Pore formation kinetics in membranes, determined from the

release of marker molecules out of liposomes or

LIBRARY

SEP11 X990o
Gerhard Schwarz and Charles H. Robert
Department of Biophysical Chemistry, Biocenter of the University, CH-4056 Basel, Switz{rlancl4A/0 f

ABSTRACT We discuss the efflux of entrapped marker material from liposomes or cells through pores in the membrane, being
monitored by the time course of a certain signal F(e.g., fluorescence emission). This is expressed in terms of an appropriate
normalized function of time, the so-called efflux function E(t). Under conditions frequently encountered in practice the
measured E(t) can be easily related to the forward rate of pore formation if the liposomes/cells are monodisperse in size. In
the basic case of a time-independent rate law it turns out that E(t) must be single exponential. Deviations from such a simple
functional behavior might be due to a fairly broad distribution of liposome/cell sizes and/or a more complicated pore
formation mechanism. A relevant evaluation of original data is demonstrated making use of experimental results obtained
with small unilamellar lipid vesicles where pores are induced by the antibiotic peptide alamethicin. This includes the
application of a general method to eliminate the effect of a given liposome/cell size distribution.

INTRODUCTION THEORETICAL

Any solute marker material which is entrapped at higher Efflux function and rate law
concentration inside liposomes or biological cells tends to The marker substance (for example, a self-quenching
flow out into the surrounding medium once sufficiently fluorescent dye such as carboxyfluorescein or calcein) is
large pores in the membrane are opened. Accordingly, envisaged to exhibit a signal F (e.g., an optical property
measurements of this efflux process provide possibilities to such as fluorescent emission), changing between an initial
demonstrate channel formation and to determine the value F. (for intact cells) and a final value F.,, which is
associated kinetics. A special experimental procedure reached when all cells have been infected with pores and
which has become quite popular takes advantage of a the efflux is thus complete. From the observed F in the
liposome- or cell-encapsulated self-quenching fluorescent measuring period we can determine a normalized efflux
dye which fluoresces when released (1). More recently function,
this method has been employed to study the membrane
action of colicin El (2), hemolysin (3), and tetanus toxin E(t) = (F - F)/(Fo. - Fj), (1)

(4), respectively. We have begun using this technique to that describes the measured efflux in the course of time.
explore the rate of pore formation caused by the antibiotic Initially, let us consider the cells (or liposomes) to be all
peptide alamethicin (20 amino acids). The present article of the same size but not necessarily spherical. The given
is primarily concerned with a theoretical analysis of internal cell volume is to be characterized by an equiva-
pertinent data so that the underlying rate law can be lent spherical cell radius, R (corresponding to a sphere
established. Our approach is generally applicable to cases whose volume is equal to the actual volume of the cell
where the release of marker material through a suffi- interior). An additional basic parameter related to cell
ciently long-lived single pore is comparatively fast with size is the number of lipid molecules, v, in a single cell
regard to the overall efflux (see Appendix). A quantitative membrane. The total number of cells in the given system
discussion of experiments with alamethicin acting on lipid may then be expressed as
vesicles is given as an illustration of the inherent potential
for practical work. N = nLNA/v, (2)

with nL being the total amount (i.e., number of moles) of

lipid in all the membranes (NA, Avogadro's number).
Address correspondence to Dr. Gerhard Schwarz, Department of Once a pore is opened in an individual cell membrane, a
Biophysical Chemistry, Biocenter of the University, Klingelbergstrasse concentration gradient of marker molecules between the
70, CH-4056 Basel, Switzerland. inner and outer bounds of that pore will result in a

Biophys. J. e Biophysical Society 0006-3495190109/577/07

0006-3495/90/09/577/07 $2.00 577
Volume 58 September 1990 577-583
diffusion-mediated efflux of these molecules. A detailed of uniform size
analysis (see Appendix) reveals that this process will
follow a single exponential time course. The relevant ko= (dp/dt),=o = I * Vp, (5b)
relaxation time ro can often be much smaller than the measuring the rate at the very beginning of the experi-
time seen to elapse in the observed change of the signal F. ment.
Then the rate-limiting process will necessarily be the pore
formation rather than the diffusion of marker material
flowing out of the cells. These cases are of practical Basic mechanism
interest and will be investigated here with regard to pores If vp remains constant during the whole period of measur-
having sufficiently long lifetimes (>> r.) so that the very able changes of F (i.e., up top values of ~-3-4), we simply
first pore formed in a cell membrane gives rise to have
essentially complete release of the encapsulated marker
substance (to the level of its external concentration). E(t) = exp (-k.t). (6)
Under these circumstances we consider the situation In other words, the observed release of marker material is
evolving after a pore forming agent is added to a suspen- subject to a single exponential time function. This can be
sion of N individual cells. Let Ni be the number of cells expected to apply in the case of the following basic model
infected with i pores (i = 0, 1, 2, ) after a time t.
. . .
mechanism. A pore-forming substance B is hypothesized
Since the total number of pores a given cell could to associate very rapidly with the membrane and then to
experience is very large, the Ni are evidently subject to a turn slowly into pores as expressed by the reaction scheme
Poisson probability distribution. Thus, in particular, the
number of intact cells (no pores) becomes No = N - (fast) k
Bf Baa mm Ba -
(7a)
exp (-p), where p(t) is the average number of pores so
far ever formed per cell. All the other cells (i 2 1) will be It involves the free (i.e., nonassociated) and associated
clear of the marker material due to the fast efflux once a monomeric states, Bf and Ba, respectively. P is an open
pore is formed. Since they are all of the same size, each
pore composed of m monomers. The kp stands for a rate
cell contributes the same share to the change of F, and the constant according to a possible nth order rate law
efflux function is simply equal to NO/N, or (n c m),
E(t) = e-P('). (3) v = kp na (7b)
The accordingly defined pore number p(t) can be readily (ra, molar ratio of Ba per lipid). We assume the number of
related to the forward rate of pore formation, defined by open pores to be always much smaller than the number of
vp= (drp/dt)forward. (4a) Ba molecules. Then ra will reach a constant value almost
immediately by virtue of the fast partitioning equilibrium
The quantity rp = np/nL (molar ratio of pores per lipid) is Bf Ba resulting in ra = (r/a) * Cf, where Cf is the
-

chosen as the appropriate concentration variable for pores concentration of Bf, and r stands for an appropriate
in the membrane. The reverse rate (describing inactiva- partition coefficient; possible nonideal interactions of Ba
tion of pores) will naturally be irrelevant because the are described by the activity coefficient a (5). Owing to
measured signal is not affected when a pore eventually mass conservation we have Cf + racL CB (CL, lipid =

gets closed. Integration of Eq. 4a and division by N (see concentration; CB, total concentration of B) and can
Eq. 2) results in the general relation therefore readily calculate ra in terms of CB and CL. So one
arrives at a single exponential efflux function as formu-
p(t) = tfvp dt. (4b) lated in Eq. 6, where now

This pore number function can be readily determined ko = Pk cn, kk=kp * [r/(a+FcL)]".
I
(8)
from the experimental efflux function. According to Eq. 3 This will be strictly proportional to CB as long as mutual
a plot of ln E(t) vs. t directly leads to p(t). The slope of interactions of Ba can be neglected (a 1). ,

this curve
dp/dt v * vp
= (Sa) Effect of cell size distribution
reflects the actual forward rate of pore formation at the In case the efflux function is found to be not single
given instant of time. The initial slope may be used to exponential, in other words, if appreciable curvature is
define a first-order rate constant related to single vesicles seen when ln E(t) is plotted vs. t, a more complicated
-

