basisforthebroadsubstratespecificityoftwoacyl-CoAdehydrogenasesFadE5frommycobacteria
XiaoboChena,b,JiayueChena,BingYana,WeiZhanga,LukeW.Guddatc,XiangLiua,1,andZiheRaoa,b,d,e
aStateKeyLaboratoryofMedicinalChemicalBiology,FrontiersScienceCenterforCellResponses,CollegeofLifeSciences,NankaiUniversity,300071Tianjin,China;bShanghaiInstituteforAdvancedImmunochemicalStudiesandSchoolofLifeScienceandTechnology,ShanghaiTechUniversity,201210Shanghai,China;cSchoolofChemistryandMolecularBiosciences,TheUniversityofQueensland,Brisbane,4072QLD,Australia;dLaboratoryofStructuralBiology,TsinghuaUniversity,100084Beijing,China;andeNationalLaboratoryofBiomacromolecules,ChineseAcademyofSciencesCenterforExcellenceinBiomacromolecules,InstituteofBiophysics,ChineseAcademyofSciences,100101Beijing,China
EditedbyDinshawJ.Patel,MemorialSloanKetteringCancerCenter,NewYork,NY,andapprovedMay28,2020(receivedforreviewFebruary14,2020)
FadE,anacyl-CoAdehydrogenase,introducesunsaturationtocarbonchainsinlipidmetabolismpathways.Here,wereportthatFadE5fromMycobacteriumtuberculosis(MtbFadE5)andMycobacteriumsmegmatis(MsFadE5)playrolesindrugresistanceandexhibitbroadspecificityforlinearacyl-CoAsubstratesbuthaveapreferenceforthosewithlongcarbonchains.Here,thestructuresofMsFadE5andMtbFadE5,inthepresenceandabsenceofsubstrates,havebeendetermined.Theserevealthemolecularbasisforthebroadsubstratespecificityoftheseenzymes.FadE5interactswiththeCoAregionofthesubstratethroughalargenumberofhydrogenbondsandanunusualπ–πstackinginteraction,allowingtheseenzymestoeptbothshort-andlong-chainsubstrates.Residuesinthesubstratebindingcavityreorienttheirsidechainstomodatesubstratesofvariouslengths.Longercarbon-chainsubstratesmakemorenumeroushydrophobicinteractionswiththeparedwiththeshorter-chainsubstrates,resultinginapreferenceforthistypeofsubstrate.
tuberculosis|mycobacteria|acyl-CoAdehydrogenase|fattyacid
Tuberculosis(TB)isaleadingcauseofhumanfatalities,withnewtherapiesurgentlyneededtobatthisdisease.ThecellwallbiosynthesispathwayinMycobacteriumtuberculosis(Mtb)isawell-establishedsourceofmoleculartargetsfordrugdevelopment
(1).monlyusedanti-TBdrugsareknowntoinhibitthebiosynthesisofcellponents.Forexample,isoniazidandethionamideareinhibitorsofmycolicacidsynthesis,andethambutolandcycloserineareinhibitorsofarabinogalactanandpeptidoglycansynthesis,respectively
(2).Structuresofseveraldrugtargetsthatareinvolvedinthebiosynthesisandtransportofcellponents,suchasarabinosyltransferasesandMmpL3,havebeendeterminedrecently(3,4).ThearabinosyltransferasesEmbA,EmbB,andEmbCareregardedasthetargetsforethambutol.MmpL3isresponsibleforthetransportofmycolicacids.SQ109,apromisinganti-TBdrug,bindsinsideMmpL3.Structuralinformationofdrugtargetswoulddefinitelypromotethestructure-guideddrugdesign. TheM.tuberculosiscellwallposedofathicklipidcoat,withthelipidcontentountingfor∼60%ofthedryweightofthecellwall.ItspresencecontributestothevirulenceofM.tuberculosis(5,6).Previousstudieshaveshownthatpathogenicmycobacteriacanusefattyacidsastheircarbonsource
(7).Significantly,cholesterolalonecanbeusedastheonlycarbonsource(8,9).IntheM.tuberculosisgenome,morethan250genesareidentifiedtobeinvolvedinlipid,fattyacid,andsterolmetabolism(10).ThepathwaybywhichlipidsaremetabolizedisthereforeofgreatsignificanceinunderstandingthegrowthandvirulenceofM.tuberculosisinthehost. Inlipidmetabolismpathways,acyl-CoAdehydrogenase(ACD)introducesunsaturationintofattyacids,withthecofactoracyl-CoAconvertingtoenoyl-CoA(11).Althoughthereare35annotatedfadEgenesinM.tuberculosis,notallarefunctional ACDs.TwoFadEproteinsencodedbyadjacentfadEgenesareheterotetrameric.Inthoseexamples,aheterotetramerisrequiredtobindtheFADcofactor(12).Theexistenceofsixheterotetramers,fadE23-fadE24,fadE28-fadE29,fadE26-fadE27,fadE31-fadE32,fadE31-fadE33,andfadE17-fadE18,encodedbyfadEgeneshavebeenreported.MostoftheseFadEenzymescharacterizedtodateinM.tuberculosisfunctionincholesterolcatabolismandplayrolesinthedehydrogenationofcholesterolsubstratesthroughβ-oxidation(13).FadE28-FadE29catalyzesthedehydrogenationof3-oxo-4-pregnene-20-carboxyl-CoA(3-OPC-CoA)(14),whileFadE26-FadE27catalyzestheoxidationof3-oxo-cholest-4-en-26oylCoA(13)andFadE34isactivetowardsteroidCoAesterscontainingfive-carbonchains(15).FadE14,characterizedasanovelacyl–acylcarrierproteindehydrogenaseratherthananacylCoAdehydrogenase,isinvolvedinthemycobactinbiosynthesispathway(16),andFadE9isanACDwhosesubstratesarebranchedchainaminoacids(17). Inarecentstudy,84drug-resistance–associatedgeneswereidentifiedthroughgenomesequencingof161differentM.tuberculosisisolates(18).Amongthese,severalwell-knowndrug-targetgenes(gyrA,ethA,rpoB,rpoC,embB,pncA,katG,thyA,andrpsL) Significance Thelipidcontentountsforapproximately60%ofthedryweightofthecellwallofpathogenicmycobacteriawithMycobacteriumtuberculosishavingmorethan250genesinvolvedinfattyacidmetabolism.Previousstudieshaveshownthatacyl-CoAdehydrogenase(ACD),whichintroducesunsaturationintofattyacids,exhibitsstrictsubstratespecificitytowarddifferentCoAthioestergroups.Here,weidentifiedauniqueACDmemberinmycobacteriathatexhibitsbroadsubstratespecificity.Furthermore,wedeterminedcrystalstructuresoftheenzymeandenzyme–plexestoexplainthebroadsubstraterecognitionobservedinthissystem.GiventheimportanceofFadE5infattyacidmetabolism,thesenewstructuresareexcellentplatformsforrationalstructurebasedantituberculosisdrugdiscovery. Authorcontributions:
Z.R.designedresearch;
X.C.,
J.C.,
B.Y.,andW.Z.performedresearch;
X.C.,
X.L.,andZ.R.analyzeddata;andX.C.,
L.W.G.,
X.L.,andZ.R.wrotethepaper. Theauthorsdeclarepetinginterest. ThisarticleisaPNASDirectSubmission. PublishedunderthePNASlicense. Datadeposition:TheatomiccoordinateshavebeendepositedintheProteinDataBank,(PDBIDcodes6KPT,6KRI,6KS9,6KSA,6KSB,6KSE,6LPY,and6LQ0–6LQ8).1Towhomcorrespondencemaybeaddressed.Email:liux@. Thisarticlecontainssupportinginformationonlineat/lookup/suppl/doi:10.1073/pnas.2002835117/-/DCSupplemental. FirstpublishedJune29,2020. DownloadedatInstituteofBiophysicsonAugust30,2020 16324–16332|PNAS|July14,2020|vol.117|no.28 /cgi/doi/10.1073/pnas.2002835117 wereidentified.However,othergenes,includingfadE5,werealsoidentified.ThefadE5geneislocatedonaclusteridentifiedasthefadE5-gap–likegenomicregion(19),whichharborsfadE,pks,papA,fadD,pE,mmpL,andgap-likegenes,andisbroadlydistributedacrossmycobacteria.InMycobacteriumsmegmatis(Ms),thesegeneproductsaresuggestedtoplayrolesinthebiosynthesisoftrehalosepolyphleates,afamilyoftrehalose-basedlipids(19).MsFadE5,actingasanACD,issuggestedtointroduceunsaturationintothefattyacylchainproducedfromtheFASI(fattyacidsynthaseI)system,andthechainisthenextendedandmodifiedbypolyketidesynthases(Pks)(20,21).InM.tuberculosis,similargeneclustershavealsobeenobserved.Sulfolipids(SL)andpolyacyltrehaloses(PAT)arebothlipidsinthecellwallofM.tuberculosis.Geneclusterscontainingpks,papA,fadD,andmmpLarealsoinvolvedintheformationandexportofSLandPAT(22,23).FadE5proteinsarealsosuggestedtoplayrolesinthebiosynthesisofthelipidsinthecellwall. ACDsareidentifiedbythespecificsubstratestheyutilize.ManyACDmembersinvolvedinthemetabolismoflipidsandaminoacidshavebeencharacterized,includingshort-chainacylCoAdehydrogenase(SCAD),medium-chainacyl-CoAdehydrogenase(MCAD),long-chainacyl-CoAdehydrogenase(LCAD),verylong-chainacyl-CoAdehydrogenase(VLCAD),isovaleryl-CoAdehydrogenase(IVD),isobutyryl-CoAdehydrogenase(IBD),branchedshortchainacyl-CoAdehydrogenase(BSCAD),andglutaryl-CoAdehydrogenase(GDH)(24).Ineukaryotes,differentACDmembersexhibitstrictsubstratespecificitiestowarddifferentCoAthioestergroups,whichresultsfromtheirdistinctactive-sitearchitectures.InthoseACDsthatcatalyzethedehydrogenationoflinearacyl-CoA,thelengthofthesubstratecarbonchainthattheenzymecaneptislimitedbythelengthofthesubstratebindingcavity.ThecavitylengthsofSCAD,MCAD,andVLCADare8,12,and24Å,andtheyhaveoptimalspecificitiesforfattyacyl-CoAswithchainsthatcontain4,8,and16carbonatoms,respectively(25–27).MostACDsformposedoffour∼43-kDasubunits.TwoexceptionsareVLCAD,whichisahomodimerconsistingoftwo∼67-kDasubunits(27,28).Theotheristheα2β2heterotetramerencodedbyfadEgenesinM.tuberculosis(13,14). Herein,weisolatedFadE5fromM.tuberculosis(MtbFadE5)andM.smegmatis(MsFadE5).MsFadE5hasbeenshowntobeinvolvedinresistingdruguptake,mostlikelybyaffectingcellwallpermeability.BothMtbFadE5andMsFadE5showbroadselectivityforlinearacyl-CoAsubstrateswithchainsvaryinginlengthsfrom4to22carbons.Importantly,itssubstratespecificityisquitedifferentfromotherhomologs,mostofwhichhaveastrictsubstraterequirement.Tounderstandthisunusualproperty,wedeterminedthecrystalstructuresofMtbFadE5andMsFadE5plexwithvarioussubstratesandMtbFadE5plexwithC18CoA.ThesestructuresprovidethemolecularbasisforthebroadsubstrateselectivityofMtbFadE5andMsFadE5. Results DrugResistanceAssaysofMsFadE5.ToverifytherolesofFadE5incellwalllipidbiosynthesis,drug-resistanceassayswereperformed(Fig.1).Inthisexperiment,thefadE5geneintheM.smegmatisgenomewasdisruptedbysubstitutionofaninternalsegmentwithahygromycin-resistancegene.VerificationofgenedisruptionwasperformedbyPCRfollowedbysequencingofthePCRproducts.TheMsfadE5genewasoverexpressedinthefadE5-deletedcellsplementationstudies.Fig.1AshowsthegrowthrateisthesameforthefadE5-deletedcellsandthewild-typecells.Whentreatedwithethambutol(2μgmL−1)andstreptomycin(0.1μgmL−1),first-lineandsecond-linedrugs,cellslackingfadE5exhibiteddecreaseddrug-resistanceparedwithwild-type.WhenthefadE5genewasoverexpressedinthefadE5-deletedcells,drugresistancetoethambutolwasgainedanddrugresistancetostreptomycinwasevenmoreevident(Fig.1BandC). SubstrateSelectivityofMtbFadE5andMsFadE5.Full-lengthMtbFadE5andMsFadE5wereproducedinE.coliforanalysis.Gelfiltrationrevealedthattheybothformhomodimersinsolution.Thiswasfurtherconfirmedbysedimentationequilibriumanalyticalultracentrifugation(AUC)(SIAppendix,Fig.S1BandC).AsequencealignmentbetweenMtbFadE5,MsFadE5,andotherACDsisshowninSIAppendix,Fig.S1A.ACDsareclassifiedordingtothesubstratestheyutilize.MostACDsform DownloadedatInstituteofBiophysicsonAugust30,2020 BIOCHEMISTRY Fig.1.Drug-resistanceassaysinM.smegmatis.(A)Growthcurvesforwild-typeM.smegmatis[Ms(WT)],MsfadE5-deletedstrain[Ms(ΔfadE::hyg)],and MsfadE5-deletedstrainwithoverexpressionofMsfadE5[Ms(pVV16::fadE5)].