Structure function relationships of interleukin 3 An analysis based on the function and binding characteristics of a series of interspecies chimera of gibbon and murine interleukin 3

Structure-function relationships of interleukin-3.

An analysis based on the function and binding

characteristics of a series of interspecies

chimera of gibbon and murine interleukin-3.

K Kaushansky, … , R E VanDyke, M K Pearce

J Clin Invest.

1992;

90(5)

:1879-1888.

https://doi.org/10.1172/JCI116065

.

IL-3 is a glycoprotein cytokine involved in the hematopoietic response to infectious,

immunologic, and inflammatory stimuli. In addition, clinical administration of recombinant

IL-3 augments recovery in states of natural and treatment-related marrow failure. IL-IL-3 acts by

binding to high affinity cell surface receptors present on hematopoietic cells. To determine

the site(s) at which IL-3 binds to it receptor, we analyzed a series of interspecies chimera of

the growth factor for species-specific receptor binding and biological activity. The results

suggest that IL-3 binds to its receptor and triggers a proliferative stimulus through two

noncontiguous helical domains located near the amino terminus and the carboxy terminus

of the molecule. To corroborate these findings, we have also mapped the binding epitopes

of 10 mAb of human or murine IL-3, and have defined four distinct epitopes. Two of these

epitopes comprise the amino-terminal receptor binding domain. A third epitope corresponds

to the carboxy-terminal receptor interactive domain, and the fourth epitope, apparently not

involved in the interaction of IL-3 and its receptor, lies between these sites. And on the basis

of sandwich immunoassays using pairs of these mAbs, the two receptor interactive regions

appear to reside in close juxtaposition in the tertiary structure of the molecule. These results

provide a correlation of the structure-function relationships of IL-3 that should prove useful in

evaluating the details […]

Research Article

Find the latest version:

http://jci.me/116065/pdf

Structure-Function Relationships of Interleukin-3

AnAnalysis Based on the Function and Binding Characteristics of a Series of

_Interspecies

Chimera of Gibbon and Murine Interleukin-3

KennethKaushansky,*Sara G. Shoemaker,*VirginiaC.Broudy,*NancyL. Lin,*Jeffrey V. Matous,*Edward M. Alderman,*

JaneD.

Aghajanian,*

Pamela J.Szklut,t Robert E.VanDyke,OMichael K. Pearce,' and John S. Abrams'

*DivisionofHematology, University of Washington, Seattle, Washington 98195;

*Genetics

Institute, Cambridge, Massachusetts 02140-2387; and

ODNAX

ResearchInstitute,PaloAlto, California 94304

Abstract

IL-3 is aglycoprotein cytokine involved in the hematopoietic response toinfectious, immunologic, and inflammatory stimuli. Inaddition, clinical administrationof recombinant IL-3 aug-ments recovery in states of natural and treatment-related marrow failure.IL-3actsby binding to highaffinitycell surface receptors present on hematopoietic cells. To determine the site(s)atwhichIL-3 bindstoitsreceptor,weanalyzedaseries ofinterspecies chimera ofthegrowth factor for species-specific receptor binding and biological activity. The results suggest that IL-3bindstoitsreceptor andtriggersaproliferative stimu-lusthrough two noncontiguous helical domains located near the amino terminusand the carboxy terminus of the molecule. To corroborate these findings,we have also mapped the binding epitopes of 10mAbof human or murine IL-3, and have defined four distinct epitopes. Two of these epitopes comprise the amino-terminal receptorbinding domain. A third epitope corre-sponds to the carboxy-terminal receptor interactive domain, and thefourthepitope, apparently not involved in the interac-tion ofIL-3and its receptor,lies between these sites. And on thebasisofsandwichimmunoassaysusing pairs of these mAbs, thetwo receptor interactive regions appear to reside in close juxtapositionin thetertiary structure of the molecule. These resultsprovideacorrelation ofthe structure-function relation-ships of IL-3that should proveusefulin evaluating the details ofIL-3-IL-3receptorinteractionand in the rational design of clinically usefulderivatives ofthisgrowthfactor.(J. Clin. In-vest. 1992.90:1879-1888.) Keywords:hematopoiesis* amphi-pathic-epitope*cytokines-multi-CSF

Introduction

IL-3 or

multi-colony-stimulating

factor

(CSF)'

is an acidic glycoprotein involved inthesurvivalandself-renewal of

hema-topoietic

stem cells, the

proliferation

and differentiation of

Addresscorrespondence andreprintrequeststoKennethKaushansky,

M.D.,Department ofMedicine,Division ofHematology

RM-l0,

Uni-versity of Washington, Seattle,WA98195.

Receivedfor publication24March1992andinrevisedform2 June 1992.

1.Abbreviations usedinthis paper:FICA,Freund'sincomplete

adju-vant; GM-CSF, granulocyte-macrophage colony-stimulating factor;

NHIA,N-hydroxysuccinimideesterof iodoacetic acid.

committed progenitorsandinthefunctionalactivation of

ma-ture leukocytes ( 1-3). In thisway, IL-3 is thought to be in-volved in the host response to noxious stimuli. In addition, recent studies have implicated IL-3 in the pathogenesis of asthma andhypersensitivity(3-5), andin murine and human leukemia(6-8).

IL-3 actsbybindingtohigh affinity cell surface receptors, whicharenormally present only on cells of hematopoietic lin-eage (8, 9). Recently, low affinity receptors for human and murineIL-3 were cloned(10, 1 1). On the basis ofconserved cysteine residues and afive amino acidWSXWS region, the IL-3receptorwasthoughttobe a memberofagrowingfamily ofpolypeptides that includesthe receptorsfor IL-2,IL-4, IL-5, IL-6, IL-7, granulocyte-macrophage (GM)-CSF, granulocyte CSF, and

erythropoietin

(12). Based on this homology and receptorcross-linking data, itwasproposedthat the IL-3 recep-tor,likethe IL-2andIL-6 receptors,is composed ofatleasttwo

subunits (12, 13). This

supposition

wasrecently confirmed

by

Kitamuraetal.,whodemonstrated thata

non-ligand-binding

subunit ofthe humanGM-CSFreceptor

(

14)alsoinduces

high

affinityIL-3bindingwhenexpressed withthe low

affinity

IL-3 receptor( 11 ). Thisfindinghas alsoprovided

potential

mecha-nismstoexplain the reciprocal competition displayed by IL-3 and GM-CSF forbindingtocells thatbear receptorsforbothof these

cytokines(13,

15, 16).

Despite its pivotalrole inhematopoiesis, little isknownof the structural features ofIL-3 thatareresponsible forreceptor bindingandbiological

activity.

Initial studies usingtruncated

synthetic

polypeptides of murine IL-3 suggestedthatthefirst disulfide bond andtheamino-terminal half ofthemoleculeare

responsible

for biological activity (17, 18). Recently, two groups havereportedonthebinding sites of neutralizingmAbs ofhumanIL-3 (19, 20). Together, 13mAbsweremappedto

residuesbetweenGln29 and Glu50, helpingtolocalizethesite of thisamino-terminal domain of

biological activity.

Onestudy hasreportedon carboxy-terminal truncation mutantsof

hu-man IL-3 (19). Although the 10

carboxy-terminal

residues weredispensable,theactivities ofmore severetruncation mu-tants werereducedthreetofour orders of

magnitude,

making

it likelythatresidues

immediately

upstreamofthis

position

are alsoinvolved in receptor interaction.

A numberof additional strategieshave been used tomap the

receptor-interactive

residues of

biologically

active

polypep-tides.Mostdependontheactivitiesof

internally

deleted

poly-peptides.However, these methodscannot

distinguish

between

loss ofactivity caused byalterations in secondaryor

tertiary

structureand that causedbylossofareceptor interactive site. Recently,we

developed

astrategyofsubstitution

mutagenesis

basedonthestructural

similarity

but

biological

distinctiveness ofthe human andmurine analogsofa

hematopoietic

growth

Structure-FunctionRelationships ofInterleukin-3 1879

J.Clin.Invest.

©The AmericanSociety for Clinical Investigation,Inc.

0021-9738/92/11/1879/10 $2.00

factor, GM-CSF (21 ). By generating cross-species chimera and testing them for species specific function, receptor-interactive residues were identified. Human (or gibbon) and murine IL-3

arealsospecies-specific: They donotstimulate the growth of progenitor cells or the function of mature cells from the alter-natespecies, and they donot cross-react atthe receptor level. Monoclonal antibodiestohumanormurineIL-3 also react in a

species specific fashion. To further explorethe structure-func-tion relationships of IL-3, we used this species specificity to mapthe sites of receptorbindingandbiological activityof pri-mate and murine IL-3. Using this approach, two nonlinear sites of IL-3-IL-3 receptor interactionwereidentified. The first

site appears to be sufficient for receptor binding, but inade-quate to transmit a proliferative stimulus. The second site is necessary for biological activity. Wealsoshowthat these two

sites correlate with thebinding epitopesof severalneutralizing mAbs ofIL-3, but not with nonneutralizing mAb epitopes. Together, these findings supportthehypothesisthat the recep-torbinding domain ofIL-3 is composed of at least two

ele-ments from distinct regionsof the molecule andbeginto pro-vide anunderstanding of theeventsthatlead to aproliferative stimulusinhematopoiesis.

Methods

Generation ofchimeric IL-3 cDNA expression vectors. Full-length

cDNA clones forgibbon IL-3 (Dr. Steve Clark,Genetics Institute, Cambridge, MA)and murine IL-3(Dr.FrankLee,DNAXInstitute,

PaloAlto,CA)weresubjectedtosite-directed mutagenesis,as

previ-ously described(21),tointroduce useful restriction sitesor todirectly

alterthecodingsequences(Fig. 1). RecombinanthybridcDNA

ex-pressionvectors wereproducedbyligatingrestrictionfragmentsof the mutantgibbonormurine IL-3 cDNA into the mammalian cell expres-sionvectorpDX(22) bystandardtechniques.Eachmutation and

liga-tionwasverified bysingle-strandedorplasmidDNAsequencingusing thedideoxynucleotidechain termination method(23).

Expressionof chimericIL-3molecules. BHK cells wereobtained from American Type Culture Collection(Rockville,MD) and main-tained in Dulbecco'sMEMsupplementedwith 5% FCS and antibi-otics. Cellsweretransfectedwith thechimeric IL-3expressionvectors andadihydrofolatereductase(DHFR)expressionvector(24) by the calcium phosphate method when 30% confluent, as previously de-scribed(2 1).After 3 d, the cellsweredivided and cultured in 250nM methotrexate. Individual colonies were selected, expanded, and

as-sayedforspecies-specificIL-3 mRNAexpressionby slot blotanalysis using full-lengthcDNAclonesforgibbonormurine IL-3 labeled by random priming (Multiprime DNA Labeling System; Amersham

Corp., Arlington Heights, IL). Cell lines expressing high levels of

IL-3-specific RNAwerethen selected for analysis. To generate

bio-synthetically"S-labeled IL-3molecules, the celllines expressing chi-mericproteinsweregrownin DMEMdevoid of cysteine and methio-nine, and supplemented with2% dialyzed FCS, antibiotics, and 50

,gCi/ml "5S-Cys

+Met(Express;NewEnglandNuclear, Boston, MA). The medium was then harvested after 24-36 h.

Quantitative Western blotanalysis.Thelevelofspecificproteinwas determined by quantitativeWesternblotting (21, 22).Aliquots ofcon-ditioned culture medium from cell lines expressing chimeric IL-3 pro-teinsweredenatured, size fractionated through 12% polyacrylamide

gels, transferredtonitrocellulose, and probed with a 1:50 dilution of a ratantiserum raised against a synthetic peptide composed of the 16 amino-terminal residues of human IL-3 (peptide 4, reference 25). Sincethis antiserum reacts equally well with IL-3 derived from bacte-rialormammalian cell expression systems, human IL-3 purified from recombinantEscherichiacoliservedas aquantitative control.An125I goatanti-rat

_Ig

antiserumwasusedtodetectimmune

_complexes

on

the blot. After autoradiography, blotswere analyzed by phosphor-imaging (Molecular Dynamics, Sunnyvale, CA) to quantitate the amountof'25I-detecting reagent present.

Hematopoietic progenitor cell assays. Human marrow cells were obtained fromconsenting healthy adult volunteers, underaprotocol approved by the University of Washington (Seattle, WA). Nonadher-enthumanmarrowmononuclear cellswereusedtoassay chimeric IL-3 molecules for their ability to stimulate the growth of granulocyte-macrophagecolony-formingcells, aspreviouslydescribed (21). The murineIL-3-dependent cell line FDCP1wasobtained from American Type Cuture Collection andwasmaintained in DMEM with 10% FCS and 50 U/ml murine GM-CSF. The FDCP1 cellswereusedtoassess theability of IL-3 moleculestostimulate murine colony growth, and for murine IL-3 receptorbindingassays,asdescribed below. The

col-ony-forming capacityofFDCP1 cells has been showntocorrelate pre-cisely with the ability of CSFtostimulate progenitor cell growth from murine marrow cells(26). These resultswere alsoconfirmed using murinemarrowcells,aspreviouslydescribed (21).

Receptor bindingassays. The human erythroleukemia cell line OCIMI (27)wasmaintainedin RPMI 1640 medium supplemented with 10%FCS andpenicillin/streptomycin. 250 pg of'251-humanIL-3

(Amersham Corp.)wasmixed with 3X I0O OCIMI cells in 100

,d

of

bindingbuffer(RPMI1640 mediumsupplementedwith 50mMHepes [pH 7.4], 1%BSA, 10 Mg/ml cytochalasinB,and0.1% sodium azide). The mixture also contained human IL-3(100 ng),murine IL-3 (25

ng),orculture medium from cell linesexpressingthechimeric proteins

(10ng-19 ug).Aftera1-hincubationat37°C,cell-bound '25I-human IL-3wasseparatedfrom unbound '25I-IL-3bycentrifugingthe cells

through phthalateoil(28).For themurine IL-3 receptor assay, 250 pg of '25I-murineIL-3(provided byDr.Gerry Krystal,British Columbia Cancer ResearchCenter,Vancouver, BC, Canada)wasmixed with 7

X I05 FDCP1 cells in 100 ,ul ofbindingbuffer(PBSsupplementedwith 1%BSA,0.1%gelatin, and 0.02% sodium azide[29]).The mixture alsocontained unlabeled humanormurine IL-3orthe gibbon-murine IL-3hybridproteins,asdescribed above. Aftera4-hincubationat4°C

(29),the cellswerecentrifuged through phthalateoil.Thebindingdata

areexpressedasthe percentage ofcompetitioninthe presence of the IL-3hybrid proteins.Themaximalcompetitionachievable witha 100-foldexcessof unlabeled human IL-3 in the OCIMIassay(75%),ora 100-foldexcessof unlabeled murine IL-3 in the FDCP1 assay(79%)

weredefinedas100%competition. Triplicatemeasurementswere per-formed in eachexperiment.

Quantitativereceptorbinding experimentswerealsoperformed

us-ing metabolicallylabeled 35S-IL-3.Truncated cDNA that encodeda Metstartcodonfollowedbythematureaminoacidsequence ofgibbon

cDNAwassubclonedintopBluescriptand RNAwastranscribed in vitrousing T7RNApolymerase. 2Mgof RNAwastranslatedinto pro-tein in reticulocyte lysates (Promega Biotech, Madison,WI) in the presence of[35S]Met. Theradioactivelypure IL-3 was separated from

unincorporatedlabelby gelfiltration(Sephadex G-25),andthe

radio-logic specificactivityof the labeled IL-3proteinwasdetermined by self

displacement analysis(30). Theamountof3"S-IL-3 produced in a

typical200Mulreticulocyte lysatereactionwas20 ng. The35S-IL-3was then used inareceptorbindingassay. OCIM 1cells wereincubated with

varying quantitiesof3"S-gibbonIL-3(7-330 pM) with or without a

1,000-foldexcessof unlabeled IL-3 in binding buffer for 1 h at 37°C,

asdescribed above.Allreceptorbindingdata were analyzed using the LIGAND program(31).

GenerationofratantipeptideIL-3(humanand mouse) monoclonal antibodies.Asyntheticoligopeptide[(C) PNLEAFNRAVKSLQNA-SAI] correspondingtoresidues 56-74 ofthe human IL-3 molecule, was obtainedcommercially(Applied BiosystemsInc., FosterCity,CA). The terminal thiolwasusedtocovalently couple the peptide to sperm whalemyoglobin (SigmaChemical Co., St. Louis, MO), using the heterobifunctionalcrosslinkingreagentN-hydroxysuccinimideesterof iodoacetic acid(NHIA)(32). Myoglobin sulfhydrylgroupswerefirst blockedby reacting 10mlof 5 mg/ml myoglobinin0.1 Msodium bicarbonate with 1 mlof 0.12Miodoacetamide(SigmaChemicalCo.)

for 1 h at room temperature followed by overnight dialysis against 0.1 Mbicarbonate.

A 5 mg/mlsolution of peptide was prepared in 2 ml of0.1 M DTT, 50 mMTris, 2.5mMEDTA, pH 8, andincubated overnight at 370C. The reducedpeptide was applied to a 1.5 X26.5 cmgel-filtration col-umn(GF05; LKB Instruments, Inc., Gaithersburg, MD), and was eluted with 15 mM aceticacid, 5mM

fl-mercaptoethanol.

Material

elutingin thevoid volumewaspooled andlyophilized. The carboxyam-idated myoglobin recoveredfrom dialysis was iodoacetylated by reac-tionwith 0.38 mlof 5 mg/ml NHIA in tetrahydrofuran for 30 min at room temperaturewith vigorous stirring, followed by dialysis against PBS. The lyophilized peptide was resuspended in 5 ml of 10 mM

Na2B4O7,0.29 M NaCI, 1 mMEDTA, pH 8.5 (which had been N2-purged for 30min,followed byaddition of 178 mg/ml ascorbate), and mixed with an equal volume of iodoacetylated myoglobin recovered from dialysis. Theconjugation reaction was carried out at room temper-atureovernight.

ALewisrat wasimmunizedintraperitoneally with 100_Agof

pep-tide-myoglobinconjugatein CFA. Theanimalwasboosted three times with the same peptide conjugate in Freund's incomplete adjuvant (FICA), and splenocyteswerefused with P3X63-AG8.653 myeloma cells(AmericanType CultureCollection) using50% polyethylene gly-col(33). This fusion resulted in the 8A5 mAb.

The 10F4 ratanti-mouse IL-3 peptide antibody was prepared in a similarfashionusinganoligopeptide[(C) VESQGEVDPEDRYVI]-ovalbuminconjugate comprisingresidues 56-70 ofmature mouse IL-3 (theanalogous residuestothose usedfor human IL-3). This antibody has beenpreviously shown to work in Western blotting, as well as in an indirect ELISA of mouse IL-3 (34, 35).

Generationofratanti-IL-3(human andmouse)monoclonal

anti-bodies. Twoseries of fusions with differentimmunogen forms were carriedouttoobtain the neutralizing rat anti-human IL-3 antibodies used inthis study.Inthefirst fusion, male Lewis rats were immunized with 50 ug i.p. yeast expressed-recombinant IL-3 in CFA. Splenocyte fusion with theP3X63-AG8.653 myeloma was carried out4dafter the fourth boost. Anti-IL-3antibody secreting hybridomawereselected on thebasis ofscreening byindirect ELISAonIL-3-coated polyvinylchlo-ridemicrotiterplates,aswellasneutralization ofIL-3-induced prolifer-ationof the KG-I (AmericanType CultureCollection) myelomono-cyte cell line assessedcolorimetricallywith [3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazoliumbromide] by the method of Mosmann

(36).This fusion produced the IF9 and the 6G8 antibodies. A subsequent fusionwascarried outusing glutaraldehyde cross-linked(37) yeast-expressed IL-3toproduce antibodies recognizing dif-ferentepitopes,aswellas morepotentneutralizing activity. Yeast-ex-pressed IL-3wasconcentratedto0.2mMby Centricon 10 ultrafiltra-tion inPBS,0.01% Tween 20(SigmaChemical Co.). Glutaraldehyde (Sigma ChemicalCo.)wasaddedto afinal concentration of 2mM ( 10:1 Mglutaraldehyde/cytokine).Thereactionwasincubated forIh at roomtemperature, thenquenchedwithL-lysine (finalconcentration 0.1 M) foranadditional houratroomtemperature and storedat4°C.

MaleLewisrats wereimmunized intraperitoneallywith cross-linked IL-3 inCFA. Animalswereboosted twice with cross-linked IL-3 in FICA, and then with IL-3 in PBS beforesplenocytefusion. Anti-IL-3

antibody-secretinghybridomawereselected basedoninhibition of

IL-3-dependentproliferation of themultifactor TF-l erythroleukemic

cell(38). This fusionproducedthe3GI1antibody.

Thedevelopment and characterization of theneutralizingrat anti-mouseIL-3 monoclonal antibodies8F8, 19B3,and43D I have been previously described (33).ThesemAbsexclusively recognizehuman (andgibbon)ormurineIL-3,includingthat made inE.coliand

Sac-charomycescerevisiaeexpressionsystems.

Generationofmouseanti-human IL-3monoclonal antibodies. Five

BALB/CJ-Baileymice (Jackson Laboratory, BarHarbor,ME)were

immunizedwith recombinanthuman IL-3(rhIL-3)by4xI ml

injec-tions (4-5sites/injection)at2-wkintervals,as50%(vol/vol) emul-sionsof25, 12.5, 6.25,and 6.25 ug, respectively, in CFA(forfirst

injection) orFRCA (Gibco Laboratories, Grand Island, NY). Mice

werethen rested for4 wk andchallenged byinjectionof 50 ggi.p. adjuvant-freerhIL3oneach of 4 d before fusion. Immunesplenocytes

fromoneof the immunized micewerefusedtoP3X63AG8.653

mu-rine myelomacells(1:4 ratio) by standardpolyethylene

glycol-me-diated cell fusionprocedures,andplatedin 96-well tissue cultureplates

at adensityof 5X I0O cells/wellin standardhypoxanthine

aminopro-teinthymidineselection medium. Resultingsubcultureswerefed ad lib, and conditioned mediumsampleswerecollectedonday14 (post-fusion)forprimary screening by RIA(18 positive/672 tested)and ELISA(33positive/672 tested). 12subcultureswereselected for fur-therstudy,each of whichwassubjectedtothree rounds ofsubcloning

bylimitingdilution(0.4 cells/well),resultinginthe establishment of 12hybridomacelllines,includingmAbs 4.4.7 and 2.6.23.

Binding characteristics ofanti-IL-3 mAbs.Todetermine the capac-ity of the anti-human IL-3 mAbstobindtodenaturedprotein, West-ern blotswereprepared usingrecombinant human IL-3producedinE. coli. Aftertransfer,the blotsweredenatured and reducedwith 20% methanol and 30 mM

_{fl-mercaptoethanol}

at80'C for 30 min and then

iodoacetylatedwith 30 mM iodoacetic acid in the darkat20'C.This stepwasincludedtomakecertain that the transferredproteinswere

fullydenaturedbeforedetection with each mAb.

Immunoprecipitation of metabolically labeledIL-3 proteins. 1 ml of35-Met/Cys-labeled conditionedculturemediumwasmixedwith 10Mugofaffinity-purifiedantibody or 100 Mlof monoclonal antibody ascitesfluid(8A5andI1F4)andincubated for 4-16 h at 4°C. For the ratmAbs,20Mugofanti-rat Ig (Cappel Laboratories, Cochranville, PA) wasaddedfor 3 h at 4°C. For the murinemAb,rabbit anti-murine Ig wasused, and bothsetsof immune complexes were precipitated with

StaphA(Pansorbin;Calbiochem Corp., La Jolla, CA) (39).

Immuno-precipitatesweresizefractionated on 12%NaDodSO4polyacrylamide

gels, the gels were fixed, soaked in Amplify solution (Amersham

Corp.),dried,andexposedtofilmfor 2-20 d.

Identifcation

of antibodypairs recognizing spatially distinct epi-topesonhuman IL-3.Animmunoenzymetricassay format was used to

identify antibodypairsrecognizing spatially distinct epitopes on the IL-3 molecule. Polyvinylchloride"U" bottom microtiterplates were coated with 10Mg/mlin PBSfor 2 hat37°Cwithvariousrat monoclo-nalanti-IL-3 antibodies. Thesewereproduced invitro in HB102 me-dium andpurified by66% saturated ammonium sulfateprecipitation

and AcA44 gel filtration chromatography, essentially as described

(40).Plateswerewashed (three cycles) between all steps with saline-Tween20. Serial dilutions ofIL-3 (a

100-Ml

volinRPMI, 15%fetal bovineserumcontaining40 ng-5 pg/ml)wereincubated for 2 h at room temperature. Nitro-iodophenyl (NIP)-conjugated anti-IL-3 mAbswereprepared by reactingImg/mlIgG in PBS with 10MlNHIA (5 mg/ml inDMSO) forI hatroomtemperature, followedby

over-nightdialysis againstPBS. NIP-labeledantibodies (100 Ml of1

Mg/ml

in

PBS/BSA/Tween 20)wereincubated in all wells.A ratIgGIanti-NIP (J4) monoclonalantibody conjugatedtohorseradishperoxidasewas usedtodetectNIP-anti-IL-3bindingtosites other than thoseoccupied by thecoating antibody. 2,2'-azino-bis

(2-ethylbenzthiazoline-6-sul-fonicacid)wasusedaschromagen.

Computeranalysis.AsoftwarepackageprovidedbyIntelliGenetics

(Mountain View, CA)calledPC/Genewas run on anIBMPC/2

com-puter. ThesecondarystructurealgorithmsofGarnier(41), Parry(42),

Chow and Fasman(43),and Gascuel(44)wereusedto_predict a-heli-cal structure andthe program Helwheel wasused todetermine the presence ofamphipatichelixes.

Results

ChimericIL-3

protein-producing

celllines. After cotransfec-tion of

expression

vectors

encoding

DHFR andthechimeric

andmutant

proteins

illustratedin

Fig.

1,

cell lines resistantto

methotrexate were

analyzed

by

Northern blot

analysis

using

full-length

cDNA

probes

for

gibbon

IL-3 and murine IL-3.Cell

lines that

produced

RNA that

_hybridized

tothe

appropriate

ATO B5_36i, D2E23

glL3 gm2

Xho EcoRI (CIa1)

E119D126 V4N16

gmg9 gm3 _L

1gm1) f(Hpa1)

F107R 104 ATG NUC

gm l - mIL3V --- --_

(Mro1} Xho 1 Eco RI Q45NOW0V120OKj2D2D2

gm6 2207 r - ---

---;PpuMiI (mIL3)

LiGibbon 3Mouse '

Figure 1. Gibbon-murinechimera. Themapsfor each of thechimera

areshown, alongwith the methods of constructionfrom theparental

vectorspDgIL3 and pDmIL3. Sequences derivedfromgIL3are

rep-resented byopenbars, whilesequencesspecificfor murine IL-3are

shownin hatchedbars. Restriction sites introducedby mutagenesis

areshowninparenthesesbelow eachmap,andregionsmutated

di-rectlyareunderlinedwithasolid baralongwith the constructfrom

which theywerederived.

cDNA probeswere testedforspecific protein production by Western blotting (Fig. 2). As theprimaryantiserum was

di-rectedagainstaregionsharedbyall of thegibbon-murineIL-3

chimera, the amount ofan '25I-labeled detecting antibody boundtoeachprotein (determined by phosphorimaging analy-sis) providedaquantitativemeasureofthe IL-3protein present in each supernatant (purifiedrecombinant human IL-3

pro-videdaquantitative standard).The variable

M,

of the chimera

is characteristicofheterogenouslevels ofglycosylation,a

pat-tern noted ina number ofprevious studies ofhematopoietic growth factors (21, 22, 35).

