Molecular Motion and Conformation of Cholesteryl Esters in Reconstituted High Density Lipoprotein by Deuterium Magnetic Resonance*

Reconstituted high density lipoprotein has been pre- pared by sonication and preparative ultracentrifugation of mixtures containing the apoprotein of high den- sity lipoprotein, egg phosphatidylcholine, cholesteryl oleate, and acyl chain deuterated cholesteryl palmitate in aqueous buffer. The resulting structures have a size and chemical composition very similar to native high density lipoprotein. Deuterium NMR spectra and longitudinal relaxation times were obtained at approxi- mately 25 "C. The variation of the 'H NMR line width with chain position is consistent with an average con- formation such that the ester acyl chain is extended. In addition, 'H NMR line widths and longitudinal relaxa- tion times indicate that the ester acyl chains possess significant mobility. Reconstituted HDL-The amount of deuterated cholesteryl palmitate present in reconstituted HDL was determined by recording a 'H NMH spectrum for reconstituted HDL containing cholesteryl[16.16,16-'H,~] palmitate in which a capillary tube containing 30 pl of CHCI.t/C'HCI:t (9:l. v/v) was coaxially inserted into the sample tube. Sufficient time to ensure complete relaxation (25 7'') was left between successive radiofrequency pulses during data acquisition. Comparison of the integrated intensity of the C'HCL and cholesteryl ester signals, sep-arated by 260 Hz. enabled quantitation of the ester.

p To whom correspondence should be addressed.
The abbreviation used is: HDL, high density lipoprotein.
Deuterium NMR is an excellent technique for studying molecular order and dynamic structure as it monitors a nonperturbing probe. selective deuteration, combined with the low natural abundance of deuterium, gives unambiguous assignments and, since the quadrupolar interaction is so strong, dipolar interactions with neighboring nuclei may usually be ignored. Recently, it has been shown that molecular order in lipoproteins may be monitored by 'H NMR (Wassall et al., 1982). In the present study, we report ' H NMR spectra and spin-lattice relaxation times for cholesteryl palmitate, selectively deuterated along the acyl chain, incorporated into reconstituted HDL particles. Line width analysis shows that the ester acyl chains undergo considerable anisotropic molecular motion. In addition, proposed models of cholesteryl ester conformation in HDL are discussed in light of the present results.

EXPERIMENTAL PROCEDURES
Lecithin was extracted from hen egg yolks (Singleton et at., 1965) and purified on a silica gel column (Richter et al., 1977).
Deuterium-depleted water and cholesteryl oleate were purchased from Sigma, while palmitic acid and cholesterol were obtained from Fisher. Cholesterol was recrystallized from benzene before being used.
Fresh human blood (53 days old) was obtained from the Canadian Red Cross, Vancouver Branch. HDLZi was isolated in the density range of 1.125-1.210 g/ml (Havel et at., 1955), and its apoprotein and total lipids were isolated according to the methods of Scanu et at. (1969).
Preparation of Reconstituted High Density Lipoprotein-Egg phosphatidylcholine, cholesteryl oleate, and deuterated cholesteryl palmitate, in the ratio of 3:l:l (w/w), were dissolved in CHC1:r. The solvent was removed by evaporation under a stream of nitrogen and subsequent overnight pumping under high vacuum. The dried lipid was incubated a t 48 "C for 10 min in NaHCOdNa&O:j buffer (pH was adjusted to 1.063 with NaCI/KBr (Havel et al., 1955) in deute-8.6) (Scanu, 1966), prepared in deuterium-depleted water. The density rium-depleted water. The lipid dispersion was shaken on a Vortex mixer until it appeared homogeneous.
Dry apoprotein was added to the lipid dispersion (50:50; protein/ total lipid) and allowed to incubate at 48 "C for approximately 10 min or until all of the solid protein particles appeared to have dissolved. The resultant cloudy mixture was sonicated a t approximately 40-48 "C for five 1-min periods, using a Biosonic 111 probe-type sonicator. After sonication, the recombined particles were isolated in the density range of 1.063-1.210 g/ml (Hirz and Scanu, 1970) and then concentrated to approximately 8 mg of protein/ml, prior to the NMR experiments, using Millipore immersible CX-30 membrane units.

Deuterium NMR of High Density Lipoprotein 2001
Viscosity measurements of reconstituted HDL were performed using an Ostwald viscometer. The amount of protein present in reconstituted HDL was quantified according to the method of Lowry et al. (1951), using egg albumin as a standard. Phospholipid was quantified by phosphorus determination (Ames, 1966), while cholesterol esters were determined as described by Hudel and Morris (1973) or by NMR (see below).
Electron Microscopy-Lipoprotein particles were negatively stained with 2 8 ammonium molybdate, pH 8.0, placed on 200-mesh Formvar carbon-coated grids, and allowed to air dry. T h e specimens were examined in a Philips 300 electron microscope operating a t 80 kV.
Nuc'oar Magnetic Resonance-Deuterium NMH experiments were carried out a t 38.8 MHx, using a Nalorac 5.9 Tesla superconducting magnet and home-built spectrometer. Collection and Fourier transformation of the free induction decays were performed on a Nicolet BNC-I2 computer. Longitudinal relaxation times were obtained by the inversion-recovery method (Vold et al., 1968). The sample temperature was approximately 25 "C.
'"P NMR spectra were acquired at 40.5 MHz on a Varian XL-100-15 spectrometer, operating in the Fourier transform mode and interfaced to a NIC 1080 computer. Magnet field stabilization was by means of an external "F field frequency lock. Free induction decays were recorded in the presence of 'H noise-decoupling having a 1 kHz band width. In order to minimize sample heating due to the radiofrequency field (10 watts) used for decoupling either the decoupler was gated on only during acquisition of the free induction decay or. when decoupling was applied continuously, sample cooling by a gas flow system was employed. The temperature of the experiments was approximately 25 "C. Chemical shifts were measured with respect to an external H,,PO, (85%) sample.
A phase-alternating pulse sequence was used in all experiments in order to minimize base-line aberrations.
Quantification of Deuterated Cholesteryl Palmitate in Reconstituted HDL-The amount of deuterated cholesteryl palmitate present in reconstituted HDL was determined by recording a 'H NMH spectrum for reconstituted HDL containing cholesteryl[16.16,16-'H,~] palmitate in which a capillary tube containing 30 pl of CHCI.t/C'HCI:t (9:l. v/v) was coaxially inserted into the sample tube. Sufficient time to ensure complete relaxation ( 2 5 7'') was left between successive radiofrequency pulses during data acquisition. Comparison of the integrated intensity of the C'HCL and cholesteryl ester signals, separated by 260 Hz. enabled quantitation of the ester.

RESULTS
We have prepared model HDL particles by sonication and subsequent ultracentrifugation of mixtures containing HDLs apoprotein, cholesteryl oleate, deuterated cholesteryl palmitate, and egg phosphatidylcholine in aqueous buffer, similar to that described by Hirz and Scanu (1970). Electron microscopy of these structures, using the negative-staining technique (Fig. 11, shows that they have an overall spherical structure and are quite homogeneous in size. The diameter of these structures and their chemical composition is shown in Table  I. '"P NMR spectra of native HDL:, and of reconstituted HDL, prepared with egg phosphatidylcholine/cholesteryl oleate/ cholesteryl palmitate or total HDL:, lipids, respectively, are shown in Fig. 2, a-c. The '"P NMR spectrum of native HDL (Fig. 2a) is essentially identical to those published previously (Glonek et al., 1974;Henderson et al., 1975). The more intense upfield resonance at -0.6 (kO.1) ppm, which has a width at half-height Aulr2 = 6 Hz, is due to phosphatidylcholine, while the less intense downfield resonance at +O. 1 (kO.1) ppm is due to sphingomyelin. The '"P NMR spectrum of reconstituted HDL, containing egg phosphatidylcholine/cholesteryl oleate/ cholesteryl palmitate, only consists of a single resonance a t -0.6 (kO.1) ppm which is due to phosphatidylcholine and which has AuIr2 = 5 (+1) Hz (Fig. 26). T o confirm the suitability of the reconstitution method, we have also prepared reconstituted HDL using total HDL:( lipids which yields a virtually indistinguishable spectrum (Fig. 2c) to that of native HDL:t.  * Numbers in parentheses represent the standard deviation. Fig. 3, a-e, depicts 'H NMR spectra of reconstituted HDL, containing cholesteryl palmitate selectively deuterated along the acyl chain. A single Lorentzian line gives a satisfactory fit to each of these spectra which, in addition to the absorption due to deuterated cholesteryl palmitate, sometimes contains the narrow, downfield-shifted line from residual deuterium in water. Control experiments revealed that the line intensity and line width of the 'H NMR spectra of deuterated cholesteryl palmitate in reconstituted HDL are unchanged over the time span of the NMR experiments (up to approximately 48 h), indicating that no significant change in structure of the sample occurs over the length of the experiments. Table I1 lists 'H NMR line widths of selectively deuterated cholesteryl palmitate in reconstituted HDL. The value of Av,,' is approximately constant for deuterons on C2-C6 of the acyl chain, then decreases a t C11 and C12 before reaching a minimum a t C16.
Longitudinal relaxation times, TI, measured for selectively deuterated cholesteryl palmitate in reconstituted HDL, are also presented in Table 11. Due to the low amplitude of the broad methylene signals for the ester, necessitating long periods of data collection, the TI values are subject to a relatively  [5,5,6,6-'H4]; d, [11,11,12,12-'H4]; e, [16,16,. The resonance shifted by approximately 150 Hz downfield from the main peak in a-d is due to residual deuterium in water. Spectral parameters: sweep width = 5000 Hz ( a and c ) , 10,000 Hz ( b ) , 2,500 Hz (d), 1,000 Hz (e); pulse length = 18 ps (90' flip angle); number of acquisitions = 300,000 ( a ) , 100,000 ( b and c ) , 50,000 (d), 5,700 (e); size of data sets = 1024 ( a and c ) , 1024 zero-filled to 2048 ( b ) , 2048 (d and e); delay between subsequent pulses = 0.1 s ( b and c ) , 0.2 s (d), 1 s ( e ) . For the spectrum in a, a 180°-~-90" pulse sequence with T = 0.11 s was used in order to minimize the interference of the absorption of residual deuterium in water with the peak from cholesteryl [2,2-'H']

DISCUSSION
The results from Fig. 1 and Table I show that the reconstituted HDL particles, prepared in the present study, have physical properties very similar to those of native HDL. The diameter of reconstituted HDL, as shown by electron microscopy (7.8 & 1.2 nm), is within the range of 6.5-10 nm reported for native HDL . In addition, 31P NMR line widths of native HDLB and of HDL reconstituted using either total lipids or egg phosphatidylcholine/cholesteryl oleate/ cholesteryl palmitate are identical within experimental error. This indicates a similar particle size in the three systems (Burnell et al., 1980) and also suggests that there is no appreciable difference in phospholipid head group conformation (McLaughlin et aL, 1975). Moreover, the chemical composition (Table I) of reconstituted HDL closely approximates that of native HDL. Thus, on the basis of size, 31P NMR and chemical composition, we believe that these particles are good models of native HDL.
The 2H NMR linewidths for cholesteryl palmitate, selectively deuterated along the acyl chain, in reconstituted HD,L (Table 11) are, at all positions except the terminal methyl group, much larger than those for the equivalent positions of deuterated cholesteryl palmitate in phosphatidylcholine vesicles Cushley et al., 1980), which suggest that ester motion is considerably more restricted in the lipoprotein than in vesicles.
Insight into the dynamic behavior of deuterated cholesteryl palmitate in reconstituted HDL can be obtained by calculating theoretical 'H NMR linewidths for an ester acyl chain undergoing given motions. The simplest case is that of a static all-trans chain embedded in a lipoprotein particle, such that particle tumbling is the only motion responsible for line-narrowing. The ' H NMR linewidth Avllz is then given by (Abragam, 1961, p.

424)
~A v l p a 2 M~T~ (1) where M2 is the rigid lattice second moment and T , is the effective correlation time for particle tumbling. The second moment M2 is related to the static quadrupolar splitting AVQ palmitate. In this case, the delay between subsequent pulse sequences was 0.1 s. Line broadening = 20 Hz ( a and c ) Burnett and Muller, 1971) by M -~A v Q ' 4 .

q r 3 3kT
(3) where 9 is the viscosity, r is the particle radius, k is the Boltzmann constant, and T is the absolute temperature. For reconstituted HDL, with 9 = 1.13 cP, r = 40 A and T = 298 OK, we obtain ' T~ = 7.5 X lo-' s; hence, from Equations 1 and 2, a line width AvlI2 = 3000 Hz is estimated for a rigid C2H2 segment. For a deuterium attached to a terminal methyl group, it is also necessary to include the effect of fast rotation about the C-C bond joining it with the adjacent CH2 group (Valic et al., 1979), and then we calculate Av1/2 = 330 Hz.
Clearly, the experimental line widths (Table 11) for deuterated cholesteryl palmitate in reconstituted HDL are much smaller than these predicted values. Therefore, we conclude that cholesteryl ester acyl chain has significant motion inside the lipoprotein particles.
Our observations are in agreement with those obtained from '"C NMR data of native HDL by Hamilton and Cordes (1978), who described the cholesteryl esters in terms of being more "liquid-like" than "solid-like.'' In the case of native HDL, which contains triglycerides, it has been suggested that the dynamic behavior of the cholesteryl esters is similar to that in a triglyceride/cholesteryl ester mixture (Hamilton and Cordes, 1978). However, on the basis of our data, it is clear that the presence of triglycerides is not necessary for the ester chains to undergo significant motion. In addition, it should be noted that the physical state of cholesteryl oleate/cholesteryl palmitate mixtures is solid at approximately 25 "C. Therefore, we believe that the fact that cholesteryl palmitate exhibits considerable chain mobility when in reconstituted HDL particles is more probably due to the interaction of the cholesteryl ester chains in the core with the phospholipid and/or protein components of the outer layer.2 Such an interaction has been suggested to explain the absence of the ester phase transition in native HDL (Tall et al., 1977).
The two most popular models for the conformation of cholesteryl esters in HDL particles (Fig. 4) are the "horseshoe" (Structure I)  and the extended form (Structure ZI) (Laggner and Muller, 1978).
Treating the motions undergone by the ester acyl chain in the lipoprotein as symmetric about the labeled axis (Fig. 4), the quadrupolar splitting, AVQ in Equation 2 is modified by a factor (3cos2p -1}/2, where / 3 is the angle between the symmetry axis for the molecular motion and the C-'H bond, while the angular brackets represent a time average over all conformations of the molecule (Oldfield et al., 1971). Hence, if cholesteryl palmitate adopted a horseshoe conformation in reconstituted HDL, a dramatic variation of the 'H NMR line width versus chain position in the highly curved region of the acyl chain would be expected as a consequence of the changes in the average value of p. In phospholipid liposomes containing deuterated cholesteryl ester, a local minimum of qua&polar splitting is indeed observed at C4-C5 of the ester chain (Gorrissen et al,, 1981). A variation of this form is clearly not the case for the line width of deuterated cholesteryl palmitate in reconstituted HDL. On the other hand, if the ester chain ' A referee has suggested that the small size of the HDL particle might prevent the cholesteryl ester from packing as it does in the bulk phase and lead to the chain mobility. were extended, the variation of the line width versus chain position might be expected to resemble the profiie of quadrupolar splittings for the sn-1 chain in phospholipid bilayers, which is known to adopt an extended conformation (Seelig and Seelig, 1977). This is, in fact, observed for deuterated cholesteryl palmitate in reconstituted HDL (Table 11). In the present case, we observe a region of relatively constant line widths, which encompasses deuterons on C2-C6, while further down the chain the linewidths decrease until reaching a minimum at C16.
Hence, our observations are consistent with an extended conformation (Structure ZI) for cholesteryl palmitate in the core of reconstituted HDL, such as proposed previously by Laggner and Muller (1978). These authors also proposed that the cholesteryl ester molecules were radially distributed within the HDL particles. Assuming such an ordered distribution of cholesteryl palmitate in the lipoprotein, the 'H NMR line width of Equation 1 can be expressed by (Stockton et al., 1976) where S = (3cos2/? -1)/2 is the C-*H order parameter, which measures the angular excursions of the C-*H bond under consideration with respect to the radial direction. The absolute values of the order parameter, IS(, calculated from the observed *H NMR line widths, are shown in Fig. 5. For comparative purposes, we have also included the values for the sn-1 chain of 1-palmitoyl-2-oleoyl phosphatidylcholine (Seelig and Seelig, 1977). While the shape of both order parameter profiles is quite similar, the values of (SI for cholesteryl palmitate in reconstituted HDL are higher than those of the equivalent positions of 1-palmitoyl-2-oleoyl phosphatidylcholine. This indicates that the angular excursions undergone by the ester chains in lipoprotein are more restricted than those of the acyl chains in phospholipid bilayers. The highest order parameter (S = 0.38) for cholesteryl palmitate is observed on C2 of the acyl chain, which probably reflects the motional restriction imposed by the rigid cholesteryl moiety. However, the fact that all of the order parameters are less than the value of 0.5 expected for a chain undergoing only rapid rotation about its long axis suggests that significant angular fluctuations do occur. These fluctuations become of larger amplitude approaching the terminal methyl group, as evidenced by the progressive decrease of IS1 along the chain. The longitudinal relaxation times, T I , for the acyl chain of deuterated cholesteryl palmitate in reconstituted HDL are relatively constant at -15 ms for all the C2Hz segments studied, while T I for the terminal CLH3 group is much larger a t approximately 150 ms. This implies that the fast molecular motions responsible for longitudinal relaxation have approximately the same rate for the C2H2 segments but are significantly slower than those of the C'H3 group. Such a behavior has been encountered previously for deuterated cholesteryl palmitate in phospholipid vesicles Cushley et ai., 1980) and for deuterated phospholipids (Brown et al., 1979).
A quantitative interpretation of the longitudinal relaxation times of cholesteryl palmitate in reconstituted HDL is, unfortunately, difficult due to the small size of these particles. For deuterated cholesteryl palmitate in phospholipid vesicles, particle tumbling ( T~ = s) is too slow to appreciably influence T I , and fast segmental motions (wo'T,'<<~) appear to dominate longitudinal relaxation . This may not be the case for the much smaller lipoprotein particles (7, = 7.5 X lo-" s). In fact, a preliminary experiment a t 61.4 MHz performed in this laboratory has indicated that TI is somewhat larger at the higher frequency, suggesting that not all the motions responsible for longitudinal relaxation in this system are in the short correlation time limit (wO%,<<l). A detailed study of T I as a function of temperature and frequency would be required to clarify this point.