Citation
A distributive peptide cyclase processes multiple microviridin core peptides within a single polypeptide substrate

Material Information

Title:
A distributive peptide cyclase processes multiple microviridin core peptides within a single polypeptide substrate
Series Title:
Nature Communicationsvolume 9, Article number: 1780 (2018) doi:10.1038/s41467-018-04154-3
Creator:
Yi Zhang, Kunhua Li, Guang Yang, Joshua L. McBride, Steven D. Bruner & Yousong Ding
Publisher:
Nature Publishing Group
Publication Date:
Language:
English
Physical Description:
Journal Article

Notes

Abstract:
Ribosomally synthesized and post-translationally modified peptides (RiPPs) are an important family of natural products. Their biosynthesis follows a common scheme in which the leader peptide of a precursor peptide guides the modifications of a single core peptide. Here we describe biochemical studies of the processing of multiple core peptides within a precursor peptide, rare in RiPP biosynthesis. In a cyanobacterial microviridin pathway, an ATP-grasp ligase, AMdnC, installs up to two macrolactones on each of the three core peptides within AMdnA. The enzyme catalysis occurs in a distributive fashion and follows an unstrict N-to-C overall directionality, but a strict order in macrolactonizing each core peptide. Furthermore, AMdnC is catalytically versatile to process unnatural substrates carrying one to four core peptides, and kinetic studies provide insights into its catalytic properties. Collectively, our results reveal a distinct biosynthetic logic of RiPPs, opening up the possibility of modular production via synthetic biology approaches.
Acquisition:
Collected for University of Florida's Institutional Repository by the UFIR Self-Submittal tool. Submitted by Yi Zhang.

Record Information

Source Institution:
University of Florida Institutional Repository
Holding Location:
University of Florida
Rights Management:
Copyright Creator/Rights holder. Permission granted to University of Florida to digitize and display this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.

Downloads

This item is only available as the following downloads:


Full Text

PAGE 1

ARTICLEAdistributivepeptidecyclaseprocessesmultiple microviridincorepeptideswithinasingle polypeptidesubstrateYiZhang1,KunhuaLi 2,GuangYang1,JoshuaL.McBride1,StevenD.Bruner2&YousongDing 1Ribosomallysynthesizedandpost-translationallymodi edpeptides(RiPPs)areanimportant familyofnaturalproducts.Theirbiosynthesisfollowsacommonschemeinwhichtheleader peptideofaprecursorpeptideguidesthemodi cationsofasinglecorepeptide.Herewe describebiochemicalstudiesoftheprocessingofmultiplecorepeptideswithinaprecursor peptide,rareinRiPPbiosynthesis.Inacyanobacterialmicroviridinpathway,anATP-grasp ligase,AMdnC,installsuptotwomacrolactonesoneachofthethreecorepeptideswithin AMdnA.Theenzymecatalysisoccursinadistributivefashionandfollowsanunstrict N -toC overalldirectionality,butastrictorderinmacrolactonizingeachcorepeptide.Furthermore, AMdnCiscatalyticallyversatiletoprocessunnaturalsubstratescarryingonetofourcore peptides,andkineticstudiesprovideinsightsintoitscatalyticproperties.Collectively,our resultsrevealadistinctbiosyntheticlogicofRiPPs,openingupthepossibilityofmodular productionviasyntheticbiologyapproaches. DOI:10.1038/s41467-018-04154-3 OPEN 1DepartmentofMedicinalChemistry,CenterforNaturalProducts,DrugDiscoveryandDevelopment,CollegeofPharmacy,UniversityofFlorida,Gaine sville, Florida32610,USA.2DepartmentofChemistry,UniversityofFlorida,Gainesville,Florida32611,USA.Correspondenceandrequestsformaterialsshouldbe addressedtoY.D.(email: yding@cop.u .edu )NATURECOMMUNICATIONS| (2018) 9:1780 |DOI:10.1038/s41467-018-04154-3|www.nature.com/naturecommunications1 1234567890():,;

PAGE 2

Ribosomallysynthesizedandpost-translationallymodi ed peptides(RiPPs)constituteamajorclassofnaturalproductsthatareincreasinglyrecognizedfortheirbiotechnologicalandbiomedicalapplications1.Agrowinglistofdiverse post-translationalmodi cations(e.g.,macrocyclizations,heterocyclization,prenylation,etc.)2allowsbroadchemicaldiversityof RiPPs,atalowgeneticcostcomparabletononribosomalpeptides,amajorfamilyofpeptidicsecondarymetabolitesproduced throughmodularbiosynthesis3.Microviridinsareauniquefamily ofRiPPsfeaturingatricycliccage-likearchitectureandpossessing potentinhibitoryactivitiestowardtrypsin-typeserineproteases4 – 8.Dysregulationofserineproteasesplaysaprominentroleinthe developmentofmanydiseasessuchascancers,type2diabetes, pulmonarydisease,Alzheimer ’ sdisease,andinfectiousdiseases9 10.Microviridinsofferanovelscaffoldwiththerapeutic potential.Sixteenmicroviridinanalogshavebeenisolatedfrom freshwatercyanobacteriaandshowedsubstantialchemicalvariations4 – 8.Recentenvironmentalsampling11andbioinformatics analysis12 – 14revealwidespreadoccurrenceofmicroviridinrelatedgeneclustersevenbeyondthephylumofcyanobacteria, indicatingimpressivestructuralandbiosyntheticdiversityof microviridinsawaitingthediscovery.Characterizingandharnessingthebiosyntheticlogicofthesegeneclusterscouldleadto theproductionofnovelmicroviridinanalogsforbiomedical applications15. ThebiosynthesisofmicroviridinsisinitiatedbytwoATP-grasp ligases(e.g.,MdnCandMdnB).Theysequentiallyformtwo macrolactonesandonemacrolactambondbetweensidechainsof conservedresiduesofthecorepeptide( T X1 K YP SD X2 D/EE/D ) withinthepolypeptideprecursorMdnA(Fig. 1 )12 16 – 18.ATPgraspenzymestypicallycatalyzeinter-molecularpeptideligations mainlyinprimarymetabolicpathways(e.g.,glycolysisandpurine biosynthesis)19 – 21.Intra-molecularmacrocyclizationofthe microviridincorepeptidewithbothesterandamidelinkagesis newtothisfamilyofenzymesandimportantlyrepresentsa uniquemacrocyclizationstrategyforRiPPbiosynthesis1.We recentlyelucidatedthestructuralbasisofMdnA/B/Cinthe microviridinJpathway16,andrevealedadistinct, -helicalleader peptide/processingenzymeinteraction.Otherprocessing enzymesinthemicroviridinbiosynthesisincludeanunidenti ed proteasethatcleavesofftheprocessedcorepeptide,an N -acetyltransferaseMdnDandanABCtransporterMdnE.Leveraging thetwoATP-graspligases,we16andtheDittmanngroup14 22partiallyreconstructedthebiosynthesisofmicroviridininvitro, whichcanbeausefulapproachtoproduceRiPPanalogswith desirableproperties22 – 25. Processingmultiplecorepeptideswithinasingleprecursor substrateisarareRiPPbiosynthesisstrategythathassofarbeen describedonlyintheproductionofcyanobacterialcyanobactins26,plantcyclotides27andorbitides28,andfungalustiloxins29andphomopsins30.Thecurrentunderstandingofthe underlyingbiosyntheticlogichascompletelycomefromthe seminalstudiesofcyanobactinbiosyntheticpathway25.Aheterocyclase(e.g.,TruD) rstinteractswiththeleaderpeptide primarilythroughacommonprecursorpeptiderecognitionelement31,amechanismsharedbythemajorityofcurrentlyknown RiPPsclassesbutnotmicroviridin16.Next,thisenzymemodi es naturalandunnaturalprecursorpeptideslikelyinaprocessive mannerandfollowinga C -toN directionality25 32,althoughthe natureofenzymesubstratescanaffectenzymecatalyticperformance.Despiteobviousdifferencesinthenatureofbiotransformations(e.g.,substrateandthetypeofchemical reactions),theobservedprocessivityanddirectionalityofcyanobactinheterocyclasepartiallyresemblethemodularbiosynthesisofprimaryandsecondarymetabolitesfattyacids,polyketides, andnonribosomalpeptides,whichgenerallyfollowsthe colinearityruleandreleasesthebiosyntheticintermediatesonly afterthecatalysisofthelastmodule33. Herewereportbiochemicalcharacterizationofthemacrocyclizationofamicroviridinprecursorpeptide(AMdnA)carryingthreecorepeptides.WeshowthatAMdnC,ahomologof macrolactone-formingMdnC16,convertsAMdnAintomultiple speciesrepresentingeachpredictedmacrolactonizationstageon thethreecorepeptides,andthattheprocessingpossessesa uniquecombinationofenzymaticfeaturesasthedistributive natureandtwo-leveldirectionality,offeringavaluableexample forenzymologyinvestigation.Furthermore,weprobetheplasticityofthemicroviridinbiosynthesisastheprocessingofengineeredAMdnAsubstratescarryingonetofourcorepeptidesby AMdnC,andkineticstudiesprovideusefulmechanisticinsights intotheenzymecatalyticproperties. Results AMdnCmodi esAMdnAwithmultiplemacrolactonelinkages .Ourbioinformaticsanalysisofpubliclyavailablegenomic databasediscoveredthemicroviridingeneclustermainlyfrom cyanobacteriabutalsorepresentativesfrom,forexample,Bacteroidetes(e.g., Microscillamarina )andProteobacteria(e.g., Sorangiumcellulosum )(SupplementaryFig. 1 ).Intriguingly,the precursorpeptide(AMdnA)ofoneclusterfromthe lamentous cyanobacterium Anabaena sp.PCC7120containsthreepredicted corepeptides(Fig. 2 a),whileAMdnCsharesover60%aminoacid identitywithmultipleMdnChomologs(SupplementaryFig. 2 ), suggestingthatthissystemcanlikelyofferanewglimpseintothe mechanismofmodularRiPPbiosynthesis.Wethereforeexpressedandpuri edrecombinantAMdnAwitheitheran N -or C terminalHis6-tagin Escherichiacoli (SupplementaryFig. 3 ).TagfreeAMdnAwasproducedbytheenzymaticremovalofthe N His6-tag.Partiallypuri edrecombinantAMdnCwasobtained from E coli C43(DE3)culturewhencoexpressedwithachaperoneplasmidpGro734(SupplementaryFig. 3 ).Substantial efforts35 36toremoveconcomitantproteinsfromAMdnC achievednomeaningfulimprovementofitspurity. WenextexaminedtheprocessingofAMdnAbyAMdnC. Usinghigh-resolution(HR)-MSanalysis,weobservedaclusterof speciesinthereaction,whosemolecularweights(MWs)were smallerthanAMdnAbyrepeatsof18Da(Fig. 2 b).Theformation ofoneintra-molecularlactonebondontheparentmoleculeis re ectedasthelossofonewater(18Da,dehydration ),andthis MSresult,therefore,suggeststheoccurrenceofserialenzymatic macrolactonizationsontheprecursorpeptideAMdnA.Themost abundantspeciesintheenzymereactioncontained vedehydrations(AMdnA5).Sinceuptotwomacrolactonizationsare predictedtotakeplaceononemicroviridincorepeptide,these datasuggesttheprocessingoftwoandhalfcoresbyAMdnC. AMdnCgeneratedthesameproductpro lefromAMdnA carryingthe C -His6-tag(SupplementaryFig. 4 )butnoactivity wasobservedwiththe N -His6-taggedsubstrate.Interestingly,the useoftheminimalleaderpeptideMdnA9 – 22thatactivatesMdnC in trans inourpreviouswork16resultedinupto vedehydrations onthisinactivesubstrate(SupplementaryFig. 5 ).Thisresult suggeststhatthe N -His6-tagnegativelyaffectscatalyticallycritical interactionsbetweentheleaderpeptideofAMdnAandAMdnC16andthecatalysisofAMdnCcanbecontrolledin trans ,afeature sharedwiththeprocessingofthemicroviridinprecursorpeptides containingasinglecorepeptide16 22. WefurtheroptimizedthereactionconditionsofAMdnC (SupplementaryFig. 6 ).Undertheoptimalconditions,one productspecieswithsevendehydrationsappearedinthereaction ofAMdnC,unexpectedasapredictedmaximumofsixwould resultfromfullyprocessingthethreecorepeptideswithin ARTICLENATURECOMMUNICATIONS|DOI:10.1038/s41467-018-04154-32NATURECOMMUNICATIONS| (2018) 9:1780 |DOI:10.1038/s41467-018-04154-3|www.nature.com/naturecommunications