43-o

18-o

E

E

E

E

E

'~

j

0 = 0 0 0

_E

M

Figure2.Chimeric cell linesandproteins.BHKcells cotransfected with chimeric IL-3 cDNAexpressionvectorsandaDHFRRvector

(24)wereselected in methotrexate andanalyzed by quantitative Westernblotting. Cell lines that expressed high RNA levels (by slot-blotanalysis,notshown)weregrowntoconfluence andconditioned culture mediumcollected. 60-,ulaliquots of conditionedculture

me-dium from eachcellline and 100ngof E.colihuman IL-3were size-fractionatedon a12%polyacrylamide gel, transferredto nitrocellu-lose, and probed witharatantiserumdirectedagainst the

amino-ter-minal 16residues of human IL-3 (antipeptide4,reference25).After

incubation witha '25I-labeledgoatanti-ratIgantiserum,

autoradiog-raphy and phosphorimagingwasperformed.In this experiment,60

_,1

of BHKgIL3 conditioned medium contained500ngprotein;BHK

gm9, 3,000ngprotein; BHKgm 1,2,000ngprotein;BHKgm6, 400

ngprotein;gm2, 430ngprotein; andgm3, 1,030ngprotein.

The biologic activity ofchimeric IL-3 proteins. Serial dilu-tions of the chimeric IL-3-conditioned culture mediawere

as-sayed for biologic activity in colony-formingassaysusing

hu-man or murine marrow cells (Table I). Together with the

quantitative analysis from the Western blot experiments, the specific activities of each chimeric growth factor (relative to

native human or murine IL-3) were determined. Beginning

from the amino terminus, chimericproteingm3 wasnearly fully active using murineassays.As expected from the species-specificity of gibbon and murine IL-3,gm3 failedtostimulate humanprogenitorstoformhematopoietic colonies. Because of thesequence homology between gibbon and murine IL-3 in

residues 16-22, chimeric proteingm2representsafive-amino

acid substitution mutantofgm 3 (L15to V14and SIVK22to

NMID21).

Thepresenceof thesechanges alone resulted in the

nearly complete inactivation ofchimericgm 2 in themurine marrowcellproliferationassays(Table I).

To further investigate the importance ofresidues in this region,anadditional double substitutionmutanttermed2207

wasanalyzed. Asshown in Fig. 1,the 2207mutantcontainsan

Ile-Asp dipeptide atpositions 21 and 22 inplaceof the wild typeVal-Lys dipeptide. Quantitationof mutant 2207 and the

mIL3 producing cell lines was accomplished using mAb

43D 11,an antibody that recognizesanepitopedistinct from

the site ofthe 2207 mutation.Serial dilutionanalysisof mutant

2207 (Table I) demonstrated that the doublemutation reduces

thespecific activityofmurineIL-330-fold.

The next two chimera, gm 6 and gm 1, failed to signifi-cantly stimulate colony growth by either human or murine

cells, despite thepresenceofadequateamounts

ofimmunoreac-TableI. Biological Activity of Chimeric IL-3 Proteins

Activity Relative specific activity*

Mutant Human Murine Human Murine

U/mi

gIL3 6.6x 104 0 (1) 0

gm9 6.0x 104 0 1.4x 10-1 0

gml 1.2x102 0 5.0x10-4 0

gm6 1.4x 10' 0 3.0x 10-4 0

gm2 0 2.2x10' 0 7.0X10-4

gm3 0 3.1 104 0 1.7x10-'

mIL3 0 3.8x104 0 (1)

2207 0 1.0X 103 0 3.3X 10-2

Eachprotein wassimultaneously analyzedby human and murine

marrowcolony-formingassayand byquantitativeWesternblotting. The activity resultsrepresentthemeanofatleastfourexperiments.

The data derived from Fig. 2wasusedtocalculate theamountof

gIL3 and the relativeamountsof the chimera.Thespecific activity of gIL3 wascalculatedtobe8 X 106 U/mg.Therelativeamountsof

mIL3 andmutant2207wasdeterminedby phosporimaging of the

blot shown inFig. 4 C usingaprimary antibody(43D1 1),which recognizesanepitope distinct fromthesite ofthe 2207,gm3, and gm2mutations.The colony-stimulating activitypresentinthe

con-ditionedmediumfromBHK-derivedcelllines isexpressedin U/ml

(50U=amountnecessarytostimulate half-maximal colony

forma-tionfrom105 low-density nonadherent humanmarrowcellsor7.5

X 104whole murinemarrowcells. *Comparedtonativehuman

ormurine IL-3.

1882 Kaushanskyetal.

.W.W.4W

tive protein (Fig. 2).In contrast, gm 9, whichdiffers fromgm 1 at onlysixpositions (between residues

F107

through

_EII9),

was nearlyfullyactive using human marrow progenitor cells (Table I).Thesix-aminoacid differences between gm 9(ontheleft) and gm 1 (on the right) are shown in bold type: . . .

F107RRKLKFYLKTLE19DLE

. . . to . . .

FIOIRKKLRFY

MVHLN113DLE.

Thus,tworegions,between human IL-3 resi-dues

Vall4

andAsp21 and between

PheI07

andGlu119,were nec-essary for theproliferativeactivity of the growth factor.

Receptor binding activity ofthe chimeric proteins. Receptor bindingassays wereperformed usingthe human erythroleuke-miacellline OCIMl and the murinefactor-dependentcell line FDCP1. Thesecellsdisplayhigh affinity receptors for human andmurineIL-3, respectively. As shown in Table II, chimeric proteins that containatleasttheamino-terminal45residuesof gibbonIL-3 were able to compete with human IL-3 for binding tothe IL-3receptor onOCIM1 cells.Chimericmolecules that contain atleast the 124carboxy-terminal residues ofthe mu-rine protein (residues 16-140) could compete for binding of 1231-murine IL-3toFDCP1 cells. These data suggest that some ofthe sameresidues identified by the colony-forming assays were required for receptor binding. In addition, these data implythat molecules (e.g., gm 6) that fail to stimulate a bio-logicresponse are capable, albeit with reduced affinity, ofbind-ingtotheIL-3 receptor. To be certain that these findings were not caused by differences between the human and gibbon growth factors, weadapted an in vitro translation system to produce radioactivelypure gibbon IL-3. Metabolically

"S-la-beledproteinswerethentestedin thereceptorbindingassay. Using

3S-gibbon

IL-3, OCIM1 cells werefoundtodisplay 470 receptors percell witha Kdof52 pM (Fig. 3). These results for receptor binding avidity compare favorably with that using

1251-human

IL-3 (170receptors per cell,Kd63pM) and suggest thatthe

gibbon

andhuman moleculesinteract identicallywith the human IL-3 receptor. Thediscrepancy between receptor numberidentifiedbyhuman and gibbon IL-3 likely relates to

TableII.IL-3Receptor Competition Assay

Percentspecific competition Protein/mutant Human Murine

hIL3 (100) 0

gIL3 80 0

gm 9 33 ND

gm1 44 ND

gm6 51 11

gm6* 81 ND

gm2 0 17

gm3 17 98

mIL3 ND (100)

Concentrated culture media conditionedbyBHK-derived cell lines which secrete thegibbon-murine chimeric IL-3 proteinsweretested in anOCIMIorFDCP1 IL-3 receptorcompetitionassay. The results areexpressedasthe percentage ofcompetitioninduced byeach chimericprotein.Each supernatantwasadjustedtocontain 3 ,ug of

specific proteinasdetermined byquantitativeWesternblotting(Fig.

4). *In oneexperiment,19,gof gm 6proteinwastested. The data represent themeanresults oftwoexperiments performedintriplicate.

ND, notdetermined.

A

ha

43_-.

25-_n

18-*-B

WL f10 ftl f12 fl3 114 115 0

w

w_Ul

Ul

0

z

0 m

1 2 3

L-3 BOUND (pM)

Figure 3. IL-3receptor binding assay. (A)"S-gibbon IL-3,labeled

with

35S]

Met in a reticulocyte lysate (WL=wholelysate), was sep-aratedfrom unincorporated[5S ]Met by gelfiltration. Fractions I 1-14werepooled for use in receptor binding analysis. (B) Scatchard

analysisof"5S-labeled gibbon IL-3bindingto OCIMI cells. OCIMI cells(0.4 x 106)wereincubated with varying quantities of"5S-gibbon

IL-3. The data represent the average of duplicate determinations, and were analyzed using the LIGAND program (31 ).

theuse ofdifferent preparations of

OCIMl

cells for the two

experiments.

