THE JOURNAL OF BIOL~CICAL CHEMISTRY 0 1994 by The American Society for Biochemistry and Molecular Biology, Inc.

VOl. 269, No. 37, Issue of September 16, PP. 22975-22982, 1994 Printed in U.S.A.

The Influence of Apolipoprotein Structure on the Efflux of Cellular Free Cholesterol to High Density Lipoprotein*

(Received for publication, April 11, 1994, and in revised form, June 28, 1994)

W. Sean Davidson, Sissel Lund-Katz, William J. Johnson, G. M. AnantharamaiahS, Mayakonda N. PalgunachariS, Jere P. SegrestS, George H. Rothblat, and Michael C. Phillips5 From the Medical College of Pennsylvania, Department of Biochemistry, Philadelphia, Pennsylvania 19129 and the $Atherosclerosis Research Unit, School of Medicine, University of Alabama at Birmingham, Birmingham, Alabama 35294

The influence of apolipoprotein conformation on the ability of high density lipoprotein (HDL) to remove cel- lular free cholesterol (FC) has not been studied in detail. To address the effects of amphipathic a-helix structure on cellular FC efflux, three class A helical peptides and apolipoprotein (apo) AI were complexed to dimyristoyl phosphatidylcholine (DMPC) to make discoidal com- plexes that were used as acceptors of cell cholesterol. The peptides consisted of an 18-amino acid, am- phipathic, a-helical peptide with the sequence DWL- KAFYLKVAEKLKEAF USA), a dimer of 18A covalently linked by a proline residue (37pA), and acetyl-MA-amide (Ac-l8A-NH,) that has a higher a-helix content than the unblocked 18A molecule. The three peptides strongly mimic the lipid-binding characteristics of the am- phipathic segments of apolipoproteins and form discoi- dal complexes with DMPC that are similar in diameter (11-12 run) to those formed by human apoAf when recon- stituted at a 2.51 (w:w) phospholipid to protein ratio. The abilities of these complexes to remove radiolabeled FC were compared in experiments using cultured mouse L-cell fibroblasts; efflux of FC from both the plasma membrane and the lysosomal pools was examined. For each of the acceptors, the removal of cholesterol from the plasma membrane and lysosomal pools was equally efficient. All four discoidal complexes were equally effi- cient cell membrane FC acceptors when compared at saturating acceptor concentrations of >200 pg of DMPC/ml of medium. However, at the same lipid concen- tration, protein-free DMPC small unilamellar vesicles (SUV) were significantly less efficient. The initial rates of FC removal from cells at saturating concentrations of acceptor particles (V,,) were 12,10,10, and 11% per h, respectively, for the complexes containing either 184 Ac-MA-NH,, 37p4 or apoAI, but only 1% cellular FC per h for the DMPC SUV. The 10-fold higher V,, for the apoproteidpeptide-containing acceptors was likely due to a reversible interaction of apoprotein or peptide with the plasma membrane that changed the lipid packing characteristics in such a way as to increase the rate of FC desorption from the cell surface. This interaction required amphipathic a-helical segments, but it was not affected by the length, number, or lipid-binding affinity of the helices. Furthermore, the efflux efficiency was not dependent on the amino acid sequence of the helical segments which suggests that this interaction is not me-

HL34343, Training Grant HL07443 from the National Institutes of * This work was supported by Program Project Grants HL22633 and

Health, and a predoctoral fellowship from the American Heart Associa- tion, Southeastern Pennsylvania Mlliate (to W. S. D.). The costs of Publication of this article were defrayed in part by the payment of page

in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. charges. This article must therefore be hereby marked “aduertisernent”

5 To whom correspondence and reprint requests should be addressed.

diated by a specific cell surface binding site. At lower acceptor concentrations of <lo0 pg of DMPC/ml of me- dium, differences in efflux efficiency were seen between the peptide-containing particles as a group and the apoAI-containing particle demonstrating that a-helix structure can affect the ability of acceptor particles to either sequester or retain cholesterol molecules.

High density lipoprotein (HDL)’ is thought to play a signif- icant role in the process of reverse cholesterol transport (1). This lipoprotein class contains particles that are heterogeneous in size, shape, and composition (2). I t is becoming clear that the apolipoprotein composition of a subspecies of HDL is particu- larly important in determining the metabolic fate of HDL par- ticles. Rader et al . (3) have demonstrated that HDL particles that contain the primary protein component of HDL, apoli- poprotein (apo) AI, without apoAII are catabolized more rapidly than those that contain both apoAI and apoAII in humans. In addition to the composition, the conformation of resident apo- lipoproteins has also been shown to modulate the metabolism of HDL particles. Spectroscopic studies reveal that the confor- mation of apoAI varies substantially when present on particles of different lipid compositions (4-81, and immunoreactivity studies show large differences in conformation between the lipid-free and lipid-bound forms of apoAI (9). Such conforma- tional changes in apoAI can affect the interaction of HDL par- ticles with 1ecithin:cholesterol acyltransferase (10). However, the effect of apolipoprotein conformation on the first step of reverse cholesterol transport, the removal of free cholesterol (FC) from the plasma membranes of nonhepatic cells by HDL, has not been studied in detail.

The exchangeable apolipoproteins associated with HDL are organized into repeated a-helical domains (11). Consequently, changes in HDL composition likely affect the conformation of these domains. An indication of the significance of apolipopro- tein conformation in FC efflux comes from studies showing that reconstituted particles that contain apoAI, apoAI1, or apoC promote efflux to different extents (12, 13). More recently, von Eckardstein et al . (14) showed that a mutant form of apoAI containing altered a-helical domains is 30% less efficient at removing FC from cultured cells than wild-type apoAI. These studies suggest that lipoproteins containing apoprotein helical domains can promote FC efflux, but the efficiency of the process may depend on the conformation of these helices on the lipopro- tein surface. Analysis of the amino acid sequence of apolipopro- teins shows that the helical domains are amphipathic in na-

* The abbreviations used are: HDL, high density lipoprotein; rHDL, reconstituted HDL; apoAI, apolipoprotein AI; BSA, bovine serum albu- min; DPPC, dipalmitoyl phosphatidylcholine; DMPC, dimyristoyl phos- phatidylcholine; FC, free cholesterol; MEM, minimal essential medium; SW, small unilamellar vesicle.