PAGE 3

AMdnA(Fig. 2 b).Twonegativecontrolswithrecombinant chaperoneGroEL(SupplementaryFig. 3 )andboiledAMdnCdid notproduceanydetectableamountofprocessedAMdnAsunder thesamereactionconditions(SupplementaryFig. 7 ).Tofurther probethecatalyticactivityofAMdnC,wemutateditsK165and D280withalanine,whicharepredictedtointeractwithMg2 +and ATP,respectively16,andarestrictlyconservedamongMdnC homologs(SupplementaryFig. 2 ).ThetworecombinantAMdnC mutantshadthesimilarlevelofimpuritytowildtype (SupplementaryFig. 3 )andwerecatalyticallyinactivetoward AMdnA(SupplementaryFig. 8 ).Thisresultsuggestedthe essentialityofMg2 +andATPtotheAMdnCreaction,thesame asotherATP-graspligases,anditeratedthattheformationof detecteddehydrationspeciesintheAMdnCreactionisnotdriven byanyimpuritypresentintheenzymesolution(Supplementary Fig. 3 ). StructuraldeterminationoftheprocessedAMdnA .WedevelopedaproteomicapproachinvolvingGluCendoproteasedigestiontostructurallycharacterizethemajorprocessedAMdnA speciesintheAMdnCreaction.GluCselectivelyandeffectively cleavespeptidebonds C -terminaltoglutamicacidresiduesand theoreticallywouldcleaveoffeachofthethreecorepeptides withinAMdnA(SupplementaryFig. 9 a).LC-HR-MSanalysisof thedigestionmixtureofintactAMdnAsubstrateidenti edthree fragments(F1-3)carryingthecorepeptideM1,M2,andM3, respectively(Fig. 2 c).Wealsoobservedonefragmentcontaining bothM1andM2(F12,Fig. 2 c).Bycontrast,bothF1andF12 almostcompletelydisappearedfromtheLCtraceofthedigested mixtureofprocessedAMdnA,andtwonewmajorpeakswith retentiontimesofabout8.0minand8.7minemerged(Fig. 2 c). HR-MSanalysisassignedthesetwonewpeaksasF22andF32,respectively,andalsoidenti edtwosmallpeakswithretentiontimesof6.6minand7.5minasF11andF123,respectively(Fig. 2 c,SupplementaryFig. 9 ).Therefore,AMdnC catalyzedthecorrespondinglossofone,twoandtwowatersfrom theM1,M2,andM3withinAMdnAasthemajorprocessed species.Theobserveddehydrationpatternsuggestsa1-2-2ring topologyofthemajorspeciesasshowninFig. 2 d.Wefurther validatedthisdeducedstructurebycomparativeMS/MSanalysis (Fig. 3 ).IntandemMStraces,multiplefragmentsweregenerated fromtheM2andM3oftheF2andF3(Fig. 3 a,c).Expectedly, thesefragmentsdisappearedinthetracesoftheF22andF32 becausetheformationoftwomacrolactonesblockedthefragmentationofthecorepeptide(Fig. 3 b,d).Thelargesizesofthe F1andF11ledtounanalyzableMS/MSfragmentationas observedinotherreports37.Finally,wequantitatedthenumberof lactonebondsinthemajorspeciesbyusingthereducingagent LiBH413.Afterreduction,onelactonebondisconvertedintotwo – OHgroups,withadiagnosticmassincreaseof4Da.A20-Da increasewasobserveduponthetreatmentofthereductant, indicatingtheexistenceof velactonesinthemajorspeciesofthe AMdnCreaction(SupplementaryFig. 10 )13. AMdnCprocessesAMdnAinadistributivefashion .TheformationofmultiplespeciesduringvariablestagesofmacrolactonizationstepsindicatesadistributivenatureofAMdnCin processingAMdnA(Fig. 2 b).Tofurtherinvestigatethiscatalytic feature,weloweredthemolarratioofsubstrate/enzymeto115:1 andthenmonitoredthereactioncoursefrom0to16h(Fig. 4 ). After0.5h,AMdnCtransformedAMdnAintothespecieswith onlyoneortwodehydrations,whilethenewspecieswiththreeto vedehydrationsappearedafter2h.Furtherextendingthe reactiontimeto16hledtoamodestshiftoftheproductpro le The formation of two lactones and one lactam ATP grasp ligases MdnC and MdnB The removal of leader peptide N -Terminal acetylation Microviridin J C N H O C O O C IST R K Y P S D W E E W O O Leader peptide I S T R K Y P S D W E E W Leader peptide NHAc O HN O HO HN O NH O HN O NH O OH NH H2N HN O O HN O HN O O N O O H N O HN O HN O O NH NH O N H O H2N+ Fig.1 SchematicrepresentationofmicroviridinJbiosynthesis.MdnCinstallstwomacrolactonesonMdnA,followedbythemacrolactamizationbyMdnB. Themodi edcoreisthenreleasedforan N -acetylationmodi cation NATURECOMMUNICATIONS|DOI:10.1038/s41467-018-04154-3ARTICLENATURECOMMUNICATIONS| (2018) 9:1780 |DOI:10.1038/s41467-018-04154-3|www.nature.com/naturecommunications3