Mappingthebindingepitopes

ofanti-IL-3

monoclonal

anti-bodies. A series of five distinct anti-human IL-3 and three distinct anti-murineIL-3 mAbswereraised againstthe recom-binantproteins. Inaddition, antipeptidemAbsraisedagainst residues 56-74ofhuman IL-3 and against residues 56-70 of murine IL-3 were alsoavailable for analysis. The capacity of thesemAbstorecognizedenatured human ormurineIL-3 in a Westernblot format isshownin Fig.4. To be certain that the IL-3proteinswerecompletelydenatured on thenitrocellulose, the blotsweretreated withNaDodSO4 and

f,-mercaptoethanol

and were acetylated after transfer.The strongsignal foreach mAb stronglysuggeststhateachantibody recognizesadiscrete, short lineararrayofIL-3-specific amino acids.

Next,several of themAbs were used to detect thechimeric IL-3

proteins

byWesternblot

analysis.

Theresults from two informative

experiments

areshown

(Fig.

4,Band

C).

Basedon

the

differential

capacity

of mAb4.4.7to

recognize

chimeragm 2andgm3, the binding

epitope

of this

antibody

mustinclude residues between

Val14

and

Asp21.

Since this mAb isnot

con-formation

dependent,

it is

likely

that this site encompasses most orallof the4.4.7

binding

site.

Furthermore,

asthismAb (andan Fab

fragment

derived from

it,

datanot

shown)

blocks

binding

of human IL-3toitsreceptoron

OCIM1

cells

(Table

III),

this site

likely comprises

atleast partofthe receptor bind-ing

domain

ofthe

growth

factor.Fromsimilar

considerations,

thedifferential reaction oftheanti-murine IL-3mAb43D11 with chimeragm 2 and gm 6 suggeststhat this

antibody

binds tomurine IL-3 betweenresidues

_{Val2l (corresponding}

to hu-man IL-3

Ile20)

and

_Pro4o

(corresponding

to human IL-3

Gln45).

Furthermore,as43D11neutralizesthe

biological

activ-ity of murine IL-3

(33),

this

region

is also involved in the interactionbetweenIL-3anditsreceptor.

To expand upon the Western

blotting

results and tobe certainthat the mAbs used inthis

study recognize

IL-3 under physiologic conditions,wenext

prepared "S-labeled

IL-3 pro-teins and

performed immunoprecipitation analyses. Fig.

5

shows the resultsobtained

using

severaloftheanti-humanand anti-murine IL-3 mAbs. Basedon their

precipitation

of

chi-meragm 6 butfailureto

recognize

gm

2,

mAbs

6G8,

lF9,

and

B

43

--25-O

18-0

@0 0 P*.

CV) _.4

C

43-0

25-_

18-I

c0 M '-_ C 0c

i~ E

E

E

E

E

0 E0

0 I,* 0

E

E

E

E

11

0 0 0

Figure4. Western blotanalyses. Recombinant human IL-3producedin E. coli(100 ng) (A),orconditioned medium(60,ul)from cells

ex-pressinggibbonIL-3(gIL3),murine IL-3(mIL3),orthe IL-3 chimera(BandC)weredenatured inNaDodSO4and size-fractionatedthrough

12% polyacrylamidegels. Afterelectrotransfer tonitrocellulose,the blotswereprobedwith the five anti-human IL-3 mAbs(A),mAb4.4.7 (B),and withmAb43D11 (C).Molecularmassmarkersareshown(kD)ontheleft.

2.6.23 were mapped to residues between

Asp21

and Gin45. Thus, theseantibodies recognizean

epitope

quite similartothe anti-murine IL-3 mAb 43D1 1. Since

6G8,

IF9, and 2.6.23 also block

binding

of human IL-3toOCIM1 cells (Table III), these results also pointtothe

importance

of this

region

inthe interaction between human IL-3 and itsreceptor.

Likeanti-human mAb 4.4.7, anti-murine mAbs8F8 and 19B3 recognized

residues

between

Leu15

(human

VaI14)

and Lys22 (human Asp21).BothofthesemAbs neutralizedmurine IL-3(33),again

confirming

the

importance

of this regionfor receptorbindingandagain, demonstrating the concordanceof the structural

organization

ofhuman andmurine IL-3.

Athird,central

epitope

wasdefined by the antipeptide anti-bodies 8A5 and I0F4.These mAbs

recognize

humanresidues 56-74 andmurine residues 56-70, and are bound to native IL-3,asjudged by their abilityto

immunoprecipitate

the

mole-TableIII.Receptor

Blocking

Capacity

of

Anti-IL-3Antibodies

Addition Competition

Saline 0

50 x hIL3 (100)

4.4.7 104

2.6.23 132

3G11 187

1F9 92

6G8 125

2.1.3 0

'25I-labeled

human IL-3 (hIL3) was added to OCIM I cells in the presenceof each of the anti-human IL-3 mAb. The percentage of

competitioncompared to a 50-fold excess of unlabeled human IL-3 is reported. Inthis experiment, 55% of the bound cpm could be competed with the unlabeled IL-3, and the experiment has been re-peated twicewith similar results. mAb 2.1.3 was raised against human GM-CSFandused as a negative control.

culefromphysiologicsolution(Fig.5). However, they did not neutralize the biological activity ofhuman or murine IL-3

(35). Thus,these mAbs defineacentraldomainnotessential for thebiological functionof IL-3.

The fourth epitopedefined in this study includeshuman residues

_PheI07

through

GluII9,

and is recognized by the anti-humanIL-3 mAb3G (Fig.5C).Again,as3G recognized

denaturedIL-3 inaWestern blot format (Fig. 4 A), these resi-dues likelyencompass most,ifnotall of its bindingepitope. And, as 3G blocksbinding of human IL-3 toits receptor (Table III),asdoesanFab fragment derived from it (datanot

shown),this result also defines residues that interact with the IL-3receptor. Asummaryof the mAb mapping studiesusing the interspecies chimera is presented in Fig. 6.

Cross-reactivityof anti-IL-3 mAb. To confirm the

assign-ment ofepitope regions usingthe interspecies chimeric

pro-teins, several pairs of mAbsweretested inanELISAformat. For the mostpart, pairs of mAbs thatwereassignedto

non-overlapping regionsofIL-3 wouldsimultaneouslybindtoIL-3 in thisassay(e.g., 3Gl / lF9 pair, Fig. 7). However,one

unex-pected findingwas the failure of 6G8 and 3G to simulta-neously bind to human IL-3. Two conclusions derive from these findings. First, although6G8 and lF9 bindtothesame

panel of IL-3 chimera, they apparently recognize subtly

differ-entepitopes. And second, the failure of6G8 and 3Gl to simul-taneously bindtoIL-3suggeststhat the epitopes for thesetwo

mAbsmayreside in closejuxtapositionin the tertiarystructure

of the molecule.

Secondarystructuralanalysis ofIL-3. Finally,tobeginto

correlate the functional domains of IL-3 with itsstructure, a

secondarystructureprediction of human IL-3 basedonseveral

differentcomputeralgorithmswasperformedandis shown in Fig. 8. Taken together, thesealgorithmspointtothepresence

of four helixes. Using the Chou and Fasmanpredictions,(that residues17-29,41-51,57-78,and103-125 ofhuman IL-3are

ahelixes) and the algorithm ofEdmunson,_{the first and fourth}

predicted helixesarehighlyamphipathic;_{i.e., contain both}

hy-1884 Kaushanskyetal.

A

43

-_-25 -_

18-_

9-egg!

A

43_-ft

18-4

C*, _ Ct C'

L _{E E E E}

_E0

_& m 0 m a a

*

a,

c -r

qp

- °

CO

E ES E E

_9

N 0 m 0 CD

V14

D21

Q45

c0

-" _{E E E}

F107

E119

S133

cv, o) T- to

J E E E

0a 01 0 m

_

-gm3 - gm2

- gm 6

Figure 5. Immunoprecipitation analysis. Conditioned medium(1 ml) from BHK-derived cell lines which produced the IL-3 chimera were immunoprecipitated with 10 g of a number of anti-IL-3 mAbs (A=6G8, B=8F8, C=_3GI1,D=8A5). The precipitates were size-fractionated by NaDodSO4 polyacrylamide gels, - gm 1 dried, and exposedtofilmfor 2-8 d. Molecularmass

markers (kD) are shown. Negative immunoprecipita-- _gm9 tionswere

_always

runwithaknownpositive

antibody

ascontrol. Each mAbwasused inatleastthree sepa-rateexperiments for each chimeric protein with simi-larresults. Theline maptotheright demonstrates thesites ofcrossoverbetweengibbonIL-3 sequence above andmurine IL-3 sequence below for each chi-meric molecule. The single letter amino acid code for thegibbon IL-3 residues is shownateachsite.

yim2

Figure 6. Binding epi-tope analysis. The amino acid sequence of humanIL-3 is depicted along withamapof the

F4*

cross-overpoints of the IL-3 chimera used in this study. The binding epitopes of the receptor blocking and neutraliz-ing mAb are shown to theleft,non-receptor blocking antibodies to therightof the map.

*Anti-murineIL-3 an-tibodies. Data on the sites of 6A5,4D4,and * 2E11arefromConlon

etal. andZiltener et al.

(50-52).

drophobicandhydrophilicfaces(Fig. 8,Band C).Next, the sequences of human and murine IL-3 in these regions were compared.Asshown, the human and murine IL-3 sequences differ

primarily

onthehydrophilicfaces.Strongsequence con-servation is presentonlyfor the(presumably)structurally

im-portant

hydrophobic

residues in these helixes. Ofnote, the two regions identifiedascritical forreceptorbinding, biological ac-tivity,and for thebindingofneutralizingmAbs are locatedon

thefirstandfourth

predicted

ahelixes.

Discussion

IL-3 is produced by activated T-lymphocytes in response to

antigenic

stimulation. Theimportanceof IL-3 inphysiologic and ina

growing

numberofpathologicstatesisbecoming

in-creasingly

clear. Andrecently, recombinant human IL-3 has been used to stimulate hematopoietic recovery in states of marrow failure.Inthis study,thestructure-function

relation-ships

of this

cytokine

have beenexplored.We report thattwo

regions

ofthe molecule,

widely

separated in the primaryamino acid sequence, are

required

forthe

biologic

activity ofIL-3. Thisconclusion is basedontwotypesofanalyses:on the recep-tor

binding

and

biological activity

of chimera of the

growth

Structure-FunctionRelationshipsofInterleukin-3 1885

B

43_-1

18-w

*

*

8F8, 19B3 4.4.7

43D114

3G11

4

1

0.7' E 0.6

en 0.5

a

t 0.4'

0 o 0.3'

ot0.2'

w

0.1-nn

CostDelect

Ab' NAb

-_--- 3G11F9

-a

.IFS

U.U . ..,-.

I °IS2lo, 103 104 10s

106

human IL-3 (ydst)conc(pglMl)

Figure7.Immunoenzymetricassayformattodetectmonoclonal

an-tibodypairsrecognizing spatiallydistinctepitopesonIL-3. The 3GI1

coating antibodyandtheIF9detecting antibody pair uniquely pro-videsasensitivetwo-sitesandwichimmunoassayfor human IL-3. The othercombinationseffectivelycrossblockimplying recognitionof

spatiallyoverlappingepitopes.

factor andonthelocation of the

binding epitopes

ofaseriesof neutralizingandnonneutralizing mAb that recognize human ormurine IL-3.

Using

aseries of

gibbon-murine

chimeric IL-3 molecules, we found that residues in both the amino-terminal half and

carboxy-terminal

half of the

growth

factorwererequired for fullbiologic

activity.

This

approach

tostructure-function analy-sis is basedonthe

assumption

thatreceptorinteractive sites ofa

proteincanbe

identified

by

substituting

regionsnotcritical for receptor

binding

with

homologous

domains of structural simi-larity.Inthiswaythe limitations of deletion

analysis (in

which removal ofstructurally

important

but

receptor-irrelevant

resi-duescannot be

distinguished

from elimination ofsites of recep-tor

interaction)

are overcome. Wehaverecently used this ap-proach to map the

receptor-interactive

residues ofthe

func-tionally

related

cytokine,

GM-CSF

(21 ),

results of whichwere

supported by

crystallographic analysis

ofthe

growth

factor,

onceit became available(45). Other successful

applications

of thisapproachincludemappingof the receptorbindingsite of a-interferons (46) and the

binding

siteof theHIVvirusonthe TcellCD4molecule (47). Recently, Chothia and co-workers

haveshown thatiftwo proteins share aslittleas 20%primary amino acid homology, their tertiarystructures are quite similar (48).Inadditionto26%aminoacid identityand 40% homol-ogybasedonconservativesubstitutions,thebiologically criti-cal

disulfide

bondof IL-3 is conserved throughout evolution (18,25 ).Furthermore, Pakula andSauerhave recently shown thatsurface side chain changeshavelittle impacton thetertiary structureofpolypeptides (49).Thus,thefindingthatnearly all of the amino acid changesbetweengibbon and murineIL-3in thetworegions found critical for biologic activityoccur at the surface of the growth factor (Fig. 8)suggeststhatthetertiary conformation of gibbon and murine IL-3arequite similarand that thefoldedstructureof thegibbon-murine hybrids used in this

study

approaches that of the native proteins.

Intheamino-terminal region, murineIL-3 residues

Leus

through Lys22 (correspondingtogibbon residues

Vall4

through

Asp21)

appear to be critical for colony-stimulating activity. Thiswasbest exemplifiedbycomparingthe colony-stimulat-ing activity ofmutants gm 3 and gm 2,whichdiffer only in five

positions

(Leuls

and

_Serj9

through Lys22.Furthermore,gibbon IL-3-specific residues in this regionwererequired for binding to the human IL-3 receptor(gm 6 vs. gm 2, TableII) and murine

IL-3-specific

amino acidswererequired for bindingto themurine IL-3receptor(gm 3vs.gm2).Inaddition,the

F6b

fragment

of

a mAb (termed 4.4.7) thatblocked bindingof humanIL-3 toitsreceptor also bound to human IL-3 in this same

region.

Similar resultswereobtained for the anti-murine IL-3mAbs 19B3and8F8.Themild reduction (sixfold)in the activity ofgm3 comparedtomurine IL-3suggestthat the first 15residues contribute inaminorway tofull biological activity. Additional residues

immediately

downstreamof this

region

also appear to beinvolved in receptorbinding.The anti-hu-manIL-3 mAbs 6G8, 2.6.23,and IF9, and theanti-murine IL-3 mAb43D11 recognizeaclosely contiguous epitope, lo-cated between human Asp21 and

Gln45

(murine Lys22 and

Pro4o).

Allof thesemAbsblockbindingto the

IL-3

receptor andneutralize the biological activity ofthe growthfactor. In

addition,

together withtwootherreports(19,20), thisstudy

A

__

1.

2. ahelix

3.

4.

B

I _ --7WA-en

C

Figure 8. Secondarystructurepredictions ofIL-3. (A)

The algorithms ofI) Garnier (41), 2)Parry(42),3)

Chou and Fasman(43), and 4) Gascuel (44)were

appliedtothe aminoacidsequenceof human IL-3. The predictions of Chou andFasmanwerenext

uti-lizedtodetermine the characteristics of each helix. The first (B) and fourth (C) helixesarepresented in

Edmunsonprojection. Thesequencesofhuman

(hIL3) and murine IL-3 (mIL3)aredepicted.

*Non-conservedaminoacids.

1886 Kaushanskyetal.

* * *

V D R

* K R16N05 T,12 N

D

109

R

N120 1418,

F _Fl13 Heiicaiwheel- L115M SequenceHIL3

D E106 position

_{105 to}

₁₂₀

1117Lill

H L118 L

K

1ohlU

'14 107 _L

mIL3 _y F

brings

to 16 the number of

neutralizing

mAbs that have been

mapped

toresidues between

Asp21

and

Ginso.

Thus, this

region

is

highly immunogenic

and

easily

accessible for

protein-pro-teininteraction.

Threeofthefour

secondary

structural

algorithms

applied

tothe

primary

structuresofhuman IL-3

predict

thatthis criti-calamino-terminal

region

is

likely

toassume ana-helical

con-formation in solution. Examinationofthehelix in the

Edmun-son

_projection

reveals thatofthefive residuesalteredbetween gm3 andgm

2,

threearesurface-oriented andtwoareburied in the

_hydrophobic

coreof the

protein.

Asthetwo

hydrophobic

changes

are

_relatively

conservative

(murine

IL-3

Ile20

toMet,

Val21

to

Ile),

thesurface

changes

likely

representcritical

recep-torinteractive residues. Thiswasconfirmed inpart

by

the

sub-stantially

reduced

activity

ofmutant2207

(murine

Val21-Lys22

tohuman

Ile20-Asp21). However, although

thisamino-terminal

region

is

important

forreceptor

binding,

it isnotsufficient for

biological activity.

Thesubstantial lossof function of chimera gm6and gm 1 is

likely

attributed

todissociation of these resi-dues andadditional critical

regions

of

IL-3,

andnotcaused

by

grossalterations instructural

integrity.

Strong

supportfor this latterconclusion is

provided

by

the

hybrid

gm

6,

amolecule

capable

of

_binding

totheIL-3receptor

despite

areduction in

biological

activity by

fourordersof

magnitude.

The second

region

of IL-3

required

for

biological activity

mapped

to residues between human IL-3

Phe107

and

_GluII9.

Thisconclusion isbasedonthedifferential

activity

ofthe gm 9 andgm 1 chimera.

Furthermore,

liketheamino-terminal

do-main,

this

region

alsobindsthe

Fab

fragment

ofaneutralizing

mAb,

3G1 1.

And,

like the

amino-terminal

functional

domain,

this

_region

isalso

predicted

to forman

amphiphilic

ahelix. Examination of the

gibbon-murine

homology

inthis helix is

even more

_striking

for

hydrophilic

substitutionsand

hydro-phobic

conservation.

Again,

these

findings

strongly

suggest that the

hydrophilic

aminoacid residues in this

region

are

re-sponsible,

along

withthoseofthefirst

helix,

forthe

species-spe-cific

activity

of IL-3

and, by extrapolation,

for the

biologic

activ-ity

ofthe

growth

factor.