22975

22976 Apolipoprotein Structure and Efflux of Cellular Cholesterol

ture; the amino acid side chains that protrude from the helix backbone form polar and nonpolar faces on opposite sides of the helix (for a review, see Ref. 15). Amino acid sequences required to mimic the amphipathic nature of a-helical domains in ex- changeable apolipoproteins have been described (16), and pep- tide analogs of such domains have been synthesized that have similar lipid-binding properties as the intact apoprotein. These peptides can form discoidal complexes with dimyristoyl phos- phatidylcholine (DMPC) that are similar in size to complexes formed by apoAI (17). Our objective in this study was to test the hypothesis that the

conformation of the amphipathic a-helical domains of apoli- poproteins on HDL particles can affect the ability of the par- ticles to remove FC from cells. To simplify the problem of as- sessing the conformation of such domains in lipoprotein particles, we adapted the strategy of utilizing synthetic pep- tides with known helical content. Reconstituted discoidal HDL particles containing DMPC were prepared using three different synthetic peptides and human apoAI, and their relative effi- ciencies in removing cellular FC were measured; DMPC small unilamellar vesicles (SUV) were used as an apoprotein-free, reference acceptor particle. The three class A amphipathic hel- ical peptides used were originally described by Segrest and Anantharamaiah and colleagues (15) and are designated 18A, 37pA, and Ac-MA-NH,. 18A is the parent peptide having 18 amino acids in the sequence DWLKAFYDKVAEKLKEXF and is approximately 50% helical when bound to lipid (17). 37pA is a dimer of 18A molecules that are covalently joined by a proline residue so that there are two helical domains separated by the proline within the molecule (18). Ac-MA-NH, is an 18A mole- cule that is chemically blocked with an acetyl group at the amino terminus and a n amide group at the carboxyl terminus; this modification results in increased helicity (-75%) when the molecule is bound to lipid (19). These peptides were chosen to study the effect of increasing the length of the helix (comparing 18A to Ac-18A-NHJ and the effect of increasing the helix num- ber per molecule (comparing 18A to 37pA). The results of this study suggest that the presence of amphipathic helical seg- ments in acceptor particles is necessary for efficient efflux of cholesterol from cells. Furthermore, these helical segments ap- pear to interact with the cell surface and facilitate FC efflux by affecting the lipid packing properties of the membrane. In ad- dition, the conformation or arrangement of the helical seg- ments can modulate the ability of extracellular acceptor par- ticles to either sequester or retain cholesterol molecules.

EXPERIMENTAL PROCEDURES

Materials Dimyristoyl phosphatidylcholine (+99% grade) was purchased from

Avanti Polar Lipids, Inc. [1,2-3H1Cholesterol, [4-14Clcholesterol, C3Hldi- palmitoyl phosphatidylcholine, and [14C]formaldehyde were obtained from DuPont NEN. [4-'4C1Cholesterol was purified by preparative TLC (Silica Gel G developed with ethyl ether). [1,2-3HlCholesterol oleate was prepared from the labeled FC as previously reported (20). Fatty acid- free bovine serum albumin was obtained from Sigma. Minimal essential medium (MEM) and bovine calf and fetal serum were from Life Tech- nologies, Inc. Media were supplemented with 50 pg/ml gentamicin. All other reagents were analytical grade.

Methods Peptide Preparation and Purification of ApoAI-Amphipathic helical

peptides 18A and 37pA were synthesized by established solid phase techniques (21). The peptides were purified by high performance liquid chromatography and lyophilized. The lyophilized peptide was stored at -20 "C until used. Ac-MA-NH, was derived from the parent 18A mole- cule as described previously (19). ApoAI was obtained from ultracen- trifugally isolated, human HDL by anion exchange chromatography on Q-Sepharose and stored at -30 "C (22). Prior to use, purified apoAI was resolubilized in 6 M guanidine HCI and dialyzed extensively against Tris buffer (0.01 M Tris, 1.0 mM EDTA, and 0.15 M NaCl, pH 8.2). The

resulting protein solution was stored for no longer than 1 day at 4 "C. Reconstitution and Characterization of PeptidelDMPC Complexes-

All peptides and apoAI were combined with DMPC at a 2.51 (w/w) phospho1ipid:protein ratio. Lipid in chloroform was dried in borosilicate tubes under nitrogen and then under a vacuum for 1 h. Tris buffer was added to give a lipid concentration of 5 mg/ml and vortexed to form a cloudy solution of multilamellar liposomes. Lyophilized peptide was dissolved in Tris buffer (1.5-2.0 mg dry weighuml) and stirred for 1 h at 4 "C. From this point, apoAI and the peptides were treated the same, and they are both designated as "protein" unless otherwise noted. After determination of the protein concentration from the absorbance at 280 nm (Az8o) (4, 5, 17, 19), the appropriate amount of protein solution was added to the liposome solution, and the volume was adjusted with Tris buffer to give a final protein concentration of 1.0 mg/ml. The vortexed solutions were sonicated a t room temperature in a Branson 2200 bath sonicator for 45 min. The resulting proteidipid complexes were char- acterized by gel filtration chromatography on a (2.5 x 100 cm) Superose 6 column that was eluted with the Tris buffer at 0.7 mumin. Fractions of 3.0 ml were analyzed by phosphorus analysis (23) and respec- tively. The amounts of protein and phospholipid associated with each peak in the chromatogram were determined by integrating the peak area by computer (PeakFit, Jandel Scientific, San Rafael, CA). In ad- dition, the complexes were characterized by negative stain electron microscopy as described by Forte and Nordhausen (24). DMPC S W s were prepared by the method of Barenholz et al. (25).

To monitor the stability of reconstituted particles under cell culture conditions, some preparations were labeled with 14C-protein and 3H- phospholipid. Prior to combining the protein with DMPC, the proteins were trace-labeled by reductive methylation with [l4C1formaldehyde as described by Jentoft and Dearborn (26). Qpical specific activities were 0.2 pCi/mg apoAI and 0.5-0.9 pCi/mg peptide. The labeled proteins were combined with DMPC containing trace levels of 13HIDPPC; typical specific activities were in the range of 0.7-0.8 pCi/mg of phospholipid.