PAGE 4

withtheformationofthespecieswiththelossofsixwaters (Fig. 4 ).Theformationofintermediateswithallpossibledehydrationeventsovertimeindicatesadistributivecatalysisof AMdnC,suggestingAMdnCbindsintactAMdnA,formsa rst macrolactone,dissociatesfromtheprocessedintermediate,and thenrecapturesitaswellasintactAMdnAasthesubstratesfor subsequentmacrolactonizations.Thisprocessisiterativetoward theformationofmoreadvancedspecies(Fig. 4 ).IntheRiPP biosynthesis,onlyfewenzymes,whichtypicallyprocesstheRiPP precursorpeptidescontainingasinglecorepeptide,possessthe distributivecatalyticproperty,e.g.,LctMinvolvedinlantipeptide biosynthesis38,OphAforthemethylationofomphalotin39 40and PsnBforthecyclizationoffourrepeatedunitsofplesiocin41.On theotherhand,TruDdemonstratedtheprocessivecatalysiswhen theunnaturalsubstratewithasinglecyanobactincorepeptide wasusedasthesubstrate32. ThedirectionalityofAMdnCinprocessingAMdnA .ThedistributivenatureofcatalysisbyAMdnCresultsintheaccumulationofintermediatesateachstageofsubstrateprocessing,and offerstheopportunitytoinvestigatethedirectionalityofthe catalysis,anobservedaspectofmodularRiPPbiosynthesis25.To probethisfeature,weterminatedthereactionsafter0,0.5,2,and 16h(Fig. 4 )andthendeterminedthelocationoflactonebondsof processedspeciesbyGluC-basedproteomicanalysis.At0.5h,we observedpeaksforF11,F22,andF31(Fig. 5 ),which collectivelyconstitutetheintermediateswiththelossofoneand twowatersasshowninFig. 4 .ThisresultsuggeststhatAMdnCis abletomacrolactonizeanyofthreecorepeptideswithinAMdnA astheinitialstep.Ontheotherhand,theF1signaldecreasedtoa greaterextentthanF2andF3,indicatingthatAMdnCfavorsthe M1intheprocessing(Fig. 5 ).Inlinewiththisobservation,the signalratioofF11/F1increasedalongthereactioncourse,and wassigni cantlyhigherthanthatoftheF22/F2atalltime points(Fig. 5 ).Furthermore,themajorityoftheM1andM2were processedafter16h,whileasubstantialamountofF3remained intact.At16h,weobservedonlyF32,whoseretentiontimewas thesameasF31intheLC-HRMSanalysis.Theseresults demonstrateanunstrictbutfavored N -toC directionalityof AMdnCinprocessingthreecorepeptideswithinAMdnA.In comparison,LctMandOphAwiththedistributivecatalyticfeatureshowtheunidirectionality,whileTruDstrictlyfollowsthe C toN directionalityinmaturatingtheunnaturalsubstratecarrying asinglecyanobactincorepeptide32.Unidirectionalcatalysisis alsotheprevailingfeatureofmodularbiosynthesisofprimaryand secondarymetabolites42 43.Inthisregard,AMdnCrepresentsa highlyvaluableexamplefortheinvestigationofprocessing directionalityinnaturalproductbiosynthesis. Inadditiontothedirectionalityintheprocessingofmultiple corepeptides,individualcorepeptideswithinAMdnAare macrocyclizedbytwomacrolactones,thereforedemonstrating topologicaldirectionality.Inanearlyreport,MdnAismodi ed rstwiththelargerandthenthesmallermacrolactone(Fig. 1 )44. ToinvestigatethisaspectintheprocessingofAMdnA,wecreated sixAMdnAalaninemutantsbyindividuallymutatingeachacidic residuerequiredfortheformationofthe rst(D51,D76,and D93)orthesecond(E53,E78,andD95)macrolactone(Fig. 2 d, SupplementaryFigs. 3 and 11 ).BothD76AandD93Amutants lostuptothreewatersintheAMdnCreactionasshowninthe LC-HR-MSanalysis(Fig. 6 a),indicatingthatthesemutations blocktheformationofthe rstandsubsequentlythesecond macrolactoneontheM2andM3,respectively(Supplementary Fig. 11 cand 11 e).Bycontrast,onlythesecondmacrolactonewas notformedontheM1,M2,andM3inthecorrespondingE53A, E78A,andD95Amutants(SupplementaryFig. 11 b, 11 d,and 11 f), whichlostupto ve,four,andfourwaters,respectively(Fig. 6 b). Wefurthercon rmedtheringtopologyofthemajorproductsin thesereactionsbytheGluC-basedproteomicanalysis(SupplementaryFig. 12 ).TheseresultsshowthatAMdnCfollowsastrict orderassequentiallyformingthelargerandsmallermacrolactonesoneachcorepeptidewithinAMdnAandtheformation 1246 1250 1254 1258 1262 1266 m / zCal. MW: 11,346.19 Obs. MW: 11,346.14[M5+9H]9+[M7+9H]9+Cal. MW: 11,256.14 Obs. MW: 11,256.181247.65 5.06.07.08.09.0Time (min) F2 F1c b dF1: F12: F2: F3: F3F32 F22F2F11F3 M1 M1 M2 M3 F12F123 M2 1261.69 [M+9H]9+1251.69 ABC transporter Acetyltransferase Unrelated or unknown genes Precursor peptide ATP grasp ligaseaAMdnF AMdnC AMdnB AMdnA AMdnE AMdnD Leader peptide -ester ester ester -ester -ester M2 M1 M3 Fig.2 AMdnCmacrolactonizesthethreecorepeptideswithinAMdnA. a A putativemicroviridingeneclusterwasidenti edfrom Anabaena sp. PCC7120.TheconservedbindingmotifoftheleaderpeptidewithinAMdnA isshadedingreenwhilethreepredictedcorepeptidesareunderlinedwith solidlines. b HR-MSanalysisdetectedanarrayofspecieswithonetoseven dehydrations( )afterincubatingAMdnAwithAMdnCatamolarratioof 45:1for16h.Themostabundantspecieslost vewatersfromAMdnA. c HPLCtracesofintact(up)andprocessed(down)AMdnAsafterGluC digestion.Keychromatographicpeakswerelabeledwiththenamesof correspondingpeptidefragmentsreleasedbyGluC. d Thededuced structureofthemostabundantspeciesintheAMdnCreaction.Keyacidic residuesinvolvedinthelactonizationsofcorepeptideswerenumbered ARTICLENATURECOMMUNICATIONS|DOI:10.1038/s41467-018-04154-34NATURECOMMUNICATIONS| (2018) 9:1780 |DOI:10.1038/s41467-018-04154-3|www.nature.com/naturecommunications

PAGE 5

ofthesecondmacrolactonedependsontheoccurrenceofthe rst one.Weobservedupto vedehydrations,insteadofthe predictedmaximumoffour,fromtheD51Amutantinthe AMdnCreaction(Fig. 6 a).FurtheranalysisoftheAMdnA sequenceledustoidentifya'quasicorepeptide'fromtheintermodularregionoftheM1andM2(Fig. 2 a,d).Thesequenceof thequasicore( T G K LRD E )hassigni cantdeviationsfromthatof thefullcorepeptide( T X1 K YP SD X2 D/EE/D )butcan presumablyformonemacrolactonebetweenitsThrandGlu residues(SupplementaryFig. 11 a).Evidently,GluCdigestionof theprocessedAMdnAD51Aidenti edF123,F22,andF32,butnotF11thatwasfoundfromthedigestedprocessed AMdnA(SupplementaryFig. 12 a).Themacrolactonizationofthe quasicorecanmaketheGluresidueinaccessibletoGluCfor releasingtheF11.Theformationofonemacrolactoneonthe quasicorecanaccountforthelossofthreewaterstogenerate F123fromtheD51A(SupplementaryFig. 12 a)andleadtothe formationofthespecieswithsevendehydrationsfromAMdnA (Fig. 2 b,SupplementaryFig. 4 ).Collectively,theseresultsreveal thetopologicaldirectionalityofAMdnCinprocessingindividual corepeptidesofAMdnA,demonstratetheindependentprocessingofeachcorepeptide,andalsosuggestnoformationofthe macrolactonesacrossdifferentcorepeptides.Importantly,the formationofmultipleintermediatesfromtheAMdnAmutants (Fig. 6 )providesnewlinesofevidenceindicatingboththe distributivenatureandtheunstrictoveralldirectionalityof AMdnCcatalysis. TheprocessingofengineeredAMdnAsubstratesbyAMdnC ThecatalyticversatilityofAMdnCtowardAMdnAalanine mutantsencouragedustofurtherassessitssubstratescope.We rstusedthenaturalmicroviridinprecursorpeptideMdnAcarryingthesinglecorepeptideandobservedtheMdnA2asthe onlyproductoftheAMdnCreactionasshowninLC-HR-MS analysis(Fig. 7 a).Incomparison,MdnCisknowntoinstalltwo macrolactonesonMdnA16.Inthiswork,wefoundthatMdnC generatedtheAMdnA1asthemajorproductandasmall amountofAMdnA2,butnootheradvancedspecies(Fig. 7 b). TheseresultssuggestthesuperiorcatalyticversatilityofAMdnC. TofurtherinvestigatethepromiscuityofAMdnC,weprepared b a c d b182+b8 y172+y162+b152+y152+b7 y7 b6 y6 b5 y5b4 b72+y4 b3 y2 b2 F22+ 100 300 500 700 900 m / z b172+b162+b152+b142+b132+b6 y6 b5y5b4 y3 b3 b2 y1F32+ 100 300 500 700 900m / z [F32]2+b18 *2+y17 *2+y16 *2+b4 b3 b2 y4 y5 y1 y2 100 300 500 700 900 b17 *2+b16 *2+b15 *2+ b14 *2+b13 *2+y6 y5 y3 y2 y72+b2 y1 [F32]2+ 100 300 500 700 900m/z A V T L K S D N E D NG GGE I D Y P F2:I V T L K F P S D D D D QP VGLE F3:O C O O CAV T L K Y P S D N E DN O G GGE DI F22: O C O O CIV T L K F P S D D D DQ O P VGLE F32:m/zb2 b3 b4 b5 b6 b7 b8 b15 b18y17 y16 y15 y7 y6 y5 y4 y2b17y1 y3 y5 y6b16 b15 b14 b13 b6 b5 b4 b3 b2y6b13 *y7 y5 y3 y2 y1b14 b15 b16 b17 b2 b18 y17 y16 b4 b3 b2y5 y4 y2 y1 Fig.3 StructuraldeterminationofAMdnA5byHRtandemMSanalysis. a HR-MS/MSspectrumofthereleasedF2ofintactAMdnA. b HR-MS/MS spectrumofthereleasedF22ofprocessedAMdnA. c HR-MS/MSspectrumofthereleasedF3ofintactAMdnA. d HR-MS/MSspectrumofthereleased F32ofprocessedAMdnA.ThesamplesweretreatedwithGluCfor16h.MacrolactonizationsontheM2andM3preventedtheformationofmultiple fragmentsfromthecorrespondingregionsofF22andF32 [M3+9H]9+ [M+9H]9+[M1+9H]9+[M2+9H]9+[M4+9H]9+[M5+9H]9+[M6+9H]9+0 h 0.5 h 2 h 16 h 124612501254125812621266 m/z 1242 Fig.4 AMdnCprocessesAMdnAinadistributivemanner.HR-MSanalysis showedtheformationofintermediateswithvaryingdegreesof dehydrationsafterincubatingAMdnAwithAMdnCfor0,0.5,2,and16h. Themolarratioofsubstrate/enzymewassetto115:1toassistthedetection ofaccumulatedintermediates NATURECOMMUNICATIONS|DOI:10.1038/s41467-018-04154-3ARTICLENATURECOMMUNICATIONS| (2018) 9:1780 |DOI:10.1038/s41467-018-04154-3|www.nature.com/naturecommunications5