September 1990
578 Biophysical Journal Volume 58
Volume 58 September 1990
mechanism of pore formation may be indicated. One must owing to our basic assumption of p(R, t) being propor-
take into account, however, that some curvature will be tional to R2. In the present case we thus arrive at an
caused by a distribution of cell sizes. Larger cells natu- apparent mean R = J7/, * Ro.
rally have a better a priori chance to be infected by an To analyze effects of size distribution effects on a
open pore and so will lose their marker material sooner measured efflux function, we first define an apparent
than smaller cells. average pore number
The fraction of cells having an inner radius between R
and R + dR may be given as (P(R) dR. The number of p*(t) -ln E(t).
=
(1 1)

lipid molecules per cell, and possibly also vp depend on

v,
For the circumstances of the above rectangular 'P(R), this
R. This generally implies the existence of a specific pore quantity can be calculated readily from Eq. 1Oa and
number function p(R, t). The amount of marker sub- plotted vs. the respective p(t) as shown in Fig. 1. If all the
stance contributing to F is proportional to R3. Then, the cells had actually the same radius R = R, the analogous
actually measured efflux function becomes plot would naturally be the trivial straight line p* = Tp.
For any given (P(R) and w(x), v(x) functions we may
E(t) = f R3 * e-p(R,) * (P(R) * dR/ R3 * (P(R) * dR, calculate specific p* vs. p curves (for further details see
(9a) Appendix).
We are now in a position to eliminate the effect of size
which is the proper average of the basic effilux function in distribution on the experimental data. For any of the
Eq. 3. apparent p*(t) taken from the measured E(t), a pertinent
For practical reasons, let us describe the size distribu- p value is determined by means of the specific p* vs. p
tion in terms of the dimensionless variable x = R/RO with
regard to some arbitrary reference radius Ro. Cells with
this radius have the pore number function po(t). Accord-
ing to Eq. 4b we can generally express p(R, t) as po(t)
multiplied by the scaling factors w(x) = v/lv and v(x) =
VP/(vp)0. We may then convert Eq. 9a to
E(t) = x3 exp [-w(x) * v(x)

po(t)]P(o(x)dx/f x3P0(x) dx. (9b)

The size distribution function is denoted (Pj(x). First we
consider a rate law being independent of cell size, i.e.,
v(x) = 1, and set w(x) = x2, which is expected to be
applicable to larger cells of fairly spherical shape (be-
cause v would be proportional to the membrane surface). FIGURE 1 Dependence of the apparent number of pores per cell p* on
Then for a rectangular size distribution up to R = Ro the mean p (see text). The straight line represents the trivial case of
uniform cell sizes. The other curves refer to a rectangular distribution of
(implying there are no cells with R > RO), Eq. 9b is spherical cell sizes with the inner radius R covering the range from 0
readily integrated, yielding the simple expression, through an upper bound Ro (see Appendix for details). The solid curve
has been calculated for a size-independent pore formation rate vp and a
E(t) = 2* [1 - (1 + po)e-Pi] /p2 cell surface being proportional to R2 (Eq. 10). The dashed and dotted
curve results from a more rigorous calculation which takes into account
1 - (2/3)po + (1/4) * po2 . . (lOa)
the given finite value of the bilayer thickness relative to the cell diameter
(v as in Eq. A3). The dashed curve considers a size-dependent value of vp
Let us compare this size-averaged efflux function with the which is proportional to R5. Conversion of an experimental p* value
efflux function E(t) of a hypothetical monodisperse cell (e.g., p* = 2.0 at t = 4.2 min for the 0.55MuM experiment in Fig. 3) into
suspension with a certain mean radius R. Expanding Eq. a p value applying to a certain mean cell size (of radius R) can be
3, we have E(t) = exp [-p(t)] 1-_p + (,/2)p2 * * , =
carried out as follows. First, one has to choose the appropriate curve
derived for the underlying size distribution function and radius depen-
where p(t) is the proper pore number function for that dence of the pore number p. For example, with the given solid curve,
mean radius. We now define R as the uniform cell radius p* = 2.0 is apparently equivalent to a p = 2.4 (see dashed arrow in Fig.
for which the corresponding E(t) will have the same 3). This is the average number of pores formed within 4.2 min in a cell of
initial slope as the actual size-averaged E(t). It follows radius R As noted in the text, in order to convert our experimental
from Eq. I Oa that we must set p*(t) into a p(t) with a constant slope, a particular p* vs. p curve is
required that fits the points indicated by the crosses. Evidently this is
quite well achieved by the dashed curve.
(2/3)po(t) =
(R/R)2'po(t) =
p(t), (lOb)