Efflux of Plasma Membrane Cholesterol-The mouse L-cell line was used for the measurements of cellular FC efflux because the cells grow readily in culture and do not synthesize FC. A deficiency of the sterol AZ4-reductase at the last step of FC synthesis results in truncation of the synthetic pathway at desmosterol(27). Consequently, these cells cannot dilute the labeled FC when depletion of cellular FC stimulates endog- enous sterol synthesis. Six days before the efflux measurement, the cells were plated in 6-well cell dishes (35 mm) a t 150,000 cells per well in bicarbonate-buffered MEM (MEM-bicarb) containing 10% fetal bo- vine serum and 50 &ml gentamicin. The cells were grown to subcon- fluency for 4 days in a humidified incubator at 37 "C with 5% CO,. Two days prior to the experiment, labeling media consisting of MEM-bicarb, 2.5% fetal bovine serum, 2.0 pCi/ml L3HIFC in ethanol, and 1.0 pdml of the Sandoz acetyl CoA:cholesterol acyltransferase inhibitor 58035 was applied to the cells for 24 h. The final ethanol concentration was always <0.2% (dv). Less than 3% of the total labeled FC was esterified by the cells during the labeling procedure and efflux time course. In the re- maining 12 h before the experiment, the cells were washed and incu- bated in a 1% solution of BSA in MEM-bicarb to equilibrate the label between the various cellular sterol pools. The efflux measurements were carried out in a humidified incubator with 5% CO, a t 37 "C. The medium used was a hybrid MEM-bicarb that contained 50 m~ HEPES. This prevented abrupt pH changes due to frequent removal of the cells from the incubator during the experiment. The efflux measurement was initiated by washing the cells with the MEM-bicarb-HEPES medium followed by application of 2 muwell of the test medium containing MEM-bicarb-HEPES, 0.5% BSA, 1.0 pg/ml 58035, and the DMPC/ peptide complex present a t the appropriate concentration. At each time point, 75 pl of medium was aspirated and passed through a 45-pm pore size, 96-well filter plate (Millipore multiscreen vacuum manifold (Mil- lipore)) to remove any floating cells. 50 pl of filtered medium was used for liquid scintillation counting. The total counts in the medium a t each time point were used to calculate the fraction of labeled FC that re- mained in the cells. After removal of efflux media, cell monolayers were washed with phosphate-buffered saline, and cellular lipids were ex- tracted in isopropyl alcohol for 24 h (28). In some experiments, cho- lesteryl methyl ether was added to the isopropyl alcohol extraction as an internal standard for mass quantitation of FC by GLC as described previously (20). The amount of radioactivity present in the cells a t the initiation of the efflux experiment was determined from identically labeled cells that were not exposed to test medium. L-cell viability was monitored throughout the incubation with regular inspections under a light microscope. In addition, the cell protein content was determined by the Markwell modification of the Lowry protein assay (29); no de- crease in cell protein was found between control and acceptor particle-

Apolipoprotein Structure and Efflux of Cellular Cholesterol 22977

exposed cells during the efflux experiments. Typical protein values were between 400 and 500 pg of cell protein per 35-mm well. Efflux of FC that originated in the lysosomal compartment was measured as described in detail by Johnson and co-workers (20, 30).

Data Analysis-The transfer of FC between cells and extracellular acceptors has been shown to be a bidirectional process in which cell and acceptor take up and release FC concomitantly (31). The kinetic anal- ysis assumes a closed system in which FC exists in one of two kinetic pools, either the cellular FC pool or the acceptor FC pool. Two kinetic pools of releasable FC have been reported in mouse L-cell membranes (32, 33). In our experiments, the faster effluxing pool has a half-time (t,) of about 2 h, whereas the slower pool has a t , of 40 to 50 h when incubated with apoAI-containing rHDL.' Because the time courses in this study are limited to periods below 6 h, the release of FC from the slow pool is negligible, and monoexponential kinetics are observed in- dicating that the cellular FC is essentially released from a single kinetic pool under these conditions. The experimental system meets the as- sumption of a closed system for FC due to the inhibition of the esteri- fication of FC label by the acetyl-CoAcholesterol acyltransferase inhib- itor and the inability of the cells to synthesize FC de nouo.

A single exponential equation that describes a bidirectional transfer between two pools was fit to the experimental data by computer (Inplot, GraphPad Inc.). The equation was of the form: Y = H,e@ + H , which was shown by comparison of the F statistic (34) to describe the efflux time course better than a biexponential equation at all acceptor concentra- tions tested. Y represents the fraction of radiolabeled FC remaining in the cells, t is the incubation time in hours, H , is a pre-exponential term that reflects the fraction of cellular FC that exists in the medium at equilibrium, g is the sum of the rate constants for efflux (k,) and influx (kt), and H, is a constant that represents the fraction of labeled cell FC that remains associated with the cells at equilibrium due to a constant retrograde flux of FC from the extracellular acceptor to the cell. HI, g , and Hz are variables that the computer can modify to fit the equation to the data. The rate constant for the efflux process (k , ) is the product of H , and g. A t,, value in hours can then be calculated as follows: t,,, = In 2/k,.

RESULTS Characterization of PeptideIDMPC Complexes-Four par-

ticles were reconstituted that contained either apoAI or each of the three peptides 18A, Ac-MA-NH,, and 37pA. Previous stud- ies using the same peptides and phospholipid have shown that the peptides make disc-shaped particles that are between 8 and 12 nm in diameter (17-19). Under the sonication conditions used, the three peptides and apoAI readily solubilize multila- mellar DMPC liposomes into small particles that do not scatter light and that form uniform rouleux when visualized by nega- tive-stain electron microscopy. Meaningful comparisons of cell FC emux rates to the apoAI and the peptide complexes depend on reliably establishing their composition, size, and homogene- ity. The sizes of peptide-containing particles cannot be deter- mined readily by nondenaturing gradient gel electrophoresis because of the low net charges associated with the peptides and the lack of appropriately charged molecular weight standards. In view of this, a calibrated (2.5 x 100 cm) Superose-6 gel filtration column was used to characterize the products of peptideDMPC reconstitutions. Fig. 1 shows the protein and phospholipid elution profiles of the four complexes. The ab- sence of a phospholipid peak around the void volume (70 ml) shows that the peptides and apoAI can completely solubilize DMPC liposomes. The peptides and apoAI form complexes con- taining phospholipid and protein with an elution volume near 105 ml. Additional peaks are detectable at or near the total volume of the column (165 ml) that represent lipid-free peptide or apoAI in monomeric or low molecular weight oligomeric forms. The apoAI/DMPC system contains the lowest amount of free protein; the 37pA and Ac-18A-NHZ systems have larger amounts, and the largest amount of free protein occurs in the 18AIDMPC mixture. The small amounts of phospholipid ap-

W. S. Davidson, G. H. Rothblat, and M. C. Phillips, unpublished results.

100

80

60 60

40 40

20

0 100

80

80

40

20

- 1 - 2 0

I I i - I I I I I 0 - - 100

80

60

- - 40

Ac-IM-NH, - -

- -

-

0 I

20 -

0. loo

80 - Apo AI

- 100

80 - -

20 20

0 - 1 ' 1 . ' I

0 25 50 75 100 125 150 175 200 0 . . . '

FIG. 1. Protein and phospholipid elution profiles of synthetic peptidelDMPC and apoAvDMpC reconstituted complexes sub- jected to gel filtration chromatography. A 2.5 x 100-cm Superose 6 (Pharmacia) gel filtration column was used to analyze solutions of re- constituted particles that were combined at a 2 5 1 phospholipid to protein (w/w) ratio. Fractions of 3.0 ml were collected, and the phos- pholipid in each fraction was quantified by analysis of phosphorus (dot- ted line), and the protein was quantified by absorbance at 280 nm (Am, solid line). The values for each component are expressed as a percent-

