Solubilization of Triolein and Cholesteryl Oleate in Egg Phosphatidylcholine Vesicles*

The incorporation of cholesteryl oleate and triolein into phospholipid vesicles was studied in cosonicated mixtures of 94 weight % egg phosphatidylcholine and 6 weight % neutral lipid (0-6% triolein and 6-0% cholesteryl oleate). 13C NMR spectroscopy was used to quantitate both neutral lipids in vesicles containing 90% isotopically substituted [c~rbonyZ-’~C]cholesteryl oleate and [~arbonyl-’~C]triolein. Vesicles were also prepared with radiolabeled cholesteryl oleate and triolein and the composition of ultracentrifugal subfrac- tions determined by chemical and radioisotopic methods. For a given starting composition, the incorpora- tion of neutral lipids into vesicles was similar for vesicles prepared and analyzed by the two methodologies. The maximum solubility in vesicles prepared at 55 “C with a single neutral lipid was 3.1 weight % triolein (2.8 mol %) and 2.3 weight % cholesteryl oleate (2.8 mol %). In sonication mixtures with both triolein and cholesteryl oleate, the incorporation of each lipid into vesicles was proportional to the starting concentration; the total incorporation of neutral lipid was 54.0% (weight or mole per cent). The solubility limits were intermediate between the theoretical cases of complete additivity and complete competition. The [ 13C]carbonyl chemical shifts showed that the carbonyl groups of the vesicle-solubilized neutral lipids were close to the vesicle

The incorporation of cholesteryl oleate and triolein into phospholipid vesicles was studied in cosonicated mixtures of 94 weight % egg phosphatidylcholine and 6 weight % neutral lipid (0-6% triolein and 6-0% cholesteryl oleate). 13C NMR spectroscopy was used to quantitate both neutral lipids in vesicles containing 90% isotopically substituted [c~rbonyZ-'~C]cholesteryl oleate and [~arbonyl-'~C]triolein. Vesicles were also prepared with radiolabeled cholesteryl oleate and triolein and the composition of ultracentrifugal subfractions determined by chemical and radioisotopic methods. For a given starting composition, the incorporation of neutral lipids into vesicles was similar for vesicles prepared and analyzed by the two methodologies. The maximum solubility in vesicles prepared at 55 "C with a single neutral lipid was 3.1 weight % triolein (2.8 mol %) and 2.3 weight % cholesteryl oleate (2.8 mol %). In sonication mixtures with both triolein and cholesteryl oleate, the incorporation of each lipid into vesicles was proportional to the starting concentration; the total incorporation of neutral lipid was 54.0% (weight or mole per cent). The solubility limits were intermediate between the theoretical cases of complete additivity and complete competition. The [ 13C]carbonyl chemical shifts showed that the carbonyl groups of the vesicle-solubilized neutral lipids were close to the vesicle surface and that excess triolein and cholesteryi oleate partitioned into an oil phase containing both triolein and cholesteryl oleate.
Cholesterol esters and triglycerides are weakly polar neutral lipids which generally form separate phases in the presence of lipids containing charged groups or having a stronger polarity. Thus, for example, in plasma lipoproteins the cholesterol esters and triglycerides partition into a hydrophobic core stabilized by a surface of phospholipid, cholesterol, and protein (1,2). However, the question of whether these weakly polar lipids are totally insoluble or are partially soluble in phospholipids is important. We have recently shown, using phase equilibria techniques (3,4) and 13C NMR spectroscopy (5, 6), that cholesterol esters (6) and triglycerides ( 5 ) are slightly soluble (up to -5 weight %) in the phospholipid surface of emulsions (3, 4) and in bilayers (5, 6).