PAGE 6

sixengineeredAMdnAvariantscontainingonetofourcore peptides,M1V1,M1V2,M1M2,M2M3V1,M2M3V2,and M123M3(Fig. 7 c,SupplementaryFig. 3 andSupplementary Table 2 ).Here,M1,M2andM3representedthe rst,second,and thirdcorepeptideofAMdnA(Fig. 2 d),whileV1andV2indicatedthepresenceandabsenceofthequasicorepeptide, respectively.LC-HR-MSanalysisrevealedthatAMdnCsuccessfullyprocessedtheunnaturalsubstratescarryingoneandtwo corepeptides,M1V1,M1M2,andM2M3V1(Fig. 7 d – f),and promoteduptoeightdehydrationsontheM123M3(Fig. 7 g), indicatingtheprocessingofthefourthcorepeptide.Wedidnot observeanyprocessedM1V2andM2M3V2species(SupplementaryFig. 13 ),suggestingthequasicorepeptideofAMdnA mightin uencethecatalysisofAMdnCinanunclearmanner. Nonetheless,theresultsdemonstratetheexceptionalsubstrate promiscuityofAMdnCandlaythebasisforthefurtherinvestigationoftheevolutionofmicroviridinbiosynthesis. KineticcharacterizationofAMdnC .Toprovideadditional mechanisticinsightsintothecatalysisofAMdnC,weperformed kineticstudies.ThedistributivecatalysisofAMdnCcompounded bytheunstrictreactiondirectionalityleadstoanumberof potentialprocessedintermediates(i.e.,thereisamaximumof9 possibleAMdnA3species).Determiningthekineticconstants ofeachspeciesusingLC-MSanalysisisacomplexproblem(each likelyhasvarying Kmand kcat)andtechnicallychallenging,dueto itslowconcentrationinthereactionandthelackofmultiple suitableinternalstandards.Instead,wesoughttomeasurethe grossreactionkinetics.Speci cally,weemployedtheHPLC analysistoaccuratelyquantitatethenetproductionofADPfrom ATPthataccountsforallphosphorylationreactionsbyAMdnC, whichleadtosubsequentmacrocyclizations(Supplementary Fig. 14 a).Controlreactionsthatwerethesameasenzymatic reactionsexceptnosubstratewereusedtosubtracttheADP producedbyautomaticATPhydrolysisoranyATPusing enzymesco-puri edwithAMdnC.Thisapproachdetermineda measured Km(AMdnA)of26.90.2MintheAMdnCreaction (Table 1 ,SupplementaryFigs. 14 band 15 a),comparablewith MdnA( Km= 23.81.2 M)thatwasdeterminedbyquantitating theconcentrationsofMdnA2intheMdnCreaction.Onthe otherhand,theapparent kcat(ATP,12.41.4min 1)ofAMdnC wasabout25-timeshigherthanMdnC(MdnA,0.470.02min 1). TofurtherunderstandthecatalyticperformanceofAMdnC,we includedvaryingconcentrationsofMdnAandM1V1inthe kineticanalysis.Interestingly,wefoundthatAMdnCdisplayedan approximatetwo-foldhigheref ciency( kcat/ Km)towardthese twounnaturalsubstrateswithasinglecorepeptidethanAMdnA whilemaintainingasimilar Kmvalue(Table 1 ,Supplementary Fig. 15 b,c).Toexaminehowtheapproach(thenetproductionof ADP)usedtodeterminetheAMdnCkineticsaffectedthemeasuredkineticconstants,weusedtheHPLCanalysistoquantitate theconcentrationsofMdnA2intheAMdnCreactionwith MdnAasthesubstrate(Fig. 7 a).Wefoundthatthe Kmvalues determinedbythesetwoapproacheswereatthesamelevel(24.4 0.8 Mvs28.11.0 M)whiletheapparent kcat(MdnA,0.7 0.1min 1)was39timeslowerthanthat(ATP,27.40.2min 1) ofthesamereactionwhendeterminedonthebasisofADP production(Table 1 ,SupplementaryFig. 15 d).Thesigni cant differenceoftwoapparent kcatvaluesofthesamereactionsuggestedthatthe nalmacrocyclizationislikelytherate-limiting stepofAMdnCreaction.Importantly,thesamelevelofmeasured KmvalueswithAMdnA,MdnAandM1V1assubstratesindicated thattheleaderpeptideofprecursorpeptidesubstratesisprimarily responsibleforthecatalyticallydetermininginteractionswith AMdnC.Indeed,thebindingaf nities( KD)ofAMdnAand M1V1withAMdnCwere2.7and2.3M,respectively,atthe samelevel(SupplementaryFig. 16 a,b).Furthermore,wecreateda quadraAMdnAmutant(AMdnAi)carryingfouralaninemutationsD51A,T59A,D76A,andD93A,whichareexpectedto correspondinglyabolishthemacrolactonizationontheM1,quasi core,M2andM3.ComparedwithAMdnAandM1V1,this inactiveanalogpossessedaslightlyhigherbindingaf nity( KD= 3.6M)withAMdnC(SupplementaryFig. 16 c).Expectedly,no macrocyclizedspecieswasproducedfromAMdnAibyAMdnC andthenetproductionofADPwasatthesamelevelofthe negativecontrol(SupplementaryFig. 17 a,b).However,we observedthatAMdnAiinhibitedtheenzymaticprocessingof AMdnAinadose-dependentmanner(SupplementaryFig. 17 c). OnlyAMdnA1wasproducedfromAMdnAwhen3.8Mof AMdnAiwereincludedinthereaction.Kineticanalysisfurther revealedAMdnAitobeacompetitiveinhibitorasitincreasedthe KmvaluesofAMdnAwhilethechangesofobserved kcatvalues werenotstatisticallysigni cant(Table 1 ).Therefore,AMdnAi competeswithAMdnAandprocessedspeciesintheinteractions withAMdnC.Assuch,itinhibitsthedistributivecatalysisof AMdnCwithacalculated Kiof1.30.2 M(Supplementary Fig. 17 d).Furthermore,ourkineticstudiesrevealthatthepresenceofmultiplecorepeptideswithinAMdnAdoesnotenhance therateofenzyme-catalyzedATPhydrolysis(Table 1 ).For processivecatalysis,theconcatenationofmultiplemodi cation siteswithinasinglesubstratecouldincreasethecatalyticef ciencyasreducingthesearchdimensionofenzymetoone( N to C or C to N ),asshownwiththerestrictionenzyme Eco RIina previousreport45. Discussions Multiplemodi cationsonthesubstratebythesameenzymecan occurintwoalternativecatalyticmodes,processivityordistributivity46 47.Processivecatalystsassociatewiththeirsubstrates andthenperformmultipleroundsofreactionsuntilthe Time (min)Time (min)Time (min) F1 F2 F3 0 h 0.5 h 2 h 16 h F11 F22 F31 F31 F32 6.76 6.64 7.76 8.08 8.36 8.69 6.57.57.58.58.5 Fig.5 OveralldirectionalityofAMdnCinprocessingAMdnA.Extractedion chromatogramsofthefragmentsofintactandprocessedAMdnAsat differenttimepointsrevealedtheunstrictbutfavored NtoC overall directionalityofAMdnC.F31andF32hadthesameretentiontimesin LC-HR-MSanalysis ARTICLENATURECOMMUNICATIONS|DOI:10.1038/s41467-018-04154-36NATURECOMMUNICATIONS| (2018) 9:1780 |DOI:10.1038/s41467-018-04154-3|www.nature.com/naturecommunications

