Interaction of monocytes with cocultures of human aortic wall cells involves interleukins 1 and 6 with marked increases in connexin43 message

Interaction of monocytes with cocultures of

human aortic wall cells involves interleukins 1

and 6 with marked increases in connexin43

message.

M Navab, … , D C Drinkwater, A M Fogelman

J Clin Invest.

1991;

87(5)

:1763-1772.

https://doi.org/10.1172/JCI115195

.

Medium from cocultures of human aortic endothelial cells (HAEC) and smooth muscle cells

(HASMC) taken from the same donor contained approximately two- to fourfold more

macrophage colony-stimulating factor, granulocyte/macrophage colony-stimulating factor,

and up to 5.1-fold more transforming growth factor beta than could be accounted for by the

sum of the activities of media from equivalent numbers of HAEC and HASMC cultured

separately. After pulse labeling, immunoprecipitated [35S]fibronectin and [14C]collagen

were also found to be substantially increased in the coculture compared to the sum of

HAEC and HASMC cultured separately. The cocultivation of HAEC and HASMC resulted in

a 2.7-fold increase in connexin43 messenger RNA. When direct physical contact between

HAEC and HASMC was prevented by a membrane that was permeable to medium, the

levels of [35S]fibronectin and [14C]collagen in the coculture were significantly reduced.

Monocytes cultured alone contained low levels of [35S]fibronectin and [14C]collagen but

when added to the coculture there was up to a 22-fold increase in [35S]fibronectin and a

1.9-fold increase in [14C]collagen compared to the coculture alone. The increase in

fibronectin was prevented in the presence of neutralizing antibody to interleukin 1 and

antibody to interleukin 6 by 45% and 67%, respectively. Addition of monocytes to cocultures

also induced the levels of mRNA for connexin43 by 2.8-fold. We conclude that the

interaction of HAEC, HASMC, and monocytes […]

Research Article

Interaction of

Monocytes

with Cocultures

of Human Aortic Wall Cells

Involves

Interleukins 1 and 6

with

Marked

Increases in Connexin43

_Message

Mahamad Navab,*FengLiao,* GregoryP.Hough,*Lon A.Ross,*BrianJ. VanLenten,*TripathiB.Rajavashisth,t

Aldons J.Lusis,*Hillel Laks,' Davis C.Dnnkwater,'andAlanM.Fogelman*

Division ofCardiology, DepartmentofMedicine,*and Division of CardiothoracicSurgery, Department ofSurgery,5 SchoolofMedicine, University ofCalifornia,LosAngeles, California 90024-1679;andDivision ofMedicalGenetics,*

Harbor-UCLAMedical Center, Torrance, California90502-2064

Abstract

Medium from cocultures of human aortic

endothelial

cells (HAEC) and smooth muscle cells

(HASMC)

taken

from

the samedonor contained

approximately

two- to

fourfold

more mac-rophage colony-stimulating factor,

granulocyte/macrophage

colony-stimulating factor,

and up to

5.1-fold

more transforming growth

factor

fi

than could be accounted

for

by the sum of the

activities

of

media

from

equivalent

numbers of HAEC and

HASMC

cultured separately.

After

pulse labeling, immuno-precipitated

I35Slfibronectin

and

_{I'4Cicollagen}

werealsofound tobe

substantially increased

in the

coculture

compared tothe sumof

HAEC

and

HASMC

cultured separately. The coculti-vation of HAEC and HASMC resulted in a 2.7-fold increase in

connexin43

messenger RNA.When

direct physical

contact be-tween

HAEC

and

HASMC

wasprevented by a membrane that was permeable to medium, the levels of

['Sfibronectin

and

114Cjcollagen

in the coculturewere

significantly

reduced. Mono-cytescultured alone contained low levels of

_{PI5Slfibronectin}

and

['4Cjcollagen

butwhen addedtothe coculture therewasupto a

22-fold increase

in

I3Slfibronectin

and a

1.9-fold increase

in

[14Cjcollagen

compared

tothe coculture alone. The increase in

fibronectin

wasprevented in the presence of

neutralizing

anti-body

to

interleukin

1 and

antibody

to

interleukin 6

by 45% and

67%, respectively. Addition

of monocytes to cocultures also in-duced the levels of mRNA for connexin43 by 2.8-fold. We con-cludethat the

interaction

of

HAEC, HASMC,

and monocytes in coculture can result in marked increases in the levels of sev-eral

biologically important

molecules and that

increased

gap

junction formation

between the cells and

interleukins

1 and6 may bepartially responsible

for

these changes. (J. Clin. Invest. 1991.

87:1763-1772.)

Key

words:

collagen

-

colony-stimulating

factor *

endothelial

cell * fibronectin * smooth muscle

Introduction

Under

normal

conditions

thereare

only

a

limited

number of

smooth muscle cells (SMC)1 and

monocytes present in the

sub-Addressreprint requests to Dr. Navab, Division of Adult Cardiology, Room 47-123, Center for the Health Sciences, UCLA School of Medi-cine,Los Angeles, CA 90024-1679.

Receivedfor publication9 March 1990 and inrevisedform 7 Jan-uary 1991.

1.Abbreviationsused in this paper:COL/SMC,human aortic smooth muscle cell layers grownon afilter and covered withacollagen layer;

endothelialspace

(1).

However,

in thecourseof

development

ofan atherosclerotic lesion there isa marked

migration

of monocytes and SMC into the subendothelial space

(2).

The presence of endothelial cells

(EC),

SMC,

and monocytes in suchaconfinedspace

might

be

predicted

tofavor cellular inter-actions. In tissue

culture,

the interaction of EC and SMC has been showntoleadtoincreased low

_{density lipoprotein (LDL)}

receptor

activity (3),

increased

lysosomal cholesteryl

esterase

activity (4),

and reducedratesof SMC

_{proliferation (5).}

The coculture of SMC and EC has also been shown toalter the

amount and the

composition

of the extracellular matrix

_(6).

Recently

two laboratories

(7, 8)

have shown that bovine EC and

pericytes

or

_SMC,

when cultured

together, produced

active

transforming growth factor-fl (TGF-f,)

in the culture medium. Both EC

(9)

and SMC

(10)

have been shownto express message for

connexin43,

amember ofthe

family

ofrelatedgap

junction

proteins.

These

junctions

aremembranestructuresinvolved in intercellular communication and have been

implicated

in

buffering

of

cytoplasmic

ions

(1

1), synchronization

of cellular behavior suchascontraction of

myocardial

cells

(12),

growth

control and

embryonic

differentiation

(13),

and suppression of deleterious effects of somatic cell mutations ina

variety

of

en-zymes

(14).

The induction of connexin43 as a result of the interaction of

EC, SMC,

and monocytesin coculture hasnot

been

previously reported.

In thepresent reportwedemonstrate that the coculture ofadult human aortic EC

(HAEC)

with adult human aortic SMC

(HASMC)

taken from thesamedonor re-sulted in increased levels ofseveral

biologically important

mole-cules when the cellswerecultured under conditions

permitting

direct cell-cellcontact.

Moreover,

weshow that the addition of human monocytes tothe coculture markedly enhanced the levels of these matrixcomponents. We alsopresentevidence

indicating

that cell-cell interaction results in increased induc-tion of mRNA for connexin43 and that interleukins 1 and 6

(IL-

1 and

IL-6)

areinvolved in increased levels of matrix pro-teins.

Methods

Materials. Tissue culture medium, reagents, fetal bovine serum(FBS), andpooledhumanserum(HS)wereobtained fromsourcesdescribed

(15).

Microporous filters, Transwells,andplasticwarewere

_purchased

EC/COL, endothelial cells grown on a layerof collagen on a

filter;

EC/COL/SMC,amultilayerofsmooth musclecells grown ona

filter,

covered withalayerof collagen, withanendothelial monolayer grown

onthe surface of collagen; FBS, fetal bovine serum; FN,fibronectin; GM-CSF,granulocyte/macrophage colony-stimulating factor; HA, human aortic;HI,heat-inactivated; HS, pooled human serum;IDME, Iscoves modified Eagle's medium; M-CSF, macrophage colony-stimu-lating factor; RI,radiolabeled andimmunoprecipitated;SMC, human aortic smooth muscle cells;TGF-fl,transforming growth factor-,B.

Interaction of J. Clin. Invest.

