Lipopolysaccharide (LPS) binding protein and soluble CD14 function as accessory molecules for LPS induced changes in endothelial barrier function, in vitro

Lipopolysaccharide (LPS)-binding protein and

soluble CD14 function as accessory molecules

for LPS-induced changes in endothelial barrier

function, in vitro.

S E Goldblum, … , J Pugin, P S Tobias

J Clin Invest.

1994;

93(2)

:692-702.

https://doi.org/10.1172/JCI117022

.

Bacterial LPS induces endothelial cell (EC) injury both in vivo and in vitro. We studied the

effect of Escherichia coli 0111:B4 LPS on movement of 14C-BSA across bovine pulmonary

artery EC monolayers. In the presence of serum, a 6-h LPS exposure augmented (P <

0.001) transendothelial 14C-BSA flux compared with the media control at concentrations >

or = 0.5 ng/ml, and LPS (10 ng/ml) exposures of > or = 2-h increased (P < 0.005) the flux. In

the absence of serum, LPS concentrations of up to 10 micrograms/ml failed to increase

14C-BSA flux at 6 h. The addition of 10% serum increased EC sensitivity to the LPS

stimulus by > 10,000-fold. LPS (10 ng/ml, 6 h) failed to increase 14C-BSA flux at serum

concentrations < 0.5%, and maximum LPS-induced increments could be generated in the

presence of > or = 2.5%. LPS-binding protein (LBP) and soluble CD14 (sCD14) could each

satisfy this serum requirement; either anti-LBP or anti-CD14 antibody each totally blocked

(P < 0.00005) the LPS-induced changes in endothelial barrier function. LPS-LBP had a

more rapid onset than did LPS-sCD14. The LPS effect in the presence of both LBP and

sCD14 exceeded the effect in the presence of either protein alone. These data suggest that

LBP and sCD14 each independently functions as an accessory molecule for LPS

presentation to […]

Research Article

Find the latest version:

Lipopolysaccharide (LPS)-Binding Protein and Soluble

CD14 Function

as

Accessory Molecules for LPS-induced

Changes

in

Endothelial Barrier Function,

In

Vitro

Simeon E. Goldblum, Terrence W. Brann, Xueda Ding, Jerome Pugin, and Peter S. Tobias

DivisionofInfectious Diseases, Department ofMedicine, Veterans Affairs Medical Center, University ofMaryland School ofMedicine, Baltimore, Maryland 21201; andDepartmentofImmunology, Scripps ResearchInstitute, LaJolla, California92073

Abstract

BacterialLPS induces endothelial cell(EC) injury both in vivo andinvitro. We studied the effect of

Escherichia

coli 0111:B4 LPS on movement of '4C-BSA across bovinepulmonary artery EC monolayers. In the presence of serum, a 6-h LPS exposure augmented (P < 0.001) transendothelial

14C-BSA

flux com-pared with the media control at concentrations 2 0.5 ng/ml, andLPS(10ng/ml) exposures of 22-hincreased(P<0.005) theflux.Intheabsence of serum, LPS concentrations of upto

10

Ag/ml

failed to increase

14C-BSA

fluxat6h.Theaddition of 10% serum increased ECsensitivity to theLPS stimulus by > 10,000-fold. LPS (10ng/ml, 6 h) failedtoincrease

14C-BSA

flux at serum concentrations <

0.5%,

and maximum LPS-in-ducedincrements could be

generated

inthe presence of . 2.5%. LPS-binding protein

(LBP)

and soluble CD14

(sCD14)

could each

satisfy

this serum

requirement;

either

anti-LBP

or anti-CD14

antibody

each

totally

blocked

(P

<

_0.00005)

the LPS-in-duced changes in endothelial barrier function. LPS-LBP hada

morerapid onset than did LPS-sCD14. The LPS

effect

in the presence of both LBP and sCD14 exceeded the

effect

in the presenceof either proteinalone.These data suggest that LBP

andsCD14eachindependently functionsas anaccessory mole-culefor LPS

presentation

tothe

non-CD14-bearing

endothe-lial surface.However, in the presence ofserumbothmolecules

are

required. (J.

Clin. Invest. 1994.

93:692-702.) Key

words:

endotoxin

*adult

respiratory

distress

syndrome

* vascular

per-meability

Introduction

Gram-negative bacteremia

andits attendant endotoxemiacan

serve as clinical antecedents for adultrespiratorydistress syn-drome

(1).

Bacterial LPS induces acute

pulmonary

vascular

endothelial cell

(EC)

I

_injury

_in

_experimental

_animals

_(2).

Mul-Portions of this workwerepresentedattheAmericanFederation of Clinical Research nationalmeetinginBaltimore,MDon1May1992. Addresscorrespondence toSimeon E.Goldblum, M.D.,Medical Service (111), Veterans Affairs Medical Center, 10 North Greene

Street,Baltimore,MD21201.

Receivedforpublication13 November 1992 and inrevisedform11 October 1993.

1. Abbreviations used in thispaper:CRP, C-reactiveprotein; DAF, decay-accelerating factor; EC, endothelial cell; HS, human serum;

HUVEC, human umbilical vein endothelial cell; LBP,LPS-binding

protein;LDH,lactatedehydrogenase;sCD14,soluble CD14.

The Journal of ClinicalInvestigation, Inc. Volume 93, February 1994, 692-702

tiple host responseshavebeen postulatedasmechanisms for LPS-induced lung injury, including: systemicand pulmonary

hemodynamic alterations, complement cleavageproducts,

ara-chidonatemetabolites,platelet-activating factor, granulocytes, toxicoxygenintermediates, andproteinases (2),aswellas

cy-tokines(3). LPS has been shown to augment themovementof

'251I-albumin

acrossbovine pulmonary arterial ECmonolayers

inthe absence of hydrostaticpressure changes, granulocyte ef-fectorcells, alveolar macrophages, or numerous nonendothe-lial-derived host mediators (4). Some LPS-induced EC re-sponsesare serumdependent (4,5). Recently,anLPS-binding protein (LBP) has been demonstrated in the seraof various species (6-9). LBP bindsto thelipid A component of LPS (10), and this LPS-LBP complex recognizes CD14 on host cellsof monocyte/macrophagelineage (11). The endothelium isreportedly a non-CD14-bearing surface (12, 13). Monocyte-derived CD14, which is attached to the plasma membrane via a phosphatidylinositol glycan anchor (14, 15), can also be pres-ent in soluble form in normal serum (14, 16). This soluble CD14(sCD 14) may participateinLPS presentation to the EC (17,18).Inthis report, anendotoxin-sensitive pulmonary

vas-cular EC line(4, 5)wasusedtofocus on thedirect effect ofLPS

on endothelial barrierfunction and theinfluence ofserum con-stituent(s)on thisLPS-EC interaction.Morespecifically,we

present evidence that the acute phase protein, LBP, and the monocyte

differentiation antigen,

CD 14, can each function as accessory molecules for

LPS-induced

changes in endothelial

barrierfunction.Thatmonocyticcells shed or release a mole-cule intotheintravascularcompartment which in concert with

an

hepatocyte-synthesized protein

canregulate LPS presenta-tiontothe

non-CD

14-bearing

ECsurface is novel.

Methods

Reagents. LPS phenol-extracted from Escherichia coli serotype 011 1:B4(Sigma Chemical Co.,St.Louis,MO) was suspended in PBS

at 1 mg/ml, andthisstocksolutionwasstored at -20°C.Rabbit and human LBPwerepurified fromacutephase sera as described (6, 7). Purified goatpreimmune IgGaswell as polyclonal anti-human LBP

fromimmune serum were preparedusingammonium sulfate precipita-tion(50%)andDEAE-cellulosechromatographyatpH 7.7,according

tostandard methods(19).The goat had been immunized with human LBP expressed usingthebaculovirus, SF-9 insect cell system (20).