PAGE 7

formationof nalproducts,whileadistributivecatalystdissociatesfromtheprocessedintermediatesaftereachtransformation.Incomparison,theprocessivecatalysisincreasesthe effectivemolarityofthesubstratebyenforcingtheone-dimension movementofthecatalyst(e.g.,either N to C or C to N directionalityonpeptidicsubstrates),whilethefreediffusioninthreedimensionalspaceisassociatedwithdistributiveenzymessuchas GCN5thatacetylatemultiplelysineresiduesonasinglehistone48andglycogensynthasekinase-3inphosphorylatingmicrotubuleassociatedtauprotein49.Accommodatingly,processivecatalysis requiresanonspeci csubstraterecognition,whichallowsunidirectionalslidingofcatalyst.UsingDNApolymeraseasan example,thebindingislargelymediatedbyelectrostaticinteractionsbetweentheenzymeandthephosphatebackboneand minorgrooveoftheDNAtemplate50.Inthemicroviridinbiosynthesis,werevealedthespeci cbindingbetweenthe -helical regionoftheleaderpeptideandthe 9 10loopofMdnC,which allostericallyactivatestheenzyme16.Thisfeaturedinteractionis likelypreservedforthecatalysisofAMdnC,sinceahomology modeledstructure,includingthe 9 10loop,highlyresembles MdnC(PDBcode 5IG9 ,SupplementaryFig. 18 ).Furthermore, theresultsofserialexperimentsinthisworksupportedthe determiningroleoftheleaderpeptidefortheenzymaticprocessingofcorepeptide,forexample,theMdnA9 – 22activated AMdnCtoprocessanotherwise-inactiveAMdnAwiththe N His6tag(SupplementaryFig. 5 );AMdnCwasabletoprocess MdnAandengineeredAMdnAvariants(Fig. 7 );andAMdnAi competitivelyinhibitedthecatalysisofAMdnC(Supplementary Fig. 16 ).Ontheotherhand,theleaderpeptide/enzymeinteractionmayhaveamerecontributiontothedistributivecatalytic modeofAMdnC;obviously,theMdnA9 – 22isunabletoactasa hingetoconstraintheslidingoftheenzymeonthe N -His6tagged AMdnAtogeneratemultipleprocessedspecies(Supplementary Fig. 5 ). Thedirectionalityinprocessingasinglecorepeptidehasbeen elucidatedinseveralRiPPclasses,suchasthiazole/oxazolemodi edmicrocins(TOMMs)51andlantipeptide38 52 53.The N to C38 54 55, C to N51 52,andnonlinearbutstill-orderedprocessings53allhavebeenreported.Inthiswork,AMdnCdemonstratedatwo-leveldirectionality.Theformationoftwolactones onindividualcoreswithinAMdnAfollowsastringentorder (Fig. 6 ),comparabletothe N to C directionalityofsinglecore peptidesystems,whileprocessingthethreecorepeptideswithin AMdnAbyAMdnCshowedanunstrictdirectionality(Fig. 5 ).As acomparison,theunidirectionalitywasobservedinthemodi cationoftheunnaturalsubstratewithasinglecyanobactincore byTruD32andinthemodularbiosynthesisoffattyacids,polyketides,andnonribosomalpeptides33.Thedirectionalityfeature ofAMdnCmaybelinkedwithitsdistributivecatalysis.Onthe otherhand,whetherandhowotherbiosyntheticenzymes(e.g., AMdnB)mayin uencethemacrolactonizationsofAMdnAis unknown. DistributivecatalysishasbeenreportedwithseveralRiPPs processingenzymesincludingmicrocinB17synthetases54,lantipeptideprocessingenzymesNisB55 56,LctM,HalM238,and LabKC52,OphAforthe N -methylationofomphalotin39,and recentlycharacterizedATP-graspenzymePsnB41.Theseenzymes generallyprocessthesubstratesthatcarryonlyasinglecore peptide,anddisplayacertaindegreeofdirectionality57.Biochemicalstudieswithunnaturalandnaturalcyanobactinprecursorpeptidessuggestthattheheterocyclasescanbeprocessive andunidirectional32andthecatalyticpropertiescanbein uencedsigni cantlybythesubstratesofthereactions58 59.Comparedwiththeseenzymes,AMdnCdistributivelyprocessedthe substratecarryingthethreecorepeptidesandpossessedthetwoleveldirectionality.ThesefeaturessuggestAMdnCasauseful exampleforadvancingtheunderstandingofRiPPsprocessing enzymesingeneral. Inthiswork,wedeterminedcatalytickineticparametersof AMdnCbyquantitatingthenetproductionofADPinthe reaction.Thisapproachcircumventedthetechnicaldif cultyto directlymeasuretheconcentrationsofeachprocessedspeciesin LC-MSanalysis,particularlywhenAMdnAwasusedasthe substrate,andhaspreviouslybeenusedtodeterminethesteady statekineticsofTruD59.Oneassumptionofthisapproach, however,isthattheATPhydrolysisistherate-limitingstepof AMdnCreaction.Ourstudiesrevealedthatthecalculated KmvaluesofAMdnA,M1V1,andMdnAbythisapproachwereat thesamelevelasthoseofMdnAintheMdnCandAMdnC reactions(Table 1 ),whichweredeterminedbyquantitatingthe MdnA2concentrationsbyHPLCanalysisinthecurrentand ourpreviouswork16.Ontheotherhand,theapparentrates( kcat) ofATPhydrolysisintheAMdnCreactionswere20 – 60times fasterthanthosedeterminedbytheHPLCmethod(apparent kcatvaluesvariedfrom0.47min 1to0.7min 1)16.Theseresults 1027.19E78A E78A+AMdnC1022.29 1020.65 1015 1025 m / z 1022.92 1028.47 1015 1025 m / z m / zD76A D76A+AMdnC1025.19 1023.56 1015 1025D93Am / z 1023.56 1028.47 1015 1025D93A+AMdnC1020.65 1019.01 1027.20E53A E53A+AMdnC1015 1025 m / zD51A D51A+AMdnC1028.47 1022.92 1020.28 1015 1025 m / z1028.47 D95A D95A+AMdnCa b[M+11H]11+* * *[M+11H]11+[M4+11H]11+[M5+11H]11+[M2+11H]11+[M3+11H]11+[M3+11H]11+[M+11H]11+[M+11H]11+[M+11H]11+[M+11H]11+[M3+11H]11+[M4+11H]11+[M4+11H]11+[M4+11H]11+[M5+11H]11+ Fig.6 TopologicaldirectionalityofAMdnCinprocessingindividualcore peptidesofAMdnA. a Threekeyacidicresiduesrequiredfortheformation ofthe rstlactoneoneachcorepeptidewithinAMdnAwereindividually substitutedwithalanine.Theresultantswereusedasthesubstratesof AMdnC. b Threekeyacidicresiduesinvolvedintheformationofthesecond lactonewereindividuallysubstitutedwithalaninetocreateAMdnA mutantsasthesubstratesofAMdnC.Themostprocessedintermediates wereshadedingray.UnrelatedminorMSpeakswerelabeledwith* NATURECOMMUNICATIONS|DOI:10.1038/s41467-018-04154-3ARTICLENATURECOMMUNICATIONS| (2018) 9:1780 |DOI:10.1038/s41467-018-04154-3|www.nature.com/naturecommunications7

