Proteoglycan form of macrophage colony stimulating factor binds low density lipoprotein

Proteoglycan form of macrophage

colony-stimulating factor binds low density lipoprotein.

S Suzu, … , F Kimura, S Kondo

J Clin Invest.

1994;

94(4)

:1637-1641.

https://doi.org/10.1172/JCI117506

.

We recently isolated a proteoglycan form of macrophage colony-stimulating factor

(PG-M-CSF) that carries a chondroitin sulfate glycosaminoglycan chain. Here, we examined the

interaction of PG-M-CSF with low density lipoprotein (LDL). When LDL preincubated with

PG-M-CSF was fractionated by molecular size sieving chromatography, it was eluted earlier

than untreated LDL. When LDL was preincubated with chondroitin sulfate-free 85-kD

M-CSF instead of PG-M-M-CSF, the elution profile of LDL remained unchanged, indicating

specific interaction between PG-M-CSF and LDL. The level of PG-M-CSF binding in the

wells of a plastic microtitration plate precoated with LDL was significant, this binding being

completely abolished by pretreatment of PG-M-CSF with chondroitinase AC, which

degrades chondroitin sulfate. The addition of exogenous chondroitin sulfate or

apolipoprotein B inhibited the binding of PG-M-CSF to LDL in a dose-dependent manner,

indicating that the interaction between PG-M-CSF and LDL was mediated by the binding of

the chondroitin sulfate chain of PG-M-CSF to LDL apolipoprotein B. PG-M-CSF was also

demonstrated in the arterial wall, and there were increased amounts of PG-M-CSF in

atherosclerotic lesions. The in vitro interaction between PG-M-CSF and LDL thus appears

to have physiological significance.

Research Article

Proteoglycan

Form of

Macrophage Colony-stimulating Factor Binds Low

Density Lipoprotein

Shinya Suzu,*iI Toshimori Inaba,* Nobuya Yanai,* Takuji Kawashima,* Nobuhiro Yamada,* Teruaki

Oka,'

Rikuo Machinami,§ Tetsuya Ohtsuki,11 Fumihiko Kimura,1 Shuichi Kondo,1 Chikao

Torikata,I

Naokazu Nagata,"

and Kazuo

Motoyoshi"I

*The Biochemical Research Laboratory, Morinaga MilkIndustry Co., Ltd., Kanagawa 228; tThe ThirdDepartment ofInternalMedicine and§The Department of Pathology, Faculty of Medicine, University of Tokyo, Tokyo 113; and

IlThe

ThirdDepartment ofInternal Medicine and 1TheDepartment of Pathology, NationalDefenseMedicalCollege, Saitama359, Japan

Abstract

We recently isolated a proteoglycan form of macrophage

colony-stimulating factor (PG-M-CSF) that carriesa

chon-droitin sulfate glycosaminoglycan chain. Here,weexamined

the interaction of PG-M-CSF with low density lipoprotein (LDL). When LDL preincubated with PG-M-CSFwas

frac-tionated by molecular size sieving chromatography,it was

elutedearlier than untreated LDL. When LDLwas

preincu-bated with chondroitin sulfate-free 85-kD M-CSF instead of PG-M-CSF, the elution profile of LDL remained

un-changed, indicating specificinteraction betweenPG-M-CSF

and LDL. The level ofPG-M-CSF bindinginthe wells ofa

plasticmicrotitrationplate precoatedwithLDLwas

signifi-cant, this binding being completely abolished by pretreat-ment of PG-M-CSF with chondroitinase AC, which de-grades chondroitin sulfate. The addition of exogeneous

chondroitin sulfateorapolipoprotein Binhibited the

bind-ing of PG-M-CSFtoLDL inadose-dependentmanner,

indi-cating that the interaction between PG-M-CSF and LDL

wasmediatedbythebindingof thechondroitin sulfate chain

of PG-M-CSF to LDL apolipoprotein B. PG-M-CSF was

also demonstrated in the arterial wall,and there were in-creased amounts of PG-M-CSF in atherosclerotic lesions.

