Higher affinity oligosaccharide ligands for E selectin

Higher-affinity oligosaccharide ligands for

E-selectin.

R M Nelson, … , O Cecconi, M P Bevilacqua

J Clin Invest.

1993;

91(3)

:1157-1166.

https://doi.org/10.1172/JCI116275

.

A series of synthetic oligosaccharides based on sialyl Lewis x (sLex; Neu5Ac alpha 2-3Gal

beta 4[Fuc alpha 3]GlcNAc) and sialyl Lewis a (sLea; Neu5Ac alpha 2-3Gal beta

1-3[Fuc alpha 1-4]GlcNAc) was used to study the binding interactions of selectins.

E-selectin-immunoglobulin fusion protein (E-selectin-Ig) bound to immobilized bovine serum albumin

(BSA)-neoglycoproteins containing sLex or sLea in a Ca(2+)-dependent manner.

Solution-phase sLex tetrasaccharide blocked this interaction by 50% at a concentration of 750 +/- 20

microM (IC50). sLea was more effective (IC50 = 220 +/- 20 microM), while nonsialylated,

nonfucosylated derivatives showed little or no activity at concentrations up to 1 mM.

Attachment of an 8-methoxycarbonyloctyl aglycone in a beta linkage to the anomeric carbon

of the GlcNAc of sLex or sLea increased their blocking activity nearly twofold. Finally,

replacement of the 2-N-acetyl substituent of the GlcNAc by an azido or amino group

resulted in substantial increases in activity, with the most potent inhibitor being amino

substituted sLea, which was 36-fold more active (IC50 = 21 +/- 3 microM) than the reducing

tetrasaccharide sLex. In contrast to results obtained with E-selectin-Ig, P-selectin-Ig binding

to immobilized BSA-sLea was blocked modestly by most oligosaccharides at 1 mM, with no

substantial differences among them. IC50 values of soluble oligosaccharides determined in

competitive binding studies accurately predicted blocking of leukocyte adhesion to

recombinant E-selectin-Ig and to […]

Research Article

Find the latest version:

http://jci.me/116275/pdf

Higher-Affinity Oligosaccharide Ligands for E-Selectin

RichardM.Nelson, Sylvia Dolich, Alejandro Aruffo,*OlivieroCecconi, and Michael P.Bevilacqua

Howard HughesMedical Institute, Division of Cellular and Molecular Medicine and Department of Pathology, University of California, San Diego, La Jolla, California 92093-0669; and *BristolMyers-SquibbPharmaceutical ResearchInstitute, Seattle, Washington 98121

Abstract

Aseries of synthetic oligosaccharides basedonsialyl Lewisx

(sLex;

Neu5Aca2-3Gal31-4[Fuca1-31GlcNAc)

and

sialyl

Lewis a(sLe';

Neu5Acca2-3Gall#1-31Fuca1-4jGlcNAc)

was used tostudythebinding interactions of selectins. E-selectin-immunoglobulin fusion protein (E-selectin-Ig) boundto immo-bilized bovine serum albumin

(BSA)-neoglycoproteins

con-taining sLex or sLea in a

Ca2"-dependent

manner.

Solution-phasesLextetrasaccharide blocked this interaction by 50%at a

concentration of 750±20

,M

(IC50).

sLe' was moreeffective

(IC50

= 220±20

,uM),

whilenonsialylated, nonfucosylated de-rivatives showed littleor noactivityatconcentrationsup to 1 mM.Attachmentofan8-methoxycarbonyloctylaglycone in a A

linkagetotheanomeric carbonof the GlcNAc of sLexor

sLew

increased their blocking activity nearly twofold. Finally, re-placementofthe2-N-acetylsubstituent oftheGlcNAc by an

azidooraminogroup resulted insubstantial increases in

activ-ity, with the most potent inhibitor being amino substituted sLea, whichwas36-fold moreactive

_(IC50

= 21±3

tM)

than the reducing tetrasaccharide sLex. In contrast to results

ob-tained with

E-selectin-Ig, P-selectin-Ig

bindingtoimmobilized BSA-sLeawasblockedmodestly bymostoligosaccharidesat 1 mM,withnosubstantial differencesamongthem.

_IC50

values ofsolubleoligosaccharides determined incompetitivebinding studies accurately predicted blocking of leukocyte adhesionto

recombinant E-selectin-Ig and tocytokine-activated endothe-lium.(J. Clin.Invest. 1993. 91:1157-1166.) Key words:

adhe-sion*carbohydrate*P-selectin (CD62)*sialylLewis a * sialyl

Lewisx

Introduction

Theselectin

family

of molecules supportsadhesion of leuko-cytes to theblood vessel wall-akeystepin theinflammatory

response toinjuryand infection. Thethree known members,

designated

E-, P-, and L-selectin (for nomenclaturesummary, seereference 1),arestructurallysimilartransmembrane

glyco-Addressreprintrequests to Dr.Michael Bevilacqua, Howard Hughes Medical Institute, Cellularand MolecularMedicine, Room 331, Uni-versity of California, San Diego, 9500 Gilman Drive, La Jolla, CA 92093-0669.

Receivedforpublication 10 August 1992 and in revised form 11 November 1992.

1. Abbreviations usedin this paper: Fuc, fucose; Gal, galactose; Glc, glucose; GlcNAc,N-acetyl glucosamine;IC50,concentration of

inhibi-proteins, each containing an amino-terminal lectin-like

do-main, anEGFrepeat,andavariable number of complement regulatory-likerepeats(2-7).E-selectintranscriptionand sur-face expression by vascular endothelial cells are transiently

upregulated(maximal at 2-6 h) after stimulation by bacterial

endotoxin and by cytokinesassociated withinflammation,

in-cluding IL-I and TNF(2, 8). Experiments in vitro have

de-monstrated that E-selectincan supporttheadhesion of PMN,

monocytes,andasubpopulationof Tlymphocytes (2, 8-18). P-selectin (CD62) is storedina-granulesanddensegranulesof

plateletsaswellasinWeibel-Palade bodies of endothelial cells, and israpidly (minutes)redistributed to the cell surface after

activation bythrombin and other mediators( 19-23),where it supportsleukocyte binding (24, 25). L-selectin is presenton

the surface ofmostlymphocytesandmediates their adhesion to high endothelial venules foundin

peripheral lymph

nodes

(reviewed in references 26 and 27). L-selectin is also foundon

thesurface ofcirculatingPMNand monocytes, and can

sup-port their adhesion to

cytokine-activated

endothelium (28-32).Recent reports havesuggested that the selectinsare

espe-cially

important

in

leukocyte rolling

onthevesselwall, a pro-cess that can precede firm attachment and extravasation

during inflammation (33-35).

Carbohydrate recognition by selectins has been demon-strated in several studies (reviewed in reference 36).

E-se-lectin binds sialylated, fucosylated lactosamine structures, notably sialyl-Lewis x (sLex,

Neu5Aca2-3Galf1-4[Fuca

1-3]

GlcNAc)' (37-41).

E-selectinmayalsobindto

oligosaccha-rides relatedtosLex and sLea in which the sialic acid is replaced byasulfategroup(42), andto

fucosylated

structuresthat lack sialic acid (43). The

pentasaccharide

lacto-N-fucopentaoseIII

(LNF-III)

contains the Lewis x (Lex) determinant and was

foundtoblockthe

rosetting

ofactivated plateletstoleukocytes,

aprocess

involving

P-selectin (44). Other investigationshave

demonstrated thatsialic acid isacomponentofsomeP-selectin ligands(45,46),and thatoligosaccharides containing sLexare

recognized by P-selectin (47,48). More recently, the recogni-tionofthe sLex structure wasextendedto murine L-selectin

(49, 50).

Human E-andP-selectin and

murine

L-selectin also bind to molecules containing sLea

(Neu5Aca2-3Gall

1-3 [Fuca1-4]GlcNAc),anisomer of sLex(50-53). L- and

P-se-lectin have also been showntointeract with sulfated

polysac-charides suchas heparin, fucoidan, and dextran sulfate, and

with sulfatides, glycosphingolipids containing monosulfated

torresultinginthereduction ofbindingto50% ofmaximum;LNF-III,

lacto-N-fucopentaose III (Gal, I-_{4 [ Fuca 1}-3]

GlcNAcO

1-3Gal

I1-4Glc); mOD450,milli-optical densityunitsat450nm;Neu5Ac, N-ace-tyl neuraminic acid (for abbreviations of oligosaccharides,seeTableI);

OPD,o-phenylenediamine dihydrochloride; selectin-Ig,selectin immu-noglobulin fusion protein (see Fig. 1,left panel); sLea, sialyl Lewisa;

sLex, sialyl Lewis x; thimerosal,

mercury-[(o-carboxyphenyl)thio]-ethyl sodiumsalt; Tween-20,polyoxyethylenesorbitanmonolaurate.

Higher-AffinityOligosaccharide Ligands for E-Selectin 1157

J. Clin.Invest.

©3

TheAmericanSocietyfor ClinicalInvestigation,Inc. 0021-9738/93/03/1 157/10 $2.00

galactose (53-57). Themoleculardetails of selectin recogni-tion of carbohydrates, both oligosaccharides and polysaccha-rides,remaintobedetermined.

