Synthetic amphipathic helical peptides that mimic apolipoprotein A I in clearing cellular cholesterol

Synthetic amphipathic helical peptides that

mimic apolipoprotein A-I in clearing cellular

cholesterol.

A J Mendez, … , J P Segrest, J F Oram

J Clin Invest.

1994;

94(4)

:1698-1705.

https://doi.org/10.1172/JCI117515

.

Clearance of excess cholesterol from cells by HDL is facilitated by the interaction of HDL

apolipoproteins with cell-surface binding sites or receptors, a process that may be important

in preventing atherosclerosis. In this study, synthetic peptides containing 18-mer

amphipathic helices of the class found in HDL apolipoproteins (class A) were tested for their

abilities to remove cholesterol and phospholipid from cultured sterol-laden fibroblasts and

macrophages and to interact with cell-surface HDL binding sites. Lipid-free peptides

containing two identical tandem repeats of class A amphipathic helices promoted

cholesterol and phospholipid efflux from cells and depleted cellular cholesterol accessible

for esterification by acyl CoA/cholesterol acyltransferase, similar to what was observed for

purified apolipoprotein A-I. Peptide-mediated removal of plasma membrane cholesterol and

depletion of acyl CoA/cholesterol acyltransferase-accessible cholesterol appeared to occur

by separate mechanisms, as the latter process was less dependent on extracellular

phospholipid. The dimeric amphipathic helical peptides also competed for high-affinity HDL

binding sites on cholesterol-loaded fibroblasts and displayed saturable high-affinity binding

to the cell surface. In contrast, peptides with a single helix had little or no ability to remove

cellular cholesterol and phospholipid, or to interact with HDL binding sites, suggesting that

cooperativity between two or more helical repeats is required for these activities. Thus,

synthetic peptides comprising dimers of a structural motif common to exchangeable

apolipoproteins can mimic apolipoprotein […]

Research Article

Find the latest version:

http://jci.me/117515/pdf

Rapid

Publication

Synthetic Amphipathic

Helical Peptides That

Mimic Apolipoprotein A-I

in

Clearing Cellular Cholesterol

Armando J. Mendez,* G. M. Anantharamaiah,t Jere P. Segrest,*and John F.Oram* *Department of Medicine RG-26, University of Washington, Seattle, Washington 98195; and

*Department

of Medicine and Biochemistry, and theAtherosclerotic Research Unit,

Universityof AlabamaatBirmingham Medical Center,Birmingham, Alabama 35294

Abstract

Clearance of excesscholesterol from cells by HDL is facili-tatedby the interaction ofHDL apolipoproteins with

cell-surface binding sites or receptors, a process that may be important in preventing atherosclerosis. In this study,

syn-theticpeptides containing18-meramphipathichelices of the class found in HDL apolipoproteins (class A) were tested for their abilities to remove cholesterol and phospholipid from culturedsterol-ladenfibroblasts and macrophages and to interact with cell-surface HDL binding sites. Lipid-free

peptides containingtwoidenticaltandem repeats of class A

amphipathichelicespromoted cholesterol and phospholipid efflux from cells and depleted cellular cholesterol accessible for esterification by acyl CoA/cholesterol acyltransferase, similartowhatwasobserved forpurified apolipoprotein A-I. Peptide-mediated removal ofplasma membrane choles-terolanddepletionofacyl CoA/cholesterol

acyltransferase-accessiblecholesterolappearedtooccurby separate mecha-nisms,asthe latter processwaslessdependenton

extracellu-larphospholipid. Thedimeric amphipathichelicalpeptides alsocompetedfor

high-affinity

HDLbindingsiteson choles-terol-loadedfibroblasts anddisplayedsaturable

high-affin-ity

binding tothecell surface. In contrast,peptides with a singlehelixhadlittleor no

ability

toremovecellular choles-terol and phospholipid, or to interact with HDL binding sites, suggesting that

cooperativity

between two or more helical repeats is required for these activities. Thus, syn-theticpeptides comprisingdimersofastructural motif com-mon toexchangeable

apolipoproteins

canmimic apolipopro-tein A-I in both binding toputative cell-surface receptors and clearing cholesterol from cells. (J. Clin. Invest. 1994.

94:1698-1705.) Key words: synthetic peptides * amphi-pathichelices-

high-density

lipoprotein-cellularcholesterol efflux * HDLreceptors

Introduction

The roleofHDLinpromotingthe removal ofexcesscholesterol fromcultured cellshasbeen wellestablished; however,

mecha-AddresscorrespondencetoJohn F.Oram, Departmentof Medicine RG-26, UniversityofWashington, Seattle,WA98195.

Receivedfor publication27April1993 andinrevisedform2 June 1994.

nisms by which this occurs remain poorly understood. HDL appears to remove cholesterol from cells by both passive and active processes. Cholesterol that is synthesized by cells or ingested by receptor-mediated endocytosis of lipoproteins can be transported rapidly to the plasma membrane where it pas-sively desorbs into the extracellular fluid and reabsorbs to phos-pholipid-containing particles ( 1, 2). When cells are overloaded with sterol, however, cholesterol accumulates within an intracel-lular pool that is accessible to the enzyme acyl CoA/cholesterol acyltransferase

(ACAT),'

thus allowing cells to store excess cholesterol as ester-rich lipid droplets. Under this condition, excretion ofexcessACAT-accessible cholesterolcanbe facili-tatedbyanactive process (3, 4) that is stimulated by the interac-tionof HDL apolipoproteins with cell-surfacebinding sites or receptors (5).

A common secondary structural motif described for the plasma lipoprotein-associated apolipoproteins is the

amphi-pathic helix (reviewed in 6 and 7). Basedontheproperties of

amphipathichelices found in the exchangeableapolipoproteins,

asynthetic peptide, designated 18A, wasdesigned and synthe-sized(8-10).This classAamphipathic helical peptide has been showntoform stablecomplexes withphospholipidssimilarto thoseformedbyapolipoproteinA-I, the major protein constit-uentofHDL (8-10). A dimer of 18A linked with aproline residue (37pA) has greater lipid affinity than 18A,presumably

dueto cooperativity provided by twocovalently linked lipid-bindingpeptides. Incontrast, an analogue of 18A(18R), with theposition of charged residues reversed along the polar face of the helix, can only marginally interact with lipid (8-10). Synthetic amphipathichelicescanmimic several of the

biologi-cal effects of apo A-I, and the potencies of these effects are related tothelipid binding affinities ofthe synthetic peptides

(11-15).The present study examines whether these synthetic

amphipathic peptides could also mimicthe abilityoflipid-free

apoA-IandHDL topromote efflux ofexcess cholesterol from cultured cells andtointeract withHDLbindingsitesonthecell surface.