NMR studies provided strong evidence that the three carbonyl groups of triolein (TO) and the single carbonyl group of cholesteryl oleate (CO) were present near the aqueous interface of the phospholipid bilayer and thus were favorably situated for enzymatic hydrolysis or binding to transfer pro-* This work was supported by United States Public Health Service Grants HL-26335 and HL-07291. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Vesicle Preparation for NMR Studies-Vesicles were prepared in essentially the same manner as described previously for vesicles with triolein only (5) and with cholesteryl oieate only (6). Compositions are given as weight % (CO + TO + PC = 100%) unless noted otherwise. Lipid mixtures (100 mg of total lipid in 1.8 ml of aqueous 0.56% KC1) were sonicated under Nz using a Branson sonifier with a microtip as before (5,6) except that all samples were maintained at 55 f 2 "C during the sonication period (45-60 min). Samples were prepared at 55 "C to assure that both triolein and cholesteryl oleate were in the liquid state (see "Results and Discussion"). Low speed centrifugation for 30 min at room temperature was used to remove titanium fragments. Following initial NMR analysis, samples were fractionated by ultracentrifugation for 4-6 h at 140,000 X g and 15 "C. The thin band of turbid material which floated to the top of the tube was removed, together with "20% of the underlying clear zone. About 1.2 ml of the clear zone was used for NMR analysis, leaving a pellet and -0.1 ml of solution on the bottom of the centrifuge tube. Samples were analyzed by thin layer chromatography which showed that 4 % unesterified fatty acid and lysolecithin were generated during aonication.
NMR Methods-Fourier transform NMR spectra were obtained at 50.3 MHz with a Bruker WP200 spectrometer system equipped with an Aspect computer system and a 10-mm 13C probe. Chemical shifts, linewidths, peak areas, and sample temperatures were measured as Phospholipid Vesicles with Triolein and Chotesteryl Oleate before (5, 6). All NMR measurements were made at 37 f 1 "C unless noted otherwise. Spin lattice relaxation times ( T I ) were measured by the fast inversion recovery method (7) and nuclear Overhauser enhancement by the method of Opella et al. (8).
Sample Preparation for Chemical and Radioisotopic Assay-Appropriate proportions of lipids (-94% PC and 6% neutral lipid (radiolabeled)) were mixed in chloroform/methanol and dried under vacuum. 12.0 ml of 0.56% KC1 solution were added to hydrate the lipids (15-30 min at 50-60 "C). Samples (100 mg of total lipid in 12.0 ml) for 30 min at power level of 100 watts at 55 * 2 "C. Following were sonicated using a macrotip with continuous sonication under N2 sonication, the sample was cooled to 24 "C and 0.3 ml were removed for determination of the exact chemical composition of the starting mixtures. The sample was then loaded into one centrifuge tube (Beckman SW 41 rotor, cellulose nitrate tube), the volume was adjusted to 11.5-12.0 ml by addition of 0.56% KCl, and the sample was spun at 30,000 rpm for -20 min (2.2 X lo6 g X min) at 24 "C. A small turbid layer generated during this centrifugation step was quantitatively resuspended along with all the supernatant. The pellet fraction and an aliquot of the supernatant were analyzed for chemical composition.
To subfractionate the vesicle preparations, 2.0-ml aliquots of the supernatant were layered under 10.0 ml of distilled H20. The tubes were marked, beginning 2.5 cm from the bottom, in 1.5-cm intervals to designate five fractions. Samples were spun in the SW 41 rotor for 24 h at 25,000 rpm (1.1 X 10" g.min). Samples were then aspirated from the centrifuge tubes in five fractions, beginning with the top layer. The lipids from each fraction were extracted by the Folch procedure (see Ref. 9) and the mass of PC determined by chemical assay and the masses of triolein and cholesteryl oleate by double label liquid scintillation counting (3, 4). The total recovery of lipids was >90% of that placed in the centrifuge tube. Table I presents the chemical compositions, determined by chemical analysis and radioisotope counting, of different ultracentrifuge fractions from one sonication mixture, with a starting composition of 94.6% PC, 4.4% TO, and 0.95% CO ("unfractionated;" row 1). The short centrifugation step, which removed sonicator tip fragments and some multilamellar liposomes, yielded a supernatant and pellet of identical composition (rows 2 and 3). Thus, there was no differential loss of lipids into the pellet, which was discarded. The compositions of the five ultracentrifuge fractions obtained after a 24-h spin of the supernatant are given in the bottom five rows. The 0.56% KC1 buffer has a background density ( p -  -

RESULTS AND DISCUSSION
Actual weight percent of total lipid in each ultracentrifugal fraction (number of milligrams/fraction divided by total milligrams in all fractions).
Sonicated mixture before centrifugation. e Supernatant mixture before ultracentrifugation. 'Pellet after 20 min of centrifugation minus titanium which was removed during the Folch extraction from the interface between the aqueous and organic layers.
1.5-ml fraction after ultracentrifugation for 24 h.   Table I are  It also contained resonances at -171.8 and -171.5 ppm from triolein in an oil phase (seen most clearly in the spectrum with the highest triolein content (5% TO, 1% CO)). Following ultracentrifugation and removal of an excess oil phase, spectra of the purified vesicles were obtained (Fig. 2,  top row, Fractionated). These spectra showed only three car-The remaining portion of the spectrum was identical with previously published spectra (5, 6). bonyl resonances, in addition to the PC carbonyl resonances, corresponding to vesicle-solubilized triolein (TOs1.3 and TOsa) and vesicle-solubilized cholesteryl oleate (COS). The peak area of the TOs peak was unaffected by the fractionation procedure; in contrast, peaks attributed to oil phase triolein and cholesteryl oleate were not detectable in the vesicle spectra. Comparison of spectra of samples before and after fractionation showed that the COS peak was obscured in spectra of unfractionated samples by the sn-1,3 carbonyl peak of oil phase triolein. In the vesicle spectra the COS, peak intensity increased with increasing per cent of cholesteryl oleate in the starting mixture (right to left, top row).
The chemical shifts and linewidth values of all carbonyl resonances were the same as, or very similar to, the values for corresponding resonances in CO/PC (6) and TO/PC vesicles    TI value of COS, which was slightly longer (1.6 f 0.1 s compared with 1.3 s). Thus, the molecular motions of the surface-located neutral lipids, as assessed by NMR linewidths, TI values, and nuclear Overhauser enhancement values, were not grossly altered in the mixed vesicles. The percentage of cholesteryl oleate and triolein solubilized in the vesicle fractions was calculated from the carbonyl peak area ratios as before (5,6). The vesicle compositions determined by NMR (given as weight per cent; total lipid = 100%) are given above the top spectra in Fig. 2. In Fig. 1, inset, the initial compositions of the NMR samples (0) and the compositions of the purified vesicles (0) are plotted. Values of the starting composition and the vesicle composition of corresponding samples are connected by a clotted line. The NMR data (Fig. I, inset) and the chemical data ( Fig. 1) defined very similar phase b~undaries.~ This result is important because the two methodologies employed each have their own advantages and limitations. The use of radiolabeled neutral lipids allowed accurate quantitation of these lipids in more dilute sonication mixtures (-10-fold more dilute than NMR sam-The only case in which compositional data differed for similar starting compositions was the data for 5% CO, 1% TO. The sample analyzed by chemical and radioisotopic methods (Fig. 1, point f ) had a higher cholesteryl oleate content than the comparable NMR sample (inset). Previous studies of vesicles with triolein (5) or with cholesteryl oleate (6) compared the compositions determined by NMR and chemical analysis of the same samples and showed a good agreement between the two methods. Therefore, the disparity for the 5% CO, 1% TO sample may reflect differences in sample preparation and/or purification. However, the present results show that, in general, both procedures yielded vesicles of similar composition (for a given starting composition). ples) and in more highly subfractionated mixtures. The NMR analysis allowed direct measurement of the vesicle composition; the single step centrifugation yielded pure vesicles without dilution of the sample, and the purity of the vesicle fraction was confirmed directly by the absence of resonances from oil phase lipids.
