ANALYTICAL
BIOCHEMISTRY
201,282-287
(1992)
The Size Dependence of Cholate-Dialyzed on Phosphatidylcholine Concentration
Vesicles
Joseph S. Tauskela,’ Matt Akler, and Michael Thompson Department of Chemistry, University of Toronto, 80 St. GeorgeStreet, Toronto, Ontario, Canada,M5S 1Al
Received
January
14,199l
The size dependence of vesicles prepared by dialysis of cholate from phosphatidylcholine (PC) dispersions has been investigated as a function of lipid concentration (at a constant applied 1ipid:detergent molar ratio of 0.7). Gel filtration of dialyzed samples produced a symmetrical profile shape, although quasielastic laser light scattering analysis of the fractions revealed an asymmetrical range of sizes about the peak for solutions containing elevated lipid concentrations. Vesicle diameters increased by approximately 20 nm for PC concentrations ranging from 10 to a maximum of 45 mg/ml. This was attributed to mixed micelle sizes being proportional to lipid concentration, since the diameters of vesicles produced from dialysis are determined by mixed micelle sizes. Before commencement of dialysis, mixed micelle sizes are proportional to lipid concentration and, although dialysis causes an increase in mixed micelle sizes, the phase ratios attained are larger for solutions containing elevated lipid concentrations. o issz Academic
Press,
Inc.
The slow dialysis of detergent from a mixed phospholipid-detergent micellar solution can produce a homogeneous population of vesicles of a predetermined size range (1,2). This is useful in an important array of applications which require these characteristics; these include model membrane studies, drug/macromolecule encapsulation, and protein-reconstitution efforts [for a review, see (3)]. On a more fundamental level, the fact that a small range of vesicle diameters can be chosen in such a reproducible manner offers insight regarding the processes involved in the transformation from a mixed micelle population to a vesicle dispersion. A mechanism has been proposed by Lasic (4) for the dialysis process,
1 Present address: sity, 3801 University
Montreal Avenue,
Neurological Institute, McGill UniverMontreal, Quebec, Canada, H3A 2B4.
in which a phospholipid-detergent micelle acts as an intermediate in the vesicle formation process. The existence of micelles was originally proposed by others (5,6) in studies of lecithin/bile salt micelles, but Lasic extended this work by hypothesizing that structural and size changes occur as the result of detergent loss. Unfortunately, direct experimental verification of the proposed structures and compositions of the mixed micelles, transition structures, and newly formed vesicles has not been clearly established due to the fact that any perturbation of the systems affects the stability and equilibrium between species (7). As a result of the complexity of the systems and the experimental difficulties in visualizing proposed species, efforts to investigate the mechanism of vesicle formation in more depth have generally proceeded in more indirect directions, with experiments designed in such a fashion that interruption of the transition process does not occur. One approach has entailed examining the reverse of the dialysis process, that is, following the vesicle to mixed micelle transition as increasing amounts of detergent are added to the system (8,9). In a more direct approach, a number of researchers have modified several experimental parameters in the detergent-dialysis technique, not only to better tailor the method toward the specific requirement (e.g., a certain average diameter), but also to obtain a more complete understanding of the processes that are directly relevant to the dialysis procedure. The effect of modifying the type of detergent (especially in terms of the critical micelle concentration) (lo), rate of detergent removal (l&12), pH (13), and applied lipid/detergent molar ratio (at constant lipid concentration) (1,2) on vesicle diameter have all produced results generally consistent with the model of Lasic. The dependence of molar ratio on vesicle size has provided the best understanding of the transition process since results obtained indicated that the order of sizes of the mixed micelles that exist prior to dialysis
282 Copyright 0 1992 All rights of reproduction
0003-2697/92 $3.00 hy Academic Press, Inc. in any form reserved.
FORMATION
OF
PHOSPHATIDYLCHOLINE
(determined by the applied molar ratio) is maintained through the dialysis procedure. A complementary analysis to this has been performed in this study, but in this case the dependence of the vesicle sizes produced on phosphatidylcholine (PC)2 concentration (at a constant applied PC:cholate molar ratio) has been examined. It is suggested that the lipid concentration dictates the sizes that mixed micelles reach during dialysis, producingvesicle diameters which are proportional to lipid concentration. EXPERIMENTAL
Reagents Lyophilized egg phosphatidylcholine was obtained from Avanti Polar Lipids Inc. (Birmingham, AL) and sodium cholate from Anachemia (Mississauga, Ontario, Canada). The exclusion chromatography gel, Sephacryl S-1000, was obtained from Sigma (St. Louis, MO) and the polybead-polystyrene microspheres used for calibration of the quasielastic laser light scattering (QELS) instrument from Polysciences Inc. (Warrington, PA). All other chemicals were of reagent grade. Preparation
of Cholate-Dialyzed
Vesicles
Preparation of the vesicles involved solubilization of egg PC with sodium cholate to form a mixed micelle solution, followed by controlled removal of detergent by dialysis. Mixed micelle solutions were prepared by dissolving the lipid in methanol with detergent and the solvent was evaporated using a rotary evaporator, leaving a thin lipid-cholate film which was then exposed to a nitrogen stream until dry. Since previous studies (1,2) have indicated that there is a size dependence of the vesicles on the 1ipid:detergent molar ratio, a constant ratio of 0.7 was used to ensure reproducibility. At this ratio only mixed micelles were present (6) so the effect of simple micelles (6) or other species (14), which are believed to exist below molar ratios of 0.6, did not have to be considered. The mixture was then dissolved in 10 mM Hepes buffer, pH 7.3, adjusted to 0.16 ionic strength with NaCl and a clear solution resulted. To avoid autooxidation and chemical degradation these solutions were kept under nitrogen and in the dark at 5°C. The solutions were dialyzed within 24 h after their preparation. Detergent removal was accomplished by using a LIPOSOMAT instrument (Dianorm Ltd., Germany). The instrument is composed of two dialyzers separated by a membrane of pure cellulose with a molecular weight cut-off of 10,000 (very high permeability type). To pre-
* Abbreviations used: PC, phosphatidylcholine; QELS, quasielastic laser light scattering; POPC, 1-palmitoyl-Z-oleoylphosphatidylcholine.
VESICLES
BY
283
DIALYSIS
pare the membrane, three soakings for 15 min in distilled water and two soakings in buffer for the same duration were performed. During operation dialysis buffer was continuously pumped through one dialyzer at a flow rate of 2.5 ml/min and was not recycled, while the mixed micellar solution was pumped through a second dialyzer at a flow rate of 0.5 ml/min. Mixed micellar sample volumes were 6.0 ml, and the minimum dialysis time varied depending on the lipid concentration. Nitrogen flowed over the solution during dialysis. Dialysis was performed at room temperature, which is above the transition temperature of egg PC (-7 to -15°C). Sodium cholate was chosen as the detergent because it is removed very quickly due to its high critical micelle concentration. After preparation, the sample was centrifuged for 30 min at 20,000 rpm at 15°C to remove any multilamellar components. Vesicle samples were sealed under a nitrogen atmosphere and kept in the dark at room temperature until use. Gel Filtration
of Vesicles
The gel filtration column employed for characterization of the vesicles was a 2.5 X 50-cm column (Bio-Rad Systems, Toronto, Canada) gravity packed to the 40-cm mark with Sephacryl S-1000 gel. Prior to packing, the gel was soaked a number of times in distilled water and then in buffer for a period of at least 12 h at 5°C. Since Huang (15) has reported that passage of phospholipid vesicles through a Sepharose column results in some adsorption of the lipid (thereby causing larger elution volumes for consecutively run samples), the Sephacryl gel was presaturated with lipid, washed, and equilibrated with buffer. The sample aliquot introduced onto the column was 1.0 ml and the effluent collected had 12 drops per fraction. The absorbance of the effluent was monitored using a 0.2-cm cell path quartz cuvette (Hellma, Toronto, Canada) by a Hewlett-Packard 8451 Diode Array uv-Vis spectrophotometer at a wavelength of 300 nm. Since the alkyl chains of the phospholipids have weak extinction coefficients at this wavelength (16), the signal detected was mainly due to light scattering from the vesicle suspension and not due to uv absorption. Quasielastic
Laser Light
Scattering
Measurements
Quasielastic laser light scattering was used to determine the mean diameter of the vesicles in solution. The device used employed a He-Ne laser (wavelength of 632.8 nm) as the source and the sample was housed in a thermally jacketed scattering chamber. A digital photon counting technique using a 64-channel autocorrelator (Langley-Ford Instruments, Amherst, MA) was employed. Analysis of the resulting digital autocorrelation function was performed on an on-line NOVA-2 minicomputer (Data General Inc., Southboro, Massachu-
284
TAUSKELA,
AKLER,
0.20
g
0.15
0 s Q) 0.10 2 2 $ 2
0.05
0.00 1 20
30
40
Fraction
50
60
70
Number
FIG. 1.
Sephacryl S-1000 gel filtration profile of egg PC vesicles prepared by cholate-dialysis method. The lipid concentration is 25 mg/ml and the lipidcholate molar ratio is 0.7. The profile shape is symmetrical about the peak, except for a small multilamellar component eluting at the void volume. Elution fractions isolated for sizing analysis (identified in Table 1) are highlighted in bold.
setts) using the method of cumulants described by Koppel (17). Polybead-polystyrene microspheres of known size were used as calibration standards. RESULTS
Before the examination of the relationship between egg PC concentration and the sizes of the vesicles produced proceeded, the dialysis conditions and sampling method were optimized. A characteristic gel filtration profile is shown in Fig. 1 for an egg PC vesicle sample (for a lipid concentration equal to 25 mg/ml) prepared by the cholate dialysis method. The procedure sometimes produced a sample having a small multilamellar component which eluted at the void volume but centrifugation usually removed most or all of this fraction. A QELS analysis of the fractions collected around the peak of the turbidity profile for most of the different PC concentrations examined is presented in Table 1. A symmetrical distribution of vesicle diameters only existed at low PC concentrations. At concentrations greater than 20 mg/ml the size difference between the prepeak and peak fractions was greater than that between the postpeak and peak fractions. This trend became more pronounced at larger concentrations, although the difference in sizes for fractions collected at and after the peak was still relatively small. These results clearly show that
AND
THOMPSON
a monomodal distribution of sizes did not exist across all fractions collected so sizing determinations were only performed for the volumes located at the profile peak. Although the appearance of a clear, slightly blue opalescent solution indicated formation of vesicles, gel filtration of the samples was used to confirm the length of time required before the sizing analysis could proceed (as well as to isolate the fractions to be analyzed). Figure 2 presents a typical series of gel filtration profiles for aliquots collected at various intervals throughout the dialysis of a mixed micellar solution (PC concentration of 40 mg/ml). For the mixed micellar solution alone (i.e., at zero time dialysis), elution yielded two broad ill-defined components at high fraction numbers which were attributed to the presence of small vesicles and mixed micelles ($18). The profile of the sample that was collected 2.0 h after initiation of dialysis was not as smooth as the others which followed, but the peak position was at a lower elution volume compared to that for the mixed micelles. Identical elution volumes for the profile peak were produced for dialysis times of greater than or equal to 3.2 h. For dialysis times of up to 6.7 h the area under the profile increased with the amount of time that the mixed micelle solutions were dialyzed. A similar analysis was performed for several of the lower PC concentrations used: constant profile shapes and areas were obtained for dialysis times ranging from 4 h for 10 mg/ml solutions to 5 h for 30 mg/ml samples. For the sizing determinations, elution was performed at least 1 h after these times to ensure that no further changes were occurring to the liposome populations. The diameters of the vesicles produced by the dialysis procedure were examined as a function of concentration of the lipid employed (using a constant molar ratio of
TABLE
1
Quasielastic Laser Light Scattering Analysis of Fractions Collected from Gel Filtration of Cholate-Dialysed PC Vesicles Size” (nm) PC concentration
bg/ml) 15 20 25 40
Prepeak fractions’ 83.8 87.7 94.4 110.2
t f + +
Peak fractions 0.6 0.0 0.9 1.2
82.0 85.9 91.5 100.1
2 + * -t
Postpeak fractions 0.4 0.2 1.2 0.8
80.3 84.3 90.4 97.7
f + + +
0.4 0.3 1.0 0.2
*Values are reported with standard deviations from repeated QELS runs. b Normalizing the fraction number where the peak occurs to be 6, prepeak fractions were l-3, peak fractions were 5-7, and postpeak fractions were 9-11. The elution profile for a solution containing 25 mg/ml PC is plotted in Fig. 1 with the fractions that were analyzed highlighted in bold.
FORMATION
OF
PHOSPHATIDYLCHOLINE
0.00 0
20
40
Fraction
60
80
100
Number
FIG. 2. Sephacryl S-1000 gel filtration profiles for egg PC vesicle samples collected at various intervals during dialysis. Dialysis times (h) are: +, 0.0; X, 2.0; 0, 3.2; A, 4.0; Cl, 6.7; and Ei, 8.7. The lipid concentration is 40 mg/ml and the 1ipid:cholate molar ratio is 0.7. Profile peak positions and areas only become constant at 3.2 and 6.7 h dialysis time, respectively.
1ipid:cholate equal to 0.7). Figure 3 shows that increases in vesicle sizes resulted as the egg PC concentration was increased, up to concentrations of approximately 45 mg/ml. At concentrations greater than this, dialysis resulted in a transformation of the clear mixed micelle solutions to very viscous milky solutions, characteristic of multilamellar dispersions. DISCUSSION
The size dependence of vesicles on lipid concentration was examined in an effort to obtain a better understanding of the processes that occur in the transition from mixed micelles to liposomes during detergent dialysis. Several measures were taken to ensure that the diameters measured by QELS analysis were representative of the particular concentration examined. The first step in this analysis was to characterize vesicle populations by gel filtration. The multilamellar component that was occasionally observed (Fig. 1) has not been reported by others who have employed this method (1,2). This may be due to the fact that the column packing material employed in this study, Sephacryl S-1000, has a much larger exclusion limit (-300 nm) (19) than the Sepharose gels that were employed elsewhere, allowing better resolution of multilamellar and vesicle components.
VESICLES
BY
285
DIALYSIS
QELS sizing determinations of gel filtered vesicle fractions revealed that a range of vesicle diameters exist for each sample prepared by detergent dialysis (Table 1). An asymmetrical distribution of vesicle diameters was observed about the peak of the turbidity profile but the range of sizes was too small to determine whether a logarithmic dependence of vesicular diameters on elution volume existed (20). Previous studies have shown such a dependence for columns packed with Sepharose 4B and Sephacryl S-1000 gels but this was for a much larger size range, in which different marker proteins and Latex beads of known diameter were employed as calibration material (21,22). Since the method of cumulants is only applicable for a monomodal population of sizes (17), three fractions of eluant collected at the turbidity peak profile were used for size determinations. The reproducibility of peak fraction position and areas was used as an indication for completion of dialysis (Fig. 2). The increase in the gel filtration profile areas for dialysis periods longer than the time required to attain maximum vesicle diameters implies that a mixed micelle component existed prior to that time. When reproducible areas were obtained, this component did not form a large fraction of the total pool. Dialysis removes 98.0-99.5% (2) or greater than 99.5% (1) of cholate when high-molecular-weight cut-off mem-
145.0 i j
130.0 I
2 '; 115.0 E ; o 100.0 z 2 +
i
85.0
70.0 L-__--.i-i_.u 0
15 PC1
30
45
60
(mg/ml)
FIG. 3. Size dependence of cholate-dialyzed vesicles on egg PC concentration at a constant lipidxholate molar ratio equal to 0.7. Diameters were determined only for those fractions collected at the peak of the gel filtration profile. Vesicle diameters increased with PC concentration up to 45 mg/ml, after which only a multilamellar component resulted.
286
TAUSKELA,
AKLER,
branes are used. Although the intention of employing gel filtration was to isolate fractions for sizing determinations, elution of liposome preparations through Sepharose, Sephadex, or Sephacryl columns causes further removal of cholate, up to 99.8% of the original amount present (l&21). Thus, it can be concluded that the variation in sizes observed as a function of concentration (Fig. 3) was not due to incomplete dialysis, nor to limitations in the QELS measurement technique. The kinetics of vesicle formation can be divided into two steps: structural changes that occur to the mixed micelles during dialysis and then the actual transformation to vesicles. The current understanding of the mixed micelle structure is that it consists of a lecithin bilayer containing detergent within both inner and outer monolayers (4,6). While there is some controversy regarding the exact form these assemblies take (23-27), it is agreed that detergent molecules are oriented with their hydrophilic sides in contact with the aqueous solvent and the opposite hydrophobic sides interacting with the acyl chains of lipid molecules. Solubilization of lipid by detergent results in the formation of relatively clear solutions. Above 45 mg/ml, dialysis resulted in simply not enough detergent being available to solubilize the large number of PC molecules present, so a multilamellar component was formed. A key concept in the mixed micelle model is a fixed molar ratio of PC:detergent at constant lipid (and therefore detergent) concentration. The result of this is that the average size of the mixed micelle depends on the relative (1,2) and absolute (23) amounts of detergent and lipid that are present in the system. In our study, all solutions had the same applied PC:cholate molar ratio of 0.7, but the actual mixed micelle phase ratios for solutions of different lipid concentrations were quite different from this value. Nichols and Ozarowski (23) has recently found that the ratio of 1-palmitoyl-2-oleoylphosphatidylcholine (POPC) to deoxycholate (and therefore the mixed micelle size) is proportional to lipid concentration. Since the diameters of vesicles that are produced depend on mixed micelle sizes (1,2), the trend observed in Fig. 3 indicates that the size of the mixed micelles that exist even before dialysis proceeds dictates the diameters of the vesicles formed (assuming complete removal of detergent), assuming the PC-cholate system behaves in a similar manner to the POPC-deoxycholate system. Loss of detergent during dialysis causes an increase in the mixed micelle sizes, since this is the only way the required shielding of the nonpolar acyl chains from H,O can be maintained. Thus, the fact that vesicle diameters are proportional to lipid concentration implies that the phase limits (the 1ipid:detergent molar ratio that is attained before formation of vesicles occurs) are larger for solutions containing elevated lipid concentrations.
AND
THOMPSON
A similar dependence of vesicle size on PC concentration has been noted in one previous study in which n-alkyl glucosides were used (lo), but the opposite trend has been observed in another (1) in which cholate was employed. In this latter study, though, a change from 6.5 to 13 mg/ml caused only a small decrease in radii from 30.0 + 2.5 nm (from ultracentrifugal diffusion measurements) to 26.6 + 1.5 nm (from QELS measurements). This small concentration range and the variability in measuring techniques may not allow a general trend to be discerned. Alternatively, it is noted that vesicle diameters (Fig. 3) are at least 20 nm larger than those observed by Zumbuehl and Weder (2) for similar PC concentrations and PC:cholate molar ratios. The larger sizes for similar lipid concentrations must be due to an effective slower dialysis rate (1) (direct comparison of rates is not possible since the systems employed are not identical). Faster kinetics not only results in smallersized vesicles but may not produce much of a difference in sizes for various PC concentrations. In summary, the mixed disc model of Lasic (4) has been extended to account for larger-diameter vesicles existing in solutions containing elevated PC concentrations. Vesicle diameters depend on the phase limit and this, combined with the fact that the PC:cholate molar ratio that exists prior to dialysis is proportional to lipid concentration, allows us to conclude that the lipid concentration dictates the phase limit for the mixed micelles. Further investigation is warranted to determine whether these findings can be generalized to other 1ipid:detergent systems. As discussed elsewhere (23), it is not clear why the distribution of free and micellar detergent is independent of lipid concentration. One explanation offered (23) is that at elevated POPC concentrations the rate of free deoxycholate association with micelles is lowered while the rate of detergent dissociation remains the same, resulting in a reduction of the amount of micellar detergent in equilibrium with free species. These observations highlight the need for an accurate structure of the mixed micelle. Unfortunately, there is some disagreement regarding the shape of lecithin-bile salt micelles, with proposals of mixed-discs (4,6), rod-shapes (24-27), and capped-rods (23) being suggested. Analysis of the dependence of vesicle sizes on the concentrations of different types of lipid and detergent may allow the more accurate model to be chosen. For instance, various lipids and detergents may be chosen according to certain criteria (possibly differences in chain length, charge, or headgroup size) and any changes in behaviour noted may be correlated with structural differences. ACKNOWLEDGMENTS Support Research
for this Council
work from of Canada
the Natural is gratefully
Sciences and Engineering acknowledged. We also
FORMATION
OF
PHOSPHATIDYLCHOLINE
thank Dr. Ross Hallet and Jackie Marsh at the University of Guelph, Ontario for the use and training on the QELS instrumentation.
VESICLES
12. Lasic, 13. King,
BY
287
DIALYSIS
D. D. (1987) J. Z’heor. R. G., and Marchbanks,
Biol. 124, 35-41. R. M.
(1982)
Biochim. Biophys.
Acta 691,183-187. 14. Muller,
REFERENCES
15. Huang,
1. Milsmann, M., Schwendener, try 512,147-155.
R., and Weder,
H.
(1978) Biochemk-
D. D. (1982)
5. Small,
D. M.,
B&him.
Penkett,
N. A., Benedek, D. D. (1988)
S. A., and Chapman,
D. (1969)
Biochim.
G. B., and Carey,
M.
J. A., Nozaki,
20. Ackers,
G. K. (1967)
21. Brunner,
J, Skrabal,
C. (1980) Biochemis-
Y., and Tanford,
C. (1983)
And. Bio-
Biochem. J. 256, l-11.
chemistry 25,2597-2605. R. A., Asanger, M., and Weder, H. G. (1981) Biothem. Biophys. Res. Commun. 100,1055-1062. 11. Fischer, T. H., and Lasic, D. D. (1984) Mol. Cryst. Liq. Cryst. Lett. 102.141-153.
J. Biol. Chem. 242,3237-3238. P., and Hauser, H. (1976) Biochim. Biophys.
Acta 455,322-331. 22.
8. Schurtenberger, P., Mazer, N. A., and Kanzig, R. (1984) in Surfactants in Solution (Mittal, K. L., and Lindman, K., Eds.), Vol. 2, pp. 841-855, Plenum, New York. 9. Almog, S., Kushnir, T., Nir, S., and Lichtenberg, D. (1986) Bio10. Schwendener,
19. Reynolds,
B&him.
them. 130,471-474.
try19,601-615. 7. Lasic,
17. Koppel, D. E. (1972) J. Chem. Phys. 57,4814-4820. 18. Kramer, R. M., Hasselbach, J., and Simenza, G. (1981) Biophys. Acta 643,233-242.
Biophys. Acta 692,501-502.
Biophys. Acta 176,178-179. 6. Mazer,
Biophys. Acta 298,
1015-1019.
2. Zumbuehl, O., and Weder, H. (1981) Biochim. Biophys. Acta 640, 252-262. 3. Schmidt, K. H., (Ed.) (1986) Liposomes as Drug Carriers, Georg Thieme Verlag, New York. 4. Lasic,
16. Batzri,
K. (1981) Biochemistry 20,404-414. C. (1969) Biochemistry 8,344-352. S., and Korn, E. D. (1973) Biochim.
Schurtenberger,
P., and Hauser,
H. (1984)
B&him.
Biophys. Acta
778,470-480. 23. Nichols, J. W., and Ozarowski, J. (1990) Biochemistry 29, 46004606. 24. Shankland, W. (1970) Chem. Phys. Lipids 4, 109-130. 25. Hjelm, R. P., Jr., and Alkan, H. (1988) J. Appl. Crystallog. 21, 858-863. 26. Hjelm, R. P., Jr., Thiyagarajan, P., and Alkan, H. (1990) Mol. Cryst. Liq. Cryst. 18OA, 155-164. 27. Walter, A., Vinson, P. K., and Talmon, Y. (1990) Biophys. J. 67, 476a.
BIOCHEMISTRY
201,282-287
(1992)
The Size Dependence of Cholate-Dialyzed on Phosphatidylcholine Concentration
Vesicles
Joseph S. Tauskela,’ Matt Akler, and Michael Thompson Department of Chemistry, University of Toronto, 80 St. GeorgeStreet, Toronto, Ontario, Canada,M5S 1Al
Received
January
14,199l
The size dependence of vesicles prepared by dialysis of cholate from phosphatidylcholine (PC) dispersions has been investigated as a function of lipid concentration (at a constant applied 1ipid:detergent molar ratio of 0.7). Gel filtration of dialyzed samples produced a symmetrical profile shape, although quasielastic laser light scattering analysis of the fractions revealed an asymmetrical range of sizes about the peak for solutions containing elevated lipid concentrations. Vesicle diameters increased by approximately 20 nm for PC concentrations ranging from 10 to a maximum of 45 mg/ml. This was attributed to mixed micelle sizes being proportional to lipid concentration, since the diameters of vesicles produced from dialysis are determined by mixed micelle sizes. Before commencement of dialysis, mixed micelle sizes are proportional to lipid concentration and, although dialysis causes an increase in mixed micelle sizes, the phase ratios attained are larger for solutions containing elevated lipid concentrations. o issz Academic
Press,
Inc.
The slow dialysis of detergent from a mixed phospholipid-detergent micellar solution can produce a homogeneous population of vesicles of a predetermined size range (1,2). This is useful in an important array of applications which require these characteristics; these include model membrane studies, drug/macromolecule encapsulation, and protein-reconstitution efforts [for a review, see (3)]. On a more fundamental level, the fact that a small range of vesicle diameters can be chosen in such a reproducible manner offers insight regarding the processes involved in the transformation from a mixed micelle population to a vesicle dispersion. A mechanism has been proposed by Lasic (4) for the dialysis process,
1 Present address: sity, 3801 University
Montreal Avenue,
Neurological Institute, McGill UniverMontreal, Quebec, Canada, H3A 2B4.
in which a phospholipid-detergent micelle acts as an intermediate in the vesicle formation process. The existence of micelles was originally proposed by others (5,6) in studies of lecithin/bile salt micelles, but Lasic extended this work by hypothesizing that structural and size changes occur as the result of detergent loss. Unfortunately, direct experimental verification of the proposed structures and compositions of the mixed micelles, transition structures, and newly formed vesicles has not been clearly established due to the fact that any perturbation of the systems affects the stability and equilibrium between species (7). As a result of the complexity of the systems and the experimental difficulties in visualizing proposed species, efforts to investigate the mechanism of vesicle formation in more depth have generally proceeded in more indirect directions, with experiments designed in such a fashion that interruption of the transition process does not occur. One approach has entailed examining the reverse of the dialysis process, that is, following the vesicle to mixed micelle transition as increasing amounts of detergent are added to the system (8,9). In a more direct approach, a number of researchers have modified several experimental parameters in the detergent-dialysis technique, not only to better tailor the method toward the specific requirement (e.g., a certain average diameter), but also to obtain a more complete understanding of the processes that are directly relevant to the dialysis procedure. The effect of modifying the type of detergent (especially in terms of the critical micelle concentration) (lo), rate of detergent removal (l&12), pH (13), and applied lipid/detergent molar ratio (at constant lipid concentration) (1,2) on vesicle diameter have all produced results generally consistent with the model of Lasic. The dependence of molar ratio on vesicle size has provided the best understanding of the transition process since results obtained indicated that the order of sizes of the mixed micelles that exist prior to dialysis
282 Copyright 0 1992 All rights of reproduction
0003-2697/92 $3.00 hy Academic Press, Inc. in any form reserved.
FORMATION
OF
PHOSPHATIDYLCHOLINE
(determined by the applied molar ratio) is maintained through the dialysis procedure. A complementary analysis to this has been performed in this study, but in this case the dependence of the vesicle sizes produced on phosphatidylcholine (PC)2 concentration (at a constant applied PC:cholate molar ratio) has been examined. It is suggested that the lipid concentration dictates the sizes that mixed micelles reach during dialysis, producingvesicle diameters which are proportional to lipid concentration. EXPERIMENTAL
Reagents Lyophilized egg phosphatidylcholine was obtained from Avanti Polar Lipids Inc. (Birmingham, AL) and sodium cholate from Anachemia (Mississauga, Ontario, Canada). The exclusion chromatography gel, Sephacryl S-1000, was obtained from Sigma (St. Louis, MO) and the polybead-polystyrene microspheres used for calibration of the quasielastic laser light scattering (QELS) instrument from Polysciences Inc. (Warrington, PA). All other chemicals were of reagent grade. Preparation
of Cholate-Dialyzed
Vesicles
Preparation of the vesicles involved solubilization of egg PC with sodium cholate to form a mixed micelle solution, followed by controlled removal of detergent by dialysis. Mixed micelle solutions were prepared by dissolving the lipid in methanol with detergent and the solvent was evaporated using a rotary evaporator, leaving a thin lipid-cholate film which was then exposed to a nitrogen stream until dry. Since previous studies (1,2) have indicated that there is a size dependence of the vesicles on the 1ipid:detergent molar ratio, a constant ratio of 0.7 was used to ensure reproducibility. At this ratio only mixed micelles were present (6) so the effect of simple micelles (6) or other species (14), which are believed to exist below molar ratios of 0.6, did not have to be considered. The mixture was then dissolved in 10 mM Hepes buffer, pH 7.3, adjusted to 0.16 ionic strength with NaCl and a clear solution resulted. To avoid autooxidation and chemical degradation these solutions were kept under nitrogen and in the dark at 5°C. The solutions were dialyzed within 24 h after their preparation. Detergent removal was accomplished by using a LIPOSOMAT instrument (Dianorm Ltd., Germany). The instrument is composed of two dialyzers separated by a membrane of pure cellulose with a molecular weight cut-off of 10,000 (very high permeability type). To pre-
* Abbreviations used: PC, phosphatidylcholine; QELS, quasielastic laser light scattering; POPC, 1-palmitoyl-Z-oleoylphosphatidylcholine.
VESICLES
BY
283
DIALYSIS
pare the membrane, three soakings for 15 min in distilled water and two soakings in buffer for the same duration were performed. During operation dialysis buffer was continuously pumped through one dialyzer at a flow rate of 2.5 ml/min and was not recycled, while the mixed micellar solution was pumped through a second dialyzer at a flow rate of 0.5 ml/min. Mixed micellar sample volumes were 6.0 ml, and the minimum dialysis time varied depending on the lipid concentration. Nitrogen flowed over the solution during dialysis. Dialysis was performed at room temperature, which is above the transition temperature of egg PC (-7 to -15°C). Sodium cholate was chosen as the detergent because it is removed very quickly due to its high critical micelle concentration. After preparation, the sample was centrifuged for 30 min at 20,000 rpm at 15°C to remove any multilamellar components. Vesicle samples were sealed under a nitrogen atmosphere and kept in the dark at room temperature until use. Gel Filtration
of Vesicles
The gel filtration column employed for characterization of the vesicles was a 2.5 X 50-cm column (Bio-Rad Systems, Toronto, Canada) gravity packed to the 40-cm mark with Sephacryl S-1000 gel. Prior to packing, the gel was soaked a number of times in distilled water and then in buffer for a period of at least 12 h at 5°C. Since Huang (15) has reported that passage of phospholipid vesicles through a Sepharose column results in some adsorption of the lipid (thereby causing larger elution volumes for consecutively run samples), the Sephacryl gel was presaturated with lipid, washed, and equilibrated with buffer. The sample aliquot introduced onto the column was 1.0 ml and the effluent collected had 12 drops per fraction. The absorbance of the effluent was monitored using a 0.2-cm cell path quartz cuvette (Hellma, Toronto, Canada) by a Hewlett-Packard 8451 Diode Array uv-Vis spectrophotometer at a wavelength of 300 nm. Since the alkyl chains of the phospholipids have weak extinction coefficients at this wavelength (16), the signal detected was mainly due to light scattering from the vesicle suspension and not due to uv absorption. Quasielastic
Laser Light
Scattering
Measurements
Quasielastic laser light scattering was used to determine the mean diameter of the vesicles in solution. The device used employed a He-Ne laser (wavelength of 632.8 nm) as the source and the sample was housed in a thermally jacketed scattering chamber. A digital photon counting technique using a 64-channel autocorrelator (Langley-Ford Instruments, Amherst, MA) was employed. Analysis of the resulting digital autocorrelation function was performed on an on-line NOVA-2 minicomputer (Data General Inc., Southboro, Massachu-
284
TAUSKELA,
AKLER,
0.20
g
0.15
0 s Q) 0.10 2 2 $ 2
0.05
0.00 1 20
30
40
Fraction
50
60
70
Number
FIG. 1.
Sephacryl S-1000 gel filtration profile of egg PC vesicles prepared by cholate-dialysis method. The lipid concentration is 25 mg/ml and the lipidcholate molar ratio is 0.7. The profile shape is symmetrical about the peak, except for a small multilamellar component eluting at the void volume. Elution fractions isolated for sizing analysis (identified in Table 1) are highlighted in bold.
setts) using the method of cumulants described by Koppel (17). Polybead-polystyrene microspheres of known size were used as calibration standards. RESULTS
Before the examination of the relationship between egg PC concentration and the sizes of the vesicles produced proceeded, the dialysis conditions and sampling method were optimized. A characteristic gel filtration profile is shown in Fig. 1 for an egg PC vesicle sample (for a lipid concentration equal to 25 mg/ml) prepared by the cholate dialysis method. The procedure sometimes produced a sample having a small multilamellar component which eluted at the void volume but centrifugation usually removed most or all of this fraction. A QELS analysis of the fractions collected around the peak of the turbidity profile for most of the different PC concentrations examined is presented in Table 1. A symmetrical distribution of vesicle diameters only existed at low PC concentrations. At concentrations greater than 20 mg/ml the size difference between the prepeak and peak fractions was greater than that between the postpeak and peak fractions. This trend became more pronounced at larger concentrations, although the difference in sizes for fractions collected at and after the peak was still relatively small. These results clearly show that
AND
THOMPSON
a monomodal distribution of sizes did not exist across all fractions collected so sizing determinations were only performed for the volumes located at the profile peak. Although the appearance of a clear, slightly blue opalescent solution indicated formation of vesicles, gel filtration of the samples was used to confirm the length of time required before the sizing analysis could proceed (as well as to isolate the fractions to be analyzed). Figure 2 presents a typical series of gel filtration profiles for aliquots collected at various intervals throughout the dialysis of a mixed micellar solution (PC concentration of 40 mg/ml). For the mixed micellar solution alone (i.e., at zero time dialysis), elution yielded two broad ill-defined components at high fraction numbers which were attributed to the presence of small vesicles and mixed micelles ($18). The profile of the sample that was collected 2.0 h after initiation of dialysis was not as smooth as the others which followed, but the peak position was at a lower elution volume compared to that for the mixed micelles. Identical elution volumes for the profile peak were produced for dialysis times of greater than or equal to 3.2 h. For dialysis times of up to 6.7 h the area under the profile increased with the amount of time that the mixed micelle solutions were dialyzed. A similar analysis was performed for several of the lower PC concentrations used: constant profile shapes and areas were obtained for dialysis times ranging from 4 h for 10 mg/ml solutions to 5 h for 30 mg/ml samples. For the sizing determinations, elution was performed at least 1 h after these times to ensure that no further changes were occurring to the liposome populations. The diameters of the vesicles produced by the dialysis procedure were examined as a function of concentration of the lipid employed (using a constant molar ratio of
TABLE
1
Quasielastic Laser Light Scattering Analysis of Fractions Collected from Gel Filtration of Cholate-Dialysed PC Vesicles Size” (nm) PC concentration
bg/ml) 15 20 25 40
Prepeak fractions’ 83.8 87.7 94.4 110.2
t f + +
Peak fractions 0.6 0.0 0.9 1.2
82.0 85.9 91.5 100.1
2 + * -t
Postpeak fractions 0.4 0.2 1.2 0.8
80.3 84.3 90.4 97.7
f + + +
0.4 0.3 1.0 0.2
*Values are reported with standard deviations from repeated QELS runs. b Normalizing the fraction number where the peak occurs to be 6, prepeak fractions were l-3, peak fractions were 5-7, and postpeak fractions were 9-11. The elution profile for a solution containing 25 mg/ml PC is plotted in Fig. 1 with the fractions that were analyzed highlighted in bold.