Methods

Lipoprotein andapolipoprotein isolation. LDL and HDL3 (hereafter referredto asHDL)werepreparedbysequential ultracentrifugationin thedensityintervals 1.019to1.063and 1.125to1.21g/ml, respectively. LDLwasacetylated bythemethod ofGoldsteinetal.(16)and labeled

1.Abbreviationused in thispaper:ACAT, acylCoA/cholesterol

acyl-transferase. J. Clin. Invest.

©The AmericanSocietyfor ClinicalInvestigation,Inc. 0021-9738/94/10/1698/08 $2.00

with [3H]cholesteryl linoleate asdescribed (17).HDL wasiodinated by theiodine monochloride method describedbyBilheimer(18).apo A-I waspurifiedfromHDL aspreviously described (4). Purified apo A-Iexhibited a single band when examined by sodiumdodecylsulfate polyacrylamide gel electrophoresis with Coomassie bluestaining.

Phospholipidvesicles. Vesicles werepreparedbythe cholate dial-ysis method (19). For cell cultureexperiments,the vesicles were added tothemediumimmediatelyprior toaddingthemedium to cells.

Peptides. Peptides were synthesized andpurifiedasdescribed pre-viously (9). The primary sequences of the peptides used in this study are shown in Table I.

Cellculture. Human skin fibroblastswerecultured and maintained as previously described (4, 5, 18). Usually 70,000 cells/2 ml medium in 35 mm dishes or 15,000cells/l ml medium in 16mmwells were seededand grown to confluencebeforeuseinexperiments. Cellswere loaded with nonlipoproteincholesterolby incubation for 48 h with DME containing 2 mg/ml fatty acid-free BSA (Sigma Chemical Co., St. Louis,MO) and 30

_Msg/ml

cholesterol. Cellswerethenincubated over-night (16-18 h) in DME containing 1 mg/ml BSA and0.2 sCi/ml [3H]cholesterol(40-60Ci/mmol,AmershamCorp.,Arlington Heights, IL) tolabel plasma membrane cholesterol. Forexperimentscontaining [3H]choline-labeled phospholipids (see below), cells were labeled with0.1 uCi/ml[14C]cholesterol(50-60 Ci/mmol,Amersham Corp.). Choline-containing phospholipids werelabeled by including 1 tCi/ml [3H]choline chloride (75-85 Ci/mmol, Amersham Corp.) duringthe overnight incubation offibroblasts. After incubation with radiolabel, celllayerswererinsed five times withPBS containing 0.1%BSA(PBS/ BSA) before the additionof experimental media.

Resident mouse peritoneal macrophage cultures wereestablished from unstimulated Swiss Webster micebythe method of Edelson and Cohn(20). Peritonealcells were harvestedfrom 20-25mice inPBS, pooled at4°C, collected by centrifugation (400g, 10min, 4°C), and washedoncein DME. Cellswereplated in DMEcontaining20%(vol/ vol)FBS at 1 x 106cells/0.5ml/16mmwell. After incubation for2 h at37°C, nonadherent cells were removed by washing three times with DME.Adherent cells were maintained in culture with DMEcontaining 20% total bovineserumfor 5 dbeforeuseinexperiments. Macrophages werecholesterol loaded by incubation with 50

jig/ml

of[3H]cholesteryl linoleate-labeledacetylated-LDL for 24 h. Cells were then washed two times with serum-free DME and incubatedovernightwith serum-free DMEtoallow equilibration ofcholesterol pools.

Cholesterol andphospholipid effluxfrom cells.After theappropriate labeling protocol,cellswereincubatedat37°Cwith DMEcontaining1 mg/ml BSA and the indicated additions.At the indicated times, the efflux media were collected and celllayerswererinsed three times with PBS. Mediaandcells were stored frozen at-20°Cuntil extraction for lipid andproteincontent. Efflux mediawere extractedbythe method of Folchetal.(21)and,in someexperiments (labeled with cholesterol), counteddirectly afterabriefcentrifugation to remove cell debris. Cho-lesterol-labeledcell layerswereextracted withhexane/isopropanol(3:2, vol/vol) asdescribed (5, 17). Cells containing labeled phospholipids wereextracted with 1 ml ofisopropanolfor 1 h, and then with hexane/ isopropanolasabove.

Sterolspecieswereseparatedbythinlayer chromatographyonsilica gel Gplates developed in hexane/diethyl ether/methanol/acetic acid (120:30:10:1.5, vol:vol:vol:vol). Choline-containing phospholipids were separated by thin layer chromatography on silica gel H plates developedinchloroform/methanol/water (65:35:4, vol:vol:vol). Lipid spots were identified by staining with 12 vapor and comigration with standards. After allowing 12 stainto dissipate, appropriate spots were taken for determination ofsterol mass and/or scintillation counting. Labeledcholesterol efflux and labeledphospholipidefflux were calcu-lated as the percent of label appearing in the media, relative to the total (media+ cell)radioactivityfor each lipid.

Cell cholesterol esterification. To assess the relative activity of ACATfollowing incubationwithtestmedium, cellswerewashed once with PBS and incubated for 1 hat 370Cwith DMEcontaining9_qM '4C-oleate (50-60mCi/mmol,AmershamCorp.)boundto3

piM

BSA

(5). Cellswerechilled onice and washed twice with PBS/BSA and twice withPBS. Cellswerestoredfrozenat-20'Cuntil extraction for lipidandproteinasdescribedabove. Celllipidswereseparatedbythin layer chromatographyasdescribed abovetodeterminecholesterylester radioactivity. Incorporationof radiolabel into cholesterol esters

repre-sentscholesterolesterification by ACAT, and results were_quantitated as picomoles of ['4C]oleate incorporated into cholesteryl esters per milligramof cellprotein.

Cell-surfacebindingofHDLandpeptides. Thespecificityof

high-affinityHDLbindingsitesonfibroblastswastestedusingasequential competitive binding assay (22). Cholesterol-loaded fibroblasts were washed twice with PBS/BSA and chilled oniceduring athird wash. Cellswerethen incubated at0C (on ice) with Hepes-buffered DME containing1mg/mlBSA and theindicated concentration ofcompetitor. After3h,the cellswererapidlywashed three times with ice-cold PBS/ BSA and incubated for 1 hat0CwithHepes-bufferedDMEcontaining 1 mg/mlBSAand 5

_/sg/ml

of'"I-HDL. Specific activityof'25I-HDL ranged from200 to 300cpm/ngHDLproteinwith less than 2% of label associated withlipids.Cellswerewashed five times with ice-cold PBS/ BSAandtwice with PBS. Celllayersweredissolved in 0.1 MNaOH, andaliquotsweretakentoquantitate radioactivityandprotein.