Besides the two

_{regions required}

for receptor

binding

and

biological activity,

the_present

_study

andtwo

previous

studies havedemonstrated that theextremeamino-terminal and ex-treme

_{carboxy-terminal regions}

ofmurine

_IL-3,

aswellasthe central domain defined

by

the

_antipeptide

mAbs 8A5 and

1OF4,

are

_dispensable

for

biological activity.

Conlon and

co-workers,

and Zilteneret al. have

reported

independently

on twomAbsdirected

against

thesixamino-terminal residues of murine IL-3

(50, 51).

These antibodies failed to neutralize IL-3

bioactivity.

Ziltener and

colleagues

have also described

twomAbsthat

_mapped

very

closely

tothe

_carboxy

terminus of murine

IL-3,

residues

Ser,30

through

ArgI35

(52).

Neither of thesemAbsneutralizedthe

_bioactivity

of IL-3invitro.Taken

together,

thepresentand

previous

studies ofanti-IL-3 mAbs andantisera

_point

to_{the presence of several}

_neutralizing

epi-topesthatclusterinto two

_regions,

surrounded

_by

three

epi-topes which bindto

_regions

ofIL-3

_apparently

not

_immediately

involved inreceptor interaction.

To confirm the

binding

site

assignments

basedonthe

West-ernblot and

_{immunoprecipitation}

analyses,

manyofthe mAbs

weretested in

pairs

inasandwichELISA.As

expected,

mAbs 6G8 and

_IF9,

which

_appeared

to

_recognize

identical

_epitopes

intheimmunoprecipitation analysis,failedtosimultaneously

bindtohuman IL-3 in the ELISA(Fig. 7). Surprisingly, how-ever, although 1F9 bound to human IL-3 in the presence of 3G11, 6G8 failed to bind to IL-3 in the presence of 3GI 1. Two conclusions derive from these findings. First, 1F9 and 6G8

mustrecognize subtlydifferentepitopeswithin theregion

be-tween

Asp21

andGln45; andsecond, thebinding epitopes of 6G8 and 3G11, which are widely separated in the primary amino acid sequence of human IL-3(Asp21to

Gin45

and

_Phe,07

to

_GluII9,

respectively), likelyreside in closejuxtaposition in thetertiarystructureof the molecule. This latter conclusion is quitesimilartothefindings reportedinanalysesof GM-CSF, IL-2, and IL-6 (39, 53-55). And like the findings reported herein forIL-3, the activereceptorbinding sites ofthese cyto-kinesarealso composedof

amphipathic

helixes.

Finally,

the

ability

ofgm6tobindto the IL-3 receptor and itsnearfailuretoinduce signal transduction pointstothe im-portanceofthe fourth predicted helix in thislatter process. The

carboxy-terminal

domain may act to induce a proliferative stimulus inrespondingcellsbybindingto asignal-transducing

region

oftheIL-3receptor,bycausingreceptorinternalization, by inducing a conformational change in the receptor, or by othermechanisms. Additionalstudieswill berequiredto deter-mine the

precise

mechanism of this effect.

However,

the

capac-ity

todissociatereceptorbindingandsignaltransduction may find clinical

utility

as anIL-3 antagonist. Theimportanceof IL-3 inanumberof

pathologic

processes,

including

hypersensi-tivity

reactions

(4)

and leukemogenesis (6, 7), suggest that alternativetherapiesbasedontheprinciples of rationaledrug

design

may beforthcoming.

Acknowledaments

The authorswish to thank Zenaida Sisk for preparation of the

manu-script,StevenClark for the gibbon IL-3 cDNA, Frank Lee for the mu-rine IL-3 cDNA, Thalia Papayannopoulou for the OCIM 1 cell line, Margot Gibson for assistance with phosphorimaging, and Gerry Krys-talfor the '25ImIL-3.

References

1.Kobayashi, M.,B. H. VanLeeuwen, S. Elsbury,M. E.Martinson,I.G.

Young,and A. J.Hapel.1989.Interleukin-3 is significantlymoreeffectivethan

other colony-stimulating factors in long-term maintenance of human bone marrowderived colonyforming cells in vitro. Blood. 73:1836-1841.

2.Bot,F.J., L.vanEijk, P. Schipper, and B. Lowenberg. 1989. Effectsof human interleukin-3 on granulocytic colony-forming cells in human bone marrow.Blood. 73:1157-1160.

3.Rothenberg,M.E., W. F. Owen,D.S.Silberstein,J.Woods,R. J. Sober-man, K. F.Austen, andR. L.Stevens. 1988. Humaneosinophilshaveprolonged

survival,enhanced functionalproperties,andbecomehypodensewhenexposed

tohumaninterleukin 3. J. Clin.Invest.81:1986-1992.

4.Kita, H.,0.Tsukasa,Y.Okubo,D.Weiler,J. S.Abrams,and G. J. Gleich.

1991.Granulocyte/macrophage colony-stimulatingfactorandinterleukin3

re-lease fromhumanperipheralbloodeosinophilsandneutrophils.J.Exp. Med. 174:745-748.

5.Valent, P.,K.Geissler,C.Sillaber,K.Lechner,and P.Bettelheim. 1990.

Why clinicians shouldbeinterestedinInterleukin-3.Blut. 61:338-345.

6.Ihle, J.,K.Morishita,D.Parker,C.Bartholomew,D.Askew,A.Buchberg,

N.Jenkins,N.Copeland,and Y.Weinstein. 1989.Mechanismsin

transforma-tion ofIL3-dependenthematopoieticstemcells. Microbiol.Immunol. 149:59-69. 7.Grimaldi,J.C.,and T.C.Meeker. 1989. Thet(5;14)Chromosomal

trans-locationin a caseofacutelymphocyticleukemiajoinsthe interleukin-3 gene to theimmunoglobulin heavychain gene. Blood. 73:2081-2085.

8. Park,L. S., P. E. Waldron,D. Friend,H. M. Sassenfeld, V.Price,D.

Anderson,D.Cosman,R.G. Andrews,1. D.Bernstein,and D. L. Urdal. 1989.

Interleukin-3,GM-CSF,and G-CSFreceptorexpressiononcell lines andprimary

leukemia cells: Receptorheterogeneityandrelationshiptogrowthfactor

respon-siveness. Blood.74:56-65.

9. Park, L. S., D. Friend, V. Price, D. Anderson, J. Singer, K. S. Prickett, and D.Urdal. 1989.Heterogeneity ofhumaninterleukin-3 receptors. J. Biol. Chem. 264:5420-5427.

10. Itoh, N., S. Yonehara, J. Schreurs, D. Gorman, K. Maruyama, A. Ishii, I. Yahara, K-I. Arai, and A. Miyajima. 1990. Cloning of an interleukin-3 receptor gene: a memberofadistinct receptor gene family. Science (Wash.DC). 247:324-327.

1 1.Kitamura, T., N. Sato, K.Arai,and A. Miyajima. 1991. Expression clon-ingof the human IL-3 receptor cDNA reveals a sharedPsubunit forthe human IL-3and GM-CSF receptors. Cell. 66:1165-1174.

12. Bazan, J. F. 1990.Structuraldesign and molecular evolution ofa cytokine receptorsuperfamily.Proc. Nall. Acad. Sci. USA. 87:6934-6938.

13. Budel, L.,0.Elbaz, H. Hoogerbrugge, R. Delwel, L. Mahmoud, B. Lo-wenberg, and I. Touw. 1990. Common binding structure for granulocyte macro-phage colonystimulatingfactor andinterleukin3onhumanmyeloidleukemia

cells and monocytes. Blood. 75:1439-1445.

14.Hayashida,K., T. Kitamura, D. M. Gorman, K. Arai, T. Yokota, and A. Miyajima. 1990. Molecularcloningofasecond subunit ofthe receptor for human granulocyte-macrophagecolony-stimulatingfactor(GM-CSF):reconstitution of ahigh affinityGM-CSF receptor. Proc. Nail. Acad. Sci. USA. 87:9655.

15.Taketazu, F., S. Chiba, K. Shibuya, T. Kuwaki, H. Tsumura, K. Miya-zono, K.Miyagawa,and F. Takaku. 1991. IL-3specificallyinhibits GM-CSF

bindingtohigher affinityreceptor. J. Cell. Physiol. 146:251-257.

16. Lopez, A., J.Eglinton,D.Gillis,L.Park,S.Clark,and M. Vadas. 1989. Reciprocal inhibitionof bindingbetween interleukin 3 and

granulocyte-macro-phagecolony-stimulating factortohumaneosinophils. Proc. Nall. Acad. Sci. USA.86:7022-7026.

17.Clark-Lewis, I., R. Aebersold, H. Ziltener, J. Schrader, L. Hood, and S. Kent.1986. Automated chemicalsynthesisofaprotein growth factor for

hemo-poieticcells,interleukin-3.Science(Wash.DC). 231:134-139.

18. Clark-Lewis, I., L. Hood, and S. Kent. 1988. Role of disulfide bridges in

determiningthebiological activityof interleukin 3. Proc. Natl.Acad. Sci. USA.

85:7897-7901.

19. Dorssers, L. C. J.,M.C.Mostert,H.Burger,C.Janssen,P. J.Lemson, R. vanLambalgen, G.Wagemaker,and R. W.vanLeen.1991.Receptorand anti-body interactions of humaninterleukin-3 characterizedbymutationalanalysis.

J.Biol. Chem.266:21310-21317.

20. Lokker, N. A., U.Strittmatter,C.Steiner,B.Fagg,P.Graff,H.P.Kocher,

and G. Zenke. 1991. Mappingtheepitopesofneutralizinganti-human IL-3 monoclonalantibodies. J. Immunol. 146:893-898.

21.Kaushansky,K., S.Shoemaker,S.Alfaro,andC. Brown. 1989. The

hema-topoietic activityofgranulocyte-macrophage colony-stimulatingfactor is

depen-dent upontwodistinctregionsof the molecule:afunctionalanalysisbased upon the activitiesofhybrid growth factors.Proc.Natl.Acad. Sci. USA. 86:1213-1217. 22.Kaushansky, K., P. O'Hara, C. Hart, J. Forstrom,and F.Hagen. 1987. Role ofcarbohydratein thefunction ofhumangranulocyte-macrophage colony-stimulatingfactor.Biochemistry.26:4861-4867.

23.Sanger, F., S. Nicklen, andA.R.Coulson. 1977.DNAsequencingwith

chain-terminatinginhibitors. Proc.Natl.Acad. Sci. USA. 74:5463-5467. 24.Berkner, K., S. Busky, E.Davie,C.Hart,M.Insley,W.Kiesel,A.Kumar, M.Murray,P.O'Hara, R.Woodburg,and F.Hagen.1986.Isolation/expression ofcDNAsencodinghuman factor VII.CSHSymposiumonQuantitalion Biol-ogy. 51:531-541.

25.Otsuka, T.,A.Miyajima,N.Brown, K. Otsu, J. Abrams, S. Saeland, C.

Caux,R. de WaalMalefijt,J.deVries,P. Meyerson,etal. 1988.Isolationand

characterizationofanexpressiblecDNAencodinghuman IL-3.Induction ofIL-3 mRNA inhumanTcell clones.J. Immunol.140:2288-2295.

26.Hapel,A.J.,H.S.Warren, andD. A. Hume.1984.Different

colony-stim-ulatingfactorsaredetectedby the interleukin-3 dependent cell lines FDC-PI and 32D cl-23.Blood.64:786-790.

27.Papayannopoulou, Th., B. Nakamoto, S. Kurachi, M. Tweeddale, and H. Messner. 1988. Surfaceantigen profileandglobinphenotype of two new human

erythroleukemia lines-characterization and interpretations. Blood. 72:1029-1038.

28. Broudy, V. C.,N.Lin, J. Egrie, C. de Haen, T. Weiss, Th. Papayannopou-lou,and J. W.Adamson. 1988. Identification of the receptor for erythropoietin

onhuman andmurineerythroleukemiacells and modulation by phorbol ester anddimethyl sulfoxide. Proc.Nail.Acad. Sci. USA. 85:6513-6517.

29.Murthy, S. C., P. H. B. Sorensen, A. L.-F. Mui, and G. Krystal. 1989. Interleukin-3 down-regulates its own receptor. Blood. 73:1180-1187.

30.Calvo,J.C., J. P. Radicella, and E. H. Charreau. 1983. Measurement of

specificradioactivities in labeled hormones by self-displacement analysis.

Bio-chem.J.212:259-264.

31. Munson, P. J., and D. Rodbard. 1980. LIGAND: a versatile computerized

approach for characterization of ligand-binding systems. Anal. Biochem.

107:220-239.

32.Rector, E. S., R.J.Schwenk, K. S. Tse, and A. H. Sehon. 1978. A method for the preparation of protein-protein conjugates of predetermined composition. J. Immunol.Methods. 24:321-336.

33. Abrams, J. S., and M. K. Pearce. 1988. Development of rat anti-mouse interleukin 3 monoclonal antibodies which neutralize bioactivity in vitro. J. Im-munol. 140:131-137.

34. Mosmann, T. R.,J. S. Abrams, N. Arai, M. W. Bond, F. D.Lee, A. Miyajima, S. Miyatake, D. M. Rennick, J. Schreurs, C. A. Smith, et al. 1987. Mouse interleukin-3 recombinant cDNA cloning, expression and characteriza-tion. In Recombinant Lymphokines and Their Receptors, Immunology Series 35.S.Gillis, editor. Marcel Dekker, New York, pp. 217-240.

35. Lee, R.,J.Abrams, N. Arai, A. Miyajima, S. Miyatake, T. Mosmann, D. Rennick, J. Schreurs,C. Smith, Y. Takabe, etal. 1987.

Theexpressionandcharac-terizationof recombinant mouseIL3.Lymphokines. 15:163-182.

36. Mosmann, T. 1983. Rapid colorimetric assay for cellular growth and survival: application to proliferation and cytotoxicity assays. J. Immunol. Meth-ods. 65:55-63.

37.Reichlin, M., A. Nisonoff, and E. Margoliash. 1970. Immunological activ-ity of cytochrome C. III. Enhancement of antibody detection and immune re-sponse initiation by cytochrome C polymers. J. Biol. Chem. 245:947-954.

38.Kitamura, K., Y-F. Piao, K. Miyazono, A. Urabe, and F. Takaku. 1989. Establishment and characterization of a unique human cell line that proliferates dependently onGM-CSF,IL-3, or erythropoietin. J. Cell. Physiol. 140:323-334. 39.Brown, C. B., C. E. Hart, D. M. Curtis, M. C. Baily, and K. Kaushansky. 1990. The neutralizing monoclonal antibodies against human granulocyte-mac-rophage colony-stimulating factor recognize the receptor binding domain of the molecule. J. Immunol. 144:2184-2189.

40.Chretien, I., A. Van Kimmenade, M. K. Pearce, J. Banchereau, and J. S. Abrams. 1989. Development of polyclonal and monoclonal antibodies for immu-noassay and neutralization of human interleukin-4. J. Immunol. Methods.

117:67-81.

41.Garnier, J. 1978. Analysis of the accuracy and implications of simple methods for predicting the secondary structure of globular proteins. J. Mol. Biol. 120:97-120.

42.Parry, D., E. Minasian, and S. Leach. 1988.Conformationalhomologies among the cytokines: interleukins and colony-stimulating factors.

MolecularRec-ognition. 1:107-110.

43.Chou, P. Y., and G. D. Fasman. 1978. Prediction of the secondary struc-ture of proteins from their amino acid sequence. Adv. Enzymol.Relat.Areas Mol. Biol. 47:45-148.

44.Gascuel,O.,and J. L. Golmard. 1988. A simple method for predicting the secondary structure of globular proteins; implications and accuracy. Comput.

Appl. Biosci. 4:357.

45. Diederichs, K., T. Boone, and P. A. Karplus. 1991. Novel fold and puta-tive receptor binding site of granulocyte-macrophage colony-stimulating factor. Science (Wash.DC).254:1779-1782.

46. Raj, N., R. Israeli, K. Kelley, S. Leach, E. Minasian, K. Sikaris, D. Parry, and P. Petha. 1988. Synthesis, antiviral activity, and conformational characteriza-tion of mouse-human alpha-interferon hybrids. J. Biol. Chem. 263:8943-8952. 47. Landau, N. R., M. Warton, and D. R. Littman. 1988. The envelope glycoprotein of the human immunodeficiency virus binds to the immunoglobu-lin-like domain of CD4.Nature(Lond.). 334:159-162.

48. Chothia, C., and A. M. Lesk. 1986. The relation between the divergence of sequence and structure in proteins. EMBO (Eur. Mol. Biol. Organ)J. 5:823-826.

49.Pakula, A. A., and R. T. Sauer. 1989. Genetic analysis of protein stability and function. Annu. Rev.Genet.23:289-310.

50. Conlon, P.J.,K. H. Luk, L. S. Park, C.J.March, T. P. Hopp, and D.L.

Urdal. 1985. Generation of anti-peptide monoclonal antibodies which recognize matureCSF-2A(IL 3) protein.J.Immunol. 135:328-332.

5 1. Ziltener, H., I. Clark-Lewis, B. de St. Groth, L. Hood, S. Kent, and J. Schrader. 1987. Antipeptide antibodies define the NH2-terminal structure of the pan-specific hemopoietin interleukin 3. J. Immunol. 138:1105-1008.

52. Ziltener, H., I. Clark-Lewis, B. de St. Groth, P. Orban, L. Hood, S. Kent, andJ.Schrader. 1988. Monoclonal antipeptide antibodies recognizeIL-3and neutralize its bioactivity in vivo. J. Immunol. 140:1182-1187.

53. Brandhuber, B.J.,T.Boone, W. C. Kenney, and D. B. McKay. 1987. Three-dimensional structure of interleukin-2. Science (Wash. DC).

238:1707-1709.

54. Brakenhoff,J.,P.J.M.Hart, and L. A. Aarden. 1989. Analysis of human IL-6 mutants expressed in Escherichia coli biologic activities are not affected by deletion of amino acids 1-28. J. Immunol. 143:1175-1182.

55.Ida, N., S. Sakurai, T. Hosaka, K. Hosoi, T. Kunitomo, T. Shimazu, T. Maruyama, Y. Matsuura, and M. Kohase. 1989. Establishment of strongly neu-tralizing monoclonal antibody to humaninterleukin-6and its epitope analysis. Biochem. Biophys. Res. Commun. 165:728-734.