• Aucun résultat trouvé

Curvature instability in membranes

N/A
N/A
Protected

Academic year: 2021

Partager "Curvature instability in membranes"

Copied!
11
0
0

Texte intégral

(1)

HAL Id: jpa-00210231

https://hal.archives-ouvertes.fr/jpa-00210231

Submitted on 1 Jan 1986

HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.

Curvature instability in membranes

S. Leibler

To cite this version:

S. Leibler. Curvature instability in membranes. Journal de Physique, 1986, 47 (3), pp.507-516.

�10.1051/jphys:01986004703050700�. �jpa-00210231�

(2)

507

Curvature instability in membranes S. Leibler (*)

Groupe de Physique des Solides (+) ; Ecole Normale Supérieure, 24, rue Lhomond, 75231 Paris Cedex 05, France (Reçu le 21 aout 1983, accepté le 31 octobre 1985 )

Résumé.

2014

Un modèle thermodynamique simple a été proposé il y a quelques temps [1, 2] pour décrire les pro-

priétés physiques de divers systèmes dans lesquels l’énergie de courbure élastique joue un rôle important. Ce modèle repose sur les notions de rigidité effective et de courbure spontanée. Nous considérons ici le cas de mem-

branes bi-couches et généralisons le modèle pour des situations où de petites molécules adsorbées

2014

une source possible de courbure spontanée non nulle

2014

peuvent diffuser dans la membrane. Nous montrons que dans cer- taines conditions celles-ci peuvent entièrement destabiliser la membrane. Cette « instabilité de courbure » peut aider à comprendre certains changements de forme observés dans les membranes réelles, tels que l’echinocytose

de globules rouges.

Abstract.

2014

A simple, thermodynamical model was proposed some time ago [1, 2] to describe the physical pro- perties of various systems for which the curvature elastic energy plays an important role. The basis of this model is provided by the notions of effective rigidity and spontaneous curvature. Here we consider the case of bilayer

membranes and generalize the model for situations where small adsorbed molecules

2014

a possible source of non-

zero spontaneous curvature

2014

can diffuse within the membrane. We show that under certain conditions they

destabilize the membrane completely. This « curvature instability » can help to explain certain observed shape

transformations of real membranes, such as the echinocytosis of red blood cells.

J. Physique 47 (1986) 507-516 MARS 1986,

Classification Physics Abstracts 87.20201368.15201305.90

1. Introduction

One can imagine an idealized membrane as consisting

of two homogeneous layers of phospholipids oriented

with their polar heads towards the exterior of the membrane. Such a lipid bilayer can be in fluid state [3],

which means that the molecules diffuse freely within

the layers.

This representation is of course a crude approxi-

mation of real biological membranes, which in fact possess a quite complicated structure, including hetero- geneities in their composition, small intercalated molecules ’such as cholesterol, and many intramem- brane proteins. To describe them, one should rather

use fluid mosaic models [4], much richer and also

more complicated

Yet, the idealized lipidic bilayers, which can be produced and studied in the form of artificial vesicles, play an important role as a starting point for model- ling real biological membranes, especially if one is

interested in their simple physical properties (such as shape transformations, elasticity, mutual interactions, transport, etc.). They should be thought of as some

kind of homogenization of the plasma membranes.

Several years ago a simple model was proposed [1, 2, 5] to explain the mechanical and thermodyna-

mical properties of such closed, bilayer membranes.

On the basis of this model the notion of the elastic curvature energy of the membrane is proposed. The system is characterized by two phenomenological

parameters : the effective rigidity K and the sponta-

neous curvature Ho. These two parameters are easily

connected with the microscopic description of the

membrane [1] : the rigidity x is connected with the

binding of the membrane, and thus with changes in the

orientation of phospholipids, whereas the spontaneous

curvature Ho takes into account the eventual asym- metry in the distribution of the constituents in two

layers.

These important notions, introduced here by W.

Helfrich, F. Brochard and others [1, 2, 5, 6], have also

been useful for the description of other physical

systems such as microemulsions [7] or micelles [8].

Generally speaking the elastic curvature energy plays

a crucial role in all systems with small surface tension.

Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphys:01986004703050700

(3)

The success of the simple phenomenological model

in the study of artificial bilayer vesicles or real biologi-

cal membranes has consisted mainly in :

(i) explaining a large class of possible shapes of

these systems [6] ;

(ii) quantitatively describing some types of unusual

dynamical behaviour of the membranes, such as the

so-called flicker phenomenon observed in red blood cells [5].

In the next section we shall give a more detailed

account of this model, its achievements and also some of its limitations. In particular one of the simplifications

of the model is the assumption of the homogeneity

of the membrane. It for instance, a pure membrane includes some kind of impurities (e.g. small interca- lated particles, a different sort of lipid molecules, etc.),

which can eventually be a source for the nonzero

value of the spontaneous curvature, then one sup- poses that their distribution is homogeneous.