To measure directbinding ofpeptides 18A and 37pA to the cell surface, tyrosylresidues were labeled with1251 bytheiodine monochlo-ride method (18, 23),and radiolabeled peptideswere extensively dia-lyzed.Thespecific activities ofthepeptides rangedfrom 800 to 2,200 cpm/pmol.Fibroblasts were incubated with 10 nM to 2jMradiolabeled peptide for 3 h at 0°C as described above for binding of

1"I-HDL.

lodination of thepeptideshad no effect onbiological activitybasedon results showing that iodination of37pA did not change its ability to inhibit cellular cholesterol esterification and that unlabeled 37pA blocked cell-surfacebindingofiodinated37pA (datanotshown).

Othermethods.Proteinconcentrations wereestimated bythe method ofLowryetal.(24) usingBSA as the standard. Statistical differences werecalculated byStudent's ttest. Kinetic analyses of peptidebinding and lipid efflux were performed using the SAAM30 computer program.

Results

Theeffects of HDL,apo A-I,andsynthetic amphipathic helical peptides on cholesterol clearance from cultured human skin fibroblasts were determined by two independent assays. First,

as an index of cholesterol efflux from the plasma membrane,

the appearance of radiolabeled cholesterol in the medium was measured after cholesterol-loaded cells were labeledovernight

with trace quantities of [3H]cholesterol. Second, depletion of cellular pools accessible for esterification by ACAT was esti-mated by the decrease in incorporation of

[14C]

oleate into cholesteryl esters.Results show thatHDLand lipid-freeapo A-I can promote [3H]cholesterol efflux and decrease cholesterol esterification (Fig. 1,A and B). Both apo A-I and HDL had maximal effects on cholesterol esterification at - 400 nM,

al-thoughHDLhada greatercapacity to promote [3H] cholesterol efflux,presumably because of exchange between HDL choles-terol and plasma membrane tracer (1, 2).

1200 800

l1000 rAc18A-NH2 60

800

I

600

-4

600

[I

400

-0

400 2 200

200

0 3 6 9 0 2 4

gMProteinorPeptide

Table 1. Amino Acid Sequence of Synthetic Peptides

Peptide Sequence

_ + +_

18A

Asp-Trp-Leu-Lys-Ala-Phe-Tyr-Asp-Lys-Val-Ala-Glu-+ +

-Lys-Leu-Lys-Glu-Ala-Phe

+ -+ - +

18R

Lys-Trp-Leu-Asp-Ala-Phe-Tyr-Lys-Asp-Val-Ala-Lys-_ - +

Glu-Leu-Glu-Lys-Ala-Phe 37pA 18A-Pro-18A

37aA 18A-Ala-18A Ac-18A-NH2 Acetyl-18A-NH2

0

x

E

0

79

w

I CO

0

Figure1. Effect of HDL, apo A-I, and synthetic amphipathic helical peptidesoncellular cholesterol effluxand esterification. Cholesterol-loaded human skin fibroblasts, labeled with[3H]cholesterol,were incu-bated with serum-free DME containing 1 mg/ml BSA and the indicated concentrations of HDL or peptide. After 6 h, the medium was removed and cells werefurtherincubatedfor 1 h with medium containing 14C-oleateto measurecholesterolesterification byACAT.Celland medium lipids were extracted and separated by thin-layer chromatography, and cholesterol andcholesterylesterradioactivitywasquantitated. Choles-terolefflux(opensymbols)wascalculatedasthe percent of total(cell andmedium) [3H]cholesterolappearingin the medium. Total

[3H]-cholesterolcounts were18,605±1,211 cpm/dish.Cholesterol esterifica-tion(closedsymbols)wascalculatedaspicomolesof'4C-oleate incorpo-ratedintocholesterylestersnormalized forcellprotein. The resultsare themeansofduplicatedishesrepresentative oftwo tofive experiments. Thecoefficientsofvariationforcholesterol efflux and cholesterol esteri-fication assayswere 5.9% and6.9%, respectively.

shown). None of the other peptides tested had any effect on

incorporation of oleate intotriglycerides comparedtocontrols. Thus, even thoughAc-18A-NH2 appeared toperturb thelipid

environment of the plasma membrane, itwas no more active than 18A or18R in promotingcholesterol efflux.

In contrast to the monomers, dimers of 18A linked with proline (37pA) oralanine (37aA) wereabletoinhibit choles-terol esterification and promote[3H]cholesterol efflux (Fig. 1, Dand F). The maximumextentofdepletion of ACAT substrate for thesepeptideswascomparabletothatforHDLand apo A-I.Time course studiesatsaturating concentrations of HDL, apo A-I, and 37pA demonstrated equivalentrates for thedecrease incholesterol esterification over16 h ofincubation, while18A remained withouteffect (datanotshown).

Thesynthetic peptides werecompared to HDLand apo A-Ifor theirabilitytopromote clearance of cholesterol from sterol-laden mouseperitonealmacrophages,acultured cell model with features common tofoam cells of atherosclerotic lesions(16).

Cellswereoverloaded with cholesterol by incubation with

ace-tylated-LDL containing [3H] cholesteryl linoleate to radiolabel cellularpools of free and esterified cholesterol by the scavenger receptor pathway (16, 25). HDL and apo A-I were able to promote effluxof [3H]cholesterol into themedium and decrease the amount ofcholesterolavailable foresterification by ACAT (Fig. 2,AandB). Peptides37pA and 37aAwerealso effective inpromoting[3H cholesterolefflux and decreasing [ 4C]oleate

20- A Figure 2. Effect of synthetic

20 A amphipathic peptides on

15 cholesteroleffluxfrom mouseperitoneal

macro-CiF( De D D D * phages.5-d-oldmacrophage

5

-cultures were incubated for 24 h withserum-freeDME 4 containing 0.5 mg/ml BSA

eX23 * _{* *} and 50_1Lg/ml

acetylated-@B** * LDL,whichhadbeen

la-? t E 2

-beled with

['H]cholesteryl

oag

.c

- * g g g g g linoleate. Cellswerewashed

and incubated with medium o

50 C for 18 htoallow equilibra-_40 _ - _ * tion of cholesterol pools. *§ 30 _, * * * * S Cellswereincubated with 8jz - - * * * - mediumcontaining 0.5 mg/

*20 ml BSA alone (control) or

I 10 * _{with the following additions:} 60 - ~-D 0.36jAMHDL, 0.36 ,M apo

50_

* *

A-I,

9.0

_/M

18A,4.5_1AM

_ X

40 37pA, and 4.5

1iM

37aA.