The void volume of the column is 70 ml, and the total volume is 165 ml. age of the maximum value for that component obtained during the run.

pearing in the total volume with the 18A and 37pA systems may represent small complexes containing a few phospholipid and peptide molecules. Table I lists the initial compositions used in the reconstitutions of all four particles and the percent of initial phospholipid and peptide that is associated with the 105-ml peak. In the 37pA, apoAI, and Ac-18A-NH2 mixtures, most of the peptide and phospholipid is associated with the major complex. In contrast, the 18Amixture has only a quarter of the total peptide associated with the major complex. This is consistent with surface chemistry studies showing that 18A has a lower affinity for phospholipid surfaces than the other pro- teins (35). It should be noted that the amount of free peptide is dependent on the particle concentration; the compositions pre- sented in Table I relate to the concentrations of complexes eluting from the column only. The protein concentration when the sample elutes from the column is typically 20 times more dilute than the initial sample and about 5 times more dilute than when the sample is diluted into cell medium for choles- terol efflux measurements. Therefore, the final compositions shown in Table I are not necessarily the same as those present in the extracellular medium during an efflux experiment. How- ever, at a given concentration, there is a higher fraction of lipid-free 18A than ofAc-18A-NH, and 37pA, the concentration of lipid-free apoAI is the least (Fig. 1).

Table I also shows estimated particle sizes as measured by gel filtration chromatography and electron microscopy as well

22978 Apolipoprotein Structure and Emux of Cellular Cholesterol

TABLE I Characterization of discoidal peptidelDMPC acceptor particles

Peptide Initial (DMPC:peptide) % peptide % DMPC Final (DMPC:peptide)' Diameter Molar composition

Mass ratio Mole ratio in complex" in comp'exb Mass ratio Mole ratio Columnd EM' (DMPC:peptide)'

nm 18A 2 .51 8:1 24 88 9.4:1 30:l 10.9 12.3 -c 2.1 350:20

8: I 77 100 3.3:1 37pA 2 .51 16:l

11:l 11.0 12.5 -c 2.1 75

310:30 75 2 5 1 16:l 10.0 11.5 * 2.4 250:15

ApoAI

Ac-18A-NH2 2.5:1

2.5:1 1OO:l 98 79 2.0:1 81:l 11.0 11.3 -c 2.1 280:3

Determined by Gaussian function fitting of protein elution profile from a Superose 6 gel filtration column monitored by A,,, (e.g. Fig. 1). Determined by analysis of phospholipid profile of same column run as in (a) monitored by analysis of phosphorus.

e Estimated composition of peak that contains peptideDMPC disc. Calculated from the elution volume from a calibrated Superose 6 gel filtration column (20.7 nm (S.D.)).

e Average major diameter of 100 particles determined from negative staining electron microscopy (-cS.D.). Determined from partial specific volume calculations using the particle diameter as determined by the gel filtration column and assuming a

bilayer thickness of 4.5 nm. The partial specific volumes of the DMPC and protein were assumed to be 0.971 mug and 0.714 mug, respectively.

as the estimated phospholipid and protein molar composition of each particle. Both methods show that the complexes are of similar size; this is consistent with previously published results for 18A and 37pA (19, 21).

Stability of PeptidelDMPC Complexes under Cell Culture Conditions-The stabilities of the peptideDMPC particles in cell culture conditions were determined. The extracellular con- centrations of both the peptide and phospholipid components over the course of a typical efflux experiment were monitored by liquid scintillation counting using 14C reductively methyl- ated peptide and trace amounts of r3H]DPPC. It has been shown that reductive methylation of lysine residues does not alter the lipid-binding properties of the peptides studied.z Fur- thermore, labeled peptides form DMPC complexes of similar size and composition as unlabeled peptides (data not shown). The dual-labeled proteidlipid complexes were applied to mouse L-cells under FC efflux conditions, and the levels of both trac- ers were monitored throughout a 6-h incubation. The percent- ages of peptide counts lost from the medium during the experi- ment were: apoAI, 2%; 37pA, 3%; Ac-18A-NHz, 0%; and MA, 4.2%. These minimal losses indicate that the peptide concen- trations remained constant in the medium during the experi- ment. The percentages of phospholipid counts lost during the experiment were: apoAI, 4%; 37pA, 5%; Ac-NA-NH,, 5%; and 18A, 25%. Gel filtration chromatography of the peptideflipid complexes before and after exposure to the cells showed no significant change in particle size during the incubation (data not shown). These data show that the apoAI, 37pA, and Ac- 18A-NH2 complexes with DMPC were quite stable when incu- bated with cells. However, the 18A/DMPC particle composition changed under the efflux experiment conditions. It follows that the lSA/DMPC complex is a complicated model for discoidal HDL particles and care needs to be taken in interpreting re- sults obtained with this system in FC efflux experiments.

Eflux of Plasma Membrane and Lysosomal Cholesterol-FC efflux from L-cells that had their plasma membranes labeled with r3H]FC was measured for each of the three peptide and apoAI/DMPC particles. Initial experiments compared each par- ticle at a concentration of 100 pg of DMPCIml of medium. All complexes removed a significant amount of cholesterol when compared to control medium (Fig. 2). GLC analysis demon- strated that the fraction of initial FC mass remaining in the cells at the end of such incubations closely paralleled the frac- tional retention of FC tracer for each type of particle (data not shown), indicating that the radiolabel properly traced the FC mass movements, Typical initial cellular FC masses ranged from 35-45 pg of FC/mg of cell protein. In order of efflux effi- ciency (from lowest to highest), the complexes have the follow- ing half-times (t1,,) for efflux: apoAI, 16.1 2 3.1; 37pA, 12.0 * 1.2; Ac-18A-NH2, 9.2 z 4.6; and 18A, 6.1 2 0.1 h. No significant differences could be detected between apoAI and 37pA or be-

1.00

0.95

0.90

0.85

0.80

0.75

0.70

0.65 1 0.60 1 I I I I I I

0 1 2 3 4 5 6

Incubation Time (h)

FIG. 2. Time course of L3Hlfree cholesterol efflux from mouse L-cell fibroblasts to synthetic peptide/DMPC and apoAVDMPC reconstituted complexes and control medium. Mouse L-cell fibro- blasts that had been trace-labeled with 13H]free cholesterol were plated in 35-mm tissue culture wells and incubated for 6 h with 2.0 ml of test medium containing 0.5% BSA and 1 pg/ml Sandoz compound 58035, alone (control medium) (8) or with 100 pg of DMPC/ml present in complexes with apoAI (D), 37pA (a), Ac-NA-NH, (O), and 18A/DMPC

lesterol that remained in the cell at the designated times. Each point (0). The vertical axis indicates the fraction of initial labeled free cho-

represents the mean of 12 cell wells except for the control medium in which each point represents 3 cell wells. The error bars represent 1 S.D. All curves were obtained by fitting the entire time course to the model for tracer equilibration between two pools.

tween 18A and Ac-MA-NH,. However, efflux of FC is signifi- cantly faster to complexes containing the latter two peptides although all four types of acceptor clear cellular FC quite effi- ciently. The equilibrium distribution of the FC between the acceptor pool and the cellular pool was similar for the four types of particles; the fractions of FC remaining in the cells at equilibrium were as follows: apoAI, 0.63 2 0.07; 37pA, 0.69 -c 0.05; Ac-18A-NH2, 0.57 2 0.04, and 18A, 0.59 -c 0.08. The initial cellular content of the other major L-cell sterol, desmosterol, was about 7 pg/mg of cell protein, accounting for 1 5 2 0 % of the total unesterified sterol pool. The availability of desmosterol for efflux was evident from the fact that desmosterol levels de- creased during incubation with the acceptor particles, but this decrease was less than that of FC. This effect was probably due to endogenous desmosterol synthesis during the experiment that partially replaced released desmosterol.

The abilities of the peptide- or apoAI-containing particles to induce efflux of intracellularly derived lysosomal FC were also compared. To achieve this, the lysosomal pool of FC was labeled via reconstituted LDL that contained [3H]cholesterol oleate (20). At a concentration of 100 pg of DMPC/ml, the particles exhibited the same order of efflux efficiency as shown in Fig. 2 (data not shown). These results demonstrate that the peptide-

Apolipoprotein Structure and E f f u x of Cellular Cholesterol 22979

140

120

100 5 .- E 80

I" x 60

40

20

0

I

t T

4 T \

Lk- L t n - I I I

0 L

50 100 150 200 ' 450 500

Concentration (pg PUml)

FIG. 3. The concentration dependence of the half-times for cholesterol efflux from mouse L-cell fibroblasts to synthetic peptide/DMPC and apoAvDMPC complexes. The incubation con- ditions were the same as those for Fig. 2 except that the acceptor concentration was vaned as shown. The acceptors shown are ApoAV DMPC (W), 37pA/DMPC (01, Ac-18A-NHflMPC (O), and 18A/DMPC (0). The t , was determined from rate constants derived from fitting 6-h time course data at each acceptor concentration to the model for tracer equilibration between two pools. Each point represents 6 cell wells, and the error bars represent 1 S.D. The values shown are corrected for the FC efflux to control medium.

and apoAI-containing complexes have the same relative ability to remove plasma membrane-derived FC as lysosomally de- rived FC.

To determine the FC efflux to each of the complexes over a range of concentrations, 6-h time courses were obtained at five different concentrations of acceptor particles. To examine the dependence of efflux on the presence of an apoprotein or model peptide, protein-free DMPC small unilamellar vesicles (SUV) were also utilized. Gel filtration chromatography demonstrated that the SUV did not fuse into multilamellar vesicles when exposed to the cells. Fig. 3 compares the t , , of FC efflux to the protein-containing acceptor particles. At acceptor concentra- tions below 100 pg of DMPC/ml, clear differences existed be- tween the peptide particles as a group and the apoAI particle with the peptide/DMPC complexes being the most effective ac- ceptors. The t,, of efflux was dependent on the particle concen- tration at low acceptor concentrations. However, when the ac- ceptor concentration exceeded about 200 pg of DMPC/ml, the t , , became largely independent of the concentration; the pep- tide- and apoAI-containing particles converged to a common tlI2 value of 6-10 h (Fig. 3). When the apoAI-containing particles were compared to apoprotein-free DMPC SUV (Fig. 4), it is clear that at high concentrations, the DMPC SUV exhibited a much longer t , value (about 50 h) than any of the protein- containing complexes. This observation persisted even when SUV concentrations of up to 10 mg/ml DMPC were used. The initial rates of cellular FC efflux as a function of acceptor par- ticle concentration were plotted on a double reciprocal plot (Fig. 5) t o more accurately determine the maximal efflux rate that can be attained with these particles. The intercept on they axis gives the reciprocal of the maximum velocity (V,,) while the x-intercept gives the reciprocal of the EC,, which is the effective concentration of acceptor required to achieve half the Vm,. Fig. 5 shows that the V,, of about 10% cell FC released per h is similar for the peptide- and apoAI-containing acceptor par- ticles, but V,,, is much lower (1-2% per h) for the DMPC SUV. To eliminate the problem of experimental error at lower con- centrations having an unduly large effect on the final regres-

200 -

175 -

150 - - 5 E

3 75 FI -

125 -

E a 100 I

- .-

50 -

25 -

0 -

T

0 2000 4000 6000 8000 10000

Concentration (pg PUml)

FIG. 4. The concentration dependence of the half-times for cholesterol efflux from mouse L-cell fibroblasts to apoAVDMPC complexes and DMPC SW. The incubation conditions were the same as those for Fig. 2 except that the acceptor concentration was vaned as shown. The acceptors shown are apoAI/DMPC (W) and DMPC SUV ( 0 ). Each point represents 3 cell wells, and the error bars represent 1 S.D. The values shown are corrected for the FC efflux to control medium.

I 0.12 7

0.09

~ 0.06

200 0.03

0.00 //p 0 2 4 6 8 1 0 1 2 1 4

v / [A] (x 1 0-4)

100

-0.015 -0.010 -0.005 0.000 0.005 0.010 0.015 0.020

l/Acceptor Concentration (ug DMPClml)

terol efflux from mouse L-cell fibroblasts against the DMPC FIG. 5. Double reciprocal plot of the initial velocity of choles-

concentration of the acceptor particles. The data shown are from the same experiments shown in Fig. 3 with the inclusion of DMPC SUV data (Fig. 4) over the same concentration range. The vertical axis shows the reciprocal of initial velocity ( u ) of efflux for the experiments shown in Fig. 3. The initial velocity is defined as the portion of the time course in which the percentage of FC remaining in the cells decreases linearly with respect to time (the first 3.5 h). The horizontal axis shows the reciprocal of the acceptor concentration. The acceptors shown are DMPC SUVs ( 0 ) and discoidal complexes of DMPC with apoAI (m), 37pA (O), Ac-18A-NH2 (01, and 18A (0). The lines are derived by linear regression of the data. The inset shows the same data plotted as pro- posed by Hofstee (36). The vertical a i s is the initial velocity, and the horizontal axis is the initial velocity divided by the concentration of acceptor particle in terms of micrograms of DMPC/ml. The values shown are corrected for the FC efflux to control medium.

