824
Langmuir 1992,8, 824-832
Surfactant-Induced Leakage and Structural Change of Lecithin Vesicles: Effect of Surfactant Headgroup Size Katarina Edwards* and Mats Almgren Department of Physical Chemistry, Uppsala University, Box 532, S-751 21 Uppsala, Sweden Received September 16,1991. I n Final Form: December 9, 1991 Low concentrations of nonionic surfactants increase the membrane permeability of lecithin vesicles. Surfactant additions above a critical surfactant/lipid molar ratio induce changes in the size and morphology of the vesicles. In this study the effects of surfactant headgroup size on the change in structure and leakage of sonicated lecithin vesicles have been investigated and compared for a number of polyethylene glycol n-dodecyl monoethers. With increasing surfactant headgroup size the leakage rate is observed to increase, and the molar compositions needed to induce the structural transitions are at the same time shifted toward lower surfactant concentrations. A mechanism based on the concept of transient channels is suggested for the surfactant-mediated increase in membrane permeability. The vesicle growth, observed at surfactant concentrations just below those inducing bilayer solubilization, is explained by use of a model involving fusion and subsequent closure of surfactant stabilized bilayer disks.
Introduction In a recent article‘ we investigated and discussed the effects of the nonionic surfactant octaethylene glycol ndodecyl monoether, C12E8, on the size and structure of small unilamellar vesicles. It was shown that the general pattern of concentration-dependentstructural transitions, already described for Triton X-1002-sand octyl is followed also in the lecithin/Cl& system. Thus, low additions of octaethylene glycol n-dodecyl monoether cause a slight swelling of the vesicle membrane but does not give rise to any major change of the vesicle size or structure. Above a critical concentration, however, the surfactant induces a transformation from small to large unilamellar vesicles. The size of the large vesicles increases with Cl2E8 concentration until a second critical ratio has been reached and cylindrical mixed micelles begin to form. At high surfactant concentrations all lecithin is solubilized into mixed micelles, and with increasing concentration a gradual transition from cylindrical to spherical micelles takes place. Furthermore, it was shown that the surfactant concentration needed both to trigger vesicle growth and to form mixed micelles depends strongly on the temperature. Increasing the temperature shifts the two critical concentrations toward lower lipid/surfactant molar ratios. In addition, the maximum aggregate size, observed at surfactant concentrations just below those where mixed micelles begin to appear, is increased at the higher temperature. The transition from bilayers to rodlike structures, and then from rods to spherical micelles, can be understood from the packing criterion of Israelachvili et al.9 This
* To whom correspondence should be addressed.
(1) Edwards, K.; Almgren, M. J. Colloid Interface Sci. 1991, 147, 1. (2) Edwards, K.; Almgren, M.; Bellare, J.; Brown, W. Langmuir 1989, 5, 181. (3) Urbaneja, M. A.; Gofii, F. M.; Alonso, A. Eur. J. Biochem. 1988, 173, 585. (4) Paternostre, M. T.; Foux,M.; Rigaud, J. L. Biochemistry 1988,27, 2668. (5) Lash, J.; Hoffman, J.; Omelyanenko, W. G.; Klibanov, A. A.; Torchilin, V.; Binder, H.; Gawrisch, K. Biochim. Biophys. Acta 1990, 1022, 171. (6) Ollivon, M.; Eidelman, 0.;Blumenthal, R.; Walter, A. Biochemistry 1988,27, 1697.
(7) Miguel, M. G.; Eidelman, 0.;Ollivon, M.; Walter, A. Biochemistry 1989,28,8921. (8) Vinson, P. K.; Talmon, Y.; Walter, A. Biophys. J. 1989,56, 669.
criterion gives, however, no guidance as to the nature of the growth of the vesicles after the first critical surfactant concentration. Fromherz et a1.1e12 and Lasic13-15have considered the mechanism of vesicle formation, in particular by the method where surfactants are removed by dilution or dialysis of lecithin-surfactant mixed micelles, i.e. the reverse of the vesicle dissolution process. A flat bilayer, stabilized by adsorbed surfactants at the rim to decrease the edge energy, is invoked as an important intermediate. At high surfactant concentrations the edge energy is reduced to zero, and the flat bilayers are stable mixed micelles, the sizes of which are determined by the amounts of lecithin and surfactant, and the distribution of the latter between the aqueous solution, the bilayer membrane, and the edge. More surfactant allows the subdivision of the bilayers into smaller disks. Experimental evidence for mixed micelles in the form of disks has been reported for certain lecithin-bile salt systems.11J6J7 When the surfactant concentration is reduced, the edge energy becomes eventually finite, at which point a closure into vesicles will be favorable but resisted by the bending elastic energy of the bilayer. Fromherz calculated freeenergy profiles for the transition from open disks to closed vesicles for various values of the “vesiculation index” VF
where RD is the radius of the disk, YM the edge tension, and k, the elastic bending modulus. For VF between 0 and 1, disks are most stable, but there is also, for finite VF values, a local minimum for the vesicles. With VF > 1 the vesicles are most stable, and for VF > 2 the local minimum for the disks disappears. Although a few examples of disks, or open bilayer structures, were found in the previous cryo-TEM studies (9) Israelachvili, J. N.; Mitchell, D. J.; Ninham, R. W. J . Chem. SOC. Faraday Trans. 1 1976, 72, 1525. (10) Fromherz, P. Chem. Phys. Lett. 1983, 94, 259. (11) Fromherz, P.; Ruppel, D. FEBS Lett. 1985, 179, 155. (12) Fromherz, P.; Rijcker, C.; Ruppel, D. Faraday Discuss. Chem. Soc. 1986, 81, 39. (13) Lasic, D. D. Biochim. Biophys. Acta 1982, 692, 501. (14) Lasic, D. D. J. Theor. Biol. 1987, 124, 35. (15) Lasic, D. D. Biochem.J. 1988,256, 1. (16) Schurtenberger, P.; Mazer, N.; Waldvogel, S.; Kiinzig, W. Biochim. Biophys. Acta 1984, 775, 111. (17) Mazer, N. M.; Benedik, G. B.; Carey, M. C. Biochemistry 1980, 19, 601.
0743-7463f9212408-0824$03.00fO 0 1992 American Chemical Society
Surfactant-Induced Leakage of Vesicles of the lecithin/C12Es and lecithin/Triton X-100 systems,lt2 we found it difficult to reconcile the actual predominance of cylindrical structures with the models of Fromherz and Lasic. In general, cylindrical structures are much more commonly formed than disk-like on micellization or solubilization in surfactant solutions. Also in the bile-salt system evidence for the presence of rodlike structures has been found recently in small-angle neutron scattering studies.l* Under certain conditions close to the vesiculation point, a scattering curve was recorded which could result from a network of entangled, flexible rods, similar to the structures found in TEM studies for lecithin/C12Es and lecithin/octyl glucoside at compositions close to the maximum in the scattering curves.1s8 The Fromherz and Lasic models for the vesicle formation are very attractive and may still be basically correct; it just has to be assumed that the rodlike micelles become the most favorable structures at a free surfactant concentration less than that needed to give zero edge tension. Since zero edge tension represents a state, in the bilayervesicle model, where infhitly large two-dimensionalstructures are formed, the onset of formation of cylinders or other structures will break the vesicle growth at a finite size. An increase in temperature tends to decrease the effective headgroup area of the nonionic surfactant and may also decrease the effective length of the lecithin acyl chains by introduction of more gauche conformers,lgwhich have effects on both the packing criterion-more surfactant is required to allow the formation of cylindrical structures-and the distribution of the surfactant between the bilayer membrane and the edge. The surfactant can be expected to be better compatible with the bilayer structure and have a somewhat reduced preference for the edge, at the higher temperature. The observations that higher surfactant concentrations are needed for the dissolution and that the vesicles grow to larger size both seem to be explicable along these lines. In order to explore the effects of the headgroup area explicitly the transformations of small unilamellar egg lecithin vesicles on the addition of C12E6, C12E5, and C12E4 have been studied by static light scattering measurement; for the former two a similar pattern as for C&8 was observed, but with large changes in the critical concentrations, whereas C12E4 gives rise to the separation of a lamellar phase, instead of inducing vesicle growth. The results are discussed with reference to the Fomherz theory. Addition of a nonionic surfactant, however, affects not only the size and structure of the vesicles but also the permeability. In an earlier study20 we investigated the effect of the nonionic surfactants Cl2E8 and Triton X-100 on the membrane permeability and showed that even very low surfactant concentrations significantly increase the leakage of a trapped water-soluble probe, i.e. 6-carboxyfluorescein. The rate of release increases strongly with surfactant concentration for samples with surfactant/lipid molar ratios below those inducing vesicle growth. At higher surfactant concentrations the release of the dye becomes almost instantaneous on the time-scale of the experiments. The mechanism of surfactant-induced vesicle leakage is far from understood and the experimental studies carried out so far have not offered enough information to distinguish between the several possible ways by which the probe molecules may pass through the membrane. Information about the effects of temperature and sur(18)Hjelm, R. P.; Thiyagaragan, P.; Sivia, D. S.; Lindner, P.; Alkan, H.; Schwahn, D. Prog. Colloid Polym. Sci. 1990,81, 225. (19) Cevc, G.; Marsh, D. Phospholipid Bilayers: Physical Principles and Models; Wiley-Interscience: New York, 1987. (20) Edwards, K.;Almgren, M. Prog. Colloid Polym. Sci. 1990,82,190.