Human sCD14 and mAb 28C5specific for humanCD14 were

ob-tained fromA.Moriartyand D.Leturcq (R.W.Johnson

Pharmaceuti-cal ResearchInstitute,LaJolla,CA).Purified murineIgG2bmyeloma

protein(SigmaChemical Co.)wasused as acontrolfor theIgG2bmAb, 28C5. Theacutephaseprotein,humanC-reactive protein(CRP) (Po-lysciences Inc.,Warrington,PA) was used as a control for LBP. Two

phosphatidylinositol-anchored proteins, human decay-accelerating

factor(DAF)provided byDr. M. E.Medof(DepartmentofPathology,

Case-Western ReserveUniversity,Cleveland, OH)andbovine

folate-bindingprotein(BiQPacificInc.,Emeryville, CA)wereusedas

trols for sCD 14. CD 14-depleted humanserum(HS)wasprepared

us-ingaffinitychromatography. Briefly,anti-human CD14 mAb 28C5 (1 mg)wasimmobilizedusingaReactigelR matrix (Pierce Chemical Co.,

Rockford, IL) and wasequilibrated with PBS. Pooled-type AB HS (Sigma Chemical Co.) ( 10 ml)waspassed through the column. sCD14 boundtothe immobilized mAbwaselutedusingglycin-HCI,pH 2.5.

AfterreequilibrationwithPBS,the first flowthroughwasagain passed throughthecolumn, and the second flowthroughwasconsideredas

CD14-depleted HS. sCD14 concentrations in the starting materialwere

2

_tg/ml

HS and<1 ng/mlintheCD14-depletedserum.

Endothelialcell culture. Bovinepulmonary arteryEC(American

Type CultureCollection, Rockville, MD)weregrownat370Cunder 5%CO2in DME(SigmaChemical Co.) andwereenriched with 20% FBS (Hyclone Laboratories Inc., Logan, UT), 4 mM L-glutamine, nonessential aminoacids, andvitaminsin the presenceof penicillin (50 U/ml)andstreptomycin (50

,g/ml)

(SigmaChemicalCo.). EC werewashedand were gently detachedwithabrief( 1-2min)trypsin (0.5 mg/ml) (Sigma Chemical Co.)exposure with gentleagitation,

followedimmediately byneutralization withFBS-containingmedium. The cellswerecounted andweresuspended in medium for immediate

seedingof assay chambers(4X 105 cells/ml)or6-well tissue culture plates (3.0 x I05 cells/ml). Culturesweredeterminedtobeendothelial by uniformmorphologyandquantitativedetermination of

angioten-sin-convertingenzymeactivity withcommerciallyavailable 'H-ben-zoyl-Phe-Ala-Prosubstrate(VentrexLaboratories, Portland, ME).

Assayoftransendothelial albuminflux.Transendothelial '4C-BSA fluxwasassayedas wehavedescribed(21 ). Polycarbonatefilters (

13-mmdiameter, 0.4-um pore size) (Nucleopore Corp., Pleasanton, CA)

weretreated with 0.5% acetic acid (50°C, 20 min), were washed in distilled H20,andwereimmersed inboiling pigskingelatin(Fisher

ScientificCo., Pittsburgh, PA)(5mg/literdistilledH20)for 60 min. The filters were then dried, were glued to polystyrene chemotactic chambers(ADAPSInc.,Dedham, MA),andweregas sterilized with

ethyleneoxide. Thesechambers,which servedasthe upper

compart-mentforthe assay chambers,wereinsertedinto wells of 24-well plates, each wellcontaining1.5 mlmedium andservingasthelower

compart-mentof the assay chamber. Each upper compartmentwasseeded with 2 x 105EC in 0.5 ml medium andwasculturedfor 72 h (37°C, 5%

C02).We used

`4C-BSA

(SigmaChemicalCo.)withaspecific activity

of 30.1 PCi/mgprotein asthetracermolecule. Thebaseline barrier function of each monolayerwasdetermined by applyinganequivalent andreproducibleamountof'4C-BSA( 1.1pmol/0.5 ml)toeach upper compartment for 1 hat37°C, after which 0.5 ml from the lower com-partmentwasaddedto4.5 mlscintillation fluid(Optifluor;Packard InstrumentsCo., Inc., Downers Grove, IL) and was counted in a liquid

scintillationanalyzer (Tri-Carb 1500; Packard Instruments Co., Inc.).

OnlyEC monolayersretaining 2 95% ofthe '4C-BSAwerestudied. ThemonolayerswerethenexposedtoLPSatvarying concentrations for increasingexposure times. Simultaneouscontrols with medium alonewereperformed.Transfer of'4C-BSAacrossEC monolayers was

againassayed.

Inotherexperiments, serum-dependence for LPS-induced changes inendothelial barrier functionwasstudied. EC were seeded onto

gela-tin-impregnatedfilters andwerecultured inDMEenriched with 20% FBSasabove.Over the last 16 hofthe 72-h incubation, the monolayers were serum-starved in DME alone. At 72 h, baseline barrier function wasestablished using the permeability tracer14C-BSAin fresh medium

containingnoFBS but supplemented with BSA (34 g/liter) yielding a final protein concentration equivalent to DME enriched with 10% FBS. TheBSAwasincluded to control for potential nonspecific pro-tein-LPS interactions andtominimizenonspecificbinding of reagents

toplastic. Then,the monolayerswereincubated with media containing LPS 10 ng/ml, increasing FBS concentrations, and reciprocally

de-creasingBSA concentrationsso thatthe samefinal protein concentra-tion wasmaintained. Endothelial barrier function again was deter-mined using 14C-BSA with the same fixed BSA concentration used above.Endothelial barrier function alsowas

assayed

for serum-starved

monolayers

after 6-h exposures to

_increasing

LPS concentrations

( I1 -'- 10 5 ng/ml) again, in the presence ofthesamefixed BSA

concen-tration usedabove, in the absence of FBS. To address whether LBPor

sCD14 could beoperative in the serum-dependent LPSeffect, rabbit LBP(6), human LBP (7), or human sCD 14 were each used in selected experiments in lieu of FBS. LBP (molwt = 60,000)wasusedat

con-centrations ( 1.2 ug/ml) equimolartoLPS100 ng/ml (estimated

aver-age molwt = 5,000) (10). Human CRP and DAF as wellasbovine

folate-binding protein were used as protein controls. To establisha

dose-response relationship between LBP and LPS-induced barrier dys-function, increasing concentrations of rabbit LBP together withafixed LPSconcentration ( 100 ng/ml) were studied. Similar studies with vary-ing concentrations of human sCD 14 and the same LPS exposure were alsoperformed. In experiments where LBPwascoadministered with sCD14, lower concentrationswereused.

In one setof experiments, serum-starved monolayers were studied for barrier function in the presence of 10% HS. The HSwasobtained from thesameindividual, andwasimmediatelyaliquoted and frozen. Themonolayers were exposedtoLPS 100ng/ml for 6 h withor

with-outpreincubation with either murine monoclonal anti-human CD 14 antibody 28C5 (10

_Atg/ml)

orgoat anti-human LBPIgG (1 mg/ml) (19). Simultaneous controls for each antibody preparation aloneas

well asforspecies-matched irrelevantantibodieswerealsoperformed.

Inanother set ofexperiments,serum-starvedmonolayerswereexposed

toLPS 100ng/mlormedia alone for 6 h in the presence of either 10% HSor 10% CD 14-depleted HS from thesamedonor.Thesetwo sera

had LBPconcentrationsof 27 and 23

_,g/ml,

respectively.

EffectofLPS on ECviability. To determinewhetherLPS-induced

changes in endothelial barrier function could beexplainedby EC in-juryorlossofviability, LPS-exposed and medium control monolayers wereserially studied for lactate dehydrogenase (LDH) release (22) and 5'Cr release (21 )as wehavedescribedpreviously. ECwereseeded into the wellsof 24-wellplates (2x10 5 cells/ well) andwereculturedfor 72 htoachieveconfluence.FortheLDHrelease assay, themedium was decanted andwasreplaced withmediumcontainingLPS 10ng/ml or medium aloneforincreasingexposure times.Supernatantswere har-vested, and the monolayers weresolubilized ( 1% Triton X- 100, 0.5 h).

LDHactivitywasassayed in themedium and in the cell lysate (22). Test sampleswere mixedwith reconstituted LD-L reagent (50 mM

lactate, and7mM NAD, pH8.9)(SigmaChemicalCo.),andAA340 nm

was measured at37°C over 1.5 minin a spectrophotometer (Gilford Response II; Ciba Corning Diagnostics, Oberlin, OH).LDH release was expressed as a percentageof total LDH (LDH in medium plus

LDHin cells). Correction was made for LDH in FBS. For the 5'Cr release assay,confluent monolayers were labeled with[5"Cr]sodium

chromate (AmershamCorp., ArlingtonHeights, IL), 12ttCi/well,for 3 hat37°C (21). The washed monolayers were incubated with either LPS 10ng/ml or medium alone for increasing exposure times after which the supernatants were counted. To differentiate between actual EC releaseof5'Crand EC detachmentinto the supernatants, aliquots of washes were centrifuged (300 g, 10 min) before counting. All washed monolayers were solubilized with 1% Triton X-100 (Sigma Chemical Co.)toinduce maximum release. The lysates were

centri-fuged,andthe supernatantswerecountedfor5'Cractivity. EC injury was expressed as:(5'Cr supernatant)/(5'Crsupernatant plus5'Crcell

lysate)X 100%.

Statistical methods. The mean response for each experimental groupwas compared with its respective control by Student's t test. Analyses of variancewereusedtocompare the mean responses among

experimentaland amongcontrolgroups.A Pvalueof<0.05 was con-sideredsignificant.

Results

Dose-dependent effect

of LPS on transendothelial '4C-BSA

flux. A 6-h LPS-exposure increased transendothelial '4C-BSA

0.24

-0.22- _El 1_4C-BSAflux with 1 0% FBS

El 14C-BSA flux without FBS 0.20

-14C-BSA fluxacross naked filters

0.18

0.16

-0.14

-0.12

0.10

-0.08

-**

. 3

0-

_t4

3

_F5

Baseline 0 0 .1 0. 1 .0 3.-0 10 5 0 10 0 5 00 10 xl1 1 0 1 0 Filters

16 0.06

-0.04

-0.02

-n(n)= 288(69) 42(10) 5(10) 18 27(6) 30 38(8) 18 22(9) 18 21(10) 20 22(11) 7(5)

E. Coli 0111:B4 LPS Concentration (ng/ml)

Figure1. Dose-dependent effectofLPSontransendothelial '4C-BSA flux. Crosshatched barsrepresent mean(±SE) transendothelial '4C-BSA

flux in picomolesperhourimmediatelyafter 6-hexposurestoincreasingconcentrationsofLPSinthepresenceof 10% FBS. Open barsrepresent

mean (±SE)

14C-BSA

fluxacrossserum-starved monolayers exposedtoincreasing concentrations of LPS for 6 h in the absence of FBS. Mean (±SE)pretreatmentbaseline transendothelial '4C-BSAfluxacrossboth standard andserum-starvedmonolayersaswellas mean(±SE) '4C-BSA fluxacrossnaked filters (stippled bar)arealso shown.n,number of monolayers studiedinthepresenceofFBS. (n), number of monolayers studiedinserum-free conditions.*Significantlyincreased compared with the simultaneous media controlatP<0.001.**Significantlyincreased

compared with the serum-free mediacontrolatP<0.0001.

pmol/h (n = 288), and the mean (±SE) '4C-BSA transfer across naked filters without EC monolayerswas0.215±0.015 pmol/h. The lowest LPS concentration that induceda

signifi-cantincrementin'4C-BSA flux compared with the media con-trolwas0.5 ng/ml. The maximum mean(±SE) '4C-BSA flux of 0.151±0.016 pmol/hwas seen with LPS 1,000 ng/ml, al-though the LPS-induced effect seemedtoplateauorsaturateat

doses > _{10 ng/ml. The LPS effect in the absence of FBS is}

discussed below.

Time-dependent effect ofLPSontransendothelial '4C-BSA flux. The effect of LPSon_{endothelial barrier function}wasalso time dependent (Fig. 2). Transendothelial '4C-BSA flux was assayedimmediatelyafterincreasingexposuretimestoLPS 10

ng/ml ormedia alone. There were no significant differences between

`4C-BSA

flux across media control monolayers

throughoutthe6-h study period. LPS10ng/ml failedtoinduce

significant incrementsin'4C-BSA flux compared with simulta-neousmedia controls after 0.5- and 1-hexposures. OnlyLPS exposures of22 h significantly increased '4C-BSA flux with furthertime-dependent increments throughout the 6-h study period.Thesestudies demonstratedanLPSstimulus-to-EC re-sponselag time of> 1 hbut <2 h.

Effect ofLPSonECinjuryordeath. Twoassays wereused

todetermine whetheranLPSexposure(10 ng/ml) that com-ptomises endothelial barrier function also might induce EC injuryordeathoverthesametime period. LPS-exposed and media control monolayers werestudied serially for LDH re-lease and

5"Cr

release (Table I). Thetotal cellcountsineither

groupdidnotchangeoverthestudy period (datanot

shown).

LDHreleasewasnotsignificantlydifferentbetween

experimen-tal and control groups throughoutthe study period. In

fact,

even LPS 100 ng/ml failedtosignificantly increase LDH

re-lease compared with media controls at 6 h (4.05±1.37% vs

2.22±0.81%).

Whenthemoresensitive

5"Cr

releaseassay was

used, LPS exposure significantlyaugmented EC

5"Cr

release

comparedwithsimultaneousmedia controlsat4 and 6 h. This assaydetectsdefects in theplasmamembrane that permit pas-sageofmolecules< _1,000D. Ifone usesLDH releaseasa more

stringent detectionsystem formembrane integrityand cell

via-bility,these datasuggestthata<6-h LPSexposureat10ng/ml

inducesEC membrane perturbation without frank loss of

via-bility.

Influence

ofserum onLPS-induced changesinendothelial barrierfunction. The effect of LPSonendothelial barrier func-tion wasserumdependent. LPS concentrations < 104ng/ml failed to increase '4C-BSA flux across serum-starved

mono-layersin the absence of FBS (Fig. 1,openbars). Only LPS 105

ng/ml significantly increased '4C-BSA flux compared with the simultaneous media control. This serum-free 6-h LPS expo-sureat 105 ng/ml significantlyincreased ECcytotoxicity

com-pared with media controls on the basis of LDH release (40.76±6.01vs8.40+1

.08%,

n= 12). The LPS-induced barrier

dysfunction induced bya high LPSconcentrationin the ab-sence of FBS was associated with EC death, unlike barrier

changesseenwith low LPS concentrations in thepresenceof FBS (Table I). Transendothelial '4C-BSA fluxacross

serum-694 S. E.Goldblum, T. W.Brann,X. Ding,J. Pugin,and P. S. Tobias 0

E 0.

x 3

.E

:

.0

C-)

I e,.".,

.21

1//

.IAl

Z

A:"

K."I

I..

I,.

I ,I."

.11

*

0.14-0.12

-

0.10-0.08

0.06

0.04-

0.02-* Baseline Barrier Function o Media Control

M

LPS, 1Ong/ml

p< 0.00005

p<0.001

p<0.001

0

N= 182

0.5 1

13 15 16 18

2 12 12 LPS Exposure time (h)

starved monolayers was measured aftera 6-h exposure to a fixed LPS concentration (10ng/ml)ormedia, bothinthe pres-ence ofincreasing FBS concentrations (Fig. 3). The mean (±SE) pretreatment transendothelial '4C-BSA flux was 0.014±0.001pmol/h(n=349).LPS

1Ong/mlfailedtosignifi-cantly increase '4C-BSA fluxatserum concentrations <0.5%

compared with the mediacontrolscontaining equivalentFBS

concentrations. The maximum transendothelial '4C-BSAflux

after LPSexposure(10ng/ml,6h)was seenin thepresenceof

2 2.5% FBS. At the intervening FBS concentrations, i.e., 0.5-2.5%, the LPS effecton'4C-BSA fluxincreased withincreasing FBSconcentrations.

RoleofLBP inLPS-induced changesinendothelial barrier function. The ability ofLBPtopromote the LPS-induced EC

response was studied (Fig. 4). LPS only in the presence of eitherFBSorLBPsignificantlyincreasedtransendothelial t4C-BSA flux. The addition ofBSA orCRP failed tosupportthe

LPS effect. When monolayerswereexposedfor6 htoafixed

LPS concentration( 100 ng/ml)inthepresenceofvarying

rab-bit LBPconcentrations, theLPS/LBPeffect increasedasthe

LBP concentration increased (Fig. 5). LBP concentrations

4

8 10

Figure 2. Time-dependent effectof

LPSontransendothelial

"4C-BSA

[I jgJ

flux. Vertical bars represent mean

(±SE) transendothelial

_'4C-BSA

flux in picomoles per hour immediately

afterincreasingexposure timesto LPS 10ng/ml (crosshatched bars)

6 andsimultaneous media controls

40 38 (open bars). Mean (±SE)

pretreat-mentbaselineis shown by the closed bar.

2 0.1 ,ug/mlin thepresenceof LPS significantly increased

al-bumin fluxcomparedtoLPS alone; LBP aloneat3,g/ml had noeffect. The maximal LPS/LBP effectwas seenin the

pres-enceof 1.0

,ug/ml

LBP,acloseapproximationof theestimated equimolar concentration of LBP for LPSat 100ng/ml(10). LBPat > 1 ,g/ml wasnotassociated with anyfurther

incre-ment inthe LPS/LBP effect. These dataarecompatible with

anoptimal LPS/ LBP ratio of 1:1 inourexperimentalsystem.

WhenHSwasusedtotake advantage of human LBP anti-bodies, anti-LBP antisera completely blocked the ability of

serumtosupporttheLPS-induced endothelial barrier dysfunc-tion (Fig. 6). In a serum-free reconstituted LPS-human sCD 14 system, anti-LBP antibody had no effect, excluding

cross-reactivity of the anti-LBP antibody with sCD 14. There-fore, the single protein LBP could substitute forserum in the

promotion of the LPS-induced changes, and in thepresenceof

serum LBPwasrequired.

Role of sCD14 in LPS-induced changes in endothelial barrierfunction. Similarly, the ability ofsCD 14toalter

LPS-in-duced changes in endothelial barrier function in the absence of

serum or LBPwas studied (Fig. 7). LPS with sCD14 in the

TableI.MeasurementsofEndothelialCellInjury orDeath after LPSExposure

Time Ih (n) 2 h (n) 4h (n) 6 h (n)

LDHrelease(%)*

Media control 4.93±1.74 (9) 1.98±0.77 (9) 2.97±0.79 (9) 2.22±0.81 (23)

LPS (10

ng/ml)

2.14±0.72 (9) 3.30±1.87 (9) 2.98±1.33 (9) 4.40±0.99 (23)

5"Crrelease(%)*

Mediacontrol 5.83±0.51 (10) 8.13±0.28 (10) 8.76±0.23 (10) 13.00±0.50 (10)

LPS (10ng/ml) 6.73±0.29 (10) 7.60±0.44 (10) 13.10±0.42§ (10) 16.26±1.33§ (10)

*_Mean

_(±SE)

_{percentageofLDHreleasecalculatedas(LDHin}medium)/(LDH in medium plusLDHincell lysate)x 100%. *_Mean

_(±SE)

percentageof5"Crrelease calculated as

51Cr

activity in supernatant/total5'Cractivities insupernatant and celllysateX 100%. §Significantly

increased compared withsimultaneousmediacontrols at P < 0.04.

0 E 0..-x

U-.0 U

:0

T

I

-",- I,,--

---0.00 .

-j

Figure 3.

Serum-depen-0.10 dent effect of LPS on

transendothelial "4C-BSA

%FBS 0 0.1 0.25 0.5 0.75 1.0 2.5 5.0 10 20 flux. Transendothelial

0.09 N(Control)= 15 17 14 18 18 19 17 19 6 5

_*4C-BSA

_fluxacross

N(LPS) = 22 19 18 27 28 24 22 17 12 12 * serum-starved

mono-0.08 . layers exposed to LPS 10

ng/ml or media for 6 h

0.07

. p ^ in thepresenceof

in-0*07

-T- creasing FBS

concentra-tions. Eachsymbol

repre-0.06 / sents mean(±SE)

tran-LPS

(1Ong/ml, 6h)

sendothelial '4C-BSAflux

0.05 * * / in the presence of a

spe-/.05

-cificFBS concentration. Mean (±SE)

pretreat-0.04- ment

baseline

transen-dothelial '4C-BSA fluxis

o.o30.03- / I ffi-I\ shown by the open bar.

1

\Inset

containsnvaluesfor

Media Controls bothLPS-exposed

mono-0.02- layers and the media

con-trols in thepresenceof

0.01 increasing FBS

concen-trations.

_{*Significantly}

in-creasedcomparedwith

0.00 .* . * ,* . *. *. * . * . *. * . thesimultaneous media

0 0.10 0.2 5 0.50 0.75 1 2.5 5 10 2 0 controls containing an

equivalentFBS

concen-0.

0.

0

0.

0.

Concentration of Fetal Bovine Serum (%) trationat-P<0.045.

Figure 4. Effect ofLBPon

LPS-induced changes in endothelialbarrier func-tion. Transendothelial

'4C-BSA fluxwas deter-minedacross

serum-.10 starved

monolayers

ex-posed for 6 hto

serum-freemedia, LPS 100

ng/mlinserum-free

me-dia, media enriched with

1.08

-10%FBS, LPS ₁₀₀ng/ml

inmedia with10%FBS,

LPS 100 ng/ml withan

equimolar concentration

lO6 ^ Z X ofrabbit LBP(1.2gg/ml)

p<o.oooos Z E inserum-freemedia,LPS

~~~~~~~~~~~~~100ng/mlwithan

equi-molarconcentrationof

i.04 P~~~~~~~0005 human LBPinserum-free

p = 0.0005 | rnmedia, LPS 100ng/ml

with human CRP(1.2

p Asg/ml),or serum-free

mediawitheitherrabbit

1.02

-or humanLBPorCRP, alone. Vertical

cross-hatchedbarsrepresent mean(±SE)

transendo-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~thelial'4C-BSAflux in

Baseine ~LPS Madia LPS MedIa LPA LPS kIdiaa iPS picomolesperhour

im-(I100 nglml) + +FBS rambWi +rabbft +huma +Kuma +huma +h"ra mediatelyafter

stimula-10%FBS LBP IBP IBP IBP CRP CRP _tion,andmean(±SE) (1.2.Pglmll (1.2pg/ml (1.2yg/m~l pretreatmentbaseline is N =188 47 34 18 22 7 23 6 12 9 10 shownbytheclosed bar.

696 S. E.Goldblum, T. WBrann,X.Ding,J.Pugin,and P.S. Tobias

0 E 0.

x

E .0

m

R

0 E 0. E CL

E

.0

0

E

E E

Baselie Moe

n. 76

Figure5. Dose-dep

endothelialbarrier f

terminedacross sen

freemedia,LPS lOC the presenceofincr LBP(3 ug/ml)aloi

(±SE)transendothe ately afterstimulati

is shown by the clo LPS in the absence

absence ofserum orLBP significantlyincreased'4C-BSAflux * .

_compared

with either LPSorsCD 14 controls. Theaddition of

BSA, DAF,

or

folate-binding protein

each failedto

support

the

LPS effect. When monolayers werepreincubatedwith sCD 14 for 0.5 h and then washed before LPStreatment,noincrease in

'4C-BSA flux could be demonstrated (datanotshown).When

monolayerswereexposedfor 6 hto afixed LPS concentration

(100

ng/ml)

in thepresenceof

varying

sCD

14

concentrations,

the LPS/sCD14effect increasedas afunctionof sCD14

con-centration (Fig. 8). sCD14 concon-centrations 2 25ng/ml in the

presenceof LPS

significantly

increased albumin flux

compared

with LPS alone, and theeffectplateaued atsCD 14 concentra-tions of .500 ng/ml. At concentrations of 2 250 ng/ml, sCD 14 in the absenceof added LPS induceda

slight

but

signifi-cantincrement in albumin flux

compared

with the media con-dia LW LPS US U~s Us LPS trol.

However,

the sCD 14 preparations used in these

experi-3.0,g/ml 100 ng/ml +LBP +LBP +LBP +LBP

0.1,ug/ml 0.3Jg/ml 1.0yg/ml 3.0og/mi ments contained 0.3-16 ngLPS/ig sCD14 (i.e., up to 4 ng

11 10 11

LPS/250

ng

sCD14).

Again, using

HS to take

advantage

of

anti-human CD14 antibodies, anti-CD14

antibody

com-endent effect ofLBP on LPS-inducedchangesin pletely blocked the ability of serum to support the LPS-induced unction. Transendothelial '4C-BSAflux was de- endothelial

barrier

dysfunction (Fig. 6). In aserum-free

recon-om-starvedmonolayers exposedfor 6 hto

serum-stituted

LPS-humanLBP system,

anti-CD

14

antibody

hadno

ng/mlinserum-freemedia,LPS 100ng/mlin

effect,

excluding cross-reactivity

of the

anti-CD14

antibody

,easing concentrationsofrabbit LBP,orrabbit _{with LBP. To further substantiate the accessory function of}

ne. Vertical crosshatchedbarsrepresentmean LBP.

_{Tolfurthersubs}

_e

_{theLaccessoryrfunce}

_of Alial'4C-BSA fluxinpicomoles perhourimmedi- sCD14, monolayers were exposed to LPS in the presence of ion,and the mean(±SE) pretreatment baseline either HS or HS depleted ofsCD 14 byimmunoaffinity chroma-)sed bar. *Significantly increasedcompared with tography, both from the same donor. The normal HS

sup-of LBP atP <0.002. ported the LPS effect, whereas the CD14-depleted HS with

Figure6. Effectof anti-LBPandanti-CD14 im-munoblockade on

LPS-induced changes in endo-thelial barrierfunction.

Transendothelial 14C-BSA fluxwasdetermined

acrossserum-starved

X | S S | T S 1 | monolayers exposedfor 6

n 0.06- h_tomedia enriched with

10% HS, LPS 100_ng/ml P 0.

inmedia with 10%HS,

E LPS in media with 10%

p<0.00M5