Schwarz and Robert Pore Formation Kinetics

Kinetics 579
curve that was calculated for the given size distribution
and appropriate w(x), v(x). The original p*(t) can so
eventually be transformed in a function p(t) which refers
to the pore opening rate of cells having the pertinent mean
radius R. The procedure will be practically demonstrated 8
below. U-
0.6
t
EXPERIMENTAL
Materials and methods 0 5 10 15 20 min

We have studied the efflux of the self-quenching dye 5(6)-carboxyfluo-

rescein (CF) from unilamellar vesicles of palmitoyloleoylphosphatidyl- FIGURE 2 Fluorescence emission F at 425 nm (390 nm excitation)
choline (POPC), induced by the pore forming antibiotic peptide relative to F_ (at completed efflux) measured as a function of time for
alamethicin. The CF was supplied by Sigma Chemical Co. (St. Louis, three alamethicin concentrations (given in micromolars).
MO), the POPC by Avanti Polar Lipids, Inc. (Birmingham, AL).
Preparation and characterization of the peptide has been described
previously (6). The vesicles were made by sonification in 50 mM dye and
separated from free dye by passing the solution over a Sephadex G50 radius according to Eq. Ob becomes R = 149 A. Assum-
column (1). Experiments were conducted at 200C and pH 7.4 (Tris ing a size-independent pore formation rate vp, we applied
buffer). In each efflux experiment a IO-Al amount of a vesicle stock the procedure described above to determine p(t) for this
solution (,5-8 mg/ml lipid) was added to a 3-ml quartz cuvette special vesicle size and eventually corrected the results for
containing a particular solution of the peptide, which is stirred continu- spontaneous leakage (determined as 5 * 10' pores per
ously. With a series 8000 fluorometer (SLM Instruments, Inc., Urbana-
Champaign, IL) an increase of fluorescence emission, F, in the course of minute). The final curves are exhibited in Fig. 3 for the
time is observed. This presumably arises from the relief of quenching of two larger peptide concentrations. The initial slope,
the CF when the dye inside the vesicles is released into the external bulk (dp/dt)0, is found to depend on the aqueous alamethicin
medium where its concentration is very much reduced. We recorded concentration with a power of -5.5, as demonstrated in
some very slow leakage of dye which occurs even with no peptide added.
It should be emphasized that titrating vesicles with peptide is not
Fig. 5.
advisable. Preliminary experiments done in this way result in very fast
initial efflux rates which were ill defined. These effects presumably are
due to transient high local concentrations of the peptide in that part of DISCUSSION
the solution where the stock peptide solution is injected. Because pore
formation is greatly dependent upon the concentration of alamethicin,
substantial errors are introduced particularly at the start of the Alamethicin has been intensively studied regarding its
experiments before stirring can establish a homogeneous concentration ability to induce voltage-gated conductance in planar
of the pore former.
The distribution of vesicle sizes was investigated with negative stain
electron microscopy on a preparation of sonicated vesicles made in
3
essentially the same way as for the efflux experiments. p __

f I"
.-
p.9. 0,,°'