Wehaveexaminedcarbohydrate binding interactions of

E-andP-selectinusing rationally designed synthetic oligosaccha-ridesin threeassaysofincreasing biological complexity. First,a

competitive ELISAwasdevelopedtoquantitate the ability of solution-phase carbohydratestoinhibit thebinding of

selectin-Igfusion proteinstoimmobilized BSA-sLexorBSA-sLea

neog-lycoproteins inapurified component system.Second, a cell-protein adhesion assay was usedto determine the ability of soluble carbohydratestoinhibit the adhesion of cells bearing carbohydrate ligandstopurified, immobilized recombinant

se-lectin-Ig fusion proteins. Third, a cell-cell adhesion assay was employedtodetermine theability of carbohydratestoinhibit the adhesion ofPMN to cytokine-activated cultured human

umbilical vein endothelialcells,known toexpressE-selectin on

their surface. Our results indicatethat E- and P-selectin have

distinctcarbohydrate recognition characteristics, and that

se-lectedmodifications of theoligosaccharides sLexand sLea re-sultinhigher-affinity ligands forE-selectin. Thesefindings

sug-gestavenuesfor thedevelopmentof novelanti-inflammatory therapeutic agents.

Methods

Proteins

Recombinantfusion proteinscomposedofextracellularportions ofthe humanselectinsjoinedtohumanimmunoglobulin heavy chain CH3,

CH2,andhinge regions (Fig.1,leftpanel)wereconstructed, expressed, andpurifiedaspreviouslydescribed(39,57, 58). These soluble

selec-tin-immunoglobulin fusion proteins (selectin-Ig) contain the signal sequence,lectin-like domain,andEGF repeatalong withsix,two,and twoofthecomplement regulatory-likemodulesforE-, P-, and

L-selec-tin-Ig, respectively. Selectin-Ig puritywasassessedby SDS-PAGE

fol-lowed bysilver staining; proteinconcentrationwasdetermined using theBradfordreagent. Molecularweightsusedfor calculation ofmolar

concentrationsare:E-selectin-Ig, 170,000; P-selectin-Igand

L-selectin-Ig, 120,000.BSAwasprotease-free grade fromBoehringer-Mannheim

Biochemicals, Indianapolis, IN. BSA-neoglycoproteins containing

sLeX, sLea,

Le[,

orLea (kindly provided byChembiomedLtd., Ed-monton,Alberta)weremadeaspreviously described (59).

Oligosac-charidesweresynthesizedwithanaliphatic aglycone (8-methoxycar-bonyloctyl; -(CH2)8CO2CH3)attached ina/3linkagetocarbon 1of

theGlcNAc(Fig. 1, right panel)(60). This modification facilitated

covalent attachmentofoligosaccharidestoBSAthroughtheformation

ofacylazides.Incorporation oftheaglyconeglycosideformsofsLeX, sLea, Lex, and Lea intotherespective BSA-neoglycoproteins(mole ratio)was9, 13, 13,and 17. Antibody G559againstP-selectinwas

provided by R. McEver, University ofOklahoma Health Science

Center, CardiovascularBiologyResearch Program, Oklahoma Medi-cal ResearchFoundation,OklahomaCity,OK.

Carbohydrates

Amajorityof the oligosaccharides used in these studieswereprovided byChembiomed, Ltd.,and wereunambiguouslysynthesized usinga

chemoenzymaticroute,purified by chromatographicmethods with the final steponBiogel P-2, and freeze-dried invacuo(P.V.Nikrad,M. A.

Kasham,K.Wlasichuk, andA.P.Venot, Chembiomed Ltd.,

manu-scriptinpreparation). In mostcases, thesecompoundswere

synthe-sized withanaliphaticaglycone(8-methoxycarbonyloctyl; -(CH2)8-CO2CH3) attached ina3linkagetothereducing sugar (59, 60) (Fig. 1,

right panel;TableI).Allof thesecompoundswerepure,asassessedby

'H-nuclear magneticresonance spectroscopy. Reducing

tetrasaccha-IVt,

2

CH21CH3

tCOOH

HCH2

aCHV|OOH

Figure 1.

ridessLex andsLea(sLex OH and sLea'OH;seeTableI)were

purchasedfromOxfordGlycosystemsLtd.(Rosedale, NY).

Oligosac-charides, heparin,andfucoidan(SigmaChemical Co.,St.Louis,MO)

werereconstitutedin Dulbecco'sphosphate-buffered saline (PBS)and storedat-80'C. Sulfatides(heterogeneous3-0 sulfatedgalactosyl

cer-amide) and cerebrosides (galactosyl ceramide) from bovine brain (Sigma Chemical Co.) were reconstituted in chloroform:methanol

(1:1 )at 10mg/mland storedat-20'C.

ELISA measuring

selectin-carbohydrate

interactions

Polystyrene 96-well microtiter ELISA plates (no. 25801, Corning

Glass, Newark,CA)werecoatedwithunconjugatedBSAor

BSA-neo-glycoproteins by incubation overnightat4VC with 1.5Mg/mlin 75,l

per wellof50 mM sodiumcarbonate/bicarbonate buffer,pH9.5. For

coatingsulfatidesorcerebrosides,therequiredvolumeofstock

solu-tionwasdriedinaglasstesttube andredissolvedin methanol (HPLC

grade)atafinalconcentrationof20,g/ml;50Mlwasadded to each microtiterplatewell andevaporatedtodryness in a370Cincubator. All coatedplateswereblockedusing200Mlper well of 20 mgBSA/mlin assaybuffer (20mMHepes, pH7.4, 150 mM NaCl, 0.25mM

thimero-sal) fora minimum of2 h. The remaining stepsofthe assaywere

performedatroomtemperature. Wellswerewashedtwice withassay

bufferfollowedbyaddition of 50Ml oftestsample(containing either

purifiedselectin-Igorselectin-Igandcarbohydrate)in assaybuffer

sup-plementedwith 10 mgBSA/ml,0.05%(vol/vol) Tween-20,andeither

2 mMCaC12or5 mMEGTA.Afterincubation for3 hon arotating platformshakerat75 rpm, theplateswerewashedtwice,and100

,l

per welladded ofperoxidase-conjugated rabbit F(ab')2 anti-human IgG (H+L) (Jackson Immunoresearch Laboratories, Inc., West Grove, PA) diluted 1:3,000in assay buffersupplemented with2 mMCaC12

and 10 mgBSA/ml.After 30 min theplateswerewashed 3times with assaybuffercontaining2 mMCaC12,followed by addition of200,l/

wellof 0.8 mgo-phenylenediamine dihydrochloride(OPD)/mlin 50 mMsodiumcitrate,50 mMsodiumphosphatebuffer, pH 5.0, contain-ing0.015%(vol/vol)hydrogen peroxide.Therateoftheenzymatic reactionwasdetermined induplicateortriplicatewellsby measuring

theabsorbanceat450nmat15-30-stime intervalsusinga

_Vm.,,

kinetic

microplate reader and Softmax software (Molecular Devices, Inc.,

MenloPark,CA).Anendpointdeterminationat490nmwasmade afterstoppingthereaction in the linear range with 50,ulperwellof4 N

H2SO4.Specificbindingwascalculatedbysubtraction of thesignalin wells coated withunderivatized BSA(nonspecific binding)from the

signalinwells coatedwithBSA-neoglycoproteins.Fordetermination ofIC50values,the datawereiterativelyfittotheequation:Fraction of maximalbinding=_{ICo .}

_(ICo

+[oligosaccharide]),using non-linear least squaresanalysissoftware(Enzfitter, Elsevier-Biosoft,Cambridge,

England).

Cell isolation and culture

sLex

OH

R

sLea

HO

AcNH-R

R

= Amino

H

I

-N - H

Azido - N = N = N

H 0

I

11

N-Acetyl - N - C - CF

H 0

I

11

N-Propionyl - N - C - C

0

11

R

=

_-(CH2)8-

C - 0-

CH3

(8-methoxycarbonyloctylglycoside)

-H

(reducingtetrasaccharide)

- _CH3

(methylglycoside)

H3

,H2 - CH3

Figure 1. (Opposite page) Structure ofE-selectin-Ig

fu-sionprotein. Shadedregions

indicate the human immu-noglobulin sequences; un-shadedportionsrepresent human E-selectindomains.

Thisfusionprotein is se-creted from transfected COS cellsas adimericmolecule,

disulfide linked in the immu-noglobulin hingeregion. Ab-breviations:L, lectin-like do-main; E,epidermal growth

factor repeat;CR, comple-mentregulatory-likedomain;

H,immunoglobulin hinge

region;CH2andCH3, im-munoglobulin heavy chain constantregions.Analogous

immunoglobulinfusion

pro-teinscontaininghuman P-and L-selectin sequences were also used in some stud-ies, and these molecules con-tainedtwoCRs in each selec-tinpolypeptide. (Above)

Structures ofsLex, sLea,and derivatives.sLex,

sLea,

and substituents used in modifi-cationsatcarbons 1 and 2 of GlcNAcaredepicted. In naturallyoccurring sLexand

sLea,R=_N-acetyl.For

re-ducingoligosaccharides, R'

=H.Themajority of oligo-saccharidesused inthe pres-entstudieswere 8-methoxy-carbonyloctylglycosides (R'

=_{(CH2)8CO2CH3)}. See Ta-bleIforacompletelistof

abbreviationsandstructures ofcompounds.

collected and washedoncewith cold PBS. Contaminating red blood cellswereremoved viahypotonic lysis.Cellviabilitywasassessedusing

trypan blue exclusion and the cellsresuspendedat aconcentrationof2

X 106cells/ml in PBS containing 0.5 mgofhuman serum albumin (HSA)/ml, 3 mMglucose, and 0.3 Uaprotinin/ml (adhesionassay

buffer).

Endothelialcells. Human umbilical vein endothelial cells

(Clone-tics Corp., SanDiego,CA)wereseeded in tissue culture wells that had beencoatedwith0.1%gelatin inwaterand allowedtogrowto con-fluence in media M199 (Gibco Laboratories,Grand Island, NY)

sup-plementedwith10% FCS, 5 Uheparin/mlandendothelial cell growth supplement (Sigma Chemical Co.)at 15qg/ml. Before adhesion as-says, endothelial cell monolayerswerewashedonce toremovecellular debris and activated with 200UTNF-a/ml(gift from Biogen,

Cam-bridge,MA) in M199 -10% FCS for4hat37°C (8).

Cell-protein

and cell-cell

adhesion assays

Leukocyte adhesion assays were performed in Nunclon MicroWell Terasaki plates (No. 136528, Nunc, Naperville, IL) containing an en-dothelial cell monolayer (see above)orcoatedwith recombinant

selec-tin-Igfusion proteins. To captureselectin-Igmolecules, wells were

in-cubatedwith 10,Iofa Ig/ml solution of10 polyclonalgoat anti-hu-man IgG (-y-chain specific) antibody (OrganonTeknika, Malverne, PA)orproteinA(Chemicon,Temecula,CA) in 50mM carbonate/bi-carbonatebuffer,pH9.5,for 1 hat roomtemperatureorovernight at

4°C.Plateswerewashedtwice with PBS byflooding,andaspirated; 5

_,l

of theappropriate selectin-Igat 10,ug/ml in PBS were added to each well and allowed to incubate for 1 h at room temperature. Plates were washed,incubated for 1 h with PBS containing1%HSA,and washed

again.Wells coatedwith selectin-Ig fusion proteins or containing con-fluent endothelial cell monolayers were incubatedwith5Alof PBS or PBScontainingcarbohydrate for 30 min at 4°C prior to the addition of PMN. PMN(2 x 106cells/ml)were added in5Alof adhesion assay buffer and allowedtoadherefor 15-30min at 4°C.Nonadherent cells were removed by washing with PBS. Adherent cells were fixed by

add-ing2.5% glutaraldehyde in PBS,andcounted over a0.24-mm2area in eachof four replicate wells. Cell counts were averaged and expressed as adherent cells per square millimeter. Adhesion ofPMNtoE-and P-se-lectin-Ig fusion proteins wasinhibitableby specific monoclonal anti-bodiesH18/7andG559, respectively. Adhesion to cytokine-activated endothelial cells was partially blocked by anti-E-selectin antibody H18/7aspreviously described (8).

TableI. Oligosaccharide Abbreviations and Structures

Abbreviation*

sLacNAc

sLex

sLex OH sLex OCH3

sLex(GlcN3) sLex(GlcNH2) sLex(GlcNHPr) 9NH2-sLex KDNLeX

FucI-6sLacNAc

sLec sLea

sLea OH

sLea(GIcN3) sLea(GlcNH2) sLea(GlcNHPr)

2-6sLea

Neu5Aca2-3 Galfl1-4GlcNAcf-OR'

Neu5Aca2-3 GalflI-4(Fuca1-3)GlcNAcfl-OR'

...GIcNAc

...GlcNAcf-OCH3 ...GlcN3,B-OR' ...GlcNH2fB-OR'

...

.GcNHPrB-OR'

9NH2Neu5Ac...GlcNAcB-OR' Neu5OH... . -OR'

Neu5Aca2-3

Galf1l-4(Fuc

a1-6)GlcNAcf-OR'

Neu5Aca2-3 Gal1 1-3GIcNAcf-OR'

Neu5Aca2-3 Gal 1-3(Fuca1-4)GlcNAc3-OR'

...GlcNAc ...

GlcN3#-OR'

...GIcNH2,B-OR' ... ...GcNHPr, OR'

Neu5Aca2-6...

GlcNAco-OR'

*_{s,Neu5Ac; Pr,C(O)CH2CH3; KDN, keto-deoxy-nonulosonic acid(Neu5OH). tDotsindicate identity withthestructureabove; R'}

=_{(CH2)8CO2CH3.}

Results

E-and P-selectin-Igbindtoimmobilized sLea and BSA-sLex. SolubleE-andP-selectin-Ig boundtoBSA-sLea

immobi-lized onmicrotiterplates inaconcentration-dependent man-ner(Fig. 2). E-selectin-Ig binding appearedtodepend entirely

onthecovalentlyattachedoligosaccharide moietyof the

neo-glycoprotein, inthatit failedto bindtoBSA alone.

Further-more, this binding was abolished in the presence of 5 mM

EGTA, consistentwith the function of E-selectinas aC-type lectin (62). In similar studies, E-selectin-Ig boundto

immobi-lized BSA-sLex inaconcentration-and Ca2 -dependent

fash-ion, but displayedverylittlebindingtoBSA-neoglycoproteins containing Lea orLex(at 100 nME-selectin-Ig, - 6 and <2

mOD450/min,

respectively).

P-selectin also bound to immobilized BSA-sLea (Fig. 2) and BSA-sLex (not shown) in a concentration-dependent

manner. Unlike E-selectin-Ig, P-selectin-Ig showed a small

amount ofbindingto immobilized BSA that had not been

de-rivatized with oligosaccharide. In a typical ELISA, 100 nM

P-selectin-Ig bindingtoBSAalonegaveasignalof 9

mOD450/

min., whereas the same concentration ofE-selectin-Ig

pro-duceda BSAsignal of 3 mOD450/min(background signal in

the absenceofselectin-Igwas2-4

mOD450/

min).This

P-selec-tin-Ig bindingtounconjugatedBSAwasapparentlyunrelated toaCa2 -dependentlectinactivitysince itwasenhanced

sever-alfold inthe presence of EGTA. With this high background bindingtoBSAsubtracted,thebindingofP-selectin-Igto

BSA-sLea (Fig. 2) and BSA-sLex was largely blocked by 5 mM

EGTA.

HumanL-selectin-Ig displayedlittleor nodetectable

bind-ingtoBSA-neoglycoconjugates containing sLea, sLex,

Lea,

or

LeXinthepresenceof 0.05% Tween-20. Somepreparationsof

L-selectin-Igdisplayed modest bindingtosLex and

BSA-sLeainthe absence ofdetergent (at 100nM,20-50

mOD450/

min), possibly reflectingthe enhanced avidity ofaggregates.

Consistent with previousreports (53, 57), both L-selectin-Ig and P-selectin-Ig bound to immobilized sulfatides (not shown).In thepresenceofEGTA, L-selectin-Ig bindingto sul-fatideswasreducedby - 50%, while P-selectin-Ig bindingwas

largely unaffected.

Sialic acidandfucosearekeycomponentsofsLea and sLex

bindingto

E-selectin-Ig.

Wehaveexamined the ability of solu-tion-phase syntheticoligosaccharides to block the binding of E-selectin-Igtosolid-phase BSA-sLea. At 500

,uM,

the

tetrasac-charides sLex and sLea withattached aliphatic aglycone (see TableI) blocked specific binding bymorethan 50%, with sLea

100

C

80

C

60 LD

U)'0

40-C.)

OD 20

0

0 20 _{40 60 80} 100

Selectin-Ig (nM)

Figure2. Specific bindingof E-andP-selectin-Igtoimmobilized BSA-sLea. An ELISAwasusedtomeasureconcentration-dependent bindingtoBSA-sLea ofE-selectin-Ig (triangles) and P-selectin-Ig (squares)inthepresenceof 2 mMCaC12 (filled symbols)or5 mM

EGTA(open symbols). Specific bindingtoBSA-sLeawasdetermined

bysubtraction ofbindingtounconjugatedBSAateachconcentration andbuffer condition. As notedinthetext,P-selectin-Igdemonstrated

anunexpectedly high bindingtounconjugatedBSA in thepresence ofEGTA. Symbolsanderrorbarsarethemeanand standard devia-tionofmeasurementsofrateofcolorgenerationfromtriplicatewells.

exhibiting higher activity than sLex (Fig. 3). The non-sialy-latedtrisaccharideLex showed no blocking activity at this con-centration, while Lea demonstrated slight activity ( 13-20%

in-hibition, four experiments).Theability ofsolution-phase

tetra-saccharides sLeaand sLeX to blockbinding of E-selectin-Igto

immobilized BSA-neoglycoproteinswasinterpretedas compet-itive inhibition.

Todeterminestructural requirementsforcarbohydrate rec-ognition by E-selectin, we tested a variety of synthetic

oligosac-charides with modifications ofthe sialic acidand/or fucose

groups. The importance of the linkage of sialic acid to the

sac-charidebackbone was demonstrated usinganisomerof sLea,

designated 2-6sLea,inwhichthesialic acid isattached to galac-tose in an a2-6ratherthan ana2-3linkage. This carbohydrate displayednosignificant blocking activityat aconcentration of

500 uM. Two additional modifications to the sialic acid of

sLex, namely attachment ofan amino group on carbon 9

(9NH2sLeX)orreplacementofthe N-acetyl group at carbon 5 by ahydroxylgroup (KDNLeX),resulted in compounds that had lessactivity than unmodified sLex.

Theimportance of fucose position and linkage insLeaand

sLeX was examined using selected oligosaccharides (Fig. 3).

sLex-and sLea-relatedtrisaccharidesthat lackfucose, namely

sLacNAc

_{(Neu5Aca2-3Galfl1-4GlcNAc}

-OR') and sLec

(Neu5Aca2-3Galf1l-3GlcNAcf-

OR') respectively, had no blockina activitv. In addition. Fuc1-6sLacNAc. acomnnound

1.2 cJ)

c

-o

-

1--CC 0.8-_

a) m

aDc 0.6

Z0

Lii -8- 0.4

0 CD - 0.2

0

-sLea sLea

-OH

sLex sLex sLex

-OH -OCH3 Solubleoligosaccharide (500I1M)

Figure4.Effect of the aliphatic aglyconeonblocking activityof sLea andsLeX.Solution-phase oligosaccharide (500 MM)wasusedtoblock the directbinding of25 nME-selectin-IgtoimmobilizedBSA-sLea. sLea and sLex(second bar ofeachpanel)weresynthesizedwith the aliphatic aglycone,(CH2)8CO2CH3 (8-methoxycarbonyloctyl),

at-tached via linkagetocarbon 1 of GlcNAc. In theexperiment

de-pictedinthe leftpanel, sLea with the aglyconewascomparedtothe reducing tetrasaccharide(sLe' OH).Theright panelcomparesthe activity ofsLexwith theaglyconetothereducingtetrasaccharide and themethyl glycoside.

structurally similar to sLex with fucose I

GlcNAc,showedonlyasmall amountofb

Higher-affinity

carbohydrate ligandsfor dition ofhydrophobicsequences tocertainI

charideshas been shown toincrease their lectins(63, 64). Themajority ofoligosacci

ourexperiments were synthesized with an

Modific

Sialicacid

-T. C

...

E

1.2

C.) c E

*_ _c 1

aD 80.6

(n

cJ to 0.4

u 0

CD

-co x m

U,)(D ()l

cN

x

-j U)

I

z 0C

x

-i z

n

Solubleoligosaccharide

Figure3. Inhibition of E-selectin-Ig bindingtoi by sLea,sLeX,andoligosaccharidesmodifiedin

Solubleoligosaccharides(500gMM)wereincubal nME-selectin-Iginmicrotiter wells coated wit

washing, boundE-selectin-Igwasdetected usin

gated antibodyand colorimetricsubstrate,asc

Meanendpoint absorbancemeasurements(±S

wellsareshown.

linked a1-6 to the (-

(CH2)8CO2CH3)

attachedin a,B-glycosidic linkage to the

dockingat500

,M.

reducing sugar (TableI). The presence of this aglycone was rE-selectin. The ad- found to measurably enhance the blocking activity of sLeaand mono-and oligosac- of sLex (Fig. 4). The methyl

f3-glycoside

of sLex binding affinity for

(sLex'OCH3)

and the reducing tetrasaccharide (sLex Iaridesemployed in -OH)appeared to have identical inhibitory activities.

Titra-ialiphatic aglycone tion experiments indicated that sLea andsLex withthe

agly-cone werenearly twofoldmoreactive than thecorresponding

reducing tetrasaccharides (sLea OH and sLex OH),as

de-,ationsof: terminedby the concentration requiredtoinhibit 50% of

E-se-Fucose lectin-Ig binding(ICso).

By examining tetrasaccharides basedonthe sLea and sLex

- ~ - structureswith substitutionsatthe2 position of GlcNAc,we

identified several compounds having substantially greater T _{blocking activity against E-selectin. Replacement ofthe}

N-ace-tyl group with eitheramino orazido moieties producedthe mostpotentblockers. As shown inFig. 5, sLea(GlcNH2) and sLex(GlcN3) atconcentrations of 500 MMalmostcompletely blocked directbinding of E-selectin-IgtoBSA-sLea.Analogues of sLea andsLexsynthesized with apropionyl substitutionat

thesameposition showed decreased

[sLea(GlcNHPr)]

or

un-changed[sLex(GlcNHPr)]blockingrelative to theparent

tetra-, saccharides. Titration of these tetrasaccharides(Fig. 6,top;

Ta-,, Z bleII) indicated thataminosubstitutionfortheN-acetylgroup

of GlcNAc reduced the

_IC50

to 21±3 MM,anapparentsixfold

C

increase inactivity. Further, thisrepresentsa36-fold increase

in activity over the reducing tetrasaccharide, sLex- OH,

(500

zM)

which hasan

IC50

of

750±20

AM.

The

analogous

azido

substi-tution displayed activity between sLea and sLea(GlcNH2),

Immobilized BSA-sLea _withan

_IC50

_{of 55+14 ,M. As noted above,sXlawas abetter}

tsialic

acid and fucose _{blocker ofE-selectin-Ig binding thanwassLex; the}

_{IC_0}

_ofsLea hBSA-sLea. After was about one-third that ofsLex,comparing either the

reduc-ig anenzyme-conju- ing sugars themselves or their aglycone glycosides. Incontrast

iescribedinMethods. to their effects onE-selectin-Ig,none of thesynthetic

oligosac-3D)fromtriplicate charides testedwereespeciallyeffectiveinblocking

P-selectin-Ig bindingtoBSA-sLea, including the amino and azido

deriva-Higker-Affinity Oligosaccharide Ligands forE-Selectin 1161

12

0)

CE

0)o 0) 0.8

°' 0 0 6

us

o: O 0.4

O 0.2

CLn

0

sLe' sLe' sLe' sLe,' sLe, sLe- sLeX sLex

lGIcN) GIcNH, ,GIcNi ,-, :CcN. cN-! |G c

Solubleoligosaccharide(500

_pIM)

Figure 5.Analoguesof sLe'andsLeXwith amino and azido

substitu-tionsatcarbon 2ofGlcNAc are potent blockersof E-selectin-Ig.

So-lution-phasesLea,

sLeX,

andderivatives(500

_yM)

wereused to block thebinding of 25nME-selectin-Igtoimmobilized BSA-sLea. Struc-turesofazido,amino,andN-propionyl constituentsusedtoreplace

theN-acetylgroupof GlcNAcin sLe' andsLeXareshown inFig.1,

rightpanel.Symbols and error bars represent the mean and standard deviationofmeasurementsmadeintriplicatewells.

tivesof sLea and sLex

(Fig.

6,

bottom).

It is noteworthy that

among

solution-phase oligosaccharide

inhibitors of P-selectin-Ig, both nonsialylated and sialylated structures had some

blocking

activity,

with the latter

being moderately

more effec-tive. Asnoted

above, however, P-selectin-Ig

exhibited a sub-stantially higher levelofbindingtoimmobilized BSA-sLex and BSA-sLea than to their

nonsialylated

counterparts BSA-Lex and BSA-Lea.In separate

experiments,

both

heparin

(a

glycos-aminoglycan composed

of

alternating,

sulfatedresidues of glu-cosamine and uronic

acid)

and fucoidan (a

homopolymer

of

sulfated L-fucose)

blocked the

binding

of

P-selectin-Ig

to BSA-sLe' but hadnoeffecton

E-selectin-Ig binding (Fig.

7).

Blocking

ofcell adhesion by

oligosaccharides

in vitro. Re-sults

ofcompetitive binding

studies described above

accurately

predicted relative effects of

solution-phase carbohydrates

on

leukocyte

adhesion

to

purified

selectin-Ig

fusion

proteins

and

to

cytokine-activated endothelial monolayers.

Ata

concentra-tion

of1 mM, sLex blocked adhesion ofPMNtoimmobilized recombinant

E-selectin-Ig

by - 50%

(Fig.

8).

sLex showed

lit-tle or no

blocking of

PMNadhesionatconcentrationsupto 1

mM, a result

consistent

with its

IC50

being

threefold

higher

than that of sLea in the

competitive binding

ELISA

(Table II).

In

further

agreementwith the

competitive binding results,

the ability of

oligosaccharides

toblock celladhesionto

E-selectin-Ig required sialic acid linked a2-3 to galactose, and fucose linkedtotheGlcNAc. Compoundswith sialic acid linkeda2-6 togalactoseaswellascompoundslacking sialic acidorfucose failedtoblockPMNadhesionatconcentrationsup to 1 mM

(data not

shown).

Further,

sLea(GlcN3)

andsLea(GlcNH2)

weremuch more active than

sLea

in blockingtheadhesion of PMN toE-selectin-Ig.

sLea(GlcNHPr),

which has an

N-pro-pionyl group substituted forthe N-acetyl group ofGlcNAc,

failedtoblockadhesionat aconcentration of1 mM(Fig. 8).In separateexperiments,PMNadhesiontoimmobilized

P-selec-tin-Ig

waspoorly blocked by soluble oligosaccharidesrelated to

sLexandsLea (not shown). Althoughseveraloligosaccharides

1.2

0)

C

a0 8

0)0

0)O.8

'

tX

r, cuo

0 CD)U

. co0.6

C _0

o coQ4

=ci)

m.

0.2

CD)

0

Solubleoligosaccharide (isM)

1.4

C 1.2

Cl

CU

0.8-CD 0

0.

(1) 0.2

0 10 100 1000

Soluble oligosaccharide

(iiM)

Figure 6.Concentration-dependent inhibitionof E- and P-selectin-Ig

bindingtoimmobilized BSA-sLea by synthetic oligosaccharides.

(Top) Datafromarepresentative competitive binding assay measur-ing theinhibition of specific E-selectin-Ig binding toBSA-sLeabysix relatedoligosaccharides.(Bottom) Inhibition of specific P-selectin-Ig

binding bythesameoligosaccharides.Assays wereperformedin the presenceof2 mMCaCl2. Symbols and bars represent the mean and range ofmeasurementsmadeinduplicatewells, except at0mM sol-ubleoligosaccharide,wherethemeanand standarddeviationof six wellsis shown. Key: o,LeX;o,Lea; . sLeX;* sLea;

A, sLea(GlcN3);

v,

sLe'(GlcNH2)

. Structuralrepresentationsofoligosaccharidesare shown in Table I andFig. 1.

(including nonsialylated and nonfucosylated structures)

blockedPMN adhesionto

P-selectin-Ig

in certain assays, the

effect was modest(typically <30% inhibition) and

inconsis-tent. Neither azido nor amino

substitutions

at carbon 2 of

GlcNAcenhanced the

ability

ofthe compounds to block leuko-cyteadhesionto

P-selectin-Ig.

In a morecomplex adhesionassay, weexamined theeffects of

solution-phase oligosaccharides

ontheadhesion ofPMNto

human umbilical veinendothelial cell

mdnolayers.

Cytokine activated endothelialcells expressE-selectinin

abundance,

and

thereby

support

leukocyte

adhesion (8). Asshown in

Fig. 9,

sLeaat aconcentration of1 mM

partially

blocked PMN

adhe-sion.

sLea(GlcNH2)

wasthemostactiveblockerin thissystem,

followed

by

sLea(GlcN3).

Moreover,sLeXwas

consistently

less

activethan sLea in

blocking

PMNadhesiontoactivated endo-thelialcell

monolayers,

typically blocking

0-30%of adhesion.

Oligosaccharide

inhibitiondatafrom thethree assayswerein

-I

I

T

I

I T

I

f

a

TableII. Comparison of InhibitoryPotency ofSoluble Oligosaccharides Assessed in E-Selectin-Ig Competitive Binding Studies

Relative

Oligosaccharide IC50* blockingactivity AMM

sLex'OH 750±20 (2) 1.0

sLex 380±70 (4) 2.0

sLea OH 220±20 (1) 3.4

sLea 130±20 (4) 5.8

sLex(GlcNH2) 77±7(1) 9.7

sLea(GlcN3) 55±14 (2) 14

sLea(GlcNH2) 21±3 (4) 36

*_{Concentrationofoligosaccharidethatresulted in 50% inhibition of}

thebinding of solution-phase E-selectin-Ig to immobilizedBSA-sLea,

asmeasured in an ELISA (see text for assay details and Fig. 6 for representative binding data).

_IC50

values were determined using non-linear least squares analysis of the data, as described in Methods. The number of titration experiments performed (n) is shown in parenthe-sesfor eacholigosaccharide.Where n > 1,IC50data represent the mean of n extractedvalues+SD;where n= 1, the extractedIC50±SE

isshown.

general agreement: inhibition ofPMN adhesion to natural

(nonrecombinant) E-selectin expressed by endothelial cells

correspondedtoinhibition ofPMNadhesionto recombinant E-selectin-Ig, andtoinhibition of solution-phase E-selectin-Ig

binding to immobilized BSA-neoglycoproteins. Ofthese as-says, the competitive bindingELISA was themostsensitive.

Anexampleofthedifferencein sensitivity ofthe assays can be seenby comparingthedegreeof inhibitionineachassay sys-tem(Figs. 5, 8,and9).

Discussion

Knowledge that theselectins contain regions homologousto

carbohydrate recognition domains of C-type lectins has

1.4

0)

.' -1.2

xs E

.'

C)

Q0

o1

_

0) a)

.' -0.8

._ CZ

-5 c

0.6-C

0.4

= _(n

a ~-0.2

Ct) 0

E-Selectin-Ig

Heparin Heparin Fucoidan

Figure 7. Inhibition ofE- andP-selectin-Ig bindingtoimmobilized BSA-sLeabysulfated polysaccharides.E- and P-selectin-Ig (25 nM)

wereincubated with0(open bars), 0.1, 1.0,and 10(increasingly

darkerbars) gg/mlofheparinorfucoidan inacompetitive binding

ELISA.Symbolsanderrorbarsrepresentthemeanandstandard

de-viation ofmeasurementsmadeintriplicate wells.

1400

ol 1200

E

E 1000

C)

-o 800

o 600

m

Z ₄₀₀

02

200

0-sLea

sLea

sLea

sLea

(GIcN3) (GIcNH2) (GIcNHPr)

Soluble

oligosaccharide (1 mM)

Figure8.Inhibition ofPMNadhesiontoimmobilizedE-selectin-Ig by synthetic oligosaccharidederivatives of sLea.Solution-phasesLea anditsamino-, azido-, andN-propionyl-substituted derivativeswere

tested for theirabilitytoblockPMNadhesiontoimmobilized E-se-lectin-Ig immobilizedonmicrotiterwells.Oligosaccharideinhibitors werepresent throughout the adhesion assay, except in the control wells(first bar). Data are expressedasthemeanandstandard devia-tionof cells counted inquadruplicatewells.Backgroundadhesionto

anonspecific immunoglobulinfusionproteinwassubtracted from allvalues, andwas<5%of adhesiontoE-selectin-Ig.

prompted anintensive efforttoidentify carbohydrate

ligands

(36).Todate,mostdataon

selectin-carbohydrate binding

in-teractionshave beengenerated

using

cellular systemsto

facili-tate thecharacterization ofnatural

ligands.

A

major

successof

thiswork has been the

recognition

thatsLex,astructurefound inabundanceon PMN andmonocytes(65-67)isaligandof

E-selectin (37-41, 5 1). Other studieshavedemonstratedthat

100

60 c0 0 .0

_0

c ₄₀

s

0

00

z 20

CL

0-sLea sLea sLea

(GIcN3) (GIcNH2)

Soluble

oligosaccharide

(1

mM)

Figure9. Inhibition of PMN adhesiontocytokine-activated endothe-lialcellsbysynthetic oligosaccharidederivatives of sLea. Solution-phasesLeaand its amino-, azido-, andN-propionyl-substituted deriv-ativesweretested for theirabilitytoblock PMN adhesionto

con-fluenthuman umbilical vein endothelial cells that had been stimulated for4 hwith TNF-a. Dataareexpressed asthemeanand

standard deviation of cells countedinquadruplicatewells.Control

wellsmeasuredadhesion in the absenceofadded oligosaccharide (first bar).

Higher-Affinity Oligosaccharide Ligands forE-Selectin 1163

_1

_1

_-.

...

..

... ... .

T.. .. ... ...

.-P-selectin-lg

IE

-I-theoligosaccharide sLea, foundon colon cancer cells,isbound

by E-selectin (39, 68, 69). Recent reports demonstrate that

human P-selectinandmurine L-selectin canalsobind carbo-hydrates bearing sLex and sLea(47, 49, 50, 53). Studies on

selectin recognition of oligosaccharides have beenparalleled by

investigations suggestingthatcertain proteins contributeto

cel-lular ligands ofL- and P-selectin. Forexample, immobilized L-selectinhasbeen usedtoaffinity purify from murinelymph

nodesa50-kD sulfated, fucosylated, and sialylated

glycopro-tein,aswell as arelated90-kDglycoprotein (70). Molecular cloning

of

the 50-kD glycoprotein revealedaheavily glycosy-lated mucin-type molecule containing two serine/threonine richdomains (71 ). Separate studies have revealed that P-selec-tin bindsasmall number of protease-sensitive sites on PMN

and HL60 cells(46). Further, P-selectin has beenused to affin-ity purify and ligand blot a glycoprotein(s) from leukocyte extracts, with an apparent

_M,

on polyacrylamide gels of

- 250,000 under nonreducing conditionsand 120,000 under

reducing conditions (72). Togetherthesestudiessuggest that

both P- and

L-selectin bind with

high

affinity

to arelatively small number of cell surface

glycoproteins.

The

function

of the protein in these ligands remainsto bedetermined.

Our

approach

has beento use

synthetic oligosaccharides

relatedtosLexandsLeatodetermine fine specificity of selectin

recognition

andto identify

high-affinity

ligands for potential

use asantiinflammatory

therapeutic

agents. Towardthis end,

we haveemphasized the measurement of binding interactions between

purified

macromolecules and their inhibitionby solu-tion-phase oligosaccharides. Moreover, we have chosen to

focus

on

oligosaccharides

that show

inhibitory activity

at

con-centrations

< 1 mM.Our

observations

supportthe

hypothesis

that

E-selectin

binding

tosLea and sLex involves the sialic

acid,

galactose, and fucose

residues,

inagreementwith

previous

re-ports

(51, 52). First, nonsialylated

and

nonfucosylated

oligo-saccharides

corresponding

toboth sLeaand sLex hadlittle or no effecton E-selectin interactionsatconcentrations up to 1 mM.Second, E-selectin failedtorecognizeastructuralisomer of sLea

(2-6sLea)

with the sialic acid linked a2-6

(rather

than a2-3) to galactose.

Finally,

a trisaccharide similar to sLex (Fuc

l-6sLacNAc) with

fucose linkeda1-6

(rather

than a1-4) hadlittleor no

activity

inourassays.Our studies also

point

to

the

importance of

the

N-acetyl

groupatcarbon 5

of

Neu5Ac because the

oligosaccharide KDNLex,

which containsa

differ-enttypeof sialic acid

lacking

thisgroup

(KDN

=

keto-deoxy-nonulosonic acid; for review of sialic acid diversity, see

refer-ence 73), was

significantly

less active in E-selectin assays

than sLex.

Priorreports(51, 52) have

presented conflicting

data on

therelative activities

of

sLeXand sLeadeterminants in cell

ad-hesionassays; we note thatusing eithersoluble

reducing

tetra-saccharides or

8-methoxycarbonyloctyl glycosides,

sLea was

themorepotentblocker of

E-selectin-Ig binding (by

approxi-mately threefold) andofE-selectin-dependent adhesion.

Fur-thermore, we have demonstratedthat

replacement

ofthe

N-acetyl (-

NHCOCH3)

group of sLex or sLea with amino (-

NH2)

orazido (-

N3)

moieties enhances their interaction

withE-selectinby anadditional four-tosixfoldin the

competi-tive ELISA(Table

II).

Incontrast,replacementof theN-acetyl

group with an N-propionyl group (- NHCOCH2CH3) de-creased theblockingactivity of sLeaand hadlittleornoeffect

on sLex

(Fig.

5). It has been

reported

thatatetrasaccharide related to sLex with glucose in

place

of GlcNAc is a better

inhibitor than sLex ofE-selectin-dependent adhesion (52)and

of

E-selectin-Ig binding

to immobilized sLex

glycolipid (49).

Taken

together,

these datasuggestthat

replacement

of the

N-acetyl

groupwithasmaller

moiety

orone

capable

of

participat-ing

inionic bondscan

strengthen binding

to

E-selectin.

The

majority

ofour

studies

were

performed using

oligosac-charides

synthesized

withan

aliphatic

aglycone

attached in

f3-glycosidic linkage

tothe

reducing

sugar. Thepresenceof this

structureenhanced the

blocking activity

of

sLex and sLea

ap-proximately

twofold.

This

phenomenon

maybe dueto

stabili-zation

of

the

carbohydrate

structure

by

the

aglycone,

making

it

abetter

ligand,

oritmaybe theresult of the

aglycone

contribut-ing

anadditionalcomponentof

binding

eitheroutsideorinside the

carbohydrate

binding

pocket.

Enhanced lectin

recognition

of

carbohydrates derivatized

with

hydrophobic

groups has been observed in othersystems. For

example,

the

elderberry

bark

lectin,

which binds

NeuSAc-Gal/GalNAc,

shows in-creased

affinity

for

glycosides

with added

_#-linked

nonpolar

aglycones

(63).

More

recently,

influenza

hemagglutinin

was

shownto

display

increased

affinity

for

monovalent

sialosides

after

substitution

with

aglycones terminating

in

aromatic

struc-tures

(64).

The

possibility

of

producing

higher-affinity

selectin

ligands

relatedtosLeaor

sLex by

the

addition of

other

nonpo-largroupsdeserves attention.

As noted in the

introduction, prior

studies have demon-strated thathuman E-andP-selectin andmurine L-selectincan

bind sLex and

sLea,

suggesting

that

they

have similaror identi-cal

_{oligosaccharide recognition properties.}

Other

investigations

havedemonstrated

differential

adhesion of certain cellsto

E-and

P-selectin

aswellas

differential

protease

sensitivity

ofputa-tive cellular

ligands (47, 48,

51,

74).

These results suggest that

E- and P-selectin may

recognize

distinct

ligands,

butdo not

reveal whether the

differences

reside in the

oligosaccharide

structure itselfor in its mode

of

presentation

(e.g.,

N-linked

sugar,

0-linked

sugar,

glycolipid),

orboth.In our

_studies,

E-and P-selectin

clearly

demonstrate differential

recognition

of

oligosaccharides

insolution. Both sLexandsLeademonstrated

substantial

blocking

activity against E-selectin-Ig,

whereas their

nonsialylated

counterparts had little or no

effect

up to

concentrations of

1 mM.

By

contrast, both the

nonsialylated

and

sialylated

structures were

found

toblock

P-selectin-Ig,

al-beit

relatively weakly.

Asassessed inboth cell

adhesion

assays

and in the

competitive binding

ELISA,

replacement

of

the

N-acetyl

group

of

GlcNAc

by

aminoorazido

moieties

profoundly

enhanced

interactions

of sLeaandsLex with

E-selectin,

buthad

nomeasurable effectontheir interactions with P-selectin.The

molecular basis for

differential

recognition

of

oligosaccharides

by

E-and

P-selectins

remainstobedetermined.

A recent

study

has demonstrated that

multimeric

com-plexes

containing

murine

L-selectin-Ig

boundtoimmobilized

sLex-glycolipid

(49).

This interactionwasinhibited

by

50%in

the presence 5 mMsLex. In our

_studies,

human

_{L-selectin-Ig}

displayed

littleornodetectable

binding

to

BSA-neoglycocon-jugates

containing

Lex,

sLeX, Lea,

orsLea inthe presenceof

Tween-20,

addedtominimize the

_possibility

of

protein

aggre-gation.

Thesecombined observations

suggest

that

_multiplicity

of

binding

may be

important

forL-selectin

recognition

of sLex

and sLea.

Nevertheless,

potential

interspecips

differences in

binding

specificity

and

affinity

need tobe addressed.

Consis-tentwith

previous

observations

(56, 57),

L-and

P-selectin-Ig

boundto immobilized sulfatides. In

_preliminary

studies,

failed to block this interaction (Drs. Nelson and Bevilacqua, unpublishedobservations).

Leukocyte adhesion to the vessel wall and extravasationare

essential for host defense.However, incertainsettingssuch as

adult respiratory distress syndrome, ischemic reperfusion in-jury, and arthritis, leukocytes cause substantialtissuedamage. Theimportance of E-selectin in leukocyte adhesion and inflam-mation is now widely accepted. In an effort toidentify

anti-in-flammatorytherapeutic agentswehave chosen to focus on solu-bleoligosaccharides relatedtosLex, thefirstdescribed ligand for E-selectin. We have demonstrated thatsolution-phasesLea isa more potentblocker of E-selectin than is sLex.

Further-more,addition of an aliphaticaglyconeincombination withan

amino substitution on theGlcNAcof sLea resulted ina com-pound with 36-fold higher activity thansLex,as measuredin a

competitive binding assay. These studies have also demon-strated differential recognition ofoligosaccharides by E- and

P-selectin. It is anticipated that high-affinity carbohydrate

li-gands for E-selectin may prove usefulinthe treatment of

hu-mandiseases.

Acknowledgments

We are grateful to the scientists atChembiomed, Ltd. (Edmonton, Alberta), including A. P. Venot, M. A. Kasham, P. V. Nikrad, R.

Ippolito,and F. Ungerfor generously providingmostof the

oligosac-charides used in these studies and forhelpfuldiscussions.Inaddition,a special thanksto A. P. Venotfor his deep interestinthis work and for critical review of themanuscript.Wethank R. McEver(Universityof Oklahoma HealthScienceCenter)forantibodyG559against

P-selec-tin,andM.Gimbrone(Brighamand Women'sHospital,Harvard Med-ical School, Boston, MA) for F(ab')2 fragments ofantibody H18/7

againstE-selectin. We thankA.Corriganfor expert assistance in

manu-script preparation. Dr. Bevilacqua is an investigator in the Howard Hughes Medical Institute and a Pew Scholar in the Biomedical

Sciences.

References

1. Bevilacqua. M.P.,E. Butcher,B. Furie, M.Gallatin,M.Gimbrone,J.

Harlan, K.Kishimoto,L.Lasky, R. McEver, J. Paulson,etal.1991.Selectins:a

familyof adhesion receptors.Cell.67:233.

2.Bevilacqua,M.P., S.Stengelin, M.A.Gimbrone,Jr., and B.Seed. 1989. Endothelial leukocyte adhesion molecule1: aninducible receptor for neutrophils related to complement regulatory proteins and lectins. Science (Wash. DC).

243:1160-1165.

3.Siegelman,M.H.,M. vandeRijn,and I. L.Weissman. 1989. Mouselymph

nodehomingreceptorcDNA clone encodesaglycoprotein revealingtandem

interactiondomains.Science(Wash.DC).243:1165-1172.

4.Johnston,G. I., R. G.Cook,and R.P.McEver. 1989.CloningofGMP- 140,

agranule membraneproteinof platelets and endothelium: sequence similarityto

proteinsinvolved in cell adhesionandinflammation. Cell. 56:1033-1044. 5. Lasky, L.A.,M.S.Singer,T. A.Yednock, D. Dowbenko,C.Fennie,H.

Rodriguez, T. Nguyen, S. Stachel,andS.D.Rosen. 1989. Cloning of a

lympho-cytehomingreceptorrevealsalectindomain.Cell.56:1045-1055.

6.Tedder,T.F., C. M. Isaacs, T. J. Ernst, G. D. Demetri, D.A.Adler,and C. M.Disteche. 1989. IsolationandchromosomallocalizationofcDNAs

encod-inganovel humanlymphocytecell surface molecule, LAM-I. J.Exp. Med.

170:123-133.

7.Camerini,D.,S. P.James,I.Stamenkovic, andB.Seed.1989.Leu-8/TQl

isthe humanequivalent of theMel-14 lymphnode homing receptor.Nature

(Lond.).342:78-82.

8.Bevilacqua,M.P.,J.S. Pober,D. L.Mendrick,R. S. Cotran,andM. A.

Gimbrone,Jr.1987. Identification ofaninducible endothelial-leukocyte

adhe-sion molecule. Proc. Natl.Acad.Sci. USA. 84:9238-9242.

9.Hession, C.,L.Osborn,D.Goff, G. Chi-Rosso,C.Vassallo,M. Pasek, C.

Pittack, R.Tizard,S.Goelz, K.McCarthy,etal. 1990. Endothelialleukocyte

adhesion molecule1:directexpressioncloningandfunctional interactions.Proc.

Natl.Acad.Sci. USA.87:1673-1677.

10.Graber, N.,T.V.Gopal,D.Wilson,L. D.Beall,T.Polte,andW.

New-man.1990. T cells bindtocytokine-activatedendothelial cells viaanovel,

induc-iblesialoglycoprotein and endothelial leukocyte adhesion molecule-. J1. Im-munol. 145:819-830.

11.Picker,L.J., T.K.Kishimoto, C. W.Smith,R. A.Warnock,and E. C. Butcher. 1991. ELAM-Iisanadhesion molecule forskin-homingT cells.Nature

(Lond.).349:796-799.

12.Shimizu,Y., S. Shaw, N. Graber, T. V.Gopal,K. J.Horgan,G. A.van Seventer,and W. Newman. 1991. Activation-independent binding of human

memoryTcellstoadhesion molecule ELAM- 1. Nature (Lond.). 349:799-802. 13.Carlos, T.,N.Kovach,B.Schwartz,M. Rosa, B. Newman, E. Wayner, C. Benjamin, L. Osborn, R. Lobb, and J. Harlan. 1991. Humanmonocytesbindto twocytokine-induced adhesive ligandsoncultured human endothelial cells:

endo-thelial-leukocyteadhesion molecule-I and vascular cell adhesion molecule-I. Blood.77:2266-2271.

14.Shimizu, Y., W. Newman,T.V.Gopal, K. J. Horgan, N. Graber, L. D.

Beall,G.A. vanSeventer,andS. Shaw. 1991. Four molecularpathwaysof T cell adhesion to endothelial cells: roles of LFA-1, VCAM- 1, and ELAM- 1 and

changesinpathway hierarchyunder different activation conditions. J. Cell Biol.

113:1203-1212.

15. Hakkert,B.C., T. W. Kuijpers, J. F. M. Leeuwenberg, J. A. Van Mourik, and D. Roos. 1991.Neutrophil andmonocyteadherencetoand migrationacross

monolayersof cytokine-activated endothelial cells: the contribution of CD 18, ELAM-1,and VLA4.Blood. 78:2721-2726.

16.Luscinskas,F.W., M. I.Cybulsky, J. M. Kiely, C. S. Peckins, V. M. Davis,

andM. A.Gimbrone, Jr. 1991. Cytokine-activated humanendothelial

mono-layerssupportenhancedneutrophil transmigration viaamechanism involving both endothelial-leukocyte adhesionmolecule- I and intercellular adhesion

mole-cule-i.J. Immunol. 146:1617-1625.

17.Kuijpers, T.W.,B.C. Hakkert, M. Hoogerwerf, J. F. M. Leeuwenberg, and D. Roos. 1991. Role of endothelial leukocyte adhesion molecule- I and

plate-let-activatingfactor in neutrophil adherencetoIL-1-prestimulated endothelial cells: endothelial leukocyte adhesion molecule-l-mediated CD18 activation. J. Immunol. 147:1369-1376.

18. Leeuwenberg, J.F.M.,T.M.A. A.Jeunhomme, and W.A.Buurman. 1992.RoleofELAM- Iin adhesion of monocytes to activated human endothelial cells. Scand. J. Immunol. 35:335-341.

19.Hsu-Lin, S. C., C. L. Berman, B. C. Furie, D. August, and B. Furie. 1984.

Aplatelet membraneprotein expressed during platelet activation and secretion. Studies usingamonoclonal antibody specific for thrombin-activated platelets. J. Biol. Chem. 259:9121-9126.

20. Stenberg, P. E., R. P. McEver,M. A.Shuman, Y. V. Jacques, andD.F. Bainton. 1985.A platelet alpha-granule membrane protein (GMP-140)is

ex-pressedontheplasma membrane after activation. J. Cell Biol. 101:880-886. 21.Bonfanti, R.,B.C.Furie, B. Furie, and D. D. Wagner. 1989. PADGEM (GMP140) isacomponentof Weibel-Palade bodies of human endothelial cells. Blood. 73:1109-1112.

22. McEver, R. P., J. H. Beckstead, K. L. Moore, L. Marshall-Carlson, and D.F.Bainton. 1989. GMP-140, a platelet alpha-granule membrane protein, is alsosynthesized by vascular endothelial cells and is localized in Weibel-Palade bodies. J. Clin. Invest. 84:92-99.

23.Israels, S. J.,J.M.Gerrard, Y. V. Jacques, A. McNicol, B. Cham, M. Nishibori, and D. F. Bainton. 1992. Platelet dense granule membranes contain both granulophisin and P-selectin (GMP- 140). Blood. 80:143-152.

24. Larsen, E., A. Celi, G. E. Gilbert, B. C. Furie, J. K. Erban, R. Bonfanti, D.D.Wagner,and B.Furie. 1989.PADGEMprotein:areceptor that mediates the interaction of activated platelets with neutrophils and monocytes. Cell. 59:305-312.

25. Geng, J-G., M. P. Bevilacqua, K. L. Moore, T. M. McIntyre, S. M. Pres-cott,J. M. Kim, G. A. Bliss, G. A. Zimmerman, and R. P. McEver. 1990. Rapid neutrophil adhesion to activated endothelium mediated by GMP-140. Nature

(Lond.). 343:757-760.

26. Stoolman, L. M. 1989. Adhesion molecules controlling lymphocyte mi-gration. Cell. 56:907-910.

27. Picker, L. J., and E. C. Butcher. 1992. Physiological and molecular mecha-nisms of lymphocyte homing. Annu. Rev. Immunol. 10:561-591.

28. Lewinsohn, D. M., R. F. Bargatze, and E. C. Butcher. 1987. Leukocyte-en-dothelial cellrecognition: evidence of a common molecular mechanism shared byneutrophils, lymphocytes, and other leukocytes. J. Immunol. 138:4313-4321. 29. Kishimoto, T. K., M. A. Jutila, E. L. Berg, and E. C. Butcher. 1989. Neutrophil Mac-IandMEL- 14 adhesion proteins inversely regulated by chemo-tactic factors. Science (Wash. DC). 245:1238-1241.

30. Smith, C. W., T. K. Kishimoto, 0. Abbass, B. Hughes, R. Rothlein, L. V. McIntire, E. Butcher, and D. C. Anderson. 1991. Chemotactic factors regulate lectin adhesion molecule1(LECAM- 1 )-dependent neutrophil adhesion to cyto-kine-stimulated endothelial cells in vitro. J. Clin. Invest. 87:609-618.

31.Spertini, O., F. W. Luscinskas, G. S. Kansas, J. M. Munro, J. D. Griffin,

M. A.GimbroneJr.,and T. F. Tedder. 1991.Leukocyteadhesion molecule-I

(LAM- 1,L-selectin)interacts withaninducible endothelialcellligandtosupport

leukocyteadhesion.J.Immunol. 147:2565-2573.

32. Spertini,O., F. W. Luscinskas, M. A. Gimbrone, and T. F. Tedder. 1992. Monocyte attachment to activated human vascular endothelium in vitro is me-diated by leukocyte adhesion molecule-I (L-selectin) under nonstatic conditions.

J.Exp.Med.175:1789-1792.

33. Ley, K., P. Gaehtgens, C. Fennie, M. S. Singer, L. A. Lasky, and S. D. Rosen. 1991.Lectin-likecelladhesion molecule 1 mediates leukocyte rolling in

mesentericvenules in vivo.Blood.77:2553-2555.

34. vonAndrian,U. H., J. D.Chambers, L. M. McEvoy, R. F. Bargatze, K.-E. Arfors, and E. C. Butcher. 1991. Two-step model of leukocyte-endothelial cell interactionin inflammation:distinct roles for LECAM- I and the leukocyte02

integrinsinvivo.Proc.Natl. Acad.Sci. USA.88:7538-7542.

35. Lawrence, M. B., and T. A.Springer. 1991.Leukocytes roll on a selectin at

physiologic flowrates: distinction fromandprerequisite for adhesionthrough

integrins. Cell. 65:859-873.

36. Bevilacqua, M. P., and R. M. Nelson. 1993.Selectins. J. Clin. Invest. 91:379-387.

37. Lowe, J.B.,L.M.Stoolman, R. P. Nair, R. D. Larsen, T. L. Berhend, and R. M. Marks. 1990.ELAM-l-dependentcell adhesion to vascularendothelium

determined by atransfectedhumanfucosyltransferasecDNA.Cell.63:475484. 38.Phillips,M.L., E. Nudelman, F. C. Gaeta, M. Perez, A. K.Singhal,S.

Hakomori,and J.C. Paulson. 1990. ELAM-Imediatescelladhesionby recogni-tionofacarbohydrate ligand, sialyl-Lex. Science.250:1130-1132.

39.Walz,G., A.Aruffo, W. Kolanus, M. Bevilacqua, and B. Seed. 1990.

Recognition byELAM-I ofthesialyl-Lex determinantonmyeloidand tumor cells.Science(Wash.DC).250:1132-1135.

40.Goelz,S. E.,C.Hession,D.Goff,B.Griffiths,R.Tizard,B.Newman,G.

Chi-Rosso,and R.Lobb. 1990.ELFT:agenethatdirectstheexpressionofan

ELAM-I ligand.Cell. 63:1349-1356.

41.Tiemeyer, M., S. J.Swiedler,M.Ishihara,M. Moreland,H.

Schwein-gruber, P.Hirtzer,and B. K.Brandley. 1991. Carbohydrateligands for endothe-lial-leukocyteadhesionmolecule 1. Proc.Natl. Acad.Sci. USA. 88:1138-1142.

42.Yuen, C. T., A. M. Lawson, W.Chai,M.Larkin,M.S.Stoll, A. C. Stuart, F. X.Sullivan,T. J.Ahern,andT.Feizi. 1992. Novelsulfatedligands forthecell

adhesionmoleculeE-selectinrevealedbythe neoglycolipid technologyamong

0-linked oligosaccharideson an ovariancystadenoma glycoprotein.

Biochemis-try.31:9126-9131.

43.Larkin,M., T. J. Ahern, M. S.Stoll,M.Shaffer,D.Sako,J.O'Brien,C. T. Yuen,A.M. Lawson, R. A. Childs,K. M. Barone,etal. 1992. Spectrum of sialylatedandnonsialylated fuco-oligosaccharides bound bythe

endothelial-leu-kocyteadhesionmoleculeE-selectin.J. Biol. Chem. 267:13661-13668.

44.Larsen,E., T.Palabrica,S.Sajer,G. E.Gilbert,D. D.Wagner,B.C.Furie,

and B. Furie. 1990.PADGEM-dependent adhesion of plateletstomonocytes and

neutrophilsismediatedby alineage-specific carbohydrate,LNF III(CD 15). Cell. 63:467-474.

45.Corral,L., M. S.Singer,B.A.Macher,andS. D. Rosen. 1990.

Require-mentforsialic acid onneutrophilsin aGMP-140(PADGEM)mediated adhesive interaction with activated platelets. Biochem.Biophys.Res.Commun.

172:1349-1356.

46. Moore, K.L.,A.Varki,and R. P. McEver. 1991. GMP-140bindsto a

glycoproteinreceptoronhumanneutrophils:evidenceforalectin-like interac-tion. J.Cell Biol. 112:491-499.

47.Polley,M.J.,M. L.Phillips,E.Wayner,E.Nudelman,A.K.Singhal,S.I.

Hakomori,and J.C. Paulson. 1991.CD62andendothelialcell-leukocyte

adhe-sionmolecule I(ELAM- I)recognizethesamecarbohydrate ligand,sialyl-lewis

x. Proc.Natl. Acad.Sci. USA.88:6224-6228.

48. Zhou,Q.,K. L.Moore,D. F.Smith,A.Varki,R. McEver,and R. D.

Cummings. 1991. Theselectin GMP-140bindstosialylated, fucosylated

lactos-aminoglycansonbothmyeloidandnonmyeloidcells. J.Cell Biol. 115:557-564.

49.Foxall, C.,S. R.Watson,D.Dowbenko,C.Fennie,L.A.Lasky,M.Kiso,

A. Hasegawa, D. Asa, and B. K.Brandley. 1992. The three members ofthe

selectin receptorfamilyrecognizeacommoncarbohydrate epitope,thesialyl

lewis-xoligosaccharide. J.CellBiol. 117:895-902.

50.Berg,E. L.,J.Magnani, R.A.Warnock, M.K. Robinson,andE. C.

Butcher. 1992.Comparison of L-selectinandE-selectinligandspecificities:the

L-selectin canbind the E-selectin ligandssialyl-Lex andsialyl-Le'. Biochem. Biophys.Res.Commun. 184:1048-1055.

51.Berg,E.L.,M. K.Robinson, 0.Mansson,E.C.Butcher,and J.L.

Mag-nani. 1991.Acarbohydratedomaincommon tobothsialylLe'andsialylLeX is

recognizedby the endothelial cell leukocyte adhesion moleculeELAM- 1.J. Biol. Chem. 23:14869-14872.

52. Tyrrell, D., P. James, N. Rao, C.Foxall, S. Abbas,F. Dasgupta, M.

Nashed, A. Hasegawa, M. Kiso, D. Asa, et al. 1991. Structural requirements for the carbohydrateligand ofE-selectin. Proc. Natl.Acad. Sci. USA. 88:10372-10376.

53. Handa, K., E. D. Neudelman, M. R. Stroud, T. Shiozawa, and S. Hako-mori. 1991.Selectin GMP-140 (CD62; PADGEM) binds tosialosyl-Leaand sialosyl-Le', andsulfatedglycans modulate this binding. Biochem. Biophys. Res. Commun. 181:1223-1230.

54. Yednock, T. A., E. C. Butcher, L. M. Stoolman, and S. D. Rosen. 1987. Receptorsinvolvedin lymphocyte homing: relationship between a

carbohydrate-bindingreceptor and the MEL-14antigen.J.CellBiol. 104:725-731. 55.Skinner, M.P., D. J. Fournier, R. K. Andrews, J. J. Gorman, C. N.

Chesterman,and M. C. Berndt. 1989.Characterizationof human platelet GMP-140 as aheparin-bindingprotein. Biochem.Biophys.Res.Commun. 164:1373-1379.

56. Imai, Y., D. D. True, M. S. Singer, and S. D. Rosen. 1990. Direct

demon-strationofthelectin activityofgp9Omc',alymphocytehoming receptor.J.Cell

Biol.11:1225-1232.

57. Aruffo, A., W. Kolanus, G.Walz,P.Fredman, and B. Seed. 1991.CD62/ P-selectinrecognition of myeloid and tumor cell sulfatides. Cell. 67:35-44.

58. Aruffo, A., M. T. Dietsch, H. Wan, K. E.Hellstrom,and I.Hellstrom. 1992. Granule membraneprotein140(GMP140) binds to carcinomas and

carci-noma-derivedcell lines. Proc.Natl.Acad.Sci. USA. 89:2292-2296.

59.Pinto,B.M., and D. R. Bundle. 1983. Preparation of glycoconjugates for use asartificial antigens:asimplifiedprocedure.Carbohyd.Res. 124:313-318.

60.Lemieux,R.U.,D. R.Bundle, and B. A. Baker. 1975. Theproperties ofa

"synthetic"antigen relatedto thehuman blood-group lewis a. J. Am.Chem.Soc.

97:40764083.

61.English,D., and B. R. Anderson. 1974.Single-stepseparation ofred blood

cells, granulocytesand mononuclearphagocytes ondiscontinuous density

gra-dients of Ficoll-Hypaque. J. Immunol. Methods.5:249-22254.

62.Drickamer, K. 1988. Twodistinctclassesof carbohydrate-recognition domainsinanimal lectins.J.Biol. Chem.263:9557-9560.

63.Shibuya, N., I.J.Goldstein,W. F.Broekaert, M. Nsimba-Lubaki, B. Peeters, and W. J. Peumans. 1987. The elderberry(Sambucasnigra L.) bark

lectin recognizes the NeuSAc(a2-6)Gal/GalNAc sequence. J. Biol. Chem. 262:1596-1601.

64.Toogood,P.L., P. K.Galliker, G.D.Glick,and J. R. Knowles. 1991. Monovalentsialosidesthatbind tightlytoinfluenzaAvirus. J.Med. Chem. 34:3138-3140.

65.Symington, F.W., D. L.Hedges,andS. Hakomori. 1985.Glycolipid antigens ofhumanpolymorphonuclear neutrophilsand the inducibleHL-60

myeloid leukemia line.J.Immunol. 134:2498-2506.

66.Ohmori, K.,T.Yoneda, G. Ishihara,K.Shigeta,K.Hirashima,M.Kanai,

S.Itai,T.Sasaoki,S.Arii,H.Arita,and R.Kannagi. 1989.SialylSSEA-Iantigens

asacarbohydratemarkerofhuman naturalkiller cells andimmaturelymphoid

cells.Blood.74:255-261.

67. Munro, J. M., S. K. Lo, C. Corless, M. J. Robertson, N. C. Lee, R. L.

Barnhill,D.S.Weinberg,and M. P.Bevilacqua. 1992.Expressionof sialyl-Lewis X, anE-selectinligand,ininflammation, immuneprocesses, and lymphoid tis-sues.Am.J.Pathol. 141:1397-1408.

68.Rice,G.E.,and M. P. Bevilacqua. 1989.Aninducible endothelialcell

surface glycoprotein mediates melanoma adhesion. Science (Wash. DC). 246:1303-1306.

69.Takada, A.,K.Ohmori,N.Takahashi,K.Tsuyuoka,A.Yago,K.Zenita,

A.Hasegawa,and R.Kannagi.1991.Adhesion ofhumancancercellstovascular endothelium mediated by carbohydrate antigen, sialyl Lewis A. Biochem. Biophys.Res.Commun. 179:713-719.

70.Imai, Y.,M.S.Singer,C.Fennie,L.A.Lasky,andS.D. Rosen. 1991. Identification ofacarbohydrate-basedendothelialligandforalymphocyte

hom-ingreceptor. J.Cell Biol. 113:1213-1221.

71.Lasky, L. A., M. S.Singer,D.Dowbenko,Y. Imai,W. J.Henzel,C.

Grimley,C.Fennie,N.Gillett,S.R.Watson,and S. D. Rosen. 1992.An

endothe-lialligandfor L-selectin isanovelmucin-likemolecule.Cell. 69:927-938.

72.Moore,K.L.,N. L.Stults,S.Diaz,D. F.Smith,R. D.Cummings,A.

Varki, and R.P. McEver.1992.Identificationofaspecificglycoprotein ligandfor

P-selectin (CD62)onmyeloidcells. J.Cell Biol.118:445-456.

73. Varki, A. 1992. Diversityofthesialic acids.Glycobiology.2:25-40.

74.Larsen, G.R.,D.Sako,T.J.Ahern,M.Shaffer,J.Erban,S.A._Sajer,R.M.

Gibson,D. D.Wagner,B.C.Furie,andB.Furie. 1992.P-selectinandE-selectin: