Translational initiation in E. coli occurs at the correct sites ...

Shine-Dalgarno motifs have the consensus sequence GGAGG and can base pair with as many as nine nt in the 3' terminal sequence of 16S rRNA ( ... Sharethisarticle Doi Copytoclipboard Citethisarticle KazukiSaito RachelGreen AllenRBuskirk (2020) TranslationalinitiationinE.colioccursatthecorrectsitesgenome-wideintheabsenceofmRNA-rRNAbase-pairing eLife9:e55002. https://doi.org/10.7554/eLife.55002 Copytoclipboard DownloadBibTeX Download.RIS Article Figuresanddata Abstract Introduction Results Discussion Materialsandmethods Dataavailability References Decisionletter Authorresponse Articleandauthorinformation Metrics Shine-Dalgarno(SD)motifsarethoughttoplayanimportantroleintranslationalinitiationinbacteria.Paradoxically,ribosomeprofilingstudiesinE.colishownocorrelationbetweenthestrengthofanmRNA’sSDmotifandhowefficientlyitistranslated.Performingprofilingonribosomeswithalteredanti-Shine-Dalgarnosequences,werevealagenome-widecorrelationbetweenSDstrengthandribosomeoccupancythatwaspreviouslymaskedbyothercontributingfactors.Usingtheantibioticretapamulintotrapinitiationcomplexesatstartcodons,wefindthatthemutantribosomesselectstartsitescorrectly,arguingthatstartsitesarehard-wiredforinitiationthroughtheactionofothermRNAfeatures.WeshowthatA-richsequencesupstreamofstartcodonspromoteinitiation.Takentogether,ourgenome-widestudyrevealsthatSDmotifsarenotnecessaryforribosomestodeterminewhereinitiationoccurs,thoughtheydoaffecthowefficientlyinitiationoccurs. Translationalinitiationisacriticalstepintheregulationofgeneexpressionthatimpactswhichproteinsaresynthesizedandtowhatextent.Unlikeeukaryoticribosomes,whichscanfromthe5’-endofmessagesandgenerallyinitiateatthefirststartcodon,bacterialribosomescaninitiateatanypositionalonganmRNA;thisisacriticalrequirementbecausemanybacterialmRNAsarepolycistronic.Bacterialribosomesmustselectthecorrectstartcodonsamidstavastexcessofpotentialsites(AUG,GUG,andtosomeextentUUG)thathavetobeignored.Notonlydoesinitiationdeterminewheretranslationoccurs(andthereforewhichproteinsaremade),inmostcasestherateofinitiationdeterminesthelevelofproteinoutput.Inbacteria,acommonstrategyforregulatingtranslationistoblockribosomerecruitmenttoanmRNAthroughtheactionofsmallRNAs(Altuviaetal.,1998;Majdalanietal.,1998;Storzetal.,2004),small-moleculebindingriboswitches(Winkleretal.,2002;MandalandBreaker,2004),andregulatoryproteins(Moineetal.,1990;Babitzkeetal.,2009). InitiationratesvaryinresponsetoseveralmRNAfeaturesthatdeterminehoweffectivelyanmRNArecruits30Ssubunitstothestartcodon.Thermodynamicallystablesecondarystructuressurroundingtheinitiationsiteprevent30Srecruitment(Halletal.,1982;deSmitandvanDuin,1990).ThekineticsofRNAfoldingandunfoldingarealsocritical(deSmitandvanDuin,2003;EspahBorujeniandSalis,2016):somestructuresexistinanunfoldedstateforsuchashortperiodoftimethat30Ssubunitscannotfindthestartcodonquicklyenoughbydiffusionalone.Inseveralwell-characterizedexamples,regionsofsingle-strandedRNAknownasstandby-sitesarefoundnearby,positioning30SsubunitsincloseproximitysothattheycanefficientlycapturethestartcodonuponunfoldingofthemRNAsecondarystructure(deSmitandvanDuin,2003;EspahBorujenietal.,2014).Interactionsof30Ssubunitsandsingle-strandedmRNAregions(especiallythosethatareAU-rich)canbemediatedthroughribosomeproteinS1(Bonietal.,1991;Komarovaetal.,2005).Boundonthebackofthe30Ssubunit,theS1proteincontainsmultipleRNA-bindingdomainsthatcanrecruitmRNAandmeltsecondarystructures(Quetal.,2012),facilitatinghybridizationof16SrRNAwithcomplementarymRNAsequencescolloquiallyknownasShine-Dalgarnomotifs. Shine-DalgarnomotifshavetheconsensussequenceGGAGGandcanbasepairwithasmanyasninentinthe3’terminalsequenceof16SrRNA(ACCUCCUUAinE.coli)referredtoastheanti-ShineDalgarnoorASD(ShineandDalgarno,1974).PairingoftheSD-ASDsequencescanrecruit30Ssubunitstothestartcodon5–10ntdownstream(SteitzandJakes,1975).SDmotifsthatdiffersignificantlyfromtheconsensusorthatarepositionedtoocloseortoofarfromthestartcodonyieldlowerlevelsofinitiation.Indeed,manyexperimentsusingreportergenesshowedthatraisingtheSD-ASDaffinityincreasesproteinoutput,demonstratingitsimportancefordeterminingtranslationlevels(HuianddeBoer,1987;Jacobetal.,1987;deSmitandvanDuin,1990;Salisetal.,2009).Inaddition,theSDmodelservesasthefoundationofpracticalbioengineeringeffortsrangingfromoptimizingexpressionofrecombinantproteinstoexpansionofthegeneticcode(RackhamandChin,2005;Salisetal.,2009). Ontheotherhand,eventhoughtheASDin16SrRNAisalmostuniversallyconservedthroughoutthebacterialkingdom(Nakagawaetal.,2010),thepercentageofgeneswithSDmotifsvarieswidelybetweenspecies.Whilewell-characterizedmodelspeciessuchasE.coliandB.subtilishaveahighpercentageofgeneswithSDmotifs(54%and78%respectively),thereislittletonoenrichmentofSDmotifsupstreamofstartcodonsinBacteriodetesandCyanobacteria(Nakagawaetal.,2010).Inaddition,althoughthemajorityofspeciesinthephylaFirmicutes,Actinobacteria,andProteobacteriahavehighpercentagesofSD-containinggenes,severalspecieshavelowpercentages,arguingthatthelossofthismechanismhasoccurredmultipletimesduringevolution(Nakagawaetal.,2010;Hockenberryetal.,2017).Thesevariationsacrossthebacterialkingdom,despitethehighconservationoftheASDelementontheribosome,raisequestionsastohowimportanttheSDmechanismisforribosomerecruitment. Ribosomeprofilingisamethodfordeepsequencingofribosome-protectedmRNAfragmentsthatallowsustodefinethepositionandnumberofribosomesboundacrossthetranscriptomeatnucleotideresolution(Ingoliaetal.,2009).ThisinformationallowsustocalculatetheribosomedensityoneachmRNAasaproxyfortheefficiencyoftranslationinitiation.Inpioneeringribosomeprofilingstudiesinbacteria,theparadoxicalobservationwasmadethatthereislittleornocorrelationbetweentheribosomeoccupancyofageneandthestrengthofitsSDmotif(calculatedusingthermodynamicalgorithmsforRNApairing),ashadbeenanticipatedbasedontheSDmodel(Lietal.,2014;Schraderetal.,2014;Li,2015;DelCampoetal.,2015).ThissurprisingobservationsuggestedthatothermRNAfeaturescouldeffectivelymasktheeffectsoftheSDcorrelationatthegenome-widelevel. ToisolatetheeffectsofSDmotifsontheglobaltranslationallandscape,weexpressed16SrRNAmutantswithaltered(non-functional)ASDsequences,purifiedmutantribosomes,andusedribosomeprofilingtoaskhowefficientlytheytranslateeachmRNAinthecell.UnlikepreviousstudiesthatvarytheSDmotifandothermRNA-specificfeatures,thisapproachallowsustospecificallyeliminatetheSD-ASDinteractionwhilekeepingmRNAsequencesandstructuresintact,sothatwecanspecificallyaskquestionsabouttheroletheSD-ASDinteractionplaysindeterminingmRNAtranslationrates.Throughthisanalysis,weobserveforthefirsttimetheeffectsofSDmotifsatthegloballevel,revealingalinearcorrelationbetweenSDstrengthandribosomeoccupancy.Wethencombinedournewprofilingapproachwithretapamulintreatmenttotrapribosomesatstartcodons(Meydanetal.,2019;Weaveretal.,2019)inordertostudytheroleofSDmotifsinselectingstartcodons.Tooursurprise,theASD-mutantribosomesselectivelyrecognizethecorrectinitiationsitesaswellaswild-typeribosomes,arguingthatthesesitesarehard-wiredforinitiationindependentoftheirSD-ASDpairingstrength.WeshowthatA-richsequencesrecentlyidentifiedbyFredrickandco-workers(Baezetal.,2019)areenrichedatannotatedstartsitescomparedtootherAUGcodonsinthetranscriptomewhereinitiationdoesnottakeplace;theseA-richsequencesarealsofoundupstreamofstartcodonsinawidevarietyofspeciesacrossthebacterialkingdom.Inaddition,mRNAstructureatannotatedstartsitesislowerthanatotherAUGcodons,facilitating30Sbinding.Together,thesestudiesrefineourunderstandingoftheroleofSDmotifsandothermRNAfeaturesindefiningtheproteomesofbacteria. StudiesoftheroleofSDmotifsinpromotingtranslationintheirnativecontextshavebeencomplicatedbythefactthatchangingthesequenceofanmRNAalsoaffectsotherdeterminantsoftranslationalregulationsuchasitsoverallstructure.ToperturbthefunctionofSDmotifsatthegloballevel,wedevelopedanewapproachinwhichwemutatetheASDin16SrRNA,purifythemutantribosomes,anduseribosomeprofilingtoaskhowefficientlytheytranslateeachmRNAinthecell.Thisstrategyprovidesuswithagenome-wideviewofthefunctionofSDmotifsininteractionswiththeunalteredtranscriptome—allofthefeaturesofanmRNAthataffectsitstranslationaremaintained,therebyisolatingtheeffectsoftheSDmotifmutation.Inthismanner,weeliminatetheSD-ASDinteractionasacontributiontomRNAtranslationratesandseehowtranslationchangesacrossthetranscriptome. Wecreatedthree16SrRNAallelesinwhichtheASDismutated(Figure1A).Twoofthesemutantsweredescribedpreviouslyintheliterature.TheASDinspecialized(S)ribosomeswasinvertedfromCCUCCtoGGAGGinapioneeringstudybydeBoerwhoshowedthatalthoughtheseS-ribosomeswererelativelyinactiveonendogenoustranscripts,theyefficientlytranslateareportergenewithacomplementarySDmotif(HuianddeBoer,1987).Inlaterstudies,CunninghamandChinusedgeneticselectionstocharacterizeadditionalSD-ASDpairsandimprovetheirselectivity,creatingorthogonal(O)ribosomeswheretheASDismutatedtoUGGGA(Leeetal.,1996;RackhamandChin,2005).RibosomeswithmutantASDmotifs(likeSandO)havebeenusedinnumerousstudiesofproteinsynthesiswheretheyselectivelytranslatereportergeneswithcomplementarySDmotifs(Rexetal.,1994;Neumannetal.,2010;Orelleetal.,2015).InadditiontothesetwoASDmutants,weconstructedathird(A)withtheASDsequenceAAAAAthatweanticipatedwouldbindmRNAmoreweaklythantheO-orS-ribosomes(giventhattheirASDsequencesareG-rich).TheMS2aptamerwasinsertedintothesethreeASDmutantstofacilitatetheirpurificationasdescribedbelow(Youngmanetal.,2004;YoungmanandGreen,2005);asacontrol,wealsocreatedanMS2-tagged16SrRNAwiththecanonicalASDsequence(C). Figure1with2supplementsseeall Downloadasset Openasset CapturingtheroleofSDmotifsbyMS2RP. (A)ASDmutationsatthe3’-endof16SrRNAarehighlightedincolor.(B)SchematicofMS2RP:polysomesarecollapsedtomonosomesbyRNaseT1digestion,MS2-taggedmonosomesarepulleddownwiththeMS2coat-protein,andmRNAisfullydigestedtoyieldribosomefootprintsthataresubjectedtodeepsequencing.(C)RT-PCRof16SrRNAfromcelllysates(L)andtheeluate(E)fromtheMS2coat-proteincolumn.(D)Scatterplotofribosomeoccupancy(RO),theratioofribosomeprofilingtoRNA-seqreads,fromMS2RPofO-ribosomesvs.C-ribosomes.Theredlineindicatesa10-foldenrichmentandthePearsoncorrelationisgiven.(E)Ribosomefootprints(inreadspermillionmappedreads)fromMS2RPofO-ribosomesandC-ribosomesonthehemAgene.ThesequenceupstreamofthestartcodonispredictedtopairwiththeASDofO-ribosomes. Thesefour16SrRNAmutantswereexpressedfromplasmidsinE.coliMG1655containingthenormalcomplementofsevenwild-typerRNAoperonstosustaingrowth.BecauseoverexpressionofASDmutantsistoxic(Jacobetal.,1987),weinducedexpressionforonly20–25minduringwhichgrowthrateswerenotaffected(Figure1—figuresupplement1A).Polysomeprofilesfromthefourmutantsweresimilar(Figure1—figuresupplement1B)suggestingthattranslationremainsrobustduringthetransientexpressionofMS2-tagged16SmutantswhethertheASDisintact(C)ormutated(S,O,andA).ApreviousstudyoforthogonalribosomessuggestedthatalteringtheASDin16SrRNAreducesrRNAprocessingefficiency,leadingtotheaccumulationofprocessingintermediates,butthatmaturerRNAscontainingASDmutationshavethecorrect3’-end(Aleksashinetal.,2019).Tolookforprocessingdefectsinoursystem,weperformedRNA-seqonaffinity-capturedMS2-taggedrRNAwithoutnucleasedigestion.AsshowninFigure1—figuresupplement1C,wedonotobservetheaccumulationofprecursorswith3’-extensionsorotherdefectsintheprocessingofthe3’-endof16SrRNA.ThisresultindicatesthatcorrectlyprocessedrRNAisproducedandshouldbeabletoformmature30Ssubunits. RT-PCRwithprimersthatdistinguishendogenous16SrRNAfromtheMS2-taggedmutantswasusedtoaskwhethertheASDmutantsarefoundinactivelytranslatingpolysomes(Figure1—figuresupplement1D).WeobservedthatthesignalfromC-ribosomesisequallystronginthelysate,light,andheavypolysomefractions.Incontrast,thesignalfromthethreeASDmutantsispresentbutweakerinthepolysomefractionsthaninthelysate.ThesedatashowthatalthoughribosomeswithmutantASDscanengageintranslation,theiractivityisimpaired,consistentwithearlierstudies.Keepingthisinmind,wefocusouranalysesnotontheirabsoluteactivitybutontheirselectivity,askingwhichmRNAstheytranslatebetterthanothermRNAs. Topurifymutantribosomes,weemployedamethodpreviouslydevelopedforinvitrobiochemicalstudiesofribosomeswithlethalmutations(Youngmanetal.,2004;YoungmanandGreen,2005):theMS2aptamerwasfusedtohelix6of16SrRNAallowingustocapturemutantribosomesthroughtheirinteractionwiththeMS2coatprotein(Figure1B).Toavoidpullingdownwild-typeribosomesboundtothesamemRNAasmutantribosomes,wefirsttreatedcelllysateswithRNaseT1tocollapsepolysomestomonosomespriortoisolatingMS2-taggedribosomes.RT-PCRrevealshowwellthispurificationstrategyworks:althoughsignalfromthewild-type16SrRNApredominatesincelllysates(lowerband,Figure1C),itisnearlyundetectableinpurifiedribosomesampleselutedfromtheMS2-coatproteincolumn.ThesedatashowthatMS2-taggedribosomescanbeisolatedwithhighpurityforribosomeprofilingstudies;werefertothisprocedureasMS2RP. Comparisonofthetranslationallandscapeofthecanonical(C)totheorthogonal(O)-mutantconfirmsthattheMS2RPstrategyiseffective.For2217geneswithadequatecoverageineachsample,wecomputedribosomeoccupancy(RO)valuesbydividingtheribosomeprofilingdensitybyRNA-seqdensity.AlthoughwerecognizethatROisnotaperfectmeasureofinitiationrates—itmayalsoreflectdifferencesinelongationinsomecases—thenumberofribosomefootprintscorrelatesstronglywithproteinlevelsinexponentiallygrowingE.colicultures(Lietal.,2014);ROthereforereportsonthelevelofproteinoutputpermRNA.WeobservedcompellingdifferencesinROvaluesformanygenesinthetwosamples(Figure1D).AninitialstraightforwardexpectationisthatgeneswithSDmotifswithhighaffinitytoorthogonal(O)ASDsequencewouldhavehighROvaluesinMS2RPdatafromO-ribosomes;indeed,weobservethatacomplementarySDmotif(UCCCG)fiventupstreamofthestartcodongivesthehemAgene10-foldhigherROwiththeO-ribosomethanwiththeC-ribosome(Figure1E).ThesamephenomenonwasobservedonrbsK(7-foldhigherRO)andmreB(10-foldhigherRO)withtheO-ribosomeandonsapA(9-foldhigherRO)andrsmH(4-foldhigherRO)withtheS-ribosome(Figure1—figuresupplement2).Ineachoftheseexamples,theincreaseinROcanbeattributedtohigherlevelsoftranslationbecausethemRNAdiffersbylessthantwo-fold.Theseexamplesarequiterare,however,becauseendogenousgeneshaveevolvedtointeractwiththecanonicalASDandsotheprobabilityoffindingasequencewithstrongcomplementaritytothemutantASDatjusttherightpositionisrelativelylow.Indeed,ourdataaremostconsistentwiththeconclusionthatallthreeASDmutantsessentiallyactasgenerallossoffunctionmutants. WenextusedMS2RPtoisolatetheeffectofSDmotifsonglobaltranslation,askingtowhatextenttheydrivetranslationunderoptimalgrowthconditions.Foreachgene,wecomputedtheSDstrengthastheinverseofthefreeenergy(-∆G)ofpairingbetweenthesequence−15to−6ntupstreamofthestartcodonandthewild-typeASD(ACCUCCU).Basedonthewell-knownroleofSDmotifsinpromotingtranslationalinitiation,theexpectationisthatgeneswithstrongaffinityshouldhavehighROvalues,andconversely,geneswithweakaffinityshouldhavelowROvalues,yieldingastrongcorrelation.However,ouranalysisofdatafromcanonical(C)ribosomesshowedonlyaveryweakcorrelation(Figure2A),consistentwithpreviousreportsfromribosomeprofilingstudies(Lietal.,2014)showingthatSDstrengthhaslittlepowertopredictribosomeoccupancyinE.coli.Strikingly,theROvaluesfromthethreeASDmutants(S,O,andA)showedarobustnegativecorrelationwithSDaffinityforthewild-typeASDsequence(Figure2BandFigure2—figuresupplement1).Inotherwords,ASDmutantribosomestranslategeneswithweakSDmotifsbetterthangeneswithstrongSDmotifs. Figure2with2supplementsseeall Downloadasset Openasset MS2RPrevealsthatSDmotifsenhancetranslationgenome-wide. Ribosomeoccupancy(RO)istheratioofribosomeprofilingtoRNA-seqreadspergene.Log10ROvaluesareplottedagainsttheSDstrength(-∆Gofpairingtothewild-typeASD)foreachgenewithMS2RPdataforC-ribosomes(A)andA-ribosomes(B).(C)Scatterplotof∆logRO(C-ribosomesminusA-ribosomes)and-∆GwherervaluesindicatePearsoncorrelations. BecausetheASDmutantsareunlikelytoparticipateinSD-ASDinteractions,ROvaluesinthesesamplesreflectthecontributionsofalltheothermRNAelementsthatpromoteinitiation.TheobservationthattheseotherelementsyieldanegativecorrelationwithSDstrengthsuggeststhattheyingeneralcounteractthepositivecorrelationcontributedbySD-ASDpairing(withwild-typeribosomes).Assuch,thesecontributionseffectivelymasktheeffectofSDmotifsinFigure2A.BycalculatingthedifferenceinRO(∆logRO)foreachgenebetweentheC-andA-ribosomes,weeffectivelysubtractallthemRNAelementsthatdetermineROindependentofSD-ASDpairing,thusisolatingtheeffectsoftheSDmotifsonmRNAtranslationrates.The∆logROtermreflectshowmuchbetteramessageistranslatedbywild-typeribosomesthanbyASDmutants.When∆logROvaluesareplottedasafunctionofSD-ASDaffinity(-∆G)usingthewild-typeASDsequence,weobserveastronglinearcorrelationwithSD-ASDaffinityforeachofthemutants(Figure2CandFigure2—figuresupplement1).Asexpected,geneswithstrongSDmotifsaretranslatedbetterbyribosomeswiththecanonicalASDthanbyASD-mutantribosomes.ThefactthatweobservethiscorrelationvalidatesourcalculationsofSDstrength;analysisofthedistanceofSDmotifsfromthestartcodonconfirmsthatgeneswiththehighest∆logROhavethestrongestSDaffinityinthe−15to−6regionasshowninpreviousstudies(Figure2—figuresupplement2).ThesedataobtainedwithMS2RPrevealforthefirsttimetheeffectofSDmotifsontranslationgenome-wide,consistentwiththeircharacterizedroleinpromotinginitiation. SDmotifsarealsowidelyheldtoplayacriticalroleinrecognizingandselectinginitiationsites(SteitzandJakes,1975).Intheanalysesdescribedsofar,wehaveusedMS2RPtoestimatetheribosomedensityoneachmRNAasaproxyforinitiationratesinordertoaddressquestionsabouthowmuchtranslationisoccurringonannotatedgenes.Thesedataarelessinformativeaboutthedegreetowhichmutantribosomesinitiateatthewrongsitesinthetranscriptome.Non-canonicalinitiationisdifficulttoobserveinE.colibecause5’-and3’-untranslatedregionsofmRNAsaregenerallyquiteshortandtranslationatalternatestartcodonswithinORFsisswampedbythesignalofelongatingribosomesfromthecanonicalstartsite.Ineukaryotes,theantibioticsharringtonineandlactimidomycinhavebeenusedwithgreatsuccesstogetherwithribosomeprofilingtoidentifysiteswheretranslationalinitiationtakesplace(Ingoliaetal.,2011;Leeetal.,2012).Thesecompoundsdonotinterferewithelongatingribosomes,allowingthemtocontinuetranslationandterminatenormally.Incontrast,theytrapnewly-initiatedribosomes,providingawayofidentifyinginitiationsitesinribosomeprofilingstudies.Twoantibioticswererecentlyshowntosimilarlyspecificallytrapinitiationcomplexesinbacteria:Onc112andretapamulin(Meydanetal.,2019;Weaveretal.,2019). TostudytheroleofSDmotifsonstartcodonselection,wetreatedcellswithretapamulinfor5minandthenusedMS2RPtoidentifystartsitesoccupiedbyribosomeswiththevariousASDsequences.Forexample,elongatingwild-type(C)ribosomesarefoundallacrossthelppgeneinuntreatedcells(Figure3A,lightgrey),whereastheyarehighlyenrichedattheannotatedstartcodoninretapamulin-treatedcells(darkgrey).Asexpected,ribosomefootprintsarenotseenatthreeinternalAUGcodons,sincethesedonotfunctionasinitiationsites.Strikingly,inretapamulin-treatedcells,theA-ribosomesalsofindthecorrectstartsite,ignoringthethreeotherAUGcodons(Figure3A,darkgreen).Inanotherexample,thegmkgene,bothC-andA-ribosomesareenrichedattheannotatedstartcodoninretapamulin-treatedcellsbutnotatseveralinternalAUGcodons(Figure3B).Inbothexamples,bothWTandmutantribosomesselectthecorrect,annotatedstartsitewhileignoringotherAUGcodons. Figure3with1supplementseeall Downloadasset Openasset LossofSD-ASDpairinghaslittleeffectonstartcodonselection. (A,B)RibosomefootprintsonlppandgmkfromMS2RPdataobtainedwithandwithoutretapamulin,anantibioticthattrapsribosomesatstartcodons.AnnotatedAUGsareindicatedbyaredbar,non-annotatedAUGsareindicatedbybluebars.(C,D)Averageribosomeprotectedfragments(RPFs)atannotatedAUGsandnon-annotatedAUGs(whereAUGstartsat1).(E)AverageRPFsatthestartcodonofgeneswhoseribosome-bindingsiteshavelittleornoaffinitytoallthreemutantASDsequences. ToanalyzetheaccuracyofstartcodonselectionbytheASDvariantsinretapamulin-treatedsamplesgenome-wide,wecomputedtheaveragenumberofribosomefootprintsacrossmanygenesalignedattheirannotatedstartcodonsoralignedatalltheotherAUGtripletsinthetranscriptome(non-annotatedAUGs).OurinitialexpectationwasthatintheabsenceofSD-ASDbasepairing,themutantribosomesmightfailtorecognizethecorrectstartsitesandbindmoreoftentootherAUGtripletsinthetranscriptome.Strikingly,boththeC-andA-ribosomesshowstronginitiationpeaksatannotatedAUGs(Figure3C),whereasthesepeaksareabsentinbothsamplesatnon-annotatedAUGs(Figure3D).Theseresultsprovideinitialevidencethatribosomescorrectlyselectannotatedstartsitesgenome-wideintheabsenceoftheSD-ASDinteraction. Tofurtherexplorethissurprisingfinding,wenextaskedhowtheaffinityofmRNA-rRNAbasepairinginfluencesinitiationatannotatedstartcodons.Weassumedthatforthemutantribosomes,basepairingwouldplaylittleornoroleininitiationbecausetheywouldlikelyhavelowaffinityforannotatedstartsitesthatevolvedtobindthewild-typeASD.Totestthisassumption,wecalculatedtheaffinityofeachmutantASDforthesequenceupstreamofthestartcodonofeachgene.WegroupedgenesintodifferentsetsbasedontheseaffinitiesandplottedtheaveragenumberofribosomefootprintsattheannotatedstartsitesasinFigure3C.InthesubsetofgeneswithnopredictedaffinityforanyofthethreeASDmutants(ΔG>−1),westillseerobustenrichmentofA,O,andSribosomesattheannotatedstartsites(Figure3E).SinceallthreeASDvariantsinitiateatannotatedstartsites,thesedataargueagainstthepossibilitythatserendipitousbase-pairingbetweenthemRNAandthemutantASDsequencesexplainsthisenrichment. Wealsoanalyzedasetofannotatedstartsiteswithstrongcalculatedaffinitytothewild-typeASD.ThesesitesareexpectedtobedependentontheSD-ASDinteraction.YetweagainobservedrobuststartpeaksforeachASDvariantribosome,indicatingthatSD-ASDpairingisdispensableforinitiationeveningeneswithstrongSDmotifs(Figure3—figuresupplement1A).Furthermore,wefoundthatinasetofsiteswithpredictedhighaffinitytotheASDoftheO-ribosome,therewasstrongenrichmentofA-andS-ribosomesatstartcodons,despitethedifferencesintheASDsequence(Figure3—figuresupplement1B).Likewise,inasetofgeneswithpredictedhighaffinitytotheASDoftheS-ribosome,therewasstrongenrichmentofO-andA-ribosomesatstartcodons(Figure3—figuresupplement1C).(ThereweretoofewgeneswithhighaffinitytotheA-richASDsequencetoperformanequivalentanalysisforA-ribosomes).Takentogether,theseanalysesshowthatannotatedinitiationsitesarehard-wiredforinitiationindependentoftheirpotentialforbasepairingbetweenthemRNAandrRNA. WenextaskedwhatrolemRNA-rRNApairingplaysininitiationatAUGtripletsinthetranscriptomethatarenotnormallyusedforinitiation(non-annotatedAUGs).Forthispurpose,weuseddatafromretapamulin-treatedcellstocalculateaninitiationscore(IS)foreachAUGtriplet,definedastheaveragenumberofreadsmappedwithin3to21ntdownstreamofanAUG(tocapturefootprintsofvarioussizes)dividedbytheaveragenumberofreadsmappedoverawiderspacing(100nt,Figure4A).Thefirstandmostgeneralfindingisthatthelog2ISvaluesfromtheC-andA-ribosomeshaveasimilardistributionwithmedianscloseto0(Figure4B),indicatingthatfootprintsfromtheA-ribosomesarenotenrichedatnon-annotatedAUGcodons.ThisresultisconsistentwiththeaveragegeneplotshowninFigure3DandwiththefactthatmostoftheseAUGcodonsdonotserveasinitiationsites.TobettercharacterizethedifferencebetweenC-andA-ribosomesininitiationatnon-annotatedAUGcodons,weselectedasubsetofsitesthateffectivelyrecruitC-ribosomesandyieldstronginitiationpeaks.Thesesiteshavelog2ISvalues > 1.5andarehighlightedinblackinFigure4B.Surprisingly,thissamesubsetofAUGcodonsalsoshowshighISvaluesforA-ribosomes(Figure4C),arguingthatSD-ASDpairingisnotthefeaturethatexplainswhyinitiationtakesplaceatthesespecificAUGtripletsandnotatothers. Figure4with1supplementseeall Downloadasset Openasset TheeffectsofSD-ASDpairingoninitiationatnon-canonicalsites. (A)Evaluationofinitiationscore,IS.(B)Initiationscoresonnon-annotatedAUGtriplets.ForC-ribosomes,thefractionwithIS >1.5iscoloredblack.(C)InitiationscoresforC-andA-ribosomesforthesetofsiteswithIS >1.5forC-ribosomes(High,coloredblackinB)andthosewithIS <1.5(Low).(D,E)ISvaluesforallfourribosometypesonthesubsetofsiteswithhighaffinityfortheASDoftheS-ribosome(CCUCC).AverageRPFsattheAUGtripletswithhighISscores(F)orlowISscores(G)fromtheS-ribosomedata. TofurthercharacterizehowSD-ASDpairingaffectsinitiationatnon-annotatedAUGtriplets,wegroupedpotentialinitiationsitesbytheiraffinityforwild-typeormutantASDsasdescribedaboveforannotatedstartsites.ForsiteswithhighaffinitytotheASDoftheS-ribosome,forexample,thedistributionofISvaluesforS-ribosomescloselyresembledtheotherthreeribosomes(Figure4E),withmedianvaluesnearzero.ThesedatashowthatthepresenceofacomplementaryShine-Dalgarno-likesequencenearanAUGcodonisnotsufficienttorecruitS-ribosomesandgeneratearobuststartcodonpeak.WeselectedthesubsetofAUGswithhighaffinitytoS-ribosomeswhereinitiationoccurswithS-ribosomes(log2IS >1.5,darkredinFigure4E).Asexpected,thesehigh-ISsitesshowstrongstartpeakswithS-ribosomes;however,theotherribosomeswithdifferentASDsequencesshowrobuststartpeaksaswell(Figure4F).Similarly,low-ISsitesthatarenottranslatedbyS-ribosomes(lightredinFigure4E)arealsonottranslatedbytheotherribosomes(Figure4G).TheobservationthatSD-ASDpairingdoesnotcontributetoinitiationatthesesiteswithhighaffinitytotheS-ribosomesalsoholdstruefornon-annotatedAUGswithhighaffinitytothewild-typeASD(Figure4—figuresupplement1).Onceagain,thesedataarguethatAUGsthatrecruitribosomesandleadtoinitiationarehard-wiredforthispurposeirrespectiveofthestrengthofthemRNA-rRNAbasepairinginteraction.Takentogether,thesedataoninitiatingribosomesshowthatmRNA-rRNAbasepairingisneithernecessarynorsufficientfortranslationalinitiation. ToprovideinsightintomRNAfeaturesotherthanSDstrengththatmightcontributetoribosomerecruitment,weaskedwhichfeaturesareenrichedatannotatedstartsites.ToavoidinterferencefromSDmotifs,weselectedonlyannotatedstartsiteswithlowaffinitytothewild-typeASD(∆G > 0)andcomparedthemtonon-annotatedAUGcodons,mostofwhichdonotleadtoinitiation.Weobservedenrichmentofadenosines(A)atmanysiteswithin15ntupstreamofthestartcodonand5ntdownstream(Figure5A). Figure5with1supplementseeall Downloadasset Openasset A-richsequencesasasignalforstartcodonselection. (A)ProbabilitylogooftheregionsurroundingannotatedAUGswithlowaffinitytothewild-typeASDsequence(∆G > 0)ascomparedwithallnon-annotatedAUGsinthetranscriptome.Enrichednucleotidesareshownabovetheaxisanddepletednucleotidesbelowtheaxis.TheheightoftheletterrepresentsthebinomialP-value.