- Email: [email protected]
Interaction of surfactants with bilayer of negatively charged lipid: effect on gel-to-liquid crystalline phase transition of dilauroylphosphatidic acid vesicle membrane
Chemistry and Physics of Lipids, 48 (1988) 189--196 Elsevier Scientific Publishers Ireland Ltd.
189
Interaction of surfactants with bilayer of negatively charged lipid: effect on gel-to-liquid-crystalline phase transition of dilauroylphosphatidic acid vesicle membrane Tohru Inoue, Tetsurou Iwanaga, Kohsuke Fukushima, Ryosuke Shimozawa and Yukio Suezaki a Department of Chemistry, Faculty o f Science, Fukuoka University, Nanakuma, Fukuoka 814-01 and aPhysics Laboratory, Saga Medical School, Nebeshima, Saga 840-01 (Japan) (Received March 7th, 1988; revised and accepted May 2nd, 1988)
The interaction of surfactants with the vesicle membrane of the negatively charged lipid, dilauroylphosphatidic acid, was investigated through their effect on the gel-to-liquid-crystaifine phase transition of the lipid bilayer. Three types of surfactants (anionic, cationic and non-ionic) with different hydrocarbon chain length were examined. (i) Anionic sodium aikylsulfates affected the phase transition temperature, T , only weakly. (ii) Non-ionic alkanoyl-N-methylglucamides decreased T monotonously with increasing concentration. The depression of T induced by these surfactants was analyzed by applying the van't Hoff model for the freezing-point depression, and the partition coefficients of the surfactants between bulk water and lipid membrane were estimated. (iii) Cationic alkyltrimethylammonium bromides affected T in a complex manner depending on the hydrocarbon chain length of the surfactants. Octyl-/tetradecyl-trimethyiammonium bromide depressed/elevated T monotonously with increasing concentration, whereas the change in T induced by decyl- and dodecyltrimethylammonium bromides was not monotonous but biphasic. This complex behavior of the phase transition temperature was well explained, based on the statistical mechanical theory presented by Suezaki et al. (Biochim. Biophys. Acta, 818 (1985) 31--37), which takes into account the interaction between surfactant molecules incorporated in the lipid membrane.
Keywords: surfactant-lipid membrane interaction; dilauroylphosphatidic acid; vesicle; phase transition.
Introduction
The interaction of surfactants with a phospholipid vesicle membrane has been widely investigated with interest in their potency as a membrane solubilizer [1,2] or membrane fusogen [2D4], and, more currently, in their utility for the preparation of lipid vesicles (detergent removal method) [5D7]. Most of these studies deal with the mixed micelles formed by the addition of surfactants at a concentration higher than the critical micelle concentration. Recently, Gofli et al. [8] have reported the perturbing effect of surfactants on the structure and prop-
erties of the lipid membrane in rather lower concentration range. Such studies may have a significance different from the above-mentioned practical interest. Since surfactants are typical amphiphiles and the amphiphilicity can be readily controlled by combining various types of head groups and hydrocarbon tails, the surfactant-lipid membrane system may become a useful model system to elucidate the action mechanism of amphiphilic drugs; e.g. alkyltrimethylammonium salts are regarded as a model of tertiary ammine local anesthetic. From this point of view, we have been studying the effect of various surfactants
Correspondence to: T. Inoue. 0009-3084/88/$03.50 © 1988 Elsevier Scientific Publishers Ireland Ltd. Published and Printed in Ireland
190
on the membrane properties of the lipid vesicle paying special attention to the gel-to-liquid-crystalline phase transition [9-- 11 ]. Studies on the interaction of small molecular ligands with lipid bilayers have so far been restricted to the membrane systems composed of zwitterionic phosphatidylcholine, except for a few reports [12]. The probable reasons are that the phosphatidyicholines are the main component of biological membranes and are readily available, and that the structure and properties of their bilayers have been relatively well characterized. However, because cell membranes are usually negatively charged, the study of the interaction of amphiphiles with the membranes of negatively charged lipids may be important for the elucidation of the detailed features of the biologically significant interaction between amphiphilic small molecules and biological membranes. Thus, in the present work, we studied the perturbing effect of surfactants on the gel-to-liquid-crystalline phase transition of the vesicle membrane of negatively charged lipid, dilauroylphosphatidic acid (DLPA). Three types of surfactants, i.e. anionic, cationic and nonionic, were examined. As expected, the perturbing effects of these surfactants were strongly dependent on the electrical charge on the head groups of the surfactants.
Branson Sonifier Model 185 at 35°C for about 15 min. The sample suspension was prepared by mixing the DLPA suspension and the surfactant solution to give a desired surfactant concentration, and by sonicating again at 35°C for about 5 min. The size of vesicles thus prepared was estimated from quasi-elastic light-scattering measurements to be about 140 nm in diameter. The DLPA concentration was kept at about 5 x 10-4 M throughout the experiments unless otherwise noted. The pH levels of the sample suspensions were in the range 7.5 _ 0.2, where DLPA has a monovalent negative charge [13]. The gel-to-liquid-crystalline phase transition of DLPA vesicle membrane was followed by the scattered-light intensity from the vesicle suspension. Details concerning the procedure have been described previously [9]. The differential scan-
~~Tm c
4J
Materials and methods
Synthetic DLPA (monosodium salt) was obtained from Sigma. The surfactants used were: (i) anionic surfactants: sodium octyl-, decyl- and dodecylsulfate (SOS, SDeS, SDS, respectively); (ii) cationic surfactants: octyl-, decyl-, dodecyland tetradecyltrimethylammonium bromide (OTABr, DeTABr, DTABr, TTABr, respectively); (iii) non-ionic surfactants: octanoyl-, nonanoyl- and decanoyl-N-methylglucamide (MEGA-8, MEGA-9, MEGA-10, respectively). The surfactants were purified as described previously [9]. Water was deionized and doubly distilled, once from alkaline potassium permanganate solution. A stock suspension of DLPA vesicle in water was prepared by sonication in the cup-horn of a
l
i
20
T
T
=o
O. I m J l s LAJ
I 20
1
J
~ 30
t
I
t
30
I 40
B
I 40
Temperature /°C Fig. 1. The gel-to-liquid-crystalline phase transition o f D L P A m e m b r a n e observed by scattered-light intensity (A) and DSC (B). Experimental conditions: (A) Additives are (a) none, (b) 4.0 x 10-3 M SDeS, (c) 2.0 x 10-3 M MEGA-9, and (d) 4.1 x 10-t M DTABr. D L P A concentration is about 5 × 10-4 M. Heating rate is 0 . 5 ° C / m i n . (B) D L P A concentration is 4.8 x 10-z M. Heating rate is 1 . 0 ° C / m i n . vesicle