(B)Growthcurvesofwild-typeM.smegmatis,MsfadE5-deletedstrain,andMsfadE5-deletedstrainwithoverexpressionofMsfadE5inthepresenceorabsenceof2μgmL−1ethambutol.(C)Growthcurvesofwild-typeM.smegmatis,MsfadE5-deletedstrainandMsfadE5-deletedstrainwithoverexpressionofMsfadE5inthepresenceorabsenceof0.1μgmL−1streptomycin.Eachbaristhe meanandSDofthreemeasurements. Chenetal. PNAS|July14,2020|vol.117|no.28|16325 homo-orposedoffour∼43-kDaprotomers.OnlyVLCADisahomodimerconsistingoftwo∼67-kDaprotomers.FadE5isalsoahomodimerconsistingoftwo66-kDaprotomers.So,weinferredthatFadE5isaVLCADatfirst. However,ourassaysshowedthatMtbFadE5andMsFadE5arebothhighlyactivetowardsubstratesofvaryinglengthsoflinearacyl-CoA,includingC4CoA,C6CoA,C10CoA,C12CoA,C16CoA,andC18CoA(Fig.2AandB).Isothermaltitration DownloadedatInstituteofBiophysicsonAugust30,2020 Fig.2.ActivityassaysforMtbFadE5andMsFadE5andITCstudies.(AandC)ActivityofMtbFadE5withC4CoA,C6CoA,C10CoA,C12CoA,C16CoA,andC18CoA.TheMtbFadE5E447AmutantisinactivewiththeC18CoAsubstrate.ITCwasusedtodetermineinteractionsbetweenMtbFadE5C10CoAandC20CoA.(BandD)ActivityandITCdataforMsFadE5. 16326|/cgi/doi/10.1073/pnas.2002835117 Chenetal. DownloadedatInstituteofBiophysicsonAugust30,2020 BIOCHEMISTRY calorimetry(ITC)wasnextperformedtodeterminetheirbindingaffinitieswiththesubstrates.ThecatalyticbaseformostACDscharacterizedtodateisaglutamateresidue.InSCADitisE368(29),whileinMCADitisE376(30)andinHsVLCADitisE422(27).SequencealignmentofMsFadE5andMtbFadE5withotherACDssuggestthatE447isthecatalyticbaseinMsFadE5andMtbFadE5.Toconfirmthisassumption,weexpressedandpurifiedE447Amutantsforbothenzymesandshowedneitherpossessedactivity(Fig.2AandB),confirmingthisresidueisthecatalyticbaserequiredfordehydrogenation.ITCwasalsoperformedusingsamplesofMtbFadE5E447AandMsFadE5E447A.Theresultsshowedthatthebindingaffinitiesincreasedwiththelengthofthecarbonchain(Fig.2CandDandSIAppendix,Fig.S2).C22CoAbindstoMtbFadE5withaKdvalueof0.16μM,whileC6CoAbindstoMtbFadE5withaKdvalueof196μ
M.SimilarvalueswereobtainedforthebindingaffinitiesofMsFadE5withthesesubstrates(Table1).Thus,bothMtbFadE5andMsFadE5showabroadsubstrateselectivityforacylCoAsbuttendtopreferthelongerratherthanshortersubstrates.Theicparameters(kcatandKm)forrepresentativesubstrateswerealsodetermined(Table2andSIAppendix,Fig.S3).Thefreeconcentrationofacyl-CoAestersisunlikelytoexceed200nMundernormalcellularconditions(31).Therefore,theeffectivenessofFadE5incellsusingthesesubstrateswouldbelow.ThismaybeafactorthatcontributestotheslowgrowthrateofM.tuberculosisandM.smegmatis. OverviewoftheMsFadE5Structure.ToinvestigatethereasonsforthebroadsubstrateselectivityofMtbFadE5andMsFadE5,crystalstructuresofbothenzymesweredetermined.Initially,wesolvedthestructureofMsFadE5(80%sequenceidentitywithMtbFadE5).ThecrystalstructureshowsthatMsFadE5formsahomodimer(Fig.3A).Virtuallyallresiduescouldbefittedtotheelectrondensitymapexceptfortheregionfrom482to489.TheNterminusofMsFadE5consistsof467residuesandsharesacanonicalfoldthatisobservedinothertetramericACDs.Theadditionalresidues,467to611,formanα-helicaldomainattheCterminusthatisnotfoundinothertetramericACDs.TheC-terminalhelicaldomainexistsspecificallyinVLCADs.InhumanVLCADtheC-terminaldomainisresponsibleforbindingtotheinnermitochondrialmembrane(32,33).Residues446to478aredisorderedinthatstructureandarethoughttoberesponsibleforanchoringtothemembrane.ThisisbasedontheobservationthatFadE5waspreviouslyidentifiedinamembranefractionofM.tuberculosis(34).InFadE5,residues482to489aredisorderedandlocatedinthesurfaceoftheC-terminaldomain,suggestingthatthisregionmaybealsoinvolvedinmembrane Table1.TheKdvaluesdeterminedbyITCSubstrate MtbFadE5MsFadE5 C6-CoAC10-CoAC12-CoAC16-CoAC18-CoAC20-CoAC22-CoAC6-CoAC10-CoAC12-CoAC16-CoAC18-CoAC20-CoAC22-CoA Kd(μM) 196±7867.1±16.0 60±9.24.4±0.674.18±0.820.19±0.070.16±0.06316±27420.1±1.324.0±7.73.45±0.310.72±0.200.07±0.050.19±0.14 anchoringandmayonlyeorderedwheninteractingwiththemembrane. ActiveSiteandSubstrateBindingCavity.AFADcofactorispresentineachoftheMsFadE5activesites(Fig.3B).Thesearelocatedatthedimerinterface.InallofthestructurallycharacterizedACDs,thereisacavityclosetotheisoalloxazineoftheboundFAD,whichharborsthecarbonchainofthesubstrate.ThecarbonchainislocatedproximaltotheN5atomoftheisoalloxazineringsinceitisinvolvedintheα-,β-desaturationofthesubstrates.Thelengthofthecavitylimitsthelengthofthesubstratecarbonchain.InSCAD(PDBIDcode1JQI),MCAD(PDBIDcode3MDD),andHsVLCAD(PDBIDcode2UXW),thelengthoftheircavitiesare8,12,and24Å,respectively(27).However,thecavitylengthinMsFadE5isapproximatelyonly12.5Å(Fig.3C).Thepocketthereforeappearstobenotlongenoughtomodatesubstrateswithlongcarbonchains.Thus,theremustbeconformationalchangeswhenthelongersubstratesbind. StructuralBasisfortheBroadSubstrateSelectivityofMsFadE5andMtbFadE5.ThestructuresoftheotherreportedACDsplexwithsubstratesweresolvedbyaddingsubstratestothenativeenzymespriortocrystallization.Inoureffortstocrystallizeenzyme–plexes,wefollowedthesamestrategy.However,wewereonlyabletogeneratecrystalswheretheproducthadbeenreleased.Superimpositionofthisstructureandtheapo-MsFadE5structurerevealssomedifferences(SIAppendix,Fig.S4B).Inparticular,thesidechainsofresiduesF294andD295exhibitconformationalchanges.ThesidechainofE447,thecatalyticbase,isreorientedtofacetotheisoalloxazineringofFAD.R301,whichelectrostaticallyinteractswithE447inapo-MsFadE5,alsoundergoesaconformationalchange.Inthisstructure,theelectrondensityforY446showsthatitadoptstwodifferentconformations:Oneconformationisthesameastheapoformandtheotherconformationwidensthebindingcavityforsubstratestobind(SIAppendix,Fig.S4A). Topreventthereleaseofthesubstrate,weobtainedstructuresofMsFadE5whereE447wasmutatedtoalanine.CrystalscouldthenbeobtainedplexwiththesubstratesC4CoA,C6CoA,C8CoA,C10CoA,C12CoA,C14CoA,C16CoA,C17CoA,C18CoA,C20CoA,andC22CoA(Fig.4andSIAppendix,Fig.S5).ThestructureoftheE447AmutantofMsFadE5plexwithC18CoAprovidesanexampletoexplorethestructuralbasisofthebroadsubstratespecificity.Intotal,thereare11hydrogenbondsbetweenresiduesintheproteinandtheCoAportionofC18CoA(Fig.5B).ConformationalchangeinthesidechainofR301isneededtoallowtheformationofahydrogenbondandthemovementofF294resultsinaπ–πstackinginteraction(3.6Å)withtheadenineoftheCoA(Fig.5C).ThisphenylalanineisnotconservedamongotherACDs,asindicatedinthesequencealignment(SIAppendix,Fig.S1).TheequivalentaromaticaminoacidisonlyfoundinGDH(PDBIDcode3MPI).However,nosimilarπ–πstackinginteractioncouldbefoundinthatprotein.Thus,thisisafeaturespecifictothisenzyme. Toclarifythecontributionoftheinteractingresidues,mutationalanalysiswascarriedoutandtheKdvaluesmeasuredbyITC(SIAppendix,Fig.S6andTableS1).FortheS171A,K225A,andR460AmutantsusingC20CoA,Kdvalueswereseveralfoldslowerthanforthewild-typeenzyme.TheD456A_R460AdoublemutantalsoshowedadecreasedparedtotheR460singlemutant.So,theseresiduesaresuggestedcontributedtothebindingaffinityoftheproteinandsubstrate.However,theR301AandF294AmutantsbothshowedanincreasedbindingaffinityforC20CoA.Wespeculatethatitwouldbeeasierfortheligandtoenterintothepocketintheabsenceofthesetwobulkysidechains.Notably,S171,K225,D456,andR460couldinteractwiththesubstratewithoutanyconformationalchanges. Chenetal. PNAS|July14,2020|vol.117|no.28|16327 Table2.Enzyme Steady-stateicparametersforMsFadE5andMtbFadE5 Substrate kcat(s−1) Km(μM) MsFadE5MtbFadE5MtbFadE5MtbFadE5 C4CoAC4CoAC6CoAC18CoA 0.53±0.061.07±0.120.80±0.140.61±0.04 285.9±53.1358.7±71.8353.0±105.6162.5±28.8 kcat/Km(M−1s−1) (1.85±0.13)×103(2.98±0.26)×103(2.27±0.28)×103(3.75±0.42)×103 However,R301andF294bothundergoconformationchangesuponsubstratebinding.Theinteractionsbetweenthetworesiduesandthesubstratearefacilitatedthroughtheconformationalchanges.So,weinferthatthesetworesiduesmaynotcontributedtotheoriginalrecognitionofthesubstratebutthestabilizationaftersubstratebinding. Assuggestedabove,inplexes,thecavitycanbelengthenedaslongersubstratesbind(Fig.6).Thisisachievedbytheresiduesalongthecavityreorientingthemselvesuponsubstratebinding.IntheMsFadE5–C4CoAstructure,thesidechainofY446rotatestoallowbinding(Fig.6A).Thus,Y446playsatypeofgatekeeperrolesinceinbothenzymes,thisresiduehastorotatetoallowthesubstratetogainess.IntheMsFadE5–C10CoAmodel,thesidechainofM134changesconformationtoallowC10CoAbinding(Fig.6B).However,intheMsFadE5–C12CoAandMsFadE5–plexes,M134undergoesalargerconformationalchangesothatthecavitypletelyopen(Fig.6CandD).Theentirelengthofthecavityishydrophobicsothatthelongercarbonchainsofthesubstratemakeadditionalhydrophobicinteractions,resultinginthepreferenceofFadE5forlong-chainsubstrates.Allinall,thesidechainsoftheresiduesalongthecavitywallguidetherouteandbindingofthecarbonchain.ThesidechainofM130inMsFadE5stretchesintothecavity,forcingthecarbonchaintotakeaturn(SIAppendix,Fig.S7).InMtbFadE5,thisresidueissubstitutedbyaglycine.AstructureoftheE447A_M130GdoublemutantofMsFadE5plexwithC16CoAshowsthatthecarbonchaintakesadifferentparedtothesingleE447mutant(SIAppendix,Fig.S7). Wewereessfulingeneratingapo-MtbFadE5crystals.However,wedidmakecrystalsoftheE447AmutantofMtbFadE5plexwithC18CoA.plexisalsoahomodimer(Fig.7A).ThermsdforallCαatomsbetweentheMtbFadE5E447A–C18CoAandMsFadE5E447A–plexesaftersuperimpositionisonly0.327Å,demonstratingthetwostructuresarevirtuallyidentical(Fig.7B).TheelectrondensityfortheCoAportionofthesubstrateisstrong.However,althoughthedensityforthecarbonchainispoor,itcouldbereasonablyfittedinthe2Fo–Fcmap(Fig.7CandD). ThehydrogenbondsbetweentheMtbFadE5andtheC18CoAaresimilartothoseobservedforMsFadE5anditssubstrate(SIAppendix,TableS2).Theπ–πstackingbetweentheF294andtheadenineofsubstrateisalsopresent.Theresiduesthathaveflexiblesidechainsinthesubstratebindingcavityintheplexadoptsimilarconformationsintheplex,suggestingthesemayundergosimilarconformationalchangeswhenthesubstrateentersthebindingcavity.Thus,itisreasonabletosuggestthatthestructuralbasisofthebroadsubstrateselectivityofMtbFadE5isthesameasthatofMsFadE5.However,thereareseveralaminoacidsubstitutionsinthebindingcavityinparedtoMsFadE5.TheseareM134andF126inMsFadE5,whicharesubstitutedbyF134andW126inMtbFadE5,respectively(Fig.7E).SIAppendix,Fig.S8showshowthesechangesimpactonthebindingmodeofC18CoA.Thesizeandorientationoftheseresiduesidechainsaretotallydifferent,whichmightshapethebindingpocketsanddecidetheconformationofthecarbonchainofC18CoA. DiscussionTodate,ithasalwaysbeensuggestedthatACDshavestrictsubstratespecificitiesandSCAD,MCAD,LCAD,andVLCADarespecificallyactiveagainstfattyacyl-CoAswith4,8,14,and16carbonsinlength,respectively(25–27).ThesizeofthesubstratebindingpocketofACDsdeterminesthemaximumlengthofthecarbonchainthattheproteincanmodate.Thewidthandlengthofthecavityarerestrictedbytheresiduesalongthecavitywall.ThelengthoftheACDH-11bindingcavityis14Å,whichrestrictsthecarbonlengthto12.icassaysalsoindicatethatACDH-11haslittleactivityagainstsubstrateswithchainscontainingmorethan12carbonatoms.ThecavitylengthislimitedbythepresenceofY344andL159.Thestructureoftheapo-formofACDH-11andthestructureofACDH-11plexwithC11CoAdisplaythesameconformationsforY344andL159.ThetemperaturefactorsofthesetworesiduesarerelativelyparedtoalloftheresiduesinACDH-11(28).ThesedatarevealedthatthebindingcavityofunboundACDH-11isnodifferenttothatwhenACDH-11isboundtothesubstrates.Similarly,inSCAD,I251limitsthelengthofthesubstratebindingcavity(29).InMCAD,Q95andE99limitthelengthofthesubstratebindingcavity(30).InthepresentMsFadE5structuresplexwithsubstrates,conformationalchangesoftheresidueslimitingthebindingcavitycouldurtoprovideenoughroomforlongersubstratestobind.Furthermore,intheMsFadE5E447A–C22CoAstructure,thecavityextendstothesurfaceoftheenzymesuggestingthereismoreroomforsubstrateslongerthanC22CoAtobind. HsVLCADshowslittleactivityforsubstrateswithchainlengthsoffewerthan12carbons(26).IntheHsVLCAD–plexstructure,onlytheelectrondensityforthecarbonchainportionofthesubstrateisobservedclearly.ThemissingdensityoftheCoAportionofthesubstrateismostlikelycausedbythehighmobilityoftheCoAmoiety.TheabsenceofhydrogenbondsbetweenHsVLCADandCoAmayresultinthe Fig.3.Overallstructureofapo-MsFadE5anditsactivesite.(A)CartoonrepresentationandConnollysurface(transparent)oftheMsFadE5dimer.Thetwosubunitsareshowninredandblue.FADisshowninstickmodelswithyellowcarbonatoms.(B)2Fo–FcelectrondensitymapforFADcontouredat1.0σ.(C)Surfacerepresentation(transparent)oftheputativesubstratebindingcavity,withFADshownasastickmodel. 16328|/cgi/doi/10.1073/pnas.2002835117 Chenetal. DownloadedatInstituteofBiophysicsonAugust30,2020 Fig.4.StructuresoftheE447AmutantsofMsFadE5plexwithsubstrates.(Left)Cartoonrepresentation(green)oftheE447AmutantofMsFadE5plexwithC4CoA(stickmodelwithpinkcarbonatoms).(Right)2Fo–Fcelectrondensitymaps(contourlevel=1.0σ)forthesubstrates. highmobility(27).WhenC11CoAintheACDH-11–plexstructureissuperimposedwiththeHsVLCAD–C14CoAstructure,onlyonehydrogenbondisobserved.SCADandMCADformfiveandfourhydrogenbondswiththeCoAofthesubstrate,respectively(28).InHsVLCAD,thebindingaffinitybetweentheenzymeandthesubstrateisdeterminedmainlybyhydrophobicinteractionsbetweentheenzymeandthelongacylchains.Therefore,HsVLCADnotonlyeptsacylCoAswithlongacylchainsasasubstratebutalsoshowsa preferenceforthem.IntheMsFadE5–plex,thereare11hydrogenbondsandarareπ–πstackinginteractionbetweenF294andtheadenineofthesubstrate.TheseadditionalbondsseemtobethereasonwhyFadE5caneptsubstrateswithshortcarbonchains.Althoughthehydrophobicinteractionsarenotstrong,theCoAportionoftheshortersubstratescanbindtotheenzyme.Toexplorewhatprecludestheanalogoushydrogenbonds,thestructuresofMsFadE5andHsVLCADpared.InHsVLCAD,theβ-sheetcontainingstrands4 DownloadedatInstituteofBiophysicsonAugust30,2020 BIOCHEMISTRY Fig.5.InteractionsbetweenC18CoAandresiduesintheE447AmutantofMsFadE5.(A)Cartoonrepresentationofthestructure.C18CoAisshownasastickmodelwithcarbonatomscoloredpink.(B)HydrogenbondsbetweenC18CoAandtheresiduesinMsFadE5areshownasdottedlines.DistancesaregiveninÅngstroms.(C)Theresiduesthatinteractwiththeproteinwithreorientedsidechains.(D)SuperimpositionoftheCoAbindingsitesintheHsVLCAD–plex(yellow)andtheE447AmutantofMsFadE5plexwithC18CoA(green).C18CoAisshownasastickmodelwithcarbonatomscoloredpink. Chenetal. PNAS|July14,2020|vol.117|no.28|16329 Fig.6.ConformationalchangesinthesubstratebindingcavitiesoftheE447AmutantofMsFadE5uponsubstratebinding.(
A,Left)Cut-awaysurfacerepresentationofthesubstratebindingcavityintheplex.C4CoAisshownasastickmodelwithcarbonatomscoloredpink.(Right)Conformationalchangesfortheresiduesalongthesubstratebindingcavitywall(green)paredwiththoseintheapostructure(blue).(B)Conformationalchangesintheplex.(C)Conformationalchangesintheplex.(D)ConformationalchangesintheC20CoAstructure. and5extendsfurtherawayfromtheCoAbindingsitethanthatinMsFadE5,andinMsFadE5,thestrandsinthesheetaremuchshorterandfacetowardtheCoAportionofthesubstratesotheresiduescaninteractwithCoA(Fig.5D).Takentogether,theapo-FadE5andplexstructuresrevealthestructuralbasisforitsbroadsubstrateselectivity.Thisenzymecaneptshortcarbonchainacyl-CoAasasubstratesinceitcanformseveralhydrogenbondsandalsohasauniqueπ–πstackingwiththeCoAportionofthesubstrate.Inaddition,FadE5canmodatelonger-chainfatty-acylCoAssincetheresiduesthatpotentiallylimitthelengthofthesubstratebindingcavitycanreorienttheirsidechains. FadE5isahomodimericACDwhosesubstratesarelinearchainfattyacids.M.tuberculosiscanmakeuseoffattyacidsofdifferentlengthsasacarbonsource.Thesefattyacidsaredegradedthroughtheβ-oxidationpathwaytoprovidetheenergyrequiredforthegrowthofM.tuberculosis.Inthefirststepofthe β-oxidationpathway,ACDintroducesunsaturationintothefattyacids,convertingacyl-CoAintoenoyl-CoA.FadE5isprobablyinvolvedinthedegradationoffattyacidsofvaryinglengths.Inmycobacteria,theFASIsystemproducesC14-C26acyl-CoAsthroughthedenovofattyacidbiosynthesis,thenunsaturationandmodificationswouldbeintroducedintotheacyl-CoAs.Insomenontuberculosismycobacteria,suchasM.smegmatis,fadE5locatesinthegeneclustersthatareresponsibleforbiosynthesisofpolyketidelipids,soFadE5mightalsobeinvolvedinintroducingunsaturationintoproductsfromFASIsystem,andthechainisthenextendedbyPks.Here,wereportthatFadE5suseC4CoAtoC22CoAassubstrates.IntheFadE5–plexstructures,wehavededucedthatthesubstratebindingcavitycouldmodatesubstrateslongerthanC22CoA.SimilarfindingshavebeenobservedfortheM.tuberculosisβ-ketoacyl-acylcarrierprotein(ACP)synthaseIII(mtbFabH)(35).MtbFabHisalinkbetweentheFASIandFASIIsystemsofM.tuberculosisandcatalyzeadecarboxylativecondensationinthefirststepofFASIIprocessing.EscherichiacoliFabHandothertypicalFabHsutilizeC2-C6acyl-CoAassubstrates,whileMtbFabHcantolerateawiderangeofC12-C20acyl-CoAsubstrates. ThesubstrateselectivityofFadE5isinordancewithitsrolethatwepredictinintroducingunsaturationintotheproductsofFASI.Moreover,theproductscanbeincorporatedasphospholipidsintothemembrane,orelongatedbyFASIIforthesynthesisofmycolicacids,orelongatedbyspecificPKSstoformplexlipidslikemycoketides,SLs,andphthioceroldimycocerosate,orsynthesizepoundssuchastriacylglycerides(36–39).Asaresult,itissuggestedthatFadE5canplaymultiplerolesinlipidmetabolisminM.tuberculosis.ThelipidsinthecellwallofM.tuberculosisandM.smegmatishavebeenshowntobeinvolvedinformationofthecellwallpermeabilitybarrier.Here,drug-resistanceassayswereperformedonwild-typeM.smegmatis,fadE5-deletedM.smegmatis,andfadE5overexpressedM.smegmatis.ThefadE5-deletedM.smegmatisshowedanobviousdecreasedresistancetoethambutolandstreptomycin.SinceweinferredthatfadE5playsanimportantroleinthebiosynthesisofthelipidsinthecellwall,thedecreaseddrugresistanceofthefadE5-deletedstrainsmayhaveresultedfromtheincreasedcellwallpermeability. Intotal,35FadEmembershavebeenidentifiedinM.tuberculosis;suchalargenumberoffadEgenesinthegenomereflectsthenecessityofM.tuberculosistoloadalargevarietyoflipids.Thus,thedevelopmentofsmall-moleculeinhibitorsagainstthisandcloselyrelatedenzymeshavestrongpotentialtobedevelopedasnewgenerationmultitargetdrugs.Thisstudythereforeadvancestheunderstandingoffattyacidmetabolisminmycobacteria,andtheavailabilityofthestructureandcharacterizationinformationofFadE5shouldfacilitatethediscoveryanddevelopmentofnewantimycobacterialagents. MaterialsandMethods ProteinExpressionandPurification.ThegenesencodingMtbFadE5andMsFadE5areRv0244candMSMEG_0406,andwerePCR-amplifiedfromM.tuberculosisH37RvandM.smegmatismc2155,respectively.Thegenefrag- mentswerethenfusedintothepET28aexpressionvectorwiththeEcoRI andHindIIIsitesbyhomologousbinationusingtheOneStepCloning KittoproduceanN-terminalHis-taggedconstruct.MutationwasintroducedintoMtbFadE5andMsFadE5bythefastmutagenesissystemfromTransGen.ThefusedvectorswerealltransformedintoE.coliBL21Rosettacellsfor expression.ProteinexpressioninLBmediasupplementedwithkanamycin(50mg/mL)at37°Cwasinducedby0.5mMisfopropyl-βyr- anoside(IPTG)whentheOD600reached0.6.Cultureswerethengrownat16°Cfor18h.CellswereharvestedbycentrifugationandresuspendedinMCAC-0buffer(25mMTris·HCl,pH8.0,500mMNaCl,and10%glycerol), followedbylysisusingsonication.Thedebriswasremovedbycentrifugation at18,000rpmfor40min.Supernatantscontainingthetargetproteinswere loadedtwiceontoaNi-NTAcolumn(GEHealthcare)equilibratedwith DownloadedatInstituteofBiophysicsonAugust30,2020 16330|/cgi/doi/10.1073/pnas.2002835117 Chenetal. DownloadedatInstituteofBiophysicsonAugust30,2020 BIOCHEMISTRY Fig.7.StructuresoftheE447AmutantofMtbFadE5plexwithC18CoAanditssubstratebindingcavity.(A)Cartoonrepresentationofthedimer.Individualsubunitsareshowninyellowandred.(B)The2Fo–FcelectrondensitymapforC18CoAcontouredat1.0σ.(C)Equivalent2Fo–FcelectrondensitymapforC18CoAcontouredat0.6σ.(D)Superimpositionoftheplex(green)andplex(redandyellow).(E)Superimpositionofthesubstratebindingcavitiesintheplex(green)andtheplex(yellow).Residuesalongthesubstratebindingcavitywallsthatdifferbetweenthetwostructuresareshowninstickmodels.TheC18CoAinMsFadE5isinpink,andC18CoAinMtbFadE5isinblue. MCAC-0buffer.ThecolumnwassubsequentlywashedwithMCAC-20buffer(MCAC-0supplementedwith20mMimidazole).Then,thetargetproteinswerethenelutedwithMCAC-500buffer(MCAC-0supplementedwith500mMimidazole)andfurtherpurifiedusingaHiTrapQcolumn(GEHealthcare)foranionexchangechromatography.Forthefinalstep,aSuperdexG200column(GEHealthcare)equilibratedwith25mMTrisHCland150mMNaCl,pH8.0,wasusedtopurifythetargetproteinsbygelfiltrationchromatography.TheelutedproteinswereevaluatedbySDS/PAGE.Selenomethionine-substitutedMsFadE5proteinwasexpressedinmethionineauxotrophicE.colistrainB834(DE3)inM9medium,aspreviouslydescribed(40). Drug-ResistanceAssay.M.smegmatisstrainswereculturedin7H9brothsupplementedwith10%(vol/vol)ADC(1,000-mLsolutioncontained9gNaCl,50gBSA,and20gglucose),0.1%(vol/vol)Tween-80,0.5%(vol/vol)glycerin,and5μgmL−1carbenicillin.Cellsweregrownat37°CtoanOD600of0.6∼0.8.ThenthecellsweredilutedtoanOD600of0.1.TheOD600wasmeasuredusing96areaplate(Corning)inthepresenceorabsenceofdrugs.DataanalyseswereperformedusingGraphPadPrism6.0. AnalyticalUltracentrifugation.AUCwascarriedoutusingaBeckmanCoulterOptimaAUCequippedwithanAn60TirotortodeterminethesedimentationvelocitiesofMtbFadE5andMsFadE5.Theproteinsweredilutedinasolutioncontaining25mMTris,pH8.0,and150mMNaCl.Thesampleswereconcentratedto1mg/mL.Sedimentationvelocitydatawerecollectedat42,000rpm.Atotalof99scanswerecollectedevery4minusinginterferenceoptics.Allmeasurementswereconductedat4°
