- Email: [email protected]
Influence of dietary fats on the fluidity of the lipid domains of rabbit plasma lipoproteins
229
Atherosclerosis, 35 (1980) 229-241 @ Elsevier/North-Holland Scientific
Publishers,
Ltd.
INFLUENCE OF DIETARY FATS ON THE FLUIDITY DOMAINS OF RABBIT PLASMA LIPOPROTEINS
ELLIOTT
BERLIN
and CALVERT
YOUNG,
OF THE LIPID
Jr.
Lipid Nutrition Laboratory, Nutrition Institute, Human Nutrition Center, United States Department of Agriculture, Beltsville, MD 20705 (U.S.A.) (Received 7 June, 1979) (Revised, received 1 November, (Accepted 5 November, 1979)
1979)
Summary
The effects of dietary stearic and other saturated fatty acids on the fluidity of the plasma lipoproteins were assessed with fluorescence polarization techniques. Rabbits were maintained on diets containing either cocoa butter, milkfat, coconut oil, or corn oil as the only source of fat. Microviscosities, r), of the lipid regions of plasma very low density lipoproteins (VLDL), low density lipoproteins (LDL), and high density lipoproteins (HDL) were determined by measuring the anisotropy of fluorescence from the probe 1,6-diphenyl-1,3,5-hexatriene. The microviscosity values followed the sequence 7nnL > ,jLnL > rjvr,nL when the lipoproteins were isolated from the plasma of rabbits fed cocoa butter, milkfat, or corn oil. HDL and LDL consist of an invariant phase in the temperature range 0-50°C regardless of diet. VLDL from rabbits fed milkfat, corn oil, or cocoa butter displayed monophasic behavior in the same range, while VLDL from rabbits fed coconut oil showed a phase transition at 31.9 f 3.7”C. Lipoproteins were less fluid in fasted than in non-fasted rabbits and VLDL and LDL from fasted milkfat-fed rabbits showed phase transitions. Despite the fatty acid compositions of the dietary fats, VLDL and LDL were more fluid from rabbits fed cocoa butter than from rabbits fed corn oil; apparently metabolism influences microviscosity. Key words:
Cholesterol - Fluidity - Lipoprotein
-Saturated
fatty acids
Introduction
Consumption of saturated fats has been implicated in the development of coronary heart disease, including both atherosclerotic [l] and thrombotic [Z] processes. Renaud and coworkers demonstrated that stearic acid was the most
230
thrombogenic fatty acid in studies with rats [ 31 and rabbits [ 41. They reported the effects of various dietary fats on coagulation factors and on the chemical composition of blood platelets. To better understand these phenomena, we have investigated the effects of such fats on the physical chemical properties of entities involved in the development of coronary heart disease, e.g. the platelets and the plasma lipoproteins. Diets differing in the ratio of polyunsaturated to saturated fats undoubtedly produce different percentages of unsaturated fatty acyl chains in the various lipids of the plasma lipoproteins [ 51. Since fatty acyl chain unsaturation is an important determinant of lipid fluidity [6], it follows that there might be an association between unsaturation in dietary fats and fluidity of the plasma lipoproteins. Similarly alterations in cholesterol/phospholipid ratios induced by variations in dietary fatty acids and/or cholesterol might influence lipid phase fluidity. There have been several studies, with calorimetry or fluorescence spectroscopy, to assess dietary influences on lipoprotein fluidity in humans [ 71 and laboratory animals [8-lo]. Tall et al. [8] used differential scanning calorimetry (DSC) in studies of low density lipoproteins (LDL) and high density lipoproteins (HDL) from atherosclerotic miniature swine on diets enriched in saturated fat and cholesterol. Hillman et al. [9] reported changes in calorimetrically observed phase transitions in LDL from rabbits maintained on diets enriched with safflower oil or hydrogenated coconut oil. Castellino et al. [lo], using fluorescence probe analysis, found a 7-fold increase in the viscosity of very low density lipoproteins (VLDL) of hypercholesterolemic rabbits. Neither of these studies with rabbits included data on more than one lipoprotein fraction. We now report on the hydrocarbon fluidity and fusion activation energies of the lipid phases in VLDL, LDL, and HDL from rabbits maintained on 5% fat diets that varied both in total saturated fatty acid content and in stearic acid content to reflect reported differences in thrombogenicity [4]. Hydrocarbon fluidity was determined in these lipoproteins from measurements of the anisotropy of fluorescence from the probe 1,6-diphenyl-1,3,5-hexatriene (DPH) incorporated into the respective lipid phases. Materials and Methods * Animals and experimental diets Forty-eight male New Zealand white rabbits weighing 1.8-2.0 kg were purchased (Camm Research Institute) and fed commercial chow (Purina Rabbit Chow) ad libitum until they weighed 3.6-4.0 kg. They were divided into 4 groups and fed ad libitum diets containing different natural fats varying in saturation level and stearic acid content. The semipurified diet for rabbits of Gaman et al. [ll] minus corn oil was used as a basal diet. The experimental diets consisted of the basal diet ingredients plus 5% of either cocoa butter, milkfat, coconut oil, or corn oil. The diets were prepared commercially (Tek* Mention of a trademark or a proprietary product does not constitute a @Wantee or WU’mntY of the product by the U.S. Department of Agriculture, and does not imply its approval to the exclusion of other products that may also be suitable.
231
lad Test Diets) in pelleted form according to these specifications. Fatty acid methyl ester analysis by gas-liquid chromatography of lipid extracts of the diets confirmed that the fatty acid pattern for each diet was identical to that of the corresponding fat. Lipoprotein isolation Blood samples were taken after the rabbits were fed the diets for 6, 10, 20, and 25 weeks. Rabbits were allowed free access to food at all times except before bleeding 3 when they were fasted overnight. Blood was drawn via cardiac puncture into plastic syringes containing Na,EDTA in saline as an anticoagulant. Not all the rabbits survived this procedure, hence the experimental groups became progressively smaller. The actual number sampled at each blood drawing appears in Figure legends. After removal of platelets by centrifugation [12] for other studies, the lipoproteins were isolated from pooled platelet poor plasma by ultracentrifugal flotation [ 13,141. Following the removal of chylomicrons and other large particles (S, > 400), plasma or plasma diluted with 0.195 molal NaCl (p = 1.006 g/ml) was centrifuged for 18 h at 16°C at 40,000 rpm in a Beckman ultracentrifuge (40.3 rotor) to concentrate VLDL at the tops of the centrifuge tubes. After removal of the first ml containing the VLDL and a second reference ml, the 4 ml infranatant was mixed with 2 ml of a NaCl/NaBr solution (p = 1.182 g/ml) and centrifuged for an additional 18 h to concentrate LDL at the tops of the tubes. After removal of the first ml containing the LDL and a second reference ml, the 4-ml infranatant was mixed with 2 ml of a NaCl/NaBr solution (p = 1.478 g/ml) and centrifuged for 24 h to concentrate the HDL fraction. The lipoprotein solutions were dialyzed against 0.005 M Tris-HCl, pH 7.20, buffer and stored at 4°C prior to chemical analysis. Chemical analysis Protein content was determined Cholesterol [16] and phosphorus extracts of the lipoprotein solutions.
