WilliamG.T.WillatsSvendErikRasmussenTinaKristensenJ - PDF document

WilliamG.T.WillatsSvendErikRasmussenTinaKristensenJørnDalgaardMikkelsenJ.PaulKnoxCentreforPlantSciences,UniversityofLeeds,Nunc,Roskilde,DaniscoCultorInnovation,Copenhagen,DenmarkSugar-coatedmicroarrays:Anovelslidesurfaceforthehigh-throughputanalysisofglycans Correspondence: 1666–1671Sugar-coatedmicroarrays1667microarrayer(Affymetrix417;SantaClara,CA,USA)withapindiameterof150m.Spotswerespaced375apartandappliedatarateoffourspots/second.Thespotvolumewasapproximately50pL.Sampleswerearrayedfrom30Laliquotscontainedin96-wellmicro-titreplateswithconicalwells(Corning,Cambridge,MA,USA).Upto42slideswerearrayedfromeachmicrotitreplateandinallcases15replicatesofeachsamplewasarrayed.Foreachsamplearray,acorrespondingregisterarraywasgeneratedinwhichcarmoisinedyewasarrayedinthesameconfigurationasthesamples.Thisenabledthespatialrelationshipsbetweenthespotstobeaccu-ratelydeterminedduringsubsequentanalysisoftheprobedarrays.Theextentofsubarrayswasdelineatedbytheinclusionofdextranconjugatedtofluorosceiniso-thiocyanate(detran/FITC;Sigma-Aldrich,Poole,UK)intoselectedwells.Arrayedslideswerestoreddryatroomtemperatureforuptothreemonthspriortoprobingandanalysis.Thesignalfromprobedarrayswascollectedusingalaserarrayscanner(Affymetrix418).Selectedarrayswereanalysedusingimageanalysissoftware;BioDiscovery,MarinadelRey,CA,USA)whichenabledtherelativesignalfromeachspottobequantified.Datawerenotcollectedfromarrayscontain-ingobviousdefectssuchasincompleteremovalofbuffersalts,orcompletesampledropoutscausedbydepletionofsamplewithinspottingringsduringarraying.2.2SamplepreparationArrayedsampleswerecomplexpolysaccharides,proteo-glycansandneo-glycoproteins.Limepecticpolysaccha-ridesampleswithknowndegrees(DE)andpatternsofmethyl-esterificationwerepreparedasdescribedpre-viously[11]fromahighlymethyl-esterifiedsample(E81)bytreatmentwithpectinmethylesterasefrom(F-seriessamples)ororange(P-seriessamples)orbybasecatalysis(B-seriessamples).Galactanfromnumtuberosum(Galactan2())andLupinusaugustifo-(Galactan1())andarabinanfromBetavulgarisbinan())weresuppliedbyMegazyme(Bray,Ireland).XylogalacturonanfromPisumsativumwasagenerousgiftfromProfessorJeanFranc¸oisThibault(INRA,Nantes,France).Arabinogalactan-proteinproteoglycans(AGP)andAGP2())werepurifiedformDaucuscarotacellcultures[12].MixturesofcomplexpolysaccharideswereextractedfromsuspensionculturedNicotianata-(lineBY2)andD.carota(lineOx6)cellsandN.tabacumleavesbyhomogenisationinabuffercon-aceticacid(CDTA).TheconditionedmediumofBY2andOx6cellswasobtainedpreparedbyfiltrationofcellcultures.Theneo-glycoprotein(1waspreparedbyconjugationoffourresiduesof(1-galactosetoBSAa3atomspacerasde-scribedpreviously[13].Allpurifiedsampleswereappliedatalevelof0.5mg/mLinPBS(0.14NaCl,2.7m7.8mO,1.5m,pH7.2)andconditionedmediaandcellandtissueextractswereappliedasthree-folddilutionsinPBS.Selectedsampleswereappliedasfive-folddilutionseriesinPBSwithastartingconcentrationof1mg/mL.2.3MonoclonalantibodyprobesGlycanmicroarrayswereprobedwithapanelofmAbswithspecificitiesforarangeofglycaneptiopes.Theanti-bodiesJIM5andJIM7bindtohomogalacturonanwithlowandhighdegreesofmethyl-esterificationrespectively[14],whilePAM1recognisesanepitopeconsistingofun-esterifiedhomogalacturonanwithadegreeofpoly-merisationofatleast30galacturonicacidresidues[14].LM5andLM6recognise(1-galactanand(1-arabinanrespectively[14].LM8recognisesxylogalac-turonan[14],andLM2recognisesaglycanepitopeofara-binogalactan-proteinsproteoglycans.Allantibodiesweregeneratedbyconventionalhybridomatechnology,exceptPAM1whichwasisolatedfromaphagedisplaysinglechainsyntheticantibodylibrary[15].2.4ProbingarraysArrayedslideswereblockedwithPBScontainingBSA(3%v/w)(3%BSA/PBS)foratleast1h.Afterwashingfor2mininPBS,slideswereincubatedinantibodysolu-tionsforatleast1h.Hybridomaantibodieswereusedas1/20dilutionofhybridomasupernatantsin3%BSA/PBS.PAM1wasusedataconcentrationof10g/mL,alsodilutedin3%BSA/PBS.Followingantibodyincubations,slideswerewashedtwicefor10mininPBScontaining0.1%v/vpolyoxyethylenesorbitanmonolaurate(Tween20,PBST),followedbyone5minwashinPBS.Slideswerethenincubatedinsolutionscontainingsecondaryantibodiesconjugatedtothefluorescentdyecyanine(Cy3).Forhybridomaantibodies,ananti-rat/Cy3conju-gatewasused(AmershamBiosciences,LittleChalfont,UK),whilePAM1bindingwasdetectedusinganti-c-myc/Cy3conjugate(Sigma).Allsecondaryantibodieswereusedas1/100dilutionsin3%BSA/PBSandincu-bationswereforatleast1h.Allprimaryandsecondaryantibodyincubationswereperformedeitherundercoverslips(usingavolumeof80L)orincustom–made28mm80mmplasticenvelopes(usingavolumeof3mL).Incubationsinenvelopestendedtoresultinsuper-iorS/Nratiosandallresultsshownwereobtainedusingenvelopeincubations.Followingincubationinsecondary 1668W.G.T.Willatsetal.Proteomicsantibodies,slideswereagainwashedinPBSTandPBSasdescribedabovethenrinsedbrieflyindeionisedwatertoremovePBSbuffersalts.Inordertoremovesurfacewater,slideswereplacedin50mLcentrifugetubesandcentrifugedfor3minat1500.Anyremainingwaterwasremovedbybrieflyholdingslidesinanairstream.Slideswerestoredatroomtemperatureinanairtightbox.3ResultsBlackpolystyreneslideswereproducedwithaphysicalsurfacemodification(knownasMaxiSorp;Nunc,Ros-kilde,Denmark)(Fig.1A).Slidesweremadetoastandardsize(25mm75mm)andwereusedinconventionalmicroarrayequipment.Themicrostructureoftheslidesurface,revealedbyscanningatomicforcemicroscopy(Fig.1B),consistsofaseriesofridgesthatincreasethe Figure1.Developmentofnovelslidesurfacesforthedirectimmobilisationofcarbohydrates.Slideswerecastinblackpolystyrene(A)withaphysicalmodificationthatcreatedasurfacewiththecapacitytoimmobilisestruc-turallyandchemicallydiverseglycans.Theareaavailableforbindingisincreasedbythepeaksandtroughsinslidesurfacemicrostructurerevealedbyscanningatomicforcemicroscopy(AFM)(B).However,theoverallslidesurfaceisextremelyflat,asshownbyAFMsectioningthroughamportionofaslidesurface(C).surfaceareaavailableforbinding.However,thegrossslidesurfaceisextremelyflatwithatypicalmaximumheightvariationoflessthan40nmover10m(Fig.1C).Moreover,becausethebindingsurfaceiscreatedbyphysicalmodificationtothepolymerratherthanbyacoating,thesurfaceisalsohighlyconsistentinnature,bothwithinandbetweenslides.AnadvantageofcreatingslidesusingtheMaxiSorpsurfaceisthatthebindingpropertiesofthismodificationhavebeencharacterisedpreviously[16,17].Bindingisestablishedbypassiveadsorptionmediatedbyhydrogenbonding,ionicbondingandvandeWaalshydrophobicinteractions.Hydrogenandionicinteractionsareprimarilyinvolvedincapturingpolarpartsofmoleculesandsecuringatightjunctionbe-tweenthesurfaceandthemolecule.Immobilisationisfurtherstabilisedbytheremovalofwatermoleculesbe-tweentherelativelyhydrophobicsurfaceandtheimmobi-lisedmolecule.Waterremovalisdrivenbyaloweringofthefreeenergyinthesystembyexchangingtheweakwaterbondswithhydrophobicbonds.Thiscombinationofbindingmechanismsallowsthestableimmobilisationofarangeofglycanstructureswithdifferingphysicalandchemicalproperties.Wetestedthesuitabilityofthissurfaceforgeneratingcarbohydratemicroarraysbyanalysingtheimmobilisa-tionofarangeofglycanstructuresderivedfrom,oroccur-ringin,plantcellwallswhichcontainsomeofthemostcomplexglycansfoundinnature.Samplesanalysedwerepolysaccharides,proteoglycansandneo-glycoproteinsaswellasplantcellextracts.Aseriesofidenticalmicro-arrayswascreatedandimmobilisationwasassessedbythebindingofapanelofpreviouslycharacterisedmAbswithspecificitytocarbohydrateepitopes[14,15](Fig.2).Forallthesamplestestedtheantibodybindingprofilesindicatedthatantigenshadbeeneffectivelyimmobilisedandthatepitopeconformationswerepreserved.TheepitopesrecognisedbytheantibodiesJIM5,JIM7andPAM1allconsistofhomogalacturonan(HG,ahomo-polymerof(1-galacturonicacid)butdifferindegreeandpatternofmethyl-esterification(DE).ThedifferentialbindingoftheseantibodiestoaseriesofmicroarrayedpecticpolysaccharideswhichdifferedonlyinDEindicatedthatsubtlepostsyntheticmodificationstocomplexpolysaccharidescanbedeterminedusingthesearrays(Fig.2C–E).Inordertobeabletoeventuallycreatecomprehensivemicroarraysofglycomesitisnecessarythattheimmobi-lisationsurfaceusedshouldbecapableofbindingnotjustpurecarbohydratestructuresbutalsoglycoproteinsandproteoglycans.Thiswastestedbyarrayingtheneo-glycoprotein(1-galactan-BSAandarabinogalac-tan-proteinproteoglycans.ThebindingofLM5andLM2 1666–1671Sugar-coatedmicroarrays1669 Figure2.Carbohydratedetectingmicroarrays.Aseriesofidenticalcarbohydratemicroarrays(A–G)werecreatedbyimmobilisingaseriesofpolysaccharides,proteoglycans,neo-glycoproteinsandplantcellextracts(listedintopline,fordetailsseeSection2.2).Fifteenreplicatesofallsampleswereapplied–fivereplicatesareshown.ThearraysincludedaseriesofpecticpolysaccharidesdifferingonlyintheirDE.TheimmobilisationofarrayedglycanswasassessedbyprobingwithapanelofmAbs(listedright)withspecificitytocarbohydrateepitopes.AntibodybindingwasdetectedbyprobingwithCy3conjugatedsecondaryantibodies.Aregisterarray(H)ofdirectlyappliedcarmoisinedyewasincludedinordertotrackthepositionofallsamplesanddirectlyimmobilisedfluorescentlylabelleddextran(*)wasusedtoestablisharrayorientation.PBSwasusedasanegativecontrol.respectivelytotheseantigensindicatedthatproteogly-cansandglycoproteinscanbeeffectivelyimmobilisedusingtheseslides(Fig.2.B,F).Thereproducibleanddetectionlimitsofantibodybindingtoimmobilisedgly-canswerealsotested(Fig.3).Thedetectionlimitsofthreeantibodiesweretestedbyprobingarrayeddilutionseriesofselectedantigensandquantifyingthesignalsobtained.Galactanpolymerandpecticpolysaccharideweredetectedtoalevelof1.6g/mLbyantibodiesLM5andJIM5(Fig.3.A,B),whilexylogalacturonanwasdetectedtoalevelof40g/mLbyantibodyLM8(Fig.3.C).Basedonaspotvolumeof50pL,adetectionlimit1.6correspondstoaminimumdetectableamountof80fg.Theabilitytodetectverylowlevelsofmaterialarrayedat 1670W.G.T.Willatsetal.Proteomics Figure3.Thedetectionlimitsofthreeantibodieswasassessedbyprobingarrayeddilutionsseries(1mg/mL–g/mL)ofgalactanpolymer(Galactan1()),pecticpolysaccharideandxylogalacturonanwithantibodiesLM5(A),JIM5(B)andLM8(C)respectively.Thesignalfromeachspotwasquantifiedandtherelativemeansofsignalsof10replicatesisindicatedaboveeachsample.AllSEMswerelessthan3.8%ofthemaximumrelativeTable1.Thedetectionlimitandspatialdensityofcarbo-hydratemicroarrayscomparedtoELISAsandimmunodotassays.Valuespertaintothebind-ingofmAbJIM5topartiallymethy-esterifiedlimepectin[22].ELISAImmunodotDetectionlimit0.1ng/mLDetectionlimit80fg5pg10ngTypicaldensityofarrayedsamplesUpto10000a)Foraspotvolumeof50pLb)Foraplatecoatingvolumeof50c)Foraspotvolumeof1veryhighdensityisasignificantadvantageofmicroarrayscomparedtoexistingmethodsofcarbohydrateanalysis,suchasELISAsandimmunodotassays(Table1).Bothintra-andinter-arrayconsistencywashigh.Intra-arrayconsistencywasassessedbycollectingsignalsfromfiveseparatesets(ondifferentslides)of15replicatesofspotsofanF-seriespecticpolysaccharidesamplewithaDEof31%(F31)asdetectedbyprobingwithantibodyJIM5.Inter-arrayconsistencywasassessedbyquantifyingthedifferenceinsignalsfromspot-pairsoftwodifferentpec-ticpolysaccharides,F31andaB-seriespecticpolysac-charidewithaDEof15%.Again,fivesetsof15replicateswerearrayedoverfiveseparateslidesandprobedwithJIM5.TheSEMintra-arrayvariationwas3.2%ofthemax-imumrelativesignal,whiletheSEMoftheinter-arrayvar-iationwas5.1%ofthemaximumrelativesignal.TheS/Nratioobtainedwasgenerallyhighandthiswasinpartduetothefactthatslideswereproducedusingblackpolymerresin.TransparentslideswithidenticalsurfacepropertieswerealsotestedbutgavesignificantlyinferiorS/Nratios(resultsnotshown).Arrayedslideswerestoreddryforuptothreemonthsbeforeprobingandanalysis.Nosignifi-cantqualitativedifferencesinsignalswereobtainedforstoredslidescomparedtofreshlyarrayedslidesandtheimmobilisationofglycansontotheslidesthereforeappearedtobestableduringprolongedstorage.4DiscussionThefundamentalimportanceofcarbohydratestructuresinbiologyandpathologyisbecomingevermoreappar-ent.Glycansareknowntoplaycentralrolesindevelop-ment,carcinogenesis,celladhesion,andimmunityandareincreasinglythesubjectoftherapeuticapproaches[18].However,theindirectlinkbetweengeneexpressionandglycanendproducts,aswellastheoftenextremecomplexityofglycanscreatesanurgentneedforthedevelopmentofhigh-throughput,directphysicalmethodsofanalysis.Wehavedescribedanovelmicroarrayslidesurfacethatcanbeusedfortheimmobilisationofdiverseglycanstructures.Thearraysarehighlyreproducible,stableandaremadeusingstandardmicroarrayequip-mentandsimpleprobingprocedures.Importantly,carbo-hydratesarearrayeddirectlywithouttheneedformodifi-cationtogeneratereactivegroups.Theutilityofthesemicroarraysisprimarilyfortherapid,sensitiveandveryhigh-throughputanalysisoftheoccurrenceofglycans,ratherthanasatoolforquantification.Webelievethattheseslidesmakeasignificantcontributiontotherangeofsurfacesavailableformicroarrayproduction[19–21]andbringintoreachthepossibilityoftheglobalanalysisofglycomes.However,asisthecaseforproteinmicro-arrays,somemajorchallengesremain.Thenomenclatureadoptedforproteinarraysprovidesausefulframeworkfordevelopingstrategiestodevelopthisrapidlyevolvingareaoftechnology[9].Twotypesofarraysaredefined:functionarrays,anddetectingarrays.Infunctionarraysthearrayedtargetsareprobedwitha 1666–1671Sugar-coatedmicroarrays1671fluorescentlylabelledbaitormixturesofbaits.Incontrast,theconstructionofdetectingarraysinvolvestheimmobi-lisationofligands(suchasantibodies)whichareusedtocapturetheirbindingpartnersfromcomplexmixtures.Bindingisdetectedeitherbythebulklabellingoftargetmixturespriortoexposuretothearray,orbyprobingforcapturedtargetswithantibodiesorotherligands.Wehavereportedherethedevelopmentofcarbohydratefunctionarraysandweenvisagethatsucharrayswillhaveseveralimportantapplications.Theseincludetheidentificationoftheglycanmoietiesrecognisedbyputa-tivecarbohydratebindingproteinsand,aswehavedemonstrated,thehigh-throughputcharacterisationofantibodyspecificities.Wealsoanticipatethatthesearraysmaybeusedtoidentifycarbohydratebindingpep-tidesbyscreeningphagedisplaypeptidelibraries.Suchpeptidesmayhaveuseastherapeuticagents,andpro-videsequencedataforuseinconvergentevolutionstudiestoidentifyendogenouscarbohydratebindingpro-teins.Animportantaimforthefutureistoarraytheoutputfromseparationsofcellextractsinordertocompareglycomesduringdevelopmentordisease.Recentadvan-cesinverysmall-scaleextractionandfractionationtech-nologieswithnanoscalefractionoutputsthatcanbedir-ectlyarrayed,makesthisarealisticgoalforthenearfuture.ThankstoYasukoKamisugi,IainManfield,FerencMar-incs,VibekeRowellandThomasVorre-Grøntvedfortheirhelpfuldiscussions.ReceivedFebruary20,20025References[1]Lander,E.S.,Nat.Genet.Suppl.,3–4.–4.Brown,P.O.,Botstein,D.,Nat.Genet.Suppl.,33–37.[3]Lipshutz,R.J.,Fodor,S.P.A.,Gingeras,T.R.,Lockhart,D.Nat.Genet.Suppl.,20–24.[4]Debouck,C.,Goodfellow,P.N.,Nat.Genet.Suppl.[5]Cunningham,M.J.,J.Pharmacol.Toxicol.Methods,291–300.[6]Tomlinson,I.M.,Holt,L.J.,GenomeBiol.,1004.1–[7]Feizi,T.,Glycoconj.J.,553–565.[8]Haab,B.B.,Dunham,M.J.,Brown,P.O.,GenomeBiol.,1004.1–1004.13.[9]Kodadek,T.,Chem.Biol.,105–115.[10]Schreiber,S.L.,MacBeath,G.,,1760–[11]Limberg,G.,Korner,R.,Buchholt,H.C.,Christensen,T.M.I.etal.Carbohydr.Res.,293–307.[12]Knox,J.P.,Lindstead,P.J.,Peart,J.,Cooper,C.,Roberts,PlantJ.,317–326.[13]Jones,L.,Seymour,G.B.,Knox,J.P.,PlantPhysiol.,1405–1412.[14]Willats,W.G.T.,McCartney,L.,Mackie,W.,Knox,J.P.,PlantMol.Biol.,9–27.[15]Willats,W.G.T.,Gilmartin,P.M.,Mikkelsen,J.D.,Knox,J.P.,PlantJ.1999,,57–65.[16]Poulsen,L.K.,Pedersen,M.F.,Malling,H.J.,Sondergaard,I.,Weeke,B.,,173–180.[17]Kralovec,J.A.,Laycock,M.V.,Richards,R.,Usleber,E.,Toxicon,1127–1140.[18]Hirabayashi,J.,Arata,Y.,Kasai,K.-I.,[19]Wilson,D.S.,Nock,S.,Curr.Opin.Chem.Biol.[20]Benters,R.,Niemeyer,C.M.,Wöhrle,D.,Chembiochem.,686–694.[21]Wang,D.,Liu,S.,Trummer,B.J.,Deng,C.,Wang,A.,,234–235.[22]Willats,W.G.T.,Limberg,G.,Buchholt,H.C.,vanAlebeek,etal.CarbohydrateRes.,309–320.