8zcz30-+ * --- After6h,theefflux medium I=ut20 - * --- wasremoved and cells were 10- _{pulse-labeled with "C-oleate}

o0

_ to determine cholesterol es-i -r_<; terificationbyACAT. Media

O

I

andcelllipidswereextracted 0 0 _{andsubjectedtothin-layer} chromatographytodetermine medium(A)andcellular(D)[3H]

-cholesterol,cellular[3H]cholesterylesters(B),and'4C-oleate incorpo-rated into cellularcholesterylesters(C).Resultsarethe mean±SD of threedishes. Total[3H]cholesterol radioactivity was5,770±157 cpm/

dishforall dishes. Values marked withanasteriskaresignificantly different from controlsbyStudent'sttest;P< 0.005 in(B),P

<0.05in(C)and(D).In(A),all values except thosefor 18Aare

16HD

:0 5 0

~~~~~37pA

3 apoA-1

w2

I ~~~~~18A

C D

16

_{~~~~~~~~~~~~~~~~~apo}

_A-1I

ii10

0. 18AHL

0 5 10 0 0.2 0.4 0.6 0.8

jLMProteinorPeptlde

Figure3. Comparison of the effects ofHDL, apoA-I,andsynthetic peptidesoncholesterol and

phosphatidylcholine

effiuxfrom cholesterol-loaded fibroblasts. Cholesterol-cholesterol-loadedfibroblastswereincubated for 18 hwith serum-free mediumcontaining1mg/mlBSA, 0.1 MCi/ml[14C] -cholesterol,and1.0,.sOi/ml[3H ]choline chloride(tolabel cellular phos-phatidylcholine). Labeled cellswerethenincubated with the indicated concentrations of HDLorpeptidesfor 6 h. Cell and mediumlipidswere

extractedandseparated by thin-layer chromatography toquantitate

['4C]cholesterol (AandB) and[

_{3H]phosphatidylcholine}

_(CandD)

(asdescribed inMethods). Effiuxwascalculatedasthepercentof total (cellandmedium) labelappearinginthemedium. For[3H]

-phosphatidylcholineand[14C]cholesterol,totalcountswere4,049±t121

and13,337±368

cpm/idish,

respectively (±-SEM).Resultsarethemean

ofduplicatedishes. Thecoefficient of variationwas 10.0% and 9.6% forphospholipidandcholesteroleffiux assays, respectively.

incorporation

into cholesterol esters.

Peptide

18A was

again

without effect. The appearance of

[3

H]

cholesterol in the

me-dium couldbe accounted for

by

decreases inboth cellfree and esterified cholesterol

(Fig.

2, C and D). apo

A-I, 37pA,

and 37aAcauseda

significant

decrease inesterified cholesterol con-tent. HDLwas less effectiveat

cholesteryl

esterclearance, and effiux,wasdue

mostly

toaloss of freecholesterol. Itis

notewor-thy

that the decrease in cell[

3H]

cholesteryl

esters

by

the active

peptides

and HDL

strongly

correlated (r = _{0.794) with} the decrease in ACAT-mediated cholesterol esterification. These data demonstrate that apo, A-I,

37pA,

and 37aA have similar abilitiestopromote clearance ofexcesscholesterol from

macro-phages.

Peptides

18A and

37pA

were

compared

to HDL and apo A-I for their

ability

to promote cholesterol and

phospholipid

effiux from cholesterol-loaded fibroblasts

_(Fig.

3). Dose

re-sponsecurvesshowed thatHDL, apo

A-I,

and

37pA promoted

effiux ofbothradiolabeled cholesteroland

_{phosphatidylcholine.}

At

saturating

concentrations, all three acceptors were able to promote cholesterol effiuxto

nearly

thesameextent.Incontrast, HDLwasmuch lesseffectiveat

removing

cellular

phosphatidyl-choline

compared

tothe

lipid-free peptides,

presumably

because its

apolipoproteins

were

already

associated with

phospholipids.

Peptide

18A did notpromoteeffiux of either

phosphatidylcho-line orcholesterol atthe concentrations tested.

Results shown in

Fig.

3 suggestthat removalofcholesterol from cells

by

apo A-I or

37pA

may be

secondary

toformation of

_{phospholipid-rich particles}

that actas cholesterol acceptors.

To furthertest forthis

possibility,

we

compared

the

ability

of

37pA

to stimulate cellular cholesterol and

phospholipid

effiux

inthe presenceorabsenceofanextracellularsourceof

phospho-lipid.

Addition of

phosphatidyicholine

vesicles alonetothe

me-dium

promoted

efflux of[3H]cholesterol from the

plasma

mem-branes of fibroblasts (0 kLM

37pA, Fig.

4A). In contrast, the vesicles had no measurable effect on effiux of [3H

]

-phosphatidylcholine during

theincubations

(Fig.

4 B) nor on

cholesterol esterification

by

ACAT

(Fig.

4C)

following

the 6-hincubations. In thepresenceof

phosphatidylcholine

vesicles,

37pA

had no additional

ability

to stimulate [3H]cholesterol effiux

(compared

to vesicles alone) atconcentrations up to 2

MLM. This

_peptide,

however, was still

capable

of

stimulating

cellular [3H

]phosphatidylcholine

efflux and

inhibiting

choles-terolesterificationwhenvesicleswereincludedin the medium,

although

thehalf-maximal concentrationwasincreased

by

two-to threefold.This concentrationshiftmay havebeencaused

by

the

_'interaction

of

37pA

with

phospholipid

vesicles. These

re-sults supportthe conceptthatefflux of

plasma

membrane cho-lesterol

depends largely

on the presenceof extracellular

phos-pholipid particles,

whereas both effiux of

plasma

membrane

phospholipids

and removalof ACAT-accessible cholesterol oc-cur

by

aseparate mechanism,

possibly involving

the direct inter-action of

amphipathic

helices with cell-surface

binding

sites. The different mechanisms for removal of[3H]cholesterol and ACAT-accessible cholesterol suggest that these sterols reside indifferentcellular

pools.

We have

_previously

demonstrated that

binding

of HDL to

putative

cell-surface receptors is

required

to promote removal ofexcesscholesterol fromtheACAT substrate

pools,

and that intact

_{apolipoproteins}

are

required

for

binding

(5). Totestthe

possibility

thatapo A-Iandthe

synthetic peptides

are

_operating

by

the same mechanism, we

_compared

the

_ability

of different

peptides

tointeractwith

_{high-affinity}

HDL

binding

sitesonthe surfaceofcholesterol-loaded fibroblasts. Forthese studies, we

12 _A Figure4.Effect of37pAon

cho-10 lesterol efflux(A),

phosphati-8 dylcholineefflux (B),and

cho-lesterol esterification (C)in the

.CX