(B)Designofthereporterassay.ThereporterplasmidencodesmCherrywithastrongribosomebindingsite(RBS)andseparatelyGFPdownstreamofaregioncontainingastartsiteofinterest(30ntupstreamofAUGand42ntdownstream).(C)Initiationsitesusedinthereporterassay;thenumberindicatesthegenomicpositionofAUG.IntheT-andC-mutants,theA’supstreamofAUG(highlightedingreen)weresubstitutedbyTorC.(D)Resultsofthereporterassay.EachdotisthemedianofGFP/mCherryfromanindependentrunofflowcytometry.ThebargraphindicatesthemeanandSDfromfourindependenttests.NoGFP(aplasmidthatencodesmCherrybutnotGFP)servesasacontrolshowingthebaselinesignalfromcellularautofluorescence;theotherdataarenormalizedtothisratio.(E)Median(solidline)andinterquartilerange(shaded)ofmRNAstructureinSHAPE-MaPseqdatafor365annotatedstartsites(red)and7310non-annotatedAUGswithincodingsequences(blue). TotestwhethertheseA’spromotetranslation,weselectedfourmRNAswithA-richinitiationsites(andweakSDmotifs)andestablishedaGFPreporterassaytofollowtheiractivity(Figure5B).OfthesefourmRNAs(Figure5C),twocontainannotatedinitiationsiteswithlowASD-affinity,thestartcodonsfromyhbYandgsk.Wealsoselectedtworepresentativenon-annotatedAUGcodonsfoundwithinthecreAandyeiRgenes;thesesiteshavehighISvaluesinboththeC-ribosomeandO-ribosomeMS2RPdatafromretapamulin-treatedcells.ThesequencessurroundingthesefourAUGcodons(from30ntupstreamto45ntdownstream)werefusedinframetoGFPsuchthatGFPfluorescencereportsontheactivityoftheAUGofinterest.Inadditiontothewild-typesequence,weconstructedmutantsinwhichalloftheA’s15nucleotidesupstreamofAUGwerechangedtoeitherU’sorC’s(G’swereavoidedbecausetheyhavehighaffinityfortheASD).ThereferenceproteinmCherrywasalsoexpressedfromthesameplasmidwithastandardribosomebindingsite.TheGFP/mCherryratiowasthennormalizedtoacontrollackingtheGFPsequence(measuringonlycellularauto-fluorescence). WeobservedthattheGFP/mCherryratiowashigherthanbackgroundforallfourAUGcodons,showingthatallarecapableofdrivingGFPexpression(Figure5D).ThetwoannotatedstartsitesfromyhbYandgskinducedstrongerGFPexpressionthanthenon-annotatedstartsites,creA*andyeiR*.Importantly,however,thefactthatfluorescencewasobservedfromtheselatterexamplesconfirmstheresultsfromtheMS2RPdatafromretapamulin-treatedcellsshowingthattheyaretranslatedtosomeextentbywild-typeribosomes.WeobservedthatreplacementoftheA’swithU’sloweredGFPexpressioninallcasesexceptforyeiR*whichshowedtheweakestGFPexpression.AstrongereffectwasobservedbychangingtheA’stoC’s,whichledtocompletelossofGFPfluorescencefromallfourAUGcontextstested.TheseresultssupportourhypothesisthatA-richsequencesupstreamofstartcodonscontributetotheidentificationoftranslationalstartsites. TheabilityofA-richsequencestopromoteinitiationislikelynotlimitedtoE.coli:whenwecomparedthelocalcontextofAUGcodonsinannotatedstartsitesvs.non-annotatedAUGcodonsforasetofdiversebacteria,weagainsawthatA-richsequenceswereenriched(Figure5—figuresupplement1).ForE.coliandmostotherspeciesexamined,theenrichmentofA’swasweakerthantheenrichmentofG’scorrespondingtotheSDsequence,butforMycoplasmapneumoniaeandFlavobacteriumjohnsoniae,theSDsignalisnotobservedandtheretheenrichmentofA’sisparticularlystriking.A-richsequencesarehighlyconservedandmayserveasanimportantmechanismforstartsiteselectioninthesespecies,whilecontributingbroadlytomorediversespecies. Inbacteria,mRNAstructuresurroundingthestartcodonhasbeenshowninmechanisticstudiestoreduceribosomaloccupancy(Lodish,1970;deSmitandvanDuin,1990;deSmitandvanDuin,2003;EspahBorujeniandSalis,2016).Moreover,severaltranscriptome-wideanalysesofmRNAstructureinE.colishowlowerlevelsofstructuresurroundinginitiationsites(DelCampoetal.,2015;Burkhardtetal.,2017).WeaskedhowmRNAstructurediffersbetweenannotatedstartsitesandinternalAUGcodonsthatarenotannotatedasstartsites.WeuseddatafromarecentstudyofthestructureofmRNAsinvivousingSHAPEanddeepsequencing(Mustoeetal.,2018).Fromtranscriptswithsufficientcoverage,wecalculatedthemedianSHAPEreactivityovera120ntwindowsurrounding365annotatedstartsitesandcompareditto7310non-annotatedAUGs(Figure5E).Forannotatedinitiationsites,thelevelofmRNAstructureissignificantlylowerforaregion30ntinlengthonbothsidesoftheAUGcodon(showninred)aspreviouslyreported(DelCampoetal.,2015;Burkhardtetal.,2017).Incontrast,exceptforasharpdipinreactivityatthealignedAUGcodonduetosequencebias,weseethatmRNAstructureisconsistentlyhighacrossthiswindowforthesetofnon-annotatedAUGs(showninblue).ThesedifferencesmaybedueinparttotheabilityofribosomestomeltRNAstructureduringtranslation;indeed,initiationleadstotheunfoldingofRNA,whichfacilitatesinitiationbyanother30Ssubunit(EspahBorujeniandSalis,2016;Andreevaetal.,2018).But,giventhatSHAPEandDMSreactivityofmRNAsinvivoandinvitroarestronglycorrelated(Burkhardtetal.,2017;Mustoeetal.,2018),itisalsolikelythatmRNAstructureplaysacausalroleinsettinginitiationrates. Inthisstudy,weperformedribosomeprofilingonmutantribosomespurifiedusinganRNAtag,theMS2aptamer,astrategywecallMS2RP(Figure1).OriginallydevelopedforinvitrostudiesofribosomescontaininglethalrRNAmutations(Youngmanetal.,2004;YoungmanandGreen,2005),MS2-taggedribosomesalsohavepotentialtoyieldinsightsintothefunctionofkeyrRNAsequencesinvivo.InadditiontothestudiesoftheASDsequencein16SrRNAreportedhere,MS2RPcouldbeemployedtocharacterizethefunctionsofrRNAdomainsoninitiation,elongation,termination,andrecyclingatagenome-widelevelinvivo.BecauseMS2RPcanbeperformedonrRNAmutantsexpressedfromplasmids,themethodcanbeeasilytransferredtootherbacteriaortoeukaryoteswithoutalteringrDNAinthegenome.OfparticularinterestarerRNAvariantsinbacterialgenomesthatareexpresseddifferentiallyinresponsetochangesintheenvironmentandareproposedtohavedifferentspecificitiesorfunctions(Kuryloetal.,2018;Songetal.,2019).VariantrRNAalleleshavealsobeenreportedforeukaryoticcells(Parksetal.,2018);forexample,differentsmallsubunitrRNAallelesareexpressedinvariousdevelopmentalstagesinPlasmodium(Gundersonetal.,1987).Inaddition,thefunctionsofthehighlyvariablerRNAexpansionsegmentsineukaryotesarepoorlyunderstood(Spahnetal.,2001;Angeretal.,2013).MS2RPcouldbeapowerfultooltoelucidatetheactivitiesofvarioussubpopulationsofvariantormutatedrRNAs. Previousgenome-widestudiesinbacteriahaveshownlittleornocorrelationbetweenSDstrengthandribosomeoccupancy(Lietal.,2014;Schraderetal.,2014;DelCampoetal.,2015).UsingMS2RP,weareableforthefirsttimetorevealtheroleofSDmotifsinpromotinginitiationacrossthetranscriptome.Inourapproach,wemutatedtheASDontheribosomes,thusmaintainingmRNAsequenceandstructure,thusallowingustoisolatetheeffectsoftheSD:ASDinteractionontranslation.IntheabsenceofSD:ASDpairing,weobservedastrongnegativecorrelationbetweenribosomeoccupancyandtheSDstrength(calculatedbypairingwiththewild-typeASDsequence).Inotherwords,themutantribosomestranslategeneswithstrongSDmotifsworsethanthosewithweakSDmotifs(Figure2B).Therearetwopossibleexplanationsforthisnegativecorrelation.Itmaybethatthebindingofwild-typeribosomestomRNAswithstrongSDmotifsoccludestheirribosome-bindingsites,preventingmutantribosomesfrominitiatingandefficientlytranslatingthesegenes.Alternatively,mRNAstructureandotherfeaturesmayoutweightheimpactofSDmotifs,maskingtheireffects,explainingwhyconventionalribosomeprofilingstudiesfailedtoobservecorrelationsbetweenSDstrengthandribosomeoccupancy.Regardlessofwhichoftheseexplanationsiscorrect,theMS2RPstrategyallowsustosubtractthecumulativecontributiontoribosomeoccupancyofallofsuchothermRNAfeatures,andthustofocusexclusivelyonthecontributiontoribosomeoccupancyoftheSD:ASDinteractiongenome-wide.Inthisanalysis,wearenowabletoseealinearcorrelationbetweentheSDstrengthofanmRNAandproteinoutput(Figure2C). GiventhattheSDmotiffunctionsthroughawell-definedmechanismandiswidelyconservedthroughoutbacteria,ithasbeenthoughttoprovideanimportantmechanismforstartcodonselectionandtranslationaloutput.Consistentwithsuchaview,SDmotifsareunderrepresentedwithinORFsinordertoavoidspuriousinitiationatinternalstartcodons(Hockenberryetal.,2018).Strikingly,however,wefindthatribosomeswithalteredASDsstillfindthecorrectstartcodonsaboutasefficientlyaswild-typeribosomes(Figure3).StartpeaksforallfourribosometypesareobservedatannotatedstartsitesregardlessoftheaffinityoftheribosomebindingsitefortheASD.Thisshowsthatinitiationsitesarehard-wiredforinitiationbasedonmRNAfeaturesseparatefromthepotentialforSD-ASDpairing.Theseobservationsalsoholdtrueattheoccasionalnon-annotatedAUGcodonswheresomeinitiationoccurs(Figure4).ThesedataareconsistentwiththeconclusionthatSDmotifsarenotessentialfordeterminingwheretranslationstartsonmRNAsgenome-wide. What,then,areothermechanismsthatcouldbeusedforstartcodonselection?LocalmRNAstructureandRNAfoldingkineticsclearlymustplayacriticalroleinallowingribosomestofindthestartcodon.AnumberofmechanisticstudieshavedemonstratedthatRNAstructurearoundthestartcodonlowerstranslationlevels(Halletal.,1982;deSmitandvanDuin,1990;Ostermanetal.,2013;EspahBorujenietal.,2014).Studiesoffactorsthataltertheexpressionofsimplifiedreportergenes(involvingrandomizationofthe5’-UTRorcodingsequences)showthatlackofsecondarystructuresurroundingtheinitiationsitehasthemostsignificantcorrelationwithproteinoutput(Salisetal.,2009;Kudlaetal.,2009;Goodmanetal.,2013).Recenttranscriptome-wideanalysesofmRNAstructureinE.coliconfirmthatannotatedstartsiteshavelowerlevelsofmRNAstructure,asseenbyPARSonpurifiedmRNAandDMS-seqinvivo(DelCampoetal.,2015;Burkhardtetal.,2017).mRNAstructureislikelyanimportantfactorinstartsiteselection:usinghigh-resolutionSHAPE-MaPseqdata(Mustoeetal.,2018),weshowedthatannotatedAUGshavelowerlevelsofRNAstructure30ntupstreamanddownstreamwhereasinternalAUGarenotsurroundedbyregionsoflowerstructure(Figure5E). Interestingly,incomparingthesequencecontextofAUGcodonsthatareannotatedasinitiationsiteswiththosethatarenot,wefoundthatA’sareenrichedbothupstreamanddownstreamofannotatedinitiationsites(Figure5A)andweconfirmedtheirimportanceinreporterassays(Figure5B–D).Theseresultsfromendogenousinitiationsitesarereminiscentofobservationsoftheover-representationofA’sin5’-UTRsequencesselectedforstrongaffinitytotheribosomeinvitro(Gaoetal.,2016)andin5’-UTRsselectedfromrandomsequencesupstreamofareportergeneforhighlevelsoftranslationinvivo(Evfratovetal.,2017).Comparisonofannotatedstartsitesandnon-annotatedAUGsacrossseveralbacterialgenomesshowsthatthismechanismiswidespread(Figure5—figuresupplement1).AlthoughenrichmentofA’sismoresubtlethanenrichmentofG’sinE.coliandB.subtilis,inorganismsthatlackSDmotifs,suchasMycoplasmapneumoniaeandFlavobacteriumjohnsoniae,A-richmotifsmayplayanimportantroleininitiation.Indeed,inarecentstudy,Fredrickandco-workersusedribosomeprofilinginF.johnsoniaeandobservedenrichmentofA’supstreamofstartcodonsinmRNAswithhighribosomeoccupancyincomparisontogeneswitharetranslatedlessefficiently(Baezetal.,2019).Weenvisionthatthissequence,liketheShine-Dalgarnomotif,actsasatranslationalenhancer,fine-tuningtheefficiencyofinitiation. ThemechanismbywhichA-richsequencesenhanceinitiationisnotclear.TheprevalenceofA’smayalterthemRNAdynamics;A-richsequencestendtohavelesssecondarystructurethanGC-richsequences.WenotehoweverthatreplacingA’swithU’sinseveralreportersreducedtranslationlevelseventhoughtheU’saresimilarlynotexpectedtoyieldstrongstructures.AsecondpossibilityisthatribosomalcomponentsmayinteractspecificallywithA’sclosetothestartcodonthatareboundinsidetheribosomeduringinitiation.Fredrickandco-workersusedreporterassaystoshowthatmutationofaparticularAatthe−3positionreducesexpression;thisresultisintriguingbecausetheclassicKozaksequence(GCC(A/G)CCAUG)thatpromoteshighlevelsoftranslationineukaryotesalsocontainsapurineatposition−3.A-richsequenceshavebeenreportedtoenhancetranslationinavarietyofeukaryoticcontextsincludingDrosophilaandwheatgermandreticulocytelysates(RanjanandHasnain,1995;Sanoetal.,2002;Suzukietal.,2006;Pfeifferetal.,2012).ItmaybethatA-richsequencesinteractwithconservedelementsoftheribosomeacrossthedomainsoflife.A’sfurtherfromthestartcodon(10–20ntupstream)mayinteractwithbacteria-specificribosomalproteinS1.bS1preferablybindstoA/U-richsequenceelementsupstreamofSDsequences(Bonietal.,1991;Komarovaetal.,2005)andisthoughttounwindmRNAstructuretoinduceinitiation(Quetal.,2012;Duvaletal.,2013). Ourfindingshavebroadimplicationsfortheevolutionoftranslationalmechanismsinbacteria.NotallbacteriautilizeSDmotifstopromotetranslationalinitiation—SDmotifsarenotablylackinginBacteroidetesandCyanobacteria.BecausetheprevalenceofSDmotifsisafeatureofthegenomeingeneralandnotofasinglegene,itmakessensethatevolutionaryselectivepressurefororagainstSDusagewouldactatthelevelofthetranscriptome.Thenatureoftheseselectivepressuresremainsunclear,althoughHockenberryrecentlyarguedthatbacteriawithhighlevelsofSDusagetendtohavehighermaximalgrowthrates(Hockenberryetal.,2017).Futurestudieswillclarifytheevolutionaryrelationshipbetweenthegrowthenvironment,levelsofSDusageamongbacterialspecies,andtheirtranscriptome-wideeffects. Unlessotherwisespecified,cellswereculturedat37°Cin500mLofLB+ampicillin(50mg/L).IPTGwasadded(0.3mMfinal)whentheculturereachedOD600 = 0.3andcellswereharvestedbyfiltrationatOD600 = 0.5.Forprofilingwithretapamulin,cellsweregrownat37°Cin500mLofLB+ampicillintoOD600 = 0.3,inducedwithIPTG,growntoOD600 = 0.45,andthenharvestedbyfiltration5minaftertheadditionofretapamulin(100µg/mLfinal). CellswereharvestedbyfiltrationusingaKontes99mmfiltrationapparatusand0.45umnitrocellulosefilter(Whatman)andthenflashfrozeninliquidnitrogen.Cellswerelysedinlysisbuffer(20mMTrispH8.0,10mMMgCl2,100mMNH4Cl,5mMCaCl2,100U/mLDNaseI,and1mMchloramphenicol)usingaSpex6870freezermillwith5cyclesof1mingrindingat5Hzand1mincooling.Lysateswerecentrifugedat20,000gfor30minat4°Ctopelletcelldebris. BL21(DE3)cellsweretransformedwiththeplasmidpMal-c2G-MBP-MS2-His,culturedat37°CinLB+ampicillin(50mg/L)toOD600 = 0.7,andinducedwith0.3mMfinalIPTGfor4hrat37°C.Cellswereharvestedbycentrifugationandlysedonafrenchpressinthebindingbuffer(50mMNaH2PO4pH8.0,300mMNaCl,10mMimidazole,6mMBME).TheMBP-MS2proteinwaspurifiedbyFPLC(Atka,GE);afterwasheswiththebindingbuffer,itwaselutionwiththebindingbuffersupplementedwith200mMimidazole. 3mLofamyloseresin(NEB)weretransferredtoaPoly-PrepChromatographyColumn(Bio-Rad)andwashed3timeswith10mLoflysisbuffer.2.5mgofMBP-MS2-Hisproteinwereloadedontotheamyloseresin,incubatedat4°Cfor1hr,andwashedtwicewith10mLoflysisbuffer.ForMS2RP,1.5mLofcelllysateand15µLofRNaseT1(1000U/µL,Thermo)wereloadedontotheMBP-MS2resin,incubatedat4°Cfor2hr,andwashed3timeswith10mLoflysisbuffer.Theresinwasre-suspendedin1mLoflysisbufferand360µgMNasewasaddedtodigestmRNAandremovetheMS2hairpininrRNA,releasingtheribosomesfromthecolumn.Followinga2hrincubationat25°C,theflow-throughwascollected.Another2mLoflysisbufferwaspassedthroughtheresinandcollected.Theflow-throughfractionswerethencombined. 10–54%sucrosedensitygradientswerepreparedusingtheGradientMaster108(Biocomp)inthegradientbuffer(20mMTrispH8.0,10mMMgCl2,100mMNH4Cl,2mMDTT).5–20AUofE.colilysatewasloadedontopofsucrosegradientandcentrifugedinaSW41rotorat35,000rpmfor2.5hrat4°C.FractionationwasperformedonaPistonGradientFractionator(Biocomp). LibrariesforMS2RPandstandardribosomeprofilingarepreparedasinWoolstenhulmeetal.(2015) andMohammadetal.(2016).AtleasttwobiologicalreplicateswereperformedforeachMS2RPlibraryasdetailedintheGEOdatabaseentry.RNA-seqlibrarieswerepreparedwithTruSeqStrandedTotalRNAGoldfrom250ngoftotalRNAfollowingdepletionofrRNAbyRiboZerorRNARemovalKitforbacteria(Illumina).LibrarieswereanalyzedbyBioAnalyzerhighsensitivityDNAkit(Agilent)thensequencedontheHiSeq2500(Illumina). RNAwaspurifiedbyhot-phenolextraction.Thefirststrandsynthesiswasperformedwith500ngoftotalRNA,primerMS2check_R(5’-AGACATTACTCACCCGTCCGCCACTC-3’)andSuperScriptIII(Invitrogen).15cyclesofPCRamplificationwereperformedwithprimerMS2check_F70(5’-TGCAAGTCGAACGGTAACAGGAAG-3’),primerMS2check_R,andPhusionpolymerase(NEB).PCRproductswereresolvedby8%TEBgelandanalyzedbyTyphoonFLA9500(GE). MG1655cellscarryingthereporterplasmidwereculturedinLB+ampicillin(50mg/L)toearlylogphase.Cellswerediluted50-foldinTBS.GFPandmCherryfluorescenceweremeasuredonaGuavaeasyCyteflowcytometer(MilliporeSigma). ForlibrariespreparedbylinkerwithUMI(rAppNNNNNNCACTCGGGCACCAAGGAC),perfectlymatchingreads(including5’-endand3’-endUMI)wereconvertedtoasinglereadbyTally(Davisetal.,2013).3’-linkersequenceswereremovedbySkewer(Jiangetal.,2014).The5’endUMIaddedbytheRTprimerwereremovedbyseqtk.Readswerealignedusingbowtieversion1.1.2(Langmeadetal.,2009),firsttothetRNAs,rRNAs,andthessrA,ssrS,lacIandffsgenes.ReadsthatfailedtoaligntothosesequenceswerealignedtoE.coliMG1655NC_000913.2.Ribosomepositionwasassignedbythe3’-endofalignedreads.RNA-seqdatawereassignedbythe5’-endofalignedreads. Theaffinity(∆G)oftheASDandthesequenceofastartcodonwascalculatedforeachmRNAusingfree_scanwith‘-l0–b0’optiontodisallowinternalloopandinternalbulge(Nakagawaetal.,2010).Theinputsequenceswere−15and−6ntupstreamofAUGandthereversesequenceofwild-typeASD(UCCUCCA)orthemutantASDwhereappropriate. AverageSHAPEreactivitywasbasedontheSHAPE-MaPdata(Mustoeetal.,2018).AmedianoftheSHAPEreactivityfromtheregion−25to+25upstreamanddownstreamofthestartcodonwasusedasdegreeonRNAstructure. AUGcodonswereonlyincludedintheanalysisofaverageribosomedensityandinitiationscoresiftheyhadmorethan10mappedreadsinthewindowof−50upstreamand+50downstreamoftheAUG.Tocalculateaverageribosomedensity,foreachAUGwetooktherpmateachpositionacrossthiswindow,divideditbythetotalrpminthewindow,andthencomputedthemeanofthesevaluesforallAUGsincludedinthecalculation.Initiationscoreswerecomputedbytakingthemeanofreadsmappedwithin+3to+21ntdownstreamoftheAinAUGanddividingitbythemeanofreadsmappedontheregion−50to+50oftheAUG. ProbabilitylogosweregeneratedbykpLogo(WuandBartel,2017)usingitsdefaultsettings.ForFigure5A,inputandbackgroundsequencesaredescribedinthefigurelegend.ForFigure5—figuresupplement1thesetofinputsequencesconsistedofannotatedAUGsfromtheGFFfileavailableatNCBIandthesetofbackgroundsequencesconsistedofallAUGsinthegenomethatwerenotannotatedasinitiationsites. ThesequencingdataareavailableinprocessedWIGformatattheGEOusingaccessionnumberGSE135906andastherawFASTQfilesattheSRA.Custompythonscriptsusedtoanalyzethesequencingdataarefreelyavailableathttps://github.com/greenlabjhmi/2019_SDASD (Saito,2020;copyarchivedathttps://github.com/elifesciences-publications/2019_SDASD). Addacomment +Openannotations.Thecurrentannotationcountonthispageisbeingcalculated. SequencingdatahavebeendepositedintheGEOunderaccessioncodeGSE135906. Thefollowingdatasetsweregenerated SaitoK GreenR BuskirkAR (2019) NCBIGeneExpressionOmnibus IDGSE135906.TranslationalinitiationinE.colioccursatthecorrectsitesgenome-wideintheabsenceofmRNA-rRNAbase-pairing. https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE135906 AleksashinNA LeppikM HockenberryAJ KlepackiD Vázquez-LaslopN JewettMC RemmeJ MankinAS (2019) Assemblyandfunctionalityoftheribosomewithtetheredsubunits NatureCommunications10:1–13. https://doi.org/10.1038/s41467-019-08892-w PubMed GoogleScholar AltuviaS ZhangA ArgamanL TiwariA StorzG (1998) TheEscherichiacoliOxySregulatoryRNArepressesfhlAtranslationbyblockingribosomebinding TheEMBOJournal17:6069–6075. https://doi.org/10.1093/emboj/17.20.6069 PubMed GoogleScholar AndreevaI BelardinelliR RodninaMV (2018) TranslationinitiationinbacterialpolysomesthroughribosomeloadingonastandbysiteonahighlytranslatedmRNA PNAS115:4411–4416. https://doi.org/10.1073/pnas.1718029115 PubMed GoogleScholar AngerAM ArmacheJ-P BerninghausenO HabeckM SubkleweM WilsonDN BeckmannR (2013) StructuresofthehumanandDrosophila80Sribosome Nature497:80–85. https://doi.org/10.1038/nature12104 GoogleScholar BabitzkeP BakerCS RomeoT (2009) RegulationoftranslationinitiationbyRNAbindingproteins AnnualReviewofMicrobiology63:27–44. https://doi.org/10.1146/annurev.micro.091208.073514 PubMed GoogleScholar BaezWD RoyB McNuttZA ShatoffEA ChenS BundschuhR FredrickK (2019) GlobalanalysisofproteinsynthesisinFlavobacteriumjohnsoniaerevealstheuseofKozak-likesequencesindiverseBacteria NucleicAcidsResearch47:10477–10488. https://doi.org/10.1093/nar/gkz855 PubMed GoogleScholar BoniIV IsaevaDM MusychenkoML TzarevaNV (1991) Ribosome-messengerrecognition:mrnatargetsitesforribosomalproteinS1 NucleicAcidsResearch19:155–162. https://doi.org/10.1093/nar/19.1.155 PubMed GoogleScholar BurkhardtDH RouskinS ZhangY LiGW WeissmanJS GrossCA (2017) OperonmRNAsareorganizedintoORF-centricstructuresthatpredicttranslationefficiency eLife6:e22037. https://doi.org/10.7554/eLife.22037 PubMed GoogleScholar DavisMP vanDongenS Abreu-GoodgerC BartonicekN EnrightAJ (2013) Kraken:asetoftoolsforqualitycontrolandanalysisofhigh-throughputsequencedata Methods63:41–49. https://doi.org/10.1016/j.ymeth.2013.06.027 PubMed GoogleScholar deSmitMH vanDuinJ (1990) Secondarystructureoftheribosomebindingsitedeterminestranslationalefficiency:aquantitativeanalysis PNAS87:7668–7672. https://doi.org/10.1073/pnas.87.19.7668 PubMed GoogleScholar deSmitMH vanDuinJ (2003) Translationalstandbysites:howribosomesmaydealwiththerapidfoldingkineticsofmRNA JournalofMolecularBiology331:737–743. https://doi.org/10.1016/S0022-2836(03)00809-X GoogleScholar DelCampoC BartholomäusA FedyuninI IgnatovaZ (2015) SecondarystructureacrossthebacterialtranscriptomerevealsversatilerolesinmRNAregulationandfunction PLOSGenetics11:e1005613. https://doi.org/10.1371/journal.pgen.1005613 PubMed GoogleScholar DuvalM KorepanovA FuchsbauerO FechterP HallerA FabbrettiA ChoulierL MicuraR KlaholzBP RombyP SpringerM MarziS (2013) EscherichiacoliribosomalproteinS1unfoldsstructuredmRNAsontotheribosomeforactivetranslationinitiation PLOSBiology11:e1001731. https://doi.org/10.1371/journal.pbio.1001731 PubMed GoogleScholar EspahBorujeniA ChannarasappaAS SalisHM (2014) Translationrateiscontrolledbycoupledtrade-offsbetweensiteaccessibility,selectiveRNAunfoldingandslidingatupstreamstandbysites NucleicAcidsResearch42:2646–2659. https://doi.org/10.1093/nar/gkt1139 PubMed GoogleScholar EspahBorujeniA SalisHM (2016) TranslationinitiationiscontrolledbyRNAfoldingkineticsviaaribosomedraftingmechanism JournaloftheAmericanChemicalSociety138:7016–7023. https://doi.org/10.1021/jacs.6b01453 PubMed GoogleScholar EvfratovSA OstermanIA KomarovaES PogorelskayaAM RubtsovaMP ZatsepinTS SemashkoTA KostryukovaES MironovAA BurnaevE KrymovaE GelfandMS GovorunVM BogdanovAA SergievPV DontsovaOA (2017) Applicationofsortingandnextgenerationsequencingtostudy5'-UTRinfluenceontranslationefficiencyinEscherichiacoli NucleicAcidsResearch45:3487–3502. https://doi.org/10.1093/nar/gkw1141 PubMed GoogleScholar GaoR YuK NieJ LianT JinJ LiljasA SuX-D (2016) DeepsequencingrevealsglobalpatternsofmRNArecruitmentduringtranslationinitiation ScientificReports6:30170. https://doi.org/10.1038/srep30170 GoogleScholar GoodmanDB ChurchGM KosuriS (2013) CausesandeffectsofN-terminalCodonbiasinbacterialgenes Science342:475–479. https://doi.org/10.1126/science.1241934 PubMed GoogleScholar GundersonJH SoginML WollettG HollingdaleM delaCruzVF WatersAP McCutchanTF (1987) Structurallydistinct,stage-specificribosomesoccurinplasmodium Science238:933–937. https://doi.org/10.1126/science.3672135 PubMed GoogleScholar HallMN GabayJ DébarbouilléM SchwartzM (1982) AroleformRNAsecondarystructureinthecontroloftranslationinitiation Nature295:616–618. https://doi.org/10.1038/295616a0 GoogleScholar HockenberryAJ SternAJ AmaralLAN JewettMC (2017) Diversityoftranslationinitiationmechanismsacrossbacterialspeciesisdrivenbyenvironmentalconditionsandgrowthdemands MolecularBiologyandEvolution35:582–592. https://doi.org/10.1093/molbev/msx310 GoogleScholar HockenberryAJ JewettMC AmaralLAN WilkeCO (2018) Within-GeneShine-Dalgarnosequencesarenotselectedforfunction MolecularBiologyandEvolution35:2487–2498. https://doi.org/10.1093/molbev/msy150 PubMed GoogleScholar HuiA deBoerHA (1987) Specializedribosomesystem:preferentialtranslationofasinglemRNAspeciesbyasubpopulationofmutatedribosomesinEscherichiacoli PNAS84:4762–4766. https://doi.org/10.1073/pnas.84.14.4762 PubMed GoogleScholar IngoliaNT GhaemmaghamiS NewmanJRS WeissmanJS (2009) Genome-Wideanalysisinvivooftranslationwithnucleotideresolutionusingribosomeprofiling Science324:218–223. https://doi.org/10.1126/science.1168978 GoogleScholar IngoliaNT LareauLF WeissmanJS (2011) Ribosomeprofilingofmouseembryonicstemcellsrevealsthecomplexityanddynamicsofmammalianproteomes Cell147:789–802. https://doi.org/10.1016/j.cell.2011.10.002 PubMed GoogleScholar JacobWF SanterM DahlbergAE (1987) AsinglebasechangeintheShine-Dalgarnoregionof16SrRNAofEscherichiacoliaffectstranslationofmanyproteins PNAS84:4757–4761. https://doi.org/10.1073/pnas.84.14.4757 PubMed GoogleScholar JiangH LeiR DingSW ZhuS (2014) Skewer:afastandaccurateadaptertrimmerfornext-generationsequencingpaired-endreads BMCBioinformatics15:182. https://doi.org/10.1186/1471-2105-15-182 PubMed GoogleScholar KomarovaAV TchufistovaLS DreyfusM BoniIV (2005) AU-richsequenceswithin5'untranslatedleadersenhancetranslationandstabilizemRNAinEscherichiacoli JournalofBacteriology187:1344–1349. https://doi.org/10.1128/JB.187.4.1344-1349.2005 PubMed GoogleScholar KudlaG MurrayAW TollerveyD PlotkinJB (2009) Coding-sequencedeterminantsofgeneexpressioninEscherichiacoli Science324:255–258. https://doi.org/10.1126/science.1170160 PubMed GoogleScholar KuryloCM ParksMM JuetteMF ZinshteynB AltmanRB ThibadoJK VincentCT BlanchardSC (2018) EndogenousrRNASequenceVariationCanRegulateStressResponseGeneExpressionandPhenotype CellReports25:236–248. https://doi.org/10.1016/j.celrep.2018.08.093 GoogleScholar LangmeadB TrapnellC PopM SalzbergSL (2009) Ultrafastandmemory-efficientalignmentofshortDNAsequencestothehumangenome GenomeBiology10:R25. https://doi.org/10.1186/gb-2009-10-3-r25 GoogleScholar LeeK Holland-StaleyCA CunninghamPR (1996) GeneticanalysisoftheShine-Dalgarnointeraction:selectionofalternativefunctionalmRNA-rRNAcombinations RNA2:1270–1285. PubMed GoogleScholar LeeS LiuB LeeS HuangSX ShenB QianSB (2012) Globalmappingoftranslationinitiationsitesinmammaliancellsatsingle-nucleotideresolution PNAS109:E2424–E2432. https://doi.org/10.1073/pnas.1207846109 PubMed GoogleScholar LiGW BurkhardtD GrossC WeissmanJS (2014) Quantifyingabsoluteproteinsynthesisratesrevealsprinciplesunderlyingallocationofcellularresources Cell157:624–635. https://doi.org/10.1016/j.cell.2014.02.033 PubMed GoogleScholar LiGW (2015) HowdoBacteriatunetranslationefficiency? CurrentOpinioninMicrobiology24:66–71. https://doi.org/10.1016/j.mib.2015.01.001 PubMed GoogleScholar LodishHF (1970) Secondarystructureofbacteriophagef2ribonucleicacidandtheinitiationofinvitroproteinbiosynthesis JournalofMolecularBiology50:689–702. https://doi.org/10.1016/0022-2836(70)90093-8 GoogleScholar MajdalaniN CunningC SledjeskiD ElliottT GottesmanS (1998) DsrARNAregulatestranslationofRpoSmessagebyananti-antisensemechanism,independentofitsactionasanantisilenceroftranscription PNAS95:12462–12467. https://doi.org/10.1073/pnas.95.21.12462 PubMed GoogleScholar MandalM BreakerRR (2004) Generegulationbyriboswitches NatureReviewsMolecularCellBiology5:451–463. https://doi.org/10.1038/nrm1403 PubMed GoogleScholar MeydanS MarksJ KlepackiD SharmaV BaranovPV FirthAE MargusT KefiA Vázquez-LaslopN MankinAS (2019) Retapamulin-Assistedribosomeprofilingrevealsthealternativebacterialproteome MolecularCell74:481–493. https://doi.org/10.1016/j.molcel.2019.02.017 PubMed GoogleScholar MohammadF WoolstenhulmeCJ GreenR BuskirkAR (2016) ClarifyingthetranslationalpausinglandscapeinBacteriabyribosomeprofiling CellReports14:686–694. https://doi.org/10.1016/j.celrep.2015.12.073 GoogleScholar MoineH RombyP SpringerM Grunberg-ManagoM EbelJP EhresmannB EhresmannC (1990) Escherichiacolithreonyl-tRNAsynthetaseandtRNA(Thr)modulatethebindingoftheribosometothetranslationalinitiationsiteofthethrSmRNA JournalofMolecularBiology216:299–310. https://doi.org/10.1016/S0022-2836(05)80321-3 PubMed GoogleScholar MustoeAM BusanS RiceGM HajdinCE PetersonBK RudaVM KubicaN NutiuR BaryzaJL WeeksKM (2018) PervasiveregulatoryfunctionsofmRNAstructurerevealedbyHigh-ResolutionSHAPEprobing Cell173:181–195. https://doi.org/10.1016/j.cell.2018.02.034 PubMed GoogleScholar NakagawaS NiimuraY MiuraK GojoboriT (2010) Dynamicevolutionoftranslationinitiationmechanismsinprokaryotes PNAS107:6382–6387. https://doi.org/10.1073/pnas.1002036107 PubMed GoogleScholar NeumannH WangK DavisL Garcia-AlaiM ChinJW (2010) Encodingmultipleunnaturalaminoacidsviaevolutionofaquadruplet-decodingribosome Nature464:441–444. https://doi.org/10.1038/nature08817 PubMed GoogleScholar OrelleC CarlsonED SzalT FlorinT JewettMC MankinAS (2015) Proteinsynthesisbyribosomeswithtetheredsubunits Nature524:119–124. https://doi.org/10.1038/nature14862 PubMed GoogleScholar OstermanIA EvfratovSA SergievPV DontsovaOA (2013) ComparisonofmRNAfeaturesaffectingtranslationinitiationandreinitiation NucleicAcidsResearch41:474–486. https://doi.org/10.1093/nar/gks989 PubMed GoogleScholar ParksMM KuryloCM DassRA BojmarL LydenD VincentCT BlanchardSC (2018) VariantribosomalRNAallelesareconservedandexhibittissue-specificexpression ScienceAdvances4:eaao0665. https://doi.org/10.1126/sciadv.aao0665 PubMed GoogleScholar PfeifferBD TrumanJW RubinGM (2012) UsingtranslationalenhancerstoincreasetransgeneexpressioninDrosophila PNAS109:6626–6631. https://doi.org/10.1073/pnas.1204520109 PubMed GoogleScholar QuX LancasterL NollerHF BustamanteC TinocoI (2012) RibosomalproteinS1unwindsdouble-strandedRNAinmultiplesteps PNAS109:14458–14463. https://doi.org/10.1073/pnas.1208950109 PubMed GoogleScholar RackhamO ChinJW (2005) AnetworkoforthogonalribosomexmRNApairs NatureChemicalBiology1:159–166. https://doi.org/10.1038/nchembio719 PubMed GoogleScholar RanjanA HasnainSE (1995) InfluenceofCodonusageandtranslationalinitiationCodoncontextintheAcNPV-basedexpressionsystem:computeranalysisusinghomologousandheterologousgenes VirusGenes9:149–153. https://doi.org/10.1007/BF01702657 PubMed GoogleScholar RexG SurinB BesseG SchneppeB McCarthyJE (1994) ThemechanismoftranslationalcouplinginEscherichiacoli.HigherorderstructureintheatpHAmRNAactsasaconformationalswitchregulatingtheaccessofdenovoinitiatingribosomes TheJournalofBiologicalChemistry269:18118–18127. PubMed GoogleScholar Software SaitoK (2020) 2019_SDASD,version6afaf41 Github. https://github.com/greenlabjhmi/2019_SDASD SalisHM MirskyEA VoigtCA (2009) Automateddesignofsyntheticribosomebindingsitestocontrolproteinexpression NatureBiotechnology27:946–950. https://doi.org/10.1038/nbt.1568 PubMed GoogleScholar SanoK MaedaK OkiM MaédaY (2002) EnhancementofproteinexpressionininsectcellsbyalobstertropomyosincDNAleadersequence FEBSLetters532:143–146. https://doi.org/10.1016/S0014-5793(02)03659-1 PubMed GoogleScholar SchraderJM ZhouB LiGW LaskerK ChildersWS WilliamsB LongT CrossonS McAdamsHH WeissmanJS ShapiroL (2014) ThecodingandnoncodingarchitectureoftheCaulobactercrescentusgenome PLOSGenetics10:e1004463. https://doi.org/10.1371/journal.pgen.1004463 PubMed GoogleScholar ShineJ DalgarnoL (1974) The3'-terminalsequenceofEscherichiacoli16SribosomalRNA:complementaritytononsensetripletsandribosomebindingsites PNAS71:1342–1346. https://doi.org/10.1073/pnas.71.4.1342 PubMed GoogleScholar SongW JooM YeomJH ShinE LeeM ChoiHK HwangJ KimYI SeoR LeeJE MooreCJ KimYH EyunSI HahnY BaeJ LeeK (2019) DivergentrRNAsasregulatorsofgeneexpressionattheribosomelevel NatureMicrobiology4:515–526. https://doi.org/10.1038/s41564-018-0341-1 PubMed GoogleScholar SpahnCM BeckmannR EswarN PenczekPA SaliA BlobelG FrankJ (2001) Structureofthe80SribosomefromSaccharomycescerevisiae--tRNA-ribosomeandsubunit-subunitinteractions Cell107:373–386. https://doi.org/10.1016/S0092-8674(01)00539-6 PubMed GoogleScholar SteitzJA JakesK (1975) HowribosomesselectinitiatorregionsinmRNA:basepairformationbetweenthe3'terminusof16SrRNAandthemRNAduringinitiationofproteinsynthesisinEscherichiacoli PNAS72:4734–4738. https://doi.org/10.1073/pnas.72.12.4734 PubMed GoogleScholar StorzG OpdykeJA ZhangA (2004) ControllingmRNAstabilityandtranslationwithsmall,noncodingRNAs CurrentOpinioninMicrobiology7:140–144. https://doi.org/10.1016/j.mib.2004.02.015 PubMed GoogleScholar SuzukiT ItoM EzureT KobayashiS ShikataM TanimizuK NishimuraO (2006) Performanceofexpressionvector,pTD1,ininsectcell-freetranslationsystem JournalofBioscienceandBioengineering102:69–71. https://doi.org/10.1263/jbb.102.69 GoogleScholar WeaverJ MohammadF BuskirkAR StorzG (2019) Identifyingsmallproteinsbyribosomeprofilingwithstalledinitiationcomplexes mBio10:e02819. https://doi.org/10.1128/mBio.02819-18 PubMed GoogleScholar WinklerW NahviA BreakerRR (2002) ThiaminederivativesbindmessengerRNAsdirectlytoregulatebacterialgeneexpression Nature419:952–956. https://doi.org/10.1038/nature01145 PubMed GoogleScholar WoolstenhulmeCJ GuydoshNR GreenR BuskirkAR (2015) High-precisionanalysisoftranslationalpausingbyribosomeprofilinginBacterialackingEFP CellReports11:13–21. https://doi.org/10.1016/j.celrep.2015.03.014 PubMed GoogleScholar WuX BartelDP (2017) kpLogo:positionalk-meranalysisrevealshiddenspecificityinbiologicalsequences NucleicAcidsResearch45:W534–W538. https://doi.org/10.1093/nar/gkx323 PubMed GoogleScholar YoungmanEM BrunelleJL KochaniakAB GreenR (2004) Theactivesiteoftheribosomeiscomposedoftwolayersofconservednucleotideswithdistinctrolesinpeptidebondformationandpeptiderelease Cell117:589–599. https://doi.org/10.1016/S0092-8674(04)00411-8 PubMed GoogleScholar YoungmanEM GreenR (2005) Affinitypurificationofinvivo-assembledribosomesforinvitrobiochemicalanalysis Methods36:305–312. https://doi.org/10.1016/j.ymeth.2005.04.007 PubMed GoogleScholar JosephTWade ReviewingEditor;WadsworthCenter,NewYorkStateDepartmentofHealth,UnitedStates JamesLManley SeniorEditor;ColumbiaUniversity,UnitedStates JosephTWade Reviewer;WadsworthCenter,NewYorkStateDepartmentofHealth,UnitedStates AlexanderMankin Reviewer;UniversityofIllinois,Chicago,UnitedStates Intheinterestsoftransparency,eLifepublishesthemostsubstantiverevisionrequestsandtheaccompanyingauthorresponses. Acceptancesummary: ThispaperusesaninnovativetwistonribosomeprofilingtoinvestigatetheimportanceofShine-Dalgarnosequencesinbacterialtranslationinitiation.Surprisingly,thedatashowthatstrongbase-pairingbetweenthe16SribosomalRNAandthemRNAShine-Dalgarnosequenceisneithernecessarynorsufficientfortranslationinitiation.Thissuggeststhatstart-codonsare"hard-wired"intothegenome,largelyindependentofShine-Dalgarnosequence. Decisionletterafterpeerreview: [Editors’note:theauthorssubmittedforreconsiderationfollowingthedecisionafterpeerreview.Whatfollowsisthedecisionletterafterthefirstroundofreview.] Thankyouforsubmittingyourworkentitled"Shine-Dalgarnosequencesfine-tunetranslationgenome-widebutarenottheprimarydeterminantsofstart-siteselection"forconsiderationbyeLife.Yourarticlehasbeenreviewedbythreepeerreviewers,includingJoeWadeastheReviewingEditorandReviewer#1,andtheevaluationhasbeenoverseenbyJimManleyastheSeniorEditor.Thefollowingindividualinvolvedinreviewofyoursubmissionhasagreedtorevealtheiridentity:ShuraMankin(Reviewer#2). Ourdecisionhasbeenreachedafteranextensivediscussioninvolvingthethreereviewers.Thereviewerswereenthusiasticaboutpartsofthemanuscript,inparticularthemethoditself;however,therewassomedisagreementastothesignificanceofthework.MuchofthediscussionfocusedonthedatainFigure6,whichthereviewersconsideredtopresentpotentiallythemostimportantresult.Wefeltthatfurtheranalysisisneededtofullyaddressthekeyquestionsof(i)whetherannotatedstartcodonsareinherentlygoodatbindingribosomes,independentoftheSD,and(ii)whetheranSDalone(e.g.nexttoanATGinthemiddleofanORF)isinsufficienttobindaribosome.Putmoresimply,wefeltthatfurtheranalysisisrequiredtoshowthatstart-codonsare"hard-wired"intothegenome,largelyindependentofSDsequence.Wearethereforerejectingthepaperbecausetheoutcomeofthenewanalysisisunclear.Nonetheless,wewouldbewillingtoconsiderarevisedversioniftheanalysessuggestedbelow,orsomethingequivalent,providestrongersupportfortheideathatstart-codonsarehard-wired,independentoftheSDsequence. ThemainconcernswithFigure6arethat(i)thecontrolsetofORF-internalATGsisnotthebestcontrolbecausemany(most?)oftheATGsdon'thaveagoodSDsequenceforthemodifiedribosomes;and(ii)ribosomedensityatannotatedstartcodonsforthemodifiedribosomescouldbeduetoasubsetofstartcodonsthathavedecentmatchestothemodifiedASDsequence.WesuggestthatamoreappropriatecontrolsetofATGswouldbethosewithagoodpredictedmatchtothealteredASDsequence.WealsosuggestlimitingtheanalysisinFigure6CtostartcodonsthathaveapoormatchtothemodifiedASD.Anotherwaytolookatthiswouldbetocomparewhichannotatedstartcodonsarerecognizedbythedifferentmodifiedribosomes;ifallthreetypesofribosomerecognizethesamesubsetofstartcodons,it'ssafetoconcludethatthisisoccurringindependentoftheSD.Ifthese(orother)analysescanprovidestrongersupportforthe"hard-wired"model,thatwouldlikelybesufficientforpublication.Inadditiontothere-analysisofdatafromFigure6,it'simportanttoimprovetheclarityofthepaper,whichwasattimesconfusing(seethedetailedreviewsformoreinformationonthis).Additionally,reviewer#3makessomeimportantpointsaboutthecalculationofhybridizationenergies,suchasconsideringafull,9ntSDsequencewithvariablespacing.Lastly,themanuscriptwouldbenefitfromaclearerdescriptionofwhatisalreadyknownaboutfeaturesotherthanSDsequencethatcontributetotranslationinitiation(seecommentsfromreviewer3). Reviewer#1: ThispaperdescribesaninnovativeapproachtoprobetheimportanceofShine-Dalgarno(S-D)sequencesintranslationinitiationinEscherichiacoli.Byperformingribosomeprofilingonmodifiedribosomes,theauthorsareabletoobservetranslationbyribosomeswithalteredanti-S-Dsequences.ThismethodrevealsthatdespitenocorrelationbetweenS-Dstrengthandtranslationlevels,thereisacontributionofS-Dstrengththatisapparentwhenallconfoundingfactorshavebeencontrolledfor.Interestingly,thiseffectofS-Dsislostduringothergrowthconditions,althoughforcoldshockthatislargelyconsistentwithpreviouswork,anditisunclearwhatthemechanismisinstationaryphase.WhileIthinkthetopicisinterestingandtheprimarymethodisingenious,I'mnotconvincedthattheauthorshavelearnedmuchabouttherelativeimportanceofS-Dsintranslationinitiation.Astheyacknowledge,previousstudieshavefailedtoseeacorrelationbetweenS-Dstrengthandtranslationinitiationlevels,andtheimportanceofsecondarystructureandofA-richsequenceshasbeendescribedpreviously.ThefactthatpredictionsofS-DstrengthcorrelatewithtranslationinitiationlevelsoncefactorsotherthanS-Dhavebeenaccountedforindicatesthatthesepredictionsarefairlyaccurate.Thisisimportant,sinceitaccountsforthepossibilitythatthelackofcorrelationbetweenpredictedS-DstrengthandtranslationinitiationisbecauseofourinabilitytopredictS-Dstrength.However,theimpactofthisadvanceissmall.IalsohaveconcernsabouttheinterpretationofFigures6and7thatimpacttheoverallconclusions. -Thepresentationofthecoldshockdataisconfusing.TheoverallconclusionisthatS-D-dependenceislostatalmostallgenesduringcoldshock.However,asmallsubsetofgenesappearstodependstronglyontheS-D.Thedistinctionbetweentheeffectonthemajorityofgenesandtheeffectonasmallsubsetshouldbeexplainedmoreclearly.Thesimplestinterpretationofthesedataisthatmoststartcodonsarehighlystructuredduringcoldshock,butthosethataredonotrelyontheirS-Ds.Thismodelislargelyconsistentwithpreviouswork. -IdisagreewiththeinterpretationofFigure6.Thedatashowthatforthealteredribosomes,annotatedstartcodonsareusedfarmoreefficientlythanthecollectionofallotherATGsequenceswithinORFs.However,therearemanymoreATGswithinORFsthanannotatedstartcodons,andeveniftranslationreliesheavilyonS-Dsequences,youwouldexpectthatmostATGswithinORFswouldnotbeselectedbyalternativeribosomesbecauseonlyasmallsubsetwillhaveappropriateS-Dsequences,andmanymaybeweaklyexpressed.Myinterpretationofthesedataisthatalternativeribosomesdouseannotatedstartcodons,butthereisnowaytotellhowselectivelytheydothis.Amoreappropriatecomparisonwouldbeof(i)annotatedstartcodonsto(ii)ATGswithinORFswheretheATGisassociatedwithasequencethatispredictedtofunctionasagoodS-Dforthealternativeribosome. -AnotherconcernIhavewithFigure6isthatpresumablysome,andperhapsmanyoftheannotatedstartcodonswillhavegoodSDmatchesforthealternativeribosomes.Figure2Csuggeststhatthenumberwithgoodmatcheswillbefairlyhigh.IstheribosomedensityatannotatedstartcodonssimplyduetothesubsetofstartcodonsthathavereasonableSDmatchestothealteredASD?AnotherwaytothinkaboutthisistoaskwhetherthestartcodonscontributingtothesignalinFigure6CarethesamestartcodonsthatcontributetothesignalinSupplementaryFigure5A-B. -Figure7EshowstheimportanceofanA-richsequenceinthecontextofstartcodonslackingagoodS-D.SimilartoFigure6,thesedatahighlightthecontributionofnon-SDsequencestotranslationinitiation,buttheydonotprovideanyinformationabouttherelativeimportanceofthedifferentsequenceelements. Reviewer#2: Majorfindings: ThepaperofSaitoandal.examinesthecontributionofShine-Dalgarnosequence(SD)tothetranslationefficiencyinbacteria.Usingacleverapproach,theauthorsuseribosomeprofilingtocomparemRNAoccupancybywtribosomesandribosomeswiththealteredanti-SDsequence(ASD).InconfirmationofpreviousfindingsfromtheWeissmanlab,theyfindthatthegeneraltranslationefficiencydoesnotcorrelatewiththepredictedstrengthofSD-ASDinteractions.However,whenalltheotherfactorsaremasked,theyobserveastrongdependenceoftheinitiationrateonthestrengthofSD-ASDpairing.TheyalsonotedthatasubsetofgenesexpressedinthestressedcellsdependheavilyonrecognitionoftheSDsequencebytheribosomes.Oneoftheunexpected,buthighlyimportantfindingsistheobservationthattheribosomeswiththealteredASDcanneverthelesscorrectlyandselectivelyinitiatetranslationattheknownstartsitesunderscoringtheimportanceoffactorsotherthanSD-ASDinteractionsinthestartcodonselection.Importantly,thereportedworkrevealstheprevalenceofA-richmotifsintheribosomebindingsitesofthegeneswithweakSDsequencesinE.coliandotherbacteria.ThistrendbecomesespeciallyprominentinthebacterialspeciesthatdonotrelyonSD-ASDinteractionsfortranslationinitiation. Critique: Thisisaninteresting,intriguingandimportantstudy.Theresultsareniceandcleanandtheimplicationsareimportantforunravelingthefundamentalmechanismoftranslationinitiationinbacteria.Althoughthepaperisgenerallywellwritten,itwashardattimestofollowtheauthorslogicandIstronglyencouragetheauthorstotrytoclarifythemessage,whichoftenwashardtoextract. 1)Hereareseveralexamples: -Abstract:Thestatement"Werevealagenome-widecorrelationbetweentheSDstrengthandtranslationalefficiency"isfollowedby"thisglobalcorrelationislostandasubsetofgenes[…]becomes[dependent]onSDmotifsfortranslation".Thisishardtodigest. -Figure4Clegend("thestrengthoftheSDmotifsdetermineswhetherwild-typeorASDmutantsarerecruitedtomessages")issupposedtocontrastFigure4Flegend("theunstructuredSDmotifscanrecruitwildtyperibosomesmoreeffectivelythantheyrecruitASDmutants").However,theysoundnearlyidenticalandthus,donotaccuratelycommunicatethepointtheauthorsapparentlyaretryingtomake. -"geneswithstrongSDmotifaretranslatedbetterbyribosomeswithcanonicalASD":betterincomparisonwiththeASD-mutantribosomesorbetterincomparisonwiththegeneswithweakSD? 2)Aleksashinetal.,2019,haveshownthatalteringASDin16SrRNAcompromisesrRNAmaturation.Althoughthepresenceofunprocessedsequencesatthe5'and3'endoftheASD-mutant16SrRNAwouldnotlikelychangethegeneralconclusionsofthepaper,hypotheticallyitcouldaffectthefunctionalityandelongationrateofthemutantribosomes.Iamwonderingwhetherauthorshavecheckedhowwelltheirmutant16SrRNAsareprocessed.Irrespectively,IbelieveamoredetaileddiscussionofthegeneralfunctionