New comprehensive biochemistry vol 01 membrane structure

Box363,Birmingham B15 2TT, us: The initial suggestion that protein would probably be closely associated with lipid in plasma membranes and maybe also in other membranes was again a specu

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

New Comprehensive Biochemistry

ELSEVIER/NORTH-HOLLAND BIOMEDICAL PRESS

AMSTERDAM· NEW YORK OXFORD

In the former series of Comprehensive Biochemistry the contributions of membranes

to cellular biochemistry were considered in a volume entitled Cytochemistry(1964) inwhich the organelles of the cell were considered individually Since that time thestudy 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 structureand function Other aspects of membrane biochemistry will be discussed in forth-coming volumes onPhospholipids and on Membrane Transport. One of the outstand-ing features of recent research on membrane structure has been a transition from themarked polarisation of views that characterised the 1960s towards a general agree-ment during the 1970s that all membranes share one basic form of structuralorganisation The aims of this volume are to identify general features of membranestructure, to discuss in considerable detail some selected aspects that have beenstudied 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 ofcell structure and of the roles of individual membranes and organelles in particularcell functions For those who lack this background information, we would recom-mend reference to brief monographs such asMembranes 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, Glasgowand London) and Biological Membranes (R Harrison and G.G Lunt, 2nd ed., 1980,Blackie, Glasgow and London)

J.B FineanR.H MichellBirmingham, August 1980

Isolation, composition and general structure

of membranes

J.B FINEAN and R.H MICHELL

Department of Biochemistry, University of Birmingham

P.o Box363,Birmingham B15 2TT, us:

The initial suggestion that protein would probably be closely associated with lipid

in plasma membranes (and maybe also in other membranes) was again a speculativeone based on surface tension measurements and on the spontaneous association ofwater-soluble proteins with monolayers of lipid spread at an air-water interface.Although there was no relevant information on the protein components of mem-branes, Danielli and Davson proposed a general structural scheme [6] for cellmembranes which featured a bilayer of lipid coated at its aqueous interfaces withlayers of protein Their first suggestion that protein might penetrate into or throughthe lipid layer [7] was not based on any direct knowledge of membrane proteins, butwas simply a speculative attempt to account for the occurrence of facilitatedpermeation 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

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 withinthe cell was not recognised until the 1950s when improvementsin the preparation ofthin sections of tissues for examination by electron microscopy indicated a similar

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

ments (Fig.3) In particular, it was realised that: (a) under appropriate conditionssome lipids would adopt configurations other than a bilayer; (b) the fine detail ofmembrane structure as seen at high magnification in some electron micrographsappeared granular; and (c) membranes were dissociated into lipoprotein "particles"

by detergent treatment This encouraged speculation, especially by biochemists, thatmembranes might consist of laterally aggregated arrays of globular lipoprotein

"subunits" (e.g [10-12,14])

Ithas only been as a result of the relatively recent progress in characterisation ofmembrane proteins that substantial agreement on a general model of membranestructure has been reached In particular, the identification of membrane proteins inwhich substantial exposed regions are dominated by non-polar amino acid sidechains led to the realisation that such regions would be likely to associate withhydrocarbon regions of the membrane lipid phase; parts of these proteins mighttherefore be inserted deep into the membrane interior This, together with anemphasis on disordered or fluid packing of the lipid hydrocarbon chains and on freelateral diffusion of membrane components, was then featured in a new membranemodel, 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 generalcharacteristics of membranes than any previous model Itmay well be the last of thegeneralisable membrane models, because experimental work on membranes has nowadvanced to the stage at which the structural patterns of individual membranes arebeing defined in some detail [18] As a result, we now know that individualmembranes differ both in the spatial distributions of their molecular componentsand in the mobilities of these components

(a) Criteria for assessing purity

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

prepara-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 someorganelle-specific function (e.g an enzyme activity) that can be adequately studiedeven when the membrane exhibiting it exists only as a component of a membranemixture For membranes which contribute substantially to the total membranecomplement of cells, achievement of homogeneity requires a purification of only afew-fold, and appropriate techniques may not be unduly complex or difficult todevise Many membranes, however, constitute only a very small proportion of thetotal 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 ofhomogeneous material for analysis

The monitoring of membrane purification basically consists of following thepurification of the required membrane by monitoring some membrane-specificcriterion, associated with the simultaneous measurement of a variety of additionalcriteria specific for all of the possible contaminant structures Occasionally themorphology of a particular membrane structure remains sufficiently distinctive, evenafter homogenisation, for electron microscopy and/or phase contrast microscopy toprovide a reliable guide to purification (e.g mitochondria, rough endoplasmicreticulum, intestinal epithelial brush borders, secretory vesicles), but much moreoften the isolated membrane fragments do not retain a morphology that is suffi-ciently characteristic for their unequivocal identification (e.g smooth membranefragments may come from, among others, smooth endoplasmic reticulum, plasmamembrane or Golgi complex) In most cases, therefore, the progress of the requiredmembrane and of contaminants through a fractionation procedure is followed by theassay of a variety of membrane-specific or "marker" criteria; these are usuallyenzyme activities known to be confined to particular membranes in the cell understudy (see, for example, [19] and [20], section 1 of [21], Chapters 1-4 of [22]) Amembrane preparation should only be adjudged "pure"; (a) when the purificationachieved corresponds to that which would be estimated from consideration of themorphology of the parent cell, and (b) when the concentrations of all knowncontaminating membranes, as assessed by the activities of their characteristic markerenzymes, have been reduced to levelswhere it can confidently be calculated that theycontribute 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 ofRelative SpecificActivities (RSAs) of the chosen marker enzymes, these RSAs beingthe ratios which compare the specific activities in the final fraction(s) to the specificactivities in the initial homogenates [23] In interpreting RSAs, it is essential toremember that the mass contributed by a particular stucture to an isolated fraction

isa function both of the experimentally determined RSA and of the contribution ofthe particular organelle to the mass of the parent cell To illustrate this, consider asimplified cell with only two membrane systems, a plasma membrane that contains1% of the cell protein and mitochondria which contain 20% From this cell oneisolates an SO-fold purified plasma membrane fraction in which the RSA of the

mitochondrial marker enzyme remains 1.0, as in the original homogenate; 20% of thematerial in this substantially "purified" fraction is contributed by mitochondria Asecond fraction from the same cells has a mitochondrial marker RSA of 4.75, but isalso enriched 5-fold with respect to the plasma membrane marker: reference to thecomposition 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 particularmembrane, there are a number of technical obstacles to be negotiated The parentcells must be available in sufficient quantity and adequate purity, a method must bedevised 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 ionsand proteins However, on dilution in an homogenate this high protein concentra-tion is lost In addition, few cell fractionations are undertaken in predominantlyionic media since such media often cause aggregation of organelles and thus impairthe separation The most common media for subcellular fractionation are physiologi-cally iso-osmotic (approx 300 mosM) or hyperosmotic solutions of non-permeantneutral solutes such as sucrose or mannitol A notable exception to this custom isprovided by skeletal muscle, where the polymerisation of actinomycin in low ionicstrength media means that an ionic medium is sometimes (e.g [24]), though notalways [25], used for the isolation of Ca2+-pump-rich sarcoplasmic reticulum Inaddition, mammalian erythrocyte surface membranes, the most widely studied of allmembranes, are normally isolated in ionic media (either with or without a divalentcation chelator such as EDTA or EGTA), but most of these diverse media are oflower than physiological ionic strength and osmotic activity [26] Most of the time,the possible effects on membrane composition and structure of using non-physiological and non-ionic media for membrane isolation are largely ignored, butexperience with the red cell suggests that such uncritical attitudes may ultimatelyhave to be abandoned For example, erythrocyte ghosts made in media of physio-logical ionic strength and containing small concentrations of divalent cations [27](orreturned 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 thepresence of EDTA or EGTA (e.g, [28]) The former, especially after incubation at37°C to "reseal" them, tend to be resilient spheres or even somewhat biconcave, theyare impermeable to most materials which do not permeate the intact cell, theygenerate and sustain ion gradients, and they retain the "cytoskeletal" layer ofspectrin and actin at their inner surface [26,27,29] The latter, by contrast, adoptrather irregular shapes, are "floppy" and readily vesiculate, are deficient in spectrinand actin, are permeable even to macromolecules, and may carry "extra" membrane-associated proteins that have become adsorbed at low ionic strengths (e.g

haemoglobin and maybe also glyceraldehyde-3-phosphate dehydrogenase; seeChapter 5) [26,28,30,31] Such detailed information on the damaging effects oftransferring membranes into environments strikingly different from those prevailingwithin cells appears to exist only for the erythrocyte plasma membrane, but it might

be anticipated that other membranes, especially other plasma membranes, mightbehave similarly Renewed attempts to devise effective subcellular fractionationprocedures with which to isolate membranes and organelles in media of physiologi-cal ionic strength and composition might well yield remarkably interesting, andmaybe disquieting, insights

For details of appropriate isolation conditions for individual membranes, it isusually necessary to consult primary journals: leads into these, and occasionallytechnical 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 animalcells 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 varioustypes of blood cells and various cell-types which either occur naturally or can begrown in the peritoneal cavity (e.g macrophages, mast cells, polymorphonuclearleukocytes or free-living neoplastic cells such as Ehrlich ascites) Body fluidsnormally contain mixed cell populations, so a preliminary to subcellular fractiona-tion is usually the isolation of one cell type in homogeneous form: appropriatetechniques include differential and/or density gradient centrifugation, free flowelectrophoresis and differential adsorption onto some surface which differentiatesbetween 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 lin): see, for example, section VB of [35]

immunoglobu-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 intrinsiccell population Although this heterogeneity is often ignored, there has been amarked tendency in recent years for individual cell populations to be isolated fromtissues before functional studies are undertaken This has allowed, for example, theproperties of hepatoeytes [36,37] and of Kupffer cells [38] from mammalian liver to

be studied separately Although potential disadvantages of such techniques includethe smaller amounts of starting material that are usually available and the possibilitythat 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 morewidely adopted when isolating membranes for structural studies There are varioustechniques for weakening the forces or structures (e.g collagen fibrils) that hold cellstogether prior to dissociation of tissues to form cell suspensions, of which the mostuseful 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 cellsuspension 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

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 cellhomogenates may exploit any physical differences between the various particles inthe homogenate In practice, however, the great majority of separations are either bydifferential rate centrifugation, distinguishing particles of different sizes, or byisopycnic density-gradient centrifugation, with separation the result of differences inparticle densities In recent years two other techniques have been developed that arepotentially of general applicability The first is free-flow electrophoresis ([20], pp.78-86) in which a suspension of mixed membranes is carried slowly and continu-ously down a vertical curtain of flowing buffer across which an elecrical field isapplied At the bottom of the buffer curtain the various separated streams ofparticles with differing charge characteristics flow into a row of tubes Modemequipment for free-flow electrophoresis can, in a few hours, separate either cells ormembranes of different charges in substantial quantities The second relatively noveltechnique, based on membrane surface characteristics, is the phase partition of amembrane mixture in an aqueous two-phase polymer system such as 5.3% wIw

Dextran(M, 500000)/4.1%polyethylene glycol(M, 6000)/°.1M phosphate buffer,

pH 6.5 ([20], pp 71-75) Both phase partition and free-flow electrophoresis havealready found applications in separating certain types of plasma membrane fromother structures [20] Finally, there ate a number of specifically designed techniqueswhich take advantage of the singular characteristics of particular membranes: forexample, isolation of plasma membranes of phagocytic cells by retrieving thephagocytised particles and their associated phagocytic vacuole membrane (e.g [41]and [20], pp 88,89), binding of carbohydrate-rich surface membranes to bead-boundlectins [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 ofcontaminants than for many functional studies For most membranes this calls forthe use of some hybrid fractionation procedure which adopts, in sequence, morethan one of the above techniques: since it readily accommodates the largestquantities of material, differential rate centrifugation is almost invariably the first ofthese sequential steps

Even when an organelle population (e.g mitochondria, chloroplasts or secretoryvesicles) has been purified to "homogeneity" by these techniques it often retainsmore than one membrane system or else both membrane and some enclosedsolute(s) In such cases, a second equally rigorous round of particle disruption andfractionation has to be undertaken if an homogeneous membrane preparation (e.g

of mitochondrial inner membranes or lysosome membranes) is to be obtained

3 Membrane proteins and glycoproteins

Membranes are selectively permeable barriers which compartmentalise, and therebyexert considerable control over, cellular metabolism They provide the support andworking environment for a great variety of enzymes, receptors and antigens, each ofwhich interacts with soluble material in the aqueous milieu either at one or bothsurfaces of the membrane Sites may also be provided through which the membranecan interact with cytoskeletal elements, as in cell movement or during secretion, orwith extraneous surfaces (for example, in cell-cell interactions or in the interactionsbetween 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 withwhich various types of proteins and glycoproteins are associated: some penetratethrough the lipid bilayer, some are inserted into it only from one side, whilst othersare associated with the membrane in a more superficial manner which does notinvolve direct interaction with the lipid bilayer [17,46,47] Most membrane functionsare functions of the membrane proteins and glycoproteins, and the relative varietyand 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

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

(a) Extrinsic proteins

Extrinsic proteins may be removed from membrane preparations by treatment withsolutions of low ionic strength; lightly buffered water is often used, sometimes withthe addition of EDTA to chelate divalent cations Such procedures, which have beenlisted by Tanner [51], have been most fully characterised using erythrocyte mem-branes, in which slightly more than half of the dried weight is protein Thesetreatments solubilise 25-30% of this protein without destroying the basic organisa-tion of the membrane as revealed by thin-section or freeze-fracture electron micros-copy [46,52] After such treatments the membranes do, however, tend to sponta-neously vesiculate in a manner that does not occur in membrane preparations thatretain their extrinsic proteins [53] Most erythrocyte membrane preparations areisolated at ionic strengths much lower than are "physiological", and raising the ionicstrength to around the physiological range often releases additional "loosely"associated proteins: how much of this protein is genuinely a part of the membraneand how much is simply adsorbed at low ionic strengths is often a matter of somedispute For example, erythrocyte membranes do not bind haemoglobin when inphysiological media, but they bind large quantities of the protein during isolation inmedia of low ionic strength [30] Mild protein perturbing agents(e.g,chaotropic ionssuch as 1-, Cl04- and SCN- [51)) sometimes release additional "extrinsic" protein,and combinations of various techniques can solubilise as much as half of the totalprotein 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 isalways the possibility that some proteins form real functional associations withmembranes that are unable to survive the conditions chosen for a particularmembrane isolation: examples of such proteins might include components involved

in interactions with the filamentous and microtubular cytoskeleton within cells,cytosol enzymes which form "loose" associations with membranes [54], and solubleproteins that have a role in controlling membrane processes (e.g the calmodulinneeded for Ca2+ to control membrane Ca2+ -ATPase [55] or Ca2+ -stimulatedprotein 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 byappropriate detergents or by organic solvents for their liberation [46,57] Whenapplied to the erythrocyte membrane, treatment with non-ionic detergents such asTriton X-lOO can dissect from the membrane essentially all of the intrinsic proteinsand lipids, leaving a "shell" of extrinsic proteins that is relatively stable at physio-logical ionic strength [58-60] With other membranes, results are variable and thereare often some protein components for which it proves difficult to devise successfulnon-denaturing procedures for extraction with detergents Detergents appear torelease intrinsic proteins from membranes and then retain them in solution as aresult of their abilities to provide an amphiphilic coating over predominantlylipophilic areas of the protein surface ([6]; see Fig.5) The amount of detergentbound by an intrinsic protein is probably an approximate measure of the degree towhich 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 asimple procedure for distinguishing intrinsic from extrinsic proteins [62]

The lipophilic surface regions of intrinsic proteins are areas in which there is agreat 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]).

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 inglycophorin A, Fig.6 and [63]) or as a result of a particular pattern of folding ofpolypeptide 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 amembrane ranges from a mere toehold (e.g myelin basic protein: see Chapter6) toalmost complete immersion (e.g bacteriorhodopsin)

Although the structural definitions of extrinsic and intrinsic proteins appearrelatively unambiguous, the assignment of individual proteins to these categories is

based largely on the experimental conditions which liberate them from membranes.For intrinsic proteins that can be successfully solubilised, the criterion of ability tobind detergents provides an additional check on the assignments Extrinsic proteinsmay 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 toexplain the characteristic patterns of relatively loosely associated extrinsic proteins

in different membranes, one must propose that each intrinsic protein includesamongst its interactions some specific association with membrane-spanning intrinsicproteins This type of specific interaction is also emphasised by certain situationswhere the choice of the term "extrinsic protein", rather than "extrinsic polypeptide",may be misleading, since some of the conditions used to release "extrinsic" compo-nents cause the dissociation of multisubunit enzyme proteins Hence the "extrinsic"F1-ATPase of energy-coupling membranes is in reality a part of the intrinsic ATPsynthase of those membranes, and some other "extrinsic" polypeptides of the innermitochondrial membrane are subunits of the intrinsic cytochrome oxidase [57]

(c) Analysis of membrane proteins

The complex protein compositions of membranes are most readily demonstrated byanalyses in which membranes are first dissociated in a reducing medium containingsodium dodecyl sulphate (SDS), so as to unfold the constituent polypeptides andcoat them with the negatively charged detergent, and then electrophoresed inpolyacrylamide 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 atrates that provide fairly reliable estimates of their molecular weights, but this is nottrue of heavily glycosylated polypeptides such as those derived from some membraneglycoproteins The most common methods of detecting polypeptides in such gels arestaining for protein with Coomassie Blue or for some (but not all) carbohydratesubstituents with a periodic acid-Schiff stain Other more selective detection proce-dures include immunological methods [65] and autoradiography of some biologically

or chemically introduced label, such as an amino acid or a sugar, phosphate groupsintroduced into membrane proteins, labelled iodine introduced by peroxidase, orsome 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, whichprovides a profile of the major polypeptides that contribute to the proteins of amembrane This then serves as a convenient reference pattern within which to locateparticular polypeptides by their distinctive structural or functional properties Insuch a profile, however, polypeptides of similar molecular weight that are derivedfrom different proteins will not be distinguished, nor will any relationship beapparent between polypeptides that are of different sizes but which were derivedfrom a single functional protein in the membrane When greater resolution of thepolypeptide mixture is required, particularly with respect to minor polypeptides that

are obscured in the one-dimensional SDS-PAGE separation, then two-dimensionaltechniques, 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

patterns for the distribution of Coomassie blue-staining polypeptide componentsand periodic acid-Schiff-staining carbohydrate components are shown in Fig 7 Thisdistinguishes around 20 bands, as compared with a resolution of substantially morethan 200 components if similar preparations are subjected to a two-dimensionalseparation [69] Table 1 lists the identities of the proteins from which at least some ofthe polypeptides are derived.

The dominant "extrinsic" components in low ionic strength aqueous extracts oferythrocyte 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.

PAS·l and PAS-2

PAS-3 and PAS4

Apparent Jf,

by SDS-PAGE

240000 } 220000

200000 } 140000 90000-100000

78000 72000 43000 35000 29000 17000

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 com- plex 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 Mr 11 000 oligo- saccharide (erythroglycan) Major component is anion transporter About 25% is glucose transporter Also includes catalytic subunit of Na+/K+-ATPase and bind- ing 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 con- taminant

Globin monomer from absorbed cytoplasmic globin

haemo-Dimeric and monomeric glycophorin A, the major membrane sialoglycoprotein PAS-l may accumulate

in regions of membrane fusion during membrane culation [80a]

vesi-Unknown

proteins are the main components of a protein network located at the cytoplasmicsurface of the membrane: this submembrane cytoskeleton probably consists ofspectrin 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 alsocoexist with spectrin and actin in the membrane "shells" that remain after intrinsicproteins and lipids have been removed by Triton X-lOO treatment [60,71] Several ofthese 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 physiologicalionic strength and which is absent from ghosts prepared in such ionic media It hasbeen identified as glyceraldehyde-3-phosphate dehydrogenase, a major cytoplasmicenzyme At low ionic strengths it appears to be attached to the membranes throughspecific binding sites on band 3, and the fact that its binding is modulated bymetabolites related to the glycolytic pathway has been presented as a strongargument that its association with the membrane also occurs in the intact cell ([54]and Chapter5) At present, however, this cannot be said to be convincingly estab-lished

The most prominent of the polypeptides that arise from intrinsic erythrocytemembrane proteins are those present in band 3 and PAS-l (see Fig 7) The latter is adimer of glycophorin A, the principal sialoglycoprotein of the membrane: this hasbeen characterised in detail and its disposition in the membrane determined (seeFig.6 and Chapter3) Band 3 includes 25 to 30% of the total protein of themembrane There is little doubt that the major component of this band is an aniontransport protein, and this can be functionally identified by its ability to bindradioactive irreversible inhibitors of anion transport, e.g 4,4'-diisothiocyanostilbene-2,2'-disulphonic acid (DIDS) The presence of the anion transporter in band 3 is alsoindicated by the ability of band 3-enriched protein fractions to confer anion permea-bility upon lipid systems [75] Some estimates have suggested that this proteincomprises more than 90% of the material in band 3, but there is other evidence whichsuggests that the heterogeneity of this band may be greater than has sometimes beenacknowledged [51,73] In particular, studies with a covalent inhibitor of sugartransport (maltosyl isothiocyanate) suggest that about one-quarter of the poly-peptide chains migrating in band 3 arise from the glucose transport protein of thered cell [6,77]; studies with inhibitors indicate that glucose transport and aniontransport 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, ascompared with 300000 copies of the glucose transporter and almost a million aniontransporters Of the million or so band 3 polypeptides, about half carry a largeoligosaccharide chain ("erythroglycan") [80] Another function of polypeptides thatare isolated in band3 is to provide about 100000 binding sites through whichankyrin moors the spectrin-actin-band 4.1 cytoskeletal network to the membrane: atpresent this function is attributed to some of the anion transporters [60,72]

The obvious functional heterogeneity of the glycopolypeptides that contribute to

band 3 apparently contradicts a variety of studies which indicate that for thepurposes of structural analysis this band can be considered as a homogeneouspolypeptide species [51,54] This conflict might be to some extent reconciled if theanion 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 intracellularspace, except that this quantity is diminished as the fraction of the membrane areathat is occupied by the intrinsic proteins increases In the purple patches of

Halobacterium halobiummembrane each relatively small bacteriorhodopsin molecule

is accompanied by only 12-14 lipid molecules, whereas in the sarcoplasmic lum of skeletal muscle there are approaching a hundred lipid molecules for eachsubstantially larger Ca2+ -pump ATPase molecule The lipids that make up thesebilayers are diverse, varying both from membrane to membrane and from organism

reticu-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 containing group and the other two bear relatively bulky hydrocarbon-containingsubstituents.Inthe majority of organisms the long-chain hydrocarbons are linked tocarbons 1 and 2 of the glycerol by carboxylic ester, vinyl ether or saturated etherlinkages, and they may, depending on the organism and lipid, be selected from any

phosphate-of a variety phosphate-of saturated, mono- and polyunsaturated, branched or otherwisemodified structures The phosphate is usually on the 3-carbon of glycerol, andattached to it can be any of a variety of polar substituents The most commonglycerophospholipids are included in Fig.8; others include moleculesin which eitheradditional 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 ofthe lipid of intracellular organelles in animal cells and also contribute substantially

to the lipids of most other membranes

Ina small subgroup of bacteria (the Archaebacteria), of whichHalobacteria arethe best known, the stereochemical configuration of the glycerolipidsis reversed, sothat the phosphate-containing substituent is on carbon-l of the glycerol, anddihydrophytyl chains are attached through ether links to carbons 2 and 3: the majorphospholipid of these cells is an analogue of phosphatidylglycerol [83]

C·o,

NH

Galactosyl ceramide (cerebroside)

Po ,,0

CH.

HOCH2 HO

0=p-6

I o

CoO Coo Coo Coo

(b) Glyceroglycolipids

These glycerolipids [81] have been identified in substantial quantities in chloroplastmembranes, in blue-green algae and in bacteria, but they also exist in much smallerquantities elsewhere Monogalactosyldiacylglycerol (1,2-diacyl-3-p-D-galactosyl-sn-glycerol), digalactosyldiacylglycerol and sulphoquinovosyldiacylglycerol (in whichthe headgroup is a sulphated deoxysugar) are the dominant lipids of chloroplasts.Bacterial glyceroglycolipids include a greater variety of sugars As with theglycerophospholipids, the glyceroglycolipids of Halobacteria are of reversed stereo-chemical configuration and bear dihydrophytyl ether side-chains [83] Other

Archaebacteria (e.g the thermoacidophiles Thermoplasma and Sulfolobus), also clude polyisoprenyl ether lipids of this configuration, some of which even comprisetwo glycosylglycerols that are covalently linked, presumably across the membrane,through long-chain ethers [83a]

in-(c) Phosphosphingolipids

Sphingolipids contain a hydrophobic portion (a cerarnide) which is an N-acetylatedderivative (sometimes with an a-hydroxy fatty acid) of a long-chain aminoalcoholknown as a sphingoid (Fig.8) [81] Sphingomyelin, in which the group attached tothe terminal hydroxyl of cerarnide is phosphorylcholine, is a widely distributedphosphosphingolipid that is concentrated in the plasma membranes of animal cells.Other phosphosphingolipids bearing substituents similar to those of some of thecommon glycerophospholipids are widely distributed in nature, occasionally occur-ring in substantial quantities (e.g cerarnide-phosphorylinositol and more complexinositol phosphosphingolipids in yeast [84]) Some plant phosphosphingolipids areextremely 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 monosac-charide in galactosylcerarnide (cerebroside) up to oligosaccharides containing 20-60sugar residues in some of the minor glycosphingolipids that carry erythrocyte bloodgroup antigens [88- 90] Although they are generally minor membrane constituents,glycosphingolipids are concentrated in plasma membranes where they may occasion-ally be amongst the dominant lipid classes (e.g in myelin [45,91] and in intestinalepithelial plasma membranes [92])

(e) Sterols

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

in plasma membranes and in functionally related intracellular membranes (e.g Golgimembranes and secretory vesicle membranes) [95] Other membranes, such as theendoplasmic reticulum and mitochondria of most tissues, contain little cholesterol[95] In some organisms, alternative molecules can replace cholesterol (e.g stig-masterol and sitosterol in plants, ergosterol in yeast and tetrahymanol in cholesterol-depletedTetrahymena [94,96]).

(f) Chlorosulpholipids

A novel group of membrane lipids accompanies the typical lipids of mitochondriaand chloroplasts in the phytoflagellate Ochromonas danica and related organisms[97] These lipids are chlorinated derivatives (1-6 chlorines) of a 22-carbon alkyldisulphate in which one of the sulphate groups is terminal and the other oncarbon-l 4: the dominant member of this family is 2,2,1l,13,15,16-hexachloro-l,14-docosane disulphate Thus these lipids, in contrast to other membrane lipids, beartwo polar sulphate groups that are almost at opposite ends of their lipophilicbackbone How they are incorporated into the lipid bilayer of Ochromonas mem-branes is still unknown

(g) The mixed-lipid phase of membranes

Any selected biological membrane will contain an appreciable variety of differentlipid types selected from those described above Moreover, within each structuralclass of lipids there will be variations in the lipophilic groups both of the sphingoidmoieties and of the amide, ester and ether substituents; they may be of various chainlengths (predominantly 14 to 24 carbon atoms), saturated or unsaturated (up to 6cis-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 membranemay come to possess more than a hundred chemically distinguishable types of lipidmolecule, 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 overalllipid class (e.g phosphatidylcholine), might be simply a reflection of imprecise acylgroup selectivity on the part of the synthetic enzymes However, this would notaccount for the existence of lipids with a variety of hydrophilic headgroups, for thesame headgroup occurring on sphingolipids and on glycerolipids with acyl, alkenylether sidechains, or for the possession by cells of a fairly complex enzymic comple-ment that is apparently devoted to the adjustment of the fatty acid patterns ofpre-existinglipids [98]: the evolutionary conservation of these features points to theirhaving some distinctive, but as yet unknown, functions

One aspect of membrane lipid mixtures that appears to be universal is that atphysiological temperatures there is sufficient hydrocarbon chain distortion (e.g bycis-double bonds, methyl branches or cyclopropane rings) for the hydrophobic phase

Liposome (small unilcmejlor]

t -<:O

I'>LM

Fig 9 Arrangement of phospholipid molecules in artificially created model lipid systems used extensively

in membrane-related studies (adapted from (44)) BLM, Black (or bimolecular) Lipid Membrane.

of the membrane to be "fluid" rather than crystalline (e.g [99]) In some membranes

(e.g plasma membranes of animal cells) this "fluidity" is somewhat restricted by thepresence of substantial quantities of cholesterol [94,95] However, such freedom ofmotion in the hydrocarbon phase does not require a complex lipid mixture: the

lipids of Halobacteria seem to provide an appropriate environment for its membrane

proteins even thoughallof the hydrocarbon chains are similar dihydrophytyl ethers[83], and many enzymes isolated from membranes of complex lipid composition can

be persuaded to function quite happily in much simpler lipid environments [99].Almost certainly it is too simple to consider the lipids of a membrane as if theywere a simple mixture First, there is substantial evidence for the asymmetricdisposition of individual lipids on the two surfaces of some, and possibly all,membranes (see Chapter3); in this case, there is a different lipid environment ineach half of the lipid bilayer Secondly, there may sometimes be a requirement forlocal lipid phase changes or separations to occur in response to relatively smallchanges in temperature or to other physical or chemical changes that are ofphysiological significance Such changes, which are discussed further in ChapterSand [99] and [102], have frequently been observed in artificial lipid mixtures buttheir relevance to the functions of biological membranes under physiological condi-tions is not yet clear Thirdly, particular lipids may be required in membranes eitherbecause their presence is essential for the activity of particular membrane proteins(Chapter'S and [101]) or because their metabolism is specifically involved in somemembrane function (e.g the breakdown of phosphatidylinositol that is associatedwith the activity of Ca2+-mobilizing hormone receptors [103])

5 Structural analysis

In principle, structural analysis of membranes can be approached in a mannersimilar to that employed in traditional crystallographic analyses, Once one knowsthe chemical composition of the system, consideration of the physicochemicalcharacteristics of the components gives some indication of their most probablespatial relationships These are then refined so as to conform to the constraintsimposed by the structural information which emerges from direct measurements,primarily diffraction analysis (using X-rays, neutrons and electrons) and electronmicroscopy From this comes a relative low resolution picture of the generalorganisation of each type of membrane, to which detail is added by a variety ofadditional techniques These include spectroscopic approaches (NMR, ORD), dif-ferential thermal calorimetry, analyses of the effects of chemical and enzymaticmodification, inferences drawn from studies of permeability properties and so on.From thermodynamic consideration of the lipid molecules found in biologicalmembranes, it would be expected that in an aqueous environment they would formaggregates in which their hydrocarbon regions would be kept apart from the waterphase [104] This behaviour of hydrocarbon is mainly a consequence of the fact thatwater-water attraction is much greater than water-hydrocarbon attraction, leading

to a tendency for hydrocarbon to be squeezed out of the water phase: i.e, water islipophobic Attractive forces between individual hydrocarbon moieties are relativelyweak, but the substantial quantity of hydrocarbon in membrane lipids means thatthe overall cohesion of the hydrocarbon phase is substantial*.The forms adopted byaggregatedlipids in an aqueous environment are determined by the shapes, sizes andcharge characteristics of the various regions of the lipid molecules The lipidmixtures present in biological membranes, dominated as they are by diacyl phospholipids and other lipids of similar overall shape and character, can most effectively

be accommodated in a bilayer structure Indeed, when total lipid extracts frommembranes are dispersed in aqueous media they invariably form vesicular structures

in which the lipid occurs as continuous closed bilayer shells**. Some individual

• The frequently used terms hydrophobic and lipophilic are both strictly incorrect when applied to hydrocarbons, since the affinity of hydrocarbon for water is greater than the van der Waals adhesion between hydrocarbon groups In the absence of an entirely appropriate word, we will perforce adopt the term lipophilic to describe such behaviour.

•• Although vesicular lipid aggregates are usually termed liposomes, they are occasionally referred to as micelles In this context, the use of the term micelle can be justified by the fact that a micelle is strictly defined to be any aggregate form which'is in equilibrium with molecules in free solution However, this term was originally introduced to describe the molecular aggregates in soap and detergent solutions, and thus a micelle is usually thought of as a sphericalor discoidal cluster of molecules with

a hydrophilic surface and a completely non-polar interior As a result of this terminological confusion, the authors of several textbooks have fallen into the trap of drawing diacylphospholipid molecules organised into such spherical micelles, a configuration that they do not adopt To avoid such confusion, we would suggest that the term micelle should only be used to describe spherical or discoidal aggregates of the type present in aqueous dispersions of soaps, detergents or lysophospholi- pids (monoacylphospholipids) The vesicular aggregates in aqueous dispersions of diacylphospholipids and of other amphiphilic molecules of similar shape would be most conveniently identified by the term liposome (Fig, 3).

lipids show tendencies towards adoption of other configurations when studied inisolation, but there is no reason to expect such alternative configurations to be adominant feature in any natural membranes [102].

The ways in which proteins and glycoproteins are accommodated in membranesmust also be such as to thermodynamically reconcile their physical characteristicswith those 'both of the aqueous environment and of lipid assemblies with non-polarinteriors, most probably bilayers It was recognised in the late 60s and early 70s thatmany, though not all, membrane proteins exhibit substantial lipophilic character, inthat their liberation from membranes requires treatment with detergents or mildorganic solvents When isolated, they retain bound lipid or detergent.Ifthis is lostthen they tend, in aqueous environments, to aggregate so as to segregate consid-erable portions of their surface away from the water [61]

Thus it is to be expected simply from thermodynamic considerations that thedominant structural theme in biological membranes will be a lipid bilayer, with eachmembrane protein either interacting with the polar surface of the bilayer or insertingsome part of its bulk into the central hydrocarbon region, as dictated by itsparticular surface distribution of hydrophilic and lipophilic domains

The largely non-polar core in membrane interrupts, and thus compartmentalises,the aqueous milieu of living organisms Away from the immediate neighbourhood ofthese barriers organization among water molecules is probably extensive but irregu-lar, a state described as "flickering clusters" of molecules [105].Itwill also be locallymodified by solutes, particularly macromolecules It is possible that water structurebecomes more extensively organized in the immediate vicinity of the somewhatimmobilized molecular components of membranes, maybe as a layer from whichsome solute molecules may be excluded [106] NMR studies of cells have dis-tinguished a minor proportion (about 3-5%) of the cellular water that is in anexceptionally immobilized state [107,108], and differential thermal calorimetry ofisolated membranes identifies a small proportion of non-freezable water [109] There

is also non-freezable water in hydrated phospholipid systems, and this appears toconsist of water molecules that coat the polar headgroups; in phosphatidylcholinedispersions, there are about 11 immobilised water molecules per lipid molecule,equivalent to about 25% of the mass of the lipid molecules [110] In membranes,non-freezable water would be expected to be associated both with lipids and withexposed protein, especially glycoprotein A measure of the "structured" membrane-bound water may have been provided by X-ray diffraction studies which revealedthat the minimum amount of water needed to maintain the structural integrity of theerythrocyte membrane is about 20- 30% with respect to the dried weight of themembrane [111]

In some membranes there may be water-filled "pores" across the central non-polarbarrier that are formed by spanning proteins [112], and there may even be someleakage of water molecules between the hydrocarbon chains However, the amount

of water in this essentially dry central region of a membrane must be very small

(a) Electron microscopy

Electron microscopy has contributed to knowledge of membrane structure in avariety of ways Studies of thin sections of fixed, dehydrated, embedded, sectionedand stained preparations of tissues or tissue fractions have revealed a markedsimilarity in the general form of membranes in all living organisms By comparisonwith lipid and lipid: protein bilayer systems, these studies have also indicated thetypes of molecular organisation that might be involved In such preparations themost readily recognisable characteristic of a membrane is the so-called trilamellar ortrilaminar image seen in cross-sections of tissues or tissue fractions which have beenfixed with osmium tetroxide and further stained with another heavy metal com-pound (e.g uranyl and/or lead acetate) This image features two parallel electron-dense lines with an intervening region of lower electron density The overallthickness of the image is generally of the order of 10 nm, but there are relativelysmall, but significant, differences in the widths of the images of different mem-branes, and these can help to distinguish between different types of membranes incomplex tissue homogenates in which the more characteristic morphological features

of organelles have been lost The image seen in such electron micrographs ofmembranes is not directly interpretable in terms of molecular structure, since itdepicts the distribution of heavy metal stain in a grossly modified derivative of theoriginal membrane Nevertheless, the image has the same general form as thoseobtained from lipid and lipid: protein samples that are known from other studiesinitially to have featured lipid in bilayer form

In most micrographs prepared in this way the dense lines show a microgranularitywhich may be related to the distribution of membrane proteins However, there areonly a few membranes (for example, those of gap junctions [113] and the urinarybladder epithelium [114,115]) where the substructure forms a regular pattern whichcan be corroborated by other methods (Fig lOa) and is therefore of immediate help

in understanding membrane organisation

Where substructure is regular, its geometry may be demonstrated very effectively

by electron microscopy of negatively stained membrane fragments (Fig lOb) This isavery simple technique which has provided a quick method of scanning membranesfor regular structural features In this technique prominent (protruding) detail in thesurface structure is silhouetted against a dense background by flooding the surface

of a membrane with a dilute solution of a heavy metal salt and allowing the salt todry out Often, the membrane will be lightly pre-fixed with glutaraldehyde in order

to stabilise its structure against the distorting effects of the drying procedure In theresulting micrographs of face-on views of membranes the dense regions shouldrepresent channels, depressions or crevices in between protruding structural compo-nents such as protein; the latter do not absorb the stain and thus appear whiteagainst the dense background [114,115] When the images include views of the edges

of curved or folded membrane structures then "side-views" may be observed ofstructures which protrude from the membrane surface

The most elegant example of the resolution of membrane structural detail using

Fig 10 Electron micrographs of urinary bladder epithelial cell membrane (A) thin section of fixed and embedded sample X 247000; (B) negatively stained image of the external face of the membrane.

X 560000; (C) platinum-carbon replica of the externally facing freeze-fracture face of the membrane X84000 Micrographs provided by Dr J David Robertson.

Fig 11 Freeze-fracture electron micrograph (X 80(00) of a stack of collapsed erythrocyte ghosts A sequence from E to E (i.e EBCAE or EACBE) traverses a single collapsed erythrocyte ghost, thus crossing the membrane twice The sequence EBCAE starts from the extracellular aqueous space between two ghosts (E) When the frozen membrane fractures along its interior, one "half' of the membrane remains attached to the extracellular ice: this is represented by fracture face B The fracture then traverses the internal cytoplasmic water space (C) which contains fibrillar material Fracture face A then represents the "half' of the membrane that remains attached to the intracellular ice Finally, the fracture re-emerges into the extracellular aqueous medium (E).

electron microscopy of negatively stained membranes is the recent reconstruction ofthe image of gap junction membranes by a Fourier synthesis of information derivedfrom electron micrographs of uranyl acetate-stained membranes [116] Electronmicrographs recorded with the membranes tilted at various angles to the incidentelectron beam provided both phase and amplitude data suitable for image recon-struction The resulting three-dimensional electron density map (resolution 1.8 nm)revealed cylindrical arrays of six protein subunits (a connexon) in each membrane,with the connexons of adjacent cells apposed so as to form the junctions They alsoidentified two alternative configurations which might represent "open" and "closed"states of the permeability channels through the connexons (see also Chapter6).Negative staining has also been very useful in identifying surface projections on oneside of energy-transducing membranes (e.g mitochondrial inner membranes) as thecatalytic (F() portion of their proton-translocating ATP synthase (see also Chapter5).

An alternative method of demonstrating detail in the plane of the membrane byelectron microscopy is through the application of the freeze-fracture technique (Figs.10C and 11) In this method the tissue or membrane sample is frozen very rapidly to

- 196°C, and is then fractured in vacuo The freshly exposed frozen fracture facesare immediately shadowed at an angle of 20 to 40° by evaporation of a heavy metalsuch as platinum and finally a carbon support film is added, also by evaporation, tostabilise the platinum replica of the fracture face In frozen samples of tissue or of amembrane preparation the fracture follows an irregular course which frequentlyincludes cleavage along a membrane face (Figs lOC and 11) It has now beenestablished beyond all reasonable doubt that where the fracture follows the contour

of a membrane it is predominantly along a non-polar interface at the centre of themembrane Pure lipid "membranes" fractured in this way usually reveal a smoothsurface, but most biological membranes show particle-studded surfaces: these par-ticles probably arise as a result of the penetration of protein into the hydrocarboninterior of the membrane bilayer [117-119]

Crisp, clean fractures with minimum surface contamination are only obtainedconsistently with very rapid freezing in the presence of a cryo-protectant such as 30%glycerol which is presumed to prevent ice crystal formation However, ice crystalformation can be prevented by rapid freezing alone, but if glycerol is omitted thenthe quality of the replica in terms of sharpness of detail is still much reduced.Consequently, virtually all published micrographs of freeze-fracture replicas ofmembranes depict tissues that have been treated with 30% glycerol

(b) Diffraction studies

Low resolution data sufficient to establish the general form of membrane structurehave been obtained from a variety of membranes, largely by X-ray diffractionanalysis (for comprehensive reviews see [120] and [121]) Naturally occurring multi-membrane structures such as nerve myelin, chloroplast grana and retinal rods andcones yield a series of well-defined X-ray reflections dictated by the distribution of

electron density in a direction perpendicular to the plane of the membrane (i.e anelectron density profile through the membrane) Similar low-resolution X-ray dif-fraction patterns have been obtained from preparations of a variety of cell mem-branes isolated and purified by subcellular fractionation techniques After isolation,these have to be formed into multilamellar stacks by ultracentrifugation, sometimesfollowed by partial dehydration so as to reduce (and to vary) the periodicities in thestacks Fragments of isolated membrane treated in this way tend initially to bevesicular or sac-like structures which can be collapsed to form stacks of closelypacked membranes.

Derivation of the electron density profile from the recorded reflections requiresboth amplitude and phase information on the X-ray reflections Information onamplitudes may be obtained directly from the diffraction intensities, but phaseinformation is lost when the diffraction patterns are recorded and it must berecovered by indirect means The problem of phase determination is greatly sim-plified if the unit of structure is centrosymmetric; in this case the phase is simply +

or - (0 or7T). Fortunately, most membrane systems stack as centrosymmetric units,with each diffracting unit incorporating two membranes arranged back to back Thismay be accomplished by infolding of a continuous membrane (myelin, retinal discs),adherence of two apposed membrane surfaces (chloroplast grana, gap junctions) orcollapsing of membrane vesicles (sarcoplasmic reticulum)

The phases of the first few orders of diffraction from some of these metric systems have been established directly by swelling experiments in which thestructures of individual membrane pairs are not altered but the distance betweenthem is changed by swelling at an aqueous interface [122-124] This enables thescattering transform of the membrane to be sampled over a sufficient range toestablish the signs of the lower order reflections, allowing a low resolution electrondensity profile of the membrane to be computed That this takes the general formshown in Fig 12 is now beyond dispute

centrosym-Eleelron density

profiles of

450 400 350 300

Reference

electron densities Anhydrous protein

or lipid polar groups Hydrated lipid polar groups Water Paraffin (solid) Paraffin (liquid)

250

2nm 0 2nm

Fig 12 The generally observed form of electron density profiles derived from X-ray diffraction analyses

of multilamel1ar membrane systems, together with a calibration by reference to measured electron densities (from [44]).

The majority of electron density profiles obtained in such studies simply depictthe changes in relative electron densities through the thickness of a membrane, but

in a few cases absolute levels of electron density have been determined by lating the electron densities of the immersion medium (e.g with sucrose or glycerol)

manipu-By such means it has been established that the electron density at the centre of themyelin membrane [124,126] is approx 250 e/nm',a value which corresponds closely

to that of a pure liquid hydrocarbon The equivalent figure for multilayers of pairedgap junction membranes [127] approaches 300 e/nm3 which is in keeping with theextensive penetration of the lipid hydrocarbon interiors of the paired membranes bythe major gap junction protein

Diffraction data that are detailed enough to allow structural analyses at muchhigher resolution have been obtained from only a few membrane systems whichdisplay exceptional structural order either in terms of very extensive and regularstacking of membranes [125-128] or of very regular and extensive crystallinepacking of protein components within the plane of the membrane [18,64]

When the membrane stacking is extensive and regular the low-angle X-raydiffraction pattern may show well-defined "lamellar" diffraction out to the 13th to15th order reflection For membrane pairs giving a periodicity of 18-20 nm thiscould theoretically lead to a structural reconstruction with a resolution in the region

of 0.5- 1.0 nm However, the reliability of such analyses is still limited because ofresidual ambiguities in the phasing of the higher order reflections and of uncertain-ties in the measurement and computation of their intensities The phasing of thehigher order X-ray reflections generated by these centrosymmetric systems is gener-ally dependent on identifying reflections close to points at which the phase changessign (i.e intensities approach zero) and thereby demarcating groups of adjacentreflections which are of the same phase (+ or -) The relatively small number ofalternative combinations of signs for these groups can each be computed so as toprovide a structural solution which can be tested for its congruity with criticalstructural features suggested by other types of experimental data The validity of thepreferred structural solution depends on reliably measuring, correcting and phasingthe diffraction intensities and also on the application of a number of factors that areused in the interconversion between diffraction characteristics and structural param-eters Fortunately, uncertainties of phasing are usually confined to X-ray reflections

of low intensities which only have a small effect on the form of the electron densityprofile Several high resolution features of membrane electron density profiles havetherefore emerged with substantial, if not unanimous, support, and plausible struct-ural interpretations have been proposed for these For instance, the introduction ofthe higher order diffractions into the analyses of nerve myelin and artificiallystacked liver gap junction membrane pairs, both specialised forms of plasmamembranes, leads to the resolution of subsidiary shoulders or peaks on the slopes ofthe central low density trough of each membrane (Fig 13) Both of these membranesystems have a substantial cholesterol content, and comparisons with electrondensity profiles of phosphatidylcholine: cholesterol mixtures have indicated thatthese peaks could reflect the location of cholesterol Although the differences in the

heights of these peaks on the two slopes could reflect an asymmetric distribution ofcholesterol across the membrane bilayer [125], it has also been suggested that anasymmetric penetration of protein into the interior of the membrane might contrib-ute to these subsidiary peaks [126] Indeed, in the case of sarcoplasmic reticulumwhich contains little, if any, cholesterol, the subsidiary peaks in the low densityregion (Fig 13) are probably generated by protein that penetrates into the interior ofthe membrane.