This assumption is not necessarily correct for the

membranes which constituents are free to diffuse

laterally. The concentration c of the « impurities »

for instance, can differ from one point to another, as

well as vary in time. The aim of this paper is to consider such situations, and to try therefore to extend the domain of application of the simple phenomenological

model.

In section 3 we consider a membrane which contains

a certain amount of diffusing, intercalated particles.

We suppose that their concentration c is coupled to

the local (mean) curvature H of the membrane. As a

consequence the membrane acquires a nonzero spon- taneous curvature. This is not however the only effect

of the presence of the impurities. In some cases the

membrane can become unstable : the density fluc-

tuations can couple to the curvature modes and

destabilize an initially flat membrane. We consider both static and dynamical aspects of this phenomenon,

which we call the curvature instability. In both approaches we derive a simple instability criterion

which gives us an insight into the physical basis of the predicted phenomenon.

It is not impossible that the curvature instability

constitutes one of the mechanisms of the shape trans-

formations in real membranes. Shape changes occur

in connexion with many important phenomena such

as cell locomotion, fusion, secretion, endocytosis, phagocytosis etc. [9]. Thanks to its relatively simple morphology and its availability, the red blood cell constitutes a model system to study such shape changes. As we have already mentioned, even the

simple phenomenological model can explain a large

family of possible shapes of erythrocytes [6]. In se ion 4

we suggest that the well known phenomenon of echinocytosis or crenation of red blood cells can be connected with the curvature instability. In particular

we recall that echinocytosis can be produced in vitro by adding some small particles which intercalate themselves within the membrane [10].

The limitations of our approach are given in

section 5. In fact, our calculation is some kind of linear

stability analysis. To describe the real systems cor-

rectly one must include the nonlinear energy terms,

as well as more details about the microscopic structure

of the membrane. We therefore sum up in conclusion

some open questions which are beyond the scope of this paper and suggest some directions for further

experimental and theoretical studies.

2. Idealized membranes and the simple thermodyna-

mical model.

In the simple phenomenological model a membrane

is treated as a continuous, fluid, bidimensional system, characterized by the following elastic energy per unit area [1] :

where H is the mean curvature at a given point of the

membrane (H = 1/Rl + I/R2, Rl,2 being the two principal curvature radii of the surface); whereas K

and Ho are two phenomenological parameters called

rigidity and spontaneous curvature respectively. As

we have already mentioned in the introduction the

rigidity K is connected to the resistance to bending of

the membrane, while Ho takes its possible asymmetry into account.

We shall not discuss here the nature of the approxi-

mation on which this model is based. All relevant

details can be found in several papers [1, 2, 6, 11].

Let us only recall that when writing the elastic energy in the form (1), one is actually neglecting :

(i) any changes in membrane topology;

(ii) any area dilatation and is supposing the effective surface tension to be zero ;

(iii) any shear resistance of the membrane - an

assumption reasonable as concerns our problem,

but not valid in general [11, 12] ;

(iv) other possible terms such as tilting energy or any kind of nonlinear curvature-elastic stresses, etc.

[1.13];

(v) any dissipation effects within the membrane [2].

Having established the form of the effective surface energy of the membrane one can apply a standard thermodynamic treatment of the model. What is surprising about this simple model is that it seems to

describe adequately many physical properties of real /4hembranes such as erythrocytes for example :

(i) in fact Helfrich and Deuling [6] obtained a large variety of the red blood shapes by simply minimizing

the elastic free energy for a fixed surface and volume of the cell. In particular they were able to obtain the typical discoidal shape of erythrocytes (for certain negative values of the spontaneous curvature Ho)

as well as other shapes observed experimentally.

(ii) Brochard and Lennon [2] quantitatively studied

the so called flicker phenomenon in erythrocytes,

(4)

509

which consists in large fluctuations (flickering) of

active red blood cells. They measured the space and time correlations of these fluctuations by means of phase contrast microscopy. The correlation functions and the scaling laws which can be deduced from them

are in agreement with the predictions of the simple

model ! We shall return to the dynamical analysis of

the thermal fluctuations of the membrane in the next

section; let us only recall here the principal theoretical results of Brochard and Lennon concerning the

fluctuation modes of a single membrane (i.e. the limit qd >> 1, where q is the wavevector of the considered

dynamical mode and d is the thickness of the disco-

cyte).

If one supposes that the elastic energy of the mem-

brane is given by equation (1), and that the liquid

beneath the membrane (inside the red blood cell)

is a simple Newtonian fluid with viscosity and density p, then there are two fluctuation modes present in the system :

(i) a slow mode for which

(ii) a fast mode for which

Remark : the factor i present in these formulae

means that the modes are not propagating ones.

They decay exponentially with time as exp(- Wl,2 t).

The power spectrum of fluctuations is entirely

dominated by the slow mode :

One of the important results of this calculation is the fact that the membrane is stable against perturbations :

the frequencies w1 and OJ2 are both positive for all

wavevectors q, therefore all the perturbations decrease exponentially with time.

In the next section we shall see that this is no longer

true if the membrane includes some free moving molecules, which, preferring curved to flat locations,

can under certain conditions destabilize the mem-

brane. In the static approach (as that of Helfrich and

Deuling) this will correspond to the vanishing of the

effective rigidity Keff.

3. Intercalated particles and curvature instability.

The success of the simple thermodynamic model

described in the previous section would be viewed

as the demonstration that the molecules intercalated in the lipidic membrane do not necessarily influence

its physical properties in an important way. They

would at most change the phenomenological para- meters such as the rigidity x or the spontaneous

curvature Ho, but not alter the behaviour of the membrane qualitatively. Yet, this is not always the

case. To see this we shall now take into account the

possible presence of diffusing molecules, different

from the phospholipidic constituents of the membrane,

in the simple thermodynamic model (1).

Let us suppose that the interaction of the intercalated molecules with their neighbouring phospholipids is

« multiform », e.g. they interact much more strongly

with polar heads than with the lipidic chains and/or they are adsorbed in the outer layer (Fig. 1). This

means that the adsorbed molecules prefer tilted configurations of their neighbouring phospholipids, configurations which are typical for locally curved

membranes. The intercalated molecules thus couple

to the local curvature of the membrane surface.

We can therefore take this into account by including

the following interaction term in the surface energy (1)

of the system :

where 0 is the local density of intercalated molecules and A is a coupling constant.

The intercalated particles, which diffuse within the

membrane, interact with the phospholipidic consti-

tuents, and also among themselves. One must thus include the third term into the free energy functional :

where f[ Ø] can be written as :

We have supposed that there is no long range interaction between the intercalated particles, such

as dipolar forces for example. The total surface energy of the membrane now is :

since the membrane is supposed symmetric in the

absence of the intercalated particles (Ho = 0).

The analysis of the general case given by (8) and (7)

is beyond the scope of this paper. We shall rather

Fig. 1.

-

Schematic representation of how intercalated molecules (absorbed drugs, intramembrane proteins, etc.)

can be coupled to the local curvature of the bilayer mem-

brane.

(5)

consider some special situations, easier to analyse,

and then discuss possible consequences (and short- comings) of our approximation.

3.1 STATIC APPROACH.

-

Let us suppose that the functional f [0 J is symmetric in tP, that is p = a3 =

...

= 0. This is equivalent to considering only a sub-

space of a general phase diagram (K, Ho, Jl, a, a3, ...).

If we describe the position of the membrane by the

function ’(x, y), then

where A is the Laplace operator. In this formula we

have neglected the anharmonic terms [14] and have supposed that the membrane is not too crumpled (i.e. VC is small, there are neither overlaps nor over- hangings of the membrane, etc.).

Taking the Fourier transform we can write (9a) as :

We can now easily diagonalize this quadratic form

and then integrate over eigenvectors with an energy gap E(q) - a + bq2 + ... (« hard modes »). In this

way we obtain the effective energy functional :

where

and fl is the curvature corresponding to new effective

variables.

Thus, in the presence of diffusing particles, coupled

to the local curvature of the membrane, the effective

rigidity decreases. This seems indeed to agree with well known experimental observations in different

physical systems. For example in microemulsions, the

addition of cosurfactants, which intercalate into the surfactant-built interface, diminish its effective rigidity,

and thus the persistence length of the system [15].

Analogous phenomena also occur in other systems, for instance in ferroelectric liquid crystals, where polarity replaces the « impurities » concentration [16].

We also find that for A’ -+ aK some instability phenomenon takes place. To understand its nature

better, we shall now perform a dynamical analysis analogous to that of Brochard and Lennon [2] for a

« pure » membrane.

3.1 DYNAMICAL APPROACH.

-

We shall write the local density of the intercalated molecules in the

following form :

where Po is the mean value of the density of the par- ticles.

The free energy functional now takes the following

form :

where Ho is the spontaneous curvature which takes into account the presence of the intercalated molecules.

The inside of the vesicle will be supposed to be a simple Newtonian fluid We can write down the

hydrodynamical equations for the liquid under the

membrane [ 17] :

where v is the velocity of the fluid andp is local pressure.

At the surface of the membrane the velocity of the

fluid and the displacement C are connected by the equation :

The system is subject to the force P., per unit area

equal to :

Thus the boundary conditions on the surface for

hydrodynamical equations (13), (14) take the fol-

lowing form [ 15J :

Let us now consider the intercalated molecules and suppose that they can diffuse in the lipid membrane.

The equation for the time evolution of their local concentration can be written in this simple form :

where j is the « current » for intercalated particles.

(6)

511

It is directly connected with free energy Fmal + F;nt :

Again we suppose the simple form of the free energy

Fmol

The constants a and b are supposed to be positive

to exclude the situations of spontaneous demixing

or aggregation of the intercalated particles.