by the procedure of Lowry [ 171 were determined with
et al. [15]. isopropanol
Fluidity measurements Lipid phase fluidity was determined from measurements of depolarization of fluorescence from the probe molecule 1,6diphenyl-1,3,5-hexatriene (DPH) following the method pioneered by Shinitzky and coworkers [l&19]. Probe incorporation was accomplished by diluting 2 X 10m3 M DPH in tetrahydrofuran 500- to lOOO-fold into the aqueous lipoprotein solutions and incubating with agitation at 35°C for 2 h. Intensity of fluorescence polarization was measured with an Aminco-Bowman spectrophotofluorometer equipped with Glan-Thompson polarizers, thermostated cell holder, photon counter, and digital readout and printer. Temperature was controlled with a constant temperature circulator (Lauda), and was measured with a thermistor probe (Yellow Springs Instruments) immersed in the sample solution in the fluorometer cell. DPH was excited at 366 nm, and the fluorescence at 450 nm was detected through a Wratten 2A cutoff filter for wavelengths shorter than 415 nm. Light-scattering errors were minimized by
232
diluting the lipoprotein solutions until the polarization or anisotropy remained constant. The solutions were sufficiently diluted so that the scattered light intensity from unlabelled lipoprotein was less than 3% of the intensity of the emitted light from DPH-labelled lipoprotein. Fluorescence data were taken as a function of temperature from 50 to 0°C at intervals of 3-6 degrees. Microviscosities were calculated from anisotropy data by use of the approximation of Shinitzky and Barenholz [ 61: q = 2.4 r/(0.362
- r)
(1)
based on the Perrin equation:
where r and r0 are measured and limiting anisotropies, C(r) is a parameter that relates to the molecular shape of the fluorophore, T is the absolute temperature, r is the excited state lifetime of the fluorophore, and q is the microviscosity experienced by the fluorophore. The above approximation is based on the experimental values [ 18,191 of r0 = 0.362 for the limiting anisotropy of DPH excited at 366 nm, T,, = 11.4 nsec for the limiting value of the excited state lifetime of DPH at a quantum yield of unity, and C(r) = 8.6 X 10’ poise/ degree set from calibrations with DPH in liquid paraffin. The measured anisotropies were obtained from the intensities of emission polarized parallel and perpendicular to the polarized excitation by use of standard formulae including the instrumental correction factor of Azumi and McGlynn [20] as described by Chen and Bowman [ 211 for polarization studies with the Aminco-Bowman spectrophotofluorometer. Results Lipoprotein microviscosities are presented graphically in Figs. l-4 as logarithmic functions of the reciprocal of the absolute temperature. The straight lines were obtained by a least-squares fit for the exponential function: ,, = AeAn/nr
(3)
where AE is the energy of activation for viscous flow. Correlation coefficients, r, for the indicated lines were at least 0.95 and usually were >0.989. It is clear that lipid fluidity varied among the lipoprotein classes and was influenced by diet. The kind of dietary fat influenced the relative fluidities of the lipoprotein classes, the phase behavior of the lipoproteins, and the magnitude of microviscosity. The first two of these effects are most apparent between coconut oil (Fig. 4) and the other fats. At each sampling the plasma lipoproteins of rabbits fed cocoa butter, milkfat, or corn oil (Figs. l-3) decreased in fluidity according to the sequence qVLDL < qLnL < nnnL. VLDL was the most fluid lipoprotein fraction from coconut oil-fed rabbits but LDL from these rabbits was not always more fluid than HDL. At bleedings 1 and 4 (Figs. 4a and d) LDL was more fluid, at bleeding 2 (Fig. 4b) HDL was more fluid, and at bleeding 3 (Fig. 4c) they were about equal in fluidity. The phase behavior of the lipoproteins from the coconut oil-fed rabbits was
233
2b
o-
LOL
l ---
VLDL
_LOL I
AFTER74DAYS
AFTER
I
I
71 DAYS
2d
--+-h-k
--+re-+
AFTER I42 DAYS FASTINO ELDOD SAMPLE
AFTER I79 DAYS
103/T
IO 3/T
(OK-’ 1
AFTER 170 DAYS
fK-’
1
Fig. 1. Logarithmic plots of microviscosity vs. the reciprocal of the absolute temperature for plasma lipoproteins from rabbits fed cocoa butter diet. (I: 8 rabbits on diet 46 days; b: 10 rabbits cm diet 14 days: c: 3 rabbits on diet 144 days: and d: 5 rabbits on diet 179 days. Fig. 2. Logarithmic plots of microviscosity vs. the reciprocal of the absolute temperature for plasma lipoproteins from rabbits fed a diet with milkfat. (I: 9 rabbits on diet 43 days; b: 7 rabbits after 71 days on diet; c: 7 rabbits after 142 days: and d: 5 rabbits after 170 days.
0. ci s?
A------
HDL LDL
A-----liDL _LOL
c--
VLOL
.--VLDL
AFTER 4ZDAYS
I
AFTER 70 DAYS
2
AFTER
I
I
71 DAYS
Id
F
*lo.0 10.0 5.0
54 I’
IA ,d
d
4
4.z
I
J’
5.
A
J’
a?,#
2.0
/
/Lu=
I.0
AFTER I41 DAYS FASTIN BUXD ?%
-YizzF K-1)
,7’ 1.0 I/
/’ 1
3.2
I
3.4
1
3.6
AFTER l42 MYS FASTING WBDD SAMPLE 10%.
I
I
I
3.2
3.4
3.6
AFTER I63 DAYS ‘K.1 )
Fig. 3. Logarithmic plots of microviscosity vs. the reciprocal of the absolute temperature for plasma lipoproteins from rabbits fed a diet containing corn oil. (I: 5 rabbits after 42 days: b: 5 rabbits after 70 days: c: 4 rabbits after 141 days; and d: 5 rabbits after 178 days. Fig. 4. Logarfthmic plots of mfcrovfscosity vs. the reciprocal of the absolute temperature for Planma limoproteins from rabbits fed coconut oil. (I: 5 rabbits after 43 days; b: 5 rabbits after 71 days; c: 4 rabbits after 142 days: and d: 4 rabbits after 183 days on diet.
234
also unique; at each bleeding the VLDL exhibited a phase transition near 32°C. The only other incidences of phase transitions were with the VLDL and LDL isolated from a fasting blood sample from the milk-fat-fed rabbits (Fig. 2~). All other lipoprotein samples were monophasic in the temperature range 5O-O”C. The effects of diet on the magnitude of q are somewhat obscured in the data in Figs. l-4. Microviscosities of the individual lipoprotein classes often varied between bleedings of rabbits on the same diet. Furthermore, the slopes of the linear relationships shown in these Figures vary and the differences between the lipoprotein microviscosities are not the same at all temperatures. Microviscosities at an arbitrarily selected temperature, 37°C appear in Table 1. Ranges for lipoproteins isolated from non-fasting blood appear in Table la and individual data points for the fasting blood in Table lb. The data overlap but show trends. The lipoprotein fluidities do not necessarily reflect the level of saturation in the ingested fat. The VLDL and LDL fractions were more fluid from cocoa butter-fed than from the corn oil-fed rabbits. The most rigid LDL fraction was from the coconut oil-fed rabbits. Fasting the rabbits for 16 h before drawing blood influenced lipoprotein fluidity and the phase behavior of some of the lipoproteins. Fractions were more rigid from fasted than from non-fasted rabbits; the effects were less pronounced in the more dense lipoproteins. Phase changes were observed in VLDL and LDL from fasting blood samples from the milkfat-fed rabbits. The transition temperature for the VLDL from the fasted coconut oil-fed rabbits was depressed to 27°C (Fig. 4~). Some of the differences in fluidity between samples of the same lipoprotein fraction from animals on the same diet must be associated with differences in the status of the animals with respect to blood lipids. Despite consumption of identical diets, the rabbits often displayed wide TABLE la MICROVISCOSITY
DATA FOR THE LIPOPROTEIN
FRACTIONS
AT 37%
The values indicated are ranges for r). in poise. for non-fasting blood samples. Lipoprotein Diet
VLDL
LDL
HDL
Cocoa butter MiIhfat Coconut oil Corn oil
0.73-1.16 1.05-1.51 0.74-1.58 0.74-l .37
1.72-l .92 1.80-2.15 2.32-2.57 2.08-2.37
2.07-2.97 2.84-2.98 2.10-3.10 2.68-2.80
TABLE lb MICROVISCOSITY FOR EACH DIET
DATA
(37%)
FOR LIPOPROTEINS
FROM SINGLE FASTING
Lipoprotein Diet
VLDL
LDL
HDL
Cocoa butter Mikfat Coconut oil Corn oil
1.47 1.70 2.30 2.00
2.62 2.28 3.20 2.41
3.25 2.96 3.13 3.25
BLOOD SAMPLE
235
.
8
30-
.
.
l
A
0
8
m*A
.
HDL
.
25-
OV
0
q LODL
I
0
I
I
1
I
I
IO
20
3.0
4.0
50
MOLES
CHOLESTEROL
I
al /MOLES
I
I
1
7.0
SD
9.0
I
100
PHOSPHOLIPIO
Fig. 5. Effect of cholesterol on lipoprotein microviscosity. Graph of q at 37OC vs. molar cholesterol/ phospholipid ratio. 0: VLDL from cocoa butter-fed rabbits; A: VLDL from milkfat-fed rabbits; 0: VLDL from coconut oil-fed rabbits; v: VLDL from corn oil-fed rabbits. Half-filled symbols (@ A 0 0) for corresponding LDL fractions and filled symbols (0 A n v) for corresponding HDL fractions.
variations in the plasma concentrations of cholesterol, triglycerides, and phospholipids. Data pertaining to the effects of cholesterol on fluidity are shown graphically in Fig. 5 where the microviscosity parameters are plotted against the molar cholesterol to phospholipid ratios for the various lipoprotein fractions. When data for all rabbits were combined, least squares analysis yielded the straight lines in Fig. 5 for the lipoprotein fractions following the equations: (r) in poise) q = 0.753 + 0.170 (moles cholesterol/moles phospholipid) with the correlation coefficient r = 0.5716 for VLDL; 77= 2.228 + 0.003 (moles cholesterol/moles phospholipid) with r = 0.0224 for LDL; phospholipid) with r = 0.0583 T = 2.759 + 0.021 ( moles cholesterol/moles for HDL. These data and equations correspond to microviscosity values at 37°C. Similar sets of equations were obtained by least-squares analysis of microviscosity data at 25” and 4°C. Thus only VLDL data showed, over a wide temperature range, that cholesterol significantly lowered lipid fluidity, regardless of diet. Treatment of the data for each dietary group individually, however, indicated more significant effects of cholesterol (Table 2). The effects of cholesterol on lipoprotein microviscosity varied among the diets suggesting that the nature of cholesterol modification of fluidity reflects upon the fatty acid composition of the lipoproteins. Aside from the dietary effects, the mode of action of cholesterol differed among the lipoproteins. Microviscosity was always increased in VLDL, usually slightly decreased in LDL, and usually increased in HDL, by cholesterol. The energies of activation for viscous flow or the fusion activation energies, AE, were calculated from the slopes in Figs. l-4 according to equation (3):
236 TABLE 2 PARAMETERS
CALCULATED
ACCORDING
TO THE EQUATION:
q = no + m (moles cholesterol/
moles phospholipid), r) at 37% Diet
Lipoprotein
90 (poise)
m
r
Cocoa butter
VLDL LDL HDL
0.043 0.440 0.456
0.333 0.456 0.959
0.6146 0.6837 0.9974
MiIkfat
VLDL LDL HDL
0.948 2.293 2.840
0.100 -0.060 0.036
0.5160 0.7418 0.6185
coconut oil
VLDL LDL HDL
0.229 3.071 1.731
0.351 -0.074 0.418
0.8262 0.5386 0.6461
Corn oil
VLDL LDL HDL
0.942 2.365 3.075
0.192 -0.061 4.095
0.6541 0.8198 0.1391
TABLE 3a AE. FUSION ACTIVATION BLOOD SAMPLES
ENERGY
(-kcaI/mole)
FOR
LIPOPROTEINS
FROM
NONFASTING
The data inTables 3a and 3b are presented as positive numbers for clarity. but aII are negative. Lipoprotein Diet
VLDL
LDL
HDL
Cocoa butter Corn oil MiIkfat Coconut oil
6.54-7.14 6.66-7.39 6.28-7.63 high temp. phase 7.13-9.31 low temp. phase 5.05-6.03
6.82-l .08 6.42-1.60 1.40-8.38 1.32-8.41
6.98-8.85 6.31-9.87 8.25-9.06 1.18-8.77
TABLE 3b AE. FUSION SAMPLES
ACTIVATION
ENERGY
(-kcaI/mole)
FOR LIPOPROTEINS
FROM FASTING
Lipoprotein Diet
VLDL
LDL
HDL
Cocoa butter Corn oil Milkfat
6.16 6.94 high temp. 9.18 low temp. 6.38 high temp. 8.12 low temp. 6.42
6.85 6.56 high temp. phase 8.40 low temp. phase 4.61 8.82
8.12 8.04 8.52
Coconut oil
phase phase phase phase
8.89
BLOOD
237
TABLE 4 PARAMETERS
CALCULATED
moles phospholipid)
ACCORDING
TO THE EQUATION:
AE = AEo + m (moles cholesterol/
(DE is in kcal/mole)
Diet
Lipoprotein
AEo
m
r
Cocoa butter
VLDL LDL HDL
-8.47 -6.81 -6.21
+0.572 -0.035 -0.908
0.7644 0.1887 0.6641
Corn oil
VLDL LDL HDL
-1.24 -7.45 -7.46
+0.117 +0.226 --0.210
0.6827 0.6448 0.2744
Milkfat
VLDL LDL HDL
-5.18 -8.56 -8.72
4.539 +0.155 +0.008
0.6317 0.8079 0.0245
Coconut oil
VLDL. high temp. phase VLDL, low temp. phase LDL HDL
-6.64 -5.48 -8.90 -7.46
+0.227 -0.067 +0.118 -0.210
0.4584 0.1752 0.5479 0.2744
q = AeAEIRT for Newtonian flow. Values for AE are listed in Table 3 for the lipoprotein fractions from the rabbits fed different fats. Apparently diet and the status of the animal with regard to fasting have little effect on AE. The energy released tended to increase in the sequence VLDL, LDL, to HDL. Treatment of our AE data as a function of the molar cholesterol/phospholipid ratios by linear regression yielded the following equations: phospholipid) with r = 0.3882 AE =--6.58-0.236 (moles cholesterol/moles for VLDL; phospholipid) with r = 0.0490 AE = -7.50-0.014 (moles cholesterol/moles for LDL; and
-041 -1.0
I -06
I -0.6
1 -0.4
CHANGE
I -0.2
, 0
I to.2
I to.4
I to.6
I t06
I *I .o
(m) INDUCED IN AE BY CHOLESTEROL
Fig. 6. Relationship between the changes induced by cholesterol in the microviscosity activation energy for microviscosity (AE) for l VLDL, A LDL. and 0 HDL.
(q) and fusion
238
AE = -7.73-0.269 (moles cholesterol/moles phospholipid) with r = 0.3122 for HDL. The only cholesterol effect of any significance is on the degree of order of the VLDL. Linear regression data for each lipoprotein phase for each diet appear in Table 4. Apparently the effects of cholesterol on hydrocarbon chain order in these systems are complex and vary from fraction to fraction. Only the VLDL fraction was consistently affected by cholesterol, as shown in Fig. 6, where the slopes of the linear regression equations for microviscosity (Table 2) are plotted against the slopes of the AE equations (Table 4). Results similar to those of Fig. 6 were also obtained with microviscosity data at 25” and 4°C. Discussion Soutar [5] has pointed out that plasma lipids are transported in a less fluid form in the plasma of an animal with severe atherosclerosis compared to normal human plasma [ 81. Shepherd [ 221 reported that the microscopic fluidity of HDL increased after humans ate polyunsaturated fats. Soutar [ 51, however, cautioned against concluding that changes in lipoprotein fluidity are directly induced by changes in the saturation of the fats in the diet. This caution is appropriate in light of our findings that lipoprotein fluidities did not always parallel the relative viscosities of the ingested fats. The fats of this study, which were selected to provide different dietary levels of stearic acid, short chain fatty acids and polyunsaturated fatty acids, differ accordingly in their viscosities. Cocoa butter and milkfat are equally highly saturated fats but cocoa butter contains more stearic acid; coconut oil is even more highly saturated but contains primarily short chain fatty acids and little stearic acid. In contrast corn oil is high in polyunsaturated fatty acids and low in stearic acid. Despite its high level of polyunsaturated fats, corn oil did not increase lipoprotein fluidity. VLDL and LDL were most fluid when isolated from rabbits fed cocoa butter even though it is highly saturated and highest in stearic acid. Renaud and Gautheron [4] reported that cocoa butter was the most thrombogenic and among the least atherogenic of several fats fed to laboratory rabbits. They observed equivalent atherogenicities with cocoa butter and corn oil. Our data offer a possible explanation for those results. The thrombogenicity of cocoa butter might result from enhanced platelet-collagen interactions in arterial wall subendothelium. For the subsequent development of atherosclerosis, cholesterol from VLDL or LDL particles must then be deposited at the site of the initial platelet-collagen interaction. Although more study is needed, Soutar [ 51 indicated that the atherogenicity of LDL particles increased as their core lipids became less fluid. The fluidity data of Morrisett et al. [23] for human lipoproteins suggest that the role of catabolization or exchange differs significantly between lipoproteins of high and low fluidity. Possibly the particles of the more fluid lipoproteins in the blood of rabbits fed cocoa butter would have less tendency than particles of less fluid lipoproteins to deposit cholesterol in arterial walls. That would explain the low atherogenicity for cocoa butter. Our data clearly established that lipoprotein fluidity is affected by dietary fatty acids; however, further work is necessary to clarify the connections between diet, fluidity, and atherogenicity.
239
Diet influences the transition temperature between the liquid crystalline phase and other phases in lipoprotein lipids. We observed phase transitions only in VLDL from rabbits fed coconut oil or milkfat and not in the other fractions, regardless of diet. Castellino et al. [lo] reported no evidence of lipid phase transitions in fluorescence polarization studies of VLDL from rabbits fed a commercial chow with or without added cholesterol. In contrast, Hillman et al. [9] reported phase transitions in DSC studies of LDL from rabbits fed diets containing safflower or hydrogenated coconut oils. Jonas and Jung [24] reported no phase transitions in fluorescence polarization studies of bovine HDL. With DSC, Tall et al. [8] detected phase transitions in LDL and HDL of hypercholesterolemic miniature swine. Deckelbaum et al. [ 251 studied lipoproteins of normolipemic human subjects by DSC and reported reversible phase transitions with LDL but not VLDL near body temperature. Mantulin et al. [7] observed a thermal transition in DSC studies with human LDL but not HDL or VLDL. They also reported that the transition temperature was lower when the subjects consumed unsaturated fats. The published data, as well as our data, clearly show that dietary lipids affect the thermotropic behavior of the lipoproteins. Some of the published differences might be related to species. The results of the fluorescence polarization and DSC work are not necessarily at variance. There is some uncertainty in the location of the fluorescent probe in some of the lipid systems. Dale et al. [26] recently reported on the complexity of microviscosity measurements with DPH in vesicles and reference solvents. Although some uncertainty may exist in the precise definition of bilayer microviscosity, the present data do show relative differences in lipoprotein fluidity. Further experimental work is warranted to correlate fluorescence and calorimetric measurements in the same lipoprotein systems. The results of the present study, including the varied cholesterol effects, the rigidity induced by fasting, and the differences in fluidity among the lipoprotein classes, provide information pertaining to the structure of the lipoprotein molecules. The varied role of cholesterol with respect to lipoprotein fluidity and order, reflects on the complex interrelations of the different lipids in these macromolecules. Cholesterol usually [19,27] increases 7) and decreases AE in artificial and natural biological membranes. Both of these effects relate to the interposition of cholesterol between adjacent phospholipid hydrocarbon chains. Cholesterol decreases the random rotational motion of these hydrocarbon chains within the hydrophobic core of the lipid bilayer, hence increasing q. Cholesterol increases the degree of order by preventing the fusion and dissociation of hydrocarbon chains as random motion changes with temperature, thus lowering AE. These effects are most pronounced with artificial bilayers; most natural membranes usually exhibit AE values of -7 + 1 kcal/mole. Castellino et al. [lo] reported that hypercholesterolemia increased rabbit VLDL microviscosity but barely affected AE. They reported AE = -7.6 * 1.5 kcal/ mole for VLDL from rabbits with normal cholesterol and AE = -7.8 + 1.5 kcal/mole for VLDL from rabbits with high cholesterol. We have shown (Tables 2 and 4, Figs. 5 and 6) that cholesterol significantly and consistently affected fluidity of the lipid domains of only the VLDL fraction. The linearity of the VLDL data in Fig. 6, in contrast with the data of the other fractions, indicates that cholesterol molecules are interposed between fatty acid hydro-
240
carbon chains in the lipid domains of the VLDL but not of LDL. Most structural models [ 28,291 describe the lipoproteins as consisting of an inner hydrophobic core of triglycerides and cholesteryl esters. The models differ in their descriptions of the orientations of protein, phospholipid, and free cholesterol in the periphery of the lipoproteins. Little is known about the structure of VLDL, Deckelbaum et al. [25] noted that the cholesteryl esters are probably interspersed through the rest of the neutral lipid since they did not detect a separate cholesteryl ester melting transition by DSC. More is known about LDL structure [29] although a complete model is not available. LDL contains cholesteryl esters in separate domains. Much of the LDL protein is at or near the surface in association with a phospholipid monolayer. Cholesterol is probably partitioned between the surface and neutral core with a composition-controlled partition coefficient. These different cholesterol orientations bear upon cholesterol modulation of lipoprotein fluidity. Lipid protein interactions in LDL and HDL also affect lipoprotein fluidity [6]. Jonas [30] has associated a higher microviscosity of LDL over that of extracted LDL lipids with a lipid-protein interaction restricting DPH movement. Evidence that apoproteins also restrict the mobility of small molecules in the lipid domains of HDL is available from DPH fluorescence polarization [ 301, pyrene excimer fluorescence [ 231, and ESR probe partitioning [ 231. Our finding that the steroid nucleus of cholesterol is intercalated between hydrocarbon chains only in VLDL is compatible with these structural models. Our data did not enable us to distinguish between free and esterified cholesterol that might affect lipid fluidity in different ways, depending on their locations within the lipoprotein structure. Further experimental work to elucidate the nature of lipoprotein fluidity includes studies of the effects of cholesterol esterification and of the effects that may be inferred from the ratio of cholesterol to neutral lipids such as triglycerides, rather than from the ratio to phospholipids alone. Our results with fasting blood for each group of rabbits suggest that plasma triglycerides have an important role in lipoprotein fluidity. Fasting the animals lowered the fluidity of the VLDL but had less effect upon the higher density lipoproteins. Possibly VLDL fluidity is a function of triglyceride content, which is usually depressed in the fasting animal while HDL and LDL fluidities are controlled by the apoproteins. The observation that nnnL > ‘T)L~L> 1)vLnL and nnnL > 7LnL even though cholesterol is higher in LDL than in HDL demonstrated that apoproteins influence the fluidity of LDL and HDL. It is clear that HDL, despite its lower cholesterol level, was usually more rigid than LDL. Possibly the structural differences characterized by this physicochemical parameter are related to the reported differences in atherosclerotic tendencies associated with the levels of LDL and HDL cholesterol [ 311. References 1 Turpeinen. 0.. Diet and coronary events, J. Amer. Diet. Assn.. 52 (1968) 209. 2 Hornstra. G.. Dietary fats and arterial thrombosis, Haemostasis. 2 (1973/74) 21. 3 Renaud, S.. Thrombotic, atherosclerotic and lipemic effects of dietary fats in the rat, Angiology. 20 (1969) 657. 4 Renaud. S. and Gautheron. P.. Influence of dietary fats on atherosclerosis, coagulation and platelet phospholipids in rabbits, Atherosclerosis, 21 (1975) 116.
241
5 Soutar, A., Does dietary fat influence plasma lipoprotein structure? Nature (Land.), of lipid regions determined 6 Shinitzky, M. and Barenholz, Y., Fluidity parameters polarization. Biochim. Biophys. Acta. 515 (1978) 367. 7 Mantulin, 8 9 10 11 12 13 14
15 16 17 18 19 20 21 22
23
24 25 26 27 28
W.W., Shepherd,
J.,
Gotto.
Jr.,
A.M., and Pownall.
H.J.. Dietary
effects
273 (1978) 11. by fluorescence
on human lipopro-
tein composition and structure, Biophys. J., 21 (1978) 145a. of the lipoproteins of Tall, A.R., Atkinson, D.. Small, D.M. and Mahley, R.W., Characterization atherosclerotic swine, J. Biol. Chem.. 252 (1977) 7288. Hillman, G., Mabrey, S., Pitlick, F. and Engleman, D.,Dietary manipulation of cholesteryl ester melting temperatures in rabbit low density lipoproteins, Biophys. J., 17 (1977) 74a. Castellino, F.J.. Thomas, J.K. and Ploplis, V.A.. Microviscosity of the lipid domains of normal and hypercholesterolemic very low density lipoprotein, Biochem. Biophys. Res. Comm.. 75 (1977) 857. Gaman, E., Fisher, H. and Feigenbaum, A.S., An adequate purified diet for rabbits of all ages. Nutr. Repts. lntl.. 1 (1970) 35. Baenziger, N.L. and Majerus, P.W.. Isolation of human platelets and platelet surface membranes, Methods Enzymol., 31 (1974) 149. Hatch, F.T. and Lees, R.S., Practical methods for plasma lipoprotein analysis, Adv. Lipid Res.. 6 (1968) 1. Lindgren. F.T., Serum lipoproteins - Isolation and analysis with the preparative and analytical ultracentrifuge. In: R.M. Burton and F.C. Guerra (Eds.), Fundamentals of Lipid Chemistry, Bi-Science Publications, Webster Groves, MO, 1974, p. 475. Lowry. O.H., Rosebrough, N.J., Farr. A.L. and Randall, R.J., Protein measurement with the Folin phenol reagent, J. Biol. Chem.. 193 (1951) 265. Rosenthal. H., Pfluke, M. and Buscaglia. S.. A stable iron reagent for determination of cholesterol, J. Lab. Clin. Med., 50 (1957) 318. Bartlett, R., Phosphorus assay in column chromatography, J. Biol. Chem., 234 (1959) 466. Shinitzky, M. and Barenholz, Y., Dynamics of the hydrocarbon layer in liposomes of lecithin and sphingomyelin containing dicetlylphosphate, J. Biol. Chem., 249 (1974) 2652. Shinitzky, M. and lnbar, M., Microviscosity parameters and protein mobility in biological membranes, Biochim. Biophys. Acta. 433 (1976) 133. Azumi. T. and McGlynn. S.P., Polarization of the luminescence of phenanthrene, J. Chem. Physics, 37 (1962) 2413. Chew R.F. and Bowman, R.L., Fluorescence polarization - Measurement with ultraviolet-polarizing filters in a spectrophotofluorometer, Science, 147 (1965) 729. Shepherd, J., Packard, C.J., Patsch, J.R., Gotto, Jr., A.M. and Taunton. O.D.. Effects of dietary polyunsaturated and saturated fat on the properties of high density lipoproteins and the metabolism of apolipoprotein A-l. J. Clin. Invest.. 61 (1978) 1582. Morrisett. J.D., Pownall, H.J., Jackson, R.L.. Segura, R., Gotto, Jr., A.M. and Taunton. O.D., Effects of polyunsaturated and saturated fat diets on the chemical composition and thermotropic properties of human Plasma lipoproteins. In: R.T. Holman and W.H. Kunau (Eds.), Polyunsaturated Fatty Acids (American Oil Chemists Society Monograph No. 4). A.O.C.S. Publishers, Champaign, IL, 1977, P. 139. Jonas, A. and Jung. R.W., Fluidity of the lipid phase of bovine serum high density lipoprotein from fluorescence polarization measurements, Biochem. Biophys. Res. Comm., 66 (1975) 651. Deckelbaum, R.J.. Tall, A.R. and Small, D.M., Interaction of cholesterol ester and triglyceride in human plasma very low density lipoprotein, J. Lipid Res.. 18 (1977) 164. Dale. R.E., Chen, L.A. and Brand, L., Rotational relaxation of the microviscosity probe diphenylhexatriene in paraffin oil and egg lecithin vesicles, J. Biol. Chem., 252 (1977) 7500. Shattil. S.J. and Cooper, R.A., Membrane microviscosity and human platelet function. Biochemistry. 15 (1976) 4832.
Jackson, R.L., Morrisett, J.D. and Gotto. Jr., A.M., Lipoprotein structure and metabolism, Physiol. Rev., 56 (1976) 259. 29 Smith, L.C., Pownall, H.J. and Gotto, Jr., A.M., The plasma lipoproteins - Structure and metabolism, Ann. Rev. Biochem.. 47 (1978) 751. 36 Jonas, A., Microviscosity of lipid domains in human serum lipoproteins, Biochim. Biophys. Acta. 486 (1977) 10. 31 Miller, G.J. and Miller, N.E., Plasma high density lipoprotein concentration and development of ischemic heart disease, Lancet, 1 (1975) 16.
Atherosclerosis, 35 (1980) 229-241 @ Elsevier/North-Holland Scientific
Publishers,
Ltd.
INFLUENCE OF DIETARY FATS ON THE FLUIDITY DOMAINS OF RABBIT PLASMA LIPOPROTEINS
ELLIOTT
BERLIN
and CALVERT
YOUNG,
OF THE LIPID
Jr.
Lipid Nutrition Laboratory, Nutrition Institute, Human Nutrition Center, United States Department of Agriculture, Beltsville, MD 20705 (U.S.A.) (Received 7 June, 1979) (Revised, received 1 November, (Accepted 5 November, 1979)
1979)
Summary
The effects of dietary stearic and other saturated fatty acids on the fluidity of the plasma lipoproteins were assessed with fluorescence polarization techniques. Rabbits were maintained on diets containing either cocoa butter, milkfat, coconut oil, or corn oil as the only source of fat. Microviscosities, r), of the lipid regions of plasma very low density lipoproteins (VLDL), low density lipoproteins (LDL), and high density lipoproteins (HDL) were determined by measuring the anisotropy of fluorescence from the probe 1,6-diphenyl-1,3,5-hexatriene. The microviscosity values followed the sequence 7nnL > ,jLnL > rjvr,nL when the lipoproteins were isolated from the plasma of rabbits fed cocoa butter, milkfat, or corn oil. HDL and LDL consist of an invariant phase in the temperature range 0-50°C regardless of diet. VLDL from rabbits fed milkfat, corn oil, or cocoa butter displayed monophasic behavior in the same range, while VLDL from rabbits fed coconut oil showed a phase transition at 31.9 f 3.7”C. Lipoproteins were less fluid in fasted than in non-fasted rabbits and VLDL and LDL from fasted milkfat-fed rabbits showed phase transitions. Despite the fatty acid compositions of the dietary fats, VLDL and LDL were more fluid from rabbits fed cocoa butter than from rabbits fed corn oil; apparently metabolism influences microviscosity. Key words:
Cholesterol - Fluidity - Lipoprotein
-Saturated
fatty acids
Introduction
Consumption of saturated fats has been implicated in the development of coronary heart disease, including both atherosclerotic [l] and thrombotic [Z] processes. Renaud and coworkers demonstrated that stearic acid was the most
230
thrombogenic fatty acid in studies with rats [ 31 and rabbits [ 41. They reported the effects of various dietary fats on coagulation factors and on the chemical composition of blood platelets. To better understand these phenomena, we have investigated the effects of such fats on the physical chemical properties of entities involved in the development of coronary heart disease, e.g. the platelets and the plasma lipoproteins. Diets differing in the ratio of polyunsaturated to saturated fats undoubtedly produce different percentages of unsaturated fatty acyl chains in the various lipids of the plasma lipoproteins [ 51. Since fatty acyl chain unsaturation is an important determinant of lipid fluidity [6], it follows that there might be an association between unsaturation in dietary fats and fluidity of the plasma lipoproteins. Similarly alterations in cholesterol/phospholipid ratios induced by variations in dietary fatty acids and/or cholesterol might influence lipid phase fluidity. There have been several studies, with calorimetry or fluorescence spectroscopy, to assess dietary influences on lipoprotein fluidity in humans [ 71 and laboratory animals [8-lo]. Tall et al. [8] used differential scanning calorimetry (DSC) in studies of low density lipoproteins (LDL) and high density lipoproteins (HDL) from atherosclerotic miniature swine on diets enriched in saturated fat and cholesterol. Hillman et al. [9] reported changes in calorimetrically observed phase transitions in LDL from rabbits maintained on diets enriched with safflower oil or hydrogenated coconut oil. Castellino et al. [lo], using fluorescence probe analysis, found a 7-fold increase in the viscosity of very low density lipoproteins (VLDL) of hypercholesterolemic rabbits. Neither of these studies with rabbits included data on more than one lipoprotein fraction. We now report on the hydrocarbon fluidity and fusion activation energies of the lipid phases in VLDL, LDL, and HDL from rabbits maintained on 5% fat diets that varied both in total saturated fatty acid content and in stearic acid content to reflect reported differences in thrombogenicity [4]. Hydrocarbon fluidity was determined in these lipoproteins from measurements of the anisotropy of fluorescence from the probe 1,6-diphenyl-1,3,5-hexatriene (DPH) incorporated into the respective lipid phases. Materials and Methods * Animals and experimental diets Forty-eight male New Zealand white rabbits weighing 1.8-2.0 kg were purchased (Camm Research Institute) and fed commercial chow (Purina Rabbit Chow) ad libitum until they weighed 3.6-4.0 kg. They were divided into 4 groups and fed ad libitum diets containing different natural fats varying in saturation level and stearic acid content. The semipurified diet for rabbits of Gaman et al. [ll] minus corn oil was used as a basal diet. The experimental diets consisted of the basal diet ingredients plus 5% of either cocoa butter, milkfat, coconut oil, or corn oil. The diets were prepared commercially (Tek* Mention of a trademark or a proprietary product does not constitute a @Wantee or WU’mntY of the product by the U.S. Department of Agriculture, and does not imply its approval to the exclusion of other products that may also be suitable.
231
lad Test Diets) in pelleted form according to these specifications. Fatty acid methyl ester analysis by gas-liquid chromatography of lipid extracts of the diets confirmed that the fatty acid pattern for each diet was identical to that of the corresponding fat. Lipoprotein isolation Blood samples were taken after the rabbits were fed the diets for 6, 10, 20, and 25 weeks. Rabbits were allowed free access to food at all times except before bleeding 3 when they were fasted overnight. Blood was drawn via cardiac puncture into plastic syringes containing Na,EDTA in saline as an anticoagulant. Not all the rabbits survived this procedure, hence the experimental groups became progressively smaller. The actual number sampled at each blood drawing appears in Figure legends. After removal of platelets by centrifugation [12] for other studies, the lipoproteins were isolated from pooled platelet poor plasma by ultracentrifugal flotation [ 13,141. Following the removal of chylomicrons and other large particles (S, > 400), plasma or plasma diluted with 0.195 molal NaCl (p = 1.006 g/ml) was centrifuged for 18 h at 16°C at 40,000 rpm in a Beckman ultracentrifuge (40.3 rotor) to concentrate VLDL at the tops of the centrifuge tubes. After removal of the first ml containing the VLDL and a second reference ml, the 4 ml infranatant was mixed with 2 ml of a NaCl/NaBr solution (p = 1.182 g/ml) and centrifuged for an additional 18 h to concentrate LDL at the tops of the tubes. After removal of the first ml containing the LDL and a second reference ml, the 4-ml infranatant was mixed with 2 ml of a NaCl/NaBr solution (p = 1.478 g/ml) and centrifuged for 24 h to concentrate the HDL fraction. The lipoprotein solutions were dialyzed against 0.005 M Tris-HCl, pH 7.20, buffer and stored at 4°C prior to chemical analysis. Chemical analysis Protein content was determined Cholesterol [16] and phosphorus extracts of the lipoprotein solutions.
by the procedure of Lowry [ 171 were determined with
et al. [15]. isopropanol
Fluidity measurements Lipid phase fluidity was determined from measurements of depolarization of fluorescence from the probe molecule 1,6diphenyl-1,3,5-hexatriene (DPH) following the method pioneered by Shinitzky and coworkers [l&19]. Probe incorporation was accomplished by diluting 2 X 10m3 M DPH in tetrahydrofuran 500- to lOOO-fold into the aqueous lipoprotein solutions and incubating with agitation at 35°C for 2 h. Intensity of fluorescence polarization was measured with an Aminco-Bowman spectrophotofluorometer equipped with Glan-Thompson polarizers, thermostated cell holder, photon counter, and digital readout and printer. Temperature was controlled with a constant temperature circulator (Lauda), and was measured with a thermistor probe (Yellow Springs Instruments) immersed in the sample solution in the fluorometer cell. DPH was excited at 366 nm, and the fluorescence at 450 nm was detected through a Wratten 2A cutoff filter for wavelengths shorter than 415 nm. Light-scattering errors were minimized by
232
diluting the lipoprotein solutions until the polarization or anisotropy remained constant. The solutions were sufficiently diluted so that the scattered light intensity from unlabelled lipoprotein was less than 3% of the intensity of the emitted light from DPH-labelled lipoprotein. Fluorescence data were taken as a function of temperature from 50 to 0°C at intervals of 3-6 degrees. Microviscosities were calculated from anisotropy data by use of the approximation of Shinitzky and Barenholz [ 61: q = 2.4 r/(0.362
- r)
(1)
based on the Perrin equation:
where r and r0 are measured and limiting anisotropies, C(r) is a parameter that relates to the molecular shape of the fluorophore, T is the absolute temperature, r is the excited state lifetime of the fluorophore, and q is the microviscosity experienced by the fluorophore. The above approximation is based on the experimental values [ 18,191 of r0 = 0.362 for the limiting anisotropy of DPH excited at 366 nm, T,, = 11.4 nsec for the limiting value of the excited state lifetime of DPH at a quantum yield of unity, and C(r) = 8.6 X 10’ poise/ degree set from calibrations with DPH in liquid paraffin. The measured anisotropies were obtained from the intensities of emission polarized parallel and perpendicular to the polarized excitation by use of standard formulae including the instrumental correction factor of Azumi and McGlynn [20] as described by Chen and Bowman [ 211 for polarization studies with the Aminco-Bowman spectrophotofluorometer. Results Lipoprotein microviscosities are presented graphically in Figs. l-4 as logarithmic functions of the reciprocal of the absolute temperature. The straight lines were obtained by a least-squares fit for the exponential function: ,, = AeAn/nr
(3)
where AE is the energy of activation for viscous flow. Correlation coefficients, r, for the indicated lines were at least 0.95 and usually were >0.989. It is clear that lipid fluidity varied among the lipoprotein classes and was influenced by diet. The kind of dietary fat influenced the relative fluidities of the lipoprotein classes, the phase behavior of the lipoproteins, and the magnitude of microviscosity. The first two of these effects are most apparent between coconut oil (Fig. 4) and the other fats. At each sampling the plasma lipoproteins of rabbits fed cocoa butter, milkfat, or corn oil (Figs. l-3) decreased in fluidity according to the sequence qVLDL < qLnL < nnnL. VLDL was the most fluid lipoprotein fraction from coconut oil-fed rabbits but LDL from these rabbits was not always more fluid than HDL. At bleedings 1 and 4 (Figs. 4a and d) LDL was more fluid, at bleeding 2 (Fig. 4b) HDL was more fluid, and at bleeding 3 (Fig. 4c) they were about equal in fluidity. The phase behavior of the lipoproteins from the coconut oil-fed rabbits was
233
2b
o-
LOL
l ---
VLDL
_LOL I
AFTER74DAYS
AFTER
I
I
71 DAYS
2d
--+-h-k
--+re-+
AFTER I42 DAYS FASTINO ELDOD SAMPLE
AFTER I79 DAYS
103/T
IO 3/T
(OK-’ 1
AFTER 170 DAYS
fK-’
1
Fig. 1. Logarithmic plots of microviscosity vs. the reciprocal of the absolute temperature for plasma lipoproteins from rabbits fed cocoa butter diet. (I: 8 rabbits on diet 46 days; b: 10 rabbits cm diet 14 days: c: 3 rabbits on diet 144 days: and d: 5 rabbits on diet 179 days. Fig. 2. Logarithmic plots of microviscosity vs. the reciprocal of the absolute temperature for plasma lipoproteins from rabbits fed a diet with milkfat. (I: 9 rabbits on diet 43 days; b: 7 rabbits after 71 days on diet; c: 7 rabbits after 142 days: and d: 5 rabbits after 170 days.
0. ci s?
A------
HDL LDL
A-----liDL _LOL
c--
VLOL
.--VLDL
AFTER 4ZDAYS
I
AFTER 70 DAYS
2
AFTER
I
I
71 DAYS
Id
F
*lo.0 10.0 5.0
54 I’
IA ,d
d
4
4.z
I
J’
5.
A
J’
a?,#
2.0
/
/Lu=
I.0
AFTER I41 DAYS FASTIN BUXD ?%
-YizzF K-1)
,7’ 1.0 I/
/’ 1
3.2
I
3.4
1
3.6
AFTER l42 MYS FASTING WBDD SAMPLE 10%.
I
I
I
3.2
3.4
3.6
AFTER I63 DAYS ‘K.1 )
Fig. 3. Logarithmic plots of microviscosity vs. the reciprocal of the absolute temperature for plasma lipoproteins from rabbits fed a diet containing corn oil. (I: 5 rabbits after 42 days: b: 5 rabbits after 70 days: c: 4 rabbits after 141 days; and d: 5 rabbits after 178 days. Fig. 4. Logarfthmic plots of mfcrovfscosity vs. the reciprocal of the absolute temperature for Planma limoproteins from rabbits fed coconut oil. (I: 5 rabbits after 43 days; b: 5 rabbits after 71 days; c: 4 rabbits after 142 days: and d: 4 rabbits after 183 days on diet.
234
also unique; at each bleeding the VLDL exhibited a phase transition near 32°C. The only other incidences of phase transitions were with the VLDL and LDL isolated from a fasting blood sample from the milk-fat-fed rabbits (Fig. 2~). All other lipoprotein samples were monophasic in the temperature range 5O-O”C. The effects of diet on the magnitude of q are somewhat obscured in the data in Figs. l-4. Microviscosities of the individual lipoprotein classes often varied between bleedings of rabbits on the same diet. Furthermore, the slopes of the linear relationships shown in these Figures vary and the differences between the lipoprotein microviscosities are not the same at all temperatures. Microviscosities at an arbitrarily selected temperature, 37°C appear in Table 1. Ranges for lipoproteins isolated from non-fasting blood appear in Table la and individual data points for the fasting blood in Table lb. The data overlap but show trends. The lipoprotein fluidities do not necessarily reflect the level of saturation in the ingested fat. The VLDL and LDL fractions were more fluid from cocoa butter-fed than from the corn oil-fed rabbits. The most rigid LDL fraction was from the coconut oil-fed rabbits. Fasting the rabbits for 16 h before drawing blood influenced lipoprotein fluidity and the phase behavior of some of the lipoproteins. Fractions were more rigid from fasted than from non-fasted rabbits; the effects were less pronounced in the more dense lipoproteins. Phase changes were observed in VLDL and LDL from fasting blood samples from the milkfat-fed rabbits. The transition temperature for the VLDL from the fasted coconut oil-fed rabbits was depressed to 27°C (Fig. 4~). Some of the differences in fluidity between samples of the same lipoprotein fraction from animals on the same diet must be associated with differences in the status of the animals with respect to blood lipids. Despite consumption of identical diets, the rabbits often displayed wide TABLE la MICROVISCOSITY
DATA FOR THE LIPOPROTEIN
FRACTIONS
AT 37%
The values indicated are ranges for r). in poise. for non-fasting blood samples. Lipoprotein Diet
VLDL
LDL
HDL
Cocoa butter MiIhfat Coconut oil Corn oil
0.73-1.16 1.05-1.51 0.74-1.58 0.74-l .37
1.72-l .92 1.80-2.15 2.32-2.57 2.08-2.37
2.07-2.97 2.84-2.98 2.10-3.10 2.68-2.80
TABLE lb MICROVISCOSITY FOR EACH DIET
DATA
(37%)
FOR LIPOPROTEINS
FROM SINGLE FASTING
Lipoprotein Diet
VLDL
LDL
HDL
Cocoa butter Mikfat Coconut oil Corn oil
1.47 1.70 2.30 2.00
2.62 2.28 3.20 2.41
3.25 2.96 3.13 3.25
BLOOD SAMPLE
235
.
8
30-
.
.
l
A
0
8
m*A
.
HDL
.
25-
OV
0
q LODL
I
0
I
I
1
I
I
IO
20
3.0
4.0
50
MOLES
CHOLESTEROL
I
al /MOLES
I
I
1
7.0
SD
9.0
I
100
PHOSPHOLIPIO
Fig. 5. Effect of cholesterol on lipoprotein microviscosity. Graph of q at 37OC vs. molar cholesterol/ phospholipid ratio. 0: VLDL from cocoa butter-fed rabbits; A: VLDL from milkfat-fed rabbits; 0: VLDL from coconut oil-fed rabbits; v: VLDL from corn oil-fed rabbits. Half-filled symbols (@ A 0 0) for corresponding LDL fractions and filled symbols (0 A n v) for corresponding HDL fractions.
