• No results found

Haemophilus parainfluenzae expresses diverse lipopolysaccharide O-antigens using ABC transporter and Wzy polymerase-dependent mechanisms

N/A
N/A
Protected

Academic year: 2021

Share "Haemophilus parainfluenzae expresses diverse lipopolysaccharide O-antigens using ABC transporter and Wzy polymerase-dependent mechanisms"

Copied!
16
0
0

Loading.... (view fulltext now)

Full text

(1)

Haemophilus parainfluenzae expresses diverse

lipopolysaccharide O-antigens using ABC

transporter and Wzy polymerase-dependent

mechanisms

Rosanna E. B. Young, Brigitte Twelkmeyer, Varvara Vitiazeva, Peter M. Power, Elke

Schweda and Derek W. Hood

Linköping University Post Print

N.B.: When citing this work, cite the original article.

Original Publication:

Rosanna E. B. Young, Brigitte Twelkmeyer, Varvara Vitiazeva, Peter M. Power, Elke

Schweda and Derek W. Hood, Haemophilus parainfluenzae expresses diverse

lipopolysaccharide O-antigens using ABC transporter and Wzy polymerase-dependent

mechanisms, 2013, International Journal of Medical Microbiology, (303), 8, 603-617.

http://dx.doi.org/10.1016/j.ijmm.2013.08.006

Copyright: Elsevier

Open Access funded by Medical Research Council. Under a Creative Commons license

http://www.elsevier.com/

Postprint available at: Linköping University Electronic Press

http://urn.kb.se/resolve?urn=urn:nbn:se:liu:diva-103880

(2)

ContentslistsavailableatScienceDirect

International

Journal

of

Medical

Microbiology

j ou rn a l h o m e p a g e :w w w . e l s e v i e r . c om / l o c a t e / i j m m

Haemophilus

parainfluenzae

expresses

diverse

lipopolysaccharide

O-antigens

using

ABC

transporter

and

Wzy

polymerase-dependent

mechanisms

Rosanna

E.B.

Young

a,b,∗

,

Brigitte

Twelkmeyer

c

,

Varvara

Vitiazeva

c

,

Peter

M.

Power

b

,

Elke

K.H.

Schweda

c,d

,

Derek

W.

Hood

b,e

aDepartmentofStructuralandMolecularBiology,UniversityCollegeLondon,DarwinBuilding,GowerStreet,LondonWC1E6BT,UnitedKingdom bDepartmentofPaediatrics,UniversityofOxford,UnitedKingdom1

cClinicalResearchCentre,KarolinskaInstitutet,Novum,S-14186Huddinge,Stockholm,Sweden dDivisionofChemistry,IFM,LinköpingUniversity,SE-58183Linköping,Sweden

eMRCHarwell,OxfordshireOX110RD,UnitedKingdom

a

r

t

i

c

l

e

i

n

f

o

Articlehistory: Received2July2013

Receivedinrevisedform9August2013 Accepted18August2013 Keywords: Lipopolysaccharide O-antigen Host–bacterialinteraction Cellsurface Pasteurellaceae

a

b

s

t

r

a

c

t

LipopolysaccharideO-antigensarethebasisofserotypingschemesforGramnegativebacteriaandhelpto determinethenatureofhost–bacterialinteractions.Haemophilusparainfluenzaeisanormalcommensal ofhumansbutisalsoanoccasionalpathogen.Theprevalence,diversityandbiosynthesisofO-antigens wereinvestigatedinthisspeciesforthefirsttime.18/18commensalH.parainfluenzaeisolatescontain aO-antigenbiosynthesisgeneclusterflankedbyglnAandpepB,thesamepositionasthehmglocusfor tetrasaccharidebiosynthesisinHaemophilusinfluenzae.TheO-antigenlocishowdiverserestrictiondigest patternsbutfallintotwomaingroups:(1)thoseencodingenzymesforthesynthesisandtransferof Fuc-NAc4NinadditiontotheWzy-dependentmechanismofO-antigensynthesisandtransportand(2)those encodinggalactofuranosesynthesis/transferenzymesandanABCtransporter.Theother glycosyltrans-ferasegenesdifferbetweenisolates.ThreeH.parainfluenzaeisolatesfelloutsidethesegroupsandare predictedtosynthesiseO-antigenscontainingribitolphosphateordeoxytalose.IsolatesusingtheABC transportersystemencodeaputativeO-antigenligase,requiredforthesynthesisofO-antigen-containing LPSglycoforms,ataseparategenomiclocation.ThepresenceofanO-antigencontributessignificantlyto H.parainfluenzaeresistancetothekillingeffectofhumanseruminvitro.ThediscoveryofO-antigensin H.parainfluenzaeisstriking,asitscloserelativeH.influenzaelacksthiscellsurfacecomponent.

© 2013 Elsevier GmbH. All rights reserved.

Introduction

TheO-antigen(OAg)componentofcellsurface lipopolysaccha-ride(LPS) isone of themostdiversestructures foundinGram negativebacteria,differing bothwithin andbetweenspecies. It isthebasisof typingschemes formanybacterialspecies,using antiseraraisedspecificallyagainsteachOAgstructuretotestfor reactivity.InsomecasesacorrelationcanbeseenbetweenOAg serotypeand clinicalsymptomsdue tothenumerousrolesthat OAgplays in themodulationof bacterial–hostinteractions. For Escherichiacoli,over170differentOAgstructureshavebeen iden-tified(Lundborgetal.,2010):eachcontainsonetosevensugars

∗ Correspondingauthorat:DepartmentofStructuralandMolecularBiology, Uni-versityCollegeLondon,DarwinBuilding,GowerStreet,LondonWC1E6BT,United Kingdom.Tel.:+4402076792676.

E-mailaddress:r.young@ucl.ac.uk(R.E.B.Young).

1 Formeraffiliation.

perrepeatunit(O-unit),withextravariationaddedbydifferent sugarconformations,linkages,branchingpatternsand modifica-tions.Otherspeciesaremoreconservative;forexample,thereare only sevenknownserotypes forAggregatibacter actinomycetem-comitans(Kaplanetal.,2001;Takadaetal.,2010).

Haemophilusparainfluenzaeisapartofthenormalfloraofthe humanupperrespiratory tractbut hasalsobeen isolated occa-sionallyfromanincreasingnumberofdiseasesituationsincluding meningitis, septicaemia, pleural effusion, urethritis, prosthetic jointinfection,anabscessfollowingreconstructionforfacial paral-ysis,and endocarditisin patients withand without underlying heart disease (Bailey et al., 2011; Blacket al., 1988; Cremades et al.,2011;Darras-Joly etal.,1997; Leeet al.,2012;Lin etal., 2012;Sturm,1986).Werecentlyshowedthatincontrasttothe closely relatedspecies Haemophilusinfluenzae, H.parainfluenzae doesnotphasevarytheexpressionofitscoreLPScomponentsby thetetranucleotiderepeatmediatedslippageofLPSbiosynthesis genes(YoungandHood,2013)andatleastonestrainexpresses polymeric OAg (Vitiazeva et al., 2011). The latter observation

1438-4221/$–seefrontmatter © 2013 Elsevier GmbH. All rights reserved.

(3)

604 R.E.B.Youngetal./InternationalJournalofMedicalMicrobiology303 (2013) 603–617

concurswiththefindingsofRobertsandcolleagues(1986)that some(8/25)H.parainfluenzaeisolatesgiveladder-likeLPSprofiles usingsilver-stainedSDS-PAGE,suggestiveofmoleculescontaining OAgsof differentchainlengths. AsH.parainfluenzae OAgs have neverbeenstudiedindetail,thenumberofserotypesisunknown andnoantiserumisavailabletotestforparticularOAgstructures. TheaimofourresearchwastodeterminewhetherallH. parain-fluenzaestrainscontainsthegenesnecessaryforOAgproduction, howthe OAgsof differentstrainsare related,and whetherthe OAgsplayaroleinbacterial–hostinteractions.

ThemechanismsofOAgbiosynthesisinotherspecieshavebeen wellcharacterised.WhereascoreLPSoligosaccharidesare assem-bledontolipidA-Kdothroughthesequentialtransferofeachsugar fromits nucleotide sugar precursor, the OAg polysaccharide is alwaysaddedenbloc.Anundecaprenylphosphate(UndP)-sugar phosphotransferasetransfersthefirstsugarof theOAgontoan UndPcarrierlipid,andfurtherglycosyltransferaseenzymesaddthe subsequentsugarsfromtheirnucleotidesugarprecursors.Oneof twoalternativemechanismsisusuallythenusedtopolymeriseand translocatetheunits(reviewedbySamuelandReeves(2003)).In theWzy-dependentsystem,theOAgflippaseenzyme(Wzx)flips individualUndP-linkedO-unitsfromthecytoplasmicfacetothe periplasmicfaceoftheinnermembrane.Theunitsarethen poly-merisedbytheOAgpolymerase,Wzy,andtheresultingOAgchain isligatedtotheLPScorebytheOAgligase,WaaL.Inthissystem, themodalchainlengthisdeterminedbyafourthenzymenamed Wzz.ThealternativesystemrequiresanABCtransporter compris-ingtwopermeasesubunits(Wzm)fortranslocationandtwoATPase subunits(Wzt)todrivetheprocess.Inthis casetheentireOAg chainisassembledonthecytoplasmicfaceoftheinnermembrane usingglycosyltransferasesbeforeitstranslocationtothe periplas-micside.TheOAgisthenligatedtotheLPScorebyWaaLasbefore.It isnotknownwhetherH.parainfluenzaeusesoneofthesecommon mechanismsforOAgbiosynthesis.

TheenzymesrequiredforOAgsynthesisandassemblyare usu-allyencodedbyadistinct,co-regulatedgeneclustertermedtheOAg locus.ThecombinationofOAgenzymesexpressedbyaparticular bacteriumdeterminesthenature,orderandlinkagesofthesugars initsO-unit,soanalysisofLPSbiosynthesisgenescangreatlyaid predictionoftheOAgstructure.ThegeneticsofOAgbiosynthesis inH.parainfluenzaehaveneverbeeninvestigated.

InthispaperweidentifyanOAglocusinthecompletegenome sequenceofoneofourH.parainfluenzaecarriageisolates,strain T3T1.Investigationofthesameregionofthegenomein17other diverseH.parainfluenzaecarriageisolatesusinglongrangePCRand DNAsequencingrevealsthatthepresenceofanOAggenecluster appearstobeaubiquitousfeatureofthisspecies.SomeOAggenes couldalsobeamplifiedfromtwo‘hybrid’strainsincludedinour analyses;thesetwoisolateshavecharacteristicsofbothH. parain-fluenzaeandH.influenzae(Poweretal.,2012;Young andHood, 2013).FunctionalstudiesindicatearolefortheOAginthe interac-tionbetweenH.parainfluenzaeandhostcellsorcomponentsofthe immunesystem.ThisstudyofcommensalH.parainfluenzaeOAg lociandthecorrespondingOAgstructuresalsolaysthe ground-workforfutureserotypingandgenotypingclassificationschemes thatwouldenableresearcherstoassessthedistributionofdisease isolatesacrosstherangeofOAgsfoundincarriagestrains.

Materialsandmethods

Haemophilusstrainsandculture

TheH.parainfluenzaeandHaemophilushybridstrainswere iso-latedfromthethroatsofhealthychildrenintheUKandTheGambia andhavebeennumberedforconvenience;fullstrainnamesare

given in Table S7. Strains were grown in brain heart infusion broth(BHI)(Merck)supplementedwith2␮g/mlNADand incu-batedat37◦C for16hshaking at200rpm. Forgrowthonsolid medium,strainswereplatedonBHIagar(1%)supplementedwith 10%Levinthalsbase(McLinnetal.,1970),whichprovidesNAD,and incubatedat37◦Cfor24h.

HaemophilusgenomicDNA(gDNA)extraction

Bacteriafrom3mllogphaseculturewerepelletedby centrifug-ingat13,000×gfor2minthenwashedinPBSandresuspended in200␮lTNE(100mMNaCl,10mMTrispH8,10mMEDTA).SDS wasaddedto1%.Cellswerelysedat65◦Cfor10minthentreated withproteinaseK(500␮g/ml)at37◦Cfor2h.Thesamplewasthen mixedwith1volphenolandcentrifugedat13,000×gfor5min;the toplayer(containingDNA)wastakenintoafreshtubeandmixed with1volphenol/chloroform/isoamylalcohol(25:24:1).After cen-trifugation(13,000×g for 5min)thetop layerwasagaintaken intoafreshtubeandtheDNAwasprecipitatedwith2volethanol and 0.1vol3M NaAc.The DNA waspelleted by centrifugation (13,000×gfor10min)andwashedin70%ethanol,thenairdried andresuspendedin200␮lTEbuffer(10mMTrispH8,1mMEDTA) with50␮g/mlRNase.

Standardpolymerasechainreaction(PCR)

Forexpectedproductsizesofupto6kb,50␮lPCRswere pre-paredusing1UTaqDNAPolymerase(Invitrogen).Eachreaction also included 1× PCR Buffer (Invitrogen), approximately 40ng templategDNA,0.4␮Meachprimer(Sigma),0.4mMeachdNTP and2.5mMMgCl2.DNAwasamplifiedfor30cyclescomprising

1mineachofdenaturation(94◦C),annealing(50◦C)andextension (72◦C);forexpectedproductsof>1.5kbtheextensiontimewas increasedto3min.PCRproductswereelectrophoresedon0.8% agarosegelscontaining0.5␮g/mlethidiumbromideat100Vfor 1h, andvisualisedunderultravioletlight. Primersused forPCR analysisarelistedinTableS8.

LongrangePCR(LR-PCR)anddigests

Fifty-microlitre LR-PCRs were performed using the Expand long range PCR kit (Roche) following the manufacturer’s instructions. OAg loci were amplified from H. parainfluen-zae gDNA using primers 5-GAGACTGCGGTAGTCGATCC-3 and 5-CCATCACTTGGTTTGATGCT-3,whicharespecificforthe locus-flankinggenesglnAandpepB,respectively.Anextensiontimeof 15minincycle1,risingto22minbycycle30,wasfoundtobe sufficient for theamplification of productsof upto 20kb. Five microlitresofeachLR-PCRproductwasdigestedwithMfeI(NEB). LR-PCRproductsanddigestswererunon0.7%agarosegelsat20V for36h.

Generalcloningmethods

Restriction enzymes(NEB) and T4 DNA ligase (Roche)were usedasperthemanufacturers’instructions toconstruct recom-binantplasmids.Plasmidswereamplifiedbythetransformationof chemicallycompetentE.coli(Sambrooketal.,1989)andselection onLBagar(1%tryptone,0.5%yeastextract,0.5%NaCl,1%agar) withtheappropriate antibiotic(100␮g/ml ampicillin, 50␮g/ml kanamycin or 300␮g/ml erythromycin) and incubated at 37◦C for24h.Colonieswerepickedandplasmidsextractedusingthe alkalinelysismethod(Sambrooketal.,1989).Thepresenceand orientationoftheplasmidinsertwasdeterminedbydigestionwith appropriaterestrictionenzymes.

(4)

DisruptionoflgtF,waaL,wbaPandwcfSgenes

TodisruptspecificLPSbiosynthesisgenes,thetargetregionof DNAwasamplifiedfromHaemophilusgDNAbystandardPCRand clonedinE.coliusingthepSCAorpSCA-amp-kanvectorsystem (Stratagene)followingmanufacturer’sguidelines.Theplasmidwas cutwitharestrictionenzymespecificforasequenceinthereading frametobedisruptedandadrugresistancecassettewith compat-ibleendswasinserted.Uptakesignalsequences(USS)werealso includedinsomeconstructswiththeaimoffacilitatinguptakeof theplasmidDNAbyH.parainfluenzae.TheinclusionofUSSina plasmidappearstoincreasetransformationratesinH.influenzae (Mitchelletal.,1991)andthesameUSSisalsofoundthroughout theH.parainfluenzaeT3T1genome.PlasmidsaredescribedinTable S10andtheprimersusedintheirconstructionarelistedinTable S11.

TransformationofH.parainfluenzae

Thesuccessofvarioustransformationmethodswasfoundto be highly strain-dependent, and the method used to generate eachmutantisdetailedinTableS9.Eachtransformationwasfirst attemptedwith0.5␮glinearisedplasmidDNAusingthestatic aer-obicincubationmethodofGromkovaandGoodgal (1979),with 20mMMgSO4.Aftershakingat37◦Cfor5h,thetransformation

mixturewasplatedontoBHIwithappropriateantibioticselection (15␮g/mlkanamycinor20␮g/mlerythromycin)andincubatedat 37◦Cfor24h.Putativetransformantcolonieswerecheckedforthe mutantgenotypeusingPCR analysis.Ifthis didnot yield trans-formantsinthedesiredstrainbackground,thesamemethodwas usedbutwiththeDNA sourceas 3␮gchromosomal DNAfrom amutantfromadifferentstrainbackgroundthathad been suc-cessfullytransformed.Toreducetheriskthatextrarecombination eventsbetweenthedonorandrecipientgenomescouldaffect phe-notypicresults,atleastthreeindependentcloneswereanalysed wheneverchromosomaldonorDNAwasusedfortransformation.

The genome sequence strain, H. parainfluenzae T3T1, could notbetransformedusingthestaticaerobicmethodbutan elec-troporation protocol adapted from that of Mason et al. (2003)

wassuccessful.Fiftymillilitresof BHIbrothinoculated withan overnightculturetogiveastartingOD600measurementof0.10was

incubatedat37◦Cwithshaking.Whentheculturehadreachedan OD600of0.35(150min)itwaschilledonicefor30min.All

fur-therstepswerecarriedoutat4◦C.Cellswerepelletedfor10min at4200×gandwashed3timeswith0.5×SG(1×=15%glycerol, 272mMsucrose,pH7.4)toincreasetheircompetence.Afterthe finalcentrifugation,cellswereresuspendedin500␮l1×SG.Forty microlitresofcompetentcellsweremixedwith1␮gcircular plas-midDNAandweresubjectedtoelectroporationat2.5kV,200and 25␮F(BioRadGenePulser),withrecoveryin1mlBHIfor90minat 37◦Cwithshaking.TransformationswereplatedonBHIagarwith antibioticsasdescribedabove.

CloningandsequencingOAgloci

OAglociwereamplifiedfromH.parainfluenzaestrains13,17, 20and30usingLR-PCR(seeabove).Each12–19kbproductwas digestedusingEcoRVandHaeIIIinseparatereactionsthencleaned byethanolprecipitationanddissolvedinH2O.Thedigested

frag-mentswereligatedtoHincII-cut,phosphatase-treatedpBluescript (Stratagene) with T4 DNA ligase (Roche) and cloned in E. coli DH5␣.ColonieswereselectedonLBagar+100␮g/mlampicillin with40␮g/mlX-Gal (5-bromo-4-chloro-3-indolyl-beta-d-galacto-pyranoside)toallowblue/whitescreeningofinserts.Colonieswere picked into 1ml LB+100␮g/ml ampicillinand grown for plas-midextraction.RestrictiondigestionusingEcoRIandXhoIallowed

clonestobecategorisedaccordingtotheirinsertsize;oneclone containing each EcoRV or HaeIII restrictionfragment wasthen sequencedusingprimersM13-for-20(5 -GTAAAACGACGGCCAGT-3)andM13-rev-24(5-AACAGCTATGACCATG-3)whichbindtothe pBluescriptpartofeachconstructandreadintotheinsertregion. TheoverlappingEcoRVandHaeIIIfragmentsequenceswere assem-bledintocontigsusingVectorNTIContigExpress(Invitrogen),with additionalPCRand sequencingacrossthegapsbetweencontigs enablingfullassemblyoftheOAglocisequences.Forcompletionof thestrain30locussequence,asecondroundofcloningwascarried outusingXmnIandSspIfragments,whichwerethensequenced asabove.AllDNAsequencingwascarriedoutbytheWeatherall InstituteofMolecularMedicineSequencingService,JohnRadcliffe Hospital,Oxford,usinganABI-3730DNAanalyserwithBigDye Ter-minatorv3.1(AppliedBiosystems).

TricineSDS-PAGEforvisualisationofLPS

H.parainfluenzaecolonieswereresuspendedinPBStoequalised optical densities (approximately 109cells/ml). The suspensions

werediluted1:1in2×dissociationbuffer(125mMTrispH6.8,20% glycerol,4%SDS,10%mercaptoethanol,0.004%bromophenolblue). FollowingproteinaseKtreatment(50␮g/ml)at60◦Cfor3hand denaturationat100◦Cfor5min,30␮lsampleswerefractionated ontricineSDS-PAGEgels(Lesseetal.,1990).LPSwasvisualisedby stainingwithsilver(GEHealthcare)followingthemanufacturer’s instructions.

Resistancetothebactericidaleffectsofhumansera

ThesurvivalofH.parainfluenzaestrainsinhumanserapooled from15to18donorswasanalysedusingamethodsimilartothat ofHoodetal.(1999).ColoniesgrownonBHIagarweresuspended inPBS-BG(PBS+0.1%glucose(wt/vol)+0.05mMMgCl2+0.09mM

CaCl2),anda1/20dilutionwaspreparedin1%SDS,0.1MNaOHto

measuretheopticaldensityat260nm.Thestartingsuspensionwas adjustedtotheequivalentofOD260=0.8thendiluted1/20,000-fold.

20%pooledhumanserum(PHS)inPBS-BGwasseriallydiluted1:1 acrosssevenwellsofaflatbottomedpolystyrene96wellplateso thateachwellcontained50␮lof20–0.32%PHS;theeighthwell contained50␮lof 20%PHSthat had beendecomplementedby heatingat56◦Cfor30min.Fiftymicrolitresofthediluted bacte-rialsuspension(approximately2000c.f.u.)wasaddedtoeachwell, givingfinalPHSconcentrationsof10–0.16%.After1hat37◦C,25␮l fromeachwellwasspreadonBHIagarandincubatedfor24hat 37◦C.Theresultingcolonieswerecountedtodeterminethelevel ofbacterialsurvivalineachconcentrationofPHS.