From equation (20) we get the simple equation

for the time evolution of c(x, y) :

The first term is just a simple diffusion term;

the diffusion constant D has been measured for diffe- rent molecules inside phospholipid bilayers, for pro- teins it is of the order of 10-9 cm’ s-1. The second term, of entropic origin, expresses the fact that, in

the absence of all coupling with curvature H, the

intercalated molecules would prefer homogeneous

distribution over the membrane. Finally the last

term in equation (18) originates from the interaction

Fint with the mean curvature H(x, y) present in equation (17) : the intercalated molecules induce the curvature, but are also « attracted » by curved regions.

We can now study the stability of the membrane

against fluctuations. Let us consider a simple sinu-

soidal perturbation :

Equation (21) then takes on a simple form :

Which by taking the Laplace transform

The boundary conditions (17) and the intramem- brane continuity equation (21 a) applied to the hydro- dynamical equations (13), (14) induce the dispersion

relation for the dynamical modes (22).

If we introduce the reduced variables [17] :

then this equation can be expressed as :

which is a rather complicated relation between the wavevector q of the considered mode and the

frequency m (remark that S = S(q, w), y = y(q),

p = lt(q)).

-

From equation (24) we can naturally -obtain the results of Brochard and Lennon (2), (3) of slow and fast (stable) modes OJ! 2’ if we suppose that the inter- calated molecules are weakly coupled to the cur-

vature (A = 0) or prevented from diffusing in the

membrane (D = 0).

Let us consider the slow mode (1 » S solution

of equation (24) which dominates the spectrum.

We have :

that is :

We must realize that the constant 1/F is very small. In fact the diffusion constant of the liquid

inside the vesicle D is of order of 0.1 cm2 s-1, thus 1/F = D/D ~ 10-8 1. We shall also suppose that :

which is a very realistic assumption for real systems such as erythrocytes. We can therefore rewrite equa- tion (25) as :

from which we get :

Let us first observe that S is always real which means that the mode remains non-propagating : e - St, S real.

There are two branches for the mode (01 corresponding to two signs in (26) :

(7)

The w+ branch is positive for all values of q. On the contrary the OJ - branch can become negative and thus

the slow mode becomes unstable.

In fact from equation (27) we get the following assymptotic behaviour of OJ- :

We therefore conclude that for :

the dynamical mode becomes unstable for wavevectors q :

Equation (29) is the simple condition of the insta-

bility of the membrane in the presence of intercalated, diffusing molecules. It connects the static quantities

for the membrane A, a, and K. We call this instability

the curvature instability. Figure 2 schematically

summarizes the dispersion relation (27) for the slow mode cvi. Let us recall that in the absence of the intercalated particles w1 is of the order 10 s-1 [2]

for q - Tc/1 gm.

We have thus generalized the simple thermo- dynamical model of bilayer membranes for the case

when the concentration of the intercalated particles

can change locally. We have seen that the presence within the membrane of adsorbed molecules, coupled

to the local curvature and diffusing inside the layers,

have two main consequences on the behaviour of the system :

(i) it changes the effective spontaneous curvature of the membrane, since, as we can see from equa- tion (9b) : Ho -+ Ho + AipolK and it can therefore modify the shape of the cell continuously, as predicted by the calculations of Helfrich and Deuling [6]

(e.g. provoke the stomatocyte into discocyte trans- formation) ;

A similar conclusion was also drawn by Evans

within the framework of a different model [18]; in

that model, one assumes that the two membrane

layers are unconnected and then by alterning their

relative surface chemical equilibrium one induces

spontaneous curvature into the system;

(ii) it can trigger off the thermodynamical insta- bility of the membrane : the concentration of inter- calated molecules increases in the regions of higher

curvature and can provoke the changes in the shape

of the membrane. In particular the regions with higher

concentration of « impurities » will curve strongly.

The final shape of the membrane will of course depend

on many different factors, such as nonlinear inter- actions within the membrane, the possible presence of long range forces, the morphological details of the membrane (e.g. the presence of intramembrane

proteins) etc. We shall return to this point in the

next section.

Fig. 2.

-

Instability of the slow dynamic mode mi in the presence of intercalated particles, coupled to the local

curvature (for A2> aK).

The idea that curvature and molecular segregation

can be coupled seems to be of more general interest,

and has already appeared in other studies [19]. For

instance 2 D mixtures of smectic phases of different symmetry can undergo lateral phase separation,

which leads to the appearance of domains. The domain structure is accompanied by a variation

of local curvature. A simple analysis of this pheno-

menon done by Gebhardt, Gruler and Sackmann

was in fact based on the notion of curvature energy [20].

The condition for the instability to take place

has a very simple form in our model : A2 > xa. In fact all this is quite intuitive : the instability is more likely to occur if the particles are strongly coupled to

the membrane curvature (A big) and for membranes

with a small rigidity constant K (which get curved

more easily). The constant a depends on several physical parameters : the temperature T of the system, the mean concentration OPO of the intercalated par-

ticles, etc. Its presence in the instability condition suggests the importance of curvature instability near

the consolute point for the mixture of molecules

in the membrane.