191
ning calorimetry (DSC) measurements were made for pure DLPA vesicle in order to obtain the enthalpy change associated with the phase transition by the Seiko-Denshi DSC system (model SSC). Typical traces of the scattered-light intensity with temperature and the DSC thermogram are shown in Fig. 1. A drastic change in the scattered-fight intensity was observed corresponding to the phase transition. The midpoint temperature of the transition, T e was determined by drawing straight lines as indicated by dotted lines in Fig. 1. The measurements were repeated at least three times for a given sample; a good reproducibility (typically within _+ 0.1 °C) was obtained for T m. T m of pure DLPA obtained for 18 runs was 32.0°C with standard deviation of 0.2°C, which is in agreement with the literature [14]. The transition temperatures measured by light-scattering and DSC are different by about 1 °C. In the DSC experiment, rather concentrated DLPA (about 5 × 10-2 M) was employed. The lower transition temperature obtained by DSC is probably attributed to the high ionic strength of the DLPA suspension. In fact, the transition temperature obtained by light-scattering for 5 × 10-~ M DLPA was decreased to 30.8°C by the addition of 2 × 10-2 M NaCI. The enthaipy change associated with the phase transition of DLPA vesicle membrane was estimated to be 3.4 kcal/mol from the peak area of the DSC thermogram, which agrees with the literature [14]. It should be noted here that the addition of surfactants at high concentration led to the reduced amplitude of the change in the scattered-light intensity accompanied by deformation of the transition curve. Thus, the surfactants were added within the concentration range where normal phase transition behavior was observed. Results and discussion Anionic surfactants
The effect of anionic surfactants on the phase transition temperature of DLPA vesicle mem-
32
I"~o ~D""~O-O~o
30
~J o
(a) 0--
I
32
o
[]
0
O
O --------~O
2
4
(b)
~
30 6
(c)
D ~ [] ~
D
~
D
----a__ rn ----~.
I
I
I
l
2
3
Cs
/ 10-3f4
Fig. 2. Plot of T~e the phase transition temperature of DLPA vesicle membrane, against C~ the concentration of added anionic surfactants. Surfactants are (a) SOS, (b) SDeS and (c) SDS.
brane is shown in Fig. 2, where Tm is plotted against the added surfactant concentration, C . The change in T induced by the addition of these surfactants is small in the measured concentration range, although the decreasing tendency of T is observable, especially when SDS is added. This is in contrast to the case of dipalmitoylphosphatidylcholine (DPPC) - - anionic surfactant systems [ 9 ] , where a significant decrease in Tm was observed by the addition of these surfactants at much lower concentration ranges. The weak interaction between anionic surfactants and DLPA may be understood in terms of the electrostatic repulsion between negative charge of DLPA and anionic surfactants. Non-ionic surfactants
Figure 3 illustrates the relation between T and added surfactant concentration for D L P A - MEGA series surfactant systems. T decreases almost linearly with increasing C with the deviation from the linearity at high surfactant concentration range. Furthermore, the decreasing rate
192
32
T"=T"°
•
\
2.
\o\
~
\ 0
l
I 2
I 3
I 4
I 5
t 6
Cs / 10-3M Fig. 3. Plot of T against C for DLPA--non-ionic surfactant systems. Surfactants are MEGA-8 (A), MEGA-9 (El) and MEGA-IO (V).
of T increases with the hydrocarbon chain length of the surfactants. The depression of the phase transition temperature of phospholipid vesicle membranes induced by the small molecular ligands has been successfully analyzed in terms of the van't H o f f model for the freezing-point depression [15-17]. According to this model, the depression of the phase transition temperature of the lipid bilayer is attributed to the lowering of the chemical potential of lipid molecule in a liquid-crystalline state membrane due to mixing with the additive molecules. Then, the degree of the temperature depression is proportional to the mole fraction of additives in the lipid bilayer for the case of sufficiently low additive concentration. When the additives are partitioned between bulk water and a lipid bilayer, it follows that the partition coefficient of the additives between these phases can be estimated from the relation between the phase transition temperature and the additive concentration. Based on this treatment, T is related to the additive concentration by the following equation under the assumption that the partitioning and mixing of additives is negligible for gel state membrane compared with liquid-crystalline state membrane [9]
R T2~,o K All 55.5+CLK C'
(I)
where T , 0 is the transition temperature without additives, AH the enthalpy change associated with the phase transition, R the gas constant, C L and Cs the total concentrations of lipid and additives in molarity, respectively, and K the partition coefficient defined as K = x A ' / x A where x A' and x A are the mole fractions of the additives in liquid-crystalline membrane and water phases, respectively. According to Eqn. (1) one can estimate the partition coefficient from the slope of the straight line of T vs. C plot with the knowledge of M-/. Equation (1) was applied to the DLPA - - MEGA systems, and the values of K were estimated using the slopes of the linear portions at low concentration region in Fig. 3 and the values of T~,0 = 305 K and M-/ = 3.4 kcal/ tool. The results are as follows: K = 1.12 x 103 (MEGA-8), 2.27 x 10a (MEGA-9) and 7.51 x 103 (MEGA- 10). Figure 4 shows the relation between the partition coefficient and the hydrocarbon chain length of the surfactant, where log K is plotted against the carbon number, Arc, in the hydrocarbon chain. For comparison, the results obtained
4.5
J C~ O
3.5
/ 2.5
2
'
-'
~
,0
Nc
Fig. 4. Relation between log K and N , the carbon number in the hydrocarbon chain of MEGA series surfactants. Lipids are DLPA ( I ) and DPPC (E3).
193
with DPPC [9] are also included in the figure. It may be regarded that log K increases with N in an essentially linear fashion for both cases of DLPA and DPPC. The slope provides a contribution of about R T per methylene unit to the transfer free energy from the bulk water to the lipid membrane. This suggests [18] that the hydrophobic interaction plays an important role for the incorporation of MEGA series surfactants into the membrane of DLPA vesicle as well as DPPC vesicle. The effect of a difference in a polar head group of the lipid is seen in somewhat larger K values obtained with DLPA compared with those obtained with DPPC. This may be interpreted as reflecting the difference in the interaction between the hydrophilic group of MEGA series surfactants and the head groups of the lipids.