variations in the plasma concentrations of cholesterol, triglycerides, and phospholipids. Data pertaining to the effects of cholesterol on fluidity are shown graphically in Fig. 5 where the microviscosity parameters are plotted against the molar cholesterol to phospholipid ratios for the various lipoprotein fractions. When data for all rabbits were combined, least squares analysis yielded the straight lines in Fig. 5 for the lipoprotein fractions following the equations: (r) in poise) q = 0.753 + 0.170 (moles cholesterol/moles phospholipid) with the correlation coefficient r = 0.5716 for VLDL; 77= 2.228 + 0.003 (moles cholesterol/moles phospholipid) with r = 0.0224 for LDL; phospholipid) with r = 0.0583 T = 2.759 + 0.021 ( moles cholesterol/moles for HDL. These data and equations correspond to microviscosity values at 37°C. Similar sets of equations were obtained by least-squares analysis of microviscosity data at 25” and 4°C. Thus only VLDL data showed, over a wide temperature range, that cholesterol significantly lowered lipid fluidity, regardless of diet. Treatment of the data for each dietary group individually, however, indicated more significant effects of cholesterol (Table 2). The effects of cholesterol on lipoprotein microviscosity varied among the diets suggesting that the nature of cholesterol modification of fluidity reflects upon the fatty acid composition of the lipoproteins. Aside from the dietary effects, the mode of action of cholesterol differed among the lipoproteins. Microviscosity was always increased in VLDL, usually slightly decreased in LDL, and usually increased in HDL, by cholesterol. The energies of activation for viscous flow or the fusion activation energies, AE, were calculated from the slopes in Figs. l-4 according to equation (3):
236 TABLE 2 PARAMETERS
CALCULATED
ACCORDING
TO THE EQUATION:
q = no + m (moles cholesterol/
moles phospholipid), r) at 37% Diet
Lipoprotein
90 (poise)
m
r
Cocoa butter
VLDL LDL HDL
0.043 0.440 0.456
0.333 0.456 0.959
0.6146 0.6837 0.9974
MiIkfat
VLDL LDL HDL
0.948 2.293 2.840
0.100 -0.060 0.036
0.5160 0.7418 0.6185
coconut oil
VLDL LDL HDL
0.229 3.071 1.731
0.351 -0.074 0.418
0.8262 0.5386 0.6461
Corn oil
VLDL LDL HDL
0.942 2.365 3.075
0.192 -0.061 4.095
0.6541 0.8198 0.1391
TABLE 3a AE. FUSION ACTIVATION BLOOD SAMPLES
ENERGY
(-kcaI/mole)
FOR
LIPOPROTEINS
FROM
NONFASTING
The data inTables 3a and 3b are presented as positive numbers for clarity. but aII are negative. Lipoprotein Diet
VLDL
LDL
HDL
Cocoa butter Corn oil MiIkfat Coconut oil
6.54-7.14 6.66-7.39 6.28-7.63 high temp. phase 7.13-9.31 low temp. phase 5.05-6.03
6.82-l .08 6.42-1.60 1.40-8.38 1.32-8.41
6.98-8.85 6.31-9.87 8.25-9.06 1.18-8.77
TABLE 3b AE. FUSION SAMPLES
ACTIVATION
ENERGY
(-kcaI/mole)
FOR LIPOPROTEINS
FROM FASTING
Lipoprotein Diet
VLDL
LDL
HDL
Cocoa butter Corn oil Milkfat
6.16 6.94 high temp. 9.18 low temp. 6.38 high temp. 8.12 low temp. 6.42
6.85 6.56 high temp. phase 8.40 low temp. phase 4.61 8.82
8.12 8.04 8.52
Coconut oil
phase phase phase phase
8.89
BLOOD
237
TABLE 4 PARAMETERS
CALCULATED
moles phospholipid)
ACCORDING
TO THE EQUATION:
AE = AEo + m (moles cholesterol/
(DE is in kcal/mole)
Diet
Lipoprotein
AEo
m
r
Cocoa butter
VLDL LDL HDL
-8.47 -6.81 -6.21
+0.572 -0.035 -0.908
0.7644 0.1887 0.6641
Corn oil
VLDL LDL HDL
-1.24 -7.45 -7.46
+0.117 +0.226 --0.210
0.6827 0.6448 0.2744
Milkfat
VLDL LDL HDL
-5.18 -8.56 -8.72
4.539 +0.155 +0.008
0.6317 0.8079 0.0245
Coconut oil
VLDL. high temp. phase VLDL, low temp. phase LDL HDL
-6.64 -5.48 -8.90 -7.46
+0.227 -0.067 +0.118 -0.210
0.4584 0.1752 0.5479 0.2744
q = AeAEIRT for Newtonian flow. Values for AE are listed in Table 3 for the lipoprotein fractions from the rabbits fed different fats. Apparently diet and the status of the animal with regard to fasting have little effect on AE. The energy released tended to increase in the sequence VLDL, LDL, to HDL. Treatment of our AE data as a function of the molar cholesterol/phospholipid ratios by linear regression yielded the following equations: phospholipid) with r = 0.3882 AE =--6.58-0.236 (moles cholesterol/moles for VLDL; phospholipid) with r = 0.0490 AE = -7.50-0.014 (moles cholesterol/moles for LDL; and
-041 -1.0
I -06
I -0.6
1 -0.4
CHANGE
I -0.2
, 0
I to.2
I to.4
I to.6
I t06
I *I .o
(m) INDUCED IN AE BY CHOLESTEROL
Fig. 6. Relationship between the changes induced by cholesterol in the microviscosity activation energy for microviscosity (AE) for l VLDL, A LDL. and 0 HDL.
(q) and fusion
238
AE = -7.73-0.269 (moles cholesterol/moles phospholipid) with r = 0.3122 for HDL. The only cholesterol effect of any significance is on the degree of order of the VLDL. Linear regression data for each lipoprotein phase for each diet appear in Table 4. Apparently the effects of cholesterol on hydrocarbon chain order in these systems are complex and vary from fraction to fraction. Only the VLDL fraction was consistently affected by cholesterol, as shown in Fig. 6, where the slopes of the linear regression equations for microviscosity (Table 2) are plotted against the slopes of the AE equations (Table 4). Results similar to those of Fig. 6 were also obtained with microviscosity data at 25” and 4°C. Discussion Soutar [5] has pointed out that plasma lipids are transported in a less fluid form in the plasma of an animal with severe atherosclerosis compared to normal human plasma [ 81. Shepherd [ 221 reported that the microscopic fluidity of HDL increased after humans ate polyunsaturated fats. Soutar [ 51, however, cautioned against concluding that changes in lipoprotein fluidity are directly induced by changes in the saturation of the fats in the diet. This caution is appropriate in light of our findings that lipoprotein fluidities did not always parallel the relative viscosities of the ingested fats. The fats of this study, which were selected to provide different dietary levels of stearic acid, short chain fatty acids and polyunsaturated fatty acids, differ accordingly in their viscosities. Cocoa butter and milkfat are equally highly saturated fats but cocoa butter contains more stearic acid; coconut oil is even more highly saturated but contains primarily short chain fatty acids and little stearic acid. In contrast corn oil is high in polyunsaturated fatty acids and low in stearic acid. Despite its high level of polyunsaturated fats, corn oil did not increase lipoprotein fluidity. VLDL and LDL were most fluid when isolated from rabbits fed cocoa butter even though it is highly saturated and highest in stearic acid. Renaud and Gautheron [4] reported that cocoa butter was the most thrombogenic and among the least atherogenic of several fats fed to laboratory rabbits. They observed equivalent atherogenicities with cocoa butter and corn oil. Our data offer a possible explanation for those results. The thrombogenicity of cocoa butter might result from enhanced platelet-collagen interactions in arterial wall subendothelium. For the subsequent development of atherosclerosis, cholesterol from VLDL or LDL particles must then be deposited at the site of the initial platelet-collagen interaction. Although more study is needed, Soutar [ 51 indicated that the atherogenicity of LDL particles increased as their core lipids became less fluid. The fluidity data of Morrisett et al. [23] for human lipoproteins suggest that the role of catabolization or exchange differs significantly between lipoproteins of high and low fluidity. Possibly the particles of the more fluid lipoproteins in the blood of rabbits fed cocoa butter would have less tendency than particles of less fluid lipoproteins to deposit cholesterol in arterial walls. That would explain the low atherogenicity for cocoa butter. Our data clearly established that lipoprotein fluidity is affected by dietary fatty acids; however, further work is necessary to clarify the connections between diet, fluidity, and atherogenicity.