(8)

513

4. Possible connection with the observed shape trans-

formations in real membranes.

The ideal physical systems to experimentally verify

the existence of the curvature instability in mem-

branes would be giant artificial vesicles. One could adsorb in such pure membranes some small particles interacting strongly with phospholipidic constituents and inducing the curvature locally. In these systems

one could in principle control the rigidity K, the diffu-

sion constant D and the thermodynamical coeffi-

cients a, b, a2,... (through the variations of the tempera- ture, the concentration 410, the chemical potential,

the nature of intravesicle fluid etc.). In a certain

range of control parameters one would eventually

observe then the changes in the shape of the vesicles characteristic for the curvature instability. To our knowledge such experiments have not yet been per- formed There are however some interesting observa-

tions in real biological membranes which can be connected with the curvature instability. As an example, we have chosen the so called crenation of the red blood cells, the analysis of which will naturally enlarge the modelization of Deuling and Helfrich.

However there exist other analogous phenomena,

such as some kind of endocytosis, for which we believe,

our considerations could apply.

In fact, the « model system » on which to study

membrane shape changes is the human erythrocyte.

It has been largely investigated for many years and is thus relatively well known [21]. This cell is one

of the simplest systems : the external bilayer mem-

brane is connected with a thin, filamentous, protein

structure, referred to as the membrane skeleton

(cytoskeleton), but does not interact otherwise with its interior. For various mechanical considerations it can be viewed as a membrane « balloon » filled with a simple, homogeneous liquid [22]. Despite

this simplicity, the human red blood cell has striking physical properties : while it is usually found under its familiar biconcave, discoid form, it can squeeze

through capillaries much smaller than its own dia- meter and rapidly revert back to its normal discoid

shape when in a wider blood vessel. However, under the action of different chemical, mechanical or bio-

logical factors it can also alter its shape completely,

and lose its fascinating properties : these changes

take place in numerous blood diseases [23], but can

also be used by the organism to get rid of the old,

« worn-out » erythrocytes from the blood circu- lation [24].

The simplest kind of shape changes in the ery-

throcyte occurs when we place it in a hypotonic

medium : its volume increases while its area remains constant, the cell thus becomes spherical and then lyses [25]. In the second class of shape modification,

the volume of the cell remains practically unmodified.

These kind of changes can, roughly speaking, take place in two different ways. Firstly the cell can

crenate [26], i.e. its membrane can sprout spicules,

and make the cell look like a small sea urchin; it is

then given the name of echinocyte. Or the cell can

become cup-shaped, and is referred to as a stoma- tocyte [27].

The early stages of these two processes are rever-

sible, but soon the transformation of discocytes

into echinocytes or stomatocytes becomes irrever- sible ; in fact the effective area of the membrane decreases and the cell becomes approximately spheri- cal, until eventually lyses. In the later stages of the

processes one often observes [28] the formation of

small microvesicules which separate off from the membrane’s spicules or invaginations.

Many different causes of the crenation trans- formation are actually known. For example it can

be provoked in vivo by metabolic deplation of the

red cell [29]. Thus when cells are incubated in the absence of glucose, the intracellular ATP is consumed and crenation takes place [30]. Because the abundance of calcium ions triggers off the deplation of the ATP

crenation can be induced by adding Ca++ [31].

The high pH of the extracellular solution [32] or simply the presence of the glass wall near the cell (so called « glass effect ») [33], which increases the local pH level are other echinocytogenic factors.

While such experiments are currently performed

in the laboratories, the molecular explanation of this interesting phenomenon is far from being definitely

established [21]. What is the most difficult to under- stand seems to be the metabolic shape control mecha- nism. The role of membrane skeleton, built from

proteins such as spectrins, actins etc. [34], in the sta-

bilization of the cell membrane shapes has been proven

beyond doubts [22]. Until recently it was believed

that the shape of erythrocytes was controlled by the phosphorylation of spectrins under the action of ATP and an endogenous kinase [35]. However,

recent experiments [36] have shown that things are

far less clear : echinocytosis occurs before the phospho- rylation of spectrins becomes apparent; moreover

none of the known properties of (pure) spectrins are sensibly perturbed by this process. Research has therefore turned to other possible explanations [21],

none of which is for the moment satisfactory. The rpost interesting one seems the hypothesis that

ATP is required to maintain the dynamic state of the cytoskeleton complex. The ATP depletion might

allow a cooperative clustering of spectrins, and thus, by locally uncoupling them from the membrane

(and the intramembrane proteins in particular), could

destabilize the membrane [37].