~~~~~~~~~~HSin thepresenceof

0104

HS pO~ng/ml+HS+ +HS+ +HS+ +HS+ usnLP huanW h~anC14 hmnCD l.anti-humanLBP( LBP anti-body(Img/mi),LPS 00

+HS ant-LP bG ani D1 muin 1pgml+ nt-CD4 O~g/l aLP erng/mlin mediawith 10% HS inthepresence of

0.02- _anti-human CD14

anti-body (10 jig/ml), LPS

100ng/mland human LBP(1.2jug/ml) in

serum-free media, LPS

0.00 _{100 ng/ml andhuman}

Bmkn Me/ia IPS IFS LFS LPS LPS AFS+ LPS+ LPS+ LPS+ anti-CD14 anti-LBP LBP(1.2 ug/ml)in

+10%HS lO0ng/mi +HS+, +HS+ +HS+ +HS+ huiaLBP humaLB huxnanCil4 huonnCil4

+HS anti-LBP gmi~G anti*CD14 munne 1.2pgmI +a6tCD14 lun0rg/mi WAUP~s serum-free media in the

~~~~~G

~~~~~~~~presenceof anti-human

N- 105 11 17 6 3 11 7 18 1 1 7 8 3 3 CD14antibody,LPS 100 ng/mland human sCDl4

100ng/mlin serum-free media,LPS 100ng/mland human sCDl4 inserum-free media in thepresenceof anti-LBPantibody,ormedia with

eitherantibodyortheirspecies-matchedcontrols alone. Vertical crosshatchedbarsrepresentmean(±SE) transendothelial '4C-BSAflux in

pi-comolesperhourimmediately after stimulation,andmean(±SE)pretreatmentbaseline isshown by the closed bar. 0.10

-0

a-

0.12-Figure 7. Effect of sCD l 4 on LPS-inducedchanges in

en--0.100.10-_-1g

~~~~~~~~~~~~~~~~~~~~~~~~~dothelial

barrier function.

_p5 0.0003- | Transendothelial 4C-BSA

E fluxwasdeterminedacross

ffi ffi serum-starved monolayers

0.08-

exposed

for6 htoserum-free

X

1 _ @ media,LPS

100

ng/ml

serum-free

_media,

human

C p< 0.00005 X | sCDl450

ng/ml,

LPS 100

ng/ml,

LPSwithsCDl450

E 0.06 ng/ml, LPS 100ng/mlwith

a0

S g human DAF50_ng/ml,LPS

100

ng/ml with bovine

fo-o 0.04- = o.ooo1

late-binding

protein (FBP)

_I _{50 ng/ml, human LBF 100}

T I g I g ng/ml,LPS 100

_ng/ml

with

LBP

100

_ng/ml,

sCD14

and

0.02_{-LBP,and LPS with both}

sCDl4and LBP. Vertical crosshatched bars represent

mean(±SE)

transendothelial

o.oo 14_{C-BSA flux inpicomoles}

hmasis Media LPS soluble LPS LIPS+ LPS LBP LPS t CD14 LPS + perhour immediately after

Control lOOngIml CD14 CD14 hlman bans 100ng/ml LBP +LBP CD14+

S0ngmim DAF50ng/mi FBP SOnglal LB stimulation,and mean (±SE)

pretreatment baseline is

N- 110 11 15 11 17 6 10 4 11 6 19 shownbytheclosedbar.

suprathreshold concentrations of LPBcouldnot(Fig. 9). LBP Kinetic analysis of LPS-induced changes in

endothelial

significantly enhanced the LPS-sCD 14-induced increments barrierfunction in the presence of LBP, sCDJ4, orboth.

LBP,

compared with those seen after exposuretoLPS-sCD 14 alone sCD 14, or both accessory molecules together were studied for (Fig. 7).Therefore, sCD14 functioned as an accessory mole- their ability to promote the LPS-inducedchanges in barrier

cule for LPS in the absence of LBP, and immunoblockade with function overtime(Fig. 10). All reagents wereintroducedat

anti-CD14 antibody prevented LPS-induced changes only in thebeginningof theincubation period and remained

through-the presence ofserum.The apparent functional differences of outthe indicated time intervals when barrier functionwas im-LBPand sCD14 in serum and purifiedreconstituted systems mediately assayed. LPS in the presence of either LBPorboth

are discussed below. LBP and sCD14 (50 ng/ml) significantlyincreased

'4C-BSA

0.0 * *

* Baseline Barrier Function

El CD14*

EI...sl & g *

Figure8. Dosedependenteffetof

sCD

14

X~~~

X onLPS-inducedchangesinendothelial

0.04- _{T t * barrierfunction. Transendothelial}'4C-BSA

= I T fluxwasdeterminedacrossserum-starved

E1 1 E 1 l iw

~~~~~~~~T

| | r _onolayersexposedfo 6 htoserum-free

8

CD + LP

_(ngmi)media,

lO OLPS

_ng/ml

increasing

sCD14

T

EIT

g;1I

Xlg

l

~~~~~~concentrations,

orthesesamesCDl4

con-0.W02- centrations in thepresenceofLPS 100ng/

^

Fl

111

ffi||

g|E

|

~~~~~~~~ml.

Vertical bars_representmean

(+SE)

*||

_*

|

|||

| l

WlB

| l

lS|

M |

glgl

| l |

~~~~~~~~per

l S

~~~~~~transendothelial

'4C-BSA

flux inpicomoles hour

_immediately

afterstimulation, .*0

and mean

(+SE)

pretreatmentbaselineis

0.* shownbytheclosed bar.*Significantly

in-creasedcompared wethsCDe4 ntheafs N-122 7 10 9 10 9 10 7 10 9 10 9 10 5 7 senceof LPSatP<0.02.

**Significantly

increasedcomparedwiththemediacontrol

Human soluble CD14 (ng/ml) atP<0.002.

698 S.E.Goldblum, T. W.Brann,X.Ding,J.

Pugin,

andP.S. Tobias

flux compared with LPS

aloneatall time

points.

LPS inthe

presence

of sCD14 alone (50

or

500

ng/ml)

did not

signifi-cantly increase '4C-BSA flux until

6 h.At

6

h,

LPS in the pres-enceof bothLBPand sCD14(50

ng/ml)

induced

significantly

greater

'4C-BSA flux

than did LPS with eitherLBP orsCD14

alone.

p<0.0001

I

Baseline Media LPS 100 ng/ml Media LPS 1ong/ml

+10%HS +10%HS +10%HS +10%HS CD14-depleted CD14-depleted

Figure9.EffectofCD14-depletiononserum-dependentLPS-induced

changesinendothelialbarrier function. Transendothelial '4C-BSA

fluxwasdeterminedacrossserum-starvedmonolayers exposedfor6

htomedia enriched with 10%HS,LPS 100ng/mlinmediawith

10% HS, mediawith 10%CD14-depletedHS from thesamedonor,

orLPS100ng/mlwith 10% CD14-depletedHS.TheLBP

concen-trationsinthe HS andCD14-depletedHSwere27and 23,ug/ml,

re-spectively. Vertical bars representmean(±SE)transendothelial

'4C-BSA fluxinpicomolesperhourimmediatelyafterstimulation,and

themean(±SE)pretreatment baseline is shownbytheclosedbar.

Discussion

Inthisreport,wehavedemonstratedthatLPSextracted from E. coli0 111 :B4 induces dose-, time-, and serum-dependent

increments intransendothelial'4C-BSA fluxacrossbovine pul-monaryartery EC monolayers. In thepresenceof 10% FBS,a 6-h LPSexposure at concentrations aslow as 0.5 ng/ml

in-creased

`4C-BSA

flux. Meyricketal.(4) havepreviouslyshown that E.coli055:B5LPSatonefixed concentration (10

gg/ml)

augmentsmovementof '25I-BSAacrosssimilarlypassaged bo-vine pulmonaryarteryEC cells culturedon filters, also in the presenceof 10% FBS. TheirLPS concentrationwas 2 1,000-foldhigher than those used inourstudy.

Basedonthe dose-responserelationshipbetween LPS con-centration and change in endothelial barrier function, LPS 10

ng/ml in thepresenceof FBSwasusedtostudy the time profile for theLPS-induced effect. A minimum LPSexposureof 2 h was necessaryfor immediate changes. Using LPS10 g/ml ina

similar experimental system, Meyrick et al. (4) found in-creasedtransendothelial albumin fluxonlyat3 h. This 2-3-h stimulus-to-response lag time is similarto that described for

NValues 0.20

0

Ea

x

-c ._

0

I-l

0 lhr 2hr 4hr 6hr

0 1 2 3 4 5 6

Time

(h)

Figure 10. Kinetic analysis ofaccessorymolecule(s)regulationofLPS-inducedchanges in endothelialbarrierfunction.Transendothelial

"4C-BSA fluxwasdeterminedacrossserum-starved monolayers immediately afterincreasingexposuretimestoserum-freemedia, LPS 100ng/ml

serum-freemedia, LPS with LBP (1.2gg/ml),LPSwith sCD14 (50or500 ng/ml), and LPS withboth LBP andsCD14(50 ng/ml). Each

symbolrepresentsthemean(±SE) transendothelial '4C-BSA flux inpicomolesperhourimmediatelyafter increasingexposuretimes.

0.08-a

E a.

x

C

E

0

'r~

006

0.04

0.02

pcj.O001

I~~~~

I

certainotherLPS-inducedEC responses, including hyperadhe-siveness for neutrophils (23),and increased expression of

plas-minogen activator inhibitor 1 (24), IL-i (25), IL-6 (18, 26), IL-8 (18), and PDGF (27). The time requirements for the

LPS-induced changes in endothelial barrier function are not

unlike those described after eitherrTNFa (21 ) or rIL- 1 (28) exposures.Compared withLPS, most otherestablished media-tors of increased endothelial permeability, including

hista-mine, bradykinin,andleukotrieneB4, alldisplayamuch more

rapidonsetof action(29).

Todetermine whether endothelial barrier dysfunction

pro-duced by a c6-h LPS exposure at 10 ng/ml was mediated through EC

injury,

two

cytotoxicity

assays wereused. LPS (10

ng/ml) failedtoincrease

endothelial

cellLDHreleaseover6 h.

LPS failedtoincrease51Cr release after1 or 2h, but after4 and

6 h, 51Cr release was minimally but significantly increased. LPS-induced EC cytotoxicity has been well described and is highly dependentonLPS

concentration

and exposuretime,the presence

of

serum, aswellasthe

species

and

possibly

anatomi-cal

origin

of the target EC (5, 30-32). Most of these earlier studies used higher LPS concentrations and longer LPS expo-suretimes (4, 5, 30, 31, 33, 34).In our system, LPS induced changes in endothelial barrier functionatdosesandexposure

times ( 10

ng/ml

X2h), which couldnotbeascribedto

cytotox-icityorlossof viabilityasmeasured by theseassaysystems.On the

basis of

trypan blue

exclusion

(data not shown) and in-creased

5"Cr

release, LPS induced sublethal cell

injury

by 4

and 6 h.

The LPS-induced changes in

endothelial

barrier function

wereprofoundlyserumdependent. LPS (10 ng/ml, 6 h) failed

toincrease

14C-BSA flux

atserum

concentrations

<

0.5%,

and maximum LPS-induced increments could be

generated

in the

presence of . 2.5% FBS. The

addition

of 10% FBS increased EC

sensitivity

tothe LPS stimulus by>

10,000-fold.

LPS-in-duced EC

cytotoxicity (4, 35)

anddetachment

(5)

aswellas

prostacyclin

synthesis (4) have all been shown to be serum

dependent. LPS

functionally

interacts with numerous serum

constituents;

it activates the alternative

complement pathway,

and the

intrinsic

clotting

and kallikrein-kinin systems

(36),

and

directly

bindstoHageman factor

(36), high

and low den-sity

lipoproteins (37),

theacute

phase

protein,

LPB

(6, 10),

and sCD14

(18).

We have now demonstratedserum factors

necessaryfor LPS-induced

endothelial

barrier

dysfunction.

In our bovine EC system, either rabbit or human LBP

alone, in the absenceof otherserum

constituents, promoted

the

serum-dependent

LPS-induced

changes

in

endothelial

barrier function. A

dose-response

relationship

between the LBP

concentration

and

this

LPS effectwasdemonstrated. The

optimal LPS/LBP

ratiowas 1:1. In

addition,

anti-human

LBPantibody completely blocked the LPS-inducedeffect in the presenceof HS. ThesedatastronglysuggestthatLBP can

functionas anaccessorymolecule forLPS

presentation

to the ECsurface. LBPispresent in normalrabbitserumin

concen-trations of<0.5 ,ug/ml,

increasing

- 100-fold 24 hafter

in-duction ofan acutephaseresponse(6, 10). Inhumans,LBP rises from 7.2±4.0 ,g/ml in the sera ofnormal subjects to

220±100

tg/ml

in acutephase sera ( 38).Theconcentration of

LBPemployedin ourstudies (1.2

_1Ag/ml)

waswellwithinthe

expected concentrationrangefoundin serumof either

species

(6, 7).Thatrabbitand human LBPaswell asHScouldeach present LPStobovineEC shouldnotbesurprising,as the

mole-cule is

highly

conservedacross

species (7).

Rabbitand human

LBP share 69% amino acid identity and 78% nucleotide iden-tity. AnECsurface receptor that LPS-LBP complexes recog-nize is not known. Toour knowledge, CD14, adifferentiation

antigen on monocytic cells ( 14, 15, 39), has not been demon-strated on EC. Wright et al. (12) found that LBP greatly en-hances binding of LPS-coated erythrocytes to monocytes and

macrophages, and that this binding could be blocked by anti-CD 14antibodies. Nosuch binding couldbedemonstratedfor human umbilical veinEC (HUVEC). Beekhuizen et al. ( 13)

found no CD14 expression on EC by flow cytometry. More recently, although CDl4 transcripts have been demonstrated

inECby polymerase chain reaction, CD14message and pro-tein could not be found using Northern analysis, ELISA,

im-munoprecipitation,orflow cytometry (Harlan, J. M., personal

communication).

There are two studies that are not consistent with our obser-vation that LBPcan support LPS presentationtothe

EC.

Pugin et al. ( 18)foundthat LPS-LBP was unable to induceHUVEC

expression of adhesion molecules and cytokineproduction. It is conceivablethatEC from different anatomic sites and spe-cies differ sufficiently in their cell-surface receptors to explain thesedifferences.Further, the proteinsynthesis-dependentEC

responses of adhesion molecule and cytokine expression used by Puginet al. ( 18)differ considerably with thecytoskeletally driven changes in barrier function (22) used in our studies.

Recently, Arditi et al. (35) reported that LBP does not support LPS-induced bovine brain microvascular EC cytotoxicity as measured by LDH release after a 24-h LPS exposure. Relevant to the EC type used in this study, Meyrick et al. (30) have previouslydemonstratedthatalthough LPS induces bovine pul-monary arteryECbarrier dysfunction, pulmonary microvascu-larECfrom the same species are relatively LPS resistant. Arditi et al. (35) never showed their LBP to be active, i.e., there was no positive control, making their negative results ambiguous. Similarly, they did not confirm their depletion of LBP from serum. Thus, it might be possible to reconcile the observed differences in behavior of LBP in these three systems, but the

explanations are not verysatisfying, and thereconciliation of thesedifferencesare underinvestigation.

sCD14also functioned dose-dependently as an accessory molecule for the LPS-inducedEC response in the absence of serum or LBP. Anti-human CD14 antibody totally blocked theLPS-inducedeffect in thepresenceofHS,but had no effect in areconstitutedserum-freeLPS-LBPsystem. In other exper-iments, HS depleted of sCD14byimmunoaffinity chromatog-raphy also failed to support the LPS effect. Thus, theCD14that

contributesto the LPS-ECinteractioninour system seems to be presentinasolubleorcirculating formandnot on the EC surface. On thebasis ofasemiquantitative ELISA,normal hu-man plasma contains sCD14at - 6

,gg/ml

(16). It has now

been shown that Haemophilus influenzae LPS induces bovine

pulmonary artery and brain EC cytotoxicity in the presence of HS that can be blocked by a murine monoclonal anti-CD14 antibody (3C

10);

noexperimentswere performed in a serum-free system (32). More recently, sCD14 has been shown to

participateinLPS-induced expressionofendothelial leukocyte

adhesionmolecule-I by HUVEC, as well as cytotoxicity of bo-vine pulmonary artery ( 17) and brainmicrovascular(32, 35) EC. Another study hasdemonstratedthatsCD14is involved in theinductionby LPSofHUVEC IL-6 andIL-8 biosynthesis,as well asintercellular adhesion molecule-I and vascular-cell

ad-hesionmolecule- Iexpression(18 ).Atpresent, nospecific

ing site on EC forLPS-sCD14 complexes is known. Human

monocytesexpress CD 14ontheirsurface, and their adherence to cytokine-stimulatedECcanbe blockedby CD 14

anti-body(13).Whether thisfinding representsanEC receptor for

CD14expressedon the monocyte surface is unknown. We found that LBPand sCD 14,two structurallyunrelated molecules, could each singularly promote LPS-induced changes in endothelial barrier function in the absence of serum.Each moleculeinfluenced the LPS effect in a

dose-de-pendent manner. Their contributions to the LPS effectwere

kinetically distinct;

LBP hada much morerapidonset ofaction than did sCD14(1 vs6h).However, in the presence of serum,

antibodies

toeitherLBP orsCD14 completely blocked the LPS

effect.

Theseresultscannotbeexplained bycross-reactivityof the

antibodies,

because antibodytoeither LBP or sCD 14 did

notcross-react with the otherantigen in a serum-free reconsti-tutedsystem.Further, immunodepletionof sCD 14 did not

sig-nificantly diminish serum LBP concentrations, and this sCD

14-depleted

serumthatcontainedsuprathreshold

concen-trations

ofLBPfailedtosupportthe LPS effect. Thus, ourdata

suggest that in a serum-free, reconstituted system, LBP and sCD14 can each independently function asaccessory mole-cules for LPS

presentation,

whereasin the presenceof serum,

both moleculesarerequired. Perhapsanadditional serum

com-ponent(s)

exists that participatesinLPS, LBP,andsCD14

in-teractions.

In serum, other sinks for LPS exist. For

example,

serum

lipoproteins avidly

bindup anddetoxifyLPS(37,

40).

Forthe ECtocompete

effectively

with theseother fates forLPS inserum, a more

effective

presentation

mechanism than

either

LBP or sCD14 alone may be

required.

In the

reconstituted

system, LBPand sCD14together enhanceLPS-induced

incre-mentsin

transendothelial

'4C-BSA fluxmore than either

pro-teindoesalone. WhetherLBPand sCD14function

additively

or

synergistically

is unclear. The optimalmolar ratios of

LBP,

sCD 14, and LPS

required

to exertaspecific biologicalresponse

ina

specific

targettissue have not yetbeen defined.

Alterna-tively,

LPS

complexed

with bothLBPandsCD14might oper-atethroughathirdpathway. Ourdataarecompatiblewithone

or morefinite receptor

populations

thatrecognizeLPS in the

presence ofanaccessory molecule(s).Whether these accessory

molecules complextoand alter the

tertiary

structureofthe LPS

ligand

to promote its

accommodation

by such an EC recep-tor(s) and/or directlyserveasdocking proteins within the EC

receptor-ligand

interaction itself is unknown. Whatever the

mechanism(s),

this ability

of

LBPand sCD14 tosupportthe

LPS effect cannot be ascribed to a nonspecific protein-LPS interaction;BSA, DAF, CRP, andfolate-binding proteineach

failedtofunctionas anaccessory molecule for LPS presenta-tiontotheEC.

Humans haveevolvedover manythousands ofyearswith bacteriaandtheirendotoxins.That cellsof

monocyte/macro-phage lineage can provide a molecule that together with an

hepatocyte-derived protein can facilitate LPS-EC interaction

demonstrates remarkable host adaptability to a potentially

lethal substance found in both gut flora and in the external environment. Throughtheir facilitation oftheLPS-EC inter-action,these accessorymoleculesmay becentralto tissue-spe-cificresponses to LPS. sCD 14 may besynthesizedand shed in

situ within

monocyte/macrophage-containing

sites where he-patocyte-synthesized LBPisexcluded. Theabilityof these ac-cessory molecules to pass through physiological barriers into a given body compartment, the rapidity of their appearance, and

their local

concentrationsallmay

modulate LPS-induced

bio-logical

responses at the

ubiquitous endothelial surface.

LBP

facilitates

LPS

priming of

neutrophils

(40),

andsCD 14

partici-patesinLPS-stimulatedastrocytomacell secretion of IL-6

(17)

and

colonic

adenocarcinoma cell

secretion

of IL-8

(18).

WhetherLBPand/ orsCD14 can serve as accessorymolecules for LPS

presentation

toother host cells isnotyetknown.

Acknowledgments

This workwassupportedin part by theOffice ofResearchand

Develop-ment, Department of Veterans Affairs, the U.S. ArmyMedical Re-search and Development Command(grant DAMD 17-91-Z-1004),

and grantsAI-25563, AI-32021, andGM-37696 from theNational

InstitutesofHealth.

References

1.Seidenfeld,J.J., D. F. Pohl, R. C. Bell, G. D. Harris, and W. G. Johanson, Jr. 1986. Incidence, site, and outcomeofinfections in patients with the adult

respiratory distresssyndrome. Am. Rev. Respir. Dis. 134:12-16.

2.Brigham, K. L., and B.Meyrick. 1986.Endotoxinand lung injury. Am. Rev. Respir. Dis. 133:913-927.

3. Marks, J. D., C. B. Marks, J. M. Luce,A.B.Montgomery, J. Turner, C.A. Metz, and J. F. Murray. 1990. Plasma tumornecrosisfactor inpatientswith

septicshock.Mortalityrate,incidence ofadultrespiratory distresssyndrome, and

effects of methylprednisolone administration.Am. Rev.Respir. Dis. 141:94-97.

4.Meyrick,B.O., U. S. Ryan, and K. L. Brigham. 1986. Directeffect ofE.coli endotoxinonstructure andpermeability ofpulmonaryendothelialmonolayers and theendotheliallayerofintimal explants. Am.J.Pathol. 122:140-151.

5.Harlan, J. M., L. A. Harker, M. A. Reidy, C. M. Gajdusek, S. M. Schwartz, andG. E.Striker. 1983.Lipopolysaccharide-mediatedbovine endothelialcell

injuryin vitro.Lab.Invest.48:269-274.

6.Tobias,P.S., K. Soldau, andR.J.Ulevitch.1986. Isolationofa lipopolysac-charide-bindingacutephasereactantfromrabbitserum.J.Exp. Med. 164:777-793.

7.Schumann,R.R., S.R.Leong, G. W.Flaggs,P.W.Gray, S. D.Wright,J.C. Mathison, P. S. Tobias, and R. J.Ulevitch. 1990. Structure and function of lipopolysaccharide binding protein.Science(Wash.DC).249:1429-1431.

8. Khemlani,L.,P.N. Bochsler,and D.0.Slauson. 1992.Purification of bovine lipopolysaccharide-binding proteins. J. Leukocyte Biol. 47( Suppl. 3):206a. (Abstr.)

9.Gallay, P.,S.Carrel,M. P.Glauser,C. Barras, R. J.Ulevitch,P. S.Tobias,

J.-D. Baumgartner, andD.Heumann. 1993.Purificationandcharacterization of

murinelipopolysaccharide-bindingprotein. Infect.Immun.61:378-383.

10.Tobias,P.S., K. Soldau, and R. J.Ulevitch.1989.Identification ofalipid

Abindingsite in the acute phase reactantlipopolysaccharide binding protein.J. Biol.Chem. 264:10867-10871.

11.Wright, S. D.,R. A.Ramos, P. S. Tobias, R. J. Ulevitch, and J. C. Mathi-son. 1990. CD 14,areceptor for complexes oflipopolysaccharide(LPS) and LPS

bindingprotein.Science(Wash.DC). 249:1431-1433.

12.Wright, S. D., P. S. Tobias, R.J. Ulevitch,and R. A. Ramos. 1989.

Lipopolysaccharide (LPS) binding protein opsonizes LPS-bearing particles for recognitionbyanovel receptor on macrophages.J.Exp.Med. 170:1231-1241.

13. Beekhuizen, H.,I.Blokland, A. J. Corsel-van Tilburg, F. Koning, and R. van Furth. 1991. CD14 contributes to the adherence of human monocytes to

cytokine-stimulated endothelialcells.J.Immunol. 147:3761-3767.

14.Haziot, A., S. Chen, E. Ferrero,M.G. Low, R. Silber, and S. M. Goyert. 1988. The monocytedifferentiationantigen, CD 14, is anchored to the cell mem-branebyaphosphatidylinositollinkage.J.Immunol. 141:547-552.

15.Simmons,D.L., S. Tan,D.G. Tenen,A.Nicholson-Weller, and B. Seed. 1989. Monocyte antigen CD14 is aphospholipidanchored membrane protein. Blood. 73:284-289.

16. Bazil, V., V. Horejsi, M.Baudys, H. Kristofova, J. L. Strominger, W. Kostka, andI.Hilgert. 1986. Biochemical characterization of a soluble form of the 53-kDa monocyte surface antigen.Eur.J.Immunol. 16:1583-1589.

17.Frey,E.A.,D.S. Miller,T.G. Jahr, A. Sundan, V.Ba2il,T.Espevik, B. B. Finlay, and S.D.Wright. 1992. Soluble CD14 participates in the response of cells tolipopolysaccharide.J.Exp.Med. 176:1665-167 1.

18.Pugin, J., C.-C. Schurer-Maly, D. Leturcq, A. Moriarty, R. J. Ulevitch, andP.S. Tobias. 1993.Lipopolysaccharide(LPS) activation of human endothe-lial andepithelialcells is mediated by LPS binding protein and soluble CD14. Proc.Natl.Acad. Sci. USA.90:2744-2748.

20. Summers, M. D.,and G. E. Smith. 1988. A Manual of Methods for Baculovirus Vectors and Insect Cell Culture Procedures. Texas Agricultural

Ex-periment Station, CollegeStation,TX. 1-57.

21. Goldblum, S.E., and W. L. Sun. 1990. Tumornecrosis factora augments pulmonary arterial transendothelial albumin flux in vitro. Am. J.Physiol. 258 (LungCell.Moi.Physiol. 2):L57-L67.

22. Goldblum, S. E.,X.Ding,T. W. Brann, and J.Campbell-Washington.

1993.Bacteriallipopolysaccharide induces actin reorganization, intercellulargap

formation,and endothelial barrierdysfunctioninpulmonaryvascular

endothe-lial cells:concurrentF-actindepolymerization andnewactinsynthesis. J.Cell. Physiol. 157:13-23.

23. Pohlman, T. H., K. A.Stanness, P. G. Beatty, H. D.Ochs,and J. M. Harlan.1986. Anendothelialcellsurfacefactor(s) inducedinvitro by lipopolysac-charide,interleukin 1, and tumornecrosis factor-a increasesneutrophil adher-enceby aCDwl8-dependentmechanism.J. Immunol. 136:4548-4553.

24. Sawdey, M., T. J.Podor, and D. J.Loskutoff. 1989. Regulation oftype I

plasminogen activator inhibitorgeneexpression inculturedbovine aortic endo-thelial cells.Inductionbytransforminggrowthfactor-f3,lipopolysaccharide,and tumornecrosis factor-a.J.Biol. Chem. 264:10396-10401.

25.Stern, D. M., I. Bank, P. P. Nawroth, J.Cassimeris,W. Kisiel,J. W. Fenton II,C. Dinarello, L.Chess,and E. A.Jaffe.1985.Self-regulation of procoag-ulant events on theendothelial cell surface.J.Exp. Med. 162:1223-1235.

26.Jirik,F.R.,T.J.Podor,T.Hirano,T.Kishimoto,D.J.Loskutoff,D. A.

Carson,and M. Lotz.1989. Bacteriallipopolysaccharideandinflammatory medi-atorsaugmentIL-6 secretionby human endothelialcells. J.Immunol.

142:144-147.

27.Albelda, S.M., J. A.Elias,E. M.Levine,and J. A. Kern.1989.Endotoxin stimulatesplatelet-derived growth factorproductionfrom culturedhuman

pul-monaryendothelialcells. Am.J.Physiol.257(LungCell.Mol.Physiol.

1):L65-L70.

28.Campbell,W. N., X. Ding,andS.E. Goldblum. 1992.Interleukin 1 augmentspulmonary arterytransendothelial albumin fluxin vitro. Am.J.

Phys-icl.263 (Lung Cell. Mol.Physiol. 7):L128-L136.

29.Grega,G.J., and E.Svensjo.1984.Pharmacology ofwaterand macromo-lecularpermeabilityintheforelimbofthedog.In Edema. N.C.StaubandA.E.

Taylor, editors.RavenPress,New York. 405-424.

30.Meyrick, B., R. Hoover, M. R. Jones, L. C. Berry, Jr., and K. L. Brigham. 1989. In vitroeffects of endotoxin on bovine and sheep lung microvascular and pulmonary arteryendothelial cells. J.Cell.Physiol. 138:165-174.

31. Harlan, J. M., L. A. Harker, G. E. Striker, and L. J. Weaver. 1983. Effects oflipopolysaccharideonhuman endothelial cellsin culture. Thromb. Res. 29:15-26.

32. Patrick, D., J. Betts, E. A. Frey, R. Prameya, K. Dorovini-Zis, and B. B. Finlay. 1992. Haemophilus influenzae lipopolysaccharide disrupts confluent monolayers of bovinebrain endothelial cells via a serum-dependent cytotoxic pathway. J.Infect.Dis. 165:865-872.

33.Gartner, S. L.,D.G.Sieckmann,Y. H.Kang, L. P. Watson, and L. D. Homer.1988.Effects of lipopolysaccharide,lipid A, lipid X, and phorbol ester on culturedbovine endothelialcells. Lab.Invest.59:181-191.

34. Paulsen, D. B., D. A.Mosier, K. D.Clinkenbeard, and A. W. Confer. 1989.DirecteffectsofPasteurella haemolyticalipopolysaccharideonbovine pul-monaryendothelialcells in vitro. Am. J. Vet.Res. 50:1633-1637.

35. Arditi, M., J. Zhou, R. Dorio, G. W. Rong, S. M. Goyert, and K. S. Kim. 1993.Endotoxin-mediatedendothelial cellinjury and activation: role of soluble CD14.Infect. Immun. 61:3149-3156.

36.Kalter,E.S.,M. R.Daha, J. W. tenCate,J.Verhoef,and B. N. Bouma. 1985.Activationandinhibition ofHagemanfactor-dependentpathwaysand the

complementsystem inuncomplicated bacteremiaorbacterialshock. J.Infect.

Dis. 151:1019-1027.

37. VanLenten, B. J., A. M. Fogelman, M. E. Haberland, and P. A. Edwards. 1986. The roleoflipoproteinsandreceptor-mediated endocytosis inthetransport

of bacteriallipopolysaccharide.Proc.Natl.Acad. Sci. USA. 83:2704-2708. 38.Tobias,P.S.,K.Soldau,L.Hatlin,R.Schumann, G. Einhorn,J.

Mathi-son, and R. J.Ulevitch.1992.Lipopolysaccharide binding protein.J.Cell.

Bio-chem.(Suppl.16C):151.

39.Ashmun, R. A.,S. C.Peiper,M. B.Rebentisch,and A. T. Look. 1987. Expression ofthe human monocytemembraneantigengp55 by murine

fibro-blastsafter DNA-mediatedgenetransfer.Blood. 69:886-892.

40.Vosbeck,K.,P. S.Tobias,H.Mueller,R. A.Allen,K.-E.Arfors,R.J.

Ulevitch,and L. A.Sklar.1990.Priming of polymorphonuclear granulocytes by lipopolysaccharidesand itscomplexes withlipopolysaccharide bindingprotein

andhigh densitylipoprotein.J.LeukocyteBiol. 47:97-104.