RESULTS 2

Measured F for three alamethicin concentrations in the

0.25-0.55 ,uM range are given in Fig. 2. The "infinite 1
time" signal F. was ascertained after addition of a
detergent (Triton X- 100) which destroys the vesicles. The
value F,, on the other hand, could be determined by
extrapolation. Having evaluated the efflux function E(t) 2 4 6 8 10 12
as defined in Eq. 1, the apparent p* (t) according to Eq. 11
was evaluated for each concentration (see Fig. 3). These
FIGURE 3 (Solid circles) Apparent p* according to the data of Fig. 2
raw data were then processed to eliminate the effects of
(for alamethicin concentrations 0.55 and 0.37 AM). (Open circles)
vesicle size and background leakage. Leakage corrected p(t) for a mean vesicle radius (R = 149 A) taking
The histogram of vesicle sizes is shown in Fig. 4. It can into account a rectangular size distribution up to R = 182 A (see
be approximated by a rectangular distribution of inner Appendix for arguments) and a size-independent vp. For the method of
vesicle radii varying up to R. 182 A (neglecting the
= converting p* to p see Fig. 1 (dashed arrow refers to the numerical
example mentioned there).
lower bound; see Appendix). The appropriate mean

580 Biophysical Journal

Biophysical Journal Volume
Volume 58
58 September

September 1990
 efflux rates). The lifetime of an open pore is of the order of
200 10 ms or more (9, 1 1) which is much longer than the ro
value of <1 ms calculated for the relaxation time of CF
effilux through a single pore in the vesicle membrane (see
0

E
Appendix). Therefore, the condition of full depletion of
100i- the dye by the very first pore is apparently met so that the
measured efflux functions can be related to the forward
rate of pore formation as pointed out above.
The apparent pore numbers per vesicle, p* (t) =
I -ln E(t), presented in Fig. 3 are averages over the given
0 10 20 30 nm
vesicle size distribution. We can convert them to p values
-- R describing the pore formation rate in a vesicle having an
appropriate mean radius R by making use of theoretical
FIGURE 4 Histogram of vesicle size distribution in our samples. Num- curves as drawn in Fig. 1 (for more details see Appendix).
bers of vesicles counted (under the electron microscope) which fall with Let us first consider a rate vp, which is independent of
their inner radius R in the given bins. The dashed curve gives a more vesicle size. Then we obtain the corrected curves of Fig. 3
realistic approximation of the distribution above 170 A. It represents the applying to an inner radius of 149 A. Their slopes become
function 100 * exp (1 - x2), where x = R/170 A. Together with a smaller and smaller in the course of time. This indicates a
rectangular distribution up to R = 170 A this was used to calculate the
modified p* vs. p curve mentioned in the Appendix. gradual decrease of vp. Let us examine a possible mecha-
nism.
It has been shown previously ( 12) that the partitioning
lipid bilayers (7-10). This phenomenon is attributed to of alamethicin between the aqueous and bilayer media
pore formation involving aggregation of incorporated takes place in <1 s. Therefore, we have in fact a very fast
peptide monomers. The estimated pore diameters of equilibration of associated and free peptide monomers as
10-25 A (7) are large enough to permit diffusional indicated in the basic reaction scheme of Eq. 7. The
passage of carboxyfluorescein molecules, which have an rate-limiting step must occur in the aggregation process.
approximate diameter of 10 A (derived from the molecu- Most simply we may imagine that in a series of individual
lar weight of 378). We can expect that such pores will also steps a decisive intermediate aggregate An+I (of n + 1
be formed in our vesicle preparations under zero-voltage monomers) is eventually formed which is the final stable
conditions (preliminary experiments with a diffusion nucleus turned immediately into an active pore P. In
potential being positive outside have resulted in faster particular, the following reactions are to be envisaged:
kn kn+ (fast) 1