© The American Society for Clinical Investigation, Inc. 0021-9738/91/05/1763/10 $2.00

from CostarCorp., Cambridge, MA, and from sources reported (16). Purified, recombinant human macrophage colony-stimulating factor (M-CSF) andgranulocyte/macrophage CSF (GM-CSF) were kindly provided by Dr.Steve Clark, Genetics Institute, Cambridge, MA. Neu-tralizing antibody to M-CSF was a gift of Dr. E. R. Stanley, Albert Einstein College of Medicine, Bronx, NY; and neutralizing antibody to GM-CSFwaskindly provided by Dr. Judith Gasson, University of California, Los Angeles.

TGF-31

(100-B) and neutralizing antibody to

TGF-0

(AB-IO-NA), to IL-l (AB-202-NA), and to IL-6 (AB-2 10-NA) were purchasedfrom R&D Systems, Minneapolis, MN. The AKR-2B cell line was agiftofDr.Harold Moses, VanderbiltUniversity, Nash-ville, TN.[35S]methionine (specific activity 1,254 Ci/mmol, SJ.1015) wasfromAmersham Corp., Arlington Heights, IL; and[U-'4CJ-proline (specific activity 250mCi/mmol, 101 17E) was from ICN Radiochemi-cals, Irvine, CA. Human fibronectin (no. 4008) was obtained from Collaborative Research,Lexington, MA. Anti-human fibronectin an-tibody(IgGfraction,no.0201-0832), was obtained from Cappel Labo-ratories, OrganonTechnika Corp., Malvern, PA; and antibody to hu-man typeIcollagen (no. AB745) was purchased from Chemicon Inter-national, Temecula, CA. Tricine mini-gels (EC 1675) were from Novex,Encinitas, CA. Protein-A Sepharose and collagen type I (C 3511) were obtainedfrom Sigma Chemical Co., St. Louis, MO. The cDNAprobes for connexins32 and 43 (17, 18) were a gift from Drs. Denise Polacek and PeterDavies of University of Chicago and Dr. David L. Paul of HarvardUniversity.

Cells andcultures. HAEC,SMC,and humanperipheralblood monocytes wereisolatedasdescribed(19, 20).Inbrief, polycarbonate filters with

1-,gm

pore diameterwerecoated withgelatinand

assem-bled in theU-shaped chambers ofatransport deviceformingseparate compartmentsoneachsideof the filter (15). HASMCweregrownon

filtersat 1.5 x I05 cells/cm2and allowedtogrowfor 2 d and cover the surface of the filter. The SMCwerethen coveredwithasolution (5

gl/cm2)

of native collagen typeI(0.2% wt/vol)which was allowedto

polymerizeandformathin gel layerat37°C (19). Thecollagen layer was subsequently washedwith culture medium at 37°C and HAEC wereseededat2.5 x 106 cells/cm2andallowedtogrow and forma

completemonolayer ofconfluent EC in 24-48 h(19).Formixed cocul-tures, theprocedure describedbyAntonelli-Orlidgeandcolleagues (7) wasfollowed where EC were addedtosubconfluent SMC and the

re-sultingcoculturewasin theform ofalayerof mixed islands(orpatches) of EC and SMC.Inallexperiments,thepropagatedHAEC,andtheir autologous(matching)HASMC(fromthesamedonor)atpassage lev-els of 5-7wereused.In someexperimentscocultureswereformedon

filters in the form of Transwells. Blood monocyteswereobtained from individualsfromalarge pool ofhealthydonorsbyamodification ofthe Recaldeprocedure(20).Thehuman monocytes isolatedbythe modi-fied Recalde procedure have been demonstratedtobecomparableto

monocytesisolated by conventionalproceduresintermsofcell viabil-ity,proteincontent,purity,and monocyte chemotaxis(19,20).A por-tionof isolated monocyteswascryopreservedingrowthmedium con-taining10%dimethylsulfoxideaccordingtostandardprocedures(21) andwerefoundtobeindistinguishablefrom

freshly

isolated monocytes in adhesion, chemotaxis, and their interaction with HAEC and HASMC.The useofcryopreservedmonocyteswasfoundtobeauseful approach forreducingtheproblemoflargevariabilityindifferent monocyte isolates from variousdonors. Cocultures,ortheir compo-nents, in tissuecultureflasksor onmicroporousfilterswereculturedas

described(19),except that in the presentstudythegrowthmedium (20% heat-inactivated FBS in medium199)wasreplaced with 10-15% heat-inactivated pooled humanserum(HIHS) in Iscove's modified

Ea-gle's medium(IDME). This resulted inimproved purity,growth pat-tern, and cellmorphology. The HAEC and HASMC isolated from the aorticspecimenswereshowntobeessentiallyfree ofcontaminating cells.Thus,the HAECwerepureasdeterminedbymorphologyandby assays for factor VIII-related

antigen,

uptakeof LDL modifiedby ma-londialdehyde,and labeled with 1, I'-dioctadecyl-3,3,3',3'-tetramethyl-indo-carbocyanine perchlorate, and forangiotensin-converting

en-zymeactivity(19).The HASMCweredeterminedtobe pureasjudged

by morphology andby immunohistochemical staining using the

mono-clonalantibody HHF-35 (19).Inallexperiments the number of HAEC and HASMCwereequivalent whether the cells were cultured sepa-rately or together. In all cases the HAEC and HASMC within each experimentwerefrom thesamedonor and aortic specimen. For collec-tion of condicollec-tioned medium forTGF-#bioassays, after three 3-h incu-bations in IDME, the cultures were incubated in IDME withoutserum

present and 24-hconditionedmedia were collected. Protease inhibitors pepstatin (1

lsg/ml),

leupeptin (1 ug/ml), and aprotinin (3

jg/ml),

plus human serum albumin (0.05% wt/vol) were added, the media were centrifugedat 1,500 g for 10min,and the supernatantsolutions were concentrated 10-fold with Centricon units (molecular weight cutoff of 10,000, Amicon Corp., Danvers, MA).Aportion of the medium was transiently acidified (22) before storageat-20'C. Experiments on the activation ofTGF-#by mixed cocultures of HAEC and HASMC were carriedoutina manneridenticaltothatreported by Antonelli-Orlidge andcolleagues (7). For CSF bioassays the medium incubated with the artery wall cells contained 2.5% FBS. After 24h, the conditioned me-diumwascollected, andasfor

_TGF-f3,

proteaseinhibitors were added, sampleswere centrifuged,and supernatant solutionswere

concen-trated and storedat-20'C.ForNorthern blot analyses, coculturesor

theirindividual components wereprepared and after interaction amongHAEC, HASMC, and monocytes, the mediawereremoved, cell layers werewashed with PBS, pH 7.4, and total cellularRNA was

isolated by lysis of cell layers inguanidinium isothiocyanate, phenol-chloroformextraction,andprecipitation (23).10 yg ofeach RNA prep-aration from coculturesorcomponentswasdenatured andsubjectedto

electrophoresison a1.2%formaldehydeagarosegel.Thiswasfollowed byblottingontonylonfilters and UVcross-linking (24).TGF-(3

North-ernanalysiswasperformedasdescribed (24). The blotswerethen washedatafinalstringency of 0.5XSSC/0.1% SDS, at 65°C (IX SSC: 0.15M

NaCl/

15mMsodiumcitrate),andwereautoradiographed.For connexinNorthernanalyses, the blotswereprehybridizedfor 1 hat

65°C,in 20mlsolution of0.75MNaCI,0.15MTris-HCO (pH8.0),10

mM EDTA, 0.1%Ficoll/0.1% polyvinylpyrolidone/0.1% SDS, 0.1% bovineserumalbumin. The bufferwasthendiscarded and the hybrid-izationtocDNAprobes for connexins 43 and 32wasperformedin the

samebuffer for 16 hat55°C.Theprobeswerepreviouslylabeledto a

specific activityof - 109

cpm/,gg

(25)and 2X106cpm/mlwasused for

each blot. The blotswerethen washed for30swith 100 ml of 2XSSC,

werewashed twicewith 500mlof lx SSC for 30 min eachat room

temperature and twice with 0.5x SSC for 30 min eachat55°C,and wereautoradiographed. The autoradiographs were scanned by laser densitometry, signalswerequantitated,and thosefor connexinwere

normalized for a-tubulin values.