sion line, the V,, and EC,, terms were calculated as proposed by Hofstee (36) (Fig. 5, inset). This analysis normalizes each initial velocity for the acceptor concentration, eliminating in- appropriate weighting in the double reciprocal analysis. The results of this analysis for each particle described in Fig. 3 as well as DMPC SUV over the same concentration range are listed in Table 11. A one-way analysis of variance determined that a significant difference existed among the V,, values, and a Tukey-Kramer test showed that the DMPC SUV have a sig-

22980 Apolipoprotein Structure and Eflux of Cellular Cholesterol

TABLE I1 Comparison of V,, and EC,, values for DMPC-containing

acceptor particles

Acceptor particle: vmaXa %,’

% cell FCIh pg DMPCIml (particleslml) x IO” 18A/DMPC 12 f 2 Ac-18A-NHflMPC 10 f 2

80 f 5 110 -c 20

2.0 f 0.1 3.1 f 0.5

37pA/DMPC 10 f 1 ApoAI/DMPC

140 f 10 11 f 2 270 f 60

4.8 f 0.3

DMPC SUV 8.3 f 1.9

1 f I d 100 -c 20 0.3 f 0.06

ceptor as calculated proposed by Hofstee (see “Results”). Since the cell Maximum rate of efflux (2S.D.) at saturating concentrations of ac-

FC content was routinely 40 f 5 pg of FClmg of cell protein, a V,, value of 10%/h corresponds to a maximal efflux of about 4 pg of FClmg of cell protein.

* Effective concentration of acceptor particle (phospholipid) at which efflux is 50% of the V,,,,.

Effective concentration of acceptor particle (particle number) at which efflux is 50% of the V,=. The particle number at a given phos- pholipid concentration was calculated using the phospholipid composi- tion of each particle listed in Table I.

Statistically significant difference from all other acceptors as deter- mined by ANOVA and a Tukey-Kramer comparison test (p < 0.001).

nificantly lower V,, than all of the peptide-containing acceptor particles ( p < 0.001). The V,, values for the peptide-containing particles are not significantly different from each other ( p > 0.05). This indicates that protein-containing particles are more efficient than protein-free SUVs at removing cholesterol at saturating concentrations of acceptor. Table I1 also shows that the EC,, is higher for the apoAVDMPC particle, whereas the DMPC SUV and the peptide-containing complexes are some- what similar when expressed in terms of phospholipid concen- tration. Similarly, when the EC,, terms are normalized for particle number per ml (Table 11), it is clear that more of the apoAVDMPC particles are required to achieve half of the V,, than for any of the peptideiDMPC complexes. As expected, less SUV are required to reach this value because there is about a 10-fold higher phospholipid content per SUV than for the discs. At a given concentration of acceptor particle phospholipid, the relative particle numbers (normalized to the apoAI particle) for each complex are: apoAI, 1.0; 37pA, 1.1; Ac-NA-NH,, 0.9; 18A, 0.8; and DMPC SUV, 0.1. This demonstrates that the EC,, values are still different for the protein-containing particles when expressed in terms of particle concentrations. A direct comparison of the EC,, of the SUV to the discs is not straight- forward because the V,, values differ.

DISCUSSION

The transfer of cellular FC to HDL has been studied inten- sively, and several mechanisms have been proposed to account for the release of the various cellular pools. There is general agreement that the plasma membrane pool of FC leaves the cell via the aqueous diffision mechanism (37,381. The FC transfer by this mechanism occurs by a two-step process that involves the desorption of a FC molecule from the plasma membrane and subsequent diffusion through the aqueous phase until it collides with and is absorbed by a phospholipid-containing ac- ceptor particle (31). At dilute acceptor concentrations, the rate of cellular FC transfer to an extracellular particle depends on the rate of 1) desorption from the membrane and 2) diffusion to and sequestration into the acceptor particle. In contrast, under conditions of high acceptor concentration, the rate of FC ab- sorption into the acceptor particle is fast relative to the rate of desorption from the cell, due to the excess of acceptor particles present. Therefore, the rate of desorption from the plasma membrane becomes rate-limiting for the efflux process. Like plasma membrane FC, lysosomally derived FC leaves the cell by the same mechanism after a brief lag period during which

the product of cholesteryl ester hydrolysis is transported to the plasma membrane (20). It has been established that direct contact between stable donor and acceptor, phospholipid-con- taining particles in dilute solution is not required for transfer of cholesterol molecules by the aqueous diffusion mechanism (for a review see Ref. 31). If this situation applies to the cell surface, then in its simplest form, the aqueous diffusion mechanism does not require contact of either HDL particles or apolipopro- tein molecules with the plasma membrane. As discussed in detail below, FC efflux from mouse L-cells occurs via aqueous diffusion but the present data (Figs. 2-5) indicate that, when apoproteins are present on lipoprotein particles, the mecha- nism of efflux is more complex than the simple aqueous diffu- sion model discussed above.

Apolipoprotein-Plasma Membrane Interactions and Choles- terol Efflux-The comparison of V,, values for different accep- tor particles offers insight into possible mechanistic differences in the way that they remove cellular FC. According to the aqueous diffusion mechanism, the magnitude of desorption of FC molecules from the plasma membrane limits the overall rate of efflux at high acceptor particle concentrations. It follows that any acceptor particles that act by this mechanism should give a common V,, that directly reflects the rate of FC de- sorption from the membrane of the particular cell type. How- ever, Table I1 shows a clear difference in the maximal rate of FC efflux to high concentrations of apolipoprotein-containing ac- ceptor particles and to one that contains only lipid (DMPC SUV). This indicates that the FC efflux in the case of the apolipoprotein-containing particles is more complex than simple aqueous diffusion. The data suggest that when there are apolipoprotein-containing particles in the extracellular me- dium there is an interaction with the plasma membrane that increases the characteristic rate of desorption of FC molecules from the cell surface. Several apoprotein-membrane interac- tions have been proposed and they fall into three general cat- egories. The first encompasses the specific interactions of HDL apoproteins with cell surface binding sites. Current reports suggest that the binding of HDL to these sites primarily medi- ates the movement of newly synthesized sterols to the plasma membrane via second-messenger systems (39). The second cat- egory involves a less specific interaction of apoproteins or re- gions of apoproteins with lipid domains within the plasma membrane. For example, it has been proposed that helical re- gions of lipoprotein-bound apoAI may transiently interact with certain domains of the cell surface and thereby decrease the distance that a cholesterol molecule must diffise to be incor- porated into the acceptor particle (40). The third category re- lates to the interaction of lipoprotein-unassociated or “free” apoproteins with the cell surface. Work in this area has dem- onstrated the release of phospholipid and FC to extracellular, lipid-free, apoAI and the subsequent formation of HDL-like particles (41, 42).