Langmuir, Vol. 8, No. 3, 1992 825 factant structure may add to the understanding of the complex process.
Experimental Section Materials. Egg yolk lecithin of grade 1was purchased from Lipid Products, Nutfield, UK. The C12En (polyethylene glycol n-dodecyl monoether) fractions were bought from Nikko Chemicals, Tokyo, Japan. The products were used without further purification. HEPES (2-[4-(2-hydroxyethylene)-l-piperazinyl]ethanesulfonic acid) was from Merck, Darmstadt, Germany, and 6-carboxyfluorescein from Kodak Co., Rochester, New York. Sephadex G-50was obtained from Pharmacia, Uppsala, Sweden. Preparation of Vesicles. Small unilamellar vesicles were prepared by ultrasonic irradiation of samples containing 30 mg of lecithin in 4 mL of water. After 40-60 min of sonication with a Soniprep 150 from MSE Scientific Instruments, Crawley, England, the samples were diluted with water to the desired concentration. Static Light Scattering. was added to the sonicated and diluted vesicle solution whereafter the sample was left to equilibrate for at least 2 h. The light scattered at an angle of 90° was then measured using an experimental setup comprising a Spectraphysics He-Ne laser, Model 120, with a wavelength of 633nm,aHamamatsuphotomultiplier,andaHamamatsuphoton counter, Model C 1230. An ABC 80 computer was used for data collection and analysis. Turbidity Measurements. Concentrated surfactant solution was added in a stepwise manner to a 1.2 mM sonicated lecithin solution. After each addition the sample was allowed to equilibrate and thereafter the absorbance at 350 nm was measured, using a Cary 2400 spectrophotometer from Varian Techtron Pty Limited, and plotted against the molar composition (corrected for sample dilution and expressed in mole percent surfactant). Vesicle Leakage. The membrane permeability was studied by monitoring the fluorescence increase due to the release, and subsequent dilution of the vesicle-entrapped, self-quenching probe 6-carboxyfluorescein (6-CF). Small unilamellar vesicles were prepared by sonication in 10 mM HEPES (pH 7.4) containing 100mM 6-carboxyfluorescein. To removeuntrapped solute the vesicle solution was passed through a Sephadex G-50 column, using a 10mM HEPES buffer containing 200 mM NaCl as eluent. The vesicle preparation was thereafter diluted with buffer to a lecithin concentration of 2.4 mM and transferred to one of the driving syringes in a HI-TECH Rapid Kinetics Accesory, Model SFA-11, fast mixing apparatus from HI-TECH Scientific Limited, Salisbury, England. The second driving syringe was filled with a solution of the desired surfactant in HEPES buffer. The vesicleand surfactant solutionswerequickly ( C1zE5. An incontrovertible explanation for the biphasic nature of the leakage process has not been found, although the size polydispersity of the vesicles may be a contributing factor. The rate of the process will be characterized by its half-life only, therefore. To be able to draw any further conclusionsfrom the leakage experiments, we need a model for the process by which the probe moleculesare liberated from the aqueous compartment of the vesicle interior and released into the bulk phase. Several mechanisms may be responsible for the surfactant-mediated increase in membrane permeability. A number of surfactant monomers may for instance interact directly with the probe molecules according to the equilibrium P + nS = PS, and the probe-surfactant complex may thereafter shuttle the dye across the vesicle wall. The rate would be determined by the concentration of the carrier complex in the membrane, and the leakage rate would be expected to depend on the surfactant concentration to power n.
Langmuir, Vol. 8, No. 3, 1992 831
Surfactant-Induced Leakage of Vesicles Another, and probably more realistic, alternative is that the hydrophilic probe molecules are released through a hole, or channel, created in the membrane. The surfactant monomers may directly induce the formation of hydrophilic pores or merely stabilize transient holes. Once a channel has been created, two kinetically distinct alternatives for the release may be distinguished: either the channel stays open long enough for all the probe molecules entrapped in the vesicle to be released or else the pores constitute transient structures, allowing only a few molecules to pass before they close. The kinetics of the process should in the former case be determined by the rate of formation of a channel. If we assume the process to be of nucleation character, so that a critical number, n,, of surfactant molecules will have to assemble in the membrane to trigger the formation of a channel, then the leakage rate will be proportional to the surfactant concentration to power n,. If transient channels are involved, the rate of liberation will depend on the probability of having the channel open. In the earlier paper20we derived an expression for the leakage constant which involved the probability of having a channel in the vesicle membrane, following the treatment of Fromherz.lo Assuming that the rate of decay from an open channel of radius Rch, in a vesicle of radius R,, is given by k', = 3DJIc,2/(4R,3b)
(12)
where DOis the diffusion coefficient for carboxyfluorescein and b the bilayer thickness, the leakage rate would in the case with transient channels be given by k'l multiplied by the probability, Pch, of having a channel in the membrane Pch = BRS2exp(-Ay/kBT)
(13)
where A is the water-exposed hydrophobic area and y is the surface tension and the preexponential factor denotes the ratio between the partition functions for the vesicle with and without a channel. When a small hole is created in the membrane, the major contribution to the increase in free energy comes from the exposure of the hydrophobic surface area to the surrounding aqueous solution; the change in the bending energy is negligible. The interfacial energy is given by
&, = r A
(14) Upon surfactant adsorption to the channel wall, the interfacial energy is reduced according to the Gibbs relation, eq 8, and combining that with a Langmuir isotherm for the adsorption of surfactant, the counterpart to eq 11 for surface tension results = Yo - (kBTno/A) In (K,Ct + 1)
(15) where no is the number of adsorbed surfactant monomers a t saturation. Combining eqs 12,13, and 15 results in the following expression for the rate of probe release through transient channels 'Y
In k, = constant
+ no In (K,ct + 1)
(16)
In the application of eq 16 the values estimated in Table I1 for K , will be utilized. Figure 8 shows a double logarithmic plot of the inverse half-life of the leakage process, proportional to the leakage rate constant, versus the total surfactant concentration. If the transport of the probe across the membrane is aided by the formation of carrier complexes, or the release takes place through long-lived holes, the slope of the curves should equal n or n,, respectively. For all three surfac-
0
-1
P 3
-
-2
M
-3 L
-3.8
1
-3.5
-3.3
-2.8
-3
I&! Ct
Figure 8. Leakage rate data, plotted as log (inverse halftime) versuslog (totalsurfactant concentration),for C12E8(opencircle), (diamonds)at 20 O C and C12E8 at 50 (squares), and O C (filled circles). I
D '
--
to
-3
4
"
-
-
0.2
0.35
i rk-12.7
0.5 lg (Ksct+l)
0.65
0.8
Figure 9. Leakage rate data from Figure 8 plotted according to eq 16 using values of K , as listed in Table 11.