Cationic surfactants By the addition of cationic surfactants to DLPA vesicle suspension, T showed a complex behavior depending on the hydrocarbon chain length of the surfactants, as demonstrated in Fig. 5 [19]. Short-chained OTABr decreases T and long-chained TTABr increases it monotonously, whereas DeTABr and DTABr having an
intermediate chain length exhibit a biphasic effect on T . This behavior is quite different from that observed with DPPC vesicle membrane, where T is decreased almost linearly with the concentration regardless of the hydrocarbon chain length of the surfactants [9]. Another remarkable difference between DLPA and DPPC is the required concentration range of the surfactants for the perturbation. For DPPC, the concentration range depends strongly on the hydrocarbon chain length of the surfactants; the shorter chain, the higher concentration [9]. On the other hand, the perturbing effect on DLPA vesicle membrane appears in a similar surfactant concentration range irrespective of the hydrocarbon chain length. Furthermore, the effective concentration range for the perturbation is extremely low compared with those for DPPC vesicle membrane. These facts suggest that the electrostatic interaction between the negative charge of DLPA and the positive charge of cationic surfactants plays a dominant role for the incorporation of the cationic surfactants into DLPA vesicle membrane; the hydrophobic interaction is hidden behind the strong electrostatic interaction. In addition, because of the strong interaction, most of the surfactant added is considered to be 38
36
i - - I
- - i - -
i
•
•
34
N. a4
\
E b~
E b~
32
30
3O
[3
D _ _ E ) [3
- - D
[]
28
26
I .5
I
I
1
t .5
Cs /
I 2
10-4M
Fig. 5. Plot of T against C for DLPA--cationic surfactant systems. Surfact~ts are OTABr (A), DeTABr (121), DTABr ( I ) and TTABr (V). The solid lines were drawn according to Eqn. (2) using the parameters listed in Table I.
26
15 •
I
J 1.5
t
Co~p, /
I 2
I0-3M
Fig. 6. Plot of T against the DLPA concentration for DLPA--cationic surfactant systems. Surfactants are OTABr ( n ) and TTABr ( I ) . The molar ratios of the surfactants to DLPA were fixed at 0.06.
194
incorporated into the vesicle membrane. This is supported by the experimental results presented in Fig. 6, which shows the dependence of T~ on the DLPA concentration with the fixed ratio of the surfactant to the lipid. This figure shows that Tm is almost constant, i.e. Tm is determined by the ratio of the total concentration of surfacrants to that of DLPA regardless of the lipid concentration. Thus, the cationic surfactants are, when added to the DLPA vesicle suspension, all incorporated into the vesicle membrane rather than partitioned between the bulk water and the membrane phase. The most interesting finding in the present study is the biphasic effect of DeTABr and DTABr on the phase transition temperature of DLPA vesicle membrane. The monotonous change in Tm induced by additives is explained based on the van't Hoff model for the freezingpoint depression as mentioned before; the depression/elevation of Tm is caused by the preferential partitioning and mixing of the additives in the liquid-crystalline/gel state membrane. However, the biphasic response of Tm is difficult to explain by the simple thermodynamic treatment of the freezing-point depression. The similar biphasic behavior of the phase transition temperature of lipid bilayers induced by additives has also been reported with DPPC ml-alkanol systems [20], although the direction of the T variation is the reverse of the present case; 1-alkanols with chain length of N = 9m 13 depress the transition temperature at lower concentrations and elevate it at higher concentrations. Suezaki et al. [21] have successfully analyzed this biphasic dependence of T on the 1-alkanol concentration based on the statistical mechanical treatment. The basic idea of their theory is to introduce the intermolecular interaction between additives within the lipid membrane, which depends on the membrane state, gel or liquid-crystalline. When the interaction between additive molecules is stronger in the gel state membrane than in the liquid-crystalline state membrane, it acts to elevate the phase transition temperature, and vice versa. The intermolecular interaction can be reasonably assumed to be a quadratic function of
the additive concentration, whereas the entropic effect in the treatment of freezing-point depression is linear with respect to the additive concentration. The maximum or minimum in the Tm vs. concentration profile can be naturally derived as a result of a combination of these two effects. The theory of Suezaki et al. [21] provides the following expression for the relation between T= and the additive concentration in the membrane phase
1 [AHAx
1 +
r. = T.~
AH
X2
+ (~' - ~') i---~-~x ]
(2)
1 +xASAI&S
where the symbols have the following meanings: AH, AS, the enthalpy and entropy changes of the lipid molecule, respectively, associated with the lipid phase transition from gel-to-liquid-crystalline state; M/A, ASA, the enthalpy and entropy changes of the additive molecule, respectively, when the residing domain changes from gel-toliquid-crystalline state; e~, ~1, the interaction energy parameters between additive molecules in the gel and liquid-crystalline state of the lipid membrane, respectively; x, the molar ratio of the additive to the lipid in the membrane phase. As mentioned above, the intermolecular interaction energy between additives appears in a quadratic form in Eqn. (2). The present results obtained with DLPA - cationic surfactant systems were analyzed based on Eqn. (2). The molar ratio of the surfactant to the lipid in the membrane phase, x, was replaced by the ratio of the total concentration of the surfactant to that of the lipid, because, as mentioned before, almost all the surfactants are regarded to be incorporated into the lipid membrane. M/A, ASA and Es - ~t were assumed to be linear functions of the hydrocarbon chain length of the surfactants, i.e. AH A = AS A Et -
mh
= s,
*l =
+
ms 2
m,
where m represents the carbon number in the