The fact that the 18- or 37-residue peptide-containing par- ticles demonstrate a similar V,, to the apoAI-containing par- ticles, despite the lack of sequence homology to any region of the apoAI sequence of 243 amino acid residues, confirms pre- vious reports that the particles do not interact with a specific cell surface site (43). I t follows that the translocation of lyso- somally derived FC is also not dependent on such a site since the peptideiDMPC particles also promote efficient efflux from this pool. Therefore, the observed increase in FC efflux to ap- oprotein-containing particles most likely results from a rela- tively nonspecific interaction of the apoprotein with the plasma membrane. Such an interaction could conceivably occur in two ways. The first is that the apoprotein could remain lipoprotein- bound and still interact with the cell surface. ApoAI has been

Apolipoprotein Structure and

postulated to contain a lipophilic “hinge” domain that can “loop out” of a lipoprotein particle (15, 44); such a domain might transiently anchor an HDL particle within 5 to 30 A of the cell surface, thereby decreasing the diffusion distance for outgoing FC molecules. The layer of interfacial water between two lipid surfaces that are in such close proximity is significantly less polar than bulk phase water (45). Such an environment would further facilitate the transfer of hydrophobic molecules across this space (40). Any interaction of amphipathic helices with the phospholipid bilayer of the plasma membrane may also perturb phospholipid-cholesterol molecular interactions so that desorp- tion into the aqueous phase is facilitated (46). The peptides studied here contain one or two helical segments that are quite short (an 18-residue a-helix is 27 A long) and therefore prob- ably do not contain a domain that is capable of interacting simultaneously with two lipid surfaces. This suggests that such a hinge domain is not a requirement for efficient efflux. How- ever, it is possible that the presence of amphipathic, helical proteins in the extracellular medium modulates FC eMux by another mechanism. The second type of interaction involves the lipoprotein-unassociated protein. Table I indicates that the peptides and apoAI can exist in a lipid-free state at a concen- tration that depends on the lipid-binding properties of the pro- tein. This opens the possibility that the free protein can either promote FC and phospholipid efflux by itself (in addition to that removed by the intact lipoprotein particle) or that it can bind to the cell surface and modify the lipid packing character- istics of the membrane in such a way as to increase the de- sorption rate of FC molecules. Separate experiments under the same conditions have shown that FC emux to levels of lipid- free apoAI and the peptides that are present in the complexes reported here (see Fig. 1) is less than 1% of cellular FC.3 This indicates that, while significant FC efflux to free apoproteins does occur, the contribution from this pathway to efflux is not sufficient to account for the large difference in acceptor effi- ciency that is seen between the DMPC S W and the DMPC/ protein complexes. This difference may be due to the transient binding of the free apoproteins to the cell surface with induc- tion of a local disturbance in the lipid packing order of the cell membrane. Previous studies from this laboratory have shown that binding of apolipoprotein molecules to a phospholipid bi- layer can facilitate the desorption of FC molecules into the aqueous phase (46). A similar effect in the plasma membrane of the L-cells would explain the increased FC efflux to the apopro- tein-containing acceptor particles.

Further investigation is required to finally distinguish be- tween the above possibilities, but the nature of the peptides used in this study provides useful insights. It is clear that even a single class A amphipathic helix is sufficient to interact with the membrane to promote efficient efflux. However, neither the length of the helix (18A uersus Ac-18A-NHJ nor the number of helices per molecule (18A versus 37pA) is critical for this inter- action. It also seems that the lipid affinity of the apoprotein does not correlate with the ability to promote efflux. The rank- ing of the surface activities of the proteins (from highest to lowest) is 37pA > Ac-18A-NH, > apoAI > 18A (19, X ) , but the V,, values for eMux are the same (Table 11). Future experi- ments will take further advantage ofthe variations in structure possible with synthetic peptides; the behavior of various classes of amphipathic helix (15) will be investigated.

Effects of Acceptor Particle Structure on Cholesterol Eflux-At Iow acceptor concentrations, the rate of FC accumu- lation in the acceptor particles is affected by the on- and off- rates of FC molecules at the acceptor particle surface. Figs. 3 and 4 show that at low concentrations of acceptor phospholipid

P. Yancey and G. H. Rothblat, unpublished results.

Efflux of Cellular Cholesterol 22981

( i e . 50 pg of DMPC/ml) the efflux efficiencies of the various particles are DMPC SUV < apoAVDMPC < peptide/DMPC. The observation that the vesicles are the least efficient under these conditions is consistent with previous studies from this labora- tory (13) which concluded that the frequency of collisions be- tween dissolved FC molecules and acceptor particles is in- versely proportional to the particle radius and directly proportional to the number of acceptor particles present. SUVs have over twice the diameter of the peptide or apoAI discs, and there are about 10-fold more apoAI particles than SUVs (47) at the same phospholipid concentration (Table 11). However, this argument does not explain why the peptide/DMPC discs are more efficient acceptors than the apoftllDMPC discs (Fig. 3) at low concentrations. These particles are all of similar size and, in comparison to the SWs, there are similar numbers of par- ticles present at a given phospholipid concentration. This indi- cates that the apoAI and peptide complexes collide with an approximately equal number of FC molecules (diffusing in the aqueous space) in a given time interval. The peptide/DMPC discs must either more rapidly incorporate FC molecules (i.e. the on-rate is higher) or retain FC molecules longer once they have been incorporated into the particle (i.e. the off-rate is lower). Differentiation of these two possibilities must await bidirectional flux studies using acceptor particles that initially contain FC. However, it is clear that the way that the am- phipathic, a-helical segments are arranged in an acceptor par- ticle can modulate FC efflux from cells.

Acknowledgments-We thank Christine Ackerman, Faye Baldwin, Sheila Benowitz, and Margaret Nickel for expert technical assistance.

1.

2. 3.

4.

5.

6.

7.

8.

9.

10. 11.

12.

13.

14.

15.

16.

17.

18.

19.

20.

21. 22. 23.

24. 25.

REFERENCES Johnson, W. J., Mahlberg, F. H., Rothblat, G. H., and Phillips, M. C. (1991)

Rader, D. J., Castro, G., Zech, L. A., Fruchart, J., and Brewer, H. B. (1991) J. Fievet, C.. and Fruchart, J. C. (1991) Diabetes Metab. Reu. 7, 155-162

Sparks, D. L., Phillips, M. C., and Lund-Katz, S. (1992) J. Biol. Chem. 267,

Biochim. Biophys. Acta 1086, 273-298

Lipid Res. 32, 1849-1859

25830-25838

25839-25847 Sparks, D. L., Lund-Katz, S., and Phillips, M. C. (1992) J. B i d . Chem. 267,

Sparks, D. L., Davidson, W. S., Lund-Katz, S., and Phillips, M. C. (1993) J. Biol. Chem. 268,23250-23257

268,4798-4805

Chem. 265,22123-22129

32,6477-6484

Leroy, A,, Toohill, K. L. H., Fruchart, J. C., and Jonas, A. (1992) J. Biol. Chem.