The I3C NMR results provide insight into the structure of the excess oil phase lipids as well as the vesicle lipids, The presence at 38 "C of an oil phase cholesteryl oleate peak (-171.35 ppm) in unfractionated PC/TO/CO samples indicates that triolein and cholesteryl oleate are cosolubilized in the excess oil phase; sufficient triolein must be present to depress the liquid +liquid crystalline transition of cholesteryl oleate from 47 "C (11) to 538 "C or to abolish this transition altogether. (Previously, with sonicated mixtures of PC and cholesteryl oleate, the oil phase cholesteryl oleate peak was not observed at 38 "c, but was observed after heating to 51 "C (6).) Furthermore, all (>go%) of the cholesteryl oleate in the excess oil phase was cosolubilized with triolein, since increasing the temperature of unfractionated mixtures from 38 to 55 "C in the NMR probe did not result in an increase in the area of the cholesteryl oleate peak at 171.35 ppm. The NMR results are consistent with the oil phase compositions determined by chemical analysis and the phase diagram plots (Fig.  1) which show core contents of L 12.5% TO (CO/TO ratios of 17.0): The solubility of triolein in PC vesicles and cholesteryl oleate in PC vesicles ("3.1 and 2.3 weight % for triolein and cholesteryl oleate, respectively) was slightly higher than the limiting solubilities previously measured at 35 "C (5,6). The greater solubilities may have been a result of the higher temperature (55 "C) used during sonication in the present study. Cholesteryl oleate had a larger increase in solubility than triolein, most likely because cholesteryl oleate has a solid + liquid phase transition at 51 "C (ll), whereas triolein is liquid above 4 "C (11). We previously noted a 30% increase in the solubility of cholesteryl oleate in PC vesicles when samples were prepared above 51 "C (6): The present study showed that cholesteryl oleate and triolein had an equal maximum solubility in terms of mole per cent (-2.8) when samples of triolein and PC or cholesteryl oleate and PC were prepared at 55 "C. In Fig. 3 Fig. 3. Cosolubilization of triolein and cholesteryl oleate in the same vesicle would be expected on the basis of entropy considerations; it would be unlikely that two neutral lipids of somewhat similar physical properties would be highly segregated into separate vesicles. In addition, the neutral lipids co-mixed in the emulsion particles, as described above.
Based on the phase diagrams of Figs. 1 and 3, TO/CO/PC mixtures with starting compositions that fall in the shaded region will be a two-phase (vesicle plus H20), i.e. all the cholesteryl oleate and triolein will be soluble in the PC vesicle and no excess oil phase will form. To test this prediction, In addition, samples with cholesteryl oleate and PC which were prepared at 55 "C and examined by NMR at 55 "C without cooling had the same composition as samples prepared at 55 "C, cooled to 20 "C (for centrifugation), and examined by NMR at -35 "C (J. A. Hamilton, unpublished results). Thus, the temperature of sample preparation but not the subsequent thermal history affected the maximum solubility of cholesteryl oleate in PC.

phases \4
:ol e% eo n\ +-mole% TO samples with starting compositions of 1.0% CO, 1.0% TO, 98% PC and 1.75% CO, 1.75% TO, 96.5% PC (the latter in duplicate) were prepared as above. The initial compositions are shown in Fig. 1 as solid triangles. Although the samples were slightly turbid following sonication, the compositions of the purified vesicles determined by NMR (1.1% TO, 1.0% CO, 97.9% PC and 1.8% TO, 1.7% CO, 96.5% PC) were not significantly different from the starting compositions, as predicted.
The 13C chemical shifts of the carbonyl resonances of triolein and cholesteryl oleate in spectra of purified vesicles indicate that the carbonyl groups are exposed to the aqueous medium at the vesicle surface (5,6). Thus, both triolein and cholesterol oleate are favorably oriented for enzymatic hydrolysis and are favorably located for recognition by carrier proteins. Furthermore, no preferential solubility in PC vesicles was found for either neutral lipid with the initial relative concentrations of neutral lipids varying from equality by f5/ 1.
Studies are in progress to assess whether the presence of cholesterol and proteins in a phospholipid surface affect the solubility of cholesterol esters and triglycerides in phospholipids.