FORMATION
OF
PHOSPHATIDYLCHOLINE
0.00 0
20
40
Fraction
60
80
100
Number
FIG. 2. Sephacryl S-1000 gel filtration profiles for egg PC vesicle samples collected at various intervals during dialysis. Dialysis times (h) are: +, 0.0; X, 2.0; 0, 3.2; A, 4.0; Cl, 6.7; and Ei, 8.7. The lipid concentration is 40 mg/ml and the 1ipid:cholate molar ratio is 0.7. Profile peak positions and areas only become constant at 3.2 and 6.7 h dialysis time, respectively.
1ipid:cholate equal to 0.7). Figure 3 shows that increases in vesicle sizes resulted as the egg PC concentration was increased, up to concentrations of approximately 45 mg/ml. At concentrations greater than this, dialysis resulted in a transformation of the clear mixed micelle solutions to very viscous milky solutions, characteristic of multilamellar dispersions. DISCUSSION
The size dependence of vesicles on lipid concentration was examined in an effort to obtain a better understanding of the processes that occur in the transition from mixed micelles to liposomes during detergent dialysis. Several measures were taken to ensure that the diameters measured by QELS analysis were representative of the particular concentration examined. The first step in this analysis was to characterize vesicle populations by gel filtration. The multilamellar component that was occasionally observed (Fig. 1) has not been reported by others who have employed this method (1,2). This may be due to the fact that the column packing material employed in this study, Sephacryl S-1000, has a much larger exclusion limit (-300 nm) (19) than the Sepharose gels that were employed elsewhere, allowing better resolution of multilamellar and vesicle components.
VESICLES
BY
285
DIALYSIS
QELS sizing determinations of gel filtered vesicle fractions revealed that a range of vesicle diameters exist for each sample prepared by detergent dialysis (Table 1). An asymmetrical distribution of vesicle diameters was observed about the peak of the turbidity profile but the range of sizes was too small to determine whether a logarithmic dependence of vesicular diameters on elution volume existed (20). Previous studies have shown such a dependence for columns packed with Sepharose 4B and Sephacryl S-1000 gels but this was for a much larger size range, in which different marker proteins and Latex beads of known diameter were employed as calibration material (21,22). Since the method of cumulants is only applicable for a monomodal population of sizes (17), three fractions of eluant collected at the turbidity peak profile were used for size determinations. The reproducibility of peak fraction position and areas was used as an indication for completion of dialysis (Fig. 2). The increase in the gel filtration profile areas for dialysis periods longer than the time required to attain maximum vesicle diameters implies that a mixed micelle component existed prior to that time. When reproducible areas were obtained, this component did not form a large fraction of the total pool. Dialysis removes 98.0-99.5% (2) or greater than 99.5% (1) of cholate when high-molecular-weight cut-off mem-
145.0 i j
130.0 I
2 '; 115.0 E ; o 100.0 z 2 +
i
85.0
70.0 L-__--.i-i_.u 0
15 PC1
30
45
60
(mg/ml)
FIG. 3. Size dependence of cholate-dialyzed vesicles on egg PC concentration at a constant lipidxholate molar ratio equal to 0.7. Diameters were determined only for those fractions collected at the peak of the gel filtration profile. Vesicle diameters increased with PC concentration up to 45 mg/ml, after which only a multilamellar component resulted.
286
TAUSKELA,
AKLER,
branes are used. Although the intention of employing gel filtration was to isolate fractions for sizing determinations, elution of liposome preparations through Sepharose, Sephadex, or Sephacryl columns causes further removal of cholate, up to 99.8% of the original amount present (l&21). Thus, it can be concluded that the variation in sizes observed as a function of concentration (Fig. 3) was not due to incomplete dialysis, nor to limitations in the QELS measurement technique. The kinetics of vesicle formation can be divided into two steps: structural changes that occur to the mixed micelles during dialysis and then the actual transformation to vesicles. The current understanding of the mixed micelle structure is that it consists of a lecithin bilayer containing detergent within both inner and outer monolayers (4,6). While there is some controversy regarding the exact form these assemblies take (23-27), it is agreed that detergent molecules are oriented with their hydrophilic sides in contact with the aqueous solvent and the opposite hydrophobic sides interacting with the acyl chains of lipid molecules. Solubilization of lipid by detergent results in the formation of relatively clear solutions. Above 45 mg/ml, dialysis resulted in simply not enough detergent being available to solubilize the large number of PC molecules present, so a multilamellar component was formed. A key concept in the mixed micelle model is a fixed molar ratio of PC:detergent at constant lipid (and therefore detergent) concentration. The result of this is that the average size of the mixed micelle depends on the relative (1,2) and absolute (23) amounts of detergent and lipid that are present in the system. In our study, all solutions had the same applied PC:cholate molar ratio of 0.7, but the actual mixed micelle phase ratios for solutions of different lipid concentrations were quite different from this value. Nichols and Ozarowski (23) has recently found that the ratio of 1-palmitoyl-2-oleoylphosphatidylcholine (POPC) to deoxycholate (and therefore the mixed micelle size) is proportional to lipid concentration. Since the diameters of vesicles that are produced depend on mixed micelle sizes (1,2), the trend observed in Fig. 3 indicates that the size of the mixed micelles that exist even before dialysis proceeds dictates the diameters of the vesicles formed (assuming complete removal of detergent), assuming the PC-cholate system behaves in a similar manner to the POPC-deoxycholate system. Loss of detergent during dialysis causes an increase in the mixed micelle sizes, since this is the only way the required shielding of the nonpolar acyl chains from H,O can be maintained. Thus, the fact that vesicle diameters are proportional to lipid concentration implies that the phase limits (the 1ipid:detergent molar ratio that is attained before formation of vesicles occurs) are larger for solutions containing elevated lipid concentrations.