Jonas, A., Wald, J. H., Toohill, K. L., Krul, E., and Kezdy, K. E. (1990) J. Biof.

Calabresi, L., Meng, Q. H., Castro, G. R., and Marcel, Y. L. (1993) Biochemistry

Jonas, A., Kbzdy, K. E., and Wald, J. H. (1989) J. Biol. Chem. 264,48184824 Segrest, J. P., Jackson, R. L., Morrisett, J. D., and Gotto A. M. (1974) FEBS

Mahlberg, F. H., Glick, J. M., Lund-Katz, S., and Rothblat, G. H. (1991) J. Biof. Lett. 38, 247-253

DeLamatre, J., Wolfiauer, G., Phillips, M. C., and Rothblat, G. H. (1986) Chem. 266, 19930-19937

von Ekardstein, A., Castro, G., Wybranska, I, Theret, N., Duchateau, P, Biochim. Biophys. Acta 876,419428

Duverger, N, Fruchart, J. C., Ailhaud, G., and Assmann, G. (1992) J. Biol. Chem. 268,26162622

Segrest, J. P., Jones, M. K., De Loof, H., Brouillette, C. G., Venkatachalapathi, Y. V., and Anantharamaiah, G. M. (1992) J. Lipid Res. S3, 141-166

Kanellis, P., Romans, A. Y., Johnson, B. J., Kercret, H., Chiovetti, R., Allen, T. M., and Segrest, J. P. (1980) J. B i d . Chem. 266, 11464-11472

Anantharamaiah, G. M., Jones, J. L., Brouillette, C. G., Schmidt, C. F., Chung, B. H., Hughes, T. A., Bhown, A. S., and Segrest, J. P. (1985) J. Biol. Chem. 260, 10248-10255

Chung, B. H., Anantharamaiah, G. M., Brouillette, C. G., Nishida, T., and

Venkatachalapathi, Y. V., Phillips, M. C., Epand, R. M., Epand, R. F., Tytler, E. Segrest, J. P. (1985) J. Biof. Chem. 260, 1025&10262

Johnson, W. J., Chacko, G. K., Phillips, M. C., and Rothblat, G. H. (1990) J. M., Segrest, J. P., and Anantharamaiah, G . M. (1993) Proteins 16,349-359

Anantharamaiah, G. M. (1986) Methods Enzymol. 128,627447 Biol. Chem. 266,5546-5553

Weisweiler, P., Friedl, C., andungar, M. (1987) Clin. Chim. Acta 189,249-254 Sokoloff, L., and Rothblat, G. H. (1974) Proc. SOC. Exp. Bid. Med. 146, 1166-

Barenholz, Y., Gibbes, D., Litman, B. J., Goll, J., Thompson, T. E., and Carlson, Forte, T. M., and Nordhausen, R. W. (1986) Methods Enzyrnol. 128,442457 1172

F. D. (1977) Biochemistry 16,2806-2810

22982 Apolipoprotein Structure and Effzux of Cellular Cholesterol 26. JentoR, N., and Dearborn, D. G. (1983) Methods Enzymol. 91,570-579 27. Rothblat, G. H., Burns, C. H., Conner, R. L., and Landrey, J. R. (1970) Science

28. Johnson, W. J., Bamberger, M. J., Latta, R. A,, Rapp, P. E., Phillips, M. C., and 169,880-882

29. Markwell, M. A,, Haas, S. M., Bieber, L. L., and Tolbert, N. E. (1978) Anal. Rothblat, G. H. (1986) J. Biol. Chem. 261, 5766-5776

30. Krieger, M., Brown, M. S., Faust, J. R., and Goldstein, J. L. (1978) J. Biol. Biochem. 87,206-210

31. Phillips, M. C., Johnson, W. J., and Rothblat, G. H. (1987) Biochim. Biophys. Chem. 263,4093-4101

32. Schroeder, F., Jefferson, J. R., Kier, A. B., Knittel, J., Scallen, T. J. , Wood, W. Acta 908,223-276

33. Mahlberg, F., Davidson, W. S., Bernard, D., and Rothblat, G. H. (1993) in G., and Hapala, I. (1991) P m . Soc. Exp. Biol. Med. 196,235-252

Drugs Afecting Lipid Metabolism (Catapano, A. L., Gotto, A. M., Smith, L. C., and Paoletti, R., eds) pp. 493-498, Kluwer and Lorenzini, Milan

34. Motulsky, H. J., and Rasnas, L. A. (1987) FASEB J. 1,365-374 35. Garber, D. W., Venkatachalapathi, Y. V., Gupta, K. B., Ibdah, J., Phillips, M. C.,

Hazelrig, J. B., Segrest, J. P., and Anantharamaiah, G. M. (1992) Arterio- scler. Thromb. 12, 886-894

36. Christensen, H. N., and Palmer, G. A. (1974) Enzyme Kinetics, pp. 50-60, W. B.

37. Rothblat, G. H., and Phillips, M. C. (1991) Curr. Opin. Lipidol. 2,288-294 38. Hokland, B. M., Slotte, J. P., Bierman, E. L., and Oram, J. F (1993) J. Biol.

39. Oram, J. F. (1990) Curr. Opin. Lipidol. 1, 416-421 40. Rothblat, G. H., Mahlberg, F. H., Johnson, W. J., and Phillips, M. C. (1992) J.

41. Hara, H., and Yokoyama, S. (1990) J. Biol. Chem. 266,3080-3086 42. Bielicki, J. K., Johnson, W. J., Wienberg, R. B., Click, J. M., and Rothblat, G.

43. Karlin, J. B., Johnson, W. J., Benedict, C. R., Chacko, G. K., Phillips, M. C., and H. (1992) J. Lipid Res. 33, 1699-1709

44. Cheung, M. C., Segrest, J. P., Albers, J . J., Cone, J. T., Brouillette, C. G., Rothblat, G. H. (1987) J. Biol. Chem. 262,12557-12564

Chung, B. H., Kashyap, M., Glasscock, M. A., and Anantharamaiah, G. M. (1987) J. Lipid Res. 28,913-929

45. Bockris, J. O., Devanathan, M. A. V., and Muller, K. (1963)Proc. R. SOC. Lond. A 274,55-79

46. Letizia, J. Y., and Phillips, M. C. (1991) Biochemistry 30, 866-873 47. Huang, C. (1969) Biochemistry 8,344-345

Saunders Co., Philadelphia

Chem. 268,25343-25349

Lipid Res. 33, 1091-1097