Role of apolipoproteins in cellular cholesterol efflux

Bwchimica Elsevier

et Biophysics

Acta 875 (1986) 419-428

419

BBA 52098

Role of apolipoproteins in cellular cholesterol efflux J. DeLamatre Department

*, G. Wolfbauer,

of Physiology and Biochemistry,

(Revised

M.C. Phillips and G.H. Rothblat

Medical College of Pennsyloanra, 3300 Henry Avenue, Phdudelphm, PA 19129 (U.S.A.)

(Received June 18th, 1985) manuscript received September

Key words:

**

Apolipoprotein;

30th, 1985)

Cholesterol

efflux

The effects of serum apolipoproteins, particle size and concentration on the effectiveness of phosphatidylcholine (PC)-containing acceptor particles in causing release of cholesterol from cells growing in culture have been investigated. The acceptor particles were prepared by detergent-dialysis procedures and were either egg PC small unilamellar vesicles (SUV) or discoidal complexes of egg PC with apoproteins from human high-density lipoprotein (HDL). Gel filtration chromatography was employed to isolate particles of defined composition and size. The half-times (t,,*) for the unidirectional efflux of cholesterol from cells prelabeled with [ 3H]cholesterol were measured as a function of acceptor PC concentration in the extracellular medium. HDL apolipoprotein-egg PC discoidal complexes at 100 pg PC/ml gave the following t,,, values when incubated with rat FuSAH hepatoma, human HepC2 hepatoma, human GM3468 skin fibroblast, L-cell and mouse 5774 macrophage-tumor cells: 11 f 2, 22 f 5, 84 f 18, 17 f 2 and 32 f 6 h, respectively. Equivalent experiments using purified apolipoprotein A-I or the total apolipoprotein C fraction to form the egg PC complexes showed that the r,,, values for the hepatoma cells were unaltered. However, with the fibroblasts, L-cells and 5774 macrophages, the apolipoprotein C complexes gave significantly longer t,,, than complexes of egg PC with either apolipoprotein A-I or HDL apolipoprotein which gave the same t,,*. An analysis based on the theory of fast coagulation of colloid particles to describe collisions between desorbed cholesterol molecules and acceptor particles predicts that the dependence of r,,, for cholesterol efflux from a given cell to different acceptors should be normalized when the extracellular level of acceptors is expressed in terms of the product of the radius of the particle times the number concentration of acceptor particles. The decrease in t,,, for cholesterol efflux from fibroblasts when the egg PC acceptor was changed from an SUV to an apolipoprotein HDL discoidal complex is consistent with the above concepts. The primary effect of the apolipoproteins in promoting cellular cholesterol efflux seems to be the solubilization of PC so that the PC is present in the extracellular medium as many small particles.

* Present address: Department of Physiology, Louisiana State Universitv Medical Center, 1542 Tulane Avenue. New Orleans, LA 70112, U.S.A. ** To whom correspondence should be addressed. Abbreviations: PC, phosphatidylcholine; SUV, small unilamellar vesicle(s); HDL, high-density lipoprotein; VLDL, verylow-density lipoprotein; Hepes, 4-(2-hydroxyethyl)-2-piperazineethanesulfonic acid. 00052760/86/$03.50

0 1986 Elsevier Science Publishers

Introduction In order to investigate the mechanism of the initial step in the ‘reverse cholesterol transport’ of cholesterol from cells in peripheral tissue to the liver for clearance from the body [1,2], researchers have employed systems in which the unidirectional

B.V. (Biomedical

Division)

420

efflux of cholesterol from cells granting in culture can be monitored. Previous studies from this laboratory have shown. that the mechanism involves diffusion of’ cholesterol molecules which have desorbed from the plasma nlembrane through the aqueous phase until they collide with, and are absorbed by, acceptor particles present in the extra~e11ular medium 13-51. This aqueous diffusion mechanism has been confirmed in other cell systems (6-91. Different cell types release cholesterol to a given acceptor at different rates and this is apparetltly a function of the plasma membrane structure [S]. Conversely, a given cell type releases cholesterol to different acceptors at different rates 141; the quantitative aspects of these effects are unclear. Since the putative acceptor of cholesterol in reverse cholesterol transport is high-density lipoprotein (HDL) \1,2,10], there has been interest in investigating the role of apolipoproteins from human HDL in cell cholesterol efflux. It has been established that apolip~)proteins alone are poor cholesterol acceptors [4,13,12] and that addition of pbospholipid promotes cholesteroi efflux [ll--151. Comparison of the performances of apolipoproteins C-I, C-II and C-III, and A-I and A-II when complexed with various pbospholipids revealed that there was apparently not a very specific requirement [15]. Detailed interpretation of these efflux data is complicated because the apoproteil~/phospholipid stoicbiometry of the various complexes was not constant. In the present study, we use a detergent dialysis procedure to prepare complexes of egg phosphatidylcholine (PC) and either human HDL total apoprotein, apolipoprotein A-I or apolipoprotein C. The complexes have defined sizes and stoichiometries and their efficiencies in inducing cholesterol efflux from several cell types are compared. Any specific effects in the efflux process which are attributable to a particular apoprotein should be apparent in such a comparison. In addition. to further investigate the roles of apoprot~ins on acceptor particles and acceptor particle size in determining acceptor efficiency, cholesterol efflux to egg PC small unilamellar vesicles (SW?) and discoidal apoproteinPC complexes are compared. The kinetic data are interpreted by application of the theory for fast coagulation of colloid particles to estimate the