195 hydrocarbon chain o f the surfactant. The values o f four parameters in the above expressions, h, s~, s 2 and E, were estimated by minimizing the quantity S = Y ( T - Tm(exp))2 according to the procedure described previously [21], where T m and T ( e x p ) are the transition temperatures calculated f r o m Eqn. (2) and that observed experimentally, respectively. The best-fit values o f the parameters obtained are as follows: h = 0.377 kcal mol -~, s~ = 4.5 cal K "l, mol -~, s 2 = 0.76 cal K -1 mo1-1 and ~ = - 0 . 1 4 kcal mo1-1. The sensitivity o f these numerical values to the fitness m a y be evaluated by the standard deviation o f T . T h e standard deviation ( q ' ( S / N ) , where N is the n u m b e r o f data points) o f the estimated T was 0.17°C. The 10O7o variation o f one of the four fit parameters, the others being fixed at best-fit values, increased the standard deviation to 1.9°C (s~), 3.5°C (s2), 5.4°C (h) and 0.46°C (~). For each cationic surfactant, the obtained numerical values o f AHA, ASA, e& -- El, and the free energy changes, AGA, defined by
are listed in Table I. The curves shown in Fig. 5 were drawn according to Eqn. (2) using the values o f parameters listed in Table I. Agreement between the experimental data and the theoretical curves is satisfactory except the case of TTABr. It is interesting to compare the parameters in Table I with those obtained with D P P C - - 1-
alkanol systems [21]. The signs o f the parameters M/A, ASA and Es - El are reverse in these two cases. With respect to ~ - EI, it is positive for D P P C - - l - a l k a n o l s , while is negative for DLPA---cationic surfactants. This means that the interaction between 1-alkanols is stronger in the gel state than in the liquid-crystalline state m e m b r a n e of the D P P C vesicle, whereas the interaction between cationic surfactants is stronger in the liquid-crystalline state than in the gel state m e m b r a n e of the D L P A vesicle. The signs o f these parameters determine the pattern o f the biphasic change in the Tin-concentration profile; concave ( D P P C - - l - a l k a n o l s ) or convex ( D L P A - - c a t i o n i c surfactants). On the other hand, a c o m m o n aspect is seen for AG A in the two cases. The sign o f AGA changes f r o m negative to positive with increasing hydrocarbon chain length, although the chain length at which the sign changes is different. This indicates that, in general, the long-chain amphiphilic additives have a higher affinity to the gel state m e m b r a n e than to the liquid-crystalline state membrane, as their hydrocarbon chains become longer. The results obtained with D L P A - - c a t i o n i c surfactants, together with those with D P P C - - l - a l k a n ols, demonstrate that which state (gel or liquidcrystalline) is preferable for the intermolecular interaction between the additives and for the distribution of the additives, is dependent on the combination o f the lipids and the additives; i.e. what type of additives are incorporated in what type of lipid membranes. Conclusion
TABLE I Estimated values of M/A, ASA, AGA and E - E, for cationic surfactants interacting with DLPA vesicle membrane. The units are kcal tool-~ for M/A, AGA and ez - ~land cal K-~ tool-~ for ASA. Surfactant
m
M-/A
ASA
AGA
Et - E,
OTABr DeTABr DTABr TTABr
8 10 12 14
3.01 3.77 4.52 5.27
10.6 12.1 13.6 15.1
-0.215 0.074 0.364 0.653
-
1.12 1.40 1.68 1.96
The present study reveals some aspects o f the interaction o f surfactants with the vesicle membrane of the negatively charged lipid, D L P A , through their effect on the phase transition o f the lipid bilayer. The conclusions are summarized as follows. (i) The interaction o f anionic sodium alkylsulfates with D L P A vesicle m e m b r a n e is weak, as expected f r o m the electrostatic repulsion between negative charges Of the surfactants and the lipid. (ii) Non-ionic alkanoyl-N-methylglucamides depress the phase transition temperature monot-
196
onously with increasing concentration. The chain-length dependence of the partition coefficients of the surfactants between bulk water and lipid membrane, which were estimated from the depression of the phase transition temperature, suggests a significance of the hydrophobic interaction for the incorporation of the surfactants into the lipid membrane. (iii) Cationic aikyltrimethylammonium bromides affect the phase transition temperature in a complicated manner depending on the hydrocarbon chain length and the concentration. This behavior of the phase transition temperature is well explained by the statistical mechanical theory taking into account the intermolecular interaction between the surfactants in the lipid membrane. The theory reveals that the intermolecular interaction is stronger in the liquid-crystailine state than in the gel state membrane and that the surfactants with hydrocarbon chain longer than DeTABr have a higher affinity in the gel state than in the liquid-crystalline state membrane.
2
3 4 5 6 7 8
9 10 11 12 13 14
Acknowledgment
15
The authors wish to thank Professors M. Tanaka and G. Sugihara of Fukuoka University for their kind permission to use the DSC system in their laboratory. This work was supported in part by funds from the Central Research Institute of Fukuoka University.
16
References
20
1
D. Lichtenberg, R.J. Robson and E.A. Dennis (1983) Biochim. Biophys. Acta 737, 285--304.
17 18 19
21
A. Alonso, M.-A. Urbaneja, F.M. Gofii, F.G. Carmona, F.G. C~novas and J.C. G6mez-Fern~mdez (1987) Biochim. Biopbys. Acta 902, 237--246. A. Alonso, R. Saez, A. Villena and F.M. Gofii (1982) J. Membr. Biol. 67, 55--62. R. Saez, F.M. Gofii and A. Alonso (1985) FEBS Lett. 179, 311--315. M. Ueno, C. Tanford and J.A. Reynolds (1984) Biochemistry 23, 3070--3076. P. Schurtenberger, N. Mazer and W. Kanzi8 (1985) J. Phys. Chem. 89, 1042--1049. S. Almog, T. Kushnir, S. Nir and D. Lichtenber8 (1986) Biochemistry 25, 2597--2605. F.M. Gofii, M.-A. Urbaneja, J.L.R. Arrondo, A. Alonso, A.A. Durrani and D. Chapman (1986) Eur. J. Biochem. 160, 659--665. T. Inoue, K. Miyakawa and R. Shimozawa (1986) Chem. Phys. Lipids 42, 261--270. T. Inoue, T. Iwanaga, K. Fukushima and R. Shimozawa (1988) Chem. Phys. Lipids 46, 25--30. T. Inoue, K. Fukushima and R. Shimozawa (1988) Bull. Chem. Soc. Jpn. 61, 1565--1569. H.-J. Galla and J.R. Trudell (1980) Biochim. Biophys. Acta 599, 336--340. H. Triiuble and H. Eibl (1974) Proc. Natl. Acad. Sci. U.S.A. 71,214--219. K. Elamrani and A. Blume (1983) Biochemistry 22, 3305--3311. M.W. Hill (1974) Biochim. Biophys. Acta 356, 117-124. H. Kamaya, S. Kaneshina and I. Ueda (1981) Biochim. Biophys. Acta 646, 135--142. S. Kaneshina, H. Kamaya and I. Ueda (1983) J. Colloid Interface Sci. 93, 215--224. C. Tanford (1980) The Hydrophobic Effect, 2nd edn., John Wiley & Sons, New York. T. lnoue, T. lwanaga, K. Fukushima and R. Shimozawa (1988) Chem. Lett., 277--280. H. Kamaya, N. Matuhayasi and I. Ueda (1984) J. Phys. Chem. 88, 797--800. Y. Suezaki, H. Kamaya and I. Ueda (1985) Biochim. Biophys. Acta 818, 31--37.
189
Interaction of surfactants with bilayer of negatively charged lipid: effect on gel-to-liquid-crystalline phase transition of dilauroylphosphatidic acid vesicle membrane Tohru Inoue, Tetsurou Iwanaga, Kohsuke Fukushima, Ryosuke Shimozawa and Yukio Suezaki a Department of Chemistry, Faculty o f Science, Fukuoka University, Nanakuma, Fukuoka 814-01 and aPhysics Laboratory, Saga Medical School, Nebeshima, Saga 840-01 (Japan) (Received March 7th, 1988; revised and accepted May 2nd, 1988)
The interaction of surfactants with the vesicle membrane of the negatively charged lipid, dilauroylphosphatidic acid, was investigated through their effect on the gel-to-liquid-crystaifine phase transition of the lipid bilayer. Three types of surfactants (anionic, cationic and non-ionic) with different hydrocarbon chain length were examined. (i) Anionic sodium aikylsulfates affected the phase transition temperature, T , only weakly. (ii) Non-ionic alkanoyl-N-methylglucamides decreased T monotonously with increasing concentration. The depression of T induced by these surfactants was analyzed by applying the van't Hoff model for the freezing-point depression, and the partition coefficients of the surfactants between bulk water and lipid membrane were estimated. (iii) Cationic alkyltrimethylammonium bromides affected T in a complex manner depending on the hydrocarbon chain length of the surfactants. Octyl-/tetradecyl-trimethyiammonium bromide depressed/elevated T monotonously with increasing concentration, whereas the change in T induced by decyl- and dodecyltrimethylammonium bromides was not monotonous but biphasic. This complex behavior of the phase transition temperature was well explained, based on the statistical mechanical theory presented by Suezaki et al. (Biochim. Biophys. Acta, 818 (1985) 31--37), which takes into account the interaction between surfactant molecules incorporated in the lipid membrane.
Keywords: surfactant-lipid membrane interaction; dilauroylphosphatidic acid; vesicle; phase transition.
Introduction
The interaction of surfactants with a phospholipid vesicle membrane has been widely investigated with interest in their potency as a membrane solubilizer [1,2] or membrane fusogen [2D4], and, more currently, in their utility for the preparation of lipid vesicles (detergent removal method) [5D7]. Most of these studies deal with the mixed micelles formed by the addition of surfactants at a concentration higher than the critical micelle concentration. Recently, Gofli et al. [8] have reported the perturbing effect of surfactants on the structure and prop-
erties of the lipid membrane in rather lower concentration range. Such studies may have a significance different from the above-mentioned practical interest. Since surfactants are typical amphiphiles and the amphiphilicity can be readily controlled by combining various types of head groups and hydrocarbon tails, the surfactant-lipid membrane system may become a useful model system to elucidate the action mechanism of amphiphilic drugs; e.g. alkyltrimethylammonium salts are regarded as a model of tertiary ammine local anesthetic. From this point of view, we have been studying the effect of various surfactants
Correspondence to: T. Inoue. 0009-3084/88/$03.50 © 1988 Elsevier Scientific Publishers Ireland Ltd. Published and Printed in Ireland
190
on the membrane properties of the lipid vesicle paying special attention to the gel-to-liquid-crystalline phase transition [9-- 11 ]. Studies on the interaction of small molecular ligands with lipid bilayers have so far been restricted to the membrane systems composed of zwitterionic phosphatidylcholine, except for a few reports [12]. The probable reasons are that the phosphatidyicholines are the main component of biological membranes and are readily available, and that the structure and properties of their bilayers have been relatively well characterized. However, because cell membranes are usually negatively charged, the study of the interaction of amphiphiles with the membranes of negatively charged lipids may be important for the elucidation of the detailed features of the biologically significant interaction between amphiphilic small molecules and biological membranes. Thus, in the present work, we studied the perturbing effect of surfactants on the gel-to-liquid-crystalline phase transition of the vesicle membrane of negatively charged lipid, dilauroylphosphatidic acid (DLPA). Three types of surfactants, i.e. anionic, cationic and nonionic, were examined. As expected, the perturbing effects of these surfactants were strongly dependent on the electrical charge on the head groups of the surfactants.
Branson Sonifier Model 185 at 35°C for about 15 min. The sample suspension was prepared by mixing the DLPA suspension and the surfactant solution to give a desired surfactant concentration, and by sonicating again at 35°C for about 5 min. The size of vesicles thus prepared was estimated from quasi-elastic light-scattering measurements to be about 140 nm in diameter. The DLPA concentration was kept at about 5 x 10-4 M throughout the experiments unless otherwise noted. The pH levels of the sample suspensions were in the range 7.5 _ 0.2, where DLPA has a monovalent negative charge [13]. The gel-to-liquid-crystalline phase transition of DLPA vesicle membrane was followed by the scattered-light intensity from the vesicle suspension. Details concerning the procedure have been described previously [9]. The differential scan-
~~Tm c
4J
Materials and methods
Synthetic DLPA (monosodium salt) was obtained from Sigma. The surfactants used were: (i) anionic surfactants: sodium octyl-, decyl- and dodecylsulfate (SOS, SDeS, SDS, respectively); (ii) cationic surfactants: octyl-, decyl-, dodecyland tetradecyltrimethylammonium bromide (OTABr, DeTABr, DTABr, TTABr, respectively); (iii) non-ionic surfactants: octanoyl-, nonanoyl- and decanoyl-N-methylglucamide (MEGA-8, MEGA-9, MEGA-10, respectively). The surfactants were purified as described previously [9]. Water was deionized and doubly distilled, once from alkaline potassium permanganate solution. A stock suspension of DLPA vesicle in water was prepared by sonication in the cup-horn of a
l
i
20
T
T
=o
O. I m J l s LAJ
I 20
1
J
~ 30
t
I
t
30
I 40
B
I 40
Temperature /°C Fig. 1. The gel-to-liquid-crystalline phase transition o f D L P A m e m b r a n e observed by scattered-light intensity (A) and DSC (B). Experimental conditions: (A) Additives are (a) none, (b) 4.0 x 10-3 M SDeS, (c) 2.0 x 10-3 M MEGA-9, and (d) 4.1 x 10-t M DTABr. D L P A concentration is about 5 × 10-4 M. Heating rate is 0 . 5 ° C / m i n . (B) D L P A concentration is 4.8 x 10-z M. Heating rate is 1 . 0 ° C / m i n . vesicle
191
ning calorimetry (DSC) measurements were made for pure DLPA vesicle in order to obtain the enthalpy change associated with the phase transition by the Seiko-Denshi DSC system (model SSC). Typical traces of the scattered-light intensity with temperature and the DSC thermogram are shown in Fig. 1. A drastic change in the scattered-fight intensity was observed corresponding to the phase transition. The midpoint temperature of the transition, T e was determined by drawing straight lines as indicated by dotted lines in Fig. 1. The measurements were repeated at least three times for a given sample; a good reproducibility (typically within _+ 0.1 °C) was obtained for T m. T m of pure DLPA obtained for 18 runs was 32.0°C with standard deviation of 0.2°C, which is in agreement with the literature [14]. The transition temperatures measured by light-scattering and DSC are different by about 1 °C. In the DSC experiment, rather concentrated DLPA (about 5 × 10-2 M) was employed. The lower transition temperature obtained by DSC is probably attributed to the high ionic strength of the DLPA suspension. In fact, the transition temperature obtained by light-scattering for 5 × 10-~ M DLPA was decreased to 30.8°C by the addition of 2 × 10-2 M NaCI. The enthaipy change associated with the phase transition of DLPA vesicle membrane was estimated to be 3.4 kcal/mol from the peak area of the DSC thermogram, which agrees with the literature [14]. It should be noted here that the addition of surfactants at high concentration led to the reduced amplitude of the change in the scattered-light intensity accompanied by deformation of the transition curve. Thus, the surfactants were added within the concentration range where normal phase transition behavior was observed. Results and discussion Anionic surfactants
The effect of anionic surfactants on the phase transition temperature of DLPA vesicle mem-
32
I"~o ~D""~O-O~o
30
~J o
(a) 0--
I
32
o
[]
0
O
O --------~O
2
4
(b)
~
30 6
(c)
D ~ [] ~
D
~
D
----a__ rn ----~.
I
I
I
l
2
3
Cs
/ 10-3f4
Fig. 2. Plot of T~e the phase transition temperature of DLPA vesicle membrane, against C~ the concentration of added anionic surfactants. Surfactants are (a) SOS, (b) SDeS and (c) SDS.
brane is shown in Fig. 2, where Tm is plotted against the added surfactant concentration, C . The change in T induced by the addition of these surfactants is small in the measured concentration range, although the decreasing tendency of T is observable, especially when SDS is added. This is in contrast to the case of dipalmitoylphosphatidylcholine (DPPC) - - anionic surfactant systems [ 9 ] , where a significant decrease in Tm was observed by the addition of these surfactants at much lower concentration ranges. The weak interaction between anionic surfactants and DLPA may be understood in terms of the electrostatic repulsion between negative charge of DLPA and anionic surfactants. Non-ionic surfactants
Figure 3 illustrates the relation between T and added surfactant concentration for D L P A - MEGA series surfactant systems. T decreases almost linearly with increasing C with the deviation from the linearity at high surfactant concentration range. Furthermore, the decreasing rate
192
32
T"=T"°
•
\
2.
\o\
~
\ 0
l
I 2
I 3
I 4
I 5
t 6
Cs / 10-3M Fig. 3. Plot of T against C for DLPA--non-ionic surfactant systems. Surfactants are MEGA-8 (A), MEGA-9 (El) and MEGA-IO (V).
of T increases with the hydrocarbon chain length of the surfactants. The depression of the phase transition temperature of phospholipid vesicle membranes induced by the small molecular ligands has been successfully analyzed in terms of the van't H o f f model for the freezing-point depression [15-17]. According to this model, the depression of the phase transition temperature of the lipid bilayer is attributed to the lowering of the chemical potential of lipid molecule in a liquid-crystalline state membrane due to mixing with the additive molecules. Then, the degree of the temperature depression is proportional to the mole fraction of additives in the lipid bilayer for the case of sufficiently low additive concentration. When the additives are partitioned between bulk water and a lipid bilayer, it follows that the partition coefficient of the additives between these phases can be estimated from the relation between the phase transition temperature and the additive concentration. Based on this treatment, T is related to the additive concentration by the following equation under the assumption that the partitioning and mixing of additives is negligible for gel state membrane compared with liquid-crystalline state membrane [9]
R T2~,o K All 55.5+CLK C'
(I)
where T , 0 is the transition temperature without additives, AH the enthalpy change associated with the phase transition, R the gas constant, C L and Cs the total concentrations of lipid and additives in molarity, respectively, and K the partition coefficient defined as K = x A ' / x A where x A' and x A are the mole fractions of the additives in liquid-crystalline membrane and water phases, respectively. According to Eqn. (1) one can estimate the partition coefficient from the slope of the straight line of T vs. C plot with the knowledge of M-/. Equation (1) was applied to the DLPA - - MEGA systems, and the values of K were estimated using the slopes of the linear portions at low concentration region in Fig. 3 and the values of T~,0 = 305 K and M-/ = 3.4 kcal/ tool. The results are as follows: K = 1.12 x 103 (MEGA-8), 2.27 x 10a (MEGA-9) and 7.51 x 103 (MEGA- 10). Figure 4 shows the relation between the partition coefficient and the hydrocarbon chain length of the surfactant, where log K is plotted against the carbon number, Arc, in the hydrocarbon chain. For comparison, the results obtained
4.5
J C~ O
3.5
/ 2.5
2
'
-'
~
,0
Nc
Fig. 4. Relation between log K and N , the carbon number in the hydrocarbon chain of MEGA series surfactants. Lipids are DLPA ( I ) and DPPC (E3).
193
with DPPC [9] are also included in the figure. It may be regarded that log K increases with N in an essentially linear fashion for both cases of DLPA and DPPC. The slope provides a contribution of about R T per methylene unit to the transfer free energy from the bulk water to the lipid membrane. This suggests [18] that the hydrophobic interaction plays an important role for the incorporation of MEGA series surfactants into the membrane of DLPA vesicle as well as DPPC vesicle. The effect of a difference in a polar head group of the lipid is seen in somewhat larger K values obtained with DLPA compared with those obtained with DPPC. This may be interpreted as reflecting the difference in the interaction between the hydrophilic group of MEGA series surfactants and the head groups of the lipids.
Cationic surfactants By the addition of cationic surfactants to DLPA vesicle suspension, T showed a complex behavior depending on the hydrocarbon chain length of the surfactants, as demonstrated in Fig. 5 [19]. Short-chained OTABr decreases T and long-chained TTABr increases it monotonously, whereas DeTABr and DTABr having an
intermediate chain length exhibit a biphasic effect on T . This behavior is quite different from that observed with DPPC vesicle membrane, where T is decreased almost linearly with the concentration regardless of the hydrocarbon chain length of the surfactants [9]. Another remarkable difference between DLPA and DPPC is the required concentration range of the surfactants for the perturbation. For DPPC, the concentration range depends strongly on the hydrocarbon chain length of the surfactants; the shorter chain, the higher concentration [9]. On the other hand, the perturbing effect on DLPA vesicle membrane appears in a similar surfactant concentration range irrespective of the hydrocarbon chain length. Furthermore, the effective concentration range for the perturbation is extremely low compared with those for DPPC vesicle membrane. These facts suggest that the electrostatic interaction between the negative charge of DLPA and the positive charge of cationic surfactants plays a dominant role for the incorporation of the cationic surfactants into DLPA vesicle membrane; the hydrophobic interaction is hidden behind the strong electrostatic interaction. In addition, because of the strong interaction, most of the surfactant added is considered to be 38
36
i - - I
- - i - -
i
•
•
34
N. a4
\
E b~
E b~
32
30
3O
[3
D _ _ E ) [3
- - D
[]
28
26
I .5
I
I
1
t .5
Cs /
I 2
10-4M
Fig. 5. Plot of T against C for DLPA--cationic surfactant systems. Surfact~ts are OTABr (A), DeTABr (121), DTABr ( I ) and TTABr (V). The solid lines were drawn according to Eqn. (2) using the parameters listed in Table I.
26
15 •
I
J 1.5
t
Co~p, /
I 2
I0-3M
Fig. 6. Plot of T against the DLPA concentration for DLPA--cationic surfactant systems. Surfactants are OTABr ( n ) and TTABr ( I ) . The molar ratios of the surfactants to DLPA were fixed at 0.06.
194
incorporated into the vesicle membrane. This is supported by the experimental results presented in Fig. 6, which shows the dependence of T~ on the DLPA concentration with the fixed ratio of the surfactant to the lipid. This figure shows that Tm is almost constant, i.e. Tm is determined by the ratio of the total concentration of surfacrants to that of DLPA regardless of the lipid concentration. Thus, the cationic surfactants are, when added to the DLPA vesicle suspension, all incorporated into the vesicle membrane rather than partitioned between the bulk water and the membrane phase. The most interesting finding in the present study is the biphasic effect of DeTABr and DTABr on the phase transition temperature of DLPA vesicle membrane. The monotonous change in Tm induced by additives is explained based on the van't Hoff model for the freezingpoint depression as mentioned before; the depression/elevation of Tm is caused by the preferential partitioning and mixing of the additives in the liquid-crystalline/gel state membrane. However, the biphasic response of Tm is difficult to explain by the simple thermodynamic treatment of the freezing-point depression. The similar biphasic behavior of the phase transition temperature of lipid bilayers induced by additives has also been reported with DPPC ml-alkanol systems [20], although the direction of the T variation is the reverse of the present case; 1-alkanols with chain length of N = 9m 13 depress the transition temperature at lower concentrations and elevate it at higher concentrations. Suezaki et al. [21] have successfully analyzed this biphasic dependence of T on the 1-alkanol concentration based on the statistical mechanical treatment. The basic idea of their theory is to introduce the intermolecular interaction between additives within the lipid membrane, which depends on the membrane state, gel or liquid-crystalline. When the interaction between additive molecules is stronger in the gel state membrane than in the liquid-crystalline state membrane, it acts to elevate the phase transition temperature, and vice versa. The intermolecular interaction can be reasonably assumed to be a quadratic function of
the additive concentration, whereas the entropic effect in the treatment of freezing-point depression is linear with respect to the additive concentration. The maximum or minimum in the Tm vs. concentration profile can be naturally derived as a result of a combination of these two effects. The theory of Suezaki et al. [21] provides the following expression for the relation between T= and the additive concentration in the membrane phase
1 [AHAx
1 +
r. = T.~
AH
X2
+ (~' - ~') i---~-~x ]
(2)
1 +xASAI&S
where the symbols have the following meanings: AH, AS, the enthalpy and entropy changes of the lipid molecule, respectively, associated with the lipid phase transition from gel-to-liquid-crystalline state; M/A, ASA, the enthalpy and entropy changes of the additive molecule, respectively, when the residing domain changes from gel-toliquid-crystalline state; e~, ~1, the interaction energy parameters between additive molecules in the gel and liquid-crystalline state of the lipid membrane, respectively; x, the molar ratio of the additive to the lipid in the membrane phase. As mentioned above, the intermolecular interaction energy between additives appears in a quadratic form in Eqn. (2). The present results obtained with DLPA - cationic surfactant systems were analyzed based on Eqn. (2). The molar ratio of the surfactant to the lipid in the membrane phase, x, was replaced by the ratio of the total concentration of the surfactant to that of the lipid, because, as mentioned before, almost all the surfactants are regarded to be incorporated into the lipid membrane. M/A, ASA and Es - ~t were assumed to be linear functions of the hydrocarbon chain length of the surfactants, i.e. AH A = AS A Et -
mh
= s,
*l =
+
ms 2
m,
where m represents the carbon number in the