Theinvitro interaction betweenPG-M-CSF and LDL thus

appears tohave physiological significance. (J. Clin. Invest. 1994.94:1637-1641.) Keywords:colony-stimulating factor

* chondroitin sulfate *glycosaminoglycan - atherosclerosis *macrophage

Introduction

Proteoglycans (PG)' aremacromoleculescomposedof various

kinds ofglycosaminoglycan (GAG) chains and various kinds

ofcoreproteins,whose moietiesarecovalently bound with each

AddresscorrespondencetoKazuoMotoyoshi,M.D., The Third

Depart-mentofInternalMedicine,National DefenseMedicalCollege, Namiki 3-2, Tokorozawa, Saitama359, Japan.

Receivedfor publication 13 October1993 andin revisedforn 1

June1994.

1.Abbreviations usedinthispaper:GAG, glycosaminoglycan; M-CSF,

macrophage colony-stimulating factor; PG, proteoglycan.

other (1,2). Common GAG include heparansulfate, chondroi-tinsulfate, dermatan sulfate, and keratan sulfate ( 1, 2), thecore

proteins of these being different. ThePGare ahighly diverse group of macromolecules; in only a limited number has the

molecularstructurebeenestablished (1, 2).

PG arewidely distributed in animal tissues. The increased

amountsof certain arterial PG in atheroscleroticaortasandtheir

affinity for low density lipoprotein (LDL) have lead various investigatorstosuggest that these PGareinvolved in the athero-sclerotic process (3-7). Supporting evidence for the

involve-ment of the PG-LDL complex in the atherosclerotic process has beenprovided by the isolation of this complex from human vascular lesions (8). Interaction between PG and LDL is thought tobemediated by the binding of anionic GAG chains

to thepositively charged amino acid ofapolipoprotein B

(3-7). However, previous studies have also indicated that both

coreproteins and GAGareessentialfor the formation of these complexes and that the potential of PG to interact withLDL

dependsonthe natureof thecoreproteins andonthe

composi-tionsof the GAG (4, 8, 9). The interaction between PG mole-cules andLDLisimportantinthepathogenesis of atherosclero-sis, since this interaction may facilitate the trapping ofLDLin thearterial wallor causestructural alterations in theLDL pro-tein, alterations that influence the uptake oflipoprotein bythe

arterial mesenchyma (10-16). To date, however, no specific

PG molecule involvedin LDLbinding hasbeen characterized yet. To clarify the precise role of PG in the pathogenesis of atherosclerosis, it is important toidentify PG molecules with such functions.

Macrophage colony-stimulating factor (M-CSF), a growth factor for mononuclearphagocytic cells (17),was first identi-fied as a glycoprotein with a molecular mass of 85 kD (18, 19). A second M-CSF molecule, with a molecular mass of

> 200kD,andwhich carries chondroitin sulfateGAG, has since been identified byus and others (20,21); thiswas designated "proteoglycan form of M-CSF" (PG-M-CSF) (22). Human

M-CSF is producedas a522-amino acid polypeptide preceded by a 32-amino acidsignal peptide (19). Theprecursor poly-peptides arerapidly dimerized via disulfide bonds and transla-tionally glycosylated (19). Proteolytic cleavage at theresidue

around 223 produces anM-CSF subunitof 43 kD(19). The

85-kD M-CSF is ahomodimer of the 43-kD subunit (19). In

contrast, proteolytic cleavage occurring at the residue around 400, butnot atthataround223, producesaPG-M-CSF-specific

subunit with a molecular mass of 150-200 kD (20, 21). In

addition to having alonger carboxylterminus than the 43-kD

subunit, thePG-M-CSF-specificsubunit containsachondroitin sulfatechain ontheCOOH-terminal portion (21). It has also

been shown that the combination of the PG-M-CSF-specific J.Clin. Invest.

© TheAmerican Societyfor ClinicalInvestigation,Inc. 0021-9738/94/10/1637/05 $2.00

subunit with the identical subunitorthe 43-kD subunit lacking the COOH-terminal structures yielded homodimericor hetero-dimeric PG-M-CSF, respectively (22). Here, we show the ca-pacity of this newly identified PG, i.e., PG-M-CSF, to bind plasma LDL.

Methods

Materials.The 85-kD M-CSF and PG-M-CSF were purified from media conditioned with Chinese hamster ovary cells transfected with human M-CSF cDNA (19, 21, 22). Human LDL (d= 1.019-1.063 g/ml) was prepared byultracentrifugation from normal plasma. Chondroitin sulfate A(whalecartilage), chondroitin sulfateB(hog skin), chondroitin

sul-fate C (shark cartilage), and heparan sulfate (bovine kidney) were purchasedfromSeikagaku Kogyo Co., Ltd. (Tokyo, Japan); purified

human apolipoprotein B was purchased from Chemicon International, Inc. (Temecula, CA).