collision frequency between desorbed molecules and acceptor particles.

cholesterol

Materials. [7-‘H)Cholesterol, lo-20 Ci/mmol, was purchased from ICN ~adiochemicals (Irvine, CA) or New England Nuclear ( repurified on prescored TLC plates from AnalabsFoxboro (North Haven, CT). [carho.xyf-‘4C]Cholic acid ~sod~urn salt) and sodium cholate were supplied by Research Products International (Mt. Prospect, IL). L-Phosphatidylcholine (egg yolk), chromatographically purified, was obtained from ~albiochem (L~olla, CA) or from Sigma (St. Louis, MO). Cell culture supplies were pure from Flow Labs (McLean, VA) and G (Grand Island, NY). Solvents and liquid sci tion fluid were from Fisher Scientific (King of Prussia, PA). Human apolipoproteins were isolated by the following procedures. was obtained by ext 1.067-1.21 g/ml) at a con in 0.9% (w/v) NaCl/2 m EDTA (pH 7.6) with 50 vol. of ethanol/ether (3: 2, v/v) follo~ved by extensive washing with cold ether 1161. The solvent-free, vacuum-dried, DL apolipoprotein showed the following bands on sodium dodecyl sulfate (S~S)-pol~acrylamide gel electroph~resis: apolipoproteins A-I, A-II, small amounts of apolipoproteins E and C and traces of albumin. Apolipoprotein A-I was obtained by dissolving HDL apo1ipoprotein in 6 M guanid~ne-HAL and fractionating it on a Sephacryl S-200 g conditions using 7 EDTA/lO mM Tris (pH 8.6) as the elution buffer. Concentrated apolipoproteins A-I- and A-II-containing fractions were treated with 10% mercaptoethanol and rechromatographed on Sephacryl S-200. The column was eluted under reducing conditions using the above 7 M urea solution with 1 mM dithiothreitol added. The apolipoprotein A-I fractions were pooled after the separation, diaiysed in distilled water fpI-I 8) and 1yophiIized. The apolipoprotein le band on SDS-polyac~lamide gel electrophoresis. Apolipoprotein C was isolated

421

from human VLDL (d < 1.006 g/ml) by delipidation with acetone [17]. The apolipoprotein C-containing liquid phase was dried down on a rotary evaporator and extracted with ethanol/ether (1 : 3, v/v) followed by several washes with ether. The solvent-free apolipoprotein C peptides were dissolved in saline, dialysed against distilled water (pH S), and lyophilized. Examination of this apolipoprotein C preparation by SDS-polyacrylamide gel electrophoresis showed one wide band representing a combination of apolipoproteins C-I. C-II and C-III. Cholesterol acceptor particles. Discoidal complexes of HDL apolipoprotein, apolipoproteins A-I and C with egg phosphatidylcholine (PC) were prepared by the cholate dialysis procedure, essentially as described by Matz and Jonas [18]. Typically, mixtures of apoprotein/ phospholipid/ cholate at a mass ratio of 1 : 2.5 : 1.552.5 were formed in a buffer of 10 mM Tris/l mM EDTA/l mM NaN,/lSO mM NaCl (pH 8.0) at a final concentration of 5 mg PC/ml. Lipids were dried under N, and kept for 2 h in a vacuum oven. Buffer, cholate and apoprotein solution were added and the lipids dispersed by shaking. During an overnight incubation at 4°C the reaction mixture cleared and was then dialysed extensively against Tris buffer. In some preparations, this dialysis step was omitted; there were no significant differences in the final apoprotein-phospholipid complexes. In order to size the apoprotein/phospholipid particles, they were subjected to gel filtration on a Sepharose CL-4B column (1.5 X 90 cm) and eluted with 0.15 M NaC1/0.02% EDTA (pH 7). The peak fractions were concentrated in an ultrafiltration cell (Amicon Model 52). The removal of radiolabeled cholate was monitored in control experiments; the phospholipid/cholate molar ratio in the final complexes was greater than 100: 1 (cf. Ref. 18). Aliquots for electron microscopy, protein and phospholipid determination were taken and the complexes were dialysed overnight against Hepes-buffered medium. Finally, the discoidal particles were sterilized and diluted to the required concentrations. Small unilamellar vesicles (SUV) of egg PC were prepared by the cholate dialysis procedure described by Brunner et al. [19] which included essentially the same steps as described for the