AND
THOMPSON
A similar dependence of vesicle size on PC concentration has been noted in one previous study in which n-alkyl glucosides were used (lo), but the opposite trend has been observed in another (1) in which cholate was employed. In this latter study, though, a change from 6.5 to 13 mg/ml caused only a small decrease in radii from 30.0 + 2.5 nm (from ultracentrifugal diffusion measurements) to 26.6 + 1.5 nm (from QELS measurements). This small concentration range and the variability in measuring techniques may not allow a general trend to be discerned. Alternatively, it is noted that vesicle diameters (Fig. 3) are at least 20 nm larger than those observed by Zumbuehl and Weder (2) for similar PC concentrations and PC:cholate molar ratios. The larger sizes for similar lipid concentrations must be due to an effective slower dialysis rate (1) (direct comparison of rates is not possible since the systems employed are not identical). Faster kinetics not only results in smallersized vesicles but may not produce much of a difference in sizes for various PC concentrations. In summary, the mixed disc model of Lasic (4) has been extended to account for larger-diameter vesicles existing in solutions containing elevated PC concentrations. Vesicle diameters depend on the phase limit and this, combined with the fact that the PC:cholate molar ratio that exists prior to dialysis is proportional to lipid concentration, allows us to conclude that the lipid concentration dictates the phase limit for the mixed micelles. Further investigation is warranted to determine whether these findings can be generalized to other 1ipid:detergent systems. As discussed elsewhere (23), it is not clear why the distribution of free and micellar detergent is independent of lipid concentration. One explanation offered (23) is that at elevated POPC concentrations the rate of free deoxycholate association with micelles is lowered while the rate of detergent dissociation remains the same, resulting in a reduction of the amount of micellar detergent in equilibrium with free species. These observations highlight the need for an accurate structure of the mixed micelle. Unfortunately, there is some disagreement regarding the shape of lecithin-bile salt micelles, with proposals of mixed-discs (4,6), rod-shapes (24-27), and capped-rods (23) being suggested. Analysis of the dependence of vesicle sizes on the concentrations of different types of lipid and detergent may allow the more accurate model to be chosen. For instance, various lipids and detergents may be chosen according to certain criteria (possibly differences in chain length, charge, or headgroup size) and any changes in behaviour noted may be correlated with structural differences. ACKNOWLEDGMENTS Support Research
for this Council
work from of Canada
the Natural is gratefully
Sciences and Engineering acknowledged. We also
FORMATION
OF
PHOSPHATIDYLCHOLINE
thank Dr. Ross Hallet and Jackie Marsh at the University of Guelph, Ontario for the use and training on the QELS instrumentation.
VESICLES
12. Lasic, 13. King,
BY
287
DIALYSIS
D. D. (1987) J. Z’heor. R. G., and Marchbanks,
Biol. 124, 35-41. R. M.
(1982)
Biochim. Biophys.
Acta 691,183-187. 14. Muller,
REFERENCES
15. Huang,
1. Milsmann, M., Schwendener, try 512,147-155.
R., and Weder,
H.
(1978) Biochemk-
D. D. (1982)
5. Small,
D. M.,
B&him.
Penkett,
N. A., Benedek, D. D. (1988)
S. A., and Chapman,
D. (1969)
Biochim.
G. B., and Carey,
M.
J. A., Nozaki,
20. Ackers,
G. K. (1967)
21. Brunner,
J, Skrabal,
C. (1980) Biochemis-
Y., and Tanford,
C. (1983)
And. Bio-
Biochem. J. 256, l-11.
chemistry 25,2597-2605. R. A., Asanger, M., and Weder, H. G. (1981) Biothem. Biophys. Res. Commun. 100,1055-1062. 11. Fischer, T. H., and Lasic, D. D. (1984) Mol. Cryst. Liq. Cryst. Lett. 102.141-153.
J. Biol. Chem. 242,3237-3238. P., and Hauser, H. (1976) Biochim. Biophys.
Acta 455,322-331. 22.
8. Schurtenberger, P., Mazer, N. A., and Kanzig, R. (1984) in Surfactants in Solution (Mittal, K. L., and Lindman, K., Eds.), Vol. 2, pp. 841-855, Plenum, New York. 9. Almog, S., Kushnir, T., Nir, S., and Lichtenberg, D. (1986) Bio10. Schwendener,
19. Reynolds,
B&him.
them. 130,471-474.
try19,601-615. 7. Lasic,
17. Koppel, D. E. (1972) J. Chem. Phys. 57,4814-4820. 18. Kramer, R. M., Hasselbach, J., and Simenza, G. (1981) Biophys. Acta 643,233-242.
Biophys. Acta 692,501-502.
Biophys. Acta 176,178-179. 6. Mazer,
Biophys. Acta 298,
1015-1019.
2. Zumbuehl, O., and Weder, H. (1981) Biochim. Biophys. Acta 640, 252-262. 3. Schmidt, K. H., (Ed.) (1986) Liposomes as Drug Carriers, Georg Thieme Verlag, New York. 4. Lasic,
16. Batzri,
K. (1981) Biochemistry 20,404-414. C. (1969) Biochemistry 8,344-352. S., and Korn, E. D. (1973) Biochim.
Schurtenberger,
P., and Hauser,
H. (1984)
B&him.
Biophys. Acta
778,470-480. 23. Nichols, J. W., and Ozarowski, J. (1990) Biochemistry 29, 46004606. 24. Shankland, W. (1970) Chem. Phys. Lipids 4, 109-130. 25. Hjelm, R. P., Jr., and Alkan, H. (1988) J. Appl. Crystallog. 21, 858-863. 26. Hjelm, R. P., Jr., Thiyagarajan, P., and Alkan, H. (1990) Mol. Cryst. Liq. Cryst. 18OA, 155-164. 27. Walter, A., Vinson, P. K., and Talmon, Y. (1990) Biophys. J. 67, 476a.