Formationof PG-M-CSF and LDL complex. Two types of

associa-tions between PG-M-CSF andLDL wereexplored.

First,solublecomplexeswereobtainedby mixingsolutions ofLDL with solutions ofM-CSF.LDL(100_ogprotein), 85-kD M-CSF (100

jig),

orPG-M-CSF(100

_jig

protein)wasdissolved in 1 mlof

phos-phate-buffered saline (PBS).LDL(100

_jig)

wasmixed with 85-kD

M-CSF(100,ug)orPG-M-CSF(100

_jig)

and dissolvedin 1 mlof PBS.

These solutions were left at 370C for 2 h. A 10-Ml aliquot of each

solutionwasloadedon aTSKG3000SW column(ToyoSoda,Tokyo, Japan) and then eluted with 10 mMTris-HCl buffer(pH 7.4) containing 150 mMNaCl. Fractions (400

_0l/fraction)

were monitored for

ab-sorbanceat280nm(23) and assayed for M-CSFcontentusingan

M-CSF-specific enzyme-linkedimmunosorbent assay(ELISA) (22, 24). Second,thebinding of PG-M-CSFtoimmobilizedLDLwas

exam-ined. LDL(10Mgprotein/mlinPBS) wasaddedto thewells (0.1 ml/

well) ofamicrotiter plate (3950; Costar Corp., Cambridge,MA), and theplatewasincubated for 1 hat37°C(7,22). The wells werethen washed three times with PBS.Allwashingstepsdescribedbelow were donein thesame manner.After thewashing, the uncoated sites of the

wellswereblockedby incubationwith 3%bovineserumalbumin (BSA)

inPBS for 90min.After furtherwashing, samples dissolvedin

interac-tionbuffer(10mMTris-HClbuffer,pH 7.0, containing 1.5% BSA and 10mMCaCl2)werethenaddedtothe coatedwells, and incubationwas

carriedoutfor 1hat37°C.WeincludedCaCl2 (10 mM)in the interac-tionbuffer,sinceithad beenreportedinanearlierstudythatoptimal bindingbetweenPGand LDLwasobtained in thepresenceof divalent

cations (7). Insomeexperiments,PG-M-CSFwaspretreated with

chon-droitinaseAC( 10mU/ml, EC 4.2.2.4;SeikagakuKogyoCo., Ltd.)(12, 13) and then addedtothewells,as outlinedabove. Potent inhibitors such asGAGs, soluble LDL, or apolipoprotein B were addedtothe

PG-M-CSF solutionimmediately beforethemixturewasaddedtothe wells. TokeepapolipoproteinBsoluble,weincludedsodium

deoxycho-lateintheinteractionbuffer,at afinalconcentration of 10mM,tothe mixture ofapolipoproteinBandPG-M-CSF, accordingtothe

manufac-turer's instructions. Thewells werethenwashed, anti-M-CSF rabbit IgGdissolved in PBScontaining 1.5%BSAwasadded, and the plate

was further incubated for 1 h at 37°C. The plate was then washed, horseradish peroxidase-conjugated goatIgG against rabbit IgG

(Bio-RadLaboratories, Richmond, CA), dissolvedinPBScontaining 1.5%

BSA, wasadded,andtheplatewasincubated for another 1 hat37°C.

After afinalwashing, freshly prepared O-phenylenediamine

dihydro-chloridesolutionwasadded,and the colordevelopmentof the solutions in thewellswasmeasured with absorbanceat492 nm,usingamicroplate reader(Bio-Rad Laboratories).

Characterization of arterialM-CSF.ThepurifiedM-CSF

prepara-tions(1

Mg

each) weredialyzed against bufferA(25 mMTris-HCl,

pH7.4,containing 0.5% CHAPS and 6Murea)and thenappliedto a 1-mlDEAE-Sephacel (PharmaciaLKBBiotechnology Inc., Piscataway, NJ) columnpreviously equilibrated with buffer A. The column was

washedstepwiselywith 2ml ofbufferA,containing 0.1, 0.2, 0.3, 0.4,

0.5, 0.6, 0.7,or0.8 M NaCl. Thefractions,at a1:1,000dilution, were then assayed for M-CSF content(22, 23).ThepurifiedM-CSF prepara-tions were alsoanalyzedbygradient (4-20% acrylamide)SDS-PAGE under nonreducedconditions, followed by immunoblotting (21, 24).

Aortas wereobtained at autopsyfrom five patientswithatherosclerosis, afterfamily members had given their informed consent. The intimal

layers ofthe normal and atheroscleroticregions were peeledoff and

weighed. 10 mlof ice-coldextractionbuffer(25mMTris-HCl buffer, pH 7.4, containing 6M urea,1MNaCl, 0.5% CHAPS, 10mMEDTA, 1mMphenylmethylsulfonyl fluoride, 5mMbenzamidine-HCl,and 10 mMaminocaproic acid)waspouredover1gof the tissues. Thetissues were thenfinely minced with scissorsand extracted for 24 h at40C. Aftercentrifugation, the supernatants(tissue extracts) were collected. Each extract or normal serum (1 ml) wasdialyzed against bufferA,

loadedonto a2.5-mlDEAE-Sephacel column, and then eluted

step-wisely with 5 mlbuffer, as outlined above. Thefractions, at a 1:10

dilution,werethenassayedfor M-CSF contentwith anELISA.

Results

If PG-M-CSF and LDL formedacomplex,it would beexpected

that this complex would have a greater molecular size than

noncomplexedLDL. Toconfirmthis, wecomparedthe elution

profilesof LDL and LDLpreincubatedwith PG-M-CSF in PBS

on a TSK G3000SW column (Toso, Tokyo, Japan). Taking advantageof thenatureoflipoproteins, i.e.,theirrelatively high

value ofabsorbanceat280 nm (23),weanalyzedeach fraction for LDLcontentbymeasuringabsorbanceat280nm.As shown inFig. 1 a, LDLpreincubatedwith PG-M-CSFwas eluted in theinclusive volume of thecolumn,withanelution time earlier thanthatof untreated LDL,indicatingthe formation ofa

com-plexbetween PG-M-CSF and LDL.Acetylated LDL

preincu-bated with PG-M-CSF was also eluted earlier than untreated

acetylatedLDL (data not shown). In contrast, LDL

preincu-bated with 85-kD M-CSF had the sameelutionprofileas that of untreated LDL (Fig. 1 a). Each fraction was also assayed

for M-CSFcontentby ELISA(Fig. 1 b). LDL-treated 85-kD M-CSF had the same elution position as the 85-kD M-CSF, indicating no specific interaction between 85-kD M-CSF and LDL. On the other hand, PG-M-CSF mixed with LDL was

elutedas abroadpeakwithalater elution time than the untreated PG-M-CSF.

Wenextexamined, using amodified ELISA (22, 24),the

binding of PG-M-CSF to LDL previously adsorbed onto the wellsofaplasticmicrotitrationplate.Therewasno

significant

bindingofPG-M-CSFtowellsnotcoated withLDL(datanot

shown),and therewas no

significant binding

of the 85-kD

M-CSFtowells coated with

LDL,

while PG-M-CSF exhibitedan

appreciable capacitytobind LDL(Fig.2a). Pretreating PG-M-CSF with chondroitinaseAC, which

degrades

thechondroitin sulfate chains ofPG-M-CSF,completelyabolished thebinding (Fig. 2 a). When added to the interaction mixture, various GAGspecies,

including

threechondroitinsulfate

preparations,

inhibited PG-M-CSF

binding

totheimmobilized LDL

(Fig.

2

b). Heparansulfate also inhibited the

binding,

but its inhibition

wasless effective thanthatof chondroitin sulfate GAG

(Fig.

2

b). As shown in Fig. 2 c, the interaction of PG-M-CSF with immobilized LDLwaspartlyinhibitedbytheadditionofsoluble

LDL;

apolipoprotein

B,a

protein

present inLDL,alsoexhibited efficientinhibitoryactivity.Takentogether,these

findings

indi-catethat the interaction between PG-M-CSF andLDLwas me-diatedbythe

binding

of PG-M-CSF chondroitin sulfatetoLDL

30N

'U2 0- * *

I0 0

0

b~~~~~~

oPG

* LDL|PG o85kD * LDL | 8kD

"200-C

10

U.

100

0 10 2 0 30 40

Fraction no.