tants the regression lines suggest an exponent between 7 and 10 for the dependence on the c ~ n c e n t r a t i o n . ~ ~ When the results are plotted according to eq 16, as presented in Figure 9, the number of surfactant molecules needed to saturate the channel calculates to between 10 and 13. For comparison the leakage induced by octaethylene glycol n-dodecyl monoether, the most potent surfactant, was also investigated at 50 "C. Some results from these measurements are plotted together with the results at the lower temperature in Figures 7, 8, and 9. Depending on the surfactant concentration, the increase in temperature may either increase or decrease the rate of probe liberation. For surfactant concentrations below 30 mol '% (0.51 mM) the release is considerably faster at the higher temperature. The leakage rate after addition of 24 mol '% C12Ee may be used to illustrate the temperature effect in the lower concentration range: at 20 "C it takes about 70 s before an intensity level corresponding to half that obtained at complete release is reached, whereas this level is obtained only 15 s after mixing when the temperature is raised to 50 "C. As the surfactant concentration exceeds 30 mol 5% , the pattern changes, however, and the rate of release becomes faster at the lower temperature. This change, which is evident in Figure 7, comes as no surprise. The very fast release of dye observed above 30 mol % ,at 20 "C, is expected in this concentration range where the (34) The results on ClzEa induced vesicle leakage of 6-CFpresented in ref 20 are not consistent with the experimental findings in this paper. The different appearance of the leakage curves is probably due to the use of a polydisperse surfactant preparation (obtained as a result of degradation of the poly(oxyethy1ene)chains%)in the previousexperimenta. (35) Donbrow, M. In Nonionic Surfactants. Physical Chemistry;Surfactant Science Series; Schick, M. J., Ed.; Marcel Decker: New York, 1987; Vol. 23, p 1011.
832 Langmuir, Vol. 8, No. 3, 1992
small vesicles begin to transform into larger structures, and in a previous article1we showed that about 10 mol % more surfactant is needed at 50 OC to initiate the vesicle growth. The plots in Figures 8 and 9 indicate, in addition to a faster leakage rate at the higher temperature, a weaker concentration dependence of the constant. This speaks against the carrier mechanism as it seems unlikely that only half the number of surfactants needed to transport the probe molecule at 20 "C are required when the temperature is raised to 50 "C. The observed decrease in the leakage rate with decreasing PEO chain length provides further evidence against the idea of the dye being shuttled across the vesicle membrane by means of a carrier complex. The less hydrophobic surfactants would, contrary to the experimental findings, be expected to transverse the membrane with more ease than those with long PEO chains. The results in Figures 8 and 9 cannot otherwise discriminate between the different mechanisms. If, however, we require the leakage mechanism to be compatible with the suggestedmechanism for the growth of the vesicles in region 11, then the formation of long-lived pores by a nucleation mechanism must be ruled out. The exposure of the hydrophobic edges could only occur if the edge tension had been reduced to almost zero, i.e. inside region 11. We are left, therefore with the concept of transient pores having comparatively low surface coverage. The number of adsorbed surfactants at saturation suggested by the results in Figure 9, i.e. about 10 surfactants correspond to very small pores. By use of the saturation numbers tabulated in Table 11,the perimeter of the channel
Edwards and Almgren
would be less than 2 nm, or the diameter as small as 0.5 nm, which seems impossible: All 10 surfactant molecules could not be accommodated within such a small volume. For a cylindrical channel with a length of 3.5 nm to contain the hydrated head oups of 10Surfactants, the radius must be from 8.5 to 6.7 for headgroups with 8 to 5 EO units, each with one hydration water (EO volume = 0.1 nm3 36). Could a carboxyfluorescein probe pass through sucha filled narrow channel? I t is probably possible; it is known from fluorescencequenching studies in reversed micelles of Triton X-100, containing less than one water per EO group, that probes and quenchers as R ~ ( b p y ) 3and ~ + methylviologen are readily dissolved and very mobile in such a milieu.37 Thus, a surfactant-filled pore might well provide a channel for the passage of entrapped substances, but the application of the present model, with a Langmuir type of isotherm for the adsorption, may appear rather odd. However, in most of region I the number of adsorbed surfactants is far less than at the saturation limit, and the Langmuir isotherm may well be a good approximation. Qualitatively, the interpretation makes sense, but for an acceptable quantitative description much remains to be done.
Acknowledgment. This work was supported by the Swedish Natural Science Research Council. Registry No. HEPES, 7365-45-9;C12E6, 3055-95-6;C12E6, 3055-96-7;C12Es,3055-98-9;6-carboxyfluorescein,3301-79-9. (36) Nagarajan, R.;Ganesh, K. J. Chem. Phys. 1989,90, 5843. (37) Almgren,M.; van Stam, J.; Swarup, S.;Lofroth, J. Langmuir 1986, 2, 432.