~~~~~~presence

ofphospholipid

vesi-4 cles. Parallel cultures of choles-2 terol-loaded human skin

fibro-________________ blastswere labeled with either B 0.2,.sCilml [3H]cholesterolor

6

_~~~~~~1.0

_pCi/ml

_[3H]choline

chlo-~~~

_~ride

~~~~~ _{for 18 h}asdescribed in

06 _{Methods. The cells}werethen in-3Jcubated for 6 hat370(2with the

2 indicated concentrations of

a.

I

~~~~~~37pA

inthe presence(closed

j

~~~~~symbols)

orabsence(open

sym--E 2.0 C bols)of 100

,ig/ml

of phospha-tidylcholinevesicles added to e1.5

_~~~~~~the

mediumimmediately prior

a

1.0-1.0

~~~~~to

~~~~~~tion,

the_oneincubations._{setof dishes}After incuba-_was

E0. pulsed with 14C-oleate to

mea-C

~~~~~~~~sure

cholesterolesterificationas 0 _{0 0 1 2 3 4} described forFig. 1.

_Effiux

me-1M

37pA

dia and celllayerswereanalyzed

forcholesterol,cholesterylester, and

_{phosphatidylcholine}

radioactivity.For[3_H

_{]phosphatidylcholine}

_and

[3H]cholesterol,totalcounts were13,848± 1,469and 14,114±135cpm/

-Figoure5.Interaction of synthetic amphipathic helical peptides with cell-surface HDL bind-ing sites. Cholesterol-loaded fibroblasts were chilledoniceand incu-bated with ice-cold se-rum-free medium con-taining 1 mg/ml BSA and the indicated con-centrations of 37pA, 18A, 18R, apoA-I,or HDL.After 3 h, dishes were rinsed three times with PBS/BSA, 1 ml of serum-freemedium con-taining5

_Mg/ml

"MI-la-beled HDLwasaddedtoeachdish, and incubationswerecontinued for 1honice. Cellswerethen washed five times with PBS/BSA and twice

withPBS and dissolved in 0.1 M NaOH, and aliquotsweretaken for

125I and protein determinations. Resultsaremeansof duplicate dishes (3.8% coefficient of variation) representingatleastthreeseparate

exper-iments. In the absence of competitors, cells bound 33-3ng12"I-labeled

HDL/mg cell protein (n= 6).

usedasequential competitive binding protocoltoassayfor the

abilityof differentpeptidestoblockhigh-affinityHDLbinding. Cellswerefirstexposed at0°C toincreasingconcentrations of unlabeled competitor and, after removal of unbound competitor,

were then incubated at0°C with

'"I-HDL

at a concentration nearitshigh-affinity Kd. Because high-affinityHDLbindingis

essentially irreversible at0°C (26, 27), moleculesthat bindto these sites with similar kinetics as HDLshould remain bound

duringthesecond 0°C incubationandblock high-affinity bind-ingof'251I-HDL.Itwasnecessarytouse asequential competition

assayrather thana moreconventional simultaneous incubation

tominimize the interaction ofpeptidesand

1"I-HDL

insolution, whichmay produce artifacts due to the displacement of

'"I-apolipoproteinsfrom HDLparticles (10, 23).Asimilar

sequen-tialcompetition assayhas been used previouslytoidentify

re-ceptorligandsinplasmathat interact with otherplasma compo-nents(22).

Results using the sequential competitive binding assay

showed that neither 18Anor18R could prevent thesubsequent bindingof '25I-HDL (Fig.5A).In contrast,apoA-Iand37pA effectivelyblocked sitesrecognized by '251-HDL(Fig. 5,Aand B).These sites werealso blockedby37aA(datanotshown). The concentrations for half-maximal inhibition of

1"I-HDL

binding by HDL, apo A-I, and 37pA were 3 X 10-8 M, 1

x 10-8M,and 5X 10-8M,respectively.Theseresults indicate that high-affinity HDL binding sites on cholesterol-loaded

fi-broblasts also interact withapoA-I, 37pA, and37aA, butnot

with 18A and 18R.

We also examined direct binding of iodinated 37pA and 18Ato cholesterol-loaded fibroblasts (Fig. 6). Both peptides boundextensivelytocells,but with different saturation kinetics. Over a concentration range of 0.01 to 2.0 ,iM, 18Abinding

wasnearlylinear. Athigherconcentrations(not shown), some

saturability was evident, but thesehigh-capacity binding sites hadarelatively low-affinity (apparent Kdof10-5 M). In

con-trast, computeranalysis of

"25I-37pA

bindingindicated that the concentration curvebest fita single saturablebinding

compo-nent with a Kd of 2 x

10'

M,

plus

a linear

(nonspecific)

component.Thus,37pA appears to interact with saturable cell-surfacebinding sites that haveamuchhigher affinity than those for 18A.

We characterized thespecificity of the 37pA binding sites by competitive binding studies. To assess competition by HDL, weused the sequential competitive binding assay to avoid po-tential interactions between iodinated 37pA and lipoprotein par-ticles. At a concentration of iodinated 37pA near its Kd for high-affinity binding (10-7 M), pretreatment of cholesterol-loaded fibroblasts with unlabeled37pA and 37aA blocked 40% of the iodinated 37pA binding in a concentration-dependent and satu-rable manner (Fig. 7). Pretreatment with 18A had no effect on 37pA binding, even at concentrations of 18A that exhibited twofold greater total cellularbinding than 37pA (Fig. 6). In contrast, pretreatment of cells with apo A-I or HDL blocked approximately 20% of the 37pA binding sites (half that blocked by 37pA). These results suggest that 37pA is interacting with multiple high-affinity binding sitesoncholesterol-loaded fibro-blasts,none of which interactwith 18Awith thesame kinetic properties. A subset of these sites, however, appears to bind both HDL and apo A-I, consistent with results showing that 37pA interacts with thehigh-affinity HDL binding sites.

Discussion

Resultspresented in this report demonstrate that the synthetic peptides 37pA and 37aA, composed of two 18-mer class A amphipathic helical repeats (6-11), promoted cholesterol ef-flux andinhibited cellular cholesterol esterification in both fi-broblasts andmacrophages. These effectsweresimilartothose mediated by lipid-free apo A-I and HDL, except that HDL had a greatercapacity topromote efflux ofplasma membrane

cholesterol, probably dueto exchange of cholesterol between

1.5

037pA

'1.0

-0.5

Cl

0.0

0.0 0.5 1.0 1.5 2.0

pM

1251-ppUft

Figure6.Cell-surfacebindingofiodinated18A and37pAto cholesterol-loadedfibroblasts.Cholesterol-loaded humanskinfibroblastswere incu-batedwithincreasingconcentration of iodinatedpeptidefor 3 hat40C. Then,cellswerewashed,andradioactivitywasdeterminedasdescribed inFig.5. Thespecificactivities of"SI-labeled 18Aand "MI-labeled

37pAwere810 and2,100cpm/pmol,respectively.Each value represents themeanofduplicatedishes(5.5% coefficient ofvariation).

1702 A.J.Mendez, G.M.Anantharamaiah, J.P. Segrest, and J.F. Oram

A

o_37pA

.9 ~~~~~*18A

v18R

100

80

60

100

40

20

0 50 100 150 2C

B

037pA

o3apoA-I

*HDL3

0 2 4 6 8 10

cholesterol.Conversely, addition of37pAtomediumcontaining phospholipid vesicles increased cellular phosphatidylcholine ef-flux severalfold anddepletedoverhalf of the ACAT-accessible cholesterol when there was no further increase in cholesterol efflux from the

plasma

membrane.

One explanation for these findings is that, of these three lipidtransport processes,only phospholipid efflux anddepletion

ofACAT-accessible cholesterolinvolve interactions of amphi-pathic helices withcell-surface binding sites. Promotion of cho-lesterolefflux from the plasma membranemaybeasecondary o

_37pA

responsetoformation of membrane-derived

_phospholipid

com-V 37aA plexesgenerated byreversible cell-surfacebinding of peptides * 18A orlipid-freeapolipoproteins. Thesenewly formedparticlesmay o apoA-I act as sterol acceptors and enhance passive diffusion of choles-U

_HDL3

_{terolfrom the cell surface (1, 2). This two-steplipidtransport} 0.0 0.2 0.4 0.8 0.8 1.0 1.2 has been postulated for promotion ofcholesterol efflux from cells by various lipid-free apolipoproteins (25, 28-31). It is

jIM ProteinorPeptide likely, however, that removal of plasma membrane phospholipid InteractionofHDL, apo A-I, andsynthetic peptideswith cell is significant only for those apolipoproteins that are lipid-defi-7pA binding sites. Cholesterol-loaded fibroblasts were chilled cient. In contrast, the interaction ofapolipoproteins with cells d incubatedwith ice-cold serum-freemediumcontaining1 depletes ACAT-accessible cholesterol even when the apolipo-SAand the indicated concentrations of37pA, 37aA, 18A, apo proteins are associated with HDL phospholipid. This process DL. After3h,dishes were rinsed threetimes withPBS/BSA, may be initiated by the release of intracellular signals which ,rum-free medium containing 0.5

_jig/ml

(0.11 M) "NI-labeled activate a cholesterolexcretory pathway (3, 4).

sadded to eachdish,and incubationswere continued for 1 h Competitive binding studies showed that apo A-I, 37pA, ellsweretreated andradioactivitydetermined as described in and 37aA blocked theinteraction of HDL with its high-affinity tal cellbindingof'"I-labeled37pA(specific activityof2,100

_binding

_{sites on cholesterol-loaded fibroblasts, while 18A}had 1) in the absenceofcompetitorwas 64pmol/mgcell

protein,

no effecton HDL binding. These results strongly suggest that le representsthemeanOf duplicate dishes(3.5%/ coefficient

)n).

the interaction of HDL with its cell-surface

binding

site is medi-ated by two or more tandem repeats of class A amphipathic helices withinHDLapolipoproteins. Thus, peptides containing two of these amphipathic helicescan mimic apo A-I in both roteinparticle and cells. Additionally, 37pA and 37aA stimulating lipid transport from cells and in binding to high-)moteefflux of cellular [3H]phosphatidylcholine, simi- affinity HDL binding sites. The higher efficacy of the dimeric effects of apo A-I, while HDL hadalimitedcapacity peptides comparedtothemonomers cannotbeexplained byan

pholipid

efflux. In contrast tothepeptide dimers, pep- extension of the polar and nonpolar faces along the helical axis, taining only a single 18-meramphipathic helix (18A for linkage of the two 18-mer helices bya single amino acid

8A-NH2)had littleor noabilitytostimulate cholesterol causes a350 rotation of these faces. Moreover, linkage by

pro-;pholipid efflux and to deplete cells of substrate for line compared to alanine causes a "kink" between the axes, olesterification by ACAT.Thus, although thesemono- yet 37pA and 37aA were equally active. It is most likely that ebeen showntointeract withlipids and have biological cooperativity between two or more amphipathic helices is re-8-15), theywerealmostcompletely inactive in mim- quiredfor optimum binding to high-affinity sites on the cell ko A-Ior HDLinstimulating lipid transport from cells. surface, as has been described previously for the interaction of sureof fibroblaststoapoA-I or37pA increased efflux HDL apolipoproteins with cells (28, 32). This property may plasmamembrane ofboth cholesterol and phosphati- confer specificity for lipid-poor subclasses of HDL particles ieinparallel to depletion of ACAT-accessible choles- which have morehelices exposed for cellular interactions (7,

plying

that these three lipid transport processes are 33, 34).

by thesame cellularinteractions. Several other lines Studies withiodinated peptides revealed that both 18A and of

evidence, however,

suggest that theseprocessesareoccurring

by independent

mechanisms.First,removal of cholesterol from either the

plasma

membraneorACAT substratepoolsdoesnot

require

simultaneous release ofplasmamembrane

phospholip-ids,

asevidenced

by

results

showing

thatHDLstimulatedefflux of bothsources of cholesteroltothesameextent as 37pAand apo A-I but had

markedly

less

ability

to promote efflux of cellular

phospholipids. Second,

efflux of radiolabeled choles-terol from the

plasma

membranecanbedissociated from both

phospholipid

efflux anddepletionof ACAT-accessible choles-terol. For

example,

addition of phospholipid vesicles to the medium increasedcholesterol efflux from theplasmamembrane

severalfold within 6

h,

but there was no significant effecton

phosphatidylcholine

efflux or

depletion

of ACAT-accessible

37pA bind extensively to the surface of cholesterol-loaded fi-broblasts,butonly 37pA bindstosaturable, high-affinitysites. High-affinity binding of 37pAwasnotblockedby 1 8A andwas

only partially blocked by HDL and apo A-I, suggesting that noneof these37pA binding sites interactswith 18A and onlya

subsetinteracts withHDLapolipoproteins. Therefore,although both apoA-Iand 37pA arerecognized by HDL binding sites,

37pAappearstohaveuniquepropertiesthatpromoteits binding

toadditional cell-surface sites that do notbind 18A, apo A-I,

orHDL.

Previous studiesshowed thatmodifying HDLsoas to

de-crease its interaction with cellular high-affinity binding sites

also reduces its ability to remove excess cellular cholesterol

(4, 5, 35-37), suggestingthat thesetwoprocesses areclosely

I.

^.100

0

C

0

) 80

0 co

0 40

r-_

It2 V- 20

0

Figure7. surface 3

onicean mg/mlB A-I,orH]

1ml ofse 37pAwas onice. Cc Fig.5. To cpmlpmol Each valu of variatic

thelipopi could prc lartothe forphosi tidescon and Ac-l and phos cholester mershavt

activity(

ickingap Expor from the dylcholin

terol, imj

mediated

F

-coupled. The findings that both high-affinity binding to cells and removal of cellular cholesterol requiretwo or moretandem amphipathic helices further support this concept. The current study, however, does not provide direct evidence thatHDL,apo A-I, and dimeric peptides are removing cellular lipids through their interactions with a common binding site. The interactions ofamphipathic helices with the plasma membrane are complex and appearto involve multiple binding sites, and the relative contribution of each of these sites to lipid transport processes is unknown. In addition, comparisons ofsaturation curves re-vealed that the apparent half-maximal concentrations for inhibi-tion of HDL binding were approximately fivefold lower than those for removing cholesterol. However, this may reflect some of the special conditions of these assays, particularly the differ-ent temperatures used (370C vs. 0C). The kinetics of high-affinityHDLbinding is highly temperature sensitive, in that the dissociation rateis much faster at 370C than at 0C (26, 27). Moreover, the active conformation of lipid-free amphipathic peptides in solution may vary with temperature. Despite these differences between assays,therelativeefficacy of the peptides was similar for each process. The half-maximal molar concen-trations for stimulation of cholesterol efflux and inhibition of HDLbinding were approximately fivefold higher for 37pA than for apo A-I. This may be relatedto thefact that apoA-Ihas a molecularmass sixtimes that of 37pA and contains four times the number of amphipathic helices. These and other peptide analogues will be useful tools for testing the hypothesis that HDLapolipoproteins stimulate clearance of cellular cholesterol through the cooperative interaction of their amphipathic helices withcell-surface binding sites or receptors.

Numerouspopulationstudies have shownaninverse corre-lation between plasmaHDLlevels and riskfor coronary heart disease, suggesting that HDL protects against atherosclerosis. This may be related to the ability of HDL apolipoproteins to stimulate clearance of cholesterol from peripheral cells,

particu-larly those of the artery wall. The finding thatsynthetic peptides

with structural features ofHDLapolipoproteins canmimic the biological activity of HDL raises the possibility that specific

agonists could bedeveloped forthe cellularcholesterol excre-tory pathway. Theseagonistsmay havetherapeuticbenefits for theprevention of coronaryheartdisease.

Acknowledgments

Wewould liketoacknowledge Rosario Bowen for technical support, and Julie Kirk and PatBurnsforassistance inpreparationof themanuscript. This study was supported by National Institutes of Health grants HL-31194,HL-18645,and HL-34343.

References

1.Phillips,M.C.,W. J.Johnson, and G. H. Rothblat. 1987. Mechanisms and consequences of cellular cholesterol exchange andtransfer. Biochim. Biophys. Acta. 906:223-276.

2.Johnson, W. J., F. H.Mahlberg,G. H.Rothblat,and M.C.Phillips.1991. Cholesterol transport between cells andhigh-density lipoproteins.Biochim. Bio-phys.Acta. 1085:273-298.

3. Theret, N., C. Delbart, G. Aguie, J. D. Fruchart, G. Vassaux, and G. Ailhaud.1990. Cholesteroleffluxfromadiposecells iscoupledtodiacylglycerol productionand proteinkinase C activation.Biochem. Biophys. Res. Commun. 173:1361-1368.

4.Mendez,A.J., J. F. Oram, and E. L. Bierman. 1991. Protein kinase Cas amediatorofhigh density lipoprotein receptor-dependenteffluxof intracellular cholesterol. J.Biol.Chem. 266:10104-10111.

5. Oram, J. F., A. J. Mendez, J. P. Slotte, and T. F. Johnson. 1991. High density lipoprotein apolipoproteins mediate removal of sterol from intracellular pools but not from plasma membranes of cholesterol-loaded fibroblasts. Arte-rioscler. Thromb. 11:403-414.

6. Segrest, J. P., H. DeLoof, J. G. Dohlman, C. G. Brouillette, and G. M. Anantharamaiah. 1990. Amphipathic helix motifs: classes and properties. Pro-teins.8:103-117.

7.Segrest, J. P., M. K. Jones, H. DeLoof, C. G. Brouillette, Y. V. Venkatacha-lapathi, and G. M. Anantharamaiah. 1992. The amphipathic helix in the exchange-ableapolipoproteins: a review of secondary structure and function. J. Lipid Res. 33:141-166.

8.Kanellis, P., A. Y. Romans, B. J. Johnson, H. Kercret, R. Chiovetti, Jr., T. M. Allen, and J. P. Segrest. 1980. Studies of synthetic peptide analogs of the amphipathic helix. Effect of charged amino acid topography on lipid affinity. J. Biol. Chem. 255:11464-11472.

9. Anantharamaiah, G. M., J. L. Jones, C. G. Brouillette, C. F. Schmidt, B. H. Chung, T. A. Hughes, A. S. Bhown, and J. P. Segrest. 1985. Studies of synthetic peptide analogs of the amphipathic helix. J. Biol. Chem.

260:10248-10255.

10. Chung, B. H., G. M. Anantharamaiah, C. G. Brouillette, T. Nishida, and J. P. Segrest. 1985. Studies of synthetic peptide analogs of the amphipathic helix: correlation of structure with function. J.BioLChem. 260:10256-10262.

11.Epand,R.M., A. Gawish, M. Iqbal, K. B. Gupta, C.-H. Chen, J. P. Segrest, and G. M. Anantharamaiah. 1987. Studies of synthetic peptide analogs of the amphipathichelix. J.Bio.Chem. 262:9389-9396.

12. Joregenson, E. V., G. M. Anantharamaiah, J. P. Segrest, J. T. Gwynne, and S. Handwerger. 1989.Synthetic amphipathic peptidesresemblingapolipoproteins stimulate the release of human placenta lactogen. J. Biol. Chem. 264:9215-9219. 13.Blackburn, W. D., J. G. Dohlman, Y. V. Venkatachalapathi, D.J.Pillion, W.J.Koopman, J. P. Segrest, and G. M. Anantharamaiah. 1991. Apolipoprotein A-Idecreases neutrophil degranulation and superoxide production. J. Lipid Res. 32:1911-1918.

14.Owens, R. J., G. M.Anantharamaiah,J. B.Kahlon, R. V. Srinivas, R. W. Compans,and J. P.Segrest.1990.ApolipoproteinA-Iand itsamphipathic peptide analogues inhibit humanimmunodeficiencyvirus-inducedsynctiumformation. J. Clin. Invest. 86:1142-1150.

15. Venkatachalapathi, Y. V., M. C. Phillips, R. M. Epand, R. F. Epand, E. M.Tytler, J. P. Segrest, and G. M. Anantharamaiah. 1993. The effect of end group modificationonthepropertiesofclass Aamphipathic peptides.Proteins. 15:349-359.

16.Goldstein, J. L., Y. K. Ho, S. K. Basu, and M. S. Brown. 1979. Binding siteonmacrophagesthatmediateuptakeanddegradationofacetylatedlowdensity lipoprotein, producingmassive cholesteroldeposition.Proc.Nati.Acad. Sci. USA. 76:333-337.

17. Oram, J. F. 1986. Receptor-mediated transport of cholesterol between culturedcells andhigh density lipoproteins.MethodsEnzymol. 129:645-659.

18.Bilheimer,D.W.,S.Eisenberg,and R. I.Levy.1972. Themetabolism of very lowdensity lipoprotein proteins.Biochim.Biophys.Acta.260:212-221.

19.Albers,J.J., C-H.Chen,and A. G. Lacko. 1986.Isolation, characterization, and assay oflecithin-cholesterolacyltransferase.MethodsEnzymol. 129:691-704. 20. Edelson, P. J., and Z. A. Cohn. 1976. Purification and cultivation of monocytes and macrophages. In In Vitro Methods in Cell Mediated and Tumor Immunity. B. R.Bloom and J. R.David, editors. AcademicPress,NewYork. 333-340.

21.Folch, J.,M.Lees,and G. H.Sloane-Stanley.1957. Asimplemethod for theisolation andpurificationof totallipidsfromanimal tissues. J. Biol. Chem. 226:497-509.

22.Bowen-Pope, D.F.,and R. Ross. 1985. Methods forstudyingthe platelet-derivedgrowthfactor receptor. MethodsEnzymol. 109:69-100.

23.Garber, D.W.,Y. V.Venkatachalapathi, K. B.Gupta,J.Ibdah,M.C. Phillips,J.B.Hazelrig,J. P.Segrest,and G. M. Anantharamaiah. 1992. Turnover ofsyntheticclass Aamphipathic peptide analoguesofexchangeable apolipopro-teinsinrats.Arterioscler. Thromb. 12:886-894.

24.Lowry,0. H., N. J. Rosebrough, A.L.Farr, and R. J. Randall. 1951. Protein measurement with the Folinphenolreagent. J.Bio.Chem. 193:265-275. 25.Hara,H.,and S.Yokoyama.1991. Interaction of freeapolipoproteinswith macrophages. J. Biol. Chem.266:3080-3086.

26. Oram, J.F.,E.A.Brinton,and E. L. Bierman. 1983.Regulationofhigh densitylipoproteinreceptoractivityincultured human skin fibroblasts and human arterial smooth muscle cells. J. Clin. Invest. 72:1611-1621.

27. Oram, J.F.,C. J.Johnson,andT. A. Brown. 1987. Interaction ofhigh densitylipoproteinwith its receptoroncultured fibroblasts inmacrophages: evi-denceforreversiblebindingatthecell surfacewithout internalization. J. Biol. Chem. 262:2405-2410.

28.Hara, H.H.,M.Hiroko,A.Komaba,and S.Yokoyama. 1992.a-Helical requirementsfor freeapolipoproteinstogenerate HDL andtoinduce cellularlipid efflux.Lipids.27:302-304.

29.Hara, H.,and S.Yokoyama. 1992. Role ofapolipoproteinsincholesterol

effluxfrom macrophages tolipidmicroemulsion:proposal of aputativemodel for thepre-fihigh density lipoprotein pathway. Biochemistry. 31:2040-2046.

30.Bielicki,J. K, W. J.Johnson, R. B. Weinberg, J. M. Glick, andG. H. Rothblat 1992.Effluxof lipid fromfibroblaststoapolipoproteins:dependenceon elevatedlevels of cellular unesterifiedcholesteroL J. LipidRes.33:1699-1709.

31.Forte,T. M., R.Goth-Goldstein,R.W. Nordhausen, and M. R. McCall. 1993.ApolipoproteinA-I cellmembrane interaction: extracellular assembly of heterogeneous nascent HDL particles. J.LipidRes. 34:317-324.

32.Leblond, L., and Y. L. Marcel. 1991. The amphipathic alpha-helical repeats of apolipoproteinA-Iare responsible for binding of high density lipoproteins to HepG2 cells. J. Biol.Chem. 266:6058-6067.

33. Rothblat,G. H., F. H.Mahlberg,W.J.Johnson,and M.C. Phillips. 1992. Apolipoproteins,membranecholesterol domains,and theregulation of cholesterol efflux.J. Lipid Res.33:1091-1098.

34.Jonas, A.,A.Steinmetz,and L.Churgay.1993.The number ofamphipathic

a-helical segments ofapolipoproteins A-I, E,andA-IV determines the size and functional propertiesof their reconstitutedlipoprotein particles. J. Biol. Chem 268:1596-1602.

35.Brinton,E.A.,J. F.Oram,C.-H.Chen,J. J. Albers, and E. L. Bierman. 1986.Bindingofhigh density lipoproteintocultured fibroblasts after chemical alteration ofapoproteinamino acid residues. J. Biol. Chem 261:495-503.

36.Duell,P.B.,J. F.Oram,and E. L. Bierman. 1990.Nonenzymatic glycosyl-ation of HDLresultingin inhibition ofhigh-affinity bindingtocultured human fibroblasts. Diabetes. 39:1257-1263.