Bioassays. Thebiological activity due to CSFs in the conditioned media of the coculturesorthe components(i.e., equivalent numbers of HAECorHASMC cultured separately) was determined by the colony formationassayemployingmousebonemarrowcellsfor M-CSF (26) andhuman bonemarrowcellsfor GM-CSF (27). For calculating the CSFactivityin theconditioned mediafrom cell cultures, the number of colonies produced by medium from collagen alonewasconsidered

asbackgroundandwassubtractedfrom values obtained for medium from cell cultures. Bioassays for TGF-,B based on anchorage-indepen-dentgrowthandcolony formation ofAKR-2Bcells insoft agarwere

performedasdescribed (28, 29).

Metaboliclabeling ofcells. HAEC cultured on collagen (EC/COL)

on afilter(0.6-or4.5-cm2surface area), collagen layeredonHASMC (COL/SMC)on afilter,orthecombination (EC/COL/SMC)on afilter

weremetabolically labeledwith[35S]methionine(200,Ci/ml)for 2 h in methionine-free medium,orwith [14C]proline(20

1ACi/ml)

for 2 h in proline-freemedium in the presence of 2.5% HI serum. The culture mediawerecollected and the proteaseinhibitors (30) were added. The cultures were then washed with medium containing 40 zM nonradioactive methionine and incubated in this medium for 2 h. The chasemedium wascollected, protease inhibitorswereadded, the

me-diumwascentrifugedat350 g, and the supernatantswerekeptat

-20°C. The cell layers onfilters were washed three times with

phate-buffered saline (PBS) pH 7.4, and 1.0 ml of buffer A (30) was added to each of the filters which were subsequently kept atroom temperaturefor 30 min. The cell lysatestogether withmatrices were quantitatively transferred to microcentrifuge tubes containing protease inhibitorsand kept at -20'C.

Immunoprecipitationofradiolabeledfibronectin andcollagen. Be-foreimmunoprecipitation,thethawedmedia were precleared with pro-tein A-Sepharose.Trichloroacetic acid(TCA)-precipitablecounts were

thendetermined (31). To equalTCA-precipitablecounts(50-200 dl) from each sample, was added 10 MAl of anti-human fibronectin IgG fraction or 2 Mlof anti-human collagen IgG fraction and the volume wasadjusted to 500

gl

with buffer A. Immunoprecipitation and gel electrophoresis were then carried out as described (30-32). The gels were fixed, treated with Enhance, and autoradiographed. The levels of theradiolabeled andimmunoprecipitated(RI) fibronectinorcollagen were determined by direct counting of

10-Ml

aliquotsof RI samples before polyacrylamidegelelectrophoresis (PAGE), by laser densitome-try oftheautoradiogramsobtained after PAGE, and bycuttingoutof thegelsof the areas correspondingto RIfibronectin or collagen bands

onautoradiogramsfollowed by counting for their35Sor'4Ccontent.

Addition ofmonocytes tococultures. EC/COL, COL/SMC, or

EC/

COL/SMC onfilters were exposedonthebasalside (sideoppositeto

the cells), to thereference chemoattractant n-formyl-methionyl-leucyl-phenylalanine (FMLP) at a concentration of 10-'M for 60 min at 37°C. This concentrationwaspreviouslyfoundtobeoptimalfor maxi-mum monocytechemotaxis in this system(19).Human monocytes(1

X 106cells/ml.cm2)were presented to the apical side of the cell layers

onfilters and the incubationwascontinued for 60 min. This medium contained 2.5%HIautologousserum.After this step ofincubation,the medium containing the nonadherent leukocyteswasremoved.The cell layers on thefilters were washed withImlof IDMEat37°Cto remove

the cells that wereloosely attachedtotheendothelial monolayer. Hu-man monocytes werealso presented tofilters with collagen matrix alone and exposedontheoppositeside of the filtertoFMLPat a

IO-'

Mconcentration. The culturesonthefiltersweremaintained inIDME

containing 30% HI autologous serum, supplements, and antibiotics (19)for up to 5 d. Cultures were metabolically labeled with [355]-methionine or with

['4C]proline

1-5dafter the addition of monocytes, andthe levelsoffibronectin and collagenweredetermined. In addition

tothe procedure described forimmunoprecipitationof fibronectinor

collagen,asimpleimmunoadsorbtion assaywasdeveloped, basedon

reported procedures (33), forrapid detection of the levelsof radiola-beledfibronectin.Inthis assay antihuman fibronectin antibody (IgG fraction, 15.9 mg/ml) was diluted 1:1,000 in 0.1 M carbonatebuffer, pH9.6. Wells inmicrotiterplates were coatedwith 150 ul ofdiluted antibody. Wellswith nonimmune IgG orwithnoantibodywere in-cludedascontrols. Theplateswereincubatedat4°Cfor 16 h. The wells werethen washed threetimes inaPBS/Tween-20solution (33). The antibodysolutionwasaspiratedand 10Mlsamplestobe testedwere

diluted 1:10(vol/vol)in PBS supplemented with BSA (5 mg/ml) and wereapplied to wells in duplicates of100 uleach. Wellswith excess nonradioactive fibronectinwerealsoincluded as controls. After incu-bation for 16 hat 4°Cthe plateswerewashed five times with PBS-Tween andwererinsedfive times with distilledwater.The wellswere cutout, and the

I'S

content wasdetermined.Inaddition to the

stan-dardgelelectrophoresis procedure used, in later stages ofthese studiesa

tricine mini-gel system (34)wasemployed for the analysesofRI fibro-nectin, which was found to resolve the fibronectin into a pattern that wasspatially identical but qualitatively superior to that obtained by polyacrylamide gels containing 8M urea. Inaddition the mini-gel sys-tem,required only one-third the sample size, and one-tenth the length oftime compared to the standard SDS-PAGE systems. In orderto

determine the level of fibronectin in the conditioned medium ofthe samecoculturesatseveraltime intervalswithout

3"S

radiolabeling,an

established ELISA for fibronectin (33) was used.

Otherprocedures.Measurement ofcell protein content was done by themethod of Lowryetal.(35), anddetermination of cell numberwas

carriedoutby standard procedures (21).

Results

Growthfactor activity. Using the HAEC and HASMC from the

sameheart

donor, multilayer

cocultureswere

established,

con-ditioned mediawere

obtained,

and the

biological

activities of theCSFsweredetermined. For

M-CSF,

after

subtracting

the number of

colonies produced by

mediumincubated with colla-gen

alone,

the medium conditioned

by

cocultureswasfoundto

produce 2.3-fold

morecoloniesthan could be accounted for

by

those from thesamenumber ofHAEC and HASMC cultured

separately

(Fig.

1

A). Preincubation

of the conditioned me-dium with

neutralizing antibody

to M-CSF

prevented

colony

formation, indicating

that the

bioactivity

was indeed due to M-CSF (Fig. 1 A). The coculture medium also contained

3.7-fold more GM-CSF bioactivity compared to the sum of the activities in the medium from HAEC and HASMC

(Fig.

1

B).

Preincubation of the conditioned medium with neutralizing antibodytoGM-CSF prevented colony formation, indicating that the bioactivitywasin fact duetoGM-CSF (Fig. 1

B).

Northern analyses demonstrated that both the HAEC and HASMC contained mRNA for

TGF-fl.

The levels of mRNA from the

cocultures

wereatleast

equal

tothesumofthose from

thecomponents(datanotshown). AKR-2B cells form colonies specifically inresponsetoTGF-(i (28). As demonstrated in

Fig.

1

C,

theaddition

oftransiently

acidified media from the

compo-nentsdidnot

significantly

stimulate colony formation but ad-dition of transiently acidified media from the coculture hada marked stimulatory effect (upto5. 1-fold overthesumofthe

components). The colony formationwasinhibited in the pres-enceofneutralizing antibodytoTGF-f3 butnotwhenan

irrele-vant antibodywas added (Fig. 1 C). In contrast toprevious

reports of cocultures of bovine EC and SMC (7, 8) the in-creased TGF-j activityseeninthe cocultures established for

thepresentstudieswasentirely latent (i.e., required transient

acidification for expression; datanotshown). When, however, conditions employed by Antonelli-Orlidge and colleagues (7), where EC and SMCaregrownnotasmultilayers of confluent

EC and SMC butasmixed cultureswereused, the growth of

HAECwasinhibited by coculture medium indicating the pres-ence ofactive TGF-,3 in these coculture media (data not shown).

Matrixcomponents.The results

of immunoprecipitation of

radiolabeled fibronectin from the media and cell lysates of the componentsand the cocultureare demonstrated in Fig. 2. In

eachinstance equal

TCA-precipitable

countswereappliedto

eachlaneinthe gels. Therewas amarked increase (upto 3.2-foldfor the medium and 2.8-fold for the cells) in the incorpora-tion of ["Sjmethionine into RI fibronectin in the coculture (E +S, lane 3; arrow)ascomparedtothecomponents(E, lane1;

S,

lane 2). Addition of neutralizing antibodytoTGF-,Btothe cultures produced significant increases in the levels offibronec-tinintheculture medium (upto2.2- and 5. 1-fold, from EC andSMC, respectively, panel A) and in the cell layers (upto

2.9-fold and 3.2-fold from EC and SMC respectively, panel B);

compare lanes 4 and I and lanes 5 and 2. Addition of the

antibodytococultures,however, resulted ina73%decrease in

the RI fibronectin levels in the culture medium and less ofa

decrease (48%) in the cell lysates from the coculture (compare lanes 6 and 3). Immunoprecipitation of media and cell lysates from coculture (E+S)usinganonimmune IgG (NI) produced a_{much reduced signal for RI fibronectin levels (6% and 9% of}

controls inthe media andinthecelllayers, respectively);

200

100-0

EC VALUE EC/SMC VALUE EC/SMC EC/SMC SMCVALUE ANTI M-CSF IRREL.IgG

2M.

100'

0-C

L Z

0

O

I-Z

4c

I

T1

_T1

EC VALUE EC/SMC VALUE EC/SMC

+ +

SMCVALUE ANTIGM-CSF

EC/SMC

IRREL.IgG

ECVALUE EC/SMC VALUE ECISMC ECISMC

SMCVALUE ANTITGF-. IRREL.Ig

SOURCEOF CONDITIONED MEDIUM AND ADDITIONS

Figure1.Increased levels of growth factorsincoculturesofhuman

arterywallcells. ConditionedmediafromEC/COL, COL/SMC,and thecocultures,EC/COL/SMC,wereconcentrated 10-fold andwere

tested in CSForTGF-,Bassays.Thecollagen backgroundwas

sub-tractedfrom all of the values and thesumof thevalues forseparate

cultures of EC andSMC iscomparedwith thevalue forEC/SMC

coculture. In theseassaysthecolonyformation insoftagarwas al-lowedtoproceedfor 7-10dafter which thecoloniescontaining50

or morecellswerecounted. AntibodiestoM-CSF, GM-CSF, or

irrelevantIgGwerepreincubatedwith the media andwereaddedto

someofthecultures, designated EC/SMC+ANTIM-CSF, EC/SMC

+ANTIGM-CSF, orEC/SMC+IRREL.IgG. respectively.(A)

Values obtained usingamousebonemarrow assayfor M-CSF bioac-tivity. (B)Resultsobtainedemployingahumanbonemarrow assay

forGM-CSFbioactivity. The valuesshownarethe mean±SD of

trip-licatesamples fromtwo(M-CSF)andthree(GM-CSF)assays.The

valuesaregiven inunitsofCSF.Oneunit ofCSF is theamountof thegrowthfactorthatproducesonecolonyforeach 105cellsplated.

For theTGF-,3 bioassay, conditionedmediawereconcentrated,

tran-siently acidified,andpresentedtoAKR-2Bcells insoftagar(panel C).

Tosomeofthemedia (EC/SMC+ANTITGF-,8),neutralizing

anti-bodytoTGF-,B,50

_gg/ml,

andtosomeothers(EC/SMC+IRREL.

IgG),50,gg/mlofanirrelevant IgGwereadded. Colonies containing

parelanes 7 and3. Immunoprecipitation of samples from co-cultures in the presence of excess (75-100

gg/tube)

nonradioactive fibronectin also produced markedly reduced signals for RI fibronectin levels, downto11%and 16%of con-trols in themedia and in the cell layers, respectively (compare lanes 9 and 3), indicating the authenticity of the fibronectin signal. When HAEC and HASMC were separated byafilter with

0.2-,gm

porediameter in the same chamber and medium (E +

S,

filter)

permitting access of media but preventing EC-SMC contact, there was a marked decrease (down to 9% and 37% ofcontrols) in RI fibronectin levels in the media and in the celllayers, respectively (compare lanes 8 and 3).

The resultsof immunoprecipitation of type I collagen from themedium and cell lysates from cocultures and components is shown in Fig. 3. The coculture (E + S) demonstrated colla-gen levels that were higher (2.7-fold in the media, panel A and 2.4-fold in the cell layers, panel B) than could be accounted for

by

the sum

of

the

contributions

of

the

components (E +

S,

lane 3 vs. E, lane I or

S,

lane 2, arrow). The presence of neutralizing antibody to TGF-3

(+antiTGF-f3,

lanes4-6) reduced the colla-gen levels in the

coculture

medium (by 43%) and in the cell layers by 47% butmarkedly enhanced the amount of collagen

in

HASMC cultures

(1.9-fold for

the

media

and

2.3-fold for

the

cell layers;

compare lanes 5 and 2. The presence ofafilter with

0.2-gm

pore

diameter

between the HAEC and HASMC

(E

+

S,

filter,

lane 8) which permitted access of the media but pre-vented EC-SMC

contact,

reduced the collagen levels by

79%

in themedium from cocultures (E+ S) and by 63% inthe cell layers (compare lanes 8 and

3).

The addition of excess

nonradioactive

collagen to the samples

before

incubation with

anticollagen

(E+

S, COL,

lane

9) produced markedly

reduced

signals (9%

and 13% of controls for the media and the cell layers,

respectively);

compare with lane 3. Incubation of sam-pleswith nonimmune IgG (E +

S,

NI, lane 7) resulted in

mark-edly

reduced levels ofRI collagen (down to 4% and 12% of controls

for the media

and the cell

layers,

respectively;

compare

with

lane

3), indicating the

authenticity of

the

collagen

signal.

Addition

of

human blood monocytes

(M)

to the cocultures

(ES) produced

a

dramatic

increase

(up

to

22-fold)

inthe

fibro-nectin

levels

in the

culture

media

(Fig.

4

A, arrow).

The in-crease intype I

collagen produced by

the interaction of

mono-cyte-macrophages

and the cells of the artery wallwasless pro-nounced

(1.9-fold)

(Fig.

4

B).

The

results

of

immunoradioassay

for fibronectin

(Fig.

5

B)

confirmed the results

generated by

laser

densitometry

of the

autoradiograms obtained

after

gel

electrophoresis

of

immuno-precipitates

(Fig.

5

A), which

wereboth

similar

tothe results

obtained by direct determination of

radioactivity

in

aliquots

of

immunoprecipitates before applying

to

gels

(data

not

illus-trated).

Therewas

complete

agreement between the resultsin

Fig.

5 and those

obtained by

cutting

the relevant bandsoutof the

gels,

extracting

the RI

fibronectin,

and

counting

their35S content

(data

not

shown).

Gap

junction protein

mRNA. Northern

analyses

demon-strated that

interaction

of HAEC and HASMC resultedin

lev-morethan50 cellswerecounted after 10 d ofincubation. The values

shownarethe mean±SD of values fromtriplicatewells in three

AKR-2Bbioassays.

1766 Navabetal.

A

IL

(A

9

L

0

o

z

B

IL

0

t

U)

z

A

E

SE

v+S

E

S

E+S

+antiTGF-J)

NI filt. F4'N\

A..

w

-0.

.1r

6

t,!-4

1 2

3

4

5 6

E':

S

1i:+ S

1i,

S

E+S

+anltiTGF-fJ

+S ATI

filt.

F ki)

200

97

68

43

25

1 2 3

4

5

6

7

8

9

Figure2.Fibronectinlevels in cocultures of

ar-terywall cells. HAECgrown oncollagenona

filter (E, lane 1), HASMCcoveredwithathin layerof collagenonafilter(S, lane2),orthe

combination of HAEC and HASMConafilter

(E+S,lane3)weremetabolicallylabeled with

[35Slmethionine.

Some of the cultures contained

50 ug/mlofneutralizing antibodytoTGF-,B

(+antiTGF-ft,

lanes4-6).Othertreatments

in-cluded theuseof nonimmuneIgGfraction in

placeof antifibronectinantibodyduring

im-munoprecipitation(NI, lane 7),inclusion ofa

microporousfilter(filt., lane8)betweenthe

cocultured HAEC andHASMC,and addition of

anexcess(75 Mug)of unlabeled fibronectinto

somesamplesbeforeimmunoprecipitation (FN, lane9). Equal TCA-precipitablecountswere

takenand the fibronectin in the media(panel A)

and in the celllysates(panel B)wasanalyzed by immunoprecipitationfollowedbySDS-PAGEin

thepresenceof8 Murea.Theautoradiograms

shownarerepresentativeof threeexperiments.

els of mRNA for the gapjunction protein connexin43 that were2.7-foldhigher than could be accounted for by thesumof

the contributions of thecomponents(Fig. 6,compareES,lane 3vs.E,lane I and S, lane 2). Therewas nodetectablesignals

for mRNA for connexin32 (data not shown). Interaction of human monocyteswith HAEC produced enhanced levels of mRNAfor connexin43 thatwere3.7-foldhigherthan could be accounted forby thecontribution of monocytes and HAEC culturedseparately (compare ME,lane5withE,laneIandM, lane 4). Interaction with HASMC inducedanincrease in

mes-sageforconnexin43, whichwas2.1-fold that of the combined

valuesformonocytesandHASMC inseparatecultures (com-pareMS, lane 6 withS, lane 2 and

M,

lane4). Interaction of monocyteswith cocultures of HAEC and HASMC resulted ina

markedincrease in the induction ofmessagefor connexin43to

levels2.8-foldgreaterthan the combined values for cultureof

monocytesand coculture of EC and SMC (compare MES, lane 7with M, lane 4orES, lane 3).

IL-IandIL-6. Addition of neutralizing antibodytoIL-l to

cocultures of EC and SMCdidnotchange thefibronectin levels (Fig. 7). Presence of neutralizing antibodytoIL- I or neutraliz-ingantibodytoIL-6during the interaction ofmonocyteswith cocultures of HAEC and HASMC resulted in 45% and 67% reduction in the levels of immunoreactive fibronectin inthe culture media, respectively (Fig. 7), whereasthe presenceof irrelevantIgG didnotsignificantly change thefibronectin lev-els(Fig. 7). Presence ofantibodytoIL-1 togetherwithantibody

toIL-6didnotincrease the blocking effect ofanti-IL-6 alone (datanot

illustrated). Neutralizing

antibodiestoM-CSF and GM-CSF didnotproduceasignificant

change

inthe increased

kD

200

-F

++S

97

68

43

25

I S E>S S ES

K!)

200-9 7

-E+S

NI

filter

(OL

1

2

3

4

5

6

7

8

9

B

E

S

E+S

E

S

E+S

+a

it

I

i

'(GF-

f

E+S

NI

filter

C'W1-kI)

2

0 0

Usa

9 7-

;:i

1 2

3

4

5

6

7

8

9

Figure3. Collagenlevelsincoculturesofthe

arterywallcells. HAECgrownoncollagen

on afilter(E, lane1)HASMCcovered with

athinlayerofcollagenon afilter(S. lane 2)

orthe combination of the HAEC and HASMCon afilter(E+5, lane3)were

metabolicallylabeled with['4C]proline.

Neu-tralizing antibodytoTGF-fl, 50

Ag/ml,

(+antiTGF-fl,

lanes4-6)wasaddedtosome

of the cultures. Othertreatmentsincluded theuseof nonimmuneIgGfraction inplace

ofanticollagen(NI,lane7), inclusionofa

microporousfilter(filter,lane8)between the cocultured HAEC andHASMC,and addi-tion ofan excess(75 ug)of nonradioactive

collagentosomesamples priorto

immuno-precipitation (COL, lane9). CollagentypeI inthe medium(panel A)and the celllayers (panel B)wasanalyzed by

immunoprecipita-tion ofequalTCA-precipitablecounts fol-lowedbySDS-PAGE in thepresenceof 8 M

urea.Theautoradiogramshown here is

rep-resentative oftwo separateexperiments.(A)

Results for culturemedia; (B)results for cell

layer plusmatrix.

levelsof fibronectinproduced bytheinteraction ofmonocytes

and cocultures of EC and SMC(datanotshown).

Discussion

Innormalarteriesveryfew SMCarefoundinthe

subendothe-lialspace(1). Higher numbersofSMCinassociation with EC

have beenobservedinexperimental atherosclerosis (2) and in atherosclerotic humanarteries (2, 36). Accumulating evidence indicates thatsubstantial direct physicalcontactand

interac-tionoccursbetween ECand SMCinvitro(4, 6-8, 37, 38) and

invivo(38-42).Inourstudies, the interactioninvitro between HAEC andHASMCproduced atwo- to fourfoldincrease in M-CSF andGM-CSFlevels in the culture media(Fig. 1). M-CSFcansupport thegrowthand survivalof

monocyte-macro-phages in vitro evenin the absence ofserum growth factors

(43).Fromthestudy ofautopsyspecimens of humancoronary

arteries, Stary (36) speculated thatmany of the

monocyte-macrophages in the subendothelialspacewould have died

un-less theirsurvival had been supported by growth factors (36).

M-CSF produced by the interaction of EC and SMC could potentiallyactas onesuchfactor. GM-CSF could be another

one.GM-CSF has been showntoprolongthesurvival,

differ-entiation, proliferationanddevelopmentofresponsivecellsin vitro(43). Langetal.(44)demonstrated that theexpressionof

GM-CSFgeneintransgenic mice led to the accumulation of

monocyte-macrophagesintissues suchaseyes,striatedmuscle, and inperitonealcellsexpressingthisgeneandresulted in

sub-stantial tissuedamage (44).

Cocultures of HAEC andHASMC under the conditions employedinthe presentstudyledtoamarked increase in

la-tentTGF-f (Fig. 1 C). Incontrast tothe resultsof

Antonelli-Orlidge et al. (7) and Sato and colleagues (8, 48) all of the activityinourculturemediawaslatent. Sincewewereableto

reproduce the results of Antonelli-Orlidge by using growth-arrested HASMC andculturing HAEC directlymixed with themfollowingtheexactconditions described(7),we cannot

ascribe the different results to the difference in cell types. It

seems more likelythat thearchitecture ofourculture model

(which ismoresimilartothat ofarterywall in vivo) favored the

1768 Navabetal.

A

A

M

ES

EMS FNA

kD

200

_{w_}

UF

25

1

2

3

4 5

6

It is known that TGF-3 can have

different

and complex effectsonthesamecells

under

different conditions

and

canbe NI under

autocrine

regulation

(45-48).

Asshown in Figs. 2 and 3,

neutralizing antibody

to

_TGF-#t

stimulated

HAEC

and

HASMCtoproducefibronectin and collagentypeIwhen

the

cellswerecultured separately. Inclusion of thesame

antibody

~

*

together

with thesamenumberof HAEC and

HASMC

in

co-culture led to a

modest decrease

in the

production

of these

molecules. The increase

in RI fibronectin and collagen in

the

presenceof

TGF-13

antibody in HAEC and SMC is incontrast

tothe well-known effect of

_TGF-,8

onvarious celltypeswhere this growth factor induced fibronectin and collagen message expression or protein formation (47, 49-51). The stateof

growth

of the cells

and

the

extent

of

cell-cellcontact(51),

the

7 composition of extracellular matrix and its activating

potential

A

B

M

ES

EMS COL

NI

150

kD

200

10

I--z

50

93

12

34

56

7

8

Figure4. Effect of monocyte-macrophagesonRIfibronectinand

collagen levels produced byarterywall cells. Coculturesof HAEC and

HASMConmicroporous filters (ES, duplicate lanes2 and 3 inA;

duplicatelanes3and 4 inB)wereexposedonthe basalside,toa 10-7

Msolution of FMLPat370C forIh.Humanmonocytes(I X 106

cells/ml * cm2)werepresentedtofilters alone (M,laneIinA;

dupli-cate lanesIand 2 inB)ortococultures of HAEC and HASMC

(EMS, duplicate lanes4and 5inA;duplicate lanes 5and 6 inB). After 60minat370C the media containing the nonadherent

mono-cyteswereremoved. Subsequentlythe culturesweremaintained in IDME containing 30% HI autologousserumat370C for2d. The cultureswerepulselabeled with[35S]methionine(A)orwith [14C]-proline (B)asdescribedinMethods. Equal TCA-precipitablecounts weretaken andfibronectin(A)orcollagen (B)were

immunoprecipi-tated usingthecorresponding specific antibody (lanes1-6 inA;lanes 1-7 inB)ornonimmuneserum(NI, lane 7 inA;lane 8 inB)or

presenceofexcessnonradioactiveligand(75ggfibronectin, FN,lane 6inA;75ggcollagen, COL, lane 7 inB). Sampleswereanalyzed using10% tricinemini-gels (A)or7.5%polyacrylamide gels

contain-ing 8Murea(B). The autoradiograms presentedhereare

representa-tive ofsevenexperiments (for fibronectin)andtwoexperiments (for collagen).

production of latent TGF-,B. Histological studies of the multi-layer cocultures have demonstrated thepresenceofa

continu-ousand confluent endothelial monolayerontopofathinlayer

of extracellular matrix (1-2jLmthick)overconfluent layers of

SMC (19). After 2-3 d in culture ECwereoften foundtomake frequent direct physicalcontactswith the underlying SMC (un-published observations).

o

B

300

200

V-2 100

0.

0I

T

M E+S E+M+S FN NI

M E+S E+M+S FN NI

SOURCE OF MEDIUM AND ADDITIONS Figure 5. Determination of levels of radiolabeled fibronectin. (A) The autoradiograms obtained from SDS gelswereanalyzedbyalaser

densitometer, and triplicate readingswereobtained and utilizedto

calculate mean±SD for signals correspondingtomonocytes(M), cocultures of HAEC andHASMC (E+S), cocultures withmonocytes

(E+M+S), using nonimmune IgG (NI)oraddednonradioactive

fibronectin(FN). (B)

10-,g

samplesofthe chase media following the metaboliclabeling of the cocultures (E+S) withorwithout

mono-cytes(M)wereused inthe immunoradioassay developedforhuman

fibronectinasdescribedinMethods. The valuesexpressmean±SD of

triplicate samplesand the resultsarerepresentative of three

experi-ments.

Interaction Endothelial, SmoothMuscleCells and

97

(52), the TGF-3 receptor expression and its interaction with

matrix

and

modulation

of

_TGF-j#

binding to cell (53), the level ofTGF-,3 inhibitor and the composition of matrix (54, 55), and

the

combination

of

other growth

factors

(56, 57) have allbeen reported to

affect the

cellular responses

to

_TGF-f3.

According

to

Sporn and Roberts (58), this peptide

growth

factor

can have

both growth

stimulatory

and

inhibitory activity in

a

single

cell,

depending

on

the

context

of

the other

signal

molecules present. Streuliand Bissell (55) found that, in mammary epithelial cells grown on

plastic,

there was a dramatic induction of fibronectin message, whilethe induction was inhibited in the same cells grown on collagen gel. The authors noted that soluble factors such as

TGF-,3

are

increased

and/or

activated

andmediate

changes in

matrix

expression

to

varying

degrees

in

cells cul-tured on

different substratum

(55). Regardless

of

the roleof

TGF-f3,

the

coculture

of

HAEC and

HASMC clearly

led to an

increase in

the

production of

both

fibronectin

and collagen

which

depended on close physical approximation of the two

cell

types

(Figs.

2

and

3).

The

increased induction of

the

gap

junction protein

connexin43

in cocultures

of

EC and SMC

(Fig.

6)

indicates

the

possibility ofincreasedjunctional

commu-nication

and

interaction

between the EC and SMC in these

cocultures

(5, 10,

59).

In

the

presentstudy,

addition of

human monocytes

from

nine different

donors (but not from three

others)

to

the cocultures

amplified

the

production of

the

extra-cellular

matrix molecules

(Figs.

4,

5,

and

7).

Addition of

mono-cytes to cocultures

also

produced

a

marked induction

of

con-nexin43 raising the possibility

ofincreased

direct

heterotypic

cellular communications.

Upon

activation,

monocytes

demonstrate

markedly

in-creased

activities

for

numerous

factors including

for IL- 1 (60).

We have observed

that

during

theprocess of monocyte

trans-migration into

the

subendothelial

space

of

the coculture the

cells

acquire the typical

appearance

of

spread monocytes

sug-gesting

the

activation

of these

cells

(unpublished

observations,

and

reference

61).

Injection of

IL- 1

into

rats has beenreported to

produce elevated

serum

levels

offibronectin

(62).Leukocyte

interleukins

have also been reported to induce cultured EC to

produce

a

highly organized pericellular matrix

(63). IL-1

in-duces IL-6

in

certain

cell types

(64).

Hagiwara

and

co-workers

E

S

ES

M M E MS

MES

CONNEXIN43 6 r 9w 9

CK -TUBULIN

z

7-O 0 C

mr

s-().1)

0.0

-

-F-.

ES ES MES

ANTI IL-1

MES

ANTIIL-1

MES MES

ANTIIL-6 IRREL.IgG

Figure7.Effect of antibody to interleukins onfibronectin levels. Co-culturesofarterywall cellswithmonocytesand culturesof

compo-nents wereprepared and the leveloffibronectin in the culture super-natants wasdetermined inanELISA. Somecultures received 2

4g/ml

ofneutralizingantibodytoeachof IL-1,B (ES+ANTI IL-1,MES

+ANTI IL-1),orIL-6(MES+ANTIIL-6),ornonimmune IgG (MES+IRREL. IgG).The valuesaremean±SDof threeseparate

experiments.*P<0.04; **P<0.027.

(65) demonstrated that monocytes stimulated by IL-1 pro-duced increased levels of IL-6 and that in turn resulted in ele-vated levels

of fibronectin

in culturedhepatocytes. Lanser and Brown

(66)

have

shown that

increased fibronectin production

by

rat

hepatocytes

exposed

to

conditioned medium from

stimu-lated monocytes

is due

to IL-6. In the present

study,

the signifi-cant inhibition of

monocyte-induced

increase in fibronectin levels in

cocultures

of artery wall cells

by

neutralizing

antibod-ies

to IL- 1 and to

IL-6

strongly

suggests the

involvement

of thesecytokines in the monocyte-artery wall cell interactions.

Both

fibronectin

and

collagen

type I

have been implicated

in

atherogenesis

(67). Increased

levels

of

fibronectin

were seen

in

rabbit aorta 4 wk

after cholesterol

feeding (68). Moreover,

substantial levels of fibronectin have been observed in the

ex-tracellular matrix

of

fatty

streaks and in some areas of fibrous

plaque

containing large

numbers

of subendothelial

cells in hu-man

arteries

(69). Similarly,

increases in

collagen

have been shown in

several

experimental

models of atherosclerosis (70).

Additionally, atherosclerotic lesions

of human intima are known to

have

an

increased collagen

content

(7

1)

and

collagen

type I was

reported

tobe the

prominent form

inthe diseased

vessels (72).

The results

of

the

studies reported

here suggest that the

migration of SMC and

monocytes

into the

subendothelial space

of the early atherosclerotic

lesion may leadtocell-cell

interactions

that

result

in

production

of several

biologically

important

molecules that may

amplify

lesion development.

Consistent with

this notionare our recent

findings

that

high

levels

of

M-CSF mRNA are present in the atherosclerotic

le-sions of both Watanabe heritable

hyperlipidemic

and choles-terol-fed rabbits

(73).

Acknowledaments

Wethank Drs. Denise C.Polacek,PeterDavies,and David L. Paulfor 1 2 3 4 5 6 7 providing us with cDNA probes for connexins; Drs. Matt Ashby, Ju-dithBerliner,PeterEdwards, and CatherineClarke, for valuable dis-Figure 6. Connexin43 mRNA. Cocultures ofarterywall cells with 2h cussions;Dr.DavidWoofor

providing

uswithTGF-#cDNA

probe;

ofexposure tomonocytesandcultures ofcomponentswereusedto Dr.

Phillip

Koeffler for assistance with

obtaining

human bonemarrow

isolate total cellular RNA. Afterelectrophoresisand transferto

nylon

cells;FarahElahi and Sima Afrasiabi for their excellent technical assis-filters,sampleswereprobed withlabeled cDNAfor connexin43. The tance;and the members of the UCLAHeart

Transplant

Teamfor col-autoradiogramshown is fromonesuch

hybridized

Northernblot.

lecting

theaortic

specimens.

Thisworkwassupportedinpartbythe U.S.Public HealthService

grantsHL-30568,IT 32HL-074_12,and RR 865 andbytheLaubisch,

Rachel IsraelBerro,and M. K.GreyFunds.

References

1.Rhodin,J.A.G. 1980.Architectureof the vessel wall.InThe

Cardiovascu-larSystemII. D. F.Bohr,A. P.Somlyo,and H.V.Sparks, Jr.,editors.Handb.

Physiol. 1-31.

2._Ross,R.1986. Thepathogenesisofatherosclerosis:an_update.N.Engl.J. Med. 314:488-500.

3.Davies,P.F.,G.A.Truskey,H. B.Warren,S. E.O'Connor,andB. H. Eisenhaure. 1985. Metaboliccooperationbetween vascular endothelial cells and smooth muscle cells in co-culture:changesin lowdensity lipoprotein metabo-lism. J.CellBiol. 101:871-879,1985.

4.Hajjar,D.P.,A.J.Marcus,andK. A.Hajjar.1987.Interactions of arterial cells:studiesonthemechanisms of endothelial cell modulation of cholesterol metabolism in co-cultured smooth muscle cells.J.Biol.Chem.262:6976-6981. 5. HaJar,D. P.,D. J. Falcone,J. B.Amberson,and J. M.Heffon. 1985. Interaction of arterial cells. I. Endothelial cells alter cholesterol metabolism in co-cultured smooth muscle cells.J.LipidRes.26:1212-1223.

6.Merrilees,M.J.,and L.Scott.1981. Interaction of aortic endothelial and smooth muscle cells in culture:effectonglycosaminoglycanlevels. Atherosclero-sis.39:147-161.

7. _{Antonelli-Orlidge,A.,}K. B.Sunders,S. R.Smith,andP. A.D'Amore. 1989.Anactivatedform oftransforminggrowthfactor betaisproduced by

cocul-turesof endothelial cells andpericytes. Proc.Nail.Acad.Sci. USA. 86:4544-4548.

8.Sato, Y.,andD.B.Rifkin. 1989.Inhibition of endothelial cellmovement

bypericytesand smooth muscle cells:activation ofalatenttransforminggrowth

factor beta1-likemoleculebyplasmin duringcoculture. J.Cell Biol. 109:309-315.

9. _Larson,M.,C. C.Haudenschild,and E.C. Beyer. 1990.Gapjunction

messengerRNA_{expression by}vascularwallcells.Circ. Res. 66:1074-1080. 10.Lash, J. A.,E. S. Critser,and M.L.Pressler. 1990.Cloningofagap

junctional proteinfrom vascular smooth muscle andexpressionintwocell

mouse_embryos.J.Biol. Chem. 265:13113-13117.

11.Corsaro,C.M.,andB.R._Migeon.1977. Contactmediated communica-tion of oubain resistance in mammalian cells in culture. Nature (Lond.).

268:737-739.

12. Barr, L., M. M. Dewey, and W. Berger. 1965.Propagation of action potentialand thestructureof thenexusincardiacmuscle. J. Gen.Physiol. 48:797-823.

13.Loewenstein, W.R. 1966. _Permeabilityof membranejunctions.Ann. N.Y. Acad. Sci. 137:441-472.

14.Subak-Share, H.,R.R.Burk,and J. D.Pitts. 1969.Metaboliccooperation

betweenbiochemicallymarkedmammalian cells in tissue culture. J.CellScience. 4:353-367.

15.Navab, M.,G. P.Hough,J.A.Berliner,J. A.Frank,A.M.Fogelman,

M. E. Haberland, and P.A. Edwards. 1986. Rabbitbeta-migratingverylow

density lipoproteinincreasesendothelial macromolecular transport without al-teringelectricalresistance.J.Clin.Invest.78:389-397.

16.Navab, M.,G. P. Hough,B. J.VanLenten,J.A.Berliner,andA.M.

Fogelman. 1988. Lowdensity lipoproteinstransfer bacteriallipopolysaccharides

acrossendothelialmonolayersina_biologicallyactiveform.J.Clin. Invest. 81:601-605.

17.Paul,D. L. 1986. Molecularcloningof cDNA forratlivergapjunction protein.J Cell Biol. 103:123-134.

18.Beyer,E.C., Paul,D.L.,and D. A.Goodenough. 1987. Connexin43:a

proteinfromrathearthomologoustoagapjunction proteinfromliver. J.Cell

Biol. 105:2621-2629.

19.Navab, M.,G. P.Hough,L.W.Stevenson,D.C.Drinkwater,H.Laks,

andA.M._Fogelman.1988.Monocytemigrationintothe subendothelialspaceof

acocultureofadult humanaortic endothelial and smooth muscle cells. J.Clin.

Invest.82:1853-1863.

20._Fogelman,A.M.,F.Elahi,K.Sykes,B. J. VanLenten,M.C.Territo,and

J. A. Berliner. 1988. Modification oftheRecaldemethod for the isolation of human_monocytes.J.LipidRes.29:1243-1247.

21._Freshney,R. I. 1987. Cultureof Animal Cells. AlanR.Liss, Inc.,New

York.216-225,229-232.

22.Wakefield,L.M.,D.M.Smith,T.Massui,C.C.Harris,andM. B.Sporn.

1987. Distribution and modulation of the cellularreceptorfortransforming

growthfactor-beta.J.CellBiol. 105:965-97.

23.Chomczynski,P.,andS. Nicoletta.Singlestep method ofRNAisolation byacidguanidinium thiocyanate-phenol-chloroformextraction. 1987.Anal.

Bio-chem. 162:156-159.

24.Rajavashisth,T.B., R.Eng,R. K.Shadduck,A.Waheed,C.M.

Ben-Avram,J. E.Shively,andA.J.Lusis. 1987.Cloningandtissue-specific expression

ofmousemacrophagecolony stimulating factor mRNA. Proc. Nail. Acad. Sci.

USA. 84:1157-1161.

25. Feinberg, A. P., and B. Vogelstein. 1983. A technique forradiolabeling

DNA restriction endonuclease fragmentstohigh specific activity. Anal. Biochem. 132:6-13.

26.Golde,D.W.,and M. Cline. 1972. Identification of the

colony-stimulat-ing cell in humanperipheral blood. J. Clin. Invest. 51:2981-2983.

27.Metcalf, D., C. G. Begley, G. R. Johnson,N. A.Nicola,M. A.Vadas,A.F.

Lopez, D. J. Williamson, G. G. Wong, S. C. Clark, and E. A. Wang. 1986. Biologic properties in vitro of recombinant human granulocyte-macrophage col-onystimulatingfactor.Blood. 67:37-45.

28. Tucker, R. F., G. D. Shiply, H. L. Moses, and R.W.Holley. 1984. Growth inhibitor from BSC-1 cells closely relatedtoplatelet type betatransforming

growth factor. Science(Wash.DC).226:705-707.

29. Rosa, F.,A.Roberts, D.Danielpour, L. L.Dart,M. B.Sporn,and I. B. Dawid. 1988. Mesoderm induction in amphibians: the role of TGF-beta 2-like factors. Science(Wash. DC).239:783-785.

30. Edwards, P. A., S.-F. Lan, R. D.Tanaka,and A. M.Fogelman. 1983. Mevalonolactoneinhibits the rate of synthesis and enhances therateof

degrada-tion of3-hydroxy-3-methylglutaryl coenzyme A reductase in rat hepatocytes. J. Biol. Chem. 258:7272-7275.

31.Dickinson, E. S., P. A. Dennis, and L. L. Slakey. 1986. Time course of release into themedium ofnewly synthesized proteins by cultured aortic endothe-lial cells. Arteriosclerosis. 6:627-637.

32.Laemmli, U. K. Cleavage of structural proteins during the assembly ofthe headofbacteriophage T. 1970. Nature(Lond.).227:680-685.

33. Rasmussen, L. M., and L.Heickendorff. 1989. Quantification offibronec-tinin extracts of human aorta by an ELISA. Scand. J. Clin. Lab. Invest. 49:205-210.

34.Schagger, H., and G. Von Jagow. 1987. Tricine-sodium dodecyl poly-acrilamide gel electrophoresis for the separation of proteins. Anal. Biochem.

166:368-379.

35. Lowry, 0. H., M. J. Rosebrough, A. L. Farr, and R. J. Randall. 1951. Protein measurement with the Folinphenol reagent. J. Biol. Chem. 193:265-275. 36. Stary, H. 1989. Evolution and progression of atherosclerotic lesions in coronary arteries of childrenandyoung adults.Arteriosclerosis. 9 (Suppl.):l 19-32.

37. Jones, P. 1979. Construction of an artificial blood vessel wall from cul-turedendothelial and smooth muscle cells. Proc. Natl. Acad. Sci. USA. 76:1882-1886.

38.Davies, P. F., S. P. Olesen, D. E. Clapham, C. M. Morrel, and D. J. Shoen. 1988. Endothelial communication: state ofthe art lecture. Hypertension. 1 1:583-582.

39. Thoma,R. 1921. Uberdie Intima der Arterien. Virchows Arch. Pathol. Anat.Physiol. 230:1-45.

40. Bruns, R.R., and G. E. Palade. 1968. Studiesonbloodcapillaries.I. Generalorganizationof bloodcapillariesin muscle. J. Cell Biol. 37:244-276.

41.Huttner, I., M. Boutet, and R. H. More. 1973. Gapjunctions in arterial endothelium. J. Cell Biol. 57:247-252.

42.Spagnoli, L. G., S. Villaschi, L. Neri, and G. Palmieri. 1982. Gap junc-tions in myo-endothelial bridges of rabbit carotid arteries. Experientia(Basel).

38:124-125.

43.Metcalf,D. 1989. The molecular controlof cell division, differentiation commitment and maturation inhaemopoieticcells.Nature(Lond.).339:27-30. 44. Lang,R.A., D.Metcalf, R. A.Cuthbertson,I. Lyons,Stanley, E.,A.

Kelso, Kannourakis, C.,D. J.Williamson, C.K.Klinthworth, andT.J.Konda. 1987.Transgenic mice expressingahematopoietic growth factor gene (GM-CSF)

developaccumulation of macrophages, blindness, and a fatal syndrome of tissue damage.Cell.51:875-86.

45.Sporn,M.B., andA.B.Roberts. 1985.Autocrine growth factors and

cancer.Nature(Lond.).313:745-747.

46. Roberts,A.B.,M. B.Sporn, R.K.Assoian,J.M.Smith,N.S. Roche,

L.M.Wakefield,U.I.Heine,L.A.Liotta,V.Falanga, J. H. Kehrl,etal. 1986.

Transforming growthfactortype-beta:rapidinduction of fibrosis and angiogene-sis in vivo and stimulation ofcollagen formationinvitro. Proc.Natl. Acad.Sci. USA.83:4167-4171.

47.Roberts,C.J.,T. M.Birkermeier,J. J.McQuillan,S. KAkiyama,S. S. Yamada, W. T.Chen,K. M.Yamada,and J.A.McDonald.1988.Transforming growth factor-, stimulates theexpressionoffibronectinand of both subunits of the humanfibronectinreceptorby cultured humanfibroblasts.J.Biol. Chem. 263:4586-4592.

48.Sato, Y.,R.Tsuboi,R.Lyons,H.Moses,and D. B.Rifkin. 1990. Charac-terization of the activation of latentTGF-,Bby cocultures of endothelial cells and

pericytesorsmooth muscle cells:aself-regulatingsystem. J.CellBiol. 11 1:757-763.

49. Ignotz, R. A., and J. Massague. 1985.Transforming growthfactor type-betastimulates theexpressionoffibronectin andcollagenand their incorporation into the extracellularmatrix.J.Biol.Chem.261:4337-4345.

50.Penttinen,R.P.,Kobayashi, S.,andBornstein,P._Transforminggrowth

factorftincreases mRNA for matrix proteins both in the presence and absence of changes in mRNA stability. Proc. Natl. Acad. Sci. USA. 85:1105-1108.

51. Liau, G., and L. M. Chan. 1989. Regulation of extracellular matrix RNA levels incultured smooth muscle cells. J. Biol. Chem. 264:10315-10320.

52.McCaffery, T. A., D. J. Falcone, C. F. Brayton, L. A. Agarwal, F. G. P. Welt, and B. B. Weksler. 1989. Transforming growthfactor-aactivity is poten-tiated by heparin via dissociation of the transforming growth factor-f3/alpha2-macroglobulininactivecomplex. J. Cell Biol. 109:441-448.

53. Segarini, P.R.,and S. M. Seyedin. 1988. The high molecular weight receptor to transforming growthfactor-#3containsglycosaminoglycan chains. J. Biol. Chem. 263:8366-8370.

54.Madri, J. A., B. M. Pratt, and A. M. Tucker. 1988. Phenotypic modulation ofendothelial cells by transforming growth factor-,8 depends upon the composi-tion andorganization of the extracellular matrix. J. Cell Biol. 106:1375-1384.

55.Streuli, C. H., and M. J. Bissell. 1990. Expression of extracellular matrix components is regulated by substratum. J. Cell Biol. 110:1405-1415.

56.Lynch, S. E., R. B. Colvin, and H. N. Antoniades. 1989. Growth factors in woundhealing. J. Clin. Invest. 84:640-646.

57. Majack, R. A., M. W. Majesky, and L. V. Goodman. 1990. Role of PDGF-A expressionin the control of vascular smooth muscle cell growth by transforming growthfactor-#.J.Cell Biol. 11 1:239-247.

58.Sporn, M. B., and A. B. Roberts. 1988. Peptide growth factors are multi-functional. Nature(Lond.).332:217-219.

59.Eghbali,B., J. A. Kessler, and D. C. Spray. 1990. Expression of gap junc-tion channels in communicajunc-tion-incompetent cells after stable transfecjunc-tion with cDNAencoding connexin 32. Proc. Natl.Acad.Sci. USA. 87:1328-1331.

60.Dayer, J. M., B. de Rochemonteix, B. Burrus, S. Demczuk, and C. A.

Dinarello. 1986. Human recombinant interleukin 1 stimulates collagenase and prostaglandin E2 production by human synovial cells. J. Clin. Invest. 77:645-648.

61. Adams,D.O., and T. A. Hamilton. 1984. The cell biology of macrophage activation. Annu. Rev. Immunol. 2:283-298.

62.Hagiwara,T., I. Kono, K. Nemoto, H. Kashiwagi, and K. Onozaki. 1989. Recombinant interleukin- 1 triggers the increase of circulating fibronectin levels in rats. Int. Arch.Allergy Appl. Immunol. 89:376-380.

63. Montesano, R., A. Mossaz, J.-E. Ryser, L. Orci, and P. Vassalli. 1984. Leukocyteinterleukins induce cultured endothelial cells to produce a highly orga-nized glycosaminoglycan rich pericellular matrix. J. Cell Biol. 99:1706-1715.

64.Elias, J. A., and V. Lentz. 1990. IL- 1 and tumor necrosis factor synergisti-cally stimulate fibroblast IL-6 production and stabilize IL-6 messenger RNA. J.

Immunol. 145:161-166.

65.Hagiwara,T., H. Suzuki, I. Kono, H. Kashiwagi, Y. Akiyama, and K. Onozaki. 1990. Regulation offibronectin synthesis by interleukin- 1 and interleu-kin-6in rat hepatocytes. Am.J.Pathol. 136:39-47.

66. Lanser,M.E., and G. E. Brown. 1989.Stimulationof rat hepatocyte

fibronectinproductionby monocyteconditionmedium is due to interleukin 6. J. Exp.Med. 170:1781-1786.

67. Mecham,R. P., L.A.Whitehouse, D. S.Wrenn,W. C.Parks,G. C.

Griffin,R. M.Senior,E.C.Crouch,K. R.Sternmark,and N. F. Voelkel. 1987. Smooth-muscle mediatedconnectivetissueremodelinginpulmonary

hyperten-sion. Science(Wash. DC). 237:423-426.

68. Uematsu,M., J.Tanouchi,K.Ishihara,K.Fujii,Y.Toshida,Y.Doi,and T. Kamada. 1989.Hypercholesterolemiawithout mechanical interventions in-ducesearlyaccumulation offibronectinduringfattystreak formation. Arterioscle-rosis. 9:769a. (Abstr.)

69.Shekhonin,B.V.,S. P.Domogatsky,G. L.Idelson,V. E.Koteliansky,and V. S. Rukosuev. 1987. Relativedistributionoffibronectin and typeI,III, IV,V collagens in normal andatheroscleroticintima of human arteries. Atherosclero-sis.67:9-16.

70.Mayne,R. 1986.Collagenousproteinsof blood vessels. Arteriosclerosis. 6:585-593.

71.Levene, C.I., and J. C. F. Poole. 1962. Thecollagencontentofthe normal andatherosclerotic human aortic intima. Br. J. Exp.Pathol.43:469-471.

72.Smith,E. B.1965. The influenceofage and atherosclerosisonthe chemis-try of aortic intima. J.Atheroscler.Res.5:241-248.

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

A.M.Fogelman, andA.J.Lusis.1990.Modifiedlowdensitylipoproteinsinduce

endothelialcellexpressionofgranulocyteandmacrophage colony stimulating

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