Cell fusion

Ciba Foundation symposium 103

1984

Pitman London

Cell fusion

The Ciba Foundation is an international scientific and educational charity. It was established in 1947 by the Swiss chemical and pharmaceutical company of CIBA Limited-now CIBA-GEIGY Limited. The Foundation operates independently in London under English trust law.

The Ciba Foundation exists to promote international cooperation in biological, medical and chemical research. It organizes about eight international multidisciplinary symposia each year on topics that seem ready for discussion by a small group of research workers. The papers and discussions are published in the Ciba Foundation symposium series. The Foundation also holds many shorter meetings (not published), organized by the Foundation itself or by outside scientific organizations. The staff always welcome suggestions for future meetings.

The Foundation’s house at 41 Portland Place, London, WIN 4BN, provides facilities for all the meetings. Its library, open seven days a week to any graduate in science or medicine, also provides information on scientific meetings throughout the world and answers general enquiries on biomedical and chemical subjects. Scientists from any part of the world may stay in the house during working visits to London.

Cell fusion

Ciba Foundation symposium 103

1984

Pitman London

0 Ciba Foundation 1984

ISBN 0 272 79750 2

Published in February 1984 by Pitman Publishing Ltd., 128 Long Acre, London WC2E 9AN, UK. Distributed in North America by CIBA Pharmaceutical Company (Medical Education Division), Post Office Box 12832, Newark, NJ 07101, USA

Suggested series entry for library catalogues: Ciba Foundation symposia

Ciba Foundation symposium 103 x + 291 pages, 62 figures, 23 tables

British Library cataloguing in publication data:

Symposium on Cell Fusion (1983: Ciba Foundation, London) Cell fusion.-(Ciba Foundation symposium; 103) 1. Cel ls4ongress 2. Biophysics- Congresses I. Title 11. Evered, David 111. Whelan, Julie IV. Series 574.87’0604 1 QH611

ISBN 0-272-79750-2

Printed in Great Britain at The Pitman Press, Bath

Contents

Symposium on Cell fusion, held at the Ciba Foundation, London, 17-1 9 May

The subject for this symposium was proposed by Dr John Mayer Editors: David Evered (Organizer) and Julie Whelan

1983 \

B. A. PETHICA (Chairman) Introduction 1

V. A. PARSEGIAN, R. P. RAND and D. GINGELL Lessons for the study of membrane fusion from membrane interactions in phospholipid systems 9 Discussion 22

J. A. LUCY Fusogenic mechanisms 28 Discussion 39

A. J. VERKLEIJ, J. LEUNISSEN-BIJVELT, B. de KRUIJFF, M. HOPE and P. R. CULLIS Non-bilayer structures in membrane fusion 45 Discussion 54

U. ZIMMERMANN, J. VIENKEN, G. PILWAT and W. M. ARNOLD Electro-fusion of cells: principles and potential for the future 60 Discussion 73

B. M. SHAPIRO Molecular aspects of sperm-egg fusion 86 Discussion 95

M. J. 0. WAKELAM and D. PETTE Myoblast fusion and inositol phospholipid breakdown: causal relationship or coincidence? Discussion 113

100

E. C. COCKING Plant-animal cell fusions 119 Discussion 123

D. DOYLE and H. BAUMANN Transfer of plasma membrane proteins between cells using reconstituted membrane vesicles as shuttle vehicles 129 Discussion 144

V

vi CONTENTS

M. DAS and S. BISHAYEE Insertion of EGF receptors into target cells in the absence of fusogenic agents Discussion 158

1-50

A. LOYTER, M. TOMASI, A. G. GITMAN, L. ETINGER and 0. NUSSBAUM The use of specific antibodies to mediate fusion between Sendai virus envelopes and living cells Discussion 175

163

M. RECHSTEINER, D. CHIN, R. HOUGH, T. McGARRY, S. ROGERS, K. ROTE and L. WU What determines the degradation rate of an injected protein? 181 Discussion 196

R. J . MAYER, P. EVANS, S. RUSSELL and J . S. AMENTA Degradative fate of transplanted proteins 202 Discussion 215

J . E. CELIS Expression of mRNAs microinjected into somatic cells 220 Discussion 232

J. GUYDEN, W. GODFREY, B. DOE, F. OUSLEY and L. WOFSY Im- munospecific vesicle targeting facilitates fusion with selected cell populations 239 Discussion 249

C. NICOLAU, A. LEGRAND and P. SORIANO Liposomes for gene transfer and expression in vivo Discussion 264

254

Final general discussion Comments on the status of the bilayer concept of biomembranes 268 Mode of action of polyethylene glycol 271 Microinjection and protein degradation studies 274 Mechanism of cell fusion by viruses 275 Physiological cell fusion 277

Index of contributors 281

Subject index 283

Participants

D. ALLAN Department of Experimental Pathology, School of Medicine, University College London, University Street, London WClE 655, UK

W. M. ARNOLD Arbeitsgruppe Membranforschung am IME im Gebaude, Institut fur Chemie, Kernforschungsanlage Julich GmbH, Postfach 1913, D-5170 Julich, Federal Republic of Germany

A. ASANO Department of Biochemistry, Cancer Research Institute, Sap- poro Medical College, South 1 West 17, Sapporo 060, Japan

A. D. BANGHAM The Cottages, 17 High Green, Great Shelford, Cam- bridge, UK

H. BAUMANN Department of Cell and Tumor Biology, Roswell Park M.emoria1 Institute, Buffalo, NY 14263, USA

J. E. CELIS Division of Biostructural Chemistry, Department of Chemi- stry, Aarhus University, Langelandsgade 140, DK-8000 Aarhus C, Denmark

E. C. COCKING Department of Botany, School of Biological Sciences, University of Nottingham, University Park, Nottingham NG7 2RD, UK

M. DAS Department of Biochemistry and Biophysics, University of Penn- sylvania, School of Medicine, Philadelphia, PA 19104, USA

J. F. DICE Department of Physiology and Biophysics, Harvard Medical School, 25 Shattuck Street, Boston, MA 02115, USA

F. J. DOHERTY Department of Biochemistry, University Hospital and Medical School, Clifton Boulevard, Nottingham NG7 2UH, UK

D. GINGELL Department of Biology as Applied to Medicine, Middlesex Hospital Medical School, Mortimer Street, London W1P 7PN, UK

J. GUYDEN Department of Microbiology and Immunology, University of California, Berkeley, CA 94720, USA

vii

PARTICIPANTS ...

Vl l l

K. B. HENDIL August Krogh Institute, 13 Universitetsparken, 2100 Copenhagen 0, Denmark

A. LOYTER Department of Biological Chemistry, Institute of Life Sciences, The Hebrew University of Jerusalem, Jerusalem 91904, Israel

J. A. LUCY Department of Biochemistry, Royal Free Hospital School of Medicine, Rowland Hill Street, London NW3 2PF, UK

B. MAGGIO Department of Biological Chemistry, Faculty of Chemical Sciences, National University of Cdrdoba, SUC. 16, Casilla de Correo 61, 5016 Cordoba, Argentina

R. J. MAYER Department of Biochemistry, University Hospital and Medical School, Clifton Boulevard, Nottingham NG7 2UH, UK

C. NICOLAU Centre de Biophysique MolCculaire, Avenue de la Recher- che Scientifique, 45045 Orleans Cedex, France

V. A. PARSEGIAN Room 1007, Building 12A, National Institutes of Health, Bethesda, MD 20205, USA

B. A. PETHICA Electrobiology Inc., PO Box 682, 300 Fairfield Road, Fairfield, NJ 07006, USA

M. C. RECHSTEINER Department of Biology, 201 Biology Building, University of Utah, Salt Lake City, UT 84112, USA

R. A. SCHLEGEL Molecular and Cell Biology Program, 101 South Frear Building, Pennsylvania State University, University Park, PA 16802, USA

B. M. SHAPIRO Department of Biochemistry, University of Washington, Seattle, WA 98195, USA

A. J. VERKLEIJ Institute of Molecular Biology, Rijkuniversiteit Utrecht, Padualaan 8, PO Box 80.063, 3508 TB Utrecht, The Netherlands

M. J. 0. WAKELAM Department of Biochemistry, Imperial College of Science and Technology, Exhibition Road, London SW7 2AZ, UK

K. WILLECKE Institut fur Zellbiologie, Universitat Essen, Hufelandstr. 55, 4300 Essen 1, Federal Republic of Germany

PARTICIPANTS ix

U. ZIMMERMANN* Arbeitsgruppe Membranforschung am IME im Gebaude , Institiit fur Chemie, Kernforschungsanlage Julich GmbH, Post- fach 1913, D-5170 Jiilich, Federal Republic of Germany

* In absentia.

Introduction

B. A. PETHICA

Elcctrobiology lnc. , P. 0. Box 682, 300 Fairfield Road, Fairfield, NJ 07006, USA

In introducing the symposium I want to encourage later contributors to bring out matters of contention, and to draw on their own experience to identify the issues that we should be debating. It is not our function simply to tell each other about the latest results, but rather to pin down and attack some of the key issues. What impresses me, as a physical chemist, is that the processes of cell fusion are so much richer and more active than the kind of event that the physical chemist usually studies. I therefore want to emphasize at the outset the flux of chemical reactions-the whole business of metabolism-that attends cell fusion, over and above the events taking place in the rather dead-looking models with which physical chemists feel more comfortable.

The subject of the symposium is cell fusion, and this should be emphasized. One can make distinctions between physiological or natural cell fusions as exemplified by the myoblast, pathological cell fusion as mediated by viruses, and a group in which unnatural events like the addition of polyethylene glycol or the use of pulsing electric fields will induce cell fusion. In any one of these instances the cell fusion process is unambiguous: an ordinary microscope is all you need to tell whether you have cell fusion or not. But in reading the literature I have been struck by the fact that the mechanistic problems of cell fusion have been largely transposed to another area, called membrane fusion. As a physical chemist, I find myself unconvinced that membrane fusion, as currently defined, is necessarily part of the process of cell fusion. There is no doubt that for two cells to form one, there is a process of fusion, but the question is whether the membranes ‘fuse’. Is it essential to cell fusion that sizeable regions of two membranes come together and mix, or do they come together at a chosen macromolecule, from which ‘point’ some sequence of rearrangements and metabolic chemical events spreads out the new mem- brane? The widespread assumption that cell fusion is usually mediated by one or more cooperative mechanisms of ‘membrane fusion’ seems unnecessary. Given the intensity of the hunt and the problems with electron microscope

1984 Cell fusion. Pitman Books, London (Ciba Foundation symposium 103) p 1-8

1

2 PETHICA

artefacts, 1 find myself unconvinced by the curiously scanty and contentious evidence for mixed septa, transient pores, fused regions and so on as typical morphological intermediates in cell fusion. I am particularly unimpressed by the evidence from sonicated lipid vesicles that fusion occurs between them. To the extent that vesicle fusion is doubtful, membrane fusion between vesicles seems equally doubtful. There are other physicochemical mechan- isms than fusion which can account for changes in vesicle size distribution, a process that bears much resemblance to the Ostwald ripening process in colloids.

Here, then, we have questions of basic mechanistic interest. Does cell fusion start at a ‘point’-perhaps a macromolecule-or at a set of points? Or is fusion cooperative, as for example that two opposing bilayers come together to form a mixed structure as a precursor to cytoplasmic mixing‘? Each mechanism would entail a very different kinetic programme for the fusion process and correspondingly different activation energies. After initia- tion, is the joint membrane development a mixing of the constituents already there, or, once the cytoplasms of two cells have begun to mix, is there a joint biochemical programme whereby a new kind of membrane is synthesized? And, indeed, need one mechanism apply to different cell types or the various classes of fusogens?

Some of you wiIl know that I am a sceptic about the lipid bilayer model of cell membranes, particularly in the form of the Singer-Nicolson fluid mosaic account. For animal cells, I see no proof that the bilayer exists as a general model. For some cell types, such as the human red cell, I think there is long-standing evidence amounting to proof that it does not exist (Clifford et a1 1968). Yet, when new observations are made, there is an almost automatic tendency to fit them into the lipid bilayer model, which has almost become an intellectual Procrustean bed. With different assumptions, would we get diffe- rent insights-and a richer inspiration for experiment? Happily, there is a growth of enquiry into other models, giving macromolecules and alternative lipid arrangements a larger role in the basic membrane structures.

The question of whether the membrane after the fusion of identical cells is the same as before is worth asking. With heterofusion, the cytoplasmic and internal membranes may both be inhomogeneous, in the sense that particular regions reflect the post-fusional memory of pre-fusion genetic control factors. The mixing implied by the word ‘fusion’ could be less than supposed. This reflection prompts me to ask whether a process of ‘mitosis’ of a fused homo- or heterokaryon has ever been observed, and whether such separated cells would necessarily be the same as the pre-fusion types?

One can attempt a ‘natural history’ of cell fusion, starting with the thought that the cells are producing fluxes of free energy and matter by metabolism, with many molecules turning over and being broken down, synthesized or

INTRODUCTION 3

resynthesized. The cell is not an inert bag to which we can apply the ordinary rules of statistical mechanics or colloidal science. If we think of two cells initially isolated in suspension and we follow their route to fusion, it is evident that they have first to approach each other. The much-canvassed long-range forces, such as those involving electrical double layers, van der Waals interac- tions and steric repulsions, are certainly relevant here (Pethica 1980). I would note the neglect of another set of forces relevant to the approach of living cells, namely those arising from the gradients in the chemical potentials of metabolites near the cells. Thus, in addition to the forces appropriate to dead colloids, we should include terms that take account of metabolic activity in cells. As two cells approach, the two sets of individual flux profiles overlap, and this overlap itself constitutes a force between the cells, which can be attractive or repulsive. A simple way to visualize this intercellular force due to metabolic fluxes is to note that every living cell is surrounded by an osmotic pressure gradient which can be radially positive or negative, depending on the balance of the metabolism. Overlap of two such gradients gives a net osmotic repulsion or attraction. Nor should we overlook the interactive relation of these forces to local metabolism itself, through the dependence of the chemic- al reaction rates on the local concentrations. No quantitative calculations of intercellular forces based on flux interactions seem to be available. Until they are, biologists are well within their rights in taking the cell-interaction accounts of physical chemists with a grain of salt. One of the few related calculations available shows that the electrical double layer on the surface of a cell can be overwhelmed by the diffusion potential resulting from realistic enzyme action involving the production of ions at the cell surface (DeSimone 1977). So the mere (?) fact of metabolic activity can produce fluxes to swamp the colloid scientists’ favourite long-range forces.

The approach of two cells also depends on the curvature of the opposing membrane elements, including villi, pili etc. Correspondingly, adhesion can be much assisted by the intervention of suitable submicron particles to act as bridges (Pethica 1980). Both viruses and inorganic particles can play this role as adhesion intermediates and markedly alter the whole approach and contact profile. Once two cells are joined by a mutually contacting particle, a variety of events can lead on to fusion-membrane components spreading across the bridge, local chemical reaction shifts, and so on.

When the cells are in close apposition, all the foregoing factors operate, but the short-range stereochemical forces dominate. These are the detailed forces underlying ‘specific’ effects such as antibody reactions and type-dependent cell adhesion.

From apposition the cells move into adhesion or contact, characterized by membrane structures such as the gap junction. From this stage they can move on to functional contact, a term borrowed from Loewenstein (1981) meaning

4 PETHICA

that the two cells have electrical and diffusional interconnection for molecules of modest molecular weight. This has been well established for non-fusing tissue cells, and is observed as a prior state for the fusion of myoblasts. Functional contact is mediated by macromolecular structures, which Loewen- stein calls connexons. These are instructive concepts relevant to the final step of interest to us, namely cell fusion. Is functional contact, demonstrated for myoblasts, a typical intermediate state on the route to cell fusion, or are there routes by which the cells go directly from contact to fusion? If functional contact is a frequent precursor state for fusion, we can import into fusion analysis the methods developed by Loewenstein and others for looking at the number and quanta1 development of connexons, which might be envisaged as lipid-polymer macromolecular clusters equally belonging to both membranes, nucleating cell fusion.

This overall sketch of the ‘natural history’ of the fusion of two cells is represented in Fig. 1. Plainly, if the two cells are initially in contact with other cells, their separation processes will have to be included in the scheme. Processes such as cytokinesis or chemotaxis, the latter probably a long-range case of a diffusion-related force, are also part of the approach scheme. How far the sketch will be complete for the fusion of cytoplasmic organelles or for the passage of vesicles through the plasma membrane in secretion is not so clear. Structures such as connexons in these processes are not entirely fanci- ful, but are not established entities.

There are two further groups of interactive membrane changes that the physical chemist can usefully point out as relevant to cell contact and fusion. Firstly, even if we regard membranes just as dead multicomponenf equilib- rium structures, leaving aside the metabolic fluxes, it is necessary by a form of Le Chatelier’s principle that if the,interaction force varies with the composi- tion (or more strictly, with the chemical potentials of the membrane consti- tuents), the local concentrations of the membrane constituents must vary with the separation distance (Pethica 1980). If components in the membrane enhance attraction as the cells come together, they will tend to be concen- trated in the junction region. If there are components that enhance repulsion, they will tend to be pushed out as the cells come together. These changes will occur at rates depending on viscosity and other local resistances. To give an example, the fact of the double layer repulsion between negatively charged cells tells us that as the cells are‘ pushed together, groups like the ionized carboxylate group will tend to disappear in favour of un-ionized carboxyl, and amino groups will tend to be charged up to the cationic state. These chemical changes will be most pronounced for ionizable groups with pK values close to neutrality. For the van der Waals forces, which are usually attractive, we know from calculation that sugars attract each other more than do proteins, lipids, and water, in that order (Nir 1977). We would, therefore, expect to

INTRODUCTION

Process

5

Forces and factors

APPROACH

1 APPOSl TI ON

1 CONTACT

Electrical double layers/van der Waals interactions S ter ic /en t ro p ic interact ions METABOLIC FLUX GRADIENTS Local curvature - pil i , vi l l i , etc. Part ic I e bridging

All the above, plus short-range (stereochemical) 'spec i f i c ' interact ions

All the above, leading to formation of close membrane structures Chemical modification in contact regions Membrane potentials small

Electrical /diffusional communication CONNEXONS

CELL FUSION 'Point' initiation of common membrane Cooperative membrane fusion '? Double system of organelle membranes in heterokaryons ? Reversible ?

?

FIG. 1 . A natural history of cell fusion. Schema for the overall process of fusion of two initially separated cells.

find a weak concentration of molecules like glycoproteins in the junction region with a tendency to some local dehydration, if van der Waals forces are dominant. With steric forces, which are usually repulsive, we expect to find that highly branched flexible polymers would be pushed out of the junction. Of course, I am considering an equilibrium membrane here. When a real membrane meets polymeric resistance, it may have the means to bore through it, for example by using a neuraminidase to dispose of the opposing barrier. Obviously, since the local compositions necessarily change on approach and contact, these changes can give rise to many biochemical sequels in living cells as the local enzymic and transport activities are mod- ified. Membranes in contact are certainly not the sum of the initial mem- branes, in composition or functionality.

6 PETHICA

The second group of interactive factors to which I would like to draw attention relates to the membrane potential. We are accustomed to think that all cell membranes have a membrane potential. That is not so. The membrane potential is directly a measure of two factors, namely that there is an ionic flux across the membrane, and that the transport coefficients of each ionic species are not equal. If two membranes get close together over a sufficient area, the membrane potential must, therefore, go to zero in the contact region for contacts between two identical cells, since the concentration gradients across the membranes tend to zero in that region. Correspondingly, the membrane potentials become small in the contact region with two different kinds of cells approaching each other. Away from the contact region, of course, it depends on what the contact has done to the metabolic activity of the cells, and to the transport properties of the non-contacting membranes, as to whether the potentials go up or down, or even both, at different times in the proceedings.

There are a number of consequences to these electric potential changes (Pethica & Hall 1982). Firstly, in the contact region the membrane composi- tion undergoes changes in addition to those discussed above. If the electrical capacity of the membrane depends on the chemical potential of a given component, then the local concentration of that component varies with the square of the field. The field changes we are concerned with are more than lo5 voltscm-', so they are not small. Since the more polarizable molecules are pulled in by a field, it follows that if the field is abolished, these molecules will tend to move away. Thus one would expect that as the membrane is locally depolarized on contact, some proteins will leave and lipids will tend to remain. Secondly, we would expect to find orientation polarization changes on contact depolarization. Last, but not least, alterations in the mechanical tension of the membrane will also occur. All these field-dependent changes will trigger or modify the local metabolic processes in numerous ways. The analogues of these field-dependent membrane processes have been predicted and measured recently in monolayers (Middleton & Pethica 1981).

These two groups of simple interactive changes resulting from contact of membranes are the results of the operation of ordinary physical forces. To point them out is to emphasize the basic dynamic character of contact, even for non-living systems. Living systems are well endowed with facilities for utilizing these interactive changes to trigger new or altered biochemical events, in addition to producing the metabolic flux forces mentioned earlier.

We all tend to read the literature with our own expertise as a filter. As a committed thermodynamiter, I naturally hunted the fusion literature for thermodynamic evidence, and I find that systematic studies on the effect of temperature are rather few. An apparent Arrhenius coefficient of 18 kcal is reported for myoblast fusion (Fisher 1974). The fact that an Arrhenius plot can be used is itself striking, and'the activation energy is reminiscent of a

INTRODUCTION 7

iimple chemical reaction. It would be hard to argue that this is a cooperative effect such as the fusion of two bilayers! It may be more congenial to suggest that the activation occurs at a molecular point, or at a set of points in sequence, for the initiation of fusion. This is what one would also expect from electrical data on contacting myoblasts (Rash & Fambrough 1972), and agrees with some interpretations of the electron microscopy evidence (Bis- choff 1978, Lipton & Konigsberg 1972). Obviously, further data on tempera- ture coefficients would make a useful impact on the physical interpretation of cell fusion.

I believe that less emphasis on lipids and lipid bilayers will stimulate some new measurements on two-dimensional arrangements of proteins and lipoproteins. I tend to think of lipids in membranes rather as Hamlet in the graveyard thought of ‘poor Yorick’, a man of infinite jest who had come, alas, to stopping up a bung-hole! The lipids are multifunctional and essential to the activities of many proteins, including membrane proteins, and they are cer- tainly well suited to bung up all the membrane holes. We have been per- suaded that the lipids provide the basic membrane structure in the form of a bilayer, with the proteins spread alongside or present as membrane-spanning particles or chains. My inclination is to look for more information about lipoproteins and to see the mixed hydrophobic core of the membranes in this light. I would stress that any correlation between biomembrane processes such as cell fusion and events in a lipid monalayer, liposome or vesicle will almost certainly have a direct parallel in the behaviour of lipoproteins or of hydrophobic proteins, or portions of such proteins. This can be asserted because underlying the behaviour of lipoproteins and lyophobic proteins in aqueous systems is our old friend, the hydrophobic bond, which is also one of the main factors determining the behaviour of lipid micelles, monolayers and bilayers. The sort of correlation we find between lipid properties and mem- brane behaviour will certainly apply for other hydrophobic agents in large measure.

Let me conclude this physical chemist’s eye-view of the vast field of cell fusion with a mild apology for over-attention to the mechanism of the fusion process, and note that we shall have rich opportunity to hear out the diversity of the subject from protein biochemists, geneticists and many others in the symposium. We shall surely learn from each other, and we shall doubtless reflect on the poverty of models and mechanisms so far enunciated, by contrast with the luxuriance of the biological phenomena.

8 PETHIC A

REFERENCES

Bischoff R 1978 Myoblast fusion. In: Poste G , Nicholson GL (eds) Membrane fusion. Elsevierl North-Holland Publishing Company, Amsterdam (Cell Surface Reviews Ser, vol5) p 128-179

Clifford J, Pethica BA, Smith EG 1968 A nuclear magnetic resonance investigation of molecular motion in erythrocyte membranes. In: Bolis L, Pethica BA (eds) Membrane models and the formation of biological membranes. North-Holland Publishing Company, Amsterdam, p 19-42

DeSimone JA 1977 Perturbations in the structure of the double layer at an enzymic surface. J Theor Biol68:225-240

Fisher D 1974 Fluidity and cell fusion. In: Burton RM, Packer L (eds) Biomembranes, lipids, proteins and receptors. B.I. Science Publ. Division, Webster Groves, Missouri, p 75-93

Lipton BH, Konigsberg IR 1972 A fine-structural analysis of the fusion of myogenic cells. J Cell Biol 53:348-364

Loewenstein WR 1981 Junctional intercellular communications. The cell-to-cell membrane chan- nel. Physiol Rev 61:829-913

Middleton SR, Pethica BA 1981 Electric field effects on monolayers at the air-water interface. Faraday Symposium No. 16, p 109

Nir S 1977 Van der Waals’ interactions between surfaces of biological interest. Prog Surf Sci

Pethica BA 1980 Microbial and cell adhesion. In: Berkeley RCW et a1 (eds) Microbial adhesion

Pethica BA, Hall DG 1982 Electric field effects on membranes. J Colloid Interface Sci 85:41 Rash JE, Fambrough D 1972 Ultrastructural and electrophysiological correlates of cell coupling

8:1-58

to surfaces. Society of Chemical Industry, Ellis Horwood, Chichester, p 19-45

and cytoplasmic fusion during myogenesis in vitro. Dev Biol 30:166-186

Lessons for the study of membrane fusion from membrane interactions in phospholipid systems

V. A. PARSEGIAN*, R P. RAND** and D. GINGELLt

*Physical Sciences Laboratory, DCRT, National Institutes of Health, Bethesda, MD 20205, USA, **Department of Biological Sciences, Brock University, St Catharine’s, Ontario, L2S 3Al Canada and fDepartment of Biology as Applied to Medicine, Middlesex Hospital Medical School, London W l P 6DB, UK

Abstract. ‘Fusion’ in model systems usually refers to the decay of membrane configura- tions that are inherently unstable because of the method of preparation. Natural fusion is a controlled event during which the underlying forces and instabilities are subject to the additional effects of biochemical reactions. To understand biological fusion one must be able first to assess the interplay among these physical and chemical factors.

This paper reviews traditional measurements of electrostatic double layer and elec- trodynamic van der Waals forces acting between bilayer membranes. It also describes the much stronger hydration forces that have now been systematically studied.

An essential part of any fusion event is the ability of membrane surfaces to overcome or circumvent the hydration barrier in order to make contact. This may be accomplished through applied force, through bridging substances that displace water from the mem- brane surface, or through biochemical modification of surfaces. In model systems, destruction of the hydration layer can cause violent adhesion, membrane deformation, and rupture. Natural fusion proceeds by more subtle processes whereby interfacial forces are harnessed in ways not yet understood.

1984 Cell fusion. Pitman Books, London (Ciba Foundation symposium 103) p 9-27

‘Fusion’ is fashionable. Vital cellular functions depend on it. Many laborator- ies study it or model it. We suspect that much of its attraction lies in the assumption that biological fusion is a physical process related to the poly- morphism of lipid aggregates.

Biological fusion involves a topological transformation wherein either one closed membrane-bounded volume becomes two (endocytosis) or two be- come one (exocytosis or cell-cell fusion). There is good evidence that clean, leakless fusion occurs in cellular exocytosis, since cytoplasmic enzymes do not spill out. Whether true fusion is occurring in the many model systems remains contentious.

9

10 PARSEGIAN ET AL

We believe that the biological process is unlikely to be purely physical; cellular biochemical activity acts to modulate molecular properties and to provide the observed tight control of the fusion process. By identifying the various forces acting within and between model membranes, we can begin to see how these same forces act under the constraints set by natural systems.

In the following we shall discuss what is known of forces between model phospholipid bilayer membranes, emphasizing those situations where interac- tions are strong enough to cause deformation and rearrangement of the lipid bilayer. The common view of these forces has traditionally relied on ideas from colloid science. Specifically, it has been thought that electrostatic double layer repulsion , polymer-conferred steric stabilization, molecular bridging, and van der Waals attraction combine to explain the aggregation or adhesion of cells.

To this list one must add the enormous hydration forces that dominate most bilayer membrane interactions at distances less than 2 or 3nm. It is in fact these forces, resulting from the work of removing water from the polar groups stabilizing the membrane, that we hold to be of primary concern in the close juxtaposition of membranes that occurs during any fusion event.

Forces between phospholipid bilayers

There is now a fairly extensive literature on the interactions between bilayers in multilayer arrays (LeNeveu et a1 1976, 1977, Cowley et a1 1978, Parsegian et a1 1979, Lis et a1 1981a,b, 1982, Rand 1981, Loosley-Millman et a1 1982). Our method measures the osmotic pressure needed to remove water or salt water from the ordered array and then uses X-ray diffraction to ascertain the ensuing structural changes. This approach is limited to situations where there is a net repulsive force between the bilayers. Attractive forces can be inferred from the external pressure required to balance interbilayer repulsion, ex- trapolated to cases where the multilayer array is in equilibrium with excess water at zero applied pressure.

Under the stress of the removal of water, the bilayers not only come closer together but also become compressed laterally, packing their polar ends more tightly together and thickening the bilayer. This response demonstrates the simultaneity of bilayer interaction and deformation. It also provides a means of measuring deformability along with interaction.

The first measurements, on electrically zwitterionic egg phosphatidylcho- line (LeNeveu et a1 1976, 1977), showed a repulsive force beginning to act at 2.5 to 3.0nm separations and growing exponentially with a 0.2 to 0.3nm characteristic distance. This form of repulsion was seen down to separations of only about 0.4nm (Parsegian et a1 1979). The force appeared negligibly

MEMBRANE INTERACTIONS IN PHOSPHOLIPID SYSTEMS 11

dependent on the ionic composition of the medium and (Cowley et a1 1978) on any amount of electrical charge incorporated into the bilayers.

Because this unexpected interaction was seen even in distilled water, it was concluded that its origin was solvation of the polar head groups of the lipid molecules but at distances much larger than hydration shells had previously been suspected of reaching. Hence the designation ‘hydration force’.

Further studies on a large variety of charged and uncharged bilayers showed similar behaviour at separations less than 2.0 to 3.0 nm, regardless of ionic strength or degree of lipid charge (Lis et a1 1981a,b, 1982, Rand 1981, Loosley-Millman et a1 1982). Qualitatively similar interactions have been reported between mica surfaces hydrated by the adsorption of ions (Israelach- vili & Adams 1978, Pashley 1981, Pashley & Israelachvili 1981) or by coating with lipids (R. Horn, personal communication 1981). It now seems that exponentially decaying solvation forces are ubiquitous. They have recently been observed between DNA double helices in a wide variety of aqueous solutions (Rau et a1 1983). Phosphatidylcholine bilayers in glycol rather than water also repel exponentially over a range of 2-3 nm (P. K. T. Persson & B. A. Bergenstahl, personal communication 1983), where glycol acts as the ‘hydrating’ molecule.

In both phospholipid and DNA preparations in water, the characteristic decay length varies only slightly from one system to the next but the coefficient of the repulsive pressure is strongly dependent on the properties of the polar surface.

Marcelja and his collaborators (Marcelja & Radic 1976, Gruen & Marcelja 1982, Gruen et a1 1983) have developed an intuitive and appealing formalism to relate the observed exponential solvation repulsion to the perturbation of solvent at a polar interface. The central theme of their approach is that the surface perturbs the water immediately adjacent to it and that the successive perturbation of water away from the surface is due to water-water interac- tions. That secondary perturbation reflects properties of water as a solvent; it is these properties that lead to the exponentially decaying behaviour des- cribed above. It is the primary or surface perturbation that determines the actual strength of the force.

Beyond 2 to 3nm, solvation factors cease to dominate interactions; the forces between bilayers become exceedingly sensitive to the identity and charge of the lipid polar groups and to the ionic composition of the suspending medium. We have learnt to speak of this as the ‘electrostatic regime’, in distinction to the inner regime where hydration forces predomin- ate (Fig. 1).

Phospholipid bilayers may be charged in two ways, either by being composed of lipids with ionizable groups or by the adsorption of mobile charge to an otherwise neutral surface. In either case the decay of the

12 PARSEGIAN ET AL

REPULSIVE PRESSURE VS. SEPARATION

n N

E s Q)

c 2\ TI W

n W a 3 cn cn W [r a W

c3 0

l o 1 +2

8 r 6 I]

+I

0.0

-1

-2

> c3 tY W Z W

z 0

a z > I c3 s

I 0 20 30 40 50 60

SEPARATION (A) FIG. 1. Schematic logarithmic plot showing the influence of hydration and electrostatic and van der Waals forces between bilayers that are electrically neutral (dashed line) or charged (solid line). Right-hand scale shows the energy (not force) of the hydration interaction alone.

repulsive force with interbilayer separation in the electrostatic regime is sensitive to ionic strength, and the force increases with membrane charge density. At very low ionic strengths (Cowley et a1 1978), the electrostatic force decays much as expected from standard double layer theories. How- ever, at higher salt concentrations, certainly by the beginning of the physiolo- gical range, electrostatic repulsion decays more slowly than expected from the predictions of double layer theory. For example, for dipalmitoylphosphatidyl- choline in 30mM-CaC12, the observed exponential decay rate is some 30% slower than predicted by theory (Lis et a1 1981a). For phosphatidylglycerols or phosphatidylserines in various univalent cation solutions, the rate of decay of electrostatic repulsion is always slower than expected theoretically but depends on the cationic species (Loosley-Millman et a1 1982) that may be binding to the negatively charged bilayer surface (Eisenberg et a1 1979).

MEMBRANE INTERACTIONS IN PHOSPHOLIPID SYSTEMS 13

Radic & Marcelja (1978), using thinking similar to that used in deriving the hydration force, have argued that there is an extra energy associated with the rapid change of an electric field in water and that recognition of this factor yields a correction which acts to stretch out the electrostatic double layer and to predict a slower decay of forces. Still, the overall feature of electrostatic interactions is that they dominate repulsion only at distances so great as to render them energetically much less important than the hydration forces, which are our principal focus here.

We have said little so far about van der Waals forces. They have been detected, and they are of the magnitude expected theoretically (LeNeveu et a1 1976, 1977, Parsegian et a1 1979, Rand 1981, Lis et a1 1982). They are strong enough to overcome thermal motion and to create multilayer assem- blies of bilayers. They can be modified predictably by changing the polariza- bility of the aqueous medium through the addition of solutes (LeNeveu et a1 1977). But, except for situations involving specific attraction at molecular contact, van der Waals forces are probably too weak to play any important role in membrane fusion. At distances greater than 2nm, where long-range van der Waals attraction balances hydration or electrostatic repulsion, the net energy per unit area for parallel interaction in the energy minimum position is 0.1 erg/cm2 or less (Fig. 2).

Contact, a prerequisite for fusion, must at some stage displace water from at least some part of the membrane surface. By integrating the exponential hydration from effectively infinite separation to the few-Angstrom separation of approaching contact, one estimates the work of overcoming hydration repulsion to be extremely high, as much as 100erg/cm2. What, then, do we know of cases where hydration repulsion is overcome or in some way circumvented?

Among phospholipid systems, the most dramatic example of the disappear- ance of hydration repulsion occurs when divalent calcium ions are added to multilayers of phosphatidylserine (Portis et a1 1979, Loosley-Millman et a1 1982). Before the addition of Ca2+, phosphatidylserine bilayers repel with the electrostatic and hydration regimes characteristic of charged bilayers; after addition, the bilayers collapse, leaving virtually no water between them. A similar collapse occurs with magnesium, but at higher concentrations and with some water left between the bilayers. It appears that the binding of Ca2+ to the facing phospholipid bilayers exceeds the initial energy of hydration of the phosphatidylserine polar groups. Using measured energies of Ca2+ ion binding to single and to collapsed phosphatidylserine bilayers, one can estimate a stabilization energy of 10 to 100erg/cm2 for creating the contact between these bilayers (Parsegian & Rand 1983).

There is a useful distinction between the relatively weak, 0.1 erg/cm2 or less, minima that characterize long-range interaction and the strong

14 PARSEGIAN ET AL

n hl

E

x CT, L - a) t w

0

.01

Separation (A> 20 40 60 80

I I' I I \I - I es

I I __-- - c

c - I I c c

I I I t I

02"

I I I I I I I I

1 1 I ' I I

- I I

/ /

/ /

/ I

/ I

\ I \ / no es

-.03 FIG. 2. Linear plot of the total interaction energy between neutral (dashed line) or charged (solid line) bilayers.

interactions, 10 to 100 erg/cm2, that characterize close approach, whether energetically unfavourable (hydration, Fig. 1) or favourable (Ca2+ collapse). Conditions of strong interactions correspond with those that create strong membrane deformation, a correlation of value in the study of model fusion systems (Parsegian & Rand 1983).

Before turning to the important matters relating strong interaction and lipid deformation, we mention one other structural example of the disappear- ance of hydration repulsion, this time in natural systems. There are several instances of dimeric or tetrameric proteins that have been crystallized and whose structure has been revealed by wide-angle X-ray diffraction. Tinker & Parsegian (1978) examined the contact faces between the component mono- mers in these assemblies and found a distinction in the mode of contact, depending whether the component monomers were themselves significantly water-soluble. The contact faces of non-soluble monomers were predomi- nantly non-polar. Soluble monomers were seen to match up with an intricate arrangement of matching dipolar or monopolar charges embedded on non- polar molecular surfaces. In contact, virtually no water remained between

MEMBRANE INTERACTIONS IN PHOSPHOLIPID SYSTEMS 15

pieces. The inference is that when the pieces are separated, the surface dipoles or monopoles become hydrated and allow the monomers to disperse; when the match-up is right, though, the water is displaced by virtue of a stronger attraction between complementary sets of charges. We see this mechanism as relevant to contact between membranes that must be stable in water but that can, under the right conditions, achieve controlled contact prior to the rearrangement of component parts that leads to fusion.

Bilayer deformation

One can show formally and rigorously that some deformation must always accompany any interaction between two bodies. The extent and mode of the deformative response is not so easily given universal expression. One must take into account not only the mechanical properties of the deforming bodies but the constraints under which deformation takes place.

Two attracting, spherical phospholipid vesicles will flatten against each other. A vesicle attracted to a planar membrane will tend to cause the planar bilayer to bulge. Bilayers in a multilayer stack will thicken and their polar groups crowd together when water is withdrawn from the multilayer. Each of these responses is opposed and finally balanced by stresses created in the packing of the lipid molecules. Evans and coworkers have given detailed examples of these competitive events (Evans & Skalak, Evans & Parsegian 1983); Parsegian & Rand (1983) have related the regimes of strong and weak interaction, mentioned above, to the kinds of deformation to be expected between spherical bodies.

To begin, one may speak of an energy minimum of depth G (in erg/cm2, say) between parallel planar faces due to the balance of inter-membrane attractive and repulsive forces. The energy gained by succumbing to the energy G per unit area causes flattening to parallel membrane faces. This flattening will proceed until balanced by a force of membrane deformation. If no deformative factors other than stretching are involved, the contact angle, theta, between the flattened region of the surfaces of two equal vesicles, for example, and the rest of each vesicle is given by a version of Young’s equation,

G cos theta = 1 +-

2T where T is a membrane tension. This relation is deceptively simplistic as stated here, if one does not emphasize that T is highly sensitive to vesicular parameters.

As argued by Evans & Parsegian (1983), it is important to recognize the conditions under which the membrane is stretched in creating the contact

PARSEGIAN ET AL 16

angle theta. The tension T is by no means a constant as in the original Young formulation, where it was an interfacial tension. Evans' analysis takes great care, for example, to distinguish the condition where the vesicular volume is preserved and the membrane made to stretch, from a situation where the volume can decrease to accommodate flattening of the surface without paying the energetically expensive price of stretching the surface. The constant volume constraint permits less contact and less net attraction than does a condition of prior flaccidity, which allows flattening without stretching.

Larger vesicles can achieve a given area of contact with lower cost of deformation than can small; they are therefore more likely to maintain stable association and to display deformation.

A spherical vesicle embedding itself in a planar membrane is somewhat simpler to think about than two spheres, at least in those cases where the planar membrane is sufficiently large to maintain a constant tension T, the spherical bilayer is not deformed, and one considers only the work of stretching the planar membrane. Then the above equation becomes

G cos theta = 1 + -

T

and the angle of contact can vary continually from theta = 0, negligible attraction, to theta = 180", corresponding to full engulfment of the vesicle by the planar membrane when G = -2T.

Parsegian & Rand (1983) have shown how the capacity for deformation qualitatively increases the total energy of association between vesicles even in the regime of weak attraction, G = 0.1 erg/cm2 or less, noted above. For the kinds of attractive force encountered in strong attraction, 10 to 100erg/cm2, the forces are likely to stretch membranes to rupture, since the maximum tension supported by bilayers or natural cells appears to be only about 3dyne/cm at most (Evans & Kwok 1982).

We have already mentioned that bilayers will show another kind of deformation while being forced together. The act of removing water from a phospholipid multilayer not only brings bilayers closer together but deforms them to pack the polar groups closer together on the same surface. Parsegian et a1 (1979) (see also Rand 1981, Lis et a1 1982) showed that for free energy per molecule, g , and molecular cross-sectional area, A , the bilayer lateral pressure, dgldA, is related to the interlamellar force on each molecule, F = dglds, where s is bilayer separation, by

dg - s dg dA 2A ds'

This relation will, of course, change if the bilayers are pushed together

MEMBRANE INTERACTIONS IN PHOSPHOLIPID SYSTEMS 17

differently as, for example, in the case of lipid-coated mica sheets that are brought together as crossed cylinders (R. Horn, personal communication 1981). Then the deformation is likely to be to spread the molecules away from the region of contact. Similarly, membrane-membrane contact in cells can be expected to show exclusion of more repulsive components from, and gather- ing of attractive components to, regions of closest approach.

The qualitative point to take from the above is that the constraints on a system determine its response to membrane forces. The different constraints are: whether the vesicles change in volume; whether vesicles are big or small; whether there is a change in area during contact. Visual analogies between natural and model systems are likely to be unreliable until the operative constraints are known for both cases.

Models of fusion

One lesson that can be drawn from the foregoing is positive. The perturbation of water near contacting hydrophilic surfaces, and the resulting mutual repulsion, mean that the act of making very close contact involves a displacement whose attractive energy must be greater than the significant hydration repulsion that has been measured.

A more negative lesson is the recognition that the above-mentioned constraints on a system determine its response to membrane forces. Given the present vigorous inquiry into the behaviour of model systems and the relatively scanty information being acquired about natural systems, we gloomily suggest that it is the negative lesson that requires more attention, although we believe the more positive lesson should be used in general thinking.

Not only is it difficult to produce a model system whose constraints, as defined above, are both known and uniform, it is also difficult to define the thermodynamic equilibrium states of these systems. For example, recall that phospholipid monolayers made by solvent evaporation to produce monolayer pressure curves cannot be reproduced from one solvent to another. It is now emerging that many phospholipid vesicle systems, also made with solvents, result in structures that are not in equilibrium. For example, small unilamellar vesicles (SUVs) made of phosphatidylcholine spontaneously revert to larger structures. Large unilamellar vesicles are mechanically very fragile. Such non-equilibrium systems tend to relax in a way that reflects the state of tension into which they have been put by the method of preparation. It is therefore difficult to establish whether their behaviour is a reflection of the inherent property of the lipid or of the method of preparation. Further, the product of different methods can be expected to decay differently.

18 PARSEGIAN ET AL

Regardless of these difficulties, vesicles are the essential ingredient in all model systems and are widely regarded as paradigms for biological fusion.

The most thoroughly studied system involves the response of phosphatidyl- serine vesicles to divalent cations at millimolar concentrations. If truly leakless fusion occurs at all it must be in the early events, since the final product of the reaction is spiral multilayers with little or no interlamellar water (Portis et a1 1979). An ingenious fluorometric analysis (Wilschut et a1 1980, 1981) shows good evidence for the mixing of vesicular contents within seconds of adding Ca2+. Serious arguments for the leakage and general rupture of phosphatidylserine vesicles in Ca2+ (e.g. Ginsberg 1978, Kendall & MacDonald 1982) are cause for criticizing the use of the fluorometric method as evidence for fusion.

The powerful effect of Ca2+ derives from its capacity, mentioned above, to do the huge amount of work necessary to dehydrate the contacting faces of phosphatidylserine bilayers. The energies encountered are in fact so large that they should stretch and break vesicles of any size, a rather violent event compared to the controlled process demanded of biological fusion. The violence of this process is evident in electron micrographs showing vesicles flattening against each other (Miller & Dahl 1982, Rand et a1 1983) 100ms after the addition of Ca2+ but crushed to multilayer stacks, unrecognizable as vesicles, seconds later (Rand et a1 1983). These authors have also shown by computer-enhanced light microscopy that the post-calcium stage is at least as likely to result in vesicle breakage as in vesicle fusion.

Several other model systems are proving instructive but not yet conclusive with respect to clean fusion. Resonance energy transfer (Struck et a1 1981, Hoekstra 1982) elegantly demonstrates molecular rearrangement and redis- tribution but is unable to show conservation of vesicle contents. Multilayer liposomes of phosphatidylserine/phosphatidylcholine (PS/PC), loaded with fluorescent dye (Zimmerberg et a1 1980), effect transfer of fluorescent quanta across a planar bilayer when exposed to Ca2+ with an osmotic gradient. Alternatively, radioactive sulphate can be seen to move from small unilamel- lar vesicles across planar bilayers (Razin & Ginsburg 1980). Capacitance changes show that phosphatidylserine from liposomes incorporates into both leaflets of planar PS/PC bilayers only when there is Ca2+ in the liposome suspension (Chizmadzhev et a1 1982). Similarly, in the absence of Ca2+, a membrane charge asymmetry denoting phosphatidylserine vesicle-to- planar bilayer interaction without fusion, is seen (Cohen & Moronne 1976).

In sum: (a) Ca2+ acting on phosphatidylserine vesicles aggregates and ruptures them on a millisecond to second timescale; and (b) controlled leakless fusion in model systems still requires careful demonstration.