C.ThedatawereanalyzedusingtheprogramSedfitwithacontinuousc(s)distributionmodel(41). ITC.ThebindingaffinitiesofMtbFadE5andMsFadE5forC4CoA,C6CoA,C10CoA,C12CoA,C16CoA,C18CoA,C20CoA,andC22CoAweredeterminedusinganITC-200microcalorimeter(Microcal).Allmeasurementswereperformedat25°
C.ThemeasurementswereperformedusingtheE447AmutantsofMtbFadE5andMsFadE5.Proteinsamplesat0.04∼0.10mMinbuffercontaining25mMTris(pH8.0)and150mMNaClwereloadedinthereactioncell.Foreachbindingassay,substratesataconcentrationof0.4∼0.90mMinthesamebufferweretitratedintotheproteinsampleswithasyringe.Thetitrationconsistedof19injectionsof2.0μLevery120s(exceptforthefirstinjectionof0.4μL). EnzymeAssays.TheactivitiesofMtbFadE5andMsFadE5weremeasuredbasedonthereductionofferroceniumasanelectroneptoratanabsorbanceat300nm(extinctioncoefficientof2.75mM−1cm−1)at25°
C within1minoftheinitiationoftheassay(42,43).Thereactionmixture(100μL)contained25mMTris,pH8.0,150mMNaCl,1μMprotein,and0.5mMferroceniumhexafluorophosphate.Theassayswereinitiatedbytheadditionof0.4mMofvaryingsubstratesincludingC4CoA,C6CoA,C10CoA,C12CoA,C16CoA,andC18CoA.TheactivitiesoftheE447Amutantsweremeasuredinthesameway.TheicparametersofFadE5weredeterminedbymeasuringtheinitialvelocitiesatapHof8.0andatemperatureof25°
C.The100-μLreactionmixturewaspreparedbyadding1mMferroceniumhexafluorophosphate,dilutingdifferentsubstratestothedesiredconcentration.Thereactionwasinitiatedbyaddingenzymetoafinalconcentrationof0.8μ
M.Initialvelocitieswereexaminedatsubstrateconcentrationsfrom50to500μ
M.Michaelis–Mentenplotswereusedtodeterminetheicparameters. CrystallizationandDataCollection.Crystalsweregrownat20°Cbyvapordiffusionmixing1μLofproteinsolutionand1μLofreservoirsolution.Selenomethionine-substitutedMsFadE5crystalswereobtainedin0.19Mmagnesiumformate,15.6%(wt/vol)PEG3350,2%(vol/vol)2-propanol,20mMMES(pH6.0),and40mMCa(OAc)
2.ASAD(single-wavelengthanomalousdispersion)datasetwascollectedat100Konbeamline17UoftheShanghaiSynchrotronRadiationFacility(SSRF,China)fromacryoprotectedselenomethionine-substitutedMsFadE5crystal.CrystalsofnativeMsFadE5weregrownin0.19Mmagnesiumformate,15.6%(wt/vol)PEG3350,0.4M(NH4)2SO4,0.4%(vol/vol)PEG400,and20mMacetatepH5.5.ThedatasetwascollectedonbeamlineBL41XUoftheSPring-8synchrotronradiationfacility(Japan).CrystalsofnativeMsFadE5soakedwithsubstratesweregrownin0.1MHepessodium(pH7.0),2%vol/volPEG400,and2M(NH4)2SO4.Thisdatasetwascollectedonbeamline19UattheSSRF.CrystalsoftheE447AmutantofMsFadE5wereobtainedin0.1MBis-Tris(pH6.5)and2M(NH4)2SO4.Thedatasetwascollectedonbeamline19UattheSSRF.CrystalsoftheE447AmutantofMsFadE5plexwithvarioussubstrateswereobtainedin0.1MHepes-sodium(pH7.0),2%vol/volPEG400,2M(NH4)2SO4and1mMFADbysoakingwithCoAsinthemolarratioof1:
5.Thesedatasetswerecollectedonbeamline18Uand19UattheSSRF.CrystalsoftheE447AM130GdoublemutantofMsFadE5plexwithC16CoAwereobtainedin0.8MNaH2PO4/1.2MK2HPO4,0.1Macetate(pH4.5),1mMFAD,and1.2mMC16CoA.Thedatasetwascollectedonbeamline19UattheSSRF.CrystalsoftheE447AmutantofMtbFadE5plexwithC18CoAwereobtainedbycocrystallizationwithsubstratesatamolarratioof1:
3.Thesecrystalsweregrownin20%(wt/vol)PEG3350and2Mpotassiumnitrate.Thedatasetwascollectedonbeamline19UattheSSRF.ThedatasetswereprocessedwiththeprogramHKL2000(44)andXDSpackage(SIAppendix,TableS3). Chenetal. PNAS|July14,2020|vol.117|no.28|16331 Phasing,ModelBuilding,andRefinement.ASADdatasetfromtheselenomethionine-substitutedMsFadE5crystalwasusedtodeterminephasesusingtheprogramPhenixAutoSolWizardbytheSADmethod(45,46).Twenty-threeseleniumatomswerelocated,andafigureofmeritof0.381wasobtainedfollowingtherefinementofheavyatomparametersandphasecalculations.Then,Phenix.autobuildwasusedformodelbuilding(47).Residues482to489couldnotbetracedduetotheirpoorlyorderedelectrondensity.Thestructureofapo-MsFadE5wasdeterminedbymolecularreplacementwiththePhaser(48)moduleinCCP4(49)usingtheselenomethionine-substitutedMsFadE5modelasasearchtemplate.TheoutputmodelfrommolecularreplacementwassubsequentlysubjectedtoiterativecyclesofmanualmodeladjustmentwithCoot(50)andrefinementwithPhenix.refine(51).ThestructuresofMsFadE5mutantsplexwithsubstrateswerealldeterminedbymolecularreplacementwithPhaserusingtheapo-MsFadE5structureasthesearchtemplate.ThestructureoftheMtbFadE5mutantplexwithC18CoAwasalsodeterminedusingapoMsFadE5asatemplate.Theoutputmodelfrommolecularreplacementwas thenrebuiltwiththeprogramPhenix.autobuild.ThemodelswerefurtherrefinedwithCootandPhenix.refine.AlloftherefinedmodelswerevalidatedbyMolProbity(52).ThephasingandrefinementstatisticsaresummarizedinSIAppendix,TableS3. DataAvailability.ThecoordinatesandstructurefactorshavebeendepositedintheProteinDataBank.TheessioncodesaresummarizedinSIAppendix,TableS3. ACKNOWLEDGMENTS.WethankthestaffmembersoftheShanghaiSynchrotronRadiationFacility(China),aswellasSPring-8(Japan)fortheirhelpwithdatacollection.ThisworkwassupportedbyNationalKeyResearch&DevelopmentProgramofChinaGrant2017YFC0840300andNationalNaturalScienceFoundationofChinaGrant81520108019(toZ.R.),andFundamentalResearchFundsfortheCentralUniversities,NankaiUniversity63191431and63181333(toX.L.).
1.Z.Lou,
X.Zhang,Proteintargetsforstructure-basedanti-Mycobacteriumtuberculosisdrugdiscovery.ProteinCell1,435–442(2010).
2.K.Mdluli,
M.Spigelman,Noveltargetsfortuberculosisdrugdiscovery.Curr.Opin.Pharmacol.6,459–467(2006).
3.L.Zhangetal.,Structuresofcellwallarabinosyltransferaseswiththeanti-tuberculosisdrugethambutol.Science,10.1126/science.aba9102(2020).
4.B.Zhangetal.,CrystalstructuresofmembranetransporterMmpL3,ananti-TBdrugtarget.Cell176,636–648.e13(2019).
5.L.W.Riley,Ofmice,men,andelephants:Mycobacteriumtuberculosiscellenvelopelipidsandpathogenesis.J.Clin.Invest.116,1475–1478(2006).
6.E.C.Hett,
E.J.Rubin,Bacterialgrowthandcelldivision:Amycobacterialperspective.Microbiol.Mol.Biol.Rev.72,126–156(2008).
7.E.J.Muñoz-Elías,
J.D.McKinney,Mycobacteriumtuberculosisisocitratelyases1and2arejointlyrequiredforinvivogrowthandvirulence.Nat.Med.11,638–644(2005).
8.M.F.Wipperman,
N.S.Sampson,
S.T.Thomas,Pathogenroidrage:CholesterolutilizationbyMycobacteriumtuberculosis.Crit.Rev.Biochem.Mol.Biol.49,269–293(2014).
9.H.Bloch,
W.Segal,BiochemicaldifferentiationofMycobacteriumtuberculosisgrowninvivoandinvitro.J.Bacteriol.72,132–141(1956). 10.S.T.Coleetal.,DecipheringthebiologyofMycobacteriumtuberculosisfrompletegenomesequence.Nature393,537–544(1998). 11.D.Clark,RegulationoffattyaciddegradationinEscherichiacoli:Analysisbyoperonfusion.J.Bacteriol.148,521–526(1981). 12.M.F.Wipperman,
M.Yang,
S.T.Thomas,
N.S.Sampson,ShrinkingtheFadEproteomeofMycobacteriumtuberculosis:Insightsintocholesterolmetabolismthroughidentificationofanα2β2heterotetramericacylcoenzymeAdehydrogenasefamily.J.Bacteriol.195,4331–4341(2013). 13.M.Yangetal.,UnravelingcholesterolcatabolisminMycobacteriumtuberculosis:ChsE4ChsE5α2β2acyl-CoAdehydrogenaseinitiatesβ-oxidationof3-Oxo-cholest-4-en-26-oylCoA.ACSInfect.Dis.1,110–125(2015). 14.S.T.Thomas,
N.S.Sampson,Mycobacteriumtuberculosisutilizesauniqueheterotetramericstructurefordehydrogenationofthecholesterolsidechain.Biochemistry52,2895–2904(2013). 15.A.Ruprecht,
J.Maddox,
A.J.Stirling,
N.Visaggio,
S.Y.Seah,CharacterizationofnovelacylcoenzymeAdehydrogenasesinvolvedinbacterialsteroiddegradation.J.Bacteriol.197,1360–1367(2015). 16.R.Krithikaetal.,AiclocusrequiredforironacquisitioninMycobacteriumtuberculosis.Proc.Natl.Acad.Sci.U.S.A.103,2069–2074(2006). 17.N.Rani,
S.Hazra,
A.Singh,
A.Surolia,FunctionalannotationofputativefadE9ofMycobacteriumtuberculosisasisobutyryl-CoAdehydrogenaseinvolvedinvalinecatabolism.Int.J.Biol.Macromol.122,45–57(2019). 18.H.Zhangetal.,Genomesequencingof161MycobacteriumtuberculosisisolatesfromChinaidentifiesgenesandintergenicregionsassociatedwithdrugresistance.Nat..45,1255–1260(2013). 19.S.Burbaudetal.,Trehalosepolyphleatesareproducedbyaglycolipidbiosyntheticpathwayconservedacrossicallydistantmycobacteria.CellChem.Biol.23,278–289(2016). 20.O.A.Trivedietal.,Enzymicactivationandtransferoffattyacidsasacyl-adenylatesinmycobacteria.Nature428,441–445(2004). 21.R.Mukherjee,
D.Chatterji,Glycopeptidolipids:Immuno-modulatorsingreasymycobacterialcellenvelope.IUBMBLife64,215–225(2012). 22.J.M.Belardinellietal.,BiosynthesisandtranslocationofunsulfatedacyltrehalosesinMycobacteriumtuberculosis.J.Biol.Chem.289,27952–27965(2014). 23.L.R.Camachoetal.,AnalysisofthephthioceroldimycocerosatelocusofMycobacteriumtuberculosis.Evidencethatthislipidisinvolvedinthecellwallpermeabilitybarrier.J.Biol.Chem.276,19845–19854(2001). 24.T.Schwander,
R.McLean,
J.Zarzycki,
T.J.Erb,Structuralbasisforsubstratespecificityofinyl-CoAdehydrogenase,anunusualmemberoftheacyl-CoAdehydrogenasefamily.J.Biol.Chem.293,1702–1712(2018). 25.Y.Ikeda,
K.Okamura-Ikeda,
K.Tanaka,Purificationandcharacterizationofshortchain,medium-chain,andlong-chainacyl-CoAdehydrogenasesfromratlivermitochondria.Isolationoftheholo-andapoenzymesandconversionoftheapoenzymetotheholoenzyme.J.Biol.Chem.260,1311–1325(1985). 26.K.Izai,
Y.Uchida,
T.Orii,
S.Yamamoto,
T.Hashimoto,Novelfattyacidbeta-oxidationenzymesinratlivermitochondria.I.Purificationandpropertiesofvery-long-chainacyl-coenzymeAdehydrogenase.J.Biol.Chem.267,1027–1033(1992). 27.R.P.McAndrewetal.,Structuralbasisforsubstratefattyacylchainspecificity:Crystalstructureofhumanvery-long-chainacyl-CoAdehydrogenase.J.Biol.Chem.283,9435–9443(2008). 28.D.K.Maetal.,Acyl-CoAdehydrogenasedrivesheatadaptationbysequesteringfattyacids.Cell161,1152–1163(2015). 29.K.P.Battaileetal.,Crystalstructureofratshortchainacyl-CoAplexedwithacetoacetyl-CoA:Comparisonwithotheracyl-CoAdehydrogenases.J.Biol.Chem.277,12200–12207(2002). 30.J.J.Kim,