apoprotein-PC complexes. The ratio of egg PC/ cholate was 1 : 10 (w/w). Negative-stain electron microscopy of the different acceptor particles was performed as described previously [4]. 50-100 particles in each preparation were measured to ascertain the particle size. CeN culture. Cells of the rat hepatoma cell line Fu5AH [20,21] and the human hepatoma cell line HepG2 [22.23] were grown in 75 cm’ flasks in 15 ml minimum essential medium supplemented with 5% calf serum and 10% calf serum, respectively. L-cells were maintained in minimum essential medium and 10% calf serum. The human normal fibroblast cell GM3468 (Human Genetic Mutant Cell Repository, Camden, NJ) and the mouse macrophage-like cell line J-774 [24] were grown in Williams medium E and 10% fetal bovine serum or 10% heat-inactivated fetal bovine serum, respectively. All media contained vitamins, bicarbonate and 50 pg/ml gentamycin or 50 mu/ml penicillin, and 50 pg/ml streptomycin. All cells were grown at 37°C in 95% sir/5% CO,; routine screening showed them to be free of mycoplasma. 6 days prior to each experiment, cells were seeded in 35-mm dishes and grown for 2 days to approx. 75% confluency. The monolayers were washed free of the medium and refed with labeling medium which contained minimum essential medium or Williams medium E, solvent-extracted delipidized calf serum-protein [25] at 5 or 7.5 mg/ml, 6 pg/ml egg PC, 2.4 pg/ml unesterified cholesterol, 1% fetal bovine serum and 1 pCi/ml [ ‘HIcholesterol. Following a 2 day incubation, the radioactive medium was removed, and the cells were extensively washed and refed with fresh medium supplemented with 5 mg/ml delipidized calf serum-protein. After an overnight incubation to allow equilibration of cell [ ‘HIcholesterol pools, the protein-containing medium was removed and the cell monolayers were washed three times with 2 ml of Hepes-buffered minimum essential medium or Williams medium E immediately prior to the efflux phase of the experiment. The cholesterol efflux experiments were started by adding 2 ml of each acceptor particle concentration in Hepes-buffered minimum essential medium (FuSAH, HepG2 and L-cells) or Hepesbuffered Williams medium E (GM3468 and J-774

422

cells) to 35-mm dishes (in triplicate). A O.l-ml aliquot of the incubation medium was taken each 90 min over a 9 h period and counted in 5 ml of scintillation fluid to determine the amount of radioactive cholesterol released by the cells as described previously [4]. At the end of the incubation, the remaining medium was removed and cells were washed extensively with cold phosphatebuffered saline. Analysis of total radioactivity, cholesterol and protein were performed as follows: each culture dish containing the washed cell monolayer was incubated for 30 min with 1 ml of isopropanol [26] containing coprostanol as a gasliquid chromatography (GLC) standard. An aliquot was taken for radioactive counting and the remaining sample was removed for cholesterol determination by GLC [27]. The extracted cell monolayers were dried, after which 1 ml of 1 M NaOH was added to solubilize proteins which were then quantitated by a modified procedure of Lowry et al. [28]. Lipids were extracted by the method of Bligh and Dyer [29] or with isopropanol, as described above. Phospholipid phosphorus was measured as described by Sokoloff and Rothblat [30]. Results Cholesterol

release

from

cells

with

different

upoprotein / egg PC acceptor particles

In order to understand in detail the mechanism of cholesterol efflux from cells to phospholipid acceptors in the extracellular medium, it is necessary to employ acceptor particles of well-defined structure. In particular, it is important to utilize as homogeneous a preparation as possible and to know the chemical composition as well as the particle size and morphology. These parameters for the apoprotein-egg PC complexes used here are summarized in Table I. The discoidal complexes with characteristic phospholipid bilayer thickness of approx. 5 nm are similar to those described by Matz and Jonas [18]. The complexes were fractionated by gel filtration to give equal sizes for the three types of apoproteins utilized. Small unilamellar egg PC vesicles were also prepared by sodium cholate dialysis and have properties similarly to those described by Brunner et al. [19]. Selected acceptor properties were reexamined after incubation with either FuSAH hepatoma, HepG2

TABLE

I

CHARACTERISTICS PHATIDYLCHOLINE

OF APOLIPOPROTEIN-EGG DISCOIDAL PARTICLES

PHOS-

Apo. apolipoprotein. Results are expressed as the mean + SD. n, number of preparations. Stokes radius refers to the hydrodynamic radius of equivalent sphere obtained from elution volume on a calibrated gel-filtration column. The diameter refers to the large dimension of the discs measured from electron micrographs.

Egg PC-apoprotein Stokes’ radius (nm) Diameter

(nm)

(w/w)

Apoprotein

present

Apo HDL

Apo A-I

Apo C

2.6 + 0.6 (n=8) 6.8*1.6 (n=8) 17.6k4.7 (n =4)

3.3f0.6 (n=S) 6.7k1.4 (n=4) 17.8*4.6 (n=3)

2.7*0.6 (n=8) 7.1*1.9 (n=7) 17.3+3.7 (n=3)

hepatoma or GM3468 fibroblast cells for 9 h; their PC/apoprotein ratio, gel filtration elution volume and electron microscopic appearance were not significantly altered. The apoprotein-egg PC complexes of Table I were employed as acceptors at a concentration of 100 p-18 PC/ml to compare the half-times for cholesterol efflux (t,,,) for different cell types. Under the experimental conditions used here, unidirectional efflux of cholesterol is first-order with respect to cell-free cholesterol concentration [3-51. The rate constant (or t,,,) at a particular acceptor concentration is characteristic of the cell type and is apparently a function of plasma membrane structure [5]. As reported before [4], when the Fu5AH hepatoma cells and fibroblasts are incubated with identical acceptors at the same concentration the rate of cholesterol loss from Fu5AH cells is greater than from fibroblasts (Table II). The data in Table II (all cell monolayers were 90-100% confluent) also show that different hepatoma cells give different t,,, values because the value for HepG2 is twice that of Fu5AH cells. Furthermore, the exact nature of the apoprotein constituent of the acceptor particle is not critical in determining t,,, because HDL apolipoprotein (a mixture whose main constituents are apolipoproteins A-I and A-II) gives the same efflux as either apolipoprotein A-I or apolipoprotein C alone when complexed with egg PC. The GM3468

423

TABLE

If

COMPARISON OF HALF-TIMES FOR EFFLUX FROM DIFFERENT CELLS

CHOLESTEROL

Apo, apolipoprotein. The acceptor concentration was 100 pg egg PC/ml medium and the incubation time was 9 h at 37°C. Mean + S.D. n. number of experiments, each done in triplicate. Cell type

Half-times(h) Apo HDL-egg acceptor ’

PC Apo A-I-egg acceptor

PC Apo C-egg PC acceptor

ll& 2 (n=6) 22+ 5 (n = 5) 74+19

11* 3 (n=12) 252 4 (n=7) 131*37x (n=12) 285 1” (n=3) 48k13 * (n=?)

FuSAH

llir 2 (n=lZ) HepG2 22i 5 (n=8) GM 3468 84_+ 18 (n=l2) 17& 2 L-cell (n=3) J-774 32i 6 (n-7) a See Table I for details. * Statistically different from P c 0.01.

(n=4)

18+ 2 (n=3) 341 6 (n=4)

HDL

apolipoprotein-egg

PC at

fibroblast consistently shows the longest f,,,? irrespective of the apoprotein present in the acceptor particle. Interestingly, the three non-liver-derived cell lines, GM3468 fibroblasts, J-774 macrophages and L-cells, give the same t,,jz for complexes of HDL apolipoproteins and apolipoproteins A-I whereas the apolipoprotein C complexes give longer t,,, values for cholesterol efflux (Table II). in Table II The trends in t,,, values demonstrated at an acceptor concentration of 100 pg egg PC/ml extracellular medium are maintained when the concentration is 500 fzg PC/ml (data not shown). values presented in Table II were The II/2 obtained from the kinetics of efflux of radiolabelled cholesterol from the cells. Analysis by GLC of the mass of cholesterol remaining in the cells after the incubation conditions of Table II confirmed the tracer data in that, after 9 h, the three types of apoprotein complexes remove approx. 33% of the free cholesterol in Fu5AH cells. The variation in cholesterol mass determinations together with the low efflux from the other cells excluded the possibility of detecting mass differences after exposure to the acceptor particles for 9 h. It should be noted that under the experi-

mental conditions used in this study the proportion of total cell cholesterol present as cholesteryl ester was less than 10%. The t1/2 data at a single acceptor concentration (Table II) for Fu5AI-I and GM3468 cells, which may be considered representative of liver and peripheral cells, were expanded to include a range of acceptor concentrations (Figs. I and 2). The data in Fig. 1 demonstrate that z,,,~ for the Fu5AH cells shows a dependence on acceptor concentration similar to that reported before, becoming essentially independel~t of acceptor concentration at PC greater than 500 pg/ml and giving a minimum t,,, of approx. 3 h (cf. Ref. 4). It is also apparent that t,,, is independent of the nature of the apoprotein when HDL apolipoprotein, apolipoproteins A-I and C were used to prepare discoidal complexes. A similar dependence of fi‘,,? on the egg PC concentration in HDL apolipoprotein or apolipoprotein A-I complexes is observed with GM3468 cells (Fig. 2) although the t,,, is always longer than for FuSAH cells (cf. Ref. 4). The shortest t,,2 for GM3468 cells is about 25 h at an egg PC concentration of 1000 ,ug/ml. Unlike the situation with FuSAH cells, the behavior of apolipoprotein C-egg PC complexes is different from HDL apolipoprotein or apolipoprotein A-I complexes in that zlj2 is longer (Fig. 2). When apolipoprotein C complexes are used, about 4times the concentration of PC is required to give the same t,,, as is observed with HDL

IM

zw Concentrationof

m acceptor

4ca

x0

1003

{rg PC/mll

Fig. 1. Comparison of the half-times for cholesterol efflux from Fu5AH hepatoma cells using acceptors containing different apolipoproteins. Discoidal acceptor particles were prepared from apolipoprotein and egg phosphatidylcholine (PC) by cholate dialysis and sized by gel filtration as described in the text. X, HDL apoiipoprotein-egg PC; 0. apolipoprotein A-I-egg PC; 0, apolipoprotein C-egg PC. The half-times are mean (+ S.E.) of 2-12 experiments perfokmed in triplicate; fewer measurements were made at the highest acceptor concentration to conserve material.

Concentration of acceptor

(pg PC/ml)

Fig. 2. Comparison of the half-times of cholesterol efflux from GM3468 fibroblast monolayers observed with discoidal particles of egg PC complexed with different human apolipoproteins. The sizes and compositions of the acceptor particles are presented in Table 1. X, HDL apolipoprotein-PC (N = 3-12); 0, apolipoprotein A-I-PC (N = 2-4); 0, apolipoprotein C-PC (N = 3-12). The half-times are mean (*SE.) of the indicated number of experiments each performed in triplicate.

apolipoprotein or apolipoprotein A-I complexes. It should be noted that even at the highest concentrations of acceptors, a common minimum t,,,, is not attained. When the apoprotein/egg PC ratio was varied by +0.6 (cf. Table I) by selecting different fractions from gel filtration columns, no systematic effects on the t,,, data in Figs. 1 and 2 were detected. Effect

of acceptor

particle

coidal complexes with HDL apolipoprotein or as SUV is depicted in Fig. 3. The t,,, values for GM3468 cells reported in Fig. 3 are shorter than those in Fig. 2 and Table II because cells at a lower passage number were utilized for these particular experiments; it appears that as the passage number of this fibroblast cell line increases the t,,z for cholesterol efflux increases (Wolfbauer, G. and Rothblat, G.H., unpublished observation). It is clear that both HDL apolipoprotein/PC discs and SUV give rise to cholesterol efflux but addition of apoprotein makes the PC a more effective acceptor because, on the basis of PC concentration in the extracellular medium, the discs give lower t,,, values. The apoprotein induces major structural changes in the PC organization compared to the SUV; although the bilayer structure is retained, the PC surface exposed is greatly altered as the apoprotein/PC discs have dimensions of 13 x 5 nm compared to the 23 nm diameter of SUV, and the number of PC molecules per particle is about

‘9

size on cell cholesterol

efflux

Having explored the performance of apoproteins as acceptors for cell cholesterol by comparing HDL apolipoprotein, apolipoproteins A-I and C in discoidal complexes with similar physical properties, it was also of interest to determine how the presence of HDL apoproteins on PC acceptors modifies acceptor performance by comparing the rates of cholesterol removal obtained using either an HDL apolipoprotein/PC acceptor or an acceptor with no apoprotein. A PC small unilamaller vesicle was used as an acceptor with no apoprotein present. It is known from the literature [ll-131 that phospholipid is required to achieve efficient cholesterol efflux and that apolipoproteins alone are poor cholesterol acceptors. A comparison of egg PC performance as a cholesterol acceptor when presented to GM3468 fibroblasts as either dis-

I

0 0

ml

I

Bm m Concentration of acceptor lpg PC/ml) 4M

I

loDo

lml

Fig. 3. Comparison of the half-times for cholesterol efflux from human GM3468 fibroblast monolayers with egg phosphatidylcholine acceptor particles present in the extracellular medium as small unilamellar vesicles (A) or discoidal complexes ( X ) with the apolipoproteins of human HDL. The half-times are the mean (kS.E.) of two experiments each performed in duplicate. The small unilamellar vesicles were prepared by cholate dialysis and fractionated by gel filtration as described in the text. The mean diameter of the vesicles used in these experiments, as determined by electron microscopy and gel filtration, was 23 nm. From volumetric considerations, these particles contained about 5.3.10-‘* pg egg PC/particle. The egg PC-HDL apolipoprotein discoidal complexes used in this experiment had a stoichiometry of 2.2/l (w/w) PC/HDL apolipoprotein and mean dimensions of 13 X 5 nm by electron microscopy; the Stokes’ radius by gel filtration chromatography was 6.4 nm. From volumetric considerations, each discoidal particle contained about 5.0.10-‘7 pg PC.

425

lo-times greater for the vesicles (Fig. 3). The Appendix presents an analysis based on the theory of fast coagulation of colloid particles of the likely effects of alterations in acceptor particle size on the f;,, value for cholesterol efflux from a given cell type. Eq. 3 in the Appendix predicts that the frequency of collision between cholesterol molecules which have desorbed from the cell plasma membrane [5] and acceptor particles is proportional to the product of the radius (R,) and number concentration (C,) of acceptor particles. Consequently, the dependence on t,,? for cholesterol efflux of the extracellular level of different acceptors should be normalized when expressed in terms of the product (R.C,). Fig. 4 shows that, when the f,,, data for Fig. 3 are replotted in terms of the product (R.C>,), the t,,? for SUV and discs are normalized to a large extent. This implies that the model for cholesterol efflux and the theoretical considerations applied in the Appendix give a reasonable description of the acceptor performance of egg PC in vesicles and discoidal complexes with HDL apolipoprotein.

*

6

01 0

I

100

(Number

m of acceptor

1

/

x0

4x7

particles/unit InmImI

volumel

1

!m

Kc

x particle radius

x lo’?

Fig. 4. Dependence of the half-times for cholesterol efflux from human GM3468 fibroblast monolayers on the concentration and size of egg PC-containing acceptors in the extracellular medium. The half-times for small unilamellar vesicles (A) and egg PC-apolipoprotein discoidal complexes ( X) are taken from Fig. 3. The abscissa is plotted in terms of the product (number concentration of particles~particle Stokes’ radius) which is a reflection of the collision frequency of acceptor particles with cholesterol molecules which have desorbed from the cell plasma membrane (see Eqn. 3 in Appendix).

Discussion Phospholipids alone as SUV can promote cholesterol efflux from cells [3,4.13]. but their effectiveness on a mass basis is enhanced if the phospholipid is complexed with an apolipoprotein [l l-141. This effect has been demonstrated before for FuSAH cells by comparing SUV and spherical HDL apolipoprotein/PC particles [4]. The data in Fig. 3 extend these observations to GM3468 fibroblasts and imply that the difference between SUV and apoprotein-PC complexes applies to all cell types. The increase in acceptor efficiency could be simply a generalized effect due to the formation from a fixed mass of PC of about IO-times as many discoidal apoprotein-PC complexes as SUV. However, it is possible that the apoprotein molecules present in the discoidal complexes may also perform some specific function which enhances cholesterol efflux. Indeed, an earlier study [15] did demonstrate some differences in human apoprotein effectiveness when cholesterol was removed from Landschutz ascites cells using a variety of apoprotein-phospholipid complexes. The present investigation extends the above study to other cell types while utilizing various apoproteins associated with a single type of PC in complexes of similar size and lipid/protein stoichiometry. The use of apoprotein-PC complexes with similar physical properties (Table I) facilitates the separation of physico-chemical effects from the influence of any specific apoprotein function. The presence of a specific apolipoprotein such as apolipoprotein A-I in the acceptor particle could significantly influence efflux by (1) the apolipoprotein A-I binding to the plasma membrane of the cell, thereby raising the local concentration of acceptor particles and increasing cholesterol efflux (cf. Ref. 31) and/or (2) the apolipoprotein A-1 in the surface of the acceptor particle increasing the rate of absorption of desorbed cholesterol molecules into the acceptor. The cholesterol efflux data in Fig. 1 and Table II demonstrate that with FuSAH and HepG2 cells there is no such specific effect of apolipoproteins of the A and C families because the apoprotein-egg PC complexes are equally efficient regardless of the particular apoprotein present. The situation may be different with GM3468 fibroblasts, J-774 macrophages and

426

L-cells because the t,,, for efflux is longer when apolipoprotein C alone rather than HDL apolipoprotein or apolipoprotein A-I is present in the acceptor (Fig. 2 and Table II). One explanation for these differences could be that the three nonliver cells have receptors which recognize particles that contain apolipoprotein A-I and thus increase the local concentration of these particles near the surface of the cell. However, if the differences between apolipoprotein A-I- and apolipoprotein C-containing acceptors were solely due to an interaction with a specific binding site, one would expect the differences between the two particles to be most dramatic below receptor saturation at low acceptor concentrations. As can be seen from Fig. 2, this is not the case and the higher t,,, with apolipoprotein C acceptors is seen at all concentrations. Alternatively, the differences between the acceptors observed with some cell lines could be a consequence of the apoprotein directly influencing the composition or structure of the plasma membrane, thereby modifying the rate of cholesterol desorption from the membrane. The above evidence for binding of acceptor particles to the cell surface not being a dominant factor in the determination of the t,,, values for cholesterol efflux is consistent with the arguments summarized in the Appendix and Fig. 4. Effective removal of cholesterol from the aqueous phase immediately outside the cell plasma membrane requires conditions which optimize absorption of cholesterol molecules by the acceptor particles. This, in turn, requires maximization of the collision frequency between cholesterol molecules which have desorbed from the plasma membrane into the extracellular medium and the acceptor particles. Consequently, presentation of PC to cells as many small particles is expected to give the most effective acceptor performance because the cholesterol-collision frequency is proportional to the product of acceptor particle concentration and particle radius (cf. Eqn. 3 in the Appendix). The data in Fig. 4, showing that the enhancement in collision frequency due to the change from SUV to apoprotein-PC discs largely accounts for the differences in acceptor efficiency for these two types of particles, are consistent with the above concepts. The primary effect of the apolipoproteins in promoting cholesterol efflux from cells appears to

be the solubilization of phospholipid so that the phospholipid is present in the extracellular medium as many small particles. This solubilization of phospholipid to form discoidal complexes does not require a specific apolipoprotein and is characteristic of any of the apolipoproteins A, C, D and E which contain amphiphilic cY-helices [32]. Appendix

Acceptor particle size and unidirectional effrux of cholesterol from cells The free cholesterol content of cells decreases when suitable extracellular acceptors are incubated with cells. The mechanism of the unidirectional flow of cholesterol from tissue culture cells to acceptor particles involves diffusion of cholesterol molecules through the aqueous phase, with the overall rate being influenced by the rate of desorption from the cell plasma membrane [3-51. The efflux of much of the cell free cholesterol is firstorder with respect to the concentration of cholesterol in the plasma membrane [5]. Consequently, in a period where the cell free cholesterol content is not depleted significantly, the steadystate flux of cholesterol from the plasma membrane into a closed extracellular medium creates an effectively constant concentration of cholesterol molecules in the aqueous phase. These cholesterol molecules are absorbed rapidly by acceptor particles present at high concentrations in the extracellular medium so that this step is not ratelimiting for the overall transfer of cholesterol from cell plasma membrane to acceptor particles. However, at lower acceptor particle concentrations, the rate of absorption of cholesterol by the acceptor is reduced and can become limiting for the overall cholesterol efflux process [4]. In addition to a dependence of efflux on acceptor concentration, different types of phospholipid-containing acceptors at the same extracellular phospholipid concentration cause a cell to release cholesterol at different rates [4]. In an attempt to gain some quantitative understanding of these acceptor effects, we apply here the theory for describing coagulation of colloidal particles to relate the particle size of an acceptor to its efficiency in removing cholesterol from cells. The process of absorption of a cholesterol mole-

421

cule into an acceptor particle is analogous to the coagulation of colloidal particles, in that the rate is determined by the frequency of collisions between cholesterol molecules and acceptor particles. We assume that the probability of a collision resulting in absorption of the cholesterol molecule is unity. The collision frequency depends on the concentration of acceptor particles and cholesterol molecules, as well as the rate of their Brownian motion which is characterized by their diffusion coefficients. Thus, the collision frequency (Z,,) per unit volume of medium between cholesterol molecules and acceptor particles, with the approximation that both can be treated as spheres, is given by Eqn. 1 (see Eqn. 7.18 in Ref. 32):

Acknowledgements

Z,,=48(~~++Da)(R~+Ra)CcCa

(1)

In Eqn. 1, D = diffusion coefficient, R = hydrodynamic radius and C = particle concentration; the subscripts c and a refer to cholesterol molecules and acceptor particles, respectively. In the case of acceptor particles, such as small unilamellar phospholipid vesicles or apolipoproteinphospholipid complexes, which are multimolecular aggregates R, zs R, so that (R, + R,)- R,. The value of D for a spherical particle can be computed from the Stokes-Einstein equation where D = kT/tiqR where k is the Boltzmann constant, T is temperature and q is the viscosity of the liquid medium [33]. Since at a given temperature in a medium of constant viscosity D a l/R, it follows that, when R, zs- R, (this condition holds in this study because the average radius of a cholesterol molecule is approx. 0.6 nm (341 which is at least an order of magnitude smalfer than the radii of the acceptor particles (Table I and Fig. 3)), D, > Da and (DC + Da) = DC. Application of these conditions to equation [l] gives; Z,, = 4Dc R &Xl,

so that for a constant

GP(R,G).

cholesterol molecules which have desorbed from the plasma membrane of cells into a given volume of extracellular medium is a function of acceptor concentration and particle size. Thus, where the rate of transport of cholesterol molecules from cell to acceptor is influenced by the rate of the step involving absorption of cholesterol into the acceptor particle (i.e., by the collision frequency Z,,), the relative efflux of cellular cholesterol with different acceptors can be predicted by application of Eqn. 3. The dependence of half-times for cholesterol efflux on the extracellular level of different acceptors should be normalized when expressed in terms of the product (R,L!,).