239
Diet influences the transition temperature between the liquid crystalline phase and other phases in lipoprotein lipids. We observed phase transitions only in VLDL from rabbits fed coconut oil or milkfat and not in the other fractions, regardless of diet. Castellino et al. [lo] reported no evidence of lipid phase transitions in fluorescence polarization studies of VLDL from rabbits fed a commercial chow with or without added cholesterol. In contrast, Hillman et al. [9] reported phase transitions in DSC studies of LDL from rabbits fed diets containing safflower or hydrogenated coconut oils. Jonas and Jung [24] reported no phase transitions in fluorescence polarization studies of bovine HDL. With DSC, Tall et al. [8] detected phase transitions in LDL and HDL of hypercholesterolemic miniature swine. Deckelbaum et al. [ 251 studied lipoproteins of normolipemic human subjects by DSC and reported reversible phase transitions with LDL but not VLDL near body temperature. Mantulin et al. [7] observed a thermal transition in DSC studies with human LDL but not HDL or VLDL. They also reported that the transition temperature was lower when the subjects consumed unsaturated fats. The published data, as well as our data, clearly show that dietary lipids affect the thermotropic behavior of the lipoproteins. Some of the published differences might be related to species. The results of the fluorescence polarization and DSC work are not necessarily at variance. There is some uncertainty in the location of the fluorescent probe in some of the lipid systems. Dale et al. [26] recently reported on the complexity of microviscosity measurements with DPH in vesicles and reference solvents. Although some uncertainty may exist in the precise definition of bilayer microviscosity, the present data do show relative differences in lipoprotein fluidity. Further experimental work is warranted to correlate fluorescence and calorimetric measurements in the same lipoprotein systems. The results of the present study, including the varied cholesterol effects, the rigidity induced by fasting, and the differences in fluidity among the lipoprotein classes, provide information pertaining to the structure of the lipoprotein molecules. The varied role of cholesterol with respect to lipoprotein fluidity and order, reflects on the complex interrelations of the different lipids in these macromolecules. Cholesterol usually [19,27] increases 7) and decreases AE in artificial and natural biological membranes. Both of these effects relate to the interposition of cholesterol between adjacent phospholipid hydrocarbon chains. Cholesterol decreases the random rotational motion of these hydrocarbon chains within the hydrophobic core of the lipid bilayer, hence increasing q. Cholesterol increases the degree of order by preventing the fusion and dissociation of hydrocarbon chains as random motion changes with temperature, thus lowering AE. These effects are most pronounced with artificial bilayers; most natural membranes usually exhibit AE values of -7 + 1 kcal/mole. Castellino et al. [lo] reported that hypercholesterolemia increased rabbit VLDL microviscosity but barely affected AE. They reported AE = -7.6 * 1.5 kcal/ mole for VLDL from rabbits with normal cholesterol and AE = -7.8 + 1.5 kcal/mole for VLDL from rabbits with high cholesterol. We have shown (Tables 2 and 4, Figs. 5 and 6) that cholesterol significantly and consistently affected fluidity of the lipid domains of only the VLDL fraction. The linearity of the VLDL data in Fig. 6, in contrast with the data of the other fractions, indicates that cholesterol molecules are interposed between fatty acid hydro-
240
carbon chains in the lipid domains of the VLDL but not of LDL. Most structural models [ 28,291 describe the lipoproteins as consisting of an inner hydrophobic core of triglycerides and cholesteryl esters. The models differ in their descriptions of the orientations of protein, phospholipid, and free cholesterol in the periphery of the lipoproteins. Little is known about the structure of VLDL, Deckelbaum et al. [25] noted that the cholesteryl esters are probably interspersed through the rest of the neutral lipid since they did not detect a separate cholesteryl ester melting transition by DSC. More is known about LDL structure [29] although a complete model is not available. LDL contains cholesteryl esters in separate domains. Much of the LDL protein is at or near the surface in association with a phospholipid monolayer. Cholesterol is probably partitioned between the surface and neutral core with a composition-controlled partition coefficient. These different cholesterol orientations bear upon cholesterol modulation of lipoprotein fluidity. Lipid protein interactions in LDL and HDL also affect lipoprotein fluidity [6]. Jonas [30] has associated a higher microviscosity of LDL over that of extracted LDL lipids with a lipid-protein interaction restricting DPH movement. Evidence that apoproteins also restrict the mobility of small molecules in the lipid domains of HDL is available from DPH fluorescence polarization [ 301, pyrene excimer fluorescence [ 231, and ESR probe partitioning [ 231. Our finding that the steroid nucleus of cholesterol is intercalated between hydrocarbon chains only in VLDL is compatible with these structural models. Our data did not enable us to distinguish between free and esterified cholesterol that might affect lipid fluidity in different ways, depending on their locations within the lipoprotein structure. Further experimental work to elucidate the nature of lipoprotein fluidity includes studies of the effects of cholesterol esterification and of the effects that may be inferred from the ratio of cholesterol to neutral lipids such as triglycerides, rather than from the ratio to phospholipids alone. Our results with fasting blood for each group of rabbits suggest that plasma triglycerides have an important role in lipoprotein fluidity. Fasting the animals lowered the fluidity of the VLDL but had less effect upon the higher density lipoproteins. Possibly VLDL fluidity is a function of triglyceride content, which is usually depressed in the fasting animal while HDL and LDL fluidities are controlled by the apoproteins. The observation that nnnL > ‘T)L~L> 1)vLnL and nnnL > 7LnL even though cholesterol is higher in LDL than in HDL demonstrated that apoproteins influence the fluidity of LDL and HDL. It is clear that HDL, despite its lower cholesterol level, was usually more rigid than LDL. Possibly the structural differences characterized by this physicochemical parameter are related to the reported differences in atherosclerotic tendencies associated with the levels of LDL and HDL cholesterol [ 311. References 1 Turpeinen. 0.. Diet and coronary events, J. Amer. Diet. Assn.. 52 (1968) 209. 2 Hornstra. G.. Dietary fats and arterial thrombosis, Haemostasis. 2 (1973/74) 21. 3 Renaud, S.. Thrombotic, atherosclerotic and lipemic effects of dietary fats in the rat, Angiology. 20 (1969) 657. 4 Renaud. S. and Gautheron. P.. Influence of dietary fats on atherosclerosis, coagulation and platelet phospholipids in rabbits, Atherosclerosis, 21 (1975) 116.
241
5 Soutar, A., Does dietary fat influence plasma lipoprotein structure? Nature (Land.), of lipid regions determined 6 Shinitzky, M. and Barenholz, Y., Fluidity parameters polarization. Biochim. Biophys. Acta. 515 (1978) 367. 7 Mantulin, 8 9 10 11 12 13 14
15 16 17 18 19 20 21 22
23
24 25 26 27 28
W.W., Shepherd,
J.,
Gotto.
Jr.,
A.M., and Pownall.
H.J.. Dietary
effects
273 (1978) 11. by fluorescence
on human lipopro-
tein composition and structure, Biophys. J., 21 (1978) 145a. of the lipoproteins of Tall, A.R., Atkinson, D.. Small, D.M. and Mahley, R.W., Characterization atherosclerotic swine, J. Biol. Chem.. 252 (1977) 7288. Hillman, G., Mabrey, S., Pitlick, F. and Engleman, D.,Dietary manipulation of cholesteryl ester melting temperatures in rabbit low density lipoproteins, Biophys. J., 17 (1977) 74a. Castellino, F.J.. Thomas, J.K. and Ploplis, V.A.. Microviscosity of the lipid domains of normal and hypercholesterolemic very low density lipoprotein, Biochem. Biophys. Res. Comm.. 75 (1977) 857. Gaman, E., Fisher, H. and Feigenbaum, A.S., An adequate purified diet for rabbits of all ages. Nutr. Repts. lntl.. 1 (1970) 35. Baenziger, N.L. and Majerus, P.W.. Isolation of human platelets and platelet surface membranes, Methods Enzymol., 31 (1974) 149. Hatch, F.T. and Lees, R.S., Practical methods for plasma lipoprotein analysis, Adv. Lipid Res.. 6 (1968) 1. Lindgren. F.T., Serum lipoproteins - Isolation and analysis with the preparative and analytical ultracentrifuge. In: R.M. Burton and F.C. Guerra (Eds.), Fundamentals of Lipid Chemistry, Bi-Science Publications, Webster Groves, MO, 1974, p. 475. Lowry. O.H., Rosebrough, N.J., Farr. A.L. and Randall, R.J., Protein measurement with the Folin phenol reagent, J. Biol. Chem.. 193 (1951) 265. Rosenthal. H., Pfluke, M. and Buscaglia. S.. A stable iron reagent for determination of cholesterol, J. Lab. Clin. Med., 50 (1957) 318. Bartlett, R., Phosphorus assay in column chromatography, J. Biol. Chem., 234 (1959) 466. Shinitzky, M. and Barenholz, Y., Dynamics of the hydrocarbon layer in liposomes of lecithin and sphingomyelin containing dicetlylphosphate, J. Biol. Chem., 249 (1974) 2652. Shinitzky, M. and lnbar, M., Microviscosity parameters and protein mobility in biological membranes, Biochim. Biophys. Acta. 433 (1976) 133. Azumi. T. and McGlynn. S.P., Polarization of the luminescence of phenanthrene, J. Chem. Physics, 37 (1962) 2413. Chew R.F. and Bowman, R.L., Fluorescence polarization - Measurement with ultraviolet-polarizing filters in a spectrophotofluorometer, Science, 147 (1965) 729. Shepherd, J., Packard, C.J., Patsch, J.R., Gotto, Jr., A.M. and Taunton. O.D.. Effects of dietary polyunsaturated and saturated fat on the properties of high density lipoproteins and the metabolism of apolipoprotein A-l. J. Clin. Invest.. 61 (1978) 1582. Morrisett. J.D., Pownall, H.J., Jackson, R.L.. Segura, R., Gotto, Jr., A.M. and Taunton. O.D., Effects of polyunsaturated and saturated fat diets on the chemical composition and thermotropic properties of human Plasma lipoproteins. In: R.T. Holman and W.H. Kunau (Eds.), Polyunsaturated Fatty Acids (American Oil Chemists Society Monograph No. 4). A.O.C.S. Publishers, Champaign, IL, 1977, P. 139. Jonas, A. and Jung. R.W., Fluidity of the lipid phase of bovine serum high density lipoprotein from fluorescence polarization measurements, Biochem. Biophys. Res. Comm., 66 (1975) 651. Deckelbaum, R.J.. Tall, A.R. and Small, D.M., Interaction of cholesterol ester and triglyceride in human plasma very low density lipoprotein, J. Lipid Res.. 18 (1977) 164. Dale. R.E., Chen, L.A. and Brand, L., Rotational relaxation of the microviscosity probe diphenylhexatriene in paraffin oil and egg lecithin vesicles, J. Biol. Chem., 252 (1977) 7500. Shattil. S.J. and Cooper, R.A., Membrane microviscosity and human platelet function. Biochemistry. 15 (1976) 4832.
Jackson, R.L., Morrisett, J.D. and Gotto. Jr., A.M., Lipoprotein structure and metabolism, Physiol. Rev., 56 (1976) 259. 29 Smith, L.C., Pownall, H.J. and Gotto, Jr., A.M., The plasma lipoproteins - Structure and metabolism, Ann. Rev. Biochem.. 47 (1978) 751. 36 Jonas, A., Microviscosity of lipid domains in human serum lipoproteins, Biochim. Biophys. Acta. 486 (1977) 10. 31 Miller, G.J. and Miller, N.E., Plasma high density lipoprotein concentration and development of ischemic heart disease, Lancet, 1 (1975) 16.