101- An- I + Aql Ang An + Al

-
An+ --- P, (lla)
k-n
(d P/dt)o
where the reverse step towards An can be neglected. Then,
0/
of course,
[m in-1]
I_ t kn+1 * r1 * rn.
vp= (11 b)
Assuming a steady-state condition for A, we obtain
0
r, =
k, ri r_,,/(k- n+ kn+l * r1). ( Ic)
0.1 Thiscan be readily expressed in terms of the associated
monomer concentration, rl, if a fast preequilibrium of
0
An-I with the A, is taken into account (Kn-l being the
appropriate equilibrium constant). Then Eq. 11 b leads to
0.01 , CA [PM]
vP =
kn Kn-I * k k * r,. (lId)
-n+ kn+1 r
0.2 0.3 0.5 1.0
This rate law in terms of r1 would reflect an order between
n (for kn+i r1 >> k-n) and n + 1 (for kn+i * r1 << k-n)
FIGURE 5 Double logarithmic plot of the mean initial rate (dp/dt)o vs. The same order is expected with respect to the total
alamethicin concentration CA, for a size-independent pore formation alamethicin concentration CA, owing to the existence of a
rate. The slope is equal to 5.5.
fast partitioning of the free and associated peptide mono-

Schwarz and Robert Schwarzand Robrt Pore Kineics

Formaton Kinetics
Por Formation 58
581
mers. Our results for (dp/dt). in Fig. 5 could accordingly
APPENDIX
be explained on the basis of a pore nucleus of six
monomers. This refers, however, to the initial phase of the Efflux through a single pore
measured efflux only. To understand a drastic slowing Diffusion-mediated flow of marker molecules through an open pore can
down of vp at later times with the simple scheme 1 la, a be described according to Fick's first law, namely,
substantial decrease of r1 (and of the free alamethicin
concentration as well) in the range of minutes would be - (dn/dt) = Ap Dm(c-c')d, (Al)
necessary. However, such a phenomenon has never been with n being the amount (mol number) of the marker material at time t
observed in appropriate titration and kinetic experiments inside a given cell, c and cd standing for the internal and external
(6, 12). Moreover, it should be noted that the total concentrations, respectively, at the pore bounds (Ap, effective cross-
amount of lipid in these experiments is insufficient to section of the pore; D, diffusion coefficient of marker molecules in the
pore; d, pore length = thickness of the membrane). It is assumed that
appreciably deplete the aqueous solution phase, given a the intracellular concentration equilibrates sufficiently fast so that n =
partition coefficient of the order of 103 M' (6). Thus r, is (4ir/3) * R3 * c; c' can be taken as a constant. Under these circumstances
actually buffered due to the fast partitioning. Eq. Al is easily solved, resulting in
An alternative possibility is that the decrease of vp may
be brought about by a possible side reaction of An c - c' = (c0 - c') exp t/ro (A2a)
resulting in irreversible nonpore aggregates. One could (c, given value of co at t = 0), reflecting the intrinsic release time,
propose a slowly evolving structural alternative (com-
posed of one or more peptide monomers) which can react to= (47r/3) R3d/ApDo. (A2b)
with An so that the rate-limiting, steady-state value of rn In the case of marker ions, the pertinent pore conductance may be
(see Eq. 1 lb) is reduced. introduced to express d/ApDo (15). For a discussion of the actual time
One may raise the question that there might be an scales we set d = 4 nm (16), Ap = 0.3 nm2, and Do = 5 10-6 cm2/s
effect of cell size (or bilayer curvature, respectively) on (equivalent to a Stokes radius of 0.5 nm in an aqueous medium). Small
unilamellar lipid vesicles with R < 20 nm then would release their
the rate of pore formation (vp) in contrast to our original marker material at a rate characterized by ro < 1 ms. A constant
assumption. Because in the course of the efflux process the concentration gradient in the pore as implied by Eq. Al is in fact
larger vesicles will be depleted most rapidly, the apparent established very much faster (d2/2D, - 10-8 s). The time scale for
p(t) of Fig. 3 will naturally reflect the vp at a gradually reequilibrating any change of internal concentration is expected to be in
the range of R2/6DO 10-7 A constant level of cd outside the cell is
- s.
decreasing vesicle size. A time-independent value of vp for maintained because the solution is continuously stirred during our
a fixed radius R can in fact be reconciled with our data if a measurements. Accordingly, the conditions for the above solution of
particular p* vs. p dependence is hypothesized as indi- Eq. Al are certainly met. It must be emphasized, however, that 'r tends
cated by the crosses in Fig. 1. It is very well fitted by a to assume fairly large values when the radius increases. According to the
R3 dependence in Eq. A2b a radius of 2 ,um for instance would imply a rO
theoretical curve assuming vp to be proportional to R5 in of some 15 min. Therefore, the release of marker material may possibly
the given range of vesicle sizes (see Appendix). The be complicated by a competition between diffusion through a single open
underlying R5 dependence of vs could possibly reflect a pore and the overall pore formation kinetics. In that case a more
partition coefficient r which increases about linearly with sophisticated theoretical analysis than the present one will be required.
R as can be inferred from Eq. 8. However, when measured
r for the highly curved bilayers of small vesicles have Aspects of vesicle size distribution
been compared with those for the relatively flat bilayer Our lipid vesicle system has been found to involve a broad range of radii
structures of coarse lipid dispersions, no significant change R. This will certainly affect p(t) because of the factor v, the number of
could be noticed (13). lipid molecules per vesicle (see Eq. 4b). The latter can readily be
Thus a gradual inactivation of the pore formation rate expressed as
may be the more likely cause of the decrease of slope in
our p(t) curves. This appears to be in line with observa-
= 4r[(R + d)2/A' + R2/A"I (A3)
tions that in planar lipid membrane experiments the (AL and A" are area per lipid molecule in outer and inner surface of the
alamethicin-induced conductance runs through a maxi- membrane). For our mean vesicle size of R = 149 A, this yields v =
mum within 3-6 min after having added the peptide, and 1.06 * 104 lipid molecules if the values A4 = 74 A2, A" = 61 A2 for
phosphatidylcholines (16) are used.
then declines to a steady-state value over the next 25-30 We have approximated our size distribution data as given in Fig. 4 by
min (14). Moreover, we have preliminary efflux data a rectangular distribution function with lower and upper bounds of R =