PAGE 8

suggestthatthestabilityofsubstrate-enzymecomplexisprimarilydeterminedbytheinteractionsoftheleadpeptide/cyclase butthemacrocyclizationislikelytherate-limitingstep,which involvesthenucleophilicattackofthephosphorylatedcarboxylic acidbythe – OHgroup(SupplementaryFigure 14 a).Compared withTruD( Km= ~1 M, kcatforATPhydrolysis = 2.6min 1)59, AMdnCshowedanaround25-timeslesstightinteractionwithits substrate,butabout10-foldhigherrateofATPhydrolysis.The relativelylowstabilityofthesubstrate-AMdnCcomplexcanbe bene cialtothedistributivecatalysisofmultiplecorepeptides. Comparedwithoneglutathionesynthetasefrom Arabidopsis thaliana [ Km( EC) = 395 M, kcat(ATP) = 12.20.3s 1, basedonthenetproductionofADP]60,anATP-graspligaseof theprimarymetabolism,thecatalyticef ciencyofAMdnCwas about20timeslower,consistentwiththeprevious ndingthat enzymesofthesecondarymetabolismaretypically 30-fold slowerthantheircounterpartsofprimarypathways61. Collectively,thisworkindicatesthatAMdnCiscatalytically activetowardAMdnA.Werevealthatthemacrolactonizationsof threecorepeptideswithinAMdnArequirescontinuouscyclesof substratebindingandreleaseandfollowsthetwo-leveldirectionality.Thecombinationofthesetwocatalyticproperties representsavaluableexampletothebetterunderstandingof biosyntheticenzymesofprimaryandsecondarymetabolites.Our workfurtheroffersnewinformationaboutthemodularRiPPs biosynthesis,shedslightintotheevolutionofmicroviridinbiosyntheticsystems,andprovidesaframeworkforfuturesynthetic biologyeffortstoproducemicroviridinanalogsasserineprotease inhibitors. MethodsConstructionofexpressionplasmidsofAMdnCandmutants .The AMdnC gene (Genbank:NC_003276.1REGION: 7321-8292 )wasampli edfromisolated genomicDNAfrom Anabaena sp.PCC7120inaPCRreactionusingprimers AMdnC-FWandAMdnC-RV(SupplementaryTable 1 ).ThePCRproductwas clonedintopET28atoyieldtheexpressionconstructpET28a-AMdnCfollowing standardmolecularbiologyprotocols62.Theinsertintheconstructwassequenced toexcludeerrors.TocreateAMdnCmutants,thepET28a-AMdnCwasusedasthe templateinsite-directedmutagenesisPCRreactionswithprimersshowninSupplementaryTable 1 ConstructionofexpressionplasmidsofAMdnAandmutants .The AMdnA (NC_003276.1REGION: 9429-9731 )genewasampli edasdescribedabove. PlasmidspET28aandpET30bwereusedforexpressingAMdnAwith N -and C His6tag,respectively.TocreateAMdnAmutants,thepET28a-AMdnAwasusedas thetemplateinsite-directedmutagenesisPCRreactionswithprimersshownin SupplementaryTable 1 .TocreateAMdnAvariants,theAMdnAgenewasusedas m / zm / z [M4+12H]12+846.39 [M+12H]12+852.39 [M3+12H]12+847.89 M1M2 840 850 860 [M+11H]11+[M5+11H]11+[M8+11H]11+1171.44 1163.26 1158.35 11201130 M1V1 [M+7H]7+1131.23 [M2+7H]7+[M1+7H]7+1128.66 119012001210 m / z [M+8H]8+1210.83 M2M3V1 [M3+8H]8+[M2+8H]8+1205.83 1203.11 116011701180 m / z 1110 a d b[M+9H]9+1261.69 AMdnA AMdnA+MdnC [M2+7H]7+812.68 [M2+9H]9+1257.65 MdnA MdnA+AMdnC [M+7H]7+817.83 124612541262 806810814818 m/z m / z M2M3V1+AMdnC M123M3 M123M3+AMdnC M1M2+AMdnC M1v1+AMdnCe f g1126.09 AMdnA variants Sequence ( N to C ) M1V1 M1M2 M2M3V1 M123M3 c Fig.7 HR-MSanalysisshowedtheprocessingofnaturalandunnaturalmicroviridinprecursorpeptides. a AMdnCintroducedtwolactonesonMdnA,the precursorpeptideofmicroviridinJ. b MdnCinstalleduptotwolactonesonAMdnA. c SequencesofselectAMdnAvariants.Thecorepeptidesandquasi corepeptidewereshadedingrayandlightcyan,respectively. d ProcessedM1V1, e processedM1M2, f processedM2M3V1,and g processedM123M3in theAMdnCreactionwereshowninHR-MSspectra.Themostprocessedintermediateswereshadedingray Table1KineticparametersofAMdnCinprocessingAMdnA, MdnAandM1V1aSubstrateanalogs Km( M) kcat(min 1) kcat/ Km(min 1 M 1) M1V122.00.220.80.10.950.01 MdnA24.40.827.40.21.10.04 MdnAb28.11.00.700.10.030.001 AMdnA26.91.412.40.70.460.07 AMdnAc38.31.313.60.10.340.01 AMdnAd42.82.512.50.40.280.02 AMdnAe55.53.011.70.30.210.01 AMdnAf87.96.112.90.50.150.01aDatarepresentedmeans.d.( n 3)bkineticparametersweredeterminedbyHPLC-basedquantitationofMdnA2intheAMdnC reactionc – fAMdnCreactionscontained0.55 M,1.1 M,1.9 M,and3.8 MAMdnAi,respectively ARTICLENATURECOMMUNICATIONS|DOI:10.1038/s41467-018-04154-38NATURECOMMUNICATIONS| (2018) 9:1780 |DOI:10.1038/s41467-018-04154-3|www.nature.com/naturecommunications