In a few membranes a single protein predominates and this adopts a regular

packing pattern in the plane of the membrane (e.g the purple membrane of

Halobacterium halobium): diffractions corresponding to this pattern can be tinguished from those arising from the lamellar periodicity These then add anadditional dimension to the low resolution analysis of the membrane structure TheX-ray reflections from this membrane also give a clear indication that bacteriorho-

dis-dopsin, its sole protein, must have within its structure a-helices, the axes of which lieapproximately perpendicular to the plane of the membrane [129-131] This mem-brane has also yielded the highest resolution (0.7 nm) so far achieved in membranestructural studies, as a result of a sophisticated combination of electron diffractionand electron microscopy[18].Inthis analysis, low intensity and defocussed electronmicrographs of unstained membranes dried from 0.5% glucose provided the phaseinformation that was required for computation of a structure from the amplitudedata that had been obtained from the same sample by electron diffraction Informa-tion sufficient for a three-dimensional Fourier analysis of the structure was gathered

by recording micrographs and diffraction patterns with the membrane samples tilted

at a variety of angles with respect to the axis of the electron beam The resultingthree-dimensional electron density map identifies each bacteriorhodopsin molecule

Fig 14 Diagrammatic interpretation of the disposition of bacteriorhodopsin and lipid in the purple

membrane of Ha/obacterium ha/obium (adapted from (17,18)).

as a cluster of seven rod-like structures spanning the membrane: these are probablya-helices Since all seven helices form parts of a single polypeptide chain they must

be linked, at alternate surfaces of the membrane, by stretches of peptide chain thatare not resolved by the electron diffraction analysis Three such molecules aregrouped around a three-fold axis so that nine (3X3) a-helices form an inner ringand twelve (3X4) an outer one, with lipid bilayer filling the central (2nm diameter)space and also the other spaces between the clusters of protein molecules (Fig 14).There are approx 12 to 14 lipid molecules (mainly glycerol diether lipids withdihydrophytyl chains) per bacteriorhodopsin molecule Sequence studies of thesingle polypeptide chain of the bacteriorhodopsin molecule have led to tentativeassignments of particular peptide sequences to individual a-helices in the structure[64], and thence to a preliminary characterisation of a (proton) permeability channeldown the centre of the cluster of a-helices that constitute each individual molecule.There is thus a contrast between the proton channel through bacteriorhodopsinwhich penetrates through the interior of a single polypeptide component and thewater-filled channels (connexons) of gap junctions (see p.227) which are formedthrough the interaction of hexameric arrays of polypeptide components

References

I Nageli, C and Cramer, C (1855) Pflanzenphysiologische Untersuchungen, I Heft, Schultess, Zurich.

2 Pfeffer, W (1897) Osmotische Untersuchungen, Engelmann, Leipizig.

3 Overton, E (1899) Wschr Naturg Ges Zurich 44, 88-135.

4 Gorter, E and Grendel, F (1925) J Exp Med 41, 439-443.

5 Fricke, H (1925) Phys Rev 26,682.

6 Danielli, J.F and Davson, H (1935) J Cell Compo Physiol 5,495-508.

7 Stein, W.D and Danielli, J.F (1956) Disc Faraday Soc 21, 238-251.

8 Finean, J.B (1962) Circulation 26, 1151- 1162.

9 Robertson, J.D (1964) in M Locke (Ed.), Cellular Membranes in Development, Academic Press, New York and London, pp 1- 81.

10 Lucy, JA (1964) J Theor BioI 7, 360-373.

11 Sjostrand, F.S (1965) Symp Japan Soc Cell BioI 14 Suppl 103-125.

12 Green, D.E and Perdue, J.F (1966) Ann N.Y Acad Sci 137,667-684.

13 Lenard, J and Singer, S.J (1966) Proc Natl Acad Sci USA 56,1828-1835.

14 Benson, AA (1968) in L Bolis and B.A Pethica (Eds.), Membrane Models and the Formation of Biological Membranes, North-Holland, Amsterdam.

15 Vanderkooi, G and Green, D.E (1970) Proc Natl Acad Sci USA 66, 615-621.

16 Finean, J.B (1972) Sub-Cell Biochem 1,363-373.

17 Singer, S.J and Nicolson, G (1972) Science 175,720-731.

18 Henderson, R and Unwin, P.N.T (1975) Nature (London) 257, 28-32.

19 De Duve, C (1963) J Theoret BioI 6, 33-59.

20 Evans, W.H (1979) Lab Tech Biochem Mol BioI 7, 1-266.

21 Fleischer, S and Packer, L (Eds.) (1974) Methods in Enzymology, Vol XXXI.

22 Maddy, AH (Ed.) (1976) Methods of Biochemical Analysis, Chapman and Hall, London.

23 De Duve, C (1975) Science 189, 186-194.

24 Martonosi, A and Feretos, R (1964) J BioI Chern 239,648.

25 Meissner, G (1974) in (21), pp 238-246.

26 Schwoch, G and Passow, H (1973) Mol Cel1 Biochem 2, 197-218.

27 Billah, M.M., Finean, lB., Coleman, R and Michel1, R.H (1977) Biochim Biophys Acta 433,54-62.

28 Bramley, TA and Coleman, R (1972) Biochim Biophys Acta 290,219-228.

29 Billah, M.M., Finean, r.n., Coleman, R and Michel1, R.H (1978) Biochim Biophys Acta 465, 515-526.

30 Bramley, T., Coleman, R and Finean, J.B (1971) Biochim Biophys Acta 241, 752- 769.

31 Steck, T (1974) J Cel1BioI 62,1-19.

32 Kom, E.D (1978) Methods in Membrane Biology, Plenum, New York.

33 Glick, M.C (1976) in GA Jamieson and D.M Robinson (Eds.), Mammalian Cel1Membranes, Vol I, Butterworth, London, pp 45- 77.

34 Fleischer, S and Packer, L (Eds.) (1979) Methods in Enzymology, Vol LV.

35 Fleischer, S and Packer, L (Eds.) (1974) Methods in Enzymology, Vol XXXII.

36 Berry, M.N (1974) in [35], pp 625-633.

37 Seglen, P.O (1976) Meth Cel1 BioI 12,29-83.

38 Lentz, P.N and Di Luzio, N.R (1974) in [35], pp 647-655.

39 Hulsmann, W.e., van den Berg, J.W.O and de Jonge, H.R (1974) in [35], pp 665-673.

40 Jarett, L (1974) in [21], pp 60-71.

41 Willinger, M., Gonatas, N and Frankel, F.R (1979) J Cel1 BioI 82,45-56.

42 Walsh, F.S., Barber, B.H and Crumpton, M.J (1976) Biochemistry 15,3557-3563.

43 Luzio, P., Newby, A.C and Hales,c.x.(1976) Biochem J 154, 11-21.

44 Finean, J.B., Coleman, R and Michel1, R.H (1978) Membranes and their Cel1ular Functions, 2nd ed., Blackwel1, Oxford.

45 Jamieson, GA and Robinson, D.M (Eds.) (1977) Mammalian Cel1 Membranes, Vol 2, Butterworth, London.

46 Singer, S.J (1974) Annu Rev Biochem 43, 805-833.

47 Bretscher, M and Raff, M.e (1975) Nature (London) 258, 43-49.

48 Rawlins, FA, Villegas, G.M and Uzman, B.G (1977) in [45], pp 286-297.

49 Henderson, R (1977) Annu Rev Biophys Bioeng 6, 87-110.

50 Capaldi, RA (1977) in [45], pp 141-164.

51 Tanner, M.JA (1979) Curro Topics Membr Transp 12, I-57.

52 Marchesi, V.T., Furthmayr, H and Tomita, M (1976) Annu Rev Biochem 45, 667-698.

53 Steck, T.L (1974) J Cel1 BioI 62, 1- 19.

54 Steck, T.L (1978) J Supramol Struct 8, 311- 324.

55 Gopinath, R.M and Vincenzi, F.F (1977) Biochem Biophys Res Commun 77,1203-1209 55a Jarrett, H.W and Penniston, J.T (1977) Biochem Biophys Res Commun 77, 1210-1216.

56 Schulman, H and Greengard, P (1978) Proc Natl Acad Sci USA 75, 5432-5436.

57 Coleman, R (1980) in N Belton and e Toothil1 (Eds.), Problems of Transport and Metabolic Disease, MTP Press, Lancaster, pp 35-38.