(2)

value of C,: (3)

It should be noted that, in situations where the viscosity of the liquid medium is altered, Z,, is inversety proportional to 7. Eqn. 3 shows that the collision frequency between acceptor particles and

We thank Michael Goldfinger for expert technical assistance, and James Diven and the Pathology Department, Medical College of Pennsylvania, for help with the electron microscopy. This research was supported by NIH Program Project Grant HL22633, Training Grant HL07443 and the W.W. Smith Charitable Trust. References 1 Glomset, J.A. (1980) Adv. ht. Med. 25, 91-116 2 Norum, RR., Berg. T., Helgerud, P. and Drevon, C.A. (1983) Physiol. Rev. 63. 1343-1419 3 Phillips, M.C., McLean, L.R., Stoudt, G.W. and Rothblat, G.H. (1980) Atherosclerosis 36, 409-422 4 Rothblat, G.H. and Phillips, M.C. (1982) J. Biol. Chem. 257, 4715-4782 5 Belhni, F., Phillips, M.C., Pickell. C. and Rothbfat, G.H. (1984) Biochim. Biophys. Acta 777, 209-215 6 Bojesen, E. (1982) Nature (London) 299, 276-278 7 Wharton, S.A. and Green, C. (1982) Biochim. Biophys. Acta 711, 398-402 8 Lange, Y., Molinaro, A.L., Chauncey, T.R. and Steck, T.L. (1983) J. Biol. Chem. 258, 6920-6926 9 Slotte, J.P. and Lundberg, B. (1983) Biochim. Biophys. Acta 750.434-439 10 Eisenberg, S. (1984) J. Lipid Res. 25, 1017-1058 11 Burns, C.H. and Rothblat, G.H. (1969) B&him. Biophys. Acta 176, 616-625 12 Stein, 0. and Stein, Y. (1973) B&him. Biophys. Acta 326, 232-244 13 Stein, Y., Glangeaud, M.C., Fainaru, M. and Stein, 0. . (1975) B&him. Biophys. Acta 380, 106-118 14 Stein, O., Fainaru, M. and Stein, Y. (1979) B&him. Biophys. Acta 574. 495-504

428

IS Jackson, R.L., Gotto, A.M., Stein, 0. and Stein, Y. (1975) J. Biol. Chem. 250, 7204-7209 16 Scanu, A.M. (1966) J. Lipid Res. 7, 295-306 17 Holmquist, L. and Carlson, K. (1977) Biochim. Biophys. Acta 493. 400-409 18 Matz, C.E. and Jonas, A. (1982) J. Biol. Chem. 257, 4535-4540 19 Brunner, J.P., Skrabal, P. and Hauser, H. (1976) Biochim. Biophys. Acta 455, 322-331 20 Pitot, H.C., Peraino, C., Morse, P.A. and Potter, I.R. (1964) Nat]. Cancer Inst. Monogr. 13, 229-245 21 Rothblat, G.H., Arbogast, L.Y., Kritchevsky, D. and Naftulin, M. (1976) Lipids 11, 97-108 22 Aden, D.P., Fogel, A., Plotkin, S., Damjanov, I. and Knowles, B.B. (1979) Nature (London) 282, 615-616 23 Knowles, B.B., Howe, C.C. and Aden, D.P. (1980) Science 209, 497-499 24 Ralph, P., Prichard, J. and Cohn, M. (1975) J. Immunol. 114, 898-905

25 Rothblat, G.H.. Arbogast, L.Y.. Ouellette, L. and Howard, B.V. (1976) In Vitro 12, 554-557 26 Heider, J.G. and Boyett, R.L. (1978) J. Lipid Res. 19, 514-518 27 Rothblat, G.H. (1974) Lipids 9, 526-535 28 Markwell, M.A.K., Haas, S.M., Bieber, L.L. and Tolbert. N.E. (1978) Anal. Biochem. 87, 206-210 29 Bligh, E.G. and Dyer, W.J. (1959) Can. J. Biochem. Physiol. 37, 911-917 30 Sokoloff, L. and Rothblat, G.H. (1974) Proc. Sot. Exp. Biol. Med. 146, 1166-1172 31 Oram, J.F., Brinton, E.A. and Bierman, E.L. (1983) J. Clin. Invest. 72, 1611-1621 32 Morrisett, J.D., Jackson, R.L. and Gotto, A.M., Jr., (1977) Biochim. Biophys. Acta 472, 93-133. 33 Sheludko, A. (1966) in Colloid Chemistry, p. 219, Elsevier, Amsterdam 34 Craven, B.M. (1976) Nature (Lond.) 260, 727-729