There exists however a much simpler way to cause crenation in vitro, and that is to put the cell in an

anionic solution : the adsorption of anionic or non-

charged amphiphilic agents rapidly induce the dis-

cocyte-echinocyte transformation [10]. Several years ago, Sheetz and Singer [38] suggested that anionic molecules (which, in practice, are often small anaes-

thetic drugs) intercalate mainly into the lipid in the

(9)

exterior half of the membrane bilayer, expand that layer relative to the cytoplasmic half, and thus cause

the cell to crenate. This was the famous bilayer couples hypothesis, since the authors compared the

reaction of the membrane with the response of a bime- tallic couple to changes in temperature. Of course, this is a very simplified image, but the main idea that crenation can be triggered off by the asymmetric adsorption of amphipatic molecules nevertheless remains true. Several predictions of this hypothesis

have been confirmed experimentally by its authors,

and other groups [38, 10]. From our point of view

the adsorbed drugs are in fact the «impurities coupled to the curvature and diffusing inside the cell membrane.

Another observation also supports the hypothesis

that echinocytosis could be provoked by a certain

kind of curvature instability. During crenation one

observes important changes in the lateral distribution of the intramembrane particles, in particular the strong accumulation of some constituents in the spicules

and the microvesicles produced during crenation [10].

Of course the echinocytosis which takes place in vivo

is much more complicated We believe however that the dynamical « condensing » of the intercalated, diffusing molecules can play a certain role in other mechanisms of crenation. Let us suggest here two possible mechanisms :

(i) as it was put forward in an interesting paper by

Nabarro et al. [39] the molecules of 1, 2-diacylglycerol,

which accumulate inside the cell -!during the ATP depletion [31], could substitute for phosphatidyl-

choline in the inner layer of the membrane. Being a fusogenic lipid, and having a similar fatty acid compo- sition to the phosphatidylcholine but a smaller polar head, each molecule of 1, 2-diacylglycerol can

act as an intercalated molecule in our model, (nega- tively) coupled to the local curvature. The strong increase in the concentration of these molecules has indeed been observed in the microvesicles pro- duced during crenation [28];

(ii) there now exists strong experimental evi-

dence [21, 34] that the membrane skeleton limits the translational mobility of the erythrocyte membrane particles, in particular the intramembrane proteins.

On the other hand during crenation one observes important changes in the lateral distribution of these

particles within the membrane [40]. Therefore, if the

mechanism of dynamical cooperative clustering of spectrins [21, 37] was proved true, one can imagine

that in regions which are free of cytoskeleton the analogue of the curvature instability could take

place. In fact in these regions certain intramembrane molecules could now freely diffuse, and destabilize the membrane if they are coupled to the local cur-

vature. Of course this hypothesis needs to be checked

on the molecular level;

(iii) as we have already mentioned, changes in

Fig. 3.

-

Schematic view of the echinocytosis (or crenation transformation) in human red blood cells. Only the first,

reversible state of the process is shown.

pH level of the solution can also induce echino-

cytosis. One of the possible explanations for this phenomenon could be that changes in pH (as well as

other factors such as ion binding, adsorption onto

the membrane, etc.) modify the elastic properties

of the membrane. In particular the decrease of the

rigidity could in principle cause the curvature insta- bility, as condition (29) seems to imply. Again this is only a hypothesis needing to be verified experimentally.

5. Conclusions.

In this paper we have tried to apply the simple thermo- dynamical model of bilayers to the case when a certain

number of small intercalated molecules can diffuse inside the membrane. We have shown that these molecules can not only induce the nonzero (spon- taneous) curvature but also, under some conditions,

destabilize the membrane. As a consequence, the

shape of an artificial or a biological vesicle would

change completely.

It would of course be tempting to perform a detailed

calculation of the final form of such « destabilized » vesicles. Within the framework of their model, Helfrich

and Deuling [6] obtained shapes which indeed looked like real discocytes, stomatocytes or other red blood cells. Could one also get, by generalizing their model,

the shapes of echinocytes, or even predict the observed

separation of the microvesicles ? This should be

possible in principle, however we encounter several

difficulties which make the case we have studied much more complicated than that considered by

Helfrich and Deuling.

First, allowing the density of adsorbed molecules

to vary, one enlarges the space of trial shape functions enormously. One can of course consider much sim-

plified situations, such as the one envisaged by Nabarro

et al. [39], where the vesicles are spheres with small semispherical spicules on them, whereas the density 0

takes only two discrete values. This seems rather to be a crude approximation, but it can give some insight into physical basis of the phenomenon. One

can also hope to overcome this problem posed by

(10)

515

the large variety of possible shapes and density varia-

tions by using appropriate numerical methods.

More important however are the difficulties con-

nected with the nonlinear nature of our system. In fact nonlinearities come here from different sources :

(i) when perturbations start to grow the gradients VC become appreciable and thus the anharmonic

terms must be included into the model of Helfrich

et ale [14];

(ii) when the concentration P of the intercalated molecules increases in some regions their mutual interactions become important, and thus one must modify the energy functional (9). It should be noted,

that these interactions depend strongly on the micro- scopic nature of the considered system. They can origin from the direct electrostatic or hydrophobic forces, but also be mediated by the phospholipidic layers or even the proteins of the cytoskeleton. These

interactions are interesting since they can, in some

cases, become repulsive [41] and thus stabilize the

nonhomogeneous distribution of the intercalated

particles;

(iii) for important deformations the relationship

between stresses and strains is no longer linear, as

it was pointed out and discussed by many authors [42];

(iv) for highly curved membranes, such as spicules,

it was shown that the shear cannot be neglected [43];

(v) for modal wave lengths of the order of the cell dimension viscous dissipation in the membrane may be important.

These and other nonlinearities present in the pro-

blem, as well as the fact that the real membranes

are finite, closed objects, will determined which unstable wavevectors will contribute the most to the final state of the membrane [44]. Thus, obtaining the

final shape of destabilized vesicles seems to be quite

a hard task. Especially that one should in principle

consider a rather generic situation, defined by many

potentially pertinent parameters of the problem : p, b, a, a2, a3, ... in equation (7).

Our hypothesis that the shape changes of real

membranes could in some cases, such as echino-

cytosis considered in the last section, have something

to do with the curvature instability is thus based

only on qualitative observations. Namely, echino- cytosis can be provoked by the adsorption of small

anionic molecules (e.g. anaesthetic drugs) into the membrane, and the concentration of these molecules is increased in highly-curved spicules. To our know- ledge, there are no quantitative and systematical

measurements of crenation nor of any other analo- gous phenomena. Real systems such as red blood cells are very complex and the phenomenological

models e.g. the one built by Helfrich, Brochard and

others, are not able to describe all the physical pheno-

mena which take place there. Yet, it would be interest-

ing to see where the limits of their applicability are

and whether such limits, when found, can teach us

more about the role of different constituents of biomembranes (e.g. some intramembrane proteins

the cytoskeleton etc.). We hope that further develop-

ment of simple, phenomenological models, parallel

to the experiments carried out on simple artificial

systems can bring about some progress in this direction.

Acknowledgments.

I am most gratefully indebted to Y. Bouligand,

F. Brochard, T. Dombre, P.-G. de Gennes and L. Peliti for helpful discussions and encouragements. Part of this work was performed during the Les Houches

conference on « Disordered Systems and Biological Organization » ; I would like to acknowledge the hospitality of the organizers, as well as the financial support of DRET.

References

[1] CANHAM, P. B., J. Theor. Biol. 26 (1970) 61.

HELFRICH, W., Z. Naturforsch. 28c (1973) 693.

[2] BROCHARD, F. and LENNON, J.-F., J. Physique 36 (1975) 1035.

[3] See e.g. SACKMANN, E., in Biomembranes, ed. Chap-

mann D. (Academic Press) 1985.

[4] SINGER, S. J., Ann. Rev. Biochem. 43 (1974) 805.

[5] BROCHARD, F., DE GENNES, P.-G. and PFEUTY, P.,

J. Physique 37 (1976) 1099.

[6] DEULING, H. J. and HELFRICH, W., Biophys. J. 16 (1976) 861.

[7] SAFRAN, S. A., J. Chem. Phys. 78 (1893) 2073.

SAFRAN, S. A. and TURKEVICH, L. A., Phys. Rev. Lett.

50 (1983) 1930.

SAFRAN, S. A., TURKEVICH, L. A. and PINCUS, P., J. Physique Lett. 45 (1984) L-69.

[8] See e.g. ISREALACHVILI, J. N., MITCHELL, D. J. and NINHAM, B. W., Trans. Faraday Soc. 72 (1976)

1525.

PORTE, G., J. Phys. Chem. 87 (1983) 3541.

[9] See e.g. ALBERTS, B., BRAY, D., LEWIS, J., RAFF, M.

and WATSON, J. D., Molecular Biology of the Cell (Garland Publishing Inc.) 1983, chapters 6, 7.

[10] DEUTICKE, B., Biochim. Biophys. Acta 163 (1968) 494.

CHAILLEY, B., Thesis (1976) Université Paris VI.

[11] EVANS, E., Biophys. D. 43 (1983) 27.

EVANS, E. and SKALAK, R., Mechanics and Thermo-

dynamics of Biomembranes (CRC Press Inc.,

Boca Raton, F1) 1980.

[12] FISCHER, T. M., HAEST, C. W. M., STÖHR-LIESEN, M., SCHMID-SCHÖNBEIN, H. and SKALAK, R., Biophys.

J. 34 (1981) 409.

(11)

[13] BRAILSFORD, J. D., KORPMAN, R. A. and BULL, B. S.,

J. Theor. Biol. 86 (1980) 513.

[14] HELFRICH, W., D. Physique 46 (1985) 151.

PELITI, L. and LEIBLER, S., Phys. Rev. Lett. 54 (1985)

1690.

[15] DE GENNES, P.-G. and TAUPIN, C., J. Phys. Chem. 86 (1982) 2294.

[16] See e.g. a review article : PETROV, A. G. and BIVAS, J., Progress Surf. Sci. 18 (1984) 389 and references therein.

[17] BOUCHIAT, M. A. and MEUNIER, J., J. Physique 32 (1971) 561.

LANGEVIN, D., Thesis (1974) Université Paris VI.

[18] EVANS, E. A., Biophys. J. 14 (1974) 923.

[19] We thank one of the referees for pointing out this point to us and indicating the reference 18 to us.

[20] GEBHARDT, C., GRULER, H. and SACKMANN, E., Z.

Naturforsch. 32c (1977) 581.

[21] For an excellent review see : GRATZER, W. B., Biochem.

J. 198 (1981) 1.

[22] FUNG, Y. C. and ZWEIFACH, F. W., Ann. Rev. Fluid Mech. 3 (1971) 189.

[23] GOODMAN, S. R. and SHIPPER, K., Am. J. Physiol. 244 (1983) C121 and references therein.

[24] RAHMAN, Y. E., CERNY, E. A. and PERAINO, C., Bio-

chim. Biophys. Acta 321 (1973) 526.

[25] PONDER, E., Haemolysis and Related Phenomena (Grune

and Stratton, New York) 1948.

[26] See e.g. BRECHER, G. and BESSIS, M., Blood 40 (1972)

333.

[27] CHAILLEY, B., WEED, R. I., LEBLOND, P. F. and MAIGNE, J., Nouv. Rev. Fr. Hemat. 13 (1973) 71.

[28] ALLAN, D., BILLAH, M. M., FINEAN, J. B. and MICHELL,

R. H., Nature 261 (1970) 58.

[29] NAKAO, N., NAKAO, T. and YAMAZOE, S., Nature 187 (1960) 945.

[30] See e.g. [16] and references therein.

[31] ALLAN, D. and MICHELL, R. H., Nature 258 (1975) 348.

[32] WEED, R. I. and CHAILLEY, B., Nouv. Rev. Fr. Hemat.

12 (1972) 775.

[33] BESSIS, M. and PRENANT, M., Nouv. Rev. Fr. Hemat.

12 (1972) 351.

[34] BRANTON, D., COHEN, C. M. and TYLER, J., Cell. 24 (1981) 24.

[35] BRICHMEIER, W. and SINGER, S. J., J. Cell Biol. 73

(1977) 647.

[36] ANDERSON, J. M. and TYLER, J. M., J. Biol. Chem.

265 (1980) 1259.

BENNETT, V., J. Biol. Chem. 253 (1978) 2292.

BRENNER, S. L. and KORN, E. D., J. Biol. Chem. 255

(1980) 1670.

[37] SCHNIDLER, M., KOPPEL, D. E. and SHEETZ, M. P.,

Proc. Natl. Acad. Sci. USA 77 (1980) 1457.

SHEETZ, M. P. and CASALY, J., J. Biol. Chem. 255 (1980)

1670.

[38] SHEETZ, M. P. and SINGER, S. J., Proc. Natl. Acad. Sci.

USA 71 (1974) 4457.

[39] NABARRO, F. R. N., QUINTANILHA, A. T. and HANSON, K., in Liquid Crystals in One and Two Dimensions

(Springer-Verlag) 1980.

[40] See e.g. CHEVALIER, J., J. Microscopie 20 (1974) 247.

ELGSAETER, A. and BRANTON, D., J. Cell Biol. 63

(1974) 1018.

[41] MARCEROU, J.-P., PROST, J. and ORULER, H., Il Nuovo Cim. 3D (1984) 204.

LEIBLER, S. (in preparation).

[42] MITOV, M. D., C.R. Acad. Bulgare Sci. 31 (1978) 513.

[43] EVANS, E. A. and LA CELLE, P. L., Blood 45 (1975) 29.

KWOK, R. and EVANS, E. A., Biophys. J. 35 (1981) 637.

[44] For the general problem of the selection of the wave-

number in nonlinear problems see e.g. : Wesfreid

J. E. and Zaleski S. eds. Cellular Structures and

Instabilities (Springer-Verlag) 1984.

Références

Documents relatifs

Roughly speaking, his result says that at the first singular time of the mean curvature flow, the product of the mean curvature and the norm of the second fundamental form blows

2014 A new method is proposed to measure the curvature elastic modulus kc of a membrane, by observing the thermal induced fluctuations of the shape of a spherical

Although our previous work [1] used the superficial trace of sulci, we detect the roof of the sulci, which presents less intersection between sulci than the superficial trace of

The number of first and subsequent consultations for asthma and its association with eczema and/or hay fever has been recorded since 1999 in order to discern allergic from

Anna Maria Micheletti, Angela Pistoia, and J´ erˆ ome V´ etois, Blow-up solutions for asymptotically critical elliptic equations on Riemannian manifolds, Indiana Univ. Filippo

solutions we have to calculate the deformational energy of vesicles which have the same volume and surface area as the forms in figure 3 but differ slightly. in

Using the standard notion of holomorphic extensions of real-analytic functions on totally real submani- folds, let us define what will be the kernel of the Bergman projector, up to

We think of scalar curvature as a Riemannin incarnation of mean cur- vature (1) and we search for constraints on global geometric invariants of n-spaces X with Sc(X) > 0 that