HAL Id: jpa-00212399
https://hal.archives-ouvertes.fr/jpa-00212399
Submitted on 1 Jan 1990
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.
Stability of charged membranes
D. Bensimon, F. David, S. Leibler, A. Pumir
To cite this version:
689
LE
JOURNAL
DE
PHYSIQUE
Short Communication
Stability
of
charged
membranes
D.
Bensimon(1 ),F. David(2),S. Leibler(2)
and A.Pumir(1)
(1)
Laboratoire dePhysique
Statique,
ENS,
24 rueLhomond,
Paris75005,
France(2)
Service dePhysique Théorique,
CENSaclay,
91191 Gif/YvetteCedex,
France(Reçu
le18janvier
1990, accepté le9 février
1990)
Résumé. 2014 La contribution
électrostatique
au module d’élasticité est calculée pour une bicouchephospholipide chargée
en milieuionique.
Cettecontribution,
identique
pour une membrane conduc-trice comme isolante, esttoujours
stabilisante. On montre que cette stabilité des membranes libresest une
conséquence
de l’absence de tensionsuperficielles.
Abstract. 2014 The electrostatic contribution to thebending
elastic modulus ofcharged phospholipid
bilayers
in an ionic solution iscomputed.
It is found to be the same forconducting
andnon-conducting
membranes and isalways
stabilizing.
Thisstability
for free membranes is shown to be asimple
conse-quence of thevanishing
of thephysical
surface tension.1
Phys.
France 51( 1990)
689-695 15 AVRIL 1990,Classification
Physics
Abstracts 87.10-87.20E1. Introduction.
1fie
physical properties
andstability
ofphospholipid
bilayers
and theobjects they
form(cells
andvesicles)
is animportant problem
inbiology
andpharmacology.
Variousaspects
of theseproperties
have been studied over the years and some areby
now wellestablished,
inparticular
the Van der Waals and electrostatic intermembrane forces which form the basis of the DLVOtheory[l].
An artificial menbrane consists of a
bilayer
ofphospholipid
molecules with apolar
head(which
may becharged)
in contact with theaqueous
medium and twohydrophobic
tails which form the bulk of it. In absence of external stresses(e.g.
osmoticpressures,
filmtension,
etc.)
thearea per head of
phospholipid
molecules in a fluid membrane isoptimized,
i.e. the free energy is minimized[2,3].
That results from a balance betweenhydrophilic
andentropic
interactions which690
tend to increase the area
per
head and thehydrophobic
interactions between thelipidic
chains and the water which tend to reduce it.If thecompressibility
of the membrane is low the areaper
head will be fixed at this
optimal
value and does not fluctuate much. In thefollowing
wesuppose
that the
compressibility
isstrictly
zero; the total area of the membrane is thus fixed and the freeenergy X
is thenessentially
of elasticorigin
[4 - 6] :
Where
Rl,R2
are theprincipal
radii of curvature, Ro apossible spontaneous
curvature and x, /~the normal and Gaussian
bending
elastic moduli.In the
following,
we willinvestigate
thestability
of a membrane with surfacecharge
density
0-0 in an ionic solution characterizedby
aDebye screening length given by
x -1. We
shall be interestedmostly
in thelong wavelength
limit(X
» k
=27r/A),
where the electrostatic contribution to thefree
energy
amounts to a modification of somelocal parameters,
such as the elastic moduli. In that limit theonly
relevantlengthscale
isx-land
sinuez and /~ have units of energy (typically: x m10- 19J
[7 - 12],
by
straightforward
dimensionalanalysis
[13]
the electrostatic contribution tothe elastic modulus x is ôx =
CO"Õ/X3ê
with C a constant ofO(1)
and e the dielectric constant of the solution. There are two reasons onemay want
to gobeyond
thissimple
estimation.First,
thereare now some rather accurate mesurements
[7 - 12]
of thebending rigidity,
K, forphospholipid
membranes and it therefore becomes
possible
to makeprecise
measurements of the electrostatic contribution to x forcharged
phospholipid bilayers.
The surfacecharge
density
can be controledby
changing
thepH
(which
determines thedegree
of dissociation of thepolar groups)
and theDebye screening length by changing
the ionic concentration. The second reason is to determine thesign
of béc. If it ispositive
the membrane isrigidified,
but if it isnegative
the membrane may be destabilized atlong wavelengths by
the electrostatic interactions.In the
following
we shall consider two cases: membranes for which thein-plane conductivity
ofcharges
is much smaller than theconductivity
ofcharges
through
the solution(non-conducting
or
insulating membranes),
or membranes for which thein-plane
conductivity
is muchlarger
than the bulk ionicconductivity
of the solution(conducting membranes).
Forinsulating
membranes the localcharge
is constantduring
adeformation,
whereas forconducting
membranes it is the surfacepotential (and possibly
the totalcharge)
which is constantduring
the deformation. Forinsulating
membranes oneexpects
abending
deformation to increase the freeenergy
of the mem-brane and thus thesign
of 6/c isexpected
to bepositive:
the membrane isrigidified by
theelec-trostatic interactions. For a
conducting mem6rane,
however one mayexpect
the existence of an electrostaticinstability
similar to the electrostaticinstability
of acharged
surface[14].
This ishow-ever not the case, as the full calculation
(to
be describedbelow)
shows. In factcharged
insulating
andconducting
membranes haveidentical positive 6K !
t2. Electrostatic contribution to the free energy of a membrane.
Consider a
charged
infinite membrane of thickness d in contact with an ionic solutioncharac-terized
by a Debye screening length:
x-1.
Forconducting
membranes and forinsulating
ones withequal
charges
on bothsides,
wemay
without loss ofgenerality
consider thelimiting
case: d = 0.We
suppose
that the electric fields4>:f:on
both sides of the membrane(+
above and -below)
satisfy
the Poisson-Boltzmannequation,
or its linearized version: theDebye-Hückel equation
[1]
For a monovalent solution of ionic concentration no the
screening length
isgiven
by:
x2
=87rnoe2/kBT,W,
where eW is the relative dielectric constant of water(eW m 80).We
shallcon-sider two
types
ofboundary
conditions. If the membrane isinsulating
andincompressible
with aconstant surface
charge
density
0"0, the fields on the membrane have tosatisfy
Gauss’lawwhere n is the normal to the interface. If the membrane is
conducting
its surface is anequipoten-tial :
Let
((.c, y)be
a small deformation of a flat membrane. Ourpurpose
is too calculate the netchange
in the
free
energy 6:F of the membrane as a result of this deformation. Theprocedure
isstraight-forward :
1)
Solveperturbatively équations (2,
3)
for the field0±
up
to 0(c3) .
2) Compute
the electrostaticenergy[16]:
And
identify
thechange
6C in the electrostaticenergy
due to the deformation.3)
Keeping
the surfacecharge
or the membranepotential
constant(depending
on the caseconsidered) compute
the electrostatic contribution to thefree
energy Xeby integrating
thether-modynamic
relation[4,17]:
2.1 NON-CGNDUCTING MEMBRANES. -
Let((x, y)
be a small(0(e) )
déformation of a flatmem-brane
(Eq. (x,
y, z =((.c, y))
and (
(kx, ky)
its Fourier transform:We look for a
perturbative
solution ofequation(2):
Inserting equations (6, 7)
into theboundary
conditionequation (3a)
andsolving
theresulting
692
Where 1Îl
(k"" ky)
is the Fourier transform ofe(x, y)
=X2(2/2 - (V ( ) 2
/2 -
(V2(.
We may nowcompute
the electrostatic energyC, équation (4):
In the case of a surface in contact with an
electrolytic
solution(x # 0), part
of the electrostaticenergy is "used" to lower the
entropy
of the solutionby
partially segregating
the ions within theDebye
screening
layer.
The relevantthermodynamic quantity
is thefree
energyFe,
equation (5),
which is obtainedby
integrating equation (9):
Where we have used:
2.2 CONDUCTING MEMBRANES - The
case of conducting
membranes held at constantpotential
4>0
(or
which totalcharge
Q
isconstant)
is similar to the case ofinsulating
membranes treatedpreviously. Inserting equations (6, 7)
into the relevantboundary
condition,
equation (3b), yields:
Where,(fi
(kxky)
is the Fourier transform of0 (x, y)
= _X(2 /2
+( f
d2kq(k eik. p.
For a membrane held at a constantpotential4>o (by patch clamp
techniques,
forexample),
the electrostaticenergy
is: --- - ...-
-The second term
represents
the work done in order tokeep
thepotential
of the membranecon-stant
[16],
i.e. in order tobring
acharge
Q
= f ud 2p
from a "reservoir" atpotential
4>0.
Using
equation (11)
inequation (12)
one obtains:Let us note that in the
limitez
0 and 0"0 =-WXOO/2ir
=const.,
we recover the result for theIn the case of
electrolytic solutions x #
0,
the free energy0e
becomes:3.
Vanishing
surface tension andstability
of the membrane.The
k2 terms present
inequation
(15)
has anegative sign.
This maynaively
suggest
that theconducting charged
membrane is unstable withrespect
tolong-wavelength
modes(in opposite
tothe
insulating
case, in whichthe k2
term,Eq.(10),
ispositive).
On the other hand thek2
termis
usually
associated with thesurface
tensionwhich,
as we have mentioned in theintroduction,
isstrictly
zero for the free membrane. It is therefore necessary to reconsider thesignificance
of thevanishing
of the surface tension and itsçonsequences
in ourproblem.
The size of a
fluctuating
membrane can be ingeneral
describedby
two distinct variables: the total area A and the"projected"
areaAp,
i.e. the area of theprojection
of the membrane onthe
plane
(x, y).
Whereas for theincompressible
membrane,
whose number of molecules doesnot vary, the total area A is constant, this is not the case for
Ap.
Indeed,
for thefree
membrane there is no constraint on the value ofAp
and it fluctuâtes around its mean valueAp
> . As aconsequence
the intensivethermodynamical
variablecoupled
toAp -
thephysical
surface tensionr - vanishes
identically
[18]:
- 1
Here 0 is the total free energy of the
membrane,
which is the sum of the three contributions:(i)
the elastic energy(1), (ii)
the electrostatic term(10)
or(15)
and(iii)
the chemicalpotential
termfor the total area A:
ro is the chemical
potential
of theamphiphilic
molecules. We have thuswhere _
for
conducting
membranes forinsulating
membranesSince we consider in this
paper
the case offree,
unconstrained,
membranes,
the coefficient694
equation (18) identically
vanishes! Therefore there is noinstability
and the lowest term which contributes to the freeenergy
of thecharged
membrane is the last term ofequation (18).
It is the same both forconducting
andinsulating
membranes and isalways
positive.
The
k4term
gives
the effectiverigidity
of thecharged
membrane Ketf. It can thus be writtenas
where
where ew m 7 x
10-10 F/m
is the dielectric constant of water.Thus both
insulating
andconducting
membranes arerigidified by
electrostatic interactions. Our resultsequation (20)
are inagreement,
in thelong wavelength
limit,
with a similar andinde-pendent
perturbative analysis
of reference[19]
forconducting
membranes,
butthey disagree
with calculations of 6", for thenonconducting
case of reference[20].
We believe that thediscrepancy
is related to the fact that in reference[20]
the membrane has a finite thickness and a much smaller attenuationlength
inside themembrane,
so that the electricpotentials
on both sides of themem-brane are
decoupled.
It is not clear to us whether such anassumption
is valid forpurely
dielectric médium. We have checked that theprocedure
of reference[20],
based on thecomparison
of theelectrostatic
energies
per
unit area for acharged sphere,
acharged cylinder
and an infiniteplane,
leads,
both forinsulating
andconducting
membranes,
to the same results for the electrostaticcon-tribution to the
bending
modulusof elasticity K
than ourperturbative
analysis. (1 )
It allows also toassess the electrostatic contribution to the Gaussian modulus of
elasticity R
which isopposite
to6 "’.
4. discussion
Our
result,
equation (20),
isonly
valid within theDebye-Hückel
approximation,
i.e. forvalues of the membrane
potential
4>0
25mV. Since the values of theDebye screening length
x-1
vary
between10-6m
inpure
waterno
=10-7M)
and10-9m
inphysiological
conditions(no m
0.1M),
for Oo - 10 mv,bn
may vary between
0.2 x10-19
J inpure
water(where
it should bemeasurable)
and 2 x10-13
J inphysiological
conditions,
where it is thus notexpected
to berelevant. We also remark that the "bare" electrostatic contribution to the surface tension
(the
co-efficient of the first term in
equation (15):
6"(
=^X02)
isalways
very small
7
x10 -8
-by
7 x
10-SN/m
7b measure ô x one could extend the work ofSafinya
et al[11]
tocharged
phos-pholipids
(e.g.
DMPG).
In such asystem
by controlling
the concentration of surfactants andions,
one should be able to achieve a situation where 6", > K, i.e. where the electrostatic contribution to the
bending
rigidity
is dominant.(1)
Note added inproof :
This corrects a statement made in apreprint
version of this letter. Thesere-sults were
independently
foundby
several authors. See: PHiggs
and J.-EJoanny (private communication);
B.Duplantier,
R.E.Goldstein,
A.T Pesci and V Romero-Rochin(in
preparation),
B.Duplantier (Saclay
Acknowledgements.
We would like to
acknowledge
R.E. Goldstein and FLB. Williams for useful discussions. The work of D.B. wassupported
inpart
by
the DRET under contractno.89/1327.D.B.
acknowledges
B.
Duplantier
for useful discussions.References