58 Yu, J., Fischman, D and Steck, T.L (1973) J Supramol Struct 1,220.

59 Coleman, R., Holdsworth, G and Finean, J.B (1976) Biochim Biophys Acta 436,38-44.

60 Lux, S.E (1979) Nature (London) 281, 426-429.

61 Helenius, A and Simons, K (1975) Biochim Biophys Acta 415,29-79.

62 Helenius, A and Simons, K (1977) Proc NatI Acad Sci USA 74,529-532.

63 Furthmayr, H (1978) Nature (London) 274,519-524.

64 Engelman, D.M., Henderson, R., Mcl.achlan, A.D and Wallace, BA (1980) Proc NatI Acad Sci USA 77, 2023-2027.

65 Bjerrum, O.J and Bog-Hansen, T.e (1976) in [22], pp 378-425.

66 Hubbard, A.L and Cohn, ZA (1976) in [22], pp 427-501.

67 Ji, T.H (1979) Biochim Biophys Acta 559, 39-70.

68 Anderson, N.G and Anderson, N.L (1978) Anal Biochem 85, 331-354.

69 Rubin, R.W and Mil1ikowski, e (1978) Biochim Biophys Acta 509, 100-110.

70 Ralston, G.B (1978) Trends in Biochem Sci 3, 195-198.

71 Sheetz, M.P (1979) Biochim Biophys Acta 557, 122-134.

72 Bennett, V and Stenbuck, PJ (1979) Nature (London) 280, 468-473.

73 Knauf, R.A (1979) Curr Topics Membr Transp 12,249-363.

74 Passow, H and 'Zaiki, L (1978) in AK Solomon and M.L Karnovsky (Eds.), Molecular tion and Asymmetry in Membrane Function, Harvard University Press, Cambridge, pp 229-250.

Specialisa-75 Cabantchik, Z.I., Knauf, P.A and Rothstein, A (1978) Biochim Biophys Acta 5Specialisa-75, 239-302.

76 Mullins, R.E and Langdon, R.G (1980) Biochemistry 19, 1205-1212.

77 Jones, M.N and Nickson, J.K (1980) FEBS Lett 115, 1-8.

78 Mullins, R.E and Langdon, R.G (1980) Biochemistry 19, 1199-1204.

79 Brown, AP., Feinstein, M.B and Schaafi, R.1 (1975) Nature (London) 254, 523-525.

80 Jarnefelt, J., Rush, 1., Li, Y-T and Laine, R.A (1978) J BioI Chern 253, 8006-8009.

80a Shukla, S.D., Berriman, J., Coleman, R., Finean, J.B and Michell, R.H (1978) FEBS Lett 90, 289-292.

81 Gurr, M.1 and James, AT (1980) Lipid Biochemistry: An Introduction 3rd ed., Chapman and Hall, London.

82 Ansell, G.B., Dawson, R.M.e and Hawthorne, 1.N (Eds.) (1973) Form and Function of pids, Elsevier, Amsterdam.

Phospholi-83 Kushwaha, S.C., Kates, M and Stoeckenius, W (1976) Biochim Biophys Acta 426,703-710 83a Kagawa, Y (1980) J Membrane BioI 55, 1-8.

84 Hanson, B.A and Lester, R.L (1980) J Lipid Res 21, 309- 315.

85 Kaul, K and Lester, R.L (1975) Plant Physiol 55, 129-129.

86 Gray, G.M (1979) in O.S Estrada and e Gitler (Eds.), Perspectives in Membrane Biology, Academic Press, New York, pp 85-106.

87 Hakemori, S-1 (1975) Biochim Biophys Acta 417,55-89.

88 Koscielak, J., Miller-Podraza, H., Krauze, R and Piasek, H (1976) Eur 1 Biochem 71, 9-18.

89 Gardas, A (1976) Eur J Biochem 68,177-183.

90 Dejter-Jusynski, M., Harpaz, N., Flowers, H.M and Sharon, N (1978) Eur J Biochem 83, 363-378.

91 Rumsby, M.G and Crang, AJ (1977) in G Poste and G.L Nicholson (Eds.), Cell Surface Reviews, Vol 4, North-Holland, Amsterdam, pp 247-362.

92 Lewis, B.A., Gray, G.M., Coleman, R and Michell, R.H (1975) Biochem Soc Trans 3, 752- 753.

93 Nes, W.R (1974) Lipids 9,596-612.

94 Demel, R.A and De Kruyff, B (1976) Biochim Biophys Acta 109- 132.

95 Green, C (1977) in T.W Goodwin (Ed.), International Review of Biochemistry, Vol 14, University Park Press, Baltimore, MD, pp 101-152.

96 Thompson, G.A and Nozawa, Y (1977) Biochim Biophys Acta 472, 55-92.

97 Haines, T.H (1974) in T.W Goodwin (Ed.), MTP Int Rev Sciences: Biochemistry Series I, Vol.a, Biochemistry of Lipids, Butterworth, London, pp 271-286.

98 Lands, W.E.M and Crawford, CiG (1976) in A Martonosi (Ed.), The Enzymes of Biological Membranes, Vol 2, Plenum, New York, pp 3-85.

99 Kimelberg, H.K (1977) in G Poste and G.L Nicholson (Eds.), Cell Surface Reviews, VoU, Dynamic Aspects of Cell Surface Organisation, North-Holland, Amsterdam, pp 205-293.

100 Coleman, R (1973) Biochim Biophys Acta 300, 1-30.

101 Sanderman, H (1978) Biochim Biophys Acta 515, 209-237.

102 Cullis, P.R and De Kruyff, B (1979).Biochim Biophys Acta 559, 399-420.

103 Michell, R.H and Kirk, CJ (1981) Trends in Pharmacol Sci 4, 86-89.

109 Ladbrooke, B.D., Jenkinson, T.J., Kamat, V.B and Chapman, D (1968) Biochim Biophys Acta 164, 101-109.

110 Chapman, D and Peel, W.E (1979) in AD Keith (Ed.), The Aqueous Cytoplasm, Marcel Dekker, New York, pp 137-178.

III Finean, J.B., Coleman, R., Limbrick, AR and Green, W.G (1966) J Cell Sci 1,287-296.

112 Sha'afi, R.! (1977) in J.e Ellory and V.L Lew (Eds.), Membrane Transport in Red Cells, Academic Press, London, pp 221- 256.

113 Gilula, N.B (1978) in 1 Feldman, N.B Gilwa and J.D Pitts (Eds.), Intercellular Junctions and Synapse, Chapman and Hall, London, pp 1- 22.

114 Hicks, R.M., Ketterer, B and Warren, R.e (1974) Phil Trans Roy Soc London B 268,23-38.

115 Robertson, J.D and Vergara, J (1980) J Cell BioI 86, 514-528.

116 Unwin, P.N.T and Zampighi, G (1980) Nature 283,545-549.

117 Branton, D (1971) Phil Trans Roy Soc London B 261, 133-138.

118 Bullivant, S (1974) Phil Trans Roy Soc London B 268, 5-14.

119 Stolinski, C and Breathnach, AS (1975) Freeze-Fracture Replication of Biological Tissues, Academic Press, London.

120 Shipley, G.G (1973) Biological Membranes, VoI.2, Academic Press, London, pp 1-89.

121 Blaurock, AE (1981) Biochim Biophys Acta, in press.

122 Finean, J.B and Burge, RoE (1963) J Mol BioI 7 0 6 7 2 - 6 8 2

123 Moody, M.F (1963) Science, 142,1173-1174.

124 Blaurock, AE (1971) 1 Mol BioI 56, 35-52.

125 Caspar, D.L.D and Kirschner, D.A (1971) Nature New BioI 231,46-52.

126 Nelander, J.e and Blaurock, AE (1978) J Mol BioI 118,497-532.

127 Makowski, L., Caspar, D.L.D., Phillips, W.e and Goodenough, D.A (1977) J Cell BioI 74, 629-645.

128 Herbette, L., Marquardt, J., Scarpa, A and B1asie, J.K (1977) Biophys J 20, 245-272.

129 Henderson, R (1975) J Mol BioI 93,123-138.

130 Blaurock, AE (1975) J Mol BioI 93,139-158.

131 Blaurock, AE and King, G.! (1977) Science 196, 1101- 1104.