195 hydrocarbon chain o f the surfactant. The values o f four parameters in the above expressions, h, s~, s 2 and E, were estimated by minimizing the quantity S = Y ( T - Tm(exp))2 according to the procedure described previously [21], where T m and T ( e x p ) are the transition temperatures calculated f r o m Eqn. (2) and that observed experimentally, respectively. The best-fit values o f the parameters obtained are as follows: h = 0.377 kcal mol -~, s~ = 4.5 cal K "l, mol -~, s 2 = 0.76 cal K -1 mo1-1 and ~ = - 0 . 1 4 kcal mo1-1. The sensitivity o f these numerical values to the fitness m a y be evaluated by the standard deviation o f T . T h e standard deviation ( q ' ( S / N ) , where N is the n u m b e r o f data points) o f the estimated T was 0.17°C. The 10O7o variation o f one of the four fit parameters, the others being fixed at best-fit values, increased the standard deviation to 1.9°C (s~), 3.5°C (s2), 5.4°C (h) and 0.46°C (~). For each cationic surfactant, the obtained numerical values o f AHA, ASA, e& -- El, and the free energy changes, AGA, defined by
are listed in Table I. The curves shown in Fig. 5 were drawn according to Eqn. (2) using the values o f parameters listed in Table I. Agreement between the experimental data and the theoretical curves is satisfactory except the case of TTABr. It is interesting to compare the parameters in Table I with those obtained with D P P C - - 1-
alkanol systems [21]. The signs o f the parameters M/A, ASA and Es - El are reverse in these two cases. With respect to ~ - EI, it is positive for D P P C - - l - a l k a n o l s , while is negative for DLPA---cationic surfactants. This means that the interaction between 1-alkanols is stronger in the gel state than in the liquid-crystalline state m e m b r a n e of the D P P C vesicle, whereas the interaction between cationic surfactants is stronger in the liquid-crystalline state than in the gel state m e m b r a n e of the D L P A vesicle. The signs o f these parameters determine the pattern o f the biphasic change in the Tin-concentration profile; concave ( D P P C - - l - a l k a n o l s ) or convex ( D L P A - - c a t i o n i c surfactants). On the other hand, a c o m m o n aspect is seen for AG A in the two cases. The sign o f AGA changes f r o m negative to positive with increasing hydrocarbon chain length, although the chain length at which the sign changes is different. This indicates that, in general, the long-chain amphiphilic additives have a higher affinity to the gel state m e m b r a n e than to the liquid-crystalline state membrane, as their hydrocarbon chains become longer. The results obtained with D L P A - - c a t i o n i c surfactants, together with those with D P P C - - l - a l k a n ols, demonstrate that which state (gel or liquidcrystalline) is preferable for the intermolecular interaction between the additives and for the distribution of the additives, is dependent on the combination o f the lipids and the additives; i.e. what type of additives are incorporated in what type of lipid membranes. Conclusion
TABLE I Estimated values of M/A, ASA, AGA and E - E, for cationic surfactants interacting with DLPA vesicle membrane. The units are kcal tool-~ for M/A, AGA and ez - ~land cal K-~ tool-~ for ASA. Surfactant
m
M-/A
ASA
AGA
Et - E,
OTABr DeTABr DTABr TTABr
8 10 12 14
3.01 3.77 4.52 5.27
10.6 12.1 13.6 15.1
-0.215 0.074 0.364 0.653
-
1.12 1.40 1.68 1.96
The present study reveals some aspects o f the interaction o f surfactants with the vesicle membrane of the negatively charged lipid, D L P A , through their effect on the phase transition o f the lipid bilayer. The conclusions are summarized as follows. (i) The interaction o f anionic sodium alkylsulfates with D L P A vesicle m e m b r a n e is weak, as expected f r o m the electrostatic repulsion between negative charges Of the surfactants and the lipid. (ii) Non-ionic alkanoyl-N-methylglucamides depress the phase transition temperature monot-
196
onously with increasing concentration. The chain-length dependence of the partition coefficients of the surfactants between bulk water and lipid membrane, which were estimated from the depression of the phase transition temperature, suggests a significance of the hydrophobic interaction for the incorporation of the surfactants into the lipid membrane. (iii) Cationic aikyltrimethylammonium bromides affect the phase transition temperature in a complicated manner depending on the hydrocarbon chain length and the concentration. This behavior of the phase transition temperature is well explained by the statistical mechanical theory taking into account the intermolecular interaction between the surfactants in the lipid membrane. The theory reveals that the intermolecular interaction is stronger in the liquid-crystailine state than in the gel state membrane and that the surfactants with hydrocarbon chain longer than DeTABr have a higher affinity in the gel state than in the liquid-crystalline state membrane.
2
3 4 5 6 7 8
9 10 11 12 13 14
Acknowledgment
15
The authors wish to thank Professors M. Tanaka and G. Sugihara of Fukuoka University for their kind permission to use the DSC system in their laboratory. This work was supported in part by funds from the Central Research Institute of Fukuoka University.
16
References
20
1
D. Lichtenberg, R.J. Robson and E.A. Dennis (1983) Biochim. Biophys. Acta 737, 285--304.
17 18 19
21
A. Alonso, M.-A. Urbaneja, F.M. Gofii, F.G. Carmona, F.G. C~novas and J.C. G6mez-Fern~mdez (1987) Biochim. Biopbys. Acta 902, 237--246. A. Alonso, R. Saez, A. Villena and F.M. Gofii (1982) J. Membr. Biol. 67, 55--62. R. Saez, F.M. Gofii and A. Alonso (1985) FEBS Lett. 179, 311--315. M. Ueno, C. Tanford and J.A. Reynolds (1984) Biochemistry 23, 3070--3076. P. Schurtenberger, N. Mazer and W. Kanzi8 (1985) J. Phys. Chem. 89, 1042--1049. S. Almog, T. Kushnir, S. Nir and D. Lichtenber8 (1986) Biochemistry 25, 2597--2605. F.M. Gofii, M.-A. Urbaneja, J.L.R. Arrondo, A. Alonso, A.A. Durrani and D. Chapman (1986) Eur. J. Biochem. 160, 659--665. T. Inoue, K. Miyakawa and R. Shimozawa (1986) Chem. Phys. Lipids 42, 261--270. T. Inoue, T. Iwanaga, K. Fukushima and R. Shimozawa (1988) Chem. Phys. Lipids 46, 25--30. T. Inoue, K. Fukushima and R. Shimozawa (1988) Bull. Chem. Soc. Jpn. 61, 1565--1569. H.-J. Galla and J.R. Trudell (1980) Biochim. Biophys. Acta 599, 336--340. H. Triiuble and H. Eibl (1974) Proc. Natl. Acad. Sci. U.S.A. 71,214--219. K. Elamrani and A. Blume (1983) Biochemistry 22, 3305--3311. M.W. Hill (1974) Biochim. Biophys. Acta 356, 117-124. H. Kamaya, S. Kaneshina and I. Ueda (1981) Biochim. Biophys. Acta 646, 135--142. S. Kaneshina, H. Kamaya and I. Ueda (1983) J. Colloid Interface Sci. 93, 215--224. C. Tanford (1980) The Hydrophobic Effect, 2nd edn., John Wiley & Sons, New York. T. lnoue, T. lwanaga, K. Fukushima and R. Shimozawa (1988) Chem. Lett., 277--280. H. Kamaya, N. Matuhayasi and I. Ueda (1984) J. Phys. Chem. 88, 797--800. Y. Suezaki, H. Kamaya and I. Ueda (1985) Biochim. Biophys. Acta 818, 31--37.