins, and the relative variety
and abundance of the (glyco)protein species found in any individual membrane are
to some extent a reflection of the diversity and intensity of its biological activities.

Lipid
Myelin

Protein

~J

Erythrocyte

_

Retinal rod

_

Sorcoplasmic
reticulum

Inner
membrane of
mitochondrion

~

~I...-

_

~
1...-

Purple patches ~
Halobacterium

holobium
1...-

....

_

Fig.4. Relative proportions of lipid and protein in a range of membrane preparations (from [44)).


Isolation, composition and general structure

11

The proportion of the dried weight of various membrane preparations which is
protein varies within the range 20 to 75% (Fig.d). At the lower extreme is nerve
myelin which exhibits only a few relatively weak enzyme activities: its major
function appears to be as an electrical insulator [48]. Two very different examples of
membranes which have about three-quarters of their dried weight as protein are the
purple membrane of Halobacterium halobium and the inner mitochondrial membrane. The former possesses only a single protein, a light-driven proton pump named
bacteriorhodopsin [49], and this forms a close-packed hexagonal array in the
membrane (see Section 5b; Chapter6), whereas the protein of the latter is a highly
complex mixture of components involved in electron transport, ATP synthesis and
solute transport [50]. The interpretation in structural terms of this variability in
protein content between different membranes is not straightforward, in that it must
take account of the fact that some proteins are superficially attached at the hydrated
surfaces of membranes (extrinsic or peripheral proteins) whilst others include regions
which are inserted to a significant extent into the non-polar interior of the membrane (intrinsic or integral proteins) [46].
(a) Extrinsic proteins

Extrinsic proteins may be removed from membrane preparations by treatment with
solutions of low ionic strength; lightly buffered water is often used, sometimes with
the addition of EDTA to chelate divalent cations. Such procedures, which have been
listed by Tanner [51], have been most fully characterised using erythrocyte membranes, in which slightly more than half of the dried weight is protein. These
treatments solubilise 25-30% of this protein without destroying the basic organisation of the membrane as revealed by thin-section or freeze-fracture electron microscopy [46,52]. After such treatments the membranes do, however, tend to spontaneously vesiculate in a manner that does not occur in membrane preparations that
retain their extrinsic proteins [53]. Most erythrocyte membrane preparations are
isolated at ionic strengths much lower than are "physiological", and raising the ionic
strength to around the physiological range often releases additional "loosely"
associated proteins: how much of this protein is genuinely a part of the membrane
and how much is simply adsorbed at low ionic strengths is often a matter of some
dispute. For example, erythrocyte membranes do not bind haemoglobin when in
physiological media, but they bind large quantities of the protein during isolation in
media of low ionic strength [30]. Mild protein perturbing agents (e.g, chaotropic ions
such as 1-, Cl04- and SCN- [51)) sometimes release additional "extrinsic" protein,
and combinations of various techniques can solubilise as much as half of the total
protein from some isolated erythrocyte membrane preparations.
In addition to doubts arising from the possibility that some membrane-associated
"extrinsic" proteins may simply be adsorbed cytoplasmic constituents, there is
always the possibility that some proteins form real functional associations with
membranes that are unable to survive the conditions chosen for a particular
membrane isolation: examples of such proteins might include components involved


12

F.B. Finean and R.B. Michell

in interactions with the filamentous and microtubular cytoskeleton within cells,
cytosol enzymes which form "loose" associations with membranes [54], and soluble
proteins that have a role in controlling membrane processes (e.g. the calmodulin

needed for Ca2+ to control membrane Ca2+ -ATPase [55] or Ca2+ -stimulated
protein kinase [56], or the a-lactalbumin of the lactose synthetase complex [57]).
(b) Intrinsic proteins

Intrinsic proteins are membrane proteins that require disruption of membranes by
appropriate detergents or by organic solvents for their liberation [46,57]. When
applied to the erythrocyte membrane, treatment with non-ionic detergents such as
Triton X-lOO can dissect from the membrane essentially all of the intrinsic proteins
and lipids, leaving a "shell" of extrinsic proteins that is relatively stable at physiological ionic strength [58-60]. With other membranes, results are variable and there
are often some protein components for which it proves difficult to devise successful
non-denaturing procedures for extraction with detergents. Detergents appear to
release intrinsic proteins from membranes and then retain them in solution as a
result of their abilities to provide an amphiphilic coating over predominantly
lipophilic areas of the protein surface ([6]; see Fig.5). The amount of detergent
bound by an intrinsic protein is probably an approximate measure of the degree to
which the protein interacted with the lipophilic region of the membrane from which
it came, and the ability of such proteins to bind detergents provides the basis of a
simple procedure for distinguishing intrinsic from extrinsic proteins [62].
The lipophilic surface regions of intrinsic proteins are areas in which there is a
great predominance of exposed non-polar amino acid residues: these may arise from

Fig.5. Illustration of the mode of action of detergents in liberating intrinsic proteins from membranes.
Lipophilic portions of protein and detergent molecules are indicated in black (from [44]).


Isolation, composition and general structure

13

10

20

30

40

50

60

~~~~;~esE:~X:~~~X~~~)!I!lml;~~~;~~~~~I!I!I!1
70

eo

90

100

110

120

130

Fig. 6. Amino acid sequence of glycophorin A isolated from human erythrocyte membrane. The sequence
of lipophilic residues which traverses the lipid bilayer are indicated in black. Carbohydrate side chains are
indicated by CHO (from [63]).

relatively uninterrupted sequences of non-polar amino acids (for example, that in
glycophorin A, Fig.6 and [63]) or as a result of a particular pattern of folding of
polypeptide sequences so as to bring many non-polar residues close to one another
(as in the generation of the hydrophobic surface of bacteriorhodopsin, see p. 226 and
[64]). The extent to which such proteins are inserted into the non-polar interior of a
membrane ranges from a mere toehold (e.g. myelin basic protein: see Chapter6) to
almost complete immersion (e.g. bacteriorhodopsin).
Although the structural definitions of extrinsic and intrinsic proteins appear
relatively unambiguous, the assignment of individual proteins to these categories is


14

F.B. Finean and R.B. Michell

based largely on the experimental conditions which liberate them from membranes.
For intrinsic proteins that can be successfully solubilised, the criterion of ability to
bind detergents provides an additional check on the assignments. Extrinsic proteins
may be bound to the membrane surface through interaction with intrinsic proteins,
with lipid headgroups or with both. Warren argues in Chapter 6 that, in order to
explain the characteristic patterns of relatively loosely associated extrinsic proteins
in different membranes, one must propose that each intrinsic protein includes
amongst its interactions some specific association with membrane-spanning intrinsic
proteins. This type of specific interaction is also emphasised by certain situations
where the choice of the term "extrinsic protein", rather than "extrinsic polypeptide",
may be misleading, since some of the conditions used to release "extrinsic" components cause the dissociation of multisubunit enzyme proteins. Hence the "extrinsic"
F1-ATPase of energy-coupling membranes is in reality a part of the intrinsic ATP
synthase of those membranes, and some other "extrinsic" polypeptides of the inner
mitochondrial membrane are subunits of the intrinsic cytochrome oxidase [57].
(c) Analysis of membrane proteins

The complex protein compositions of membranes are most readily demonstrated by
analyses in which membranes are first dissociated in a reducing medium containing
sodium dodecyl sulphate (SDS), so as to unfold the constituent polypeptides and
coat them with the negatively charged detergent, and then electrophoresed in
polyacrylamide gels in the presence of an excess of SDS (see section IB of [35];
Chapter 7 of [22]).
In this technique (known as SDS-PAGE) unmodified polypeptides migrate at
rates that provide fairly reliable estimates of their molecular weights, but this is not
true of heavily glycosylated polypeptides such as those derived from some membrane
glycoproteins. The most common methods of detecting polypeptides in such gels are
staining for protein with Coomassie Blue or for some (but not all) carbohydrate
substituents with a periodic acid-Schiff stain. Other more selective detection procedures include immunological methods [65] and autoradiography of some biologically
or chemically introduced label, such as an amino acid or a sugar, phosphate groups
introduced into membrane proteins, labelled iodine introduced by peroxidase, or
some univalent or bivalent amino acid-directed chemical probe (see, for example,
Chapter 3 and [66,67]).
The most widely used form of SDS-PAGE is a one-dimensional separation, which
provides a profile of the major polypeptides that contribute to the proteins of a
membrane. This then serves as a convenient reference pattern within which to locate
particular polypeptides by their distinctive structural or functional properties. In
such a profile, however, polypeptides of similar molecular weight that are derived
from different proteins will not be distinguished, nor will any relationship be
apparent between polypeptides that are of different sizes but which were derived
from a single functional protein in the membrane. When greater resolution of the
polypeptide mixture is required, particularly with respect to minor polypeptides that


15

Isolation, composition and general structure

are obscured in the one-dimensional SDS-PAGE separation, then two-dimensional
techniques, such as that in which the proteins are separated by isoe1ectric focussing
in the first dimension and by electrophoresis in the second [68], may be used.
(d) Proteins of the erythrocyte membrane

Both the value and the limitations of the one-dimensional SDS-PAGE technique can
be conveniently illustrated from studies of the erythrocyte membrane. Typical

A

12

1.2

2.1

E
c;

0.8

0

'"

4.2

0{)

<1

4.1

0.4

5

,2.2

6

TD

4.5

2.3

0

B

PAS ·1

0.4

LIPID

E
c:

TO

o0{)

<1

0.2
PAS-2
PAS-4

o

20

40

60

80

MIGRATION, mm

Fig.7. Absorbance profiles of polyacrylamide gels obtained by electrophoresis of erythrocyte membrane
polypeptides. (A) stained with Coomassie blue; (B) stained for carbohydrate by the periodate-Schiff
method.


16

F.B. Finean and R.B. Michell

patterns for the distribution of Coomassie blue-staining polypeptide components
and periodic acid-Schiff-staining carbohydrate components are shown in Fig. 7. This
distinguishes around 20 bands, as compared with a resolution of substantially more
than 200 components if similar preparations are subjected to a two-dimensional
separation [69]. Table 1 lists the identities of the proteins from which at least some of
the polypeptides are derived.
The dominant "extrinsic" components in low ionic strength aqueous extracts of
erythrocyte membranes are the two spectrin bands (l and 2) and membrane actin
(band 5) [31,51,52,60,70]. Reference to electron micrographs suggests that these

TABLE 1
Some polypeptides and glycopeptides of the human erythrocyte membrane
For the majority of references, see the accompanying text.
Coomassie
blue-stained
band No.

Apparent Jf,
by SDS-PAGE

1
2
2.1
2.2

240000 }
220000

200000 }
140000

3

90000-100000

4.1

78000

4.2
5
6

7

72000
43000
35000
29000

H

17000

Periodic acidSchiff-stained
band No.
PAS·l and PAS-2

PAS-3 and PAS4

Identities and functions of components in band

Spectrin 1 and 2. Appear to exist in membrane mainly
as (SI :S2>2 tetramers associated with actin and band 4.1
Ankyrin (or syndein), anchors actin/spectrin/4.1 complex to some component of band 3
Catalytic subunit of Ca 2+·pump ATPase, 0.1% of total
protein
Broad band, little staining with periodic-acid-Schiff
but some polypeptides carry an M r 11 000 oligosaccharide (erythroglycan). Major component is anion
transporter. About 25% is glucose transporter. Also
includes catalytic subunit of Na+/K+-ATPase and binding sites for ankyrin, band 6 and aldolase
{ Component of spectrin/actin cytoskeleton
Acetylcholinesterase
Cytoskeletal
Actin, cytoskeletal
Glyceraldehyde-3-phosphate dehydrogenase
Cytoskeletal, possibly also some cytoplasmic contaminant
Globin monomer from absorbed cytoplasmic haemoglobin

Dimeric and monomeric glycophorin A, the major
membrane sialoglycoprotein. PAS-l may accumulate
in regions of membrane fusion during membrane vesiculation [80a]
Unknown


Isolation, composition and general structure

17

proteins are the main components of a protein network located at the cytoplasmic
surface of the membrane: this submembrane cytoskeleton probably consists of
spectrin tetramers crosslinked through actin and band 4.1 [60,70]. Further removal
of extrinsic protein then releases further bands (2.1, 2.2, 2.3, 4.1, 4.2) that also
coexist with spectrin and actin in the membrane "shells" that remain after intrinsic
proteins and lipids have been removed by Triton X-lOO treatment [60,71]. Several of
these probably have roles in the cytoskeleton, especially ankyrin (bands 2.1 and 2.2)
which serves to moor the cytoskeletal network to intrinsic polypeptides of band 3
[72]. Band6 is derived from an "extrinsic" protein that is released at physiological
ionic strength and which is absent from ghosts prepared in such ionic media. It has
been identified as glyceraldehyde-3-phosphate dehydrogenase, a major cytoplasmic
enzyme. At low ionic strengths it appears to be attached to the membranes through
specific binding sites on band 3, and the fact that its binding is modulated by
metabolites related to the glycolytic pathway has been presented as a strong
argument that its association with the membrane also occurs in the intact cell ([54]
and Chapter 5). At present, however, this cannot be said to be convincingly established.
The most prominent of the polypeptides that arise from intrinsic erythrocyte
membrane proteins are those present in band 3 and PAS-l (see Fig. 7). The latter is a
dimer of glycophorin A, the principal sialoglycoprotein of the membrane: this has
been characterised in detail and its disposition in the membrane determined (see
Fig.6 and Chapter 3). Band 3 includes 25 to 30% of the total protein of the
membrane. There is little doubt that the major component of this band is an anion
transport protein, and this can be functionally identified by its ability to bind
radioactive irreversible inhibitors of anion transport, e.g. 4,4'-diisothiocyanostilbene2,2'-disulphonic acid (DIDS). The presence of the anion transporter in band 3 is also
indicated by the ability of band 3-enriched protein fractions to confer anion permeability upon lipid systems [75]. Some estimates have suggested that this protein
comprises more than 90% of the material in band 3, but there is other evidence which
suggests that the heterogeneity of this band may be greater than has sometimes been
acknowledged [51,73]. In particular, studies with a covalent inhibitor of sugar
transport (maltosyl isothiocyanate) suggest that about one-quarter of the polypeptide chains migrating in band 3 arise from the glucose transport protein of the

red cell [6,77]; studies with inhibitors indicate that glucose transport and anion
transport are functions of different proteins [78]. Also in band 3 are component(s) of
a water channel [79] and the catalytic component of the Na+ /K + -ATPase [31,51,73],
but each erythrocyte only contains a few hundred copies of the latter polypeptide, as
compared with 300000 copies of the glucose transporter and almost a million anion
transporters. Of the million or so band 3 polypeptides, about half carry a large
oligosaccharide chain ("erythroglycan") [80]. Another function of polypeptides that
are isolated in band3 is to provide about 100000 binding sites through which
ankyrin moors the spectrin-actin-band 4.1 cytoskeletal network to the membrane: at
present this function is attributed to some of the anion transporters [60,72].
The obvious functional heterogeneity of the glycopolypeptides that contribute to


18

F.B. Finean and R.B. Michell

band 3 apparently contradicts a variety of studies which indicate that for the
purposes of structural analysis this band can be considered as a homogeneous
polypeptide species [51,54]. This conflict might be to some extent reconciled if the
anion and glucose transporters turn out to be structurally very closely related.

4. Membrane lipids
The amount of lipid in a membrane can probably be regarded as the amount needed
to provide a continuous bilayer barrier around a cell or a segregated intracellular
space, except that this quantity is diminished as the fraction of the membrane area
that is occupied by the intrinsic proteins increases. In the purple patches of
Halobacterium halobium membrane each relatively small bacteriorhodopsin molecule
is accompanied by only 12-14 lipid molecules, whereas in the sarcoplasmic reticulum of skeletal muscle there are approaching a hundred lipid molecules for each
substantially larger Ca2+ -pump ATPase molecule. The lipids that make up these

bilayers are diverse, varying both from membrane to membrane and from organism
to organism [81].
(a) Glycerophospholipids

Most membranes, from almost all types of organism, contain substantial proportions
of these lipids, in which one of the primary hydroxyls of glycerol bears a phosphatecontaining group and the other two bear relatively bulky hydrocarbon-containing
substituents. In the majority of organisms the long-chain hydrocarbons are linked to
carbons 1 and 2 of the glycerol by carboxylic ester, vinyl ether or saturated ether
linkages, and they may, depending on the organism and lipid, be selected from any
of a variety of saturated, mono- and polyunsaturated, branched or otherwise
modified structures. The phosphate is usually on the 3-carbon of glycerol, and
attached to it can be any of a variety of polar substituents. The most common
glycerophospholipids are included in Fig. 8; others include molecules in which either
additional phosphate groups or mannose residues are attached to the inositol of PI,
and also phosphatidylglycerol and its aminoacylated derivatives [81,82].
Glycerophospholipids of this stereochemicalconfiguration constitute the majority of
the lipid of intracellular organelles in animal cells and also contribute substantially
to the lipids of most other membranes.
In a small subgroup of bacteria (the Archaebacteria), of which Halobacteria are
the best known, the stereochemical configuration of the glycerolipidsis reversed, so
that the phosphate-containing substituent is on carbon-l of the glycerol, and
dihydrophytyl chains are attached through ether links to carbons 2 and 3: the major
phospholipid of these cells is an analogue of phosphatidylglycerol [83].


~

~

i:l

~.

8

~

§.

§.
§

~

OH

Coo

CoO
I

o

Cholesterol

Coo

I

I

0

0

CH2 - CH

• CH -

I

,

CH2

CoO

I

0

oI

0

I

,CH CH2
,

/

OH

,

Oap-O-

HO

'

0

CoO
I

o

I

0

o

CoO

Coo

CH-CH2

CH - CH2

,

CH2

CH2

I

I

I

, NH

6

~H-CH2

He"

CH
CH2

o

b

0=;'-6

0=p-6

o=p -0-

<;>

<;>

o

CH2

~H2

CH

CH2
,

~H2
NH2

CH2

,

O~C'
I

'NH2

I

I

,
o

I

I

~H2
CH2

CH.-~C!l- CH.

(PI)

-serine (PS) -ethanolamine
•
(PEl

NH

NH

CH'

I

CH2

o

H

-choline (PC)

~

f

,

>-~H

HO

CH
I
2

9r

,~ooc~

OH

0

CH20H)
= {HOH
CHOH
I

M,C

<;>~H

~
'"~
~

OH

0

~

~

Coo

'6 0" 'CH.

HOCH2Y

CH.

Sphingomyelin
(SM)

-inositol

O·

C·o

I

CH.

Phosphotidyl-

H

CoO

CH.-~Gl- CH.

I

-0

,

CoO

o

OH

Diphosphotidyl-glycerol (OPG)

Coo

o
0=~-6

HO~OH
OH

,

CH2

9

o-s-o
o-s-o
I
I
CH2

o,

0

CH2

1_

CaO

I

~H - CH2

CH 2

I

CoO

I

o

0

I

o

CaO

I

CH2

I

-

Coo

Galactosyl
ceramide

(cerebroside)

PO~OH
o
o

HOCH2
HO

,,0
NH-C,
CH.

HOCH2~
HO

MonosialollOR9lioside

(G1,1I)

Fig. 8. Examples of principal classes of lipids found in animal cell membranes (from [44)).

....
\0


20

F.B. Finean and R.B. Michell

(b) Glyceroglycolipids

These glycerolipids [81] have been identified in substantial quantities in chloroplast
membranes, in blue-green algae and in bacteria, but they also exist in much smaller
quantities elsewhere. Monogalactosyldiacylglycerol (1,2-diacyl-3-p-D-galactosyl-snglycerol), digalactosyldiacylglycerol and sulphoquinovosyldiacylglycerol (in which
the headgroup is a sulphated deoxysugar) are the dominant lipids of chloroplasts.
Bacterial glyceroglycolipids include a greater variety of sugars. As with the
glycerophospholipids, the glyceroglycolipids of Halobacteria are of reversed stereochemical configuration and bear dihydrophytyl ether side-chains [83]. Other
Archaebacteria (e.g. the thermoacidophiles Thermoplasma and Sulfolobus), also include polyisoprenyl ether lipids of this configuration, some of which even comprise
two glycosylglycerols that are covalently linked, presumably across the membrane,
through long-chain ethers [83a].
(c) Phosphosphingolipids

Sphingolipids contain a hydrophobic portion (a cerarnide) which is an N-acetylated
derivative (sometimes with an a-hydroxy fatty acid) of a long-chain aminoalcohol
known as a sphingoid (Fig.8) [81]. Sphingomyelin, in which the group attached to
the terminal hydroxyl of cerarnide is phosphorylcholine, is a widely distributed
phosphosphingolipid that is concentrated in the plasma membranes of animal cells.
Other phosphosphingolipids bearing substituents similar to those of some of the

common glycerophospholipids are widely distributed in nature, occasionally occurring in substantial quantities (e.g. cerarnide-phosphorylinositol and more complex
inositol phosphosphingolipids in yeast [84]). Some plant phosphosphingolipids are
extremely complex [85].
(d) Glycosphingolipids

These lipids bear sugar groupings on the terminal hydroxyl of cerarnide (Fig. 8)
[81,86,87]. These vary greatly in size and complexity, from, for example, a monosaccharide in galactosylcerarnide (cerebroside) up to oligosaccharides containing 20-60
sugar residues in some of the minor glycosphingolipids that carry erythrocyte blood
group antigens [88- 90]. Although they are generally minor membrane constituents,
glycosphingolipids are concentrated in plasma membranes where they may occasionally be amongst the dominant lipid classes (e.g. in myelin [45,91] and in intestinal
epithelial plasma membranes [92]).
(e) Sterols

Sterols (e.g. cholesterol, Fig.8) are widespread membrane constituents in a wide
variety of higher organisms and also in some microorganisms [93,94]. Cholesterol is
much the most common membrane sterol in animals, and it tends to be concentrated


Isolation, composition and general structure

21

in plasma membranes and in functionally related intracellular membranes (e.g. Golgi
membranes and secretory vesicle membranes) [95]. Other membranes, such as the
endoplasmic reticulum and mitochondria of most tissues, contain little cholesterol
[95]. In some organisms, alternative molecules can replace cholesterol (e.g. stigmasterol and sitosterol in plants, ergosterol in yeast and tetrahymanol in cholesteroldepleted Tetrahymena [94,96]).
(f) Chlorosulpholipids

A novel group of membrane lipids accompanies the typical lipids of mitochondria
and chloroplasts in the phytoflagellate Ochromonas danica and related organisms

[97]. These lipids are chlorinated derivatives (1-6 chlorines) of a 22-carbon alkyl
disulphate in which one of the sulphate groups is terminal and the other on
carbon-l 4: the dominant member of this family is 2,2,1l,13,15,16-hexachloro-l,14docosane disulphate. Thus these lipids, in contrast to other membrane lipids, bear
two polar sulphate groups that are almost at opposite ends of their lipophilic
backbone. How they are incorporated into the lipid bilayer of Ochromonas membranes is still unknown.
(g) The mixed-lipid phase of membranes

Any selected biological membrane will contain an appreciable variety of different
lipid types selected from those described above. Moreover, within each structural
class of lipids there will be variations in the lipophilic groups both of the sphingoid
moieties and of the amide, ester and ether substituents; they may be of various chain
lengths (predominantly 14 to 24 carbon atoms), saturated or unsaturated (up to 6
cis-double bonds), linear or branched and they may include other interruptions such
as cyclopropane rings [80,81].
The functional significance of this diversity, by which any individual membrane
may come to possess more than a hundred chemically distinguishable types of lipid
molecule, is yet to be comprehended. It seems possible that some of the diversity,
such as the appearance of quantitatively very minor fatty acid pairings in an overall
lipid class (e.g. phosphatidylcholine), might be simply a reflection of imprecise acyl
group selectivity on the part of the synthetic enzymes. However, this would not
account for the existence of lipids with a variety of hydrophilic headgroups, for the
same headgroup occurring on sphingolipids and on glycerolipids with acyl, alkenyl
ether sidechains, or for the possession by cells of a fairly complex enzymic complement that is apparently devoted to the adjustment of the fatty acid patterns of
pre-existing lipids [98]: the evolutionary conservation of these features points to their
having some distinctive, but as yet unknown, functions.
One aspect of membrane lipid mixtures that appears to be universal is that at
physiological temperatures there is sufficient hydrocarbon chain distortion (e.g. by
cis-double bonds, methyl branches or cyclopropane rings) for the hydrophobic phase