Epithelialcellassociationassay

The ability of Haemophilus strains to adhere to an SV40-transformedhumanbronchialepithelialcellline,16HBE14o−,was studiedusingaprotocolsimilartothatofHoodetal.(1999).This celllinehaspreviouslybeenusedtostudyotherrespiratorytract bacteria.16HBE14o−cellsweregrowntoaconfluentmonolayerin aflatbottomedpolystyrene96wellplateandwashedthreetimes withHank’sbalancedsaltsolution(HBSS;Gibco).Fiftymicrolitres of4%decomplementedPHSinDulbecco’smodifiedEaglemedium (DMEM;Gibco)wasaddedtoeachwell.BacteriagrownonBHI agarwereresuspendedin1.5mlPBS-BGandlefttosettlefor2min beforethetop1mlwastakenintoafreshtube;thisavoidslarge clumpsofcells.AsuspensionequivalenttoOD260=0.8wasmade

usingthemethodoutlinedinthesectionabove,exceptthatthe cells weredilutedwithDMEM.Fiftymicrolitresofthebacterial suspension(approximately4×107 bacteria)wasaddedtothree

(5)

606 R.E.B.Youngetal./InternationalJournalofMedicalMicrobiology303 (2013) 603–617

cells,andtheplatewasincubatedfor2.5hat37◦C inthe pres-enceof5%CO2.Tomeasureassociation,allwellswerewashedwith

HBSSthreetimestoremovenon-adherentbacteria.1%saponinwas addedtoreleasetheremainingbacteria(10minat37◦C),which werethenseriallydilutedandplatedonBHIagar.Colonieswere countedafterovernightincubation.Associationwasplottedasa percentageof thetotal numberof bacteriapresent ina control monolayerwellafterthe2.5hincubationperiod(i.e.adheredplus non-adheredbacteria).

Tomeasurebacterialuptakebytheepithelialcells,monolayers were incubated with the bacteria as above, then the medium wasreplacedwith250␮l of200␮g/mlgentamicin. Aftera fur-ther1.5hat37◦C thecells werewashed,treated withsaponin and platedas before. Gentamicin killsanybacteriaexposed on thesurfaceoftheepithelialmonolayer(thiswasconfirmedbya sensitivitytest)whilstinternalisedbacteriaareprotectedfromthe antibiotic.

Bioinformaticanalysis

TheArtemisgenomebrowserandannotationtool(Rutherford etal.,2000)wasusedtoexamineHaemophilusgenomesandOAg locussequences.Sequencehomologyanalysiswasperformedusing theNCBIbasiclocalalignmentsearchtool(BLAST)(Altschuletal., 1997)withthedefaultalgorithmparameters.Conservedprotein domainsweredetectedusingCDD(Marchler-Baueretal.,2009). Transmembranedomainswithinproteinswerepredictedusingthe TMHMMV2.0toolatwww.cbs.dtu.dk/services/TMHMM.

Sequencedata

H.parainfluenzae sequence datahave beensubmittedtothe GenBankdatabaseunderthefollowingaccessionnumbers:strain 13OAglocus,KC759394;strain20OAglocus,KC759396;strain 17OAglocus,KC759395;strain30OAglocus,KC759397;strain19 waaL(externaltotheOAglocus),KC416614;strain13waaL (exter-naltotheOAglocus),KC416615;strain19wzz(withintheOAg locus),KC416616.

Results

ManyH.parainfluenzaestrainsexhibitOAg-likeLPSpatterns The presence of OAg was investigated in the LPS of 18 H. parainfluenzaecarriage isolatesfromhealthychildren intheUK andTheGambiaand two‘hybrid’strains(Hy6andHy11)using silver-stainedtricine SDS-PAGEanalysisof proteinaseKtreated celllysates.Evenly-spacedladders ofbandstypicalof a repeat-ingoligosaccharideunit(OAg)weredetectedintheLPSprofiles of13/20(65%)oftheisolates.Itislikelythatthesebands corre-spondtoLPScoreplusOAgofvariedchainlength.Theintensity, apparentO-unitsizeandaveragemolecularweightoftheputative OAgglycoformsarereproducibleforeachstrainunderlaboratory growthconditionsbutvarygreatlybetweenstrains(Fig.1AandFig. S1),suggestingdifferencesinO-unitcompositionandchainlength regulation.

Fig.1.VisualisationofLPSandPCRinvestigationofpotentialOAglociinH.parainfluenzae.(A)ProteinaseKtreatedH.parainfluenzaecelllysateswerefractionatedbytricine SDS-PAGEandsilver-stained.Strainnumbersarelistedaboveeachlane.TheintenselowmolecularmassbandineachlaneisLPSthatdoesnotcontainOAg(LPScoreonly), whilsttheladders/smearsofbandsrepresentLPSelaboratedwithOAgofincreasingchainlength.BandspacingdependsonthesizeoftheO-unit.12.5␮loflysateatOD260=5

(strains13and15)orOD260=10(strains20,30andT3T1)wasloaded.WeakerOAg-likebandingpatternswereobservedforstrains2,8,10,14,16,17,18andHy6.(B)Long

rangePCRproductswereobtainedusingprimerstoglnAandpepB,whichflanktheOAglocus.TheoutsidelanescontainaDNAladderwithsizesasindicated.Numbers aboveeachlaneindicatetheH.parainfluenzaestrainofthetemplategDNA.Toppanel:0.4␮lofeachPCRreactionseparatedbyagarosegelelectrophoresis.Productsofa highmolecularmassarevisibleforall18trueH.parainfluenzaestrains;hybridstrainsHy6andHy11didnotyieldproductsandarenotshown.Lowerpanel:MfeIrestriction digestsofthesamelongrangePCRproducts,withfragments>1kbvisible.

(6)

AllH.parainfluenzaestrainscontainaputativeOAglocus betweenglnAandpepB

Analysis of the genome sequence of one of our OAg-expressing strains, T3T1, the first H. parainfluenzae strain to haveitsgenomesequencedandfullyassembled(WellcomeTrust SangerInstitute,Cambridge)(EMBLaccessionnumberFQ312002), revealsa putative OAgbiosynthesis locuscomprising 16 genes (PARA02720–PARA02870)over16.77kb(99.2%coding).Thisgene clusteris flankedby theglnAand pepBgenes.In many species ofGramnegativebacteriatheOAglocusisatthesamegenomic locationin differentstrains, andindeed thehmg(high molecu-larweight glycoform)locusthat isresponsiblefor theaddition of a single tetrasaccharide unit to the LPS in some H. influen-zaeisolates isalsolocatedbetweenglnAandpepB(Hoodetal., 2004).ToinvestigatewhetherotherH.parainfluenzaestrainscarry anOAglocusatthislocation,longrangePCR(LR-PCR)was car-ried out using gDNA from each of the 20 study strains with primers designed to the H. parainfluenzae T3T1 glnA and pepB genes.

PCRproductsrangingfrom12to19kbinlengthwereobtained for 18/18 true H. parainfluenzae strains (Fig. 1B, upper panel), suggestingthataseriesofgenesconsistentlyfallsbetweenglnA andpepBinthisspecies.Noproductswereobtainedforthetwo hybridstrains(Hy6andHy11).AstheendsofthePCRproducts containpart ofthe flankinggenes, theactualsize of the puta-tiveOAgloci waspredictedtobe10–17kb,withtheT3T1locus amongstthelargest.MfeIrestrictiondigestprofilesofthe18 LR-PCRproductswerealmostallunique,indicating a highlevelof nucleotidedivergencebetweentheloci(Fig.1B,lowerpanel).Only H.parainfluenzaestrains24and31hadidenticalrestriction pro-files.

SequencingoftheputativeOAglocusoffourisolates

The glnA–pepB PCR product from four of the H. parain-fluenzae isolates (strains 13, 17, 20 and 30) was digested, cloned using E. coli plasmid vectors and sequenced to investi-gatewhethertheyencodedputativeOAgbiosynthesis enzymes. The reasons for selecting these strains are detailed later. Each assembled sequence comprised 10–14 open reading frames in the glnA to pepB orientation. There is very little inter-genic DNA, suggesting that in general the genes at each locus form an operon in which the genes are co-regulated and co-transcribed.The putativeroleof each encodedprotein inthese four loci and in the H. parainfluenzae T3T1 OAg locus was explored by comparison to homologues of known function, conserved domain searches and in some cases simple ter-tiarystructure modelling.Using these methods it waspossible to predict parts of each OAg structure through bioinformatics alone. The five loci all encoded enzymes with predicted func-tions in nucleotide sugar biosynthesis, sugar transfer and OAg assembly and transport, but detailed analysis predicted highly diversesugarstructures and methodsof assemblyas described below (Fig. 2). New gene names (wajA–wajK) were obtained fromthecuratorsoftheBacterialPolysaccharideGeneDatabase (http://sydney.edu.au/science/molecularbioscience/BPGD/)for11 ofthepredictedglycosyltransferaseandacyltransferasegenes. H.parainfluenzaeT3T1synthesisesatetrasaccharideO-unitusing theWzy-dependentsystem

TheproteinsencodedbygenesPARA02760,PARA02750 and PARA02740 (Fig. 2 and Table S1) are similar to the three enzymesintheproposedpathwayforthebiosynthesisof UndP-linked FucNAc4N in E. coli Sonnei, using UDP-GlcNAc as the

precursor (Xu et al., 2002). This suggested that the first sugar of the H. parainfluenzae T3T1 O-unit could be FucNAc4N (also knownas2-acetamido-4-amino-2,4,6-trideoxygalactoseorAAT),a hypothesisthatwasconfirmedbysubsequentstructuralanalysisof LPScontainingasingleO-unit(datanotshown;Twelkmeyeretal., manuscriptinpreparation).Similargenesarealsofoundin Bac-teroidesfragilis,where FucNAc4Nisthefirstsugaroftherepeat unitforthepolysaccharideA(PS-A)capsule(Baumannetal.,1992; Coyneetal.,2001).FucNAc4Nisararesugarthathasbeenidentified aspartofvariousstructuresinonlyafewotherbacterialspecies todate,namelytheStreptococcuspneumoniaeserotype1capsule (Bentleyetal.,2006),S.pneumoniaeandStreptococcusmitis lipote-ichoicacid(Bergstrometal.,2000;Draingetal.,2006),andOAg orOAg-corelinkerstructuresinE.coliSonnei,Plesiomonas shigel-loides,BordetellaspeciesandProteusvulgaris(Arbatskyetal.,2007; Kenne etal., 1980;Preston etal., 2006;Shepherdet al.,2000). BLASTPsearchesofsequencesavailableforthesespecies(shaded inTableS1)suggestthattheFucNAc4Nbiosynthesis pathwayis highlyconserved,andonemightpredictthatotherbacteriawith thesegenessuchasPorphyromonasendodontalisandFusobacterium nucleatummayalsosynthesiseFucNAc4Naspartofa glycoconju-gate.

The proposed UndP-sugar phosphotransferase PARA02750 sharessomesequencesimilarity(33–37%aaidentity)withthe C-terminalendoftheUndP-Galphosphotransferase(WbaP)enzymes fromSalmonellaandH.influenzae,butasitappearstoaddFucNAc4N ratherthanGalastheinitialsugaroftheO-unitwewillrefertoitas WcfS,aftertheUndP-FucNAc4NphosphotransferasefromB.fragilis (72%aaidentity).InadditiontoPARA02750,whichcontainsa pre-dictedtransmembranedomain,thestrainT3T1OAglocusencodes two putativecytoplasmicglycosyltransferases:PARA02770and PARA02780.Thesearelikelytoaddthesecondandthirdsugars oftheO-unit,knowntobeGalandGalNAcrespectively,butthe sugarspecificityofeachenzymeisunclear.

Severalproteinsencoded by theH.parainfluenzae T3T1 OAg locusare predictedtorelateto themetabolismand transferof sialic acid (Neu5Ac), a common component of OAg and cap-sular polysaccharides (Fig. 2 and Table S1). The substrate for sialicacidadditionisusuallythenucleotidesugarCMP-Neu5Ac, whosebiosyntheticpathwayfromUDP-GlcNAcrequiringthefour enzymesNnaA–NnaDhasbeenwellcharacterisedforE.coli(Vimr et al.,2004).Putative nnaA–nnaD genes arepresent within the T3T1 OAg locus (PARA02830–PARA02860), sharing high levels ofsequencesimilaritywiththegenesforpolysialicacidcapsule biosynthesisinMannheimiahaemolyticaserotypeA2strains(Adlam etal.,1987).Inaddition,PARA02800isaputativeFamily52 glyco-syltransferasethatshares40%aaidentitywithwell-characterised capsule sialyltransferases fromM. haemolytica serotypeA2 and StreptococcusagalactiaeserotypeVIIIstrains. Inagreementwith theseobservations,thefourthsugaroftheH.parainfluenzaeT3T1 O-unit was foundtobe acetylated Neu5Ac(Twelkmeyer etal., manuscriptinpreparation).Theacetylationofthesialicacidresidue mightbecarriedoutbyeitherNnaD(NeuD;PARA02860)orthe putativeO-acetyltransferasePARA02820.

H.influenzaecanalsodecorateitsLPSwithNeu5Ac,whichit obtainsfromtheenvironmentusingatripartiteATP-independent periplasmic (TRAP)transporterencoded by thesiaPand siaQ/M genes(Severietal.,2005).TheabilityofH.parainfluenzaestrain T3T1tosynthesiseNeu5Acobviatestheneedtoimportthissugar, andindeedtherearenosiaPorsiaQ/MhomologuesintheT3T1 genome.

TheWzy-dependentpathwayofOAgassemblyandtransport requires an OAg flippase (Wzx), OAg polymerase (Wzy), chain lengthdeterminant(Wzz)andOAgligase(WaaL).TheH. parain-fluenzaeT3T1 OAglocusappearstoencode enzymeswitheach of these functions (Fig. 2 and Table S1). A conserved domain

(7)

608 R.E.B.Youngetal./InternationalJournalofMedicalMicrobiology303 (2013) 603–617

Fig.2.OrganisationofthefivesequencedH.parainfluenzaeOAgloci.Thestrainnameisgiventotheleftofeachdiagram.EachblockarrowrepresentsanORF,andits predictedfunction(asdiscussedinthetext)isindicatedbyitsshadingandpatternasshowninthekey.Eachlocusencodesoneoftwotransportsystems(Wzy-dependent oranABCtransporter),anUndP-sugarphosphotransferase,andaseriesofenzymesforthesynthesisandtransferofvariousOAgcomponents.Drawntothescaleindicated. GT=glycosyltransferase,OT=O-acetyltransferase.

search predicts PARA02810 to belong to the RfbX family of membraneproteinsinvolvedinOAgexport,whilsttheTMHMM transmembranemodellingalgorithmpredictsthatitcontains12 transmembranedomains, typicalof OAg flippase enzymes.The highestscoringBLASTPmatchesforPARA02790areOAgand cap-sularpolysaccharidepolymerasesfromarangeofbacterialfamilies. OAgpolymerasestypicallyshowlittleaminoacidsimilaritytoeach otherbuttheirtertiarystructureismoreconserved,withatleast10 transmembranedomainsanchoringtheproteinintheinner mem-brane(Kimetal.,2010).TMHMMpredictsPARA02790tocontain 10transmembranehelices.

ThechainlengthdistributionofH.parainfluenzaeT3T1OAgis bimodal,withmostLPSmoleculeshavingeither0–2oraround20

O-units(Fig.1A).ItsOAgchainlengthdeterminantWzz,encodedby genePARA02870,ismostcloselyrelatedtothoseofother Pasteurel-laceaegenerathatproduceOAgincludingseveralMannheimiaand Actinobacillusspecies.ThetopologypredictedforPARA02870by TMHMMisalongperiplasmicloopflankedbytwotransmembrane domains,fittingthestructurethathasbeendeterminedfor sev-eralpolysaccharideco-polymerasesincludingWzzofotherspecies (Moronaetal.,2009).

TheH.parainfluenzaeT3T1OAglocushassomenotable char-acteristics when compared to the genome as a whole. Whilst the average G+C content of the genome is 39.6%, that of the OAg locusis only 32.1%;this low %G+C is typical of OAg loci in Gram negative bacteria. In addition, the 9bp H. influenzae

(8)

uptakesignalsequence(USS)(Redfieldetal.,2006)isdistributed throughout the T3T1 genome at an average frequency of one every1.4kbbutisabsentfromtheOAglocus.Theseobservations supportthehypothesis that H.parainfluenzae hasacquired this regionofDNAthroughhorizontalgenetransfereventsrelatively recently.

TheH.parainfluenzaestrain20OAgcontainsaphosphatelinkage whoseformationiscatalysedbya‘Stealth’protein

Intragenic PCR amplification was performed on gDNA from the 20 study strains to test for the presence of some of the H. parainfluenzae T3T1 OAg locus genes (Table 1). PCR using primersdesignedtotheputativeUDP-GlcNAcdehydratasegene, PARA02740,amplifiedproductsfor7/20strains.Thisindicatedthat severalstrainsmaybecapableofdecoratingtheirLPSwithatleast oneofthesamesugarsasstrainT3T1,namelyFucNAc4N.In paral-lelwiththedeterminationoftheOAgstructurefromoneofthese strains,H.parainfluenzae20(Vitiazevaetal.,2011),wesequenced andanalyseditsfullOAglocus.Thisstrainwaschosenforanalysis duetoitshighlevelofOAgexpressionandbecauseitsshortOAg locusindicatedadifferentgeneticcompositiontothatofT3T1.

Uponassembly,theDNAsequencewasfoundtocomprise10 intactgenesspanning11.7kb(Fig.2).Theproductsofsevenofthese ORFsshare68–99%aaidentitywithproteinsencodedbythestrain T3T1OAglocus(TableS2).Thepredictedchainlengthdeterminant protein,20A,shares83%aaidentitywiththatofH. parainfluen-zaeT3T1,reflectingthesimilarmodalOAgchainlengthofthetwo strains(Fig.1A).However,thepredictedWzx(20B)andWzy(20C) proteinssharesolittleidentitywiththestrainT3T1allelesthat theycannotbealigned;liketheFucNAc4Nbiosynthesisgenesthey arecloselyrelatedtoenzymesfromBacteroidesandmayhavebeen acquiredtogetherthroughhorizontalgenetransfer.

In addition tothe WcfSprotein, which determines thefirst sugaroftheO-unitasFucNAc4N,theH.parainfluenzaestrain20 OAglocusencodestwoothertransferases(20Dand20E).Protein 20Dshares37%aaidentitywithWfgCencodedbytheOAgloci ofE.coliserogroupO152andShigelladysenteriaegroup12.These twoOAgcontainthemoiety ␣-GlcpNAc-(1→P→6)-␣-Glcp(Liu etal.,2008;Olssonetal.,2005),andWfgCispredictedto catal-ysetheunusualphosphodiesterlinkagebetweenthetwosugars (Lundborgetal.,2010).In2005,Sperisenandcolleaguesidentified anovelfamilyofproteinsthatwasconservedacrossmost eukary-otesandsomeprokaryotes(Sperisenetal.,2005).Theytermedit ‘Stealth’becausethebacterialmembersofthefamilyappearedto beinvolved inimmuneevasion,and hypothesisedthatthe pro-teinswerehexose-1-phosphoryltransferases.Whencomparedto theStealthalignmentpublishedbySperisenetal.,itbecomesclear that protein20D belongs tothis family. The conserved regions (CR)whichdefineStealtharepresentatthefollowingpositions withinthe20Dsequence:CR1=aa5–16;CR2=aa40–139;CR3=aa 223–271;CR4=aa308–343.TogetherwithitssimilaritytoWfgC, thisinformationstrengthensthecaseforgene20Dencodinga phos-phoryltransferase(i.e.anenzyme thatconnectstwosugarsvia aphosphodiesterlinkage,asseenin thestrain20 O-unit struc-ture).Thepresenceofthethreetransferasegenesandthelackof sialicacidbiosynthesis/transfergenesinthestrain20OAglocusis consistentwiththeobservedtrisaccharideO-unitstructure,which containsFucNAc4N,glucosephosphateandGalNAc(Vitiazevaetal., 2011).

WehaverecentlyshownthatdeletingwcfSinH.parainfluenzae strain20resultsinboththelossofaladderpatternonthe SDS-PAGELPSprofileandthelossofdetectableOAgusingstructural analysis,confirmingtheinvolvementofthisgeneinOAgproduction (Vitiazevaetal.,2011).

H.parainfluenzaestrain13usesanABCtransportertoadd heteropolymericOAgtoLPS

H.parainfluenzaestrain13waschosenasthenextisolatefor analysisbecauseitsOAghasadifferentchainlengthdistribution andO-unitsizetothatofstrainsT3T1or20(Fig.1A),itsglnA–pepB LR-PCRproductisintriguinglyshort(Fig.1B),anditwasnegativeby PCRamplificationforgenePARA02740suggestinganOAgstructure thatmaynotcontainFucNAc4N.Followingcloningandsequencing oftheglnA–pepBregion,BLASTPsearchesofthe10ORFsfound pre-dictedanalmostentirelydifferentsetofenzymestothoseencoded bytheH.parainfluenzaeT3T1locus,withonlytwogenesbearing anysimilaritybetweenthestrains.However,thestrain13ORFsstill encodetypicalOAgsynthesisproteinsincludingnucleotide-sugar synthesisenzymes,glycosyltransferasesandanOAgtransport sys-tem(Fig.2andTableS3).

Gene13FencodesaproteinwithsignificanthomologytotheGlf familyofUDP-galactopyranosemutases(Pfam03275).Glfcatalyses theconversionofUDP-galactopyranosetoUDP-galactofuranoseso thatgalactofuranose(Galf)canbeincorporatedintostructuressuch asthemycobacterialcellwall(Panetal.,2001)orE.coliOAg(Nassau et al.,1996).Thedifferent conformationsof Galp(6-membered ring)andGalf(5-memberedring)mayconferdifferentbiological propertiesontheresultingstructures;Galfisthermodynamically lessstableandoccursmuchlessfrequentlyinnature.Thepresence ofglfwithintheH.parainfluenzaestrain13OAglocusstrongly sug-gestedthepresenceofGalforaderivativeintheO-unit,andthis wasconfirmedbystructuralanalysis(datanotshown;Twelkmeyer etal.,manuscriptinpreparation).TheGalf-(1,3)-␤-d-GlcpNAc link-agefoundinthedisaccharideO-unitislikelytobeformedbythe glycosyltransferaseencodedbygene13G,asthisistheexact pre-dictedsugarandlinkagespecificityofE.coliWfdJwithwhichit shares36%aaidentity(Lundborgetal.,2010).Thespecificitiesofthe threeotherpredictedglycosyltransferases(WajA,WajBandWciB) andoftheWbaP-likeprotein13Ihavenotbeendetermined.

Thestrain 13OAgstructure alsocontains PEtn andO-acetyl (OAc) substituents. LPSO-acetylationhelpstoconfer resistance toantimicrobialpeptides in somespecies (Gunn, 2001).OAcis likelytobeaddedtoGalfbythepredictedO-acetyltransferaseWajC (Pfam01757),but noPEtntransferasewasidentified intheOAg locus.TheHMGunitinH.influenzaeisdecoratedwithPEtnbyan enzymeencodedoutsidethehmglocus(DerekHood,unpublished observations)andasimilarsituationmayoccurinH.parainfluenzae strain13.InspeciessuchasBordetellabronchisepticaandShigella flexneri,certainOAgsareknowntoundergolatemodificationby OAcorPEtntransferasesafterthechainhasbeentransportedto theperiplasm(AllisonandVerma,2000;Kingetal.,2009).

Strain13evidentlyusesanABC2transporter,ratherthanWzx, totransfercompletedOAgfromthecytoplasmictotheperiplasmic faceoftheinnermembrane.ThepermeaseandATPasesubunits (Wzm and Wzt)are encoded bythe firsttwo ORFs ofthe OAg locusandarecloselyrelatedtoenzymesencodedbypolysaccharide biosynthesisgeneclustersinActinobacilluspleuropneumoniaeand Aggregatibacteraphrophilus,alsomembersofthePasteurellaceae family.ModalchainlengthregulationinABCtransporter depend-entOAgsystemsispoorlyunderstood;insomeE.coliserotypes, chainterminationandtransportoccursuponmethylationofthe terminalsugar (Clarkeet al.,2004).Nomethylatedsugarswere foundintheH.parainfluenzae13OAgbystructuralanalysis (Twelk-meyeretal.,manuscriptinpreparation)andnoneoftheproteins encoded byitsOAglocusispredictedtocontainthecoiledcoil motif that is typicalof methyltransferases,suggesting that the tightOAgchainlengthdistributioninthisstrainiscontrolledby adifferentmechanism.FollowingtheABCtransporter nucleotide-bindingdomainclassificationschemeofCuthbertsonetal.(2010), theH.parainfluenzaestrain13Wztproteinfallsintophylogenetic

(9)

610 R.E.B. Young et al. / International Journal of Medical Microbiology 303 (2013) 603 617 Table1

PCRanalysistotestthedistributionofH.parainfluenzaeOAglocusgenes.PrimersweredesignedtogenesinthesequencedH.parainfluenzaeOAglociandwereusedtotestforthepresenceofeachgeneacrossthestudystrainsby PCRamplification.H.parainfluenzaestrainnumbersarelistedalongthetop,sortedaccordingtotheOAglocusgroupswhichbecameapparentduringthestudy.Theoriginofeachstrainisgivenbelowthestrainnumberaseither TheGambia(G)ortheUnitedKingdom(UK).PCRresultsarescoredaccordingtotheproductsvisiblebyagarosegelelectrophoresis.‘+’indicatesaPCRinwhichaproductoftheexpectedsizewasamplified;‘−’indicatesthatno productwasdetected.PositionsareleftemptywherenoPCRwasattempted.ThennaA–nnaDgenesareusuallyfoundtogether,soprimersweredesignedonlytonnaB(NT=nottested).Forsomegenes,asecondprimerpairwas testedonstrainsthathadgivennegativeresults.Thesealternativeprimersweredesignedtothestrain19orstrain20allele;whenaproductwasobtainedonlyforthesecondprimerpairthisismarkedas‘19’or‘20’,respectively.

Group1OAgloci Group2OAgloci Ungrouped Hybrid

Strain 2 19 20 22 24 T3T1 31 8 10 13 14 15 16 18 35 17 30 34 Hy6 Hy11 Gene Putativefunction Primers UK G UK UK G G G UK UK G UK G UK G UK G G G UK UK H.parainfluenzaeT3T1OAglocus(PARA)

02870 Chainlengthdeterminant P1/P2 − 19 − 19 19 + 19 − − − − − − − − − − − − − 02860 NnaD(NeuD) NT 02850 NnaB(NeuB) P3/P4 − + − + + + + − − − − − − − − − − − − − 02840 NnaC(NeuA) NT 02830 NnaA(NeuC) NT 02820 O-acetyltransferase P5/P6 − + − − − + − − − − − − − − − − − − − − 02810 OAgflippase P7/P8 − + 20 − − + − − − − − − − − − − − − − − 02880 Sialyltransferase P9/P10 − + − − − + − − − − − − − − − − − − − − 02790 OAgpolymerase P11/P12 − + 20 − − + − − − − − − − − − − − − − − 02780 Glycosyltransferase P13/P14 − + − − + + + − − − − − − − − − − − 02770 Glycosyltransferase P15/P16 − + 20 20 + + + − − − − − − − − − − − − − 02760 Aminotransferase P17/P18 + + + + + + + − − 02750 UndP-FucNAc4NP-transferase P17/P19 + + + + + + + − − − − − − − − − − − − − 02740 UDP-GlcNAcdehydratase P20/P21 + + + + + + + − − − − − − − − − − − − − 02730 OAgligase P22/P23 20 20 20 + + + + − − − − − − − − − − − − − 02720 (dTDP-Glcdehyd.)topepB P24/P25 + + + + + + + + + + + + + + + + − − − − Strain13OAglocus

glnAto13A(ABCpermease) P26/P27 + + + + + + + − − − −

13A ABC(permeasesubunit) P26/P28 − − − − − − − + + + + + + + − − − − − −

13B ABC(ATPasesubunit) P29/P30 − − − − − − − − + + − + − + − − − − + −

13C Glycosyltransferase P31/P32 − − − − − − − − − + − + − + − − − − − −

13D Glycosyltransferase P33/P34 − − − − − − − − − + − + − + − − − − − −

13E O-acetyltransferase P35/P36 − − − − − − − − − + − + − − − − − − − −

13F UDP-Galpmutase P37/P38 − − − − − − − + + + + + + + + − − − + +

13G Glycosyltransferase P39/P40 − − − − − − − − − + − − − + + − − − + −

13H Glycosyltransferase P41/P42 − − − − − − − + + + + + + + + − − − + +

13I UndP-sugarP-transferase P43/P44 − − − − − − − + + + + + + + + − − − + +

13Ito13J(dTDP-Glcdehyd.) P43/P45 + + + + + + + + + +

17B OAgpolymerase P46/P47 − − − − − − − − − − − − − − − + − − − −

17C OAgflippase P48/P49 − − − − − − − − − − − − − − − + − − − −

17J Ribose-5Preductase P50/P51 − − − − − − − − − − − − − − − + − − − −

30B Glc1Pthymidylyltransferase P52/P53 − − − − − − − − − − − − − − − − + + − −

(10)

groupD.MembersofthisgrouplackC-terminalextensionsand

thepolysaccharides thattheytransport donottypicallycontain

chain-terminatingmodifications.

MostH.parainfluenzaestrainshaveaT3T1-like(group1)ora

strain13-like(group2)OAglocus

ToinvestigatewhetherotherH.parainfluenzaestrainshave

simi-larOAglocitothethreesequencesthatwerenowavailable,internal

primersweredesignedtoeachgeneintheH.parainfluenzaestrain

13OAglocusandtosomeofthegenesthathadyettobestudied

fromtheH.parainfluenzaeT3T1OAglocus.Thepresenceofeach

genewastestedinthestudystrainsbyPCRamplification(Table1).

Theresultswerestrikinginthattheyimmediatelyseparated15of the18trueH.parainfluenzaestrainsintotwoclearOAglocus cate-gories.Sevenstrains,designatedasgroup1,containedhomologues ofthegenesneededforFucNAc4Nsynthesisandtransferinstrain T3T1(genesPARA02760,PARA02750andPARA02740).Asubset ofthesesevenstrainsalsogavePCRproductsforthetwoputative strainT3T1glycosyltransferasegenesand/orthesialicacid syn-thasegenennaB(neuB).ThennaB-positivestrainscorrespondto thefivelargestLR-PCRproductsinFig.1B.Moststrainsingroup1 alsoshowedevidenceforgenesrequiredfortheWzy-dependent mechanismofOAgsynthesisandtransport(Table1).

AmutuallyexclusivesetofeightstrainswerepositivefortheH. parainfluenzaestrain13UndP-sugarphosphotransferasegene13I, andfor13Hwhichwepredicttoencodeanenzymeinvolvedin thesynthesisortransferofGalf.Alloftheseisolatesexceptstrain 35alsoappeartoencodeasimilarABCtransporter(permease sub-unit)tothatofthestrain13OAglocus.Theeightisolatescontaining thestrain13-likelociweredesignatedasgroup2.Inadditionthe twohybridstrains,Hy6andHy11,shareafewofthegroup2OAg locusgenes.Group1andgroup2OAglociarefoundinboth Gam-bianandUKstrains, butthegroup2 locifromGambianstrains (H.parainfluenzae13,15and18)appeartobeparticularlyclosely related.

OAglocusmappingbyPCRamplificationofvarious combina-tionsofadjacentgenesindicatedthattheorderofthegenesthatare presentisgenerallyconservedwithinthegroups(RY,unpublished data).ItisevidentfromFig.1BthattheOAglociofH.parainfluenzae strains14and16areseveralkilobaseslongerthanthatofstrain13; perhapsthecentralregioncomprisesanovelsetof glycosyltrans-ferasegenesand/orsugarsynthesisgenesinsteadofORFs13Cto 13E.

NostrainsyieldedPCRproductsforbothgroup1andgroup2 genes:althoughstrainsT3T1and13sharethelastORFintheOAg locus(PARA02720/13J)there arenoothergenesincommon,so theexchangeofsectionsoftheOAglocusbetweengroup1and group2strainsbyhomologousrecombinationwouldbeunlikely. Thedifferent assembly systemsused bythe two groups (Wzy-dependentorABCtransporter)wouldalsoimposeconstraintson reassortment,iftheOAglocusistoremainfunctional.

Forthreestrains(H.parainfluenzae17,30and34),noPCR prod-uctswereobtainedusinganyofthegroup-specificT3T1orstrain 13primers.Toinvestigatewhetherthesestrainscontainednovel setsofOAggenes,theglnA–pepBregionsofH.parainfluenzaestrains 17and30weredigested,clonedinE.coliandsequenced.Bothof thesestrainsproduceuniqueOAg-likeladdersvisibleonaheavily loadedSDS-PAGEgel(Fig.1AandFig.S1),suggestingthattheirOAg lociarelikelytobebothnovelandfunctional.

AnalysisoftheH.parainfluenzaestrain17OAglocus

Followingsequenceassembly, theH.parainfluenzaestrain17 glnA–pepBregionwasfoundtocontain14ORFsover14.8kb(Fig.2). ThegenesencodefunctionsthataretypicalofOAgsynthesisand

Wzy-dependentassembly(TableS4).Inbrief,thelocusincludes genesencodingsixputativeglycosyltransferasesandone acyltrans-ferase,suggestinganO-unitcomprisinguptosixsugarsandatleast oneOAcgroup.Twogenesprovideevidenceforthepresenceof GalfintheOAg.Firstly,theproductofgene17Dshares87%aa iden-titywithGlffromS.pneumoniae,whereitispresentinstrainsthat includeGalfintheircapsule.Secondly,gene17Eisaputative homo-logueofpneumococcalwciB,whoseproducthasbeencategorised asaGalftransferaseandalwaysaddsthesugarviaa␤1,3linkage (Aanensenetal.,2007).

17Iand17Jaresimilartotheenzymesrequiredforthe two-stepconversionofd-ribulose-5PtoCDP-ribitol(Bauretal.,2009), which isusedtomake thepolyribitol-phosphate componentof teichoicacidintheGrampositivecellwall.Protein17Fshares61% aaidentitywithWefLinS.oralisstrainsC104andSK144,where itisproposedtotransferribitol-5-phosphatetoGalf(Yangetal., 2009).RibitolphosphateisanunusualOAgcomponentandwould contributenegativechargetotheO-unit.

AnalysisoftheH.parainfluenzaestrain30OAglocus

TheH.parainfluenzaestrain30glnA–pepBregioncontains12 ORFs (Fig. 2).The first 11 of thesecorrespond to themajority ofthe12-geneAggregatibacter (previouslyActinobacillus) actino-mycetemcomitansserotypecOAglocus,with71–94%aaidentity between gene products(Table S5). Thisgreatly aids prediction oftheH.parainfluenzaestrain30OAgstructure,astheA. actino-mycetemcomitansserotypecOAghasbeenstudiedindetailand isknowntocomprise→3)-6-deoxy- ␣-l-talose-(1,2)-6-deoxy-␣-l-talose-(1→withacetylationattheO-4positionofthefirstofthese twosugars(Nakanoetal.,1998;Shibuyaetal.,1991).

Overall, the O-unit is likely tobe very similar to that of A. actinomycetemcomitansserotypec,includingoneormore 6-deoxy-l-talopyranoseresiduesandsomedegreeofacetylationcarriedout bytheproductofgene30I.TheOAgmayalsocontainl-rhamnose, whichdiffersfrom6-deoxy-l-taloseonlyinthestereochemistryof theC4carbon.6-deoxy-l-talosehaspreviouslybeenfoundinthe OAgofthreeE.coliserotypes(Jannetal.,1995)andthe Mesorhizo-biumlotitypestrain(Russaetal.,1995).

Primers designed to the strain 30 genes 30B (rham-nose/deoxytalose synthesis pathway gene) and 30E (ABC transporter permease subunit gene) amplified PCR products from H. parainfluenzae strain 34 gDNA (Table 1), and theMfeI digestionpatternoftheLR-PCRproductfromthisstrainincludes a fragment of approximately the same molecular mass as the 30G–30LfragmentfromH.parainfluenzae 30 (6.5kb;Fig.1B).It thereforeseemslikelythatstrain34containsanOAglocusofa broadlysimilarcompositiontothatofstrain30,althoughnoOAg hasbeenobservedfortheformerstrain.Theseresultsmeanthat putativeOAgbiosynthesisgeneshavenowbeendetectedinall20 studystrains.

RecommendedprimersforthecategorisationofH.parainfluenzae OAgloci

ThediversityofOAggeneswithinthe20strainsofthestudy panelissogreatthatitisnotpracticaltodevelopaserotype nam-ingsystematpresent.However,itmaybeusefulforresearchersto categoriseclinicalisolatesbroadlybyOAggroupand/orto com-paretherestrictionpatternsoftheirOAglocitoexaminevirulence trends.Group1 andgroup2 OAgloci maybedistinguishedby twosimple polymerasechainreactionsusinggDNA;we recom-mendprimerpairP20/P21,whichamplifiesa656bpfragmentof aFucNAc4Nbiosynthesisgeneingroup1loci,andP37/P38,which

(11)

612 R.E.B.Youngetal./InternationalJournalofMedicalMicrobiology303 (2013) 603–617

amplifya537bpfragmentofaGalfbiosynthesisgeneingroup2 loci.PrimersequencesaregiveninTableS8.

IdentificationofanOAgligasegenebetweenfbaandorfH,outside theOAggenecluster

ThetransferofOAgfromtheUndPcarriertocoreLPSisusually carriedoutbyanOAgligase.Whilstallsevengroup1H. parainfluen-zaeOAglociincludeaputativeOAgligasegene(Table1),thegroup 2OAglocusofstrain13andtheungroupedOAglociofstrains17 and30donot.Wethereforeinvestigatedwhetheranyofthe unse-quencedH.parainfluenzaeOAglocicontainanOAgligasegeneina particularpositionandwhetherstrain13orotherstrainsencode anOAgligaseelsewhereinthegenome.

IntheH.influenzaehmglocus,theORFencodingtheHMGligase is at the distal end between HI0873 (annotated as rfbBbut of unknownfunction)andthelocus-flankinggenepepBandis con-vergentinorientationwiththerestofthelocus(Hoodetal.,2004). PCRanalysisdemonstratedthataligasegeneisnotfoundbetween rfbBandpepBinanyofour18H.parainfluenzaestrains(Fig.S2).

TheH.parainfluenzaestrain13genomehasnotbeensequenced, butlimitedsequencedataisavailableforstrain15(Poweretal., 2012), whichproducesOAgwithasimilarPAGEprofileandhas a similar OAg locus to strain 13 (Fig. 1A and Table 1). Using BLASTPweidentifiedaDNAsequencefragmentfromstrain15with similaritytothe3 end oftheH.parainfluenzaeT3T1OAgligase genePARA02730.PCRamplificationandfurthersequence analy-sisrevealedthattheORFwasflankedbytheLPSHepItransferase geneorfHandthefructosebis-phosphatealdolasegenefbainstrain 15;thesetwogenesareadjacentinstrainT3T1(PARA08430and PARA08440,respectively).

PCR analysiswas performedto test which H. parainfluenzae studystrainscontainastrain 15-likewaaL gene.Using internal strain15waaLprimers,a730bpproductwasamplifiedfrom13of the20studystrains(Fig.S3).ThesecomprisedalleightH. parain-fluenzaestrainswithgroup2OAgloci,thethreeungroupedstrains (17,30and34),andtwostrainswithgroup1OAgloci(2and19). PCRproductsobtainedusingprimersdesignedtofbaandorfHwere consistentwiththepresenceofaninterveninggene(i.e.waaL)at thislocusinthesame13strains(Fig.S3).

Fulllengthstrain15-likewaaLDNAsequenceswereobtained forstrains13and19andareavailableviaGenbank.When trans-lated,thesesequencessharemorehomologywithWaaLfromA. aphrophilus(57%aaidentity)thanwiththeH.influenzaeRdHMG ligase, HI0874 (42%) or the H. parainfluenzae T3T1 OAg ligase, PARA02730(34%).TMHMManalysispredictsthatthestrain13 WaaL contains 12 membrane-spanning domains and an 89 aa periplasmicloop.ThistopologyistypicalforanOAgligase:the periplasmicloopregionsofSalmonellaentericasv.Typhimurium andE.coliWaaLare predictedtobe73 and84 aa,respectively (AbeyrathneandLam,2007).

AnH.parainfluenzaestrain15mutantinwhichthenewly dis-covered OAgligase gene wasdisrupted didnot synthesise any OAg-containingglycoforms(Fig.3A),consistentwiththe hypothe-sisthatthisgeneencodestheligaseresponsiblefortheadditionof theOAgtotheLPScoreinthisstrain.

WhilstitisreassuringtoidentifytheOAgligasegeneinthe group2andungroupedH.parainfluenzaestrains,itspresencein strains2and19wasunexpectedasthesealsoincludegenesrelated toH.parainfluenzaestrain20waaLintheirOAgloci(PCRanalysis,

Table1).Apossiblescenarioisthatthesetwolineagesrecently exchangedagroup2locusforanewgroup1locus,acquiringan extraOAgligasegene intheprocess, andhave notyetlostthe originalligasegene.Alternatively,onemightbeapseudogene,or thetwoligasesmightactupon differentdonor(polysaccharide) and/or acceptor (LPS/protein) molecules in these strains. OAg

Fig.3.PhenotypicanalysisofH.parainfluenzaeOAgmutantsderivedfromstrains withgroup1orgroup2OAgloci.(A)LPSprofilesofH.parainfluenzaestrain15 anditslgtF,waaLandwbaPmutants,showingthelossofOAginallthreemutants. 12.5␮lofproteinaseKtreatedcelllysatesatOD260=1wereseparatedbytricine

SDS-PAGEandsilver-stained.OnlytheOAgregionofthegelisshown.(B)Resistanceof H.parainfluenzaestrain15(group2)anditsOAgmutantstothekillingeffectof pooledhumanserum.Resultsareshownasthesurvivalofinoculatingbacteriaasa percentageofthesurvivalina10%decomplementedserumcontrolwell.Eachdata pointrepresentsthemeanofthreereplicates;errorbarsshow±standarderrorof themean.(C)ResistanceofH.parainfluenzaestrain20(group1)anditsUndP-sugar phosphotransferasemutantstothekillingeffectofpooledhumanserum.Results areshownasthesurvivalofinoculatingbacteriaasapercentageofthesurvival ina10%decomplementedserumcontrolwell.20.18and20.19areindependent transformantsmadeusingthesameparentalstrainandplasmid.Eachdatapoint representsthemeanoftworeplicates.

ligasehomologuesareknowntoadd O-linkedsugarstocertain cellsurfaceproteinsinsomeotherspecies,e.g.pilinglycosylation inNeisseriameningitidis(Poweretal.,2006).

TheO-antigenconfersresistancetocomplement-mediatedkilling HavingestablishedthatalloftheH.parainfluenzaestrainstested containanOAggenecluster,weinvestigatedthebiologicalroles oftheO-antigensusingseveralinvitroassaysthatareproxiesfor aspectsofhostinteractions.Serumisolatedfromhumanblood con-tainscomplementcomponentsandsomeantibodies,sotheserum bactericidalassaycandetectbothclassicalandalternative comple-mentactivation.Itprimarilymeasureskillingthatismediatedby MACformation,becauseserumdoesnotcontainthemacrophages andneutrophilsthatarerequiredforopsonophagocytosis.

TheUndPsugartransferasegenesinH.parainfluenzaestrains 20and 15 (wcfSandwbaP,respectively) weredisruptedwitha

(12)

kanamycin resistance cassette by transformation with plasmid constructs.ThisresultedinthecompletelossofOAgfromboth strain20(Vitiazevaetal.,2011)andstrain15 (Fig.3A), demon-stratingthat therespectiveloci arerequiredfor OAgformation as predicted.Elimination of theOAg by disrupting wcfS, wbaP, waaLorlgtFresultsinadramaticlossofresistancetothekilling effectofcomplementinpooledhumanserumacrossa rangeof H.parainfluenzaestrains,regardlessofwhethertheOAglocuswas categorisedasgroup1orgroup2(Fig.3BandCandTableS6). EpithelialadhesionisaffectedbyextensionsfromHepIoftheLPS

Adherencetohostcellsisoneofthefirststepsrequiredforthe colonisationoftherespiratorytract,allowingbacteriatoestablish themselvesinarelativelyconstantenvironmentwhilstavoiding ciliaryclearance.ForH.influenzae,associationwithepithelialcells isthoughttobeaprerequisiteforinvasivedisease,asthe bacte-riamustmoveeitherthroughorbetweenthecellsoftheepithelial barriertoreachthebloodstream.TwoOAg-expressingH. parain-fluenzaestrainswerecompared to H.influenzae regarding their abilitytoinvadehumanbronchialepithelialcellsinvitroduring a2.5hincubation.Thepercentageoftotalbacteriathatwerefound insideepithelialcellswas1.9×10−5%forH.parainfluenzaestrain 13,3.3×10−5%forH.parainfluenzaestrain20and9.3×10−5%for H.influenzaestrainRd(Fig.4A).Thisdataisconsistentwiththe hypothesisthatofthetwospecies,H.parainfluenzaeislessableto invadeepithelialcellssomaybelesslikelytobehave pathogeni-cally.

ToexaminetheroleofLPSstructureinH.parainfluenzae adhe-sion, association assays were performed using lgtF mutants of strainswithgroup1orgroup2OAg(Fig.4BandC).Intriguingly, theeffectofablatingtheadditionofOAgandGlctoHepIofthe LPS(i.e.mutatinglgtF)wasstrain-dependent.TheH. parainfluen-zaeT3T1lgtFmutant,T3T1.2,showed15timesgreaterassociation withepithelialcellsthanwildtypeH.parainfluenzaeT3T1, suggest-ingthattheOAgimpairsadherence.ForH.parainfluenzaestrain13, whichhasadifferentOAgstructure,thelgtFmutantappearedto adherelessthanthewildtypestrain.OAgthereforeappearstoplay oppositerolesinadhesioninthetwostudystrains(seeDiscussion).

Discussion

LPSisoneofthemainstructuralcomponentsoftheouter mem-braneinGramnegativebacteria,andinmanyspeciestheexposed positionofOAgmakesitakeydeterminantofinteractionswiththe host.Wehavedemonstratedthatallof18commensalH. parain-fluenzaeisolatescontainaclusterofgenesrelatedtopolysaccharide synthesisandtransport flankedbyglnAandpepB.Thisprovides strongevidencethatthereispositiveselectioninH.parainfluenzae fortheabilitytosynthesiseapolysaccharidestructureonits sur-faceviaanUndPcarrier.MutationoftheUndPsugartransferase geneinseveralstrainsconfirmedtheinvolvementofthelocusin LPSOAgandO-unitsynthesis.

WehavedemonstratedthatitispossibletopredictsomeOAg structuraldetailsfromthegenespresentinanOAglocus.Thishas allowedustogenerateanoverviewofO-unitdiversitywithinH. parainfluenzaewithoutneedingtoanalyseeverystructureusing chemicalmethods.Manyglycosyltransferasegenesfoundinthe OAglocidonothaveclosehomologuesinotherspeciesintheNCBI databaseandmayhavenoveldonorsugar, acceptorsiteand/or linkagespecificities,makingthempotentiallyinterestingto gly-cobiologistsattemptingtosynthesiseparticularoligosaccharides (Patonetal.,2000)orcreatingpanelsofglycansfordrug discov-eryscreens.ThesugarspredictedorproventoformpartoftheseH. parainfluenzaeOAgstructuresincludeGalp,Galf,GalpNAc,GlcpNAc,

FucpNAc4N,Neu5Ac,ribitolanddeoxytalose,withOAcandPEtn additions.

TherewasanapproximatelyequaldistributionacrosstheH. parainfluenzaestrainsofthetwomostcommonOAgsynthesisand transport systems. Althoughmany bacterial speciescomprise a mixtureofWzy-dependentandABCtransporter-dependentOAg serotypes,itisunusualforthetwotypesofOAglocitomaptothe samegenomiclocationastheydoinH.parainfluenzae.Thisensures thatonlyoneOAglocusispresentineachstrain.TheuseofanABC transporterforheteropolymericOAgexport,asseeninH. parain-fluenzaestrain13,israrebutnotunprecedented(Perepelovetal., 2009;Xuetal.,2010).

AmongsttheH.parainfluenzaestrainstestedwefoundno corre-lationbetweenthetypeofOAglocusandthepresenceofparticular H.influenzae-likeoutercoreLPSbiosynthesisgenesthatwehad previouslyidentified(YoungandHood,2013)suchaslpsB,losB1 andlic2C.Thisemphasisesthehighdegreeofgeneticexchangethat mustoccurinH.parainfluenzae,particularlyinthegenesrequired for OAg synthesis. It also increasesthe likelihood that a strain colonisinganewhostexpressesanLPSstructurethathasnotbeen encounteredbythehost’simmunesystembefore,thereby extend-ingtheaveragelengthofcolonisationwhilstspecificantibodiesare produced.Indeed,ifwecombineourgeneticdataforoutercoreand OAg-relatedgenesandassumethatallstrainscanexpressatleast oneO-unitinvivo,around16ofthe18H.parainfluenzaestrainsare predictedtoexpressuniqueLPSstructures.

The diversity of H. parainfluenzae OAg structures contrasts stronglywiththeH.influenzaeHMGunit,forwhichthebiosynthetic enzymesarealsoencodedbetweenglnAandpepBinavestigialOAg locus;60%ofNTHistrainssynthesisethesametwonon-polymeric tetrasaccharideHMGstructures,andtheremaining40%donothave thehmglocus(Hoodetal.,2004).AstheH.influenzaeoutercore structureishighlyvariablewithinandbetweenstrainsduetophase variation,therequirementtodisplaydiverseOAgsforimmune eva-sionislikelyreduced.Severalothermucosalpathogens,including N.meningitidis,BordetellapertussisandCampylobacterjejuni,also lackOAgontheirLPS.

ThegenomesofoneisolateeachofHaemophilussputorumand Haemophilushaemolyticushaverecentlybeensequencedaspart oftheNIHHumanMicrobiomeProject(Petersonetal.,2009).The structure andbiosyntheticgenesofLPShave neverbeen inves-tigatedinthesespecies.BLASTPanalysis(RY,unpublisheddata) revealsthatbothgenomescontainanapparentintact polysaccha-ridebiosynthesislocusremarkablysimilartotheH.parainfluenzae strain13OAglocus,withhomologuesofwzm,wzt,glf,wfdJ,wciB, rfbBandwbaP.IntheH.sputorumCCUG13788locus,genes13Cto 13Earereplacedbyasingleglycosyltransferasegene(Fig.S4)and noligasegeneispresent.InH.haemolyticusHK386thelocus con-tainsaconvergentOAgligasegene(waaL)with78%aaidentityto theligasegene(HI0874)thatis foundinthesamepositionand orientationintheH.influenzaeRdHMGlocus(Fig.S4).Theresults ofourearlierBLASTanalyses(TablesS1andS3)suggestthat dif-ferentH.haemolyticusstrainscarryH.parainfluenzae-likegroup1 andgroup2OAg,respectively.Futureelectrophoreticand struc-turalLPSanalysesoftheseandotherHaemophilusspeciesmaywell revealthatOAgiswidespreadacrossthegenus,withH.influenzae anexceptiontotherule.

It is clear that other bacteria found in the human respira-torytractareanimportantsourceofOAgsynthesisgenesforH. parainfluenzae,althoughthedirectionofindividualhorizontalgene transfereventsisdifficulttoascertain.Geneticexchangedoesnot appeartobelimitedtogenesfromrelatedspeciesorthoseinvolved inOAgsynthesis,assomeoftheclosestpotentialhomologuesfound toH.parainfluenzaeOAglocusgeneswerecapsuleandteichoicacid synthesisgenesfromtheGrampositiverespiratorytractspecies S. pneumoniaeand Gemella haemolysans.As H.parainfluenzae is

(13)

614 R.E.B.Youngetal./InternationalJournalofMedicalMicrobiology303 (2013) 603–617

Fig.4.InteractionsbetweenepithelialcellsandHaemophilusbacteriainvitro.(A)UptakeofHaemophilusbacteriabyhumanbronchialepithelialcells.H.parainfluenzae strains13and20andH.influenzaestrainRdwereincubatedina96-wellplatewithamonolayerof16HBE14cells.Theyaxisrecordsthenumberofbacteriathatwereinside theepithelialcellsaftera2.5hincubationasapercentageofthetotalbacteriainacontrolwell.Eachbarrepresentsthemeanvalueforthreereplicates;theerrorbarsshow ±standarderrorofthemean.(B)LPSprofilesoftwoH.parainfluenzaeisolatesandtheirlgtFmutants.12.5␮lofproteinaseKtreatedcelllysatesatOD260=5(leftpanel)or

OD260=1(rightpanel)wereseparatedbytricineSDS-PAGEandsilver-stained.Labelsaboveeachlanearethewildtypestrain(bold)orlgtFmutantclonenumbers.LgtFisthe

glucosyltransferaseresponsiblefortheadditionofGlctoHepI,andastheGlcistheOAgattachmentpoint,lgtFmutantslackthisGlcresidueandOAgfromtheLPS.TheT3T1 andT3T1.2profilescontainseveralproteinbandsthatcouldnotberemovedbyproteinaseKtreatment.(C)AssociationofH.parainfluenzaestrainsT3T1and13andtheirlgtF mutantswithhumanbronchialepithelialcells.Bacteriawereincubatedina96-wellplatewithamonolayerof16HBE14cells.Theyaxesrecordthenumberofbacteriathat wereassociatedaftera2.5hincubationasapercentageofthetotalbacteriainacontrolwell.Eachbarrepresentsthemeanvalueforthreereplicates;theerrorbarsshow ±standarderrorofthemean.NotethatbothlgtFmutantsshowapproximatelythesamelevelofassociationtoepithelialcells.

reportedtocolonisethehumandigestivetractandcanbecultured fromfaecalsamples(Palmer,1981),geneticexchangewithspecies suchasB.fragilisandShigelladysenteriaemayalsooccur:thiswas reflectedin theBLASTanalysis.OAgexpressionmayin factaid colonisationofthedigestivetractbyH.parainfluenzaeasitdoes forotherbacteria(Fowleretal.,2006).

Allgroup2andungroupedH.parainfluenzaestrainswerefound toencodeaputativeOAgligaseoutsidetheOAglocus,between thegenesfba and orfH.Thiswas shownto berequiredfor the synthesisofOAg-containingLPSglycoformsinstrain15.A homo-logueofthisligasegeneisalsopresentnexttofbainthegenome sequencesofseveralotherPasteurellaceae,includingspeciesofthe

generaActinobacillus,Aggregatibacter,PasteurellaandMannheimia, suggesting thatit is eitherancestral orhasspread throughthe familyverysuccessfullybylateralgenetransfer.WhenTangand Mintz(2010)interruptedtheligasegenenexttofbainA. actino-mycetemcomitans,theyobservednotonlythelossofOAgfromthe LPSprofilebutalsoashiftintheelectrophoreticmobilityofthe adhesinEmaA,whichisusuallyglycosylatedwithwhatisthought tobethesame‘OAg’.Thisglycosylationstepwasimportantfor thestabilityofEmaA(Tangetal.,2012).ItispossiblethattheH. parainfluenzaehomologueisalsobifunctionalandcould glycosy-latebothLPSandcertainoutermembraneproteinswithidentical polysaccharides.

(14)

H.parainfluenzaemayconfersomeprotectionagainst colonisa-tionwithmorepathogenicspecies(VanHoogmoedetal.,2008)but sinceitisalsooccasionallyimplicatedindisease,opinionislikely tobedividedoverwhetheritwouldbedesirabletoretainorreduce colonisationlevels.Inanycaseitisimportantthatweunderstand thepotentialimpactofH.influenzaevaccinationprogrammeson therespiratoryflora.Wefoundcapsulesynthesisgenesinonlyone ofour18H.parainfluenzaeisolates(strain18;unpublisheddata) andthesugar-specificgeneswerenotH.influenzae-like,so capsule-basedH.influenzaevaccinesareunlikelytotargetH.parainfluenzae. RegardingpotentialNTHivaccines,thosebasedoncell-surface pro-teinsmaytargetbothspecies,whilstthosebasedontheLPSHMG unit(Sundgrenetal.,2010)wouldnotimpactdirectlyonH. parain-fluenzaeastheO-unitstructuresproducedbythisspeciescontain differentsugarandlinkagecombinationstotheHMG.

Thepresence,quantityandstructureofpolysaccharidesonthe bacterialcellsurfacehaveprofoundeffectsuponinteractionswith hostmoleculesandcells.Wedemonstratedthatinlinewith find-ingsinotherspecies(Holzeretal.,2009;Merinoetal.,1992),OAg contributestotheabilityofH.parainfluenzaetoresistthekilling effectofhumancomplement.LongOAgchainscanprevent anti-bodiesthatrecogniseconservedmembraneproteinsfromreaching theirtarget(Russoetal.,2009),whilstantibodiesbindingtotheend oftheOAgchaincantriggerdepositionofthecomplement compo-nentC3awayfromthebacterialmembrane(Goebeletal.,2008). C3depositionontheoutermembraneisrequiredfortheformation ofthemembraneattackcomplex(MAC)andfortargeting bacte-riatophagocytes.Althoughthecomplementsystemisassociated primarilywithserum,complementcomponentsarealsofoundin respiratorymucosa,especiallyduringinflammation(Hallstromand Riesbeck,2010),sotheabilityofH.parainfluenzaetosynthesiseOAg islikelytocontributetoitshighcarriagerates.

ForS.entericasv.TyphimuriumandBurkholderiacenocepacia, thepresence ofparticular OAgs ontheLPShasbeen shownto decreaseassociationwithhostcells(Holzeretal.,2009;Saldias etal.,2009).In casessuchastheseit isusuallypostulatedthat OAgmaypreventattachmentbymaskingsurfaceadhesins;this could indeed be true for H. parainfluenzae strain T3T1, which attachesmore readilytoepithelial cells inthe absenceofOAg. Incontrast,onegroupofOAgstructureshasbeenshownto con-tributetoadhesion.Humanepithelialcellssecreteamultipurpose ␤-galactoside-specificlectincalledgalectin-3whichbindstothe epithelialsurfaceandtheextracellularmatrix(Dumicetal.,2006). LPSstructuresthat containterminal␤-galactoside,suchas cer-tain OAg in Helicobacter pylori and the outer core in Neisseria gonorrhoeae,promotebacterial–hostcelladhesionbybindingto galectin-3(Fowleretal.,2006;Johnetal.,2002).TheOAgofH. parainfluenzaestrain13appearedtopromote,ratherthanreduce, adhesion to epithelial cells. This OAg contains a ␤-galactoside structure;indeed,itsO-unitbackboneispoly-N-acetyllactosamine, which(albeitwithdifferentlinkages)wastheoligosaccharidethat boundgalectin-3moststronglyfromapanelof41potential lig-andsinastudybyHirabayashiandcolleagues(2002).Thisprovides apossibleexplanationastothecontrastingeffectsofdifferentH. parainfluenzaeOAgsonepithelialattachment.

Despite the presence of a potential OAglocus, someof our H.parainfluenzae strainsdidnot synthesiseanydetectableOAg whencultivatedunder standardlaboratoryconditions(37◦C on BHIagar).RegulationofthequantityandlengthofOAgsinresponse toenvironmentalcuessuchastemperature,ironconcentrationor serumconcentrationcanallowbacteriatomodifytheirphysical andimmunologicalpropertiestoaidsurvival(Holzeretal.,2009; Jimenezet al.,2008), and itis likely thatunder certainin vivo conditionsOAgwouldbeupregulatedtodetectablelevelsinthe H.parainfluenzaestrains.However,aswehavenotsequencedall 18OAglociwecannotruleoutthepossibilitythatsomemaybe

non-functionalduetomutations.H.pyloriandB.fragilisuseDNA repeatslippageandinvertible promotersrespectively to phase-varytheirOAgstructures(Cerdeno-Tarragaetal.,2005; Sanabria-Valentin etal.,2007);we foundnoevidenceforeitherof these sequencefeaturesinthefiveH.parainfluenzaeOAglocianalysed.

Insomespecies,particularOAgserogroupsareassociatedwith increasedvirulence:forexample,mostcasesofLegionnaire’s dis-easearecausedbyserogroup1strainsofLegionellapneumophila, whicharespreadacrossdiverselineagesbutsharethesameOAg genes(Cazaletetal.,2008).ThegenomesofthreeclinicalH. parain-fluenzaeisolateshaverecentlybeensequencedaspartoftheHuman MicrobiomeProject.DataextractedfromNCBIshowthatallhave anOAggeneclusterbetweenglnAandpepB.InisolatesHK262(from theurogenitaltract)andHK2019(fromafacialskinabscess),the locusis100%identicalandencodesproteinstypicalofagroupI H.parainfluenzaeOAglocusincludingWcfS,WzzandWaaL.The OAglocussequencedataforH.parainfluenzaeATCC33392,isolated fromasepticfinger,areincompletebutitalsoencodespotential homologuesofthethreeaforementionedproteins.Thisis consis-tentwith,butinnowayproves,thehypothesisthatonlyasubset ofH.parainfluenzaeOAgserotypesiscapableofcausingdisease.

Acknowledgements

This research was supported by grants from the Medical ResearchCouncil,UK.WethankDerrickCrookandAbdelElaminfor providingtheH.parainfluenzaeandhybridstrains,RichardMoxon forhelpfuldiscussions,KathyMakepeaceforassistancewith asso-ciationassays,andJohnFranklandandTimRostronoftheWIMM DNASequencingService.GordonStevensonandPeterReevesofthe BacterialPolysaccharideGeneDatabaseprovidednewgenenames.

AppendixA. Supplementarydata

Supplementarydataassociatedwiththisarticlecanbefound, in the online version, at http://dx.doi.org/10.1016/j.ijmm.2013. 08.006.

References

Aanensen,D.M.,Mavroidi,A.,Bentley,S.D.,Reeves,P.R.,Spratt,B.G.,2007.Predicted functionsandlinkagespecificitiesoftheproductsoftheStreptococcus pneumo-niaecapsularbiosyntheticloci.J.Bacteriol.189,7856–7876.

Abeyrathne,P.D.,Lam,J.S.,2007.WaaLofPseudomonasaeruginosautilizesATPin invitroligationofOantigenontolipidA-core.Mol.Microbiol.65,1345–1359.

Adlam,C.,Knights,J.M.,Mugridge,A.,Williams,J.M.,Lindon,J.C.,1987. Produc-tionofcolominicacidbyPasteurellahaemolyticaserotypeA2organisms.FEMS Microbiol.Lett.42,23–25.

Allison,G.E.,Verma,N.K.,2000.Serotype-convertingbacteriophagesandO-antigen modificationinShigellaflexneri.TrendsMicrobiol.8,17–23.

Altschul,S.F.,Madden,T.L.,Schaffer,A.A.,Zhang,J.,Zhang,Z.,Miller,W.,Lipman, D.J.,1997.GappedBLASTandPSI-BLAST:anewgenerationofproteindatabase searchprograms.NucleicAcidsRes.25,3389–3402.

Arbatsky,N.P.,Kondakova,A.N.,Senchenkova,S.N.,Siwinska,M.,Shashkov,A.S., Zych,K.,Knirel,Y.A.,Sidorczyk,Z.,2007.Structureofanewribitolteichoic acid-likeO-polysaccharideofaserologicallyseparateProteusvulgarisstrain, TG276-1,classifiedintoanewProteusserogroupO53.Carbohydr.Res.342, 2061–2066.

Bailey,C.,Duckett,S.,Davies,S.,Townsend,R.,Stockley,I.,2011.Haemophilus parain-fluenzaeprostheticjointinfection.Theimportanceofaccuratemicrobiological diagnosisandoptionsformanagement.J.Infect.63,474–476.

Baumann,H.,Tzianabos,A.O.,Brisson,J.R.,Kasper,D.L.,Jennings,H.J.,1992. Struc-turalelucidationoftwocapsularpolysaccharidesfromonestrainofBacteroides fragilisusinghigh-resolutionNMRspectroscopy.Biochemistry31,4081–4089.

Baur,S.,Marles-Wright,J.,Buckenmaier,S.,Lewis,R.J.,Vollmer,W.,2009.Synthesis ofCDP-activatedribitolforteichoicacidprecursorsinStreptococcuspneumoniae. J.Bacteriol.191,1200–1210.

Bentley,S.D.,Aanensen,D.M.,Mavroidi,A.,Saunders,D.,Rabbinowitsch,E.,Collins, M.,Donohoe,K.,Harris,D.,Murphy,L.,Quail,M.A.,Samuel,G.,Skovsted,I.C., Kaltoft,M.S.,Barrell,B.,Reeves,P.R.,Parkhill,J.,Spratt,B.G.,2006.Genetic analy-sisofthecapsularbiosyntheticlocusfromall90pneumococcalserotypes.PLoS Genet.2,e31.

Bergstrom,N.,Jansson,P.E.,Kilian,M.,SkovSorensen,U.B.,2000.Structuresoftwo cellwall-associatedpolysaccharidesofaStreptococcusmitisbiovar1strain.

References

Related documents

46 Konkreta exempel skulle kunna vara främjandeinsatser för affärsänglar/affärsängelnätverk, skapa arenor där aktörer från utbuds- och efterfrågesidan kan mötas eller

mer hit, önskar jag att det vore höst så jag finge resa till Stockholm, och när jag väl är där och hösten kommer, önskar jag mig — Nej det skall jag inte tala om.. Men det är

mbbm= äÉîÉä= ÜáÖÜ= ÉåçìÖÜ= íç= éêÉëÉêîÉ= bbisI= ïÜÉêÉ= ~= ÇÉÅêÉ~ëÉ= áå= ëäçéÉ= ï~ë= ëÉÉå= ~ë= ÇÉJ êÉÅêìáíãÉåí= ~åÇ= ~å= áåÅêÉ~ëÉ= ~ë= éçëëáÄäÉ= êÉÅêìáíãÉåíK=

The reaction was quenched by adding Et 3 N, the organic phase was washed with brine, dried by passing through a phase separator filter, and concentrated to dryness.. The mixture

Total CO 2 emission for electric devices: At electricity part, according to information that user have entered, energy consumption for each device was calculated and saved on

The weak relationship between prots and growth can also be observed from the high levels of persistence found in prot rates Mueller (1977), whereas growth persistence is

Show that the uniform distribution is a stationary distribution for the associated Markov chain..

Sulfides occur disseminated or in veins with dark green actinolite or with chlorite (Figure 29). The grain size varies from fine- to coarse-grained for both sulfides