Figure1. Chromatographyof LDL and LDLpreincubatedwith M-CSF

on aTSKG3000SW column.LDL,85-kD M-CSF(85kD),PG-M-CSF

(PG), LDLpreincubatedwith85-kD M-CSF(LDL+85kD),orLDL preincubatedwith PG-M-CSF(LDL+PG)wereelutedon aTSK G3000SW column. The eluted fractionsweremonitoredfor absorbance

at280nm(a)orassayedfor M-CSFcontent_usinganELISA(b).No detectablepeakwasobserved when 85-kD M-CSF and PG-M-CSF fractionsweremonitoredfor absorbanceat280nm,andnosignificant amountsofM-CSFweredetected when LDL fractionswereassayed withanM-CSF-specificELISA. The void-volume fraction is No. 13.

Our

hypothesis,

that PG-M-CSF may bind LDL in the

arte-rial wall invivo,was

supported by

thedetection of PG-M-CSF in the arterial wall

_{(Fig. 3).}

The 85-kD M-CSF and PG-M-CSF showed different behavior on

DEAE-Sephacel

ion-ex-change

column

chromatography (20-22).

As shown in

Fig.

3

a, the

purified

85-kD M-CSF

(inset)

waseluted

mainly

at0.2 M NaCl

concentration,

while the

purified

PG-M-CSF

(inset)

was eluted

mainly

at 0.4 M NaCl concentration.

_Using

this

system,wecharacterized the M-CSF

species

presentin normal human serum and aortic extracts. The

majority

of serum M-CSFwaselutedat0.2 MNaCl,andtherewas afaint appearance ofM-CSFat0.4 M NaCl

(Fig.

3

b),

indicating

that the 85-kD M-CSF was the

major

M-CSF

species

inhuman serum. The

extractsof normal aortic tissue contained M-CSFata

concentra-tion of 1.66-+0.42

ng/g

tissue (n =

_5),

and the extracts of atherosclerotic intima contained

_higher

amounts of M-CSF

(4.04-+1.14

ng/g

tissue, n =

₅₎

than those from the normal

region.

The difference of the M-CSFamountbetween normal

region

and atherosclerotic

region

was

_{statistically significant}

at

the level ofP < 0.002

(Student's

ttest). The M-CSF

species

in theextract of normalaorta was elutedat 0.2 MNaCl, and therewas also asecond

_peak

observed at0.4 MNaCl, with a

0.5 a Figure 2. Binding of

PG-M-CSFtoimmobilized

E

'0.4/ LDL.Purified 85-kD

M-ob / CSF

(85kD),

PG-M-_.3t *PG_o CSF (PG), and

PG-M-PG-Chon CSFpretreatedwith c

0.2

/0o

85kD

chondroitinase AC

(PG-Chon)werediluted at the

0.1

indicated concentrations

* / _ and added tomicrotiter wells

_previously

coated

0 10 25 50 100 withLDL(a).

PG-M-M-CSF (ng/ml) CSF was mixed with he-paransulfate(HS),

b chondroitin sulfate A

oHS (CSA), chondroitin

sul-1001 *CSA CAcoditnsl

OCSB fateB (CSB),or chon-. CSC droitinsulfate C(CSC)

.' \\\ \o (b),orwithLDL or

apo-lipoproteinB(ApoB)

c 50 (c).In the case of the

mixture of

apolipopro-tein B andPG-M-CSF,

sodiumdeoxycholate

0 - wasincluded in the

solu-0 0.1 1.0 10 tion at a final

concentra-Glycosaminoglycan (jg/mI) tion of10mM, as

de-scribed inMethods.The

c presenceofsodium

10OiLDL deoxycholate atthe

con-A

\

*ApoB

centration hadnoeffect

C

onPG-M-CSF

binding

to

immobilized LDL. In bothb andc,thefinal

concentrationof

PG-M-CSFis50ng/ml andthat oftheGAG, LDL,or

apolipoproteinBis indi-00 1 100 cated on the abscissa.

O1 10 I00

Mixtures were added to

LDL or_Apo B

_(ug/ml)

wellscoated with LDL, andtheplatewas

pro-cessedasindicatedina.The level of binding inthe presenceofa

potentialinhibitorwas comparedwith the controlbindingobtained in its absence and was expressed as a percentage of the controlbinding. ThePG-M-CSF samples without thepotentinhibitorsinbandcshowed absorbances of 0.35 and 0.55, respectively.

trailingshoulder at 0.5 M NaCl (Fig. 3c). There was no sig-nificant changeintheM-CSFelutionpattern of atherosclerotic intimalextractcompared with that of the normal aorta, but both peaks containedapproximately twofold larger amounts of M-CSF thanthoseinthe normalaorta(Fig.3d).Thefirstpeakat 0.2MNaClcorrespondingto85-kD M-CSF of atherosclerotic region contained 0.77±0.32 ng M-CSF, whichwashigher than thatof normal region (0.43±0.10 ng) with astatistical

signifi-cance at the level ofP < 0.05. Similarly, the second peak at 0.4 M NaCl corresponding to PG-M-CSF of atherosclerotic region contained 0.66±0.16 ng M-CSF, whichwashigher than

that of normalregion(0.34±0.08 ng) withastatistical

signifi-cance atthelevel ofP < 0.004. These findingsindicatedthat both85-kD M-CSF and PG-M-CSFwerepresent in the arterial wall and that both types of M-CSFwereincreased in

Figure3. _{Chromatography}ofpurified M-CSFs,sera, andextractsof aortic intimaon

DEAE-Sephacel

column. Thepurified85-kD M-CSF_{(hatched bars)}andPG-M-CSF

(open

bars) (a),serafrom five normal

vol-unteers_(b),extractsof normal intima ob-tained from normal aorticregionof five

pa-tients with atherosclerosis (c),andextracts

ofintima obtained fromatherosclerotic

re-gion

from thesameindividuals(d)were

loadedonto

DEAE-Sephacel

column. The bound materialswereelutedthroughthe

col-umn_stepwisely with bufferscontaining

in-creasingconcentrations ofNaCl. The

con-centration of NaCl is indicatedonthe ab-scissa. The eluted fractionswereassayedfor M-CSFcontent,usinganELISA. The 85-kD M-CSF and PG-M-CSFwereanalyzedby SDS-PAGE,followedbyimmunoblotting(a,

inset).

Discussion

Evidence for the existence, in vivo, of the LDL-PGcomplex has been obtained in biochemical andhistochemical studies of aortic lesions (8, 10). The in vitro interaction between LDL and PG extracted from aorta has also been demonstrated in

numerous studies (3-7). However,many otheraspectsof this important interaction remain to be investigated. For example,

Vijayagopaletal. (4) found that thebindingof free GAG chains

toLDLwasweakerthanthatofthe parental PG, while Mourao

etal. (5) showed that removal of thecoreprotein of aortic PG didnotaffect the interaction of the PG with LDL. The mecha-nism whereby macrophage uptake of the LDL-PG complex occurs, however, is not clear; one group reported that LDL modified by PGwasmetabolized via apolipoprotein Breceptors

( 16), while another reported that the complexwasmetabolized via a scavenger-receptor pathway (25). These discrepancies

couldbe duetodifferences in assayconditionsorPG prepara-tions. Most ofthese studies were done withunfractionated or

partially fractionated PGextractedfromtheaorta.Inourstudy,

weclearly showed theaffinity ofPG-M-CSF and LDL (Figs. 1 and 2). To our knowledge, this study represents the first

characterization of a specific proteoglycan molecule involved in thebindingofLDL.Itis possiblethat the apparent reduction

in molecular size of PG-M-CSF preincubated with LDL (Fig. 1 b) is due to structural alterations in the linear GAG chain induced by LDL binding, since the interaction between PG-M-CSF and LDL is mediated by the binding of PG-M-CSF chondroitin sulfate GAG chains to LDL (Fig. 2, a and b).

However, the underlying mechanism in this phenomenon

re-mains to be determined. In any case, the finding supports the conclusion that PG-M-CSFand LDLinteract and forma

com-plex.

Earlier studies have demonstratedtheimportanceofM-CSF

in theatheroscleroticprocess.First,severalrecentstudies have

reportedthe elevatedexpressionof the M-CSFgenein vascular cell components ofatheroscleroticlesions,including endothelial

cells, monocytes/macrophages, and smooth muscle cells (26-29). Ourfindings presented here, i.e.,the presenceof

PG-M-CSF and the 85-kD M-PG-M-CSF in the aortic tissue and theincreased amounts of these M-CSF in atherosclerotic lesions (Fig. 3, c

and

d),

are in accordance with theseobservations. Since little

PG-M-CSFwaspresent in serum (Fig. 3 b), theincrementof

PG-M-CSF amountin atheroscleroticlesions mightreflect the

coaccumulationofPG-M-CSF withLDL. Second, ithas been

reported that the 85-kD M-CSF stimulates lipoprotein lipase

secretion frommacrophagesand that itenhances boththeuptake

and efflux ofcholesterol in human monocyte-derived

macro-phages (30). Both PG-M-CSF and the 85-kD M-CSF act on

monocytes,asdescribed previously (22); thus,itcanreasonably

be assumed that PG-M-CSF modulates LDL metabolism in

macrophages.

Although thebiologically activenature ofPG-M-CSF and itsbindingofLDLsuggestthat this molecule makesaprofound

contribution to theatherosclerotic process, this contribution is stillspeculative. However, ourstudy offersanew opportunity

tostudythenatureof theinteractionbetween PG and LDL and thesubsequent metabolismof such complexes inmacrophages.

Acknowledgment

This workwassupportedinpartbygrantsfrom the Ministry of

Educa-tion, Science and Culture,Japan.

References

1.Ruoslahti, E., and Y. Yamaguchi. 1991. Proteoglycansasmodulatorsof growth factor activities. Cell. 64:867-869.

1.5 b

M

c

LL

iA

Ui

i1.0

0.5 0.4

0)

LL An

u

0.2-0

1.0'

-.0

0)

u.5

iE

NaCI (M)

0

2. Kjellen, L., and U. Lindahl. 1991. Proteoglycans: structures and interactions. Annu. Rev.Biochem60:443-475.

3. Camejo, G., H. Acquatella, and F. Lalaguna. 1980. The interaction of lowdensity lipoproteins with arterial proteoglycans: anadditional riskfactor? Atherosclerosis. 36:55-65.

4.Vijayagopal,P.,S.R. Srinivasan, B.Radhakrishnamurthy,and G. S. Beren-son. 1981. Interaction of serum lipoproteins and a proteoglycan from bovine aorta. J. Biol.Chem. 256:8234-8241.

5.Mourlo, P. A. S., and C. A. Bracamonte. 1984. The binding of human aorticglycosaminoglycansandproteoglycans to plasma low density lipoproteins.

Atherosclerosis. 50:133-146.

6.Owens, R. T., and W. D. Wagner. 1991. Proteoglycans produced by choles-terol-enriched macrophages bind plasma low density lipoprotein. Atherosclerosis.

91:229-240.

7.Christner,J.E.,andJ. R. Baker. 1990. Acompetitiveassayoflipoprotein: proteoglycan interaction using a96-well microtitration plate. Anal. BiochemL 184:388-394.

8.Steele,R. H., W. D.Wagner,H. A.Rowe,andI.J.Edwards. 1987. Artery wall derived proteoglycan-plasma lipoprotein interaction: lipoprotein binding properties of extracted proteoglycans. Atherosclerosis. 65:51-62.

9. Wegrowski, J.,M.Moczar, and L. Robert.1986. Proteoglycan frompig aorta: comparative study of their interactions with lipoproteins. Biochem J. 235:823-831.

10.Srinivasan,S. R., B.Radhakrishanamurthy,E. R.Dalferes,Jr., and G. S. Berenson. 1979.Collagenase-solubilized glycosaminoglycancomplexof human aorticfibrousplaque lesions.Atherosclerosis. 34:105-118.

11.Toledo, 0.M.S.,and P. A. S. Mourto.1980.Sulfated glycosaminoglycans

in normal aortic wall of different mammals. Artery. 6:341-353.

12. Alavi, M., and S. Moore. 1985. Glycosaminoglycan composition and

biosynthesisin the endothelium covered neointima ofdeendotheialized rabbit aorta.Exp.Mol. Pathol.42:389-400.

13.Salisbury,G. B. J., D. P.Hajjar,and C. R. Minick. 1985. Altered

glycos-aminoglycanmetabolism ininjured arterialwall. Exp. Mol.Pathol.42:306-319.

14.Hoff,H.F.,and W. D. Wagner. 1986. Plasma low density lipoprotein

accumulation in the aortas ofhypercholesterolemicswine correlates with modifi-cations in aorticglycosaminoglycancomposition.Atherosclerosis.61:231-236.

15.Salisbury,G. B.J.,D.J.Falcone,and C. R.Minick. 1985. Insoluble low

density lipoprotein-proteoglycancomplexes enhancecholesterylester accumula-tion in macrophages. AnL J.Pathol.120:6-11.

16. Hurt-Camejo, E.,G.Camejo,B.Rosengren,0. Wiklund,and G.Bondjers. 1990.Arterialproteoglycansincrease the rate of oxidation of low density

lipopro-teins anditsuptakeby macrophages.Circulation. 82:11l-559a.(Abstr.) 17. Clark,S. C., and R. Kamen. 1987.The human hematopoietic colony-stimulatingfactors. Science (Wash. DC).236:1229-1237.

18.Motoyoshi,K., F. Takaku, H.Mizoguchi,and Y. Miura. 1978.Purification

andsomeproperties of colony-stimulatingfactor from normal human urine.Blood. 52:1012-1020.

19.Wong,G.G.,P. A.Temple,A. N.Leary,J. S.Witek-Giannotti,Y.-C.

Yang,A.B.Ciarletta, M.C.Chung,P.Murtha,R.Kriz,R. J. Kaufman,etal. 1987. Human CSF- 1: molecularcloningandexpressionof 4-Kb cDNAencoding

the humanurinary protein.Science(Wash. DC). 235:1504-1508.

20.Price,L. K.H.,H.U.Choi,L.Rosenberg,and E.R.Stanley.1992. The

predominantform of secretedcolony-stimulating factor-l is a proteoglycan. J. Biol.Chem. 267:2190-2199.

21.Suzu, S., T.Ohtsuki,N.Yanai,Z.Takatsu,T. Kawashima,F.Takaku,

N.Nagata,and K.Motoyoshi. 1992. Identification of ahighmolecularweight macrophage colony-stimulatingfactoras aglycosaminoglycan-containing species.

J.Biol. Chem 267:4345-4348.

22.Suzu, S.,T. Ohtsuki, M. Makishima, N. Yanai, T. Kawashima, N.Nagata,

and K.Motoyoshi.1992.Biological activityofaproteoglycanform ofmacrophage

colony-stimulating factor and its binding totype V collagen. J. Biol. Chem

267:16812-16815.

23.Hara, I.,andM. Okazaki. 1986.High-performance liquidchromatography

of serumlipoproteins.Methods Enzymol. 129:57-78.

24. Suzu, S.,N. Yanai, Y. Sato-Somoto, M. Yamada, T. Kawashima, T.

Hanamura,N.Nagata,F.Takaku,and K.Motoyoshi. 1991. Characterization of

macrophage colony-stimulating factor inbodyfluids by immunoblotanalysis.

Blood 77:2160-2165.

25. Vijayagopal, P., S. R. Srinivasan, B. Radhakrishnamurthy, and G. S.

Berenson. 1993. Human monocyte-derived macrophages bind low-density-lipo-protein-proteoglycan complexes by areceptor different from the low-density-lipoproteinreceptor. Biochem J. 289:837-844.

26.Rajavashisth,T.B.,A.Andalibi,M.C. Territo, J. A. Berliner, M.Navab,

A.M.Fogelman,and A. J. Lusis. 1990. Induction of endothelial cell expression of

granulocyteandmacrophage colony-stimulatingfactorsbymodifiedlow-density

lipoproteins.Nature (Lond.). 344:254-257.

27. Rosenfeld, M. E., S. Yla-Herttuala, B. A. Lipton, V. A. Ord, J. L.Lipton,

andD. Steinberg. 1992. Macrophagecolony-stimulatingfactor mRNA andprotein

inatherosclerotic lesions of rabbits and humans. Am. J. Pathol. 140:291-300. 28.Clinton, S. K., R.Underwood,L.Hayes,M. L.Sherman,D. W. Kufe, and P.Libby. 1992.Macrophage colony-stimulatingfactor geneexpression in vascular cells and in experimental and human atherosclerosis. Am. J. Pathol.

140:301-316.

29.Shimada,M., T.Inaba,H.Shimano,T.Gotoda,Y.Watanabe,K. Yama-moto, K.Motoyoshi, Y. Yazaki, and N. Yamada. 1992. Platelet-derived growth factor BB-dimer suppresses the expression of macrophagecolony-stimulating

factor in human vascular smooth muscle cells. J. Biol. Chem267:15455-15458.

30. Ishibashi, S., T. Inaba, H. Shimano, K.Harada,I.Inoue,H.Mokuno,N.

Mori,T.Gotoda,F.Takaku, and N. Yamada. 1990. Monocytecolony-stimulating