obtained for an approximately monodisperse suspension 22 and 182 A, respectively. It turned out that the averaging procedure
of large vesicles (,or 100 nm diameter) which still reflect a according to Eq. 9 was practically insensitive to very small values of R.
So we could simply start the integration at R = 0. Assuming a
definite decrease of vp in the course of time (Zong, R., size-independent vp implies v(x) = 1 in Eq. 9b. The other scaling
C. H. Robert, and G. Schwarz, unpublished results). function w(x) becomes a quadratic function as derived from Eq. A3.
Further work to clarify these points is in progress. Under these circumstances the dashed-dotted curve in Fig. 1 is obtained.

582 Biophysical Journal

Biophysical Volume 58
Volume 58 September
September 1990
1990
It is, however, suspected to underestimate the contribution of the larger and transmembrane potential on the kinetics of channel forma-
vesicles in the bin at radii of 170 through 240 A. In view of the fact that tion. Biophys. J. 55:393-405.
the actual distribution of vesicle sizes is more Gaussian than rectangular 5. Schwarz, G., S. Stankowski, and V. Rizzo. 1986. Thermodynamic
(17) we have modified our approximation appropriately above 170 A as analysis of incorporation and aggregation in a membrane: applica-
shown by the dashed tail in Fig. 4. The resulting p* vs. p curve is found to tion to the pore-forming peptide alamethicin. Biochim. Biophys.
be practically the same as for a strictly rectangular distribution where Acta. 861:141-151.
we set in Eq. 9b w(x) = x2 and v(x) = 1. This can be readily integrated 6. Rizzo, V., S. Stankowski, and G. Schwarz. 1987. Alamethicin
in a closed form (see Eq. 1Oa) and is presented as the solid curve in Fig. incorporation in lipid bilayers. A thermodynamic study. Biochem-
1. We actually used this to convert our p* to p (see Fig. 3). istry. 26:2751-2759.
Once we give up the presumption of a size-independent rate law, 7. Eisenberg, M., J. E. Hall, and C. A. Mead. 1973. The nature of the
substantial changes in the p* vs. p dependences may be generated. With voltage-dependent conductance induced by alamethin in black
a vp proportional to R5, i.e., v(x) = x5, the dashed curve in Fig. 1 will be lipid membranes. J. Membr. Biol. 14:143-176.
applicable. This in fact converts our measured p* curves intop(t), which
increase linearly with time (and refers to an appropriate mean radius of 8. Gordon, L. G. M., and D. A. Haydon. 1975. Potential-dependent
R= 158 A). conductances in lipid membranes containing alamethicin. Philos.
Trans. R. Soc. Lond. B. Biol. Sci. 270:433-447.
We thank Dr. S. Stankowski for stimulating discussions and useful 9. Boheim, G., and H. A. Kolb. 1978. Analysis of the multi-pore
advice as well as Edwin Kalb for performing the negative stain EM system of alamethicin in a lipid membrane. J. Membr. Biol.
procedures. 38:99-191.
10. Hall, J. E., I. Vodyanoy, T. M. Balasubramamian, and G. R.
This work was supported by grants No. 3.285.85/3.25230.88 from the Marshall. 1984. Alamethicin, a rich model for channel behavior.
Swiss National Science Foundation and by the North Atlantic Treaty Biophys. J. 45:233-247.
Organization (fellowship to C.H.R.). 11. Boheim, G. 1974. Statistical analysis of alamethicin channels in
black lipid membranes. J. Membr. Biol. 19:277-303.
Receivedfor publication 7 March 1990 and infinalform I May 12. Schwarz, G., H. Gerke, V. Rizzo, and S. Stankowski. 1987.
1990. Incorporation kinetics in a membrane, studied with the pore-
forming peptide alamethicin. Biophys. J. 52:685-692.
13. Stankowski, S., and G. Schwarz. 1989. Lipid-dependence of peptide-
membrane interactions. Bilayer affinity and aggregation of the
REFERENCES peptide alamethicin. FEBS (Fed. Eur. Biochem. Soc.) Lett.
250:556-560.
14. Vodyanoy, I., J. E. Hall, and V. Vodyanoy. 1988. Aggregation of
1. Weinstein, J. N., S. Yoshikami, P. Henkart, R. Blumenthal, and alamethicin molecules in the planar lipid bilayer. Biophys. J.
W. A. Hagins. 1977. Liposome-cell interaction. Transfer and 53:514a. (Abstr.)
intracellular release of a trapped fluorescent marker. Science
(Wash. DC). 195:489-492. 15. Miller, C. 1984. Ion channels in liposomes. Annu. Rev. Physiol.
46:549-558.
2. Kayalar, C., and N. Dilzgflnes. 1986. Membrane action of colicin 16. Huang, C., and J. T. Mason. 1978. Geometric packing constraints
El. Detection by the release of carboxyfluorescein and calcein in egg phosphatidylcholine vesicles. Proc. Natl. Acad. Sci. USA.
from liposomes. Biochim. Biophys. Acta. 860:51-56. 75:308-310.
3. Menestrina, G. 1988. Escherichia coli hemolysin permeabilizes 17. Stelzer, E., and H. Ruf. 1983. Characterisation of phospholipid
small unilamellar vesicles loaded with calcein by a single-hit vesicles by photon correlation spectroscopy enabling the detailed
mechanism. FEBS (Fed. Eur. Biochem. Soc.) Lett. 232:217-220. analysis of transport parameters. In Physical Chemistry of Trans-
4. Menestrina, G., S. Forti, and F. Gambale. 1989. Interaction of membrane Ion Motions. G. Spach, editor. Elsevier Science
tetanus toxin with lipid vesicles. Effects of pH, surface charge, Publishers B. V., Amsterdam. 37-43.

Schwarz and Robert Pore Formation Kinetics 583