PAGE 9

thetemplateinthePCRreactionsusingprimersshowninSupplementaryTable 1 Morespeci cally,thefragmentsofM1V1,M1V2,M1M2,andM123M3were ampli edusingtheprimerAMdnA-FWandthecorrespondingreverseprimers.To prepareM2M3V1andM2M3V2,fragmentsoftheleaderpeptideandcorepeptide domainswereseparatelyampli edfromtheAMdnAgeneandthenfusedin overlappingPCRreactions.All nalinsertsweresequencedtoexcludeerrorsand clonedasdescribedabove. InvitroreconstructionoftheAMdnCreaction .TheAMdnCreactionwascarriedoutinthemixture(100 L)containing40 MAMdnA,50mMbuffer,2mM ATP,10mMdivalentcation,and50mMKCl.Optimizationofreactionconditions includedbuffersystems(Tris,phosphate,MOPS,andHEPES),pH(4 – 9),divalent cations(Co2 +,Fe2 +,Ca2 +,Cu2 +,Zn2 +,andMg2 +),additives(5%glycerol,3mM BME,and3mMdithiothreitol),andenzymedoses(0.35 M,0.90 M,and1.6 M), whichwereperformedinasequentialandcumulativemanner.Thereactionswere incubatedat37Cfor16handterminatedbyanequalvolumeofmethanol.The resultingmixtureswerecentrifuged(14,000 g ,20min,4C)and50 Lof supernatantswereusedforHPLCandLC-MSanalysis,whoseparameterswere detailedbelow.Theoptimalreactionmixturecontained50mMHEPES,pH8.0,5% glycerol,2mMATP,10mMMgCl2,50mMKCl,40 MAMdnA,and0.9 M AMdnC. CatalytickineticanalysisofAMdnC .AlinearcurveofADPstandardwas establishedtoquantitatethenetproductionofADPintheAMdnCreactionsby HPLCanalysis.ADPandATPwereseparatedusingsolventA(30 MKH2PO4supplementedwith0.8 Mtetrabutylammoniumphosphate(TBAP),pH5.45)and solventB(acetonitrile/30 MKH2PO4,1:1,v:v,0.8 MTBAP,pH7.0).TheHPLC programincludedthefollowingsteps:thepercentageofsolventBwasmaintained at10%for0.5min,increasedto20%overaperiodof2.5min,andthenheld constantfor4min.ThepercentageofsolventBwasthenincreasedto50%overa4 minperiodandthenheldconstantfor10min.Afterit,thepercentageofsolventB wasloweredto10%over4minandthenheldconstantfor5min.VariousconcentrationsofAMdnA(0to54 M),M1V1(0to53 M),andMdnA(0to42.75 M)wereprocessedwith1.1 MAMdnCunderoptimalconditionsat37Cand thereactionswerequenchedatvarioustimepoints(2 – 40min).Fortheinhibition kineticanalysis,serialconcentrationsofAMdnAi(0,0.55,1.1,1.9,and3.8 M) wereincubatedalongwithAMdnA(0 – 37.5 M)asdescribedabove.ADPconcentrationsinthereactionsweredeterminedaftersubtractingthepeakareasof automaticATPhydrolysisinthenegativecontrol(thesameenzymaticreactions omittingsubstrate).ForLC-basedquantitationofMdnA2,weestablishedthe standardcurveofMdnAandthenmeasuredtheproductconcentrationsbasedon thepeakareas.Kineticconstantswerethencalculatedby ttingthedatatothe Michaelis – Mentenequation v = Vmax[S]/( Km+ [S]).Forcompetitiveinhibition, theequationusedwas v = Vmax[S]/[ Km(1 + [I]/ Ki) + [S]].Allreactionswere performedintriplicate,anddatarepresentedmeans.d. HPLCandhigh-resolutionLC-MSanalysis .AShimadzuProminenceUHPLC system(Kyoto,Japan) ttedwithanAgilentPoroshell120EC-C18column(2.7 m,3.050mm),coupledwithaPDAdetector,wasusedforHPLCanalysis. SolventAwasH2Ocontaining0.1%TFAandsolventBwasCH3CNcontaining 0.1%TFA.SolventBwasappliedwiththefollowinggradient:0 – 1min10%B;1 – 15 min,alineargradientto90%B;18 – 20min,alineargradientto10%Bata owrate of0.3mL/min.Thedetectionwavelengthsweresetat210,254,and365nm.Data fromLC-HR-MSandMS/MSwasobtainedusingaThermoFisherQExactive FocusmassspectrometerequippedwithelectrosprayprobeonUniversalIonMax APIsource.Acetonitrile(B)/water(A)containing0.1%formicacidwereusedas mobilephaseswithalineargradientprogram(10 – 90%solventBover15min)to separatechemicalsbytheabovereversephaseHPLCcolumnata owrateof0.3 mL/min.Apre-washphaseof15minwith10%solventBwasaddedatthe beginningofeachrun,inwhichtheelutewasdivertedtothewastebyadiverting valve.MS1wereacquiredunderFullScanmodeoftheOrbitrap,inwhichamass rangeof m / z 150 – 2000wascoveredanddatawerecollectedinthepositiveion mode.FragmentationwasintroducedbyHCDtechniquewithoptimizedcollision energyrangingfrom20to30.Foreachselectedpeptide,theionwiththehighest intensitywasselectedastheprecursorionforMS/MSanalysis.Othersettingsfor theOrbitrapscanincludedresolutionat15,000andAGCtargetat5105.Full scanmassspectraandtargetedMS/MSspectraforeachofthepre-selectedpeptides wereextractedfromtheraw lesoftheHPLC-MS/MSExperimentIIusingXcalibur ™ 2.1(ThermoScienti c). Dataavailability .Alldatasupportingthe ndingsofthisstudyareavailablefrom thecorrespondingauthoruponreasonablerequest.Datathatsupportthe ndings ofthisstudyhavebeendepositedintheProteinDataBankwiththeaccessioncode 5IG9 andinGenbankwiththeaccessioncode NC_003276.1 .Received:6May2017Accepted:23February2018 References1.Arnison,P.G.etal.Ribosomallysynthesizedandpost-translationally modi edpeptidenaturalproducts:overviewandrecommendationsfora universalnomenclature. Nat.Prod.Rep. 30 ,108 – 160(2013). 2.Ortega,M.A.&vanderDonk,W.A.Newinsightsintothebiosyntheticlogic ofribosomallysynthesizedandpost-translationallymodi edpeptidenatural products. CellChem.Biol. 23 ,31 – 44(2016). 3.McIntosh,J.A.,Donia,M.S.&Schmidt,E.W.Ribosomalpeptidenatural products:bridgingtheribosomalandnonribosomalworlds. Nat.Prod.Rep. 26 ,537 – 559(2009). 4.Okino,T.,Matsuda,H.,Murakami,M.&Yamaguchi,K.Newmicroviridins, elastaseinhibitorsfromtheblue-greenalga Microcystisaeruginosa Tetrahedron 51 ,10679 – 10686(1995). 5.Shin,H.J.,Murakami,M.,Matsuda,H.&Yamaguchi,K.MicroviridinsD-F, serineproteaseinhibitorsfromthecyanobacteriumOscillatoriaagardhii (NIES-204). Tetrahedron 52 ,8159 – 8168(1996). 6.Murakami,M.etal.Microviridins,elastaseinhibitorsfromthe cyanobacterium Nostocminutum (NIES-26). Phytochemistry 45 ,1197 – 1202 (1997). 7.Reshef,V.&Carmeli,S.Newmicroviridinsfromawaterbloomofthe cyanobacterium Microcystisaeruginosa Tetrahedron 62 ,7361 – 7369(2006). 8.Vegman,M.&Carmeli,S.Threeaeruginosinsandamicroviridinfroma bloomassemblyof Microcystis spp.collectedfroma shpondnearKibbutz LehavotHaBashan,Israel. Tetrahedron 70 ,6817 – 6824(2014). 9.Prassas,I.,Eissa,A.,Poda,G.&Diamandis,E.P.Unleashingthetherapeutic potentialofhumankallikrein-relatedserineproteases. Nat.Rev.DrugDiscov. 14 ,732 – 732(2015). 10.Turk,B.Targetingproteases:successes,failuresandfutureprospects. Nat.Rev. DrugDiscov. 5 ,785 – 799(2006). 11.Ziemert,N.,Ishida,K.,Weiz,A.,Hertweck,C.&Dittmann,E.Exploitingthe naturaldiversityofmicroviridingeneclustersfordiscoveryofnoveltricyclic depsipeptides. Appl.Environ.Microbiol. 76 ,3568 – 3574(2010). 12.Ziemert,N.,Ishida,K.,Liaimer,A.,Hertweck,C.&Dittmann,E.Ribosomal synthesisoftricyclicdepsipeptidesinbloom-formingcyanobacteria. Angew. Chem.,Int.Ed. 76,3568 – 3574(2010). 13.Philmus,B.,Christiansen,G.,Yoshida,W.Y.&Hemscheidt,T.K.Posttranslationalmodi cationinmicroviridinbiosynthesis. Chembiochem 9 3066 – 3073(2008). 14.Ahmed,M.N.etal.Phylogenomicanalysisofthemicroviridinbiosynthetic pathwaycoupledwithtargetedchemoenzymaticsynthesisyieldspotent proteaseinhibitors. ACSChem.Biol. 12 ,1538 – 1546(2017). 15.Weiz,A.R.etal.Harnessingtheevolvabilityoftricyclicmicroviridinstodissect protease-inhibitorinteractions. Angew.Chem.Int.Ed. 53 ,3735 – 3738(2014). 16.Li,K.,Condurso,H.L.,Li,G.,Ding,Y.&Bruner,S.D.Structuralbasisfor precursorprotein-directedribosomalpeptidemacrocyclization. Nat.Chem. Biol. 12 ,973 – 979(2016). 17.Hemscheidt,T.K.Microviridinbiosynthesis. MethodsEnzymol. 516 ,25 – 35 (2012). 18.Weiz,A.R.etal.Leaderpeptideandamembraneproteinscaffoldguidethe biosynthesisofthetricyclicpeptidemicroviridin. Chem.Biol. 18 ,1413 – 1421 (2011). 19.Iyer,L.M.,Abhiman,S.,Burroughs,A.M.&Aravind,L.Amidoligaseswith ATP-grasp,glutaminesynthetase-likeandacetyltransferase-likedomains: synthesisofnovelmetabolitesandpeptidemodi cationsofproteins. Mol. Biosyst. 5 ,1636 – 1660(2009). 20.Fawaz,M.V.,Topper,M.E.&Firestine,S.M.TheATP-graspenzymes. Bioorg.Chem. 39 ,185 – 191(2011). 21.Ogasawara,Y.&Dairi,T.BiosynthesisofoligopeptidesusingATP-grasp enzymes. Chem.Eur.J. 23 ,10714 – 10724(2017). 22.Reyna-Gonzalez,E.,Schmid,B.,Petras,D.,Suessmuth,R.D.&Dittmann,E. Leaderpeptide-free invitro reconstitutionofmicroviridinbiosynthesisenables designofsyntheticprotease-targetedlibraries. Angew.Chem.Int.Ed. 55 9398 – 9401(2016). 23.Oman,T.J.,Knerr,P.J.,Bindman,N.A.,Velasquez,J.E.&vanderDonk,W. A.Anengineeredlantibioticsynthetasethatdoesnotrequirealeaderpeptide onitssubstrate. J.Am.Chem.Soc. 134 ,6952 – 6955(2012). 24.Hudson,G.A.,Zhang,Z.G.,Tietz,J.I.,Mitchell,D.A.&vanderDonk,W.A. Invitrobiosynthesisofthecorescaffoldofthethiopeptidethiomuracin. J.Am. Chem.Soc. 137 ,16012 – 16015(2015).25.Sardar,D.,Lin,Z.J.&Schmidt,E.W.ModularityofRiPPenzymesenables designedsynthesisofdecoratedpeptides. Chem.Biol. 22 ,907 – 916(2015). 26.Donia,M.S.,Ravel,J.&Schmidt,E.W.Aglobalassemblylinefor cyanobactins. Nat.Chem.Biol. 4 ,341 – 343(2008). 27.Daly,N.L.,Rosengren,K.J.&Craik,D.J.Discovery,structureandbiological activitiesofcyclotides. Adv.DrugDeliv.Rev. 61 ,918 – 930(2009). 28.Shim,Y.Y.,Young,L.W.,Arnison,P.G.,Gilding,E.&Reaney,M.J.T. Proposedsystematicnomenclaturefororbitides. J.Nat.Prod. 78 ,645 – 652 (2015). NATURECOMMUNICATIONS|DOI:10.1038/s41467-018-04154-3ARTICLENATURECOMMUNICATIONS| (2018) 9:1780 |DOI:10.1038/s41467-018-04154-3|www.nature.com/naturecommunications9