M.Wang,
R.Paschke,Crystalstructuresofmedium-chainacyl-CoAdehydrogenasefrompiglivermitochondriawithandwithoutsubstrate.Proc.Natl.Acad.Sci.U.S.A.90,7523–7527(1993). 31.N.J.Faergeman,
J.Knudsen,Roleoflong-chainfattyacyl-CoAestersintheregulationofmetabolismandincellsignalling.Biochem.J.323,1–12(1997). 32.E.S.Goetzmanetal.,Expressionandcharacterizationofmutationsinhumanverylong-chainacyl-CoAdehydrogenaseusingaprokaryoticsystem.Mol..Metab.91,138–147(2007). 33.M.Souri,
T.Aoyama,
G.Hoganson,
T.Hashimoto,Very-long-chainacyl-CoAdehydrogenasesubunitassemblestothedimerformonmitochondrialinnermembrane.FEBSLett.426,187–190(1998). 34.S.Guetal.,ComprehensiveproteomicprofilingofthemembraneconstituentsofaMycobacteriumtuberculosisstrain.Mol.Cell.Proteomics2,1284–1296(2003). 35.A.K.Brownetal.,ProbingthemechanismoftheMycobacteriumtuberculosisβ-ketoacyl-acylcarrierproteinsynthaseIIImtFabH:Factors
(1).monlyusedanti-TBdrugsareknowntoinhibitthebiosynthesisofcellponents.Forexample,isoniazidandethionamideareinhibitorsofmycolicacidsynthesis,andethambutolandcycloserineareinhibitorsofarabinogalactanandpeptidoglycansynthesis,respectively
(2).Structuresofseveraldrugtargetsthatareinvolvedinthebiosynthesisandtransportofcellponents,suchasarabinosyltransferasesandMmpL3,havebeendeterminedrecently(3,4).ThearabinosyltransferasesEmbA,EmbB,andEmbCareregardedasthetargetsforethambutol.MmpL3isresponsibleforthetransportofmycolicacids.SQ109,apromisinganti-TBdrug,bindsinsideMmpL3.Structuralinformationofdrugtargetswoulddefinitelypromotethestructure-guideddrugdesign. TheM.tuberculosiscellwallposedofathicklipidcoat,withthelipidcontentountingfor∼60%ofthedryweightofthecellwall.ItspresencecontributestothevirulenceofM.tuberculosis(5,6).Previousstudieshaveshownthatpathogenicmycobacteriacanusefattyacidsastheircarbonsource
(7).Significantly,cholesterolalonecanbeusedastheonlycarbonsource(8,9).IntheM.tuberculosisgenome,morethan250genesareidentifiedtobeinvolvedinlipid,fattyacid,andsterolmetabolism(10).ThepathwaybywhichlipidsaremetabolizedisthereforeofgreatsignificanceinunderstandingthegrowthandvirulenceofM.tuberculosisinthehost. Inlipidmetabolismpathways,acyl-CoAdehydrogenase(ACD)introducesunsaturationintofattyacids,withthecofactoracyl-CoAconvertingtoenoyl-CoA(11).Althoughthereare35annotatedfadEgenesinM.tuberculosis,notallarefunctional ACDs.TwoFadEproteinsencodedbyadjacentfadEgenesareheterotetrameric.Inthoseexamples,aheterotetramerisrequiredtobindtheFADcofactor(12).Theexistenceofsixheterotetramers,fadE23-fadE24,fadE28-fadE29,fadE26-fadE27,fadE31-fadE32,fadE31-fadE33,andfadE17-fadE18,encodedbyfadEgeneshavebeenreported.MostoftheseFadEenzymescharacterizedtodateinM.tuberculosisfunctionincholesterolcatabolismandplayrolesinthedehydrogenationofcholesterolsubstratesthroughβ-oxidation(13).FadE28-FadE29catalyzesthedehydrogenationof3-oxo-4-pregnene-20-carboxyl-CoA(3-OPC-CoA)(14),whileFadE26-FadE27catalyzestheoxidationof3-oxo-cholest-4-en-26oylCoA(13)andFadE34isactivetowardsteroidCoAesterscontainingfive-carbonchains(15).FadE14,characterizedasanovelacyl–acylcarrierproteindehydrogenaseratherthananacylCoAdehydrogenase,isinvolvedinthemycobactinbiosynthesispathway(16),andFadE9isanACDwhosesubstratesarebranchedchainaminoacids(17). Inarecentstudy,84drug-resistance–associatedgeneswereidentifiedthroughgenomesequencingof161differentM.tuberculosisisolates(18).Amongthese,severalwell-knowndrug-targetgenes(gyrA,ethA,rpoB,rpoC,embB,pncA,katG,thyA,andrpsL) Significance Thelipidcontentountsforapproximately60%ofthedryweightofthecellwallofpathogenicmycobacteriawithMycobacteriumtuberculosishavingmorethan250genesinvolvedinfattyacidmetabolism.Previousstudieshaveshownthatacyl-CoAdehydrogenase(ACD),whichintroducesunsaturationintofattyacids,exhibitsstrictsubstratespecificitytowarddifferentCoAthioestergroups.Here,weidentifiedauniqueACDmemberinmycobacteriathatexhibitsbroadsubstratespecificity.Furthermore,wedeterminedcrystalstructuresoftheenzymeandenzyme–plexestoexplainthebroadsubstraterecognitionobservedinthissystem.GiventheimportanceofFadE5infattyacidmetabolism,thesenewstructuresareexcellentplatformsforrationalstructurebasedantituberculosisdrugdiscovery. Authorcontributions:
Z.R.designedresearch;
X.C.,
J.C.,
B.Y.,andW.Z.performedresearch;
X.C.,
X.L.,andZ.R.analyzeddata;andX.C.,
L.W.G.,
X.L.,andZ.R.wrotethepaper. Theauthorsdeclarepetinginterest. ThisarticleisaPNASDirectSubmission. PublishedunderthePNASlicense. Datadeposition:TheatomiccoordinateshavebeendepositedintheProteinDataBank,(PDBIDcodes6KPT,6KRI,6KS9,6KSA,6KSB,6KSE,6LPY,and6LQ0–6LQ8).1Towhomcorrespondencemaybeaddressed.Email:liux@. Thisarticlecontainssupportinginformationonlineat/lookup/suppl/doi:10.1073/pnas.2002835117/-/DCSupplemental. FirstpublishedJune29,2020. DownloadedatInstituteofBiophysicsonAugust30,2020 16324–16332|PNAS|July14,2020|vol.117|no.28 /cgi/doi/10.1073/pnas.2002835117 wereidentified.However,othergenes,includingfadE5,werealsoidentified.ThefadE5geneislocatedonaclusteridentifiedasthefadE5-gap–likegenomicregion(19),whichharborsfadE,pks,papA,fadD,pE,mmpL,andgap-likegenes,andisbroadlydistributedacrossmycobacteria.InMycobacteriumsmegmatis(Ms),thesegeneproductsaresuggestedtoplayrolesinthebiosynthesisoftrehalosepolyphleates,afamilyoftrehalose-basedlipids(19).MsFadE5,actingasanACD,issuggestedtointroduceunsaturationintothefattyacylchainproducedfromtheFASI(fattyacidsynthaseI)system,andthechainisthenextendedandmodifiedbypolyketidesynthases(Pks)(20,21).InM.tuberculosis,similargeneclustershavealsobeenobserved.Sulfolipids(SL)andpolyacyltrehaloses(PAT)arebothlipidsinthecellwallofM.tuberculosis.Geneclusterscontainingpks,papA,fadD,andmmpLarealsoinvolvedintheformationandexportofSLandPAT(22,23).FadE5proteinsarealsosuggestedtoplayrolesinthebiosynthesisofthelipidsinthecellwall. ACDsareidentifiedbythespecificsubstratestheyutilize.ManyACDmembersinvolvedinthemetabolismoflipidsandaminoacidshavebeencharacterized,includingshort-chainacylCoAdehydrogenase(SCAD),medium-chainacyl-CoAdehydrogenase(MCAD),long-chainacyl-CoAdehydrogenase(LCAD),verylong-chainacyl-CoAdehydrogenase(VLCAD),isovaleryl-CoAdehydrogenase(IVD),isobutyryl-CoAdehydrogenase(IBD),branchedshortchainacyl-CoAdehydrogenase(BSCAD),andglutaryl-CoAdehydrogenase(GDH)(24).Ineukaryotes,differentACDmembersexhibitstrictsubstratespecificitiestowarddifferentCoAthioestergroups,whichresultsfromtheirdistinctactive-sitearchitectures.InthoseACDsthatcatalyzethedehydrogenationoflinearacyl-CoA,thelengthofthesubstratecarbonchainthattheenzymecaneptislimitedbythelengthofthesubstratebindingcavity.ThecavitylengthsofSCAD,MCAD,andVLCADare8,12,and24Å,andtheyhaveoptimalspecificitiesforfattyacyl-CoAswithchainsthatcontain4,8,and16carbonatoms,respectively(25–27).MostACDsformposedoffour∼43-kDasubunits.TwoexceptionsareVLCAD,whichisahomodimerconsistingoftwo∼67-kDasubunits(27,28).Theotheristheα2β2heterotetramerencodedbyfadEgenesinM.tuberculosis(13,14). Herein,weisolatedFadE5fromM.tuberculosis(MtbFadE5)andM.smegmatis(MsFadE5).MsFadE5hasbeenshowntobeinvolvedinresistingdruguptake,mostlikelybyaffectingcellwallpermeability.BothMtbFadE5andMsFadE5showbroadselectivityforlinearacyl-CoAsubstrateswithchainsvaryinginlengthsfrom4to22carbons.Importantly,itssubstratespecificityisquitedifferentfromotherhomologs,mostofwhichhaveastrictsubstraterequirement.Tounderstandthisunusualproperty,wedeterminedthecrystalstructuresofMtbFadE5andMsFadE5plexwithvarioussubstratesandMtbFadE5plexwithC18CoA.ThesestructuresprovidethemolecularbasisforthebroadsubstrateselectivityofMtbFadE5andMsFadE5. Results DrugResistanceAssaysofMsFadE5.ToverifytherolesofFadE5incellwalllipidbiosynthesis,drug-resistanceassayswereperformed(Fig.1).Inthisexperiment,thefadE5geneintheM.smegmatisgenomewasdisruptedbysubstitutionofaninternalsegmentwithahygromycin-resistancegene.VerificationofgenedisruptionwasperformedbyPCRfollowedbysequencingofthePCRproducts.TheMsfadE5genewasoverexpressedinthefadE5-deletedcellsplementationstudies.Fig.1AshowsthegrowthrateisthesameforthefadE5-deletedcellsandthewild-typecells.Whentreatedwithethambutol(2μgmL−1)andstreptomycin(0.1μgmL−1),first-lineandsecond-linedrugs,cellslackingfadE5exhibiteddecreaseddrug-resistanceparedwithwild-type.WhenthefadE5genewasoverexpressedinthefadE5-deletedcells,drugresistancetoethambutolwasgainedanddrugresistancetostreptomycinwasevenmoreevident(Fig.1BandC). SubstrateSelectivityofMtbFadE5andMsFadE5.Full-lengthMtbFadE5andMsFadE5wereproducedinE.coliforanalysis.Gelfiltrationrevealedthattheybothformhomodimersinsolution.Thiswasfurtherconfirmedbysedimentationequilibriumanalyticalultracentrifugation(AUC)(SIAppendix,Fig.S1BandC).AsequencealignmentbetweenMtbFadE5,MsFadE5,andotherACDsisshowninSIAppendix,Fig.S1A.ACDsareclassifiedordingtothesubstratestheyutilize.MostACDsform DownloadedatInstituteofBiophysicsonAugust30,2020 BIOCHEMISTRY Fig.1.Drug-resistanceassaysinM.smegmatis.(A)Growthcurvesforwild-typeM.smegmatis[Ms(WT)],MsfadE5-deletedstrain[Ms(ΔfadE::hyg)],and MsfadE5-deletedstrainwithoverexpressionofMsfadE5[Ms(pVV16::fadE5)].(B)Growthcurvesofwild-typeM.smegmatis,MsfadE5-deletedstrain,andMsfadE5-deletedstrainwithoverexpressionofMsfadE5inthepresenceorabsenceof2μgmL−1ethambutol.(C)Growthcurvesofwild-typeM.smegmatis,MsfadE5-deletedstrainandMsfadE5-deletedstrainwithoverexpressionofMsfadE5inthepresenceorabsenceof0.1μgmL−1streptomycin.Eachbaristhe meanandSDofthreemeasurements. Chenetal. PNAS|July14,2020|vol.117|no.28|16325 homo-orposedoffour∼43-kDaprotomers.OnlyVLCADisahomodimerconsistingoftwo∼67-kDaprotomers.FadE5isalsoahomodimerconsistingoftwo66-kDaprotomers.So,weinferredthatFadE5isaVLCADatfirst. However,ourassaysshowedthatMtbFadE5andMsFadE5arebothhighlyactivetowardsubstratesofvaryinglengthsoflinearacyl-CoA,includingC4CoA,C6CoA,C10CoA,C12CoA,C16CoA,andC18CoA(Fig.2AandB).Isothermaltitration DownloadedatInstituteofBiophysicsonAugust30,2020 Fig.2.ActivityassaysforMtbFadE5andMsFadE5andITCstudies.(AandC)ActivityofMtbFadE5withC4CoA,C6CoA,C10CoA,C12CoA,C16CoA,andC18CoA.TheMtbFadE5E447AmutantisinactivewiththeC18CoAsubstrate.ITCwasusedtodetermineinteractionsbetweenMtbFadE5C10CoAandC20CoA.(BandD)ActivityandITCdataforMsFadE5. 16326|/cgi/doi/10.1073/pnas.2002835117 Chenetal. DownloadedatInstituteofBiophysicsonAugust30,2020 BIOCHEMISTRY calorimetry(ITC)wasnextperformedtodeterminetheirbindingaffinitieswiththesubstrates.ThecatalyticbaseformostACDscharacterizedtodateisaglutamateresidue.InSCADitisE368(29),whileinMCADitisE376(30)andinHsVLCADitisE422(27).SequencealignmentofMsFadE5andMtbFadE5withotherACDssuggestthatE447isthecatalyticbaseinMsFadE5andMtbFadE5.Toconfirmthisassumption,weexpressedandpurifiedE447Amutantsforbothenzymesandshowedneitherpossessedactivity(Fig.2AandB),confirmingthisresidueisthecatalyticbaserequiredfordehydrogenation.ITCwasalsoperformedusingsamplesofMtbFadE5E447AandMsFadE5E447A.Theresultsshowedthatthebindingaffinitiesincreasedwiththelengthofthecarbonchain(Fig.2CandDandSIAppendix,Fig.S2).C22CoAbindstoMtbFadE5withaKdvalueof0.16μM,whileC6CoAbindstoMtbFadE5withaKdvalueof196μ
M.SimilarvalueswereobtainedforthebindingaffinitiesofMsFadE5withthesesubstrates(Table1).Thus,bothMtbFadE5andMsFadE5showabroadsubstrateselectivityforacylCoAsbuttendtopreferthelongerratherthanshortersubstrates.Theicparameters(kcatandKm)forrepresentativesubstrateswerealsodetermined(Table2andSIAppendix,Fig.S3).Thefreeconcentrationofacyl-CoAestersisunlikelytoexceed200nMundernormalcellularconditions(31).Therefore,theeffectivenessofFadE5incellsusingthesesubstrateswouldbelow.ThismaybeafactorthatcontributestotheslowgrowthrateofM.tuberculosisandM.smegmatis. OverviewoftheMsFadE5Structure.ToinvestigatethereasonsforthebroadsubstrateselectivityofMtbFadE5andMsFadE5,crystalstructuresofbothenzymesweredetermined.Initially,wesolvedthestructureofMsFadE5(80%sequenceidentitywithMtbFadE5).ThecrystalstructureshowsthatMsFadE5formsahomodimer(Fig.3A).Virtuallyallresiduescouldbefittedtotheelectrondensitymapexceptfortheregionfrom482to489.TheNterminusofMsFadE5consistsof467residuesandsharesacanonicalfoldthatisobservedinothertetramericACDs.Theadditionalresidues,467to611,formanα-helicaldomainattheCterminusthatisnotfoundinothertetramericACDs.TheC-terminalhelicaldomainexistsspecificallyinVLCADs.InhumanVLCADtheC-terminaldomainisresponsibleforbindingtotheinnermitochondrialmembrane(32,33).Residues446to478aredisorderedinthatstructureandarethoughttoberesponsibleforanchoringtothemembrane.ThisisbasedontheobservationthatFadE5waspreviouslyidentifiedinamembranefractionofM.tuberculosis(34).InFadE5,residues482to489aredisorderedandlocatedinthesurfaceoftheC-terminaldomain,suggestingthatthisregionmaybealsoinvolvedinmembrane Table1.TheKdvaluesdeterminedbyITCSubstrate MtbFadE5MsFadE5 C6-CoAC10-CoAC12-CoAC16-CoAC18-CoAC20-CoAC22-CoAC6-CoAC10-CoAC12-CoAC16-CoAC18-CoAC20-CoAC22-CoA Kd(μM) 196±7867.1±16.0 60±9.24.4±0.674.18±0.820.19±0.070.16±0.06316±27420.1±1.324.0±7.73.45±0.310.72±0.200.07±0.050.19±0.14 anchoringandmayonlyeorderedwheninteractingwiththemembrane. ActiveSiteandSubstrateBindingCavity.AFADcofactorispresentineachoftheMsFadE5activesites(Fig.3B).Thesearelocatedatthedimerinterface.InallofthestructurallycharacterizedACDs,thereisacavityclosetotheisoalloxazineoftheboundFAD,whichharborsthecarbonchainofthesubstrate.ThecarbonchainislocatedproximaltotheN5atomoftheisoalloxazineringsinceitisinvolvedintheα-,β-desaturationofthesubstrates.Thelengthofthecavitylimitsthelengthofthesubstratecarbonchain.InSCAD(PDBIDcode1JQI),MCAD(PDBIDcode3MDD),andHsVLCAD(PDBIDcode2UXW),thelengthoftheircavitiesare8,12,and24Å,respectively(27).However,thecavitylengthinMsFadE5isapproximatelyonly12.5Å(Fig.3C).Thepocketthereforeappearstobenotlongenoughtomodatesubstrateswithlongcarbonchains.Thus,theremustbeconformationalchangeswhenthelongersubstratesbind. StructuralBasisfortheBroadSubstrateSelectivityofMsFadE5andMtbFadE5.ThestructuresoftheotherreportedACDsplexwithsubstratesweresolvedbyaddingsubstratestothenativeenzymespriortocrystallization.Inoureffortstocrystallizeenzyme–plexes,wefollowedthesamestrategy.However,wewereonlyabletogeneratecrystalswheretheproducthadbeenreleased.Superimpositionofthisstructureandtheapo-MsFadE5structurerevealssomedifferences(SIAppendix,Fig.S4B).Inparticular,thesidechainsofresiduesF294andD295exhibitconformationalchanges.ThesidechainofE447,thecatalyticbase,isreorientedtofacetotheisoalloxazineringofFAD.R301,whichelectrostaticallyinteractswithE447inapo-MsFadE5,alsoundergoesaconformationalchange.Inthisstructure,theelectrondensityforY446showsthatitadoptstwodifferentconformations:Oneconformationisthesameastheapoformandtheotherconformationwidensthebindingcavityforsubstratestobind(SIAppendix,Fig.S4A). Topreventthereleaseofthesubstrate,weobtainedstructuresofMsFadE5whereE447wasmutatedtoalanine.CrystalscouldthenbeobtainedplexwiththesubstratesC4CoA,C6CoA,C8CoA,C10CoA,C12CoA,C14CoA,C16CoA,C17CoA,C18CoA,C20CoA,andC22CoA(Fig.4andSIAppendix,Fig.S5).ThestructureoftheE447AmutantofMsFadE5plexwithC18CoAprovidesanexampletoexplorethestructuralbasisofthebroadsubstratespecificity.Intotal,thereare11hydrogenbondsbetweenresiduesintheproteinandtheCoAportionofC18CoA(Fig.5B).ConformationalchangeinthesidechainofR301isneededtoallowtheformationofahydrogenbondandthemovementofF294resultsinaπ–πstackinginteraction(3.6Å)withtheadenineoftheCoA(Fig.5C).ThisphenylalanineisnotconservedamongotherACDs,asindicatedinthesequencealignment(SIAppendix,Fig.S1).TheequivalentaromaticaminoacidisonlyfoundinGDH(PDBIDcode3MPI).However,nosimilarπ–πstackinginteractioncouldbefoundinthatprotein.Thus,thisisafeaturespecifictothisenzyme. Toclarifythecontributionoftheinteractingresidues,mutationalanalysiswascarriedoutandtheKdvaluesmeasuredbyITC(SIAppendix,Fig.S6andTableS1).FortheS171A,K225A,andR460AmutantsusingC20CoA,Kdvalueswereseveralfoldslowerthanforthewild-typeenzyme.TheD456A_R460AdoublemutantalsoshowedadecreasedparedtotheR460singlemutant.So,theseresiduesaresuggestedcontributedtothebindingaffinityoftheproteinandsubstrate.However,theR301AandF294AmutantsbothshowedanincreasedbindingaffinityforC20CoA.Wespeculatethatitwouldbeeasierfortheligandtoenterintothepocketintheabsenceofthesetwobulkysidechains.Notably,S171,K225,D456,andR460couldinteractwiththesubstratewithoutanyconformationalchanges. Chenetal. PNAS|July14,2020|vol.117|no.28|16327 Table2.Enzyme Steady-stateicparametersforMsFadE5andMtbFadE5 Substrate kcat(s−1) Km(μM) MsFadE5MtbFadE5MtbFadE5MtbFadE5 C4CoAC4CoAC6CoAC18CoA 0.53±0.061.07±0.120.80±0.140.61±0.04 285.9±53.1358.7±71.8353.0±105.6162.5±28.8 kcat/Km(M−1s−1) (1.85±0.13)×103(2.98±0.26)×103(2.27±0.28)×103(3.75±0.42)×103 However,R301andF294bothundergoconformationchangesuponsubstratebinding.Theinteractionsbetweenthetworesiduesandthesubstratearefacilitatedthroughtheconformationalchanges.So,weinferthatthesetworesiduesmaynotcontributedtotheoriginalrecognitionofthesubstratebutthestabilizationaftersubstratebinding. Assuggestedabove,inplexes,thecavitycanbelengthenedaslongersubstratesbind(Fig.6).Thisisachievedbytheresiduesalongthecavityreorientingthemselvesuponsubstratebinding.IntheMsFadE5–C4CoAstructure,thesidechainofY446rotatestoallowbinding(Fig.6A).Thus,Y446playsatypeofgatekeeperrolesinceinbothenzymes,thisresiduehastorotatetoallowthesubstratetogainess.IntheMsFadE5–C10CoAmodel,thesidechainofM134changesconformationtoallowC10CoAbinding(Fig.6B).However,intheMsFadE5–C12CoAandMsFadE5–plexes,M134undergoesalargerconformationalchangesothatthecavitypletelyopen(Fig.6CandD).Theentirelengthofthecavityishydrophobicsothatthelongercarbonchainsofthesubstratemakeadditionalhydrophobicinteractions,resultinginthepreferenceofFadE5forlong-chainsubstrates.Allinall,thesidechainsoftheresiduesalongthecavitywallguidetherouteandbindingofthecarbonchain.ThesidechainofM130inMsFadE5stretchesintothecavity,forcingthecarbonchaintotakeaturn(SIAppendix,Fig.S7).InMtbFadE5,thisresidueissubstitutedbyaglycine.AstructureoftheE447A_M130GdoublemutantofMsFadE5plexwithC16CoAshowsthatthecarbonchaintakesadifferentparedtothesingleE447mutant(SIAppendix,Fig.S7). Wewereessfulingeneratingapo-MtbFadE5crystals.However,wedidmakecrystalsoftheE447AmutantofMtbFadE5plexwithC18CoA.plexisalsoahomodimer(Fig.7A).ThermsdforallCαatomsbetweentheMtbFadE5E447A–C18CoAandMsFadE5E447A–plexesaftersuperimpositionisonly0.327Å,demonstratingthetwostructuresarevirtuallyidentical(Fig.7B).TheelectrondensityfortheCoAportionofthesubstrateisstrong.However,althoughthedensityforthecarbonchainispoor,itcouldbereasonablyfittedinthe2Fo–Fcmap(Fig.7CandD). ThehydrogenbondsbetweentheMtbFadE5andtheC18CoAaresimilartothoseobservedforMsFadE5anditssubstrate(SIAppendix,TableS2).Theπ–πstackingbetweentheF294andtheadenineofsubstrateisalsopresent.Theresiduesthathaveflexiblesidechainsinthesubstratebindingcavityintheplexadoptsimilarconformationsintheplex,suggestingthesemayundergosimilarconformationalchangeswhenthesubstrateentersthebindingcavity.Thus,itisreasonabletosuggestthatthestructuralbasisofthebroadsubstrateselectivityofMtbFadE5isthesameasthatofMsFadE5.However,thereareseveralaminoacidsubstitutionsinthebindingcavityinparedtoMsFadE5.TheseareM134andF126inMsFadE5,whicharesubstitutedbyF134andW126inMtbFadE5,respectively(Fig.7E).SIAppendix,Fig.S8showshowthesechangesimpactonthebindingmodeofC18CoA.Thesizeandorientationoftheseresiduesidechainsaretotallydifferent,whichmightshapethebindingpocketsanddecidetheconformationofthecarbonchainofC18CoA. DiscussionTodate,ithasalwaysbeensuggestedthatACDshavestrictsubstratespecificitiesandSCAD,MCAD,LCAD,andVLCADarespecificallyactiveagainstfattyacyl-CoAswith4,8,14,and16carbonsinlength,respectively(25–27).ThesizeofthesubstratebindingpocketofACDsdeterminesthemaximumlengthofthecarbonchainthattheproteincanmodate.Thewidthandlengthofthecavityarerestrictedbytheresiduesalongthecavitywall.ThelengthoftheACDH-11bindingcavityis14Å,whichrestrictsthecarbonlengthto12.icassaysalsoindicatethatACDH-11haslittleactivityagainstsubstrateswithchainscontainingmorethan12carbonatoms.ThecavitylengthislimitedbythepresenceofY344andL159.Thestructureoftheapo-formofACDH-11andthestructureofACDH-11plexwithC11CoAdisplaythesameconformationsforY344andL159.ThetemperaturefactorsofthesetworesiduesarerelativelyparedtoalloftheresiduesinACDH-11(28).ThesedatarevealedthatthebindingcavityofunboundACDH-11isnodifferenttothatwhenACDH-11isboundtothesubstrates.Similarly,inSCAD,I251limitsthelengthofthesubstratebindingcavity(29).InMCAD,Q95andE99limitthelengthofthesubstratebindingcavity(30).InthepresentMsFadE5structuresplexwithsubstrates,conformationalchangesoftheresidueslimitingthebindingcavitycouldurtoprovideenoughroomforlongersubstratestobind.Furthermore,intheMsFadE5E447A–C22CoAstructure,thecavityextendstothesurfaceoftheenzymesuggestingthereismoreroomforsubstrateslongerthanC22CoAtobind. HsVLCADshowslittleactivityforsubstrateswithchainlengthsoffewerthan12carbons(26).IntheHsVLCAD–plexstructure,onlytheelectrondensityforthecarbonchainportionofthesubstrateisobservedclearly.ThemissingdensityoftheCoAportionofthesubstrateismostlikelycausedbythehighmobilityoftheCoAmoiety.TheabsenceofhydrogenbondsbetweenHsVLCADandCoAmayresultinthe Fig.3.Overallstructureofapo-MsFadE5anditsactivesite.(A)CartoonrepresentationandConnollysurface(transparent)oftheMsFadE5dimer.Thetwosubunitsareshowninredandblue.FADisshowninstickmodelswithyellowcarbonatoms.(B)2Fo–FcelectrondensitymapforFADcontouredat1.0σ.(C)Surfacerepresentation(transparent)oftheputativesubstratebindingcavity,withFADshownasastickmodel. 16328|/cgi/doi/10.1073/pnas.2002835117 Chenetal. DownloadedatInstituteofBiophysicsonAugust30,2020 Fig.4.StructuresoftheE447AmutantsofMsFadE5plexwithsubstrates.(Left)Cartoonrepresentation(green)oftheE447AmutantofMsFadE5plexwithC4CoA(stickmodelwithpinkcarbonatoms).(Right)2Fo–Fcelectrondensitymaps(contourlevel=1.0σ)forthesubstrates. highmobility(27).WhenC11CoAintheACDH-11–plexstructureissuperimposedwiththeHsVLCAD–C14CoAstructure,onlyonehydrogenbondisobserved.SCADandMCADformfiveandfourhydrogenbondswiththeCoAofthesubstrate,respectively(28).InHsVLCAD,thebindingaffinitybetweentheenzymeandthesubstrateisdeterminedmainlybyhydrophobicinteractionsbetweentheenzymeandthelongacylchains.Therefore,HsVLCADnotonlyeptsacylCoAswithlongacylchainsasasubstratebutalsoshowsa preferenceforthem.IntheMsFadE5–plex,thereare11hydrogenbondsandarareπ–πstackinginteractionbetweenF294andtheadenineofthesubstrate.TheseadditionalbondsseemtobethereasonwhyFadE5caneptsubstrateswithshortcarbonchains.Althoughthehydrophobicinteractionsarenotstrong,theCoAportionoftheshortersubstratescanbindtotheenzyme.Toexplorewhatprecludestheanalogoushydrogenbonds,thestructuresofMsFadE5andHsVLCADpared.InHsVLCAD,theβ-sheetcontainingstrands4 DownloadedatInstituteofBiophysicsonAugust30,2020 BIOCHEMISTRY Fig.5.InteractionsbetweenC18CoAandresiduesintheE447AmutantofMsFadE5.(A)Cartoonrepresentationofthestructure.C18CoAisshownasastickmodelwithcarbonatomscoloredpink.(B)HydrogenbondsbetweenC18CoAandtheresiduesinMsFadE5areshownasdottedlines.DistancesaregiveninÅngstroms.(C)Theresiduesthatinteractwiththeproteinwithreorientedsidechains.(D)SuperimpositionoftheCoAbindingsitesintheHsVLCAD–plex(yellow)andtheE447AmutantofMsFadE5plexwithC18CoA(green).C18CoAisshownasastickmodelwithcarbonatomscoloredpink. Chenetal. PNAS|July14,2020|vol.117|no.28|16329 Fig.6.ConformationalchangesinthesubstratebindingcavitiesoftheE447AmutantofMsFadE5uponsubstratebinding.(
A,Left)Cut-awaysurfacerepresentationofthesubstratebindingcavityintheplex.C4CoAisshownasastickmodelwithcarbonatomscoloredpink.(Right)Conformationalchangesfortheresiduesalongthesubstratebindingcavitywall(green)paredwiththoseintheapostructure(blue).(B)Conformationalchangesintheplex.(C)Conformationalchangesintheplex.(D)ConformationalchangesintheC20CoAstructure. and5extendsfurtherawayfromtheCoAbindingsitethanthatinMsFadE5,andinMsFadE5,thestrandsinthesheetaremuchshorterandfacetowardtheCoAportionofthesubstratesotheresiduescaninteractwithCoA(Fig.5D).Takentogether,theapo-FadE5andplexstructuresrevealthestructuralbasisforitsbroadsubstrateselectivity.Thisenzymecaneptshortcarbonchainacyl-CoAasasubstratesinceitcanformseveralhydrogenbondsandalsohasauniqueπ–πstackingwiththeCoAportionofthesubstrate.Inaddition,FadE5canmodatelonger-chainfatty-acylCoAssincetheresiduesthatpotentiallylimitthelengthofthesubstratebindingcavitycanreorienttheirsidechains. FadE5isahomodimericACDwhosesubstratesarelinearchainfattyacids.M.tuberculosiscanmakeuseoffattyacidsofdifferentlengthsasacarbonsource.Thesefattyacidsaredegradedthroughtheβ-oxidationpathwaytoprovidetheenergyrequiredforthegrowthofM.tuberculosis.Inthefirststepofthe β-oxidationpathway,ACDintroducesunsaturationintothefattyacids,convertingacyl-CoAintoenoyl-CoA.FadE5isprobablyinvolvedinthedegradationoffattyacidsofvaryinglengths.Inmycobacteria,theFASIsystemproducesC14-C26acyl-CoAsthroughthedenovofattyacidbiosynthesis,thenunsaturationandmodificationswouldbeintroducedintotheacyl-CoAs.Insomenontuberculosismycobacteria,suchasM.smegmatis,fadE5locatesinthegeneclustersthatareresponsibleforbiosynthesisofpolyketidelipids,soFadE5mightalsobeinvolvedinintroducingunsaturationintoproductsfromFASIsystem,andthechainisthenextendedbyPks.Here,wereportthatFadE5suseC4CoAtoC22CoAassubstrates.IntheFadE5–plexstructures,wehavededucedthatthesubstratebindingcavitycouldmodatesubstrateslongerthanC22CoA.SimilarfindingshavebeenobservedfortheM.tuberculosisβ-ketoacyl-acylcarrierprotein(ACP)synthaseIII(mtbFabH)(35).MtbFabHisalinkbetweentheFASIandFASIIsystemsofM.tuberculosisandcatalyzeadecarboxylativecondensationinthefirststepofFASIIprocessing.EscherichiacoliFabHandothertypicalFabHsutilizeC2-C6acyl-CoAassubstrates,whileMtbFabHcantolerateawiderangeofC12-C20acyl-CoAsubstrates. ThesubstrateselectivityofFadE5isinordancewithitsrolethatwepredictinintroducingunsaturationintotheproductsofFASI.Moreover,theproductscanbeincorporatedasphospholipidsintothemembrane,orelongatedbyFASIIforthesynthesisofmycolicacids,orelongatedbyspecificPKSstoformplexlipidslikemycoketides,SLs,andphthioceroldimycocerosate,orsynthesizepoundssuchastriacylglycerides(36–39).Asaresult,itissuggestedthatFadE5canplaymultiplerolesinlipidmetabolisminM.tuberculosis.ThelipidsinthecellwallofM.tuberculosisandM.smegmatishavebeenshowntobeinvolvedinformationofthecellwallpermeabilitybarrier.Here,drug-resistanceassayswereperformedonwild-typeM.smegmatis,fadE5-deletedM.smegmatis,andfadE5overexpressedM.smegmatis.ThefadE5-deletedM.smegmatisshowedanobviousdecreasedresistancetoethambutolandstreptomycin.SinceweinferredthatfadE5playsanimportantroleinthebiosynthesisofthelipidsinthecellwall,thedecreaseddrugresistanceofthefadE5-deletedstrainsmayhaveresultedfromtheincreasedcellwallpermeability. Intotal,35FadEmembershavebeenidentifiedinM.tuberculosis;suchalargenumberoffadEgenesinthegenomereflectsthenecessityofM.tuberculosistoloadalargevarietyoflipids.Thus,thedevelopmentofsmall-moleculeinhibitorsagainstthisandcloselyrelatedenzymeshavestrongpotentialtobedevelopedasnewgenerationmultitargetdrugs.Thisstudythereforeadvancestheunderstandingoffattyacidmetabolisminmycobacteria,andtheavailabilityofthestructureandcharacterizationinformationofFadE5shouldfacilitatethediscoveryanddevelopmentofnewantimycobacterialagents. MaterialsandMethods ProteinExpressionandPurification.ThegenesencodingMtbFadE5andMsFadE5areRv0244candMSMEG_0406,andwerePCR-amplifiedfromM.tuberculosisH37RvandM.smegmatismc2155,respectively.Thegenefrag- mentswerethenfusedintothepET28aexpressionvectorwiththeEcoRI andHindIIIsitesbyhomologousbinationusingtheOneStepCloning KittoproduceanN-terminalHis-taggedconstruct.MutationwasintroducedintoMtbFadE5andMsFadE5bythefastmutagenesissystemfromTransGen.ThefusedvectorswerealltransformedintoE.coliBL21Rosettacellsfor expression.ProteinexpressioninLBmediasupplementedwithkanamycin(50mg/mL)at37°Cwasinducedby0.5mMisfopropyl-βyr- anoside(IPTG)whentheOD600reached0.6.Cultureswerethengrownat16°Cfor18h.CellswereharvestedbycentrifugationandresuspendedinMCAC-0buffer(25mMTris·HCl,pH8.0,500mMNaCl,and10%glycerol), followedbylysisusingsonication.Thedebriswasremovedbycentrifugation at18,000rpmfor40min.Supernatantscontainingthetargetproteinswere loadedtwiceontoaNi-NTAcolumn(GEHealthcare)equilibratedwith DownloadedatInstituteofBiophysicsonAugust30,2020 16330|/cgi/doi/10.1073/pnas.2002835117 Chenetal. DownloadedatInstituteofBiophysicsonAugust30,2020 BIOCHEMISTRY Fig.7.StructuresoftheE447AmutantofMtbFadE5plexwithC18CoAanditssubstratebindingcavity.(A)Cartoonrepresentationofthedimer.Individualsubunitsareshowninyellowandred.(B)The2Fo–FcelectrondensitymapforC18CoAcontouredat1.0σ.(C)Equivalent2Fo–FcelectrondensitymapforC18CoAcontouredat0.6σ.(D)Superimpositionoftheplex(green)andplex(redandyellow).(E)Superimpositionofthesubstratebindingcavitiesintheplex(green)andtheplex(yellow).Residuesalongthesubstratebindingcavitywallsthatdifferbetweenthetwostructuresareshowninstickmodels.TheC18CoAinMsFadE5isinpink,andC18CoAinMtbFadE5isinblue. MCAC-0buffer.ThecolumnwassubsequentlywashedwithMCAC-20buffer(MCAC-0supplementedwith20mMimidazole).Then,thetargetproteinswerethenelutedwithMCAC-500buffer(MCAC-0supplementedwith500mMimidazole)andfurtherpurifiedusingaHiTrapQcolumn(GEHealthcare)foranionexchangechromatography.Forthefinalstep,aSuperdexG200column(GEHealthcare)equilibratedwith25mMTrisHCland150mMNaCl,pH8.0,wasusedtopurifythetargetproteinsbygelfiltrationchromatography.TheelutedproteinswereevaluatedbySDS/PAGE.Selenomethionine-substitutedMsFadE5proteinwasexpressedinmethionineauxotrophicE.colistrainB834(DE3)inM9medium,aspreviouslydescribed(40). Drug-ResistanceAssay.M.smegmatisstrainswereculturedin7H9brothsupplementedwith10%(vol/vol)ADC(1,000-mLsolutioncontained9gNaCl,50gBSA,and20gglucose),0.1%(vol/vol)Tween-80,0.5%(vol/vol)glycerin,and5μgmL−1carbenicillin.Cellsweregrownat37°CtoanOD600of0.6∼0.8.ThenthecellsweredilutedtoanOD600of0.1.TheOD600wasmeasuredusing96areaplate(Corning)inthepresenceorabsenceofdrugs.DataanalyseswereperformedusingGraphPadPrism6.0. AnalyticalUltracentrifugation.AUCwascarriedoutusingaBeckmanCoulterOptimaAUCequippedwithanAn60TirotortodeterminethesedimentationvelocitiesofMtbFadE5andMsFadE5.Theproteinsweredilutedinasolutioncontaining25mMTris,pH8.0,and150mMNaCl.Thesampleswereconcentratedto1mg/mL.Sedimentationvelocitydatawerecollectedat42,000rpm.Atotalof99scanswerecollectedevery4minusinginterferenceoptics.Allmeasurementswereconductedat4°
C.ThedatawereanalyzedusingtheprogramSedfitwithacontinuousc(s)distributionmodel(41). ITC.ThebindingaffinitiesofMtbFadE5andMsFadE5forC4CoA,C6CoA,C10CoA,C12CoA,C16CoA,C18CoA,C20CoA,andC22CoAweredeterminedusinganITC-200microcalorimeter(Microcal).Allmeasurementswereperformedat25°
C.ThemeasurementswereperformedusingtheE447AmutantsofMtbFadE5andMsFadE5.Proteinsamplesat0.04∼0.10mMinbuffercontaining25mMTris(pH8.0)and150mMNaClwereloadedinthereactioncell.Foreachbindingassay,substratesataconcentrationof0.4∼0.90mMinthesamebufferweretitratedintotheproteinsampleswithasyringe.Thetitrationconsistedof19injectionsof2.0μLevery120s(exceptforthefirstinjectionof0.4μL). EnzymeAssays.TheactivitiesofMtbFadE5andMsFadE5weremeasuredbasedonthereductionofferroceniumasanelectroneptoratanabsorbanceat300nm(extinctioncoefficientof2.75mM−1cm−1)at25°
C within1minoftheinitiationoftheassay(42,43).Thereactionmixture(100μL)contained25mMTris,pH8.0,150mMNaCl,1μMprotein,and0.5mMferroceniumhexafluorophosphate.Theassayswereinitiatedbytheadditionof0.4mMofvaryingsubstratesincludingC4CoA,C6CoA,C10CoA,C12CoA,C16CoA,andC18CoA.TheactivitiesoftheE447Amutantsweremeasuredinthesameway.TheicparametersofFadE5weredeterminedbymeasuringtheinitialvelocitiesatapHof8.0andatemperatureof25°
C.The100-μLreactionmixturewaspreparedbyadding1mMferroceniumhexafluorophosphate,dilutingdifferentsubstratestothedesiredconcentration.Thereactionwasinitiatedbyaddingenzymetoafinalconcentrationof0.8μ
M.Initialvelocitieswereexaminedatsubstrateconcentrationsfrom50to500μ
M.Michaelis–Mentenplotswereusedtodeterminetheicparameters. CrystallizationandDataCollection.Crystalsweregrownat20°Cbyvapordiffusionmixing1μLofproteinsolutionand1μLofreservoirsolution.Selenomethionine-substitutedMsFadE5crystalswereobtainedin0.19Mmagnesiumformate,15.6%(wt/vol)PEG3350,2%(vol/vol)2-propanol,20mMMES(pH6.0),and40mMCa(OAc)
2.ASAD(single-wavelengthanomalousdispersion)datasetwascollectedat100Konbeamline17UoftheShanghaiSynchrotronRadiationFacility(SSRF,China)fromacryoprotectedselenomethionine-substitutedMsFadE5crystal.CrystalsofnativeMsFadE5weregrownin0.19Mmagnesiumformate,15.6%(wt/vol)PEG3350,0.4M(NH4)2SO4,0.4%(vol/vol)PEG400,and20mMacetatepH5.5.ThedatasetwascollectedonbeamlineBL41XUoftheSPring-8synchrotronradiationfacility(Japan).CrystalsofnativeMsFadE5soakedwithsubstratesweregrownin0.1MHepessodium(pH7.0),2%vol/volPEG400,and2M(NH4)2SO4.Thisdatasetwascollectedonbeamline19UattheSSRF.CrystalsoftheE447AmutantofMsFadE5wereobtainedin0.1MBis-Tris(pH6.5)and2M(NH4)2SO4.Thedatasetwascollectedonbeamline19UattheSSRF.CrystalsoftheE447AmutantofMsFadE5plexwithvarioussubstrateswereobtainedin0.1MHepes-sodium(pH7.0),2%vol/volPEG400,2M(NH4)2SO4and1mMFADbysoakingwithCoAsinthemolarratioof1:
5.Thesedatasetswerecollectedonbeamline18Uand19UattheSSRF.CrystalsoftheE447AM130GdoublemutantofMsFadE5plexwithC16CoAwereobtainedin0.8MNaH2PO4/1.2MK2HPO4,0.1Macetate(pH4.5),1mMFAD,and1.2mMC16CoA.Thedatasetwascollectedonbeamline19UattheSSRF.CrystalsoftheE447AmutantofMtbFadE5plexwithC18CoAwereobtainedbycocrystallizationwithsubstratesatamolarratioof1:
3.Thesecrystalsweregrownin20%(wt/vol)PEG3350and2Mpotassiumnitrate.Thedatasetwascollectedonbeamline19UattheSSRF.ThedatasetswereprocessedwiththeprogramHKL2000(44)andXDSpackage(SIAppendix,TableS3). Chenetal. PNAS|July14,2020|vol.117|no.28|16331 Phasing,ModelBuilding,andRefinement.ASADdatasetfromtheselenomethionine-substitutedMsFadE5crystalwasusedtodeterminephasesusingtheprogramPhenixAutoSolWizardbytheSADmethod(45,46).Twenty-threeseleniumatomswerelocated,andafigureofmeritof0.381wasobtainedfollowingtherefinementofheavyatomparametersandphasecalculations.Then,Phenix.autobuildwasusedformodelbuilding(47).Residues482to489couldnotbetracedduetotheirpoorlyorderedelectrondensity.Thestructureofapo-MsFadE5wasdeterminedbymolecularreplacementwiththePhaser(48)moduleinCCP4(49)usingtheselenomethionine-substitutedMsFadE5modelasasearchtemplate.TheoutputmodelfrommolecularreplacementwassubsequentlysubjectedtoiterativecyclesofmanualmodeladjustmentwithCoot(50)andrefinementwithPhenix.refine(51).ThestructuresofMsFadE5mutantsplexwithsubstrateswerealldeterminedbymolecularreplacementwithPhaserusingtheapo-MsFadE5structureasthesearchtemplate.ThestructureoftheMtbFadE5mutantplexwithC18CoAwasalsodeterminedusingapoMsFadE5asatemplate.Theoutputmodelfrommolecularreplacementwas thenrebuiltwiththeprogramPhenix.autobuild.ThemodelswerefurtherrefinedwithCootandPhenix.refine.AlloftherefinedmodelswerevalidatedbyMolProbity(52).ThephasingandrefinementstatisticsaresummarizedinSIAppendix,TableS3. DataAvailability.ThecoordinatesandstructurefactorshavebeendepositedintheProteinDataBank.TheessioncodesaresummarizedinSIAppendix,TableS3. ACKNOWLEDGMENTS.WethankthestaffmembersoftheShanghaiSynchrotronRadiationFacility(China),aswellasSPring-8(Japan)fortheirhelpwithdatacollection.ThisworkwassupportedbyNationalKeyResearch&DevelopmentProgramofChinaGrant2017YFC0840300andNationalNaturalScienceFoundationofChinaGrant81520108019(toZ.R.),andFundamentalResearchFundsfortheCentralUniversities,NankaiUniversity63191431and63181333(toX.L.).
1.Z.Lou,
X.Zhang,Proteintargetsforstructure-basedanti-Mycobacteriumtuberculosisdrugdiscovery.ProteinCell1,435–442(2010).
2.K.Mdluli,
M.Spigelman,Noveltargetsfortuberculosisdrugdiscovery.Curr.Opin.Pharmacol.6,459–467(2006).
3.L.Zhangetal.,Structuresofcellwallarabinosyltransferaseswiththeanti-tuberculosisdrugethambutol.Science,10.1126/science.aba9102(2020).
4.B.Zhangetal.,CrystalstructuresofmembranetransporterMmpL3,ananti-TBdrugtarget.Cell176,636–648.e13(2019).
5.L.W.Riley,Ofmice,men,andelephants:Mycobacteriumtuberculosiscellenvelopelipidsandpathogenesis.J.Clin.Invest.116,1475–1478(2006).
6.E.C.Hett,
E.J.Rubin,Bacterialgrowthandcelldivision:Amycobacterialperspective.Microbiol.Mol.Biol.Rev.72,126–156(2008).
7.E.J.Muñoz-Elías,
J.D.McKinney,Mycobacteriumtuberculosisisocitratelyases1and2arejointlyrequiredforinvivogrowthandvirulence.Nat.Med.11,638–644(2005).
8.M.F.Wipperman,
N.S.Sampson,
S.T.Thomas,Pathogenroidrage:CholesterolutilizationbyMycobacteriumtuberculosis.Crit.Rev.Biochem.Mol.Biol.49,269–293(2014).
9.H.Bloch,
W.Segal,BiochemicaldifferentiationofMycobacteriumtuberculosisgrowninvivoandinvitro.J.Bacteriol.72,132–141(1956). 10.S.T.Coleetal.,DecipheringthebiologyofMycobacteriumtuberculosisfrompletegenomesequence.Nature393,537–544(1998). 11.D.Clark,RegulationoffattyaciddegradationinEscherichiacoli:Analysisbyoperonfusion.J.Bacteriol.148,521–526(1981). 12.M.F.Wipperman,
M.Yang,
S.T.Thomas,
N.S.Sampson,ShrinkingtheFadEproteomeofMycobacteriumtuberculosis:Insightsintocholesterolmetabolismthroughidentificationofanα2β2heterotetramericacylcoenzymeAdehydrogenasefamily.J.Bacteriol.195,4331–4341(2013). 13.M.Yangetal.,UnravelingcholesterolcatabolisminMycobacteriumtuberculosis:ChsE4ChsE5α2β2acyl-CoAdehydrogenaseinitiatesβ-oxidationof3-Oxo-cholest-4-en-26-oylCoA.ACSInfect.Dis.1,110–125(2015). 14.S.T.Thomas,
N.S.Sampson,Mycobacteriumtuberculosisutilizesauniqueheterotetramericstructurefordehydrogenationofthecholesterolsidechain.Biochemistry52,2895–2904(2013). 15.A.Ruprecht,
J.Maddox,
A.J.Stirling,
N.Visaggio,
S.Y.Seah,CharacterizationofnovelacylcoenzymeAdehydrogenasesinvolvedinbacterialsteroiddegradation.J.Bacteriol.197,1360–1367(2015). 16.R.Krithikaetal.,AiclocusrequiredforironacquisitioninMycobacteriumtuberculosis.Proc.Natl.Acad.Sci.U.S.A.103,2069–2074(2006). 17.N.Rani,
S.Hazra,
A.Singh,
A.Surolia,FunctionalannotationofputativefadE9ofMycobacteriumtuberculosisasisobutyryl-CoAdehydrogenaseinvolvedinvalinecatabolism.Int.J.Biol.Macromol.122,45–57(2019). 18.H.Zhangetal.,Genomesequencingof161MycobacteriumtuberculosisisolatesfromChinaidentifiesgenesandintergenicregionsassociatedwithdrugresistance.Nat..45,1255–1260(2013). 19.S.Burbaudetal.,Trehalosepolyphleatesareproducedbyaglycolipidbiosyntheticpathwayconservedacrossicallydistantmycobacteria.CellChem.Biol.23,278–289(2016). 20.O.A.Trivedietal.,Enzymicactivationandtransferoffattyacidsasacyl-adenylatesinmycobacteria.Nature428,441–445(2004). 21.R.Mukherjee,
D.Chatterji,Glycopeptidolipids:Immuno-modulatorsingreasymycobacterialcellenvelope.IUBMBLife64,215–225(2012). 22.J.M.Belardinellietal.,BiosynthesisandtranslocationofunsulfatedacyltrehalosesinMycobacteriumtuberculosis.J.Biol.Chem.289,27952–27965(2014). 23.L.R.Camachoetal.,AnalysisofthephthioceroldimycocerosatelocusofMycobacteriumtuberculosis.Evidencethatthislipidisinvolvedinthecellwallpermeabilitybarrier.J.Biol.Chem.276,19845–19854(2001). 24.T.Schwander,
R.McLean,
J.Zarzycki,
T.J.Erb,Structuralbasisforsubstratespecificityofinyl-CoAdehydrogenase,anunusualmemberoftheacyl-CoAdehydrogenasefamily.J.Biol.Chem.293,1702–1712(2018). 25.Y.Ikeda,
K.Okamura-Ikeda,
K.Tanaka,Purificationandcharacterizationofshortchain,medium-chain,andlong-chainacyl-CoAdehydrogenasesfromratlivermitochondria.Isolationoftheholo-andapoenzymesandconversionoftheapoenzymetotheholoenzyme.J.Biol.Chem.260,1311–1325(1985). 26.K.Izai,
Y.Uchida,
T.Orii,
S.Yamamoto,
T.Hashimoto,Novelfattyacidbeta-oxidationenzymesinratlivermitochondria.I.Purificationandpropertiesofvery-long-chainacyl-coenzymeAdehydrogenase.J.Biol.Chem.267,1027–1033(1992). 27.R.P.McAndrewetal.,Structuralbasisforsubstratefattyacylchainspecificity:Crystalstructureofhumanvery-long-chainacyl-CoAdehydrogenase.J.Biol.Chem.283,9435–9443(2008). 28.D.K.Maetal.,Acyl-CoAdehydrogenasedrivesheatadaptationbysequesteringfattyacids.Cell161,1152–1163(2015). 29.K.P.Battaileetal.,Crystalstructureofratshortchainacyl-CoAplexedwithacetoacetyl-CoA:Comparisonwithotheracyl-CoAdehydrogenases.J.Biol.Chem.277,12200–12207(2002). 30.J.J.Kim,
M.Wang,
R.Paschke,Crystalstructuresofmedium-chainacyl-CoAdehydrogenasefrompiglivermitochondriawithandwithoutsubstrate.Proc.Natl.Acad.Sci.U.S.A.90,7523–7527(1993). 31.N.J.Faergeman,
J.Knudsen,Roleoflong-chainfattyacyl-CoAestersintheregulationofmetabolismandincellsignalling.Biochem.J.323,1–12(1997). 32.E.S.Goetzmanetal.,Expressionandcharacterizationofmutationsinhumanverylong-chainacyl-CoAdehydrogenaseusingaprokaryoticsystem.Mol..Metab.91,138–147(2007). 33.M.Souri,
T.Aoyama,
G.Hoganson,
T.Hashimoto,Very-long-chainacyl-CoAdehydrogenasesubunitassemblestothedimerformonmitochondrialinnermembrane.FEBSLett.426,187–190(1998). 34.S.Guetal.,ComprehensiveproteomicprofilingofthemembraneconstituentsofaMycobacteriumtuberculosisstrain.Mol.Cell.Proteomics2,1284–1296(2003). 35.A.K.Brownetal.,ProbingthemechanismoftheMycobacteriumtuberculosisβ-ketoacyl-acylcarrierproteinsynthaseIIImtFabH:Factors
声明：
该资讯来自于互联网网友发布，如有侵犯您的权益请联系我们。