PAGE 10

29.Tsukui,T.etal.Ustiloxins,fungalcyclicpeptides,areribosomallysynthesized inUstilaginoideavirens. Bioinformatics 31 ,981 – 985(2015). 30.Ding,W.etal.Biosyntheticinvestigationofphomopsinsrevealsawidespread pathwayforribosomalnaturalproductsinAscomycetes. Proc.NatlAcad.Sci. USA 113 ,3521 – 3526(2016). 31.Burkhart,B.J.,Hudson,G.A.,Dunbar,K.L.&Mitchell,D.A.Aprevalent peptide-bindingdomainguidesribosomalnaturalproductbiosynthesis. Nat. Chem.Biol. 11 ,564 – 570(2015). 32.Koehnke,J.etal.Thecyanobactinheterocyclaseenzyme:aprocessive adenylasethatoperateswithade nedorderofreaction. Angew.Chem.Int. Ed. 52 ,13991 – 13996(2013). 33.Meier,J.L.&Burkart,M.D.Thechemicalbiologyofmodularbiosynthetic enzymes. Chem.Soc.Rev. 38 ,2012 – 2045(2009). 34.Ding,Y.,Seufert,W.H.,Beck,Z.Q.&Sherman,D.H.Analysisofthe cryptophycinP450epoxidaserevealssubstratetoleranceandcooperativity. J. Am.Chem.Soc. 130 ,5492 – 5498(2008). 35.Rye,H.S.etal.Distinctactionsof cis and trans ATPwithinthedoubleringof thechaperoninGroEL. Nature 388 ,792 – 798(1997). 36.Viitanen,P.V.,Gatenby,A.A.&Lorimer,G.H.Puri edchaperonin60 (GroEL)interactswiththenonnativestatesofamultitudeof Escherichiacoli proteins. ProteinSci. 1 ,363 – 369(1992). 37.Zhang,Y.,Fonslow,B.R.,Shan,B.,Baek,M.-C.&Yates,J.R.IIIProtein analysisbyshotgun/bottom-upproteomics. Chem.Rev. 113 ,2343 – 2394 (2013). 38.Lee,M.V.etal.Distributiveanddirectionalbehavioroflantibioticsynthetases revealedbyhigh-resolutiontandemmassspectrometry. J.Am.Chem.Soc. 131 12258 – 12264(2009). 39.vanderVelden,N.S.etal.Autocatalyticbackbone N -methylationinafamily ofribosomalpeptidenaturalproducts. Nat.Chem.Biol. 13 ,833 – 835(2017). 40.Ramm,S.etal.Aself-sacri cing N -methyltransferaseistheprecursorofthe fungalnaturalproductomphalotin.Angew.Chem.Int.Ed. 56 ,9994 – 9997 (2017). 41.Lee,H.,Park,Y.&Kim,S.Enzymaticcross-linkingofsidechainsgeneratesa modi edpeptidewithfourhairpin-likebicyclicrepeats. Biochemistry 56 4927 – 4930(2017). 42.Staunton,J.&Weissman,K.J.Polyketidebiosynthesis:amillenniumreview. Nat.Prod.Rep. 18 ,380 – 416(2001). 43.Strieker,M.,Tanovic,A.&Marahiel,M.A.Nonribosomalpeptide synthetases:structuresanddynamics. Curr.Opin.Struct.Biol. 20 ,234 – 240 (2010). 44.Philmus,B.,Guerrette,J.P.&Hemscheidt,T.K.Substratespeci cityand scopeofMvdD,aGRASP-likeligasefromthemicroviridinbiosyntheticgene cluster. ACSChem.Biol. 4 ,429 – 434(2009). 45.Surby,M.A.&Reich,N.O.Contributionoffacilitateddiffusionand processivecatalysistoenzymeef ciency:Implicationsforthe Eco RI restriction-modi cationsystem. Biochemistry 35 ,2201 – 2208(1996). 46.vanDongen,S.F.M.etal.Aclamp-likebiohybridcatalystforDNAoxidation. Nat.Chem. 5 ,945 – 951(2013). 47.vanDongen,S.F.M.,Elemans,J.,Rowan,A.E.&Nolte,R.J.M.Processive catalysis. Angew.Chem.Int.Ed. 53 ,11420 – 11428(2014). 48.Kuo,M.H.etal.Transcription-linkedacetylationbyGcn5pofhistonesH3 andH4atspeci clysines. Nature 383 ,269 – 272(1996). 49.Sperber,B.R.,Leight,S.,Goedert,M.&Lee,V.M.Y.Glycogensynthase kinase-3betaphosphorylatestauproteinatmultiplesitesinintactcells. Neurosci.Lett. 197 ,149 – 153(1995). 50.Morales,J.C.&Kool,E.T.Minorgrooveinteractionsbetweenpolymerase andDNA:moreessentialtoreplicationthanWatson-Crickhydrogenbonds? J.Am.Chem.Soc. 121 ,2323 – 2324(1999). 51.Melby,J.O.,Dunbar,K.L.,Trinh,N.Q.&Mitchell,D.A.Selectivity, directionality,andpromiscuityinpeptideprocessingfroma Bacillus spAl Hakamcyclodehydratase. J.Am.Chem.Soc. 134 ,5309 – 5316(2012). 52.Krawczyk,B.,Ensle,P.,Muller,W.M.&Sussmuth,R.D.Deuteriumlabeled peptidesgiveinsightsintothedirectionalityofclassIIIlantibioticsynthetase LabKC. J.Am.Chem.Soc.134 ,9922 – 9925(2012). 53.Jungmann,N.A.,Krawczyk,B.,Tietzmann,M.,Ensle,P.&Sussmuth,R.D. DissectingreactionsofnonlinearprecursorpeptideprocessingoftheclassIII lanthipeptidecurvopeptin. J.Am.Chem.Soc. 136 ,15222 – 15228(2014). 54.Kelleher,N.L.,Hendrickson,C.L.&Walsh,C.T.Posttranslational heterocyclizationofcysteineandserineresiduesintheantibioticMicrocin B17:distributivityanddirectionality. Biochemistry 38 ,15623 – 15630(1999). 55.Ortega,M.A.etal.StructureandmechanismofthetRNA-dependent lantibioticdehydrataseNisB. Nature 517 ,509 – 512(2015). 56.Mavaro,A.etal.Substraterecognitionandspeci cityoftheNisBprotein,the lantibioticdehydrataseinvolvedinnisinbiosynthesis. J.Biol.Chem. 286 30552 – 30560(2011). 57.Yang,X.&vanderDonk,W.A.Ribosomallysynthesizedandposttranslationallymodi edpeptidenaturalproducts:Newinsightsintotherole ofleaderandcorepeptidesduringbiosynthesis. Chem.Eur.J. 19 ,7662 – 7677 (2013). 58.Sardar,D.,Pierce,E.,McIntosh,J.A.&Schmidt,E.W.Recognitionsequences andsubstrateevolutionincyanobactinbiosynthesis. ACSSynth.Biol. 4 167 – 176(2015). 59.McIntosh,J.A.&Schmidt,E.W.Marinemolecularmachines: heterocyclizationincyanobactinbiosynthesis. Chembiochem 11 ,1413 – 1421 (2010). 60.Jez,J.M.&Cahoon,R.E.Kineticmechanismofglutathionesynthetasefrom Arabidopsisthaliana J.Biol.Chem. 279 ,42726 – 42731(2004). 61.Bar-Even,A.etal.Themoderatelyef cientenzyme:evolutionaryand physicochemicaltrendsshapingenzymeparameters. Biochemistry 50 4402 – 4410(2011). 62.Sambrook,J.&Russell,D.W. MolecularCloning:ALaboratoryManual (ColdSpringHarborLaboratory,NewYork,2001).AcknowledgementsWethankProf.CPWolk(MichiganStateUniversity)forthegiftof Anabaena sp. PCC7120.WealsothankDrs.HendrikLuesch,ValeriePaul,andChrisXingforinformativediscussionsandMs.GengnanLiandDr.RanjalaRatnayakefortechnicalsupport. PartofthisworkwassupportedbyAmericaCancerSocietyInstitutionalResearchGrant (Y.D.)andtheDepartmentofMedicinalChemistryattheUniversityofFlorida.Y.D.is anAirForceOf ceofScienti cResearchYoungInvestigator.AuthorcontributionsY.Z.,S.D.B.,andY.D.conceivedtheproject.Y.Z.conductedalltheexperimentswith helpsfromK.L.,G.Y.,andJ.L.M..Y.Z.,S.D.B.,andY.D.analyzedthedataandwrotethe manuscriptwithhelpsfromK.L.,G.Y.,andJ.L.M..Allauthorsapprovedthemanuscript.AdditionalinformationSupplementaryInformation accompaniesthispaperat https://doi.org/10.1038/s41467018-04154-3 Competinginterests: Theauthorsdeclarenocompetinginterests. Reprintsandpermission informationisavailableonlineat http://npg.nature.com/ reprintsandpermissions/ Publisher'snote: SpringerNatureremainsneutralwithregardtojurisdictionalclaimsin publishedmapsandinstitutionalaf liations. OpenAccess ThisarticleislicensedunderaCreativeCommons Attribution4.0InternationalLicense,whichpermitsuse,sharing, adaptation,distributionandreproductioninanymediumorformat,aslongasyougive appropriatecredittotheoriginalauthor(s)andthesource,providealinktotheCreative Commonslicense,andindicateifchangesweremade.Theimagesorotherthirdparty materialinthisarticleareincludedinthearticle ’ sCreativeCommonslicense,unless indicatedotherwiseinacreditlinetothematerial.Ifmaterialisnotincludedinthe article ’ sCreativeCommonslicenseandyourintendeduseisnotpermittedbystatutory regulationorexceedsthepermitteduse,youwillneedtoobtainpermissiondirectlyfrom thecopyrightholder.Toviewacopyofthislicense,visit http://creativecommons.org/ licenses/by/4.0/ TheAuthor(s)2018 ARTICLENATURECOMMUNICATIONS|DOI:10.1038/s41467-018-04154-310NATURECOMMUNICATIONS| (2018) 9:1780 |DOI:10.1038/s41467-018-04154-3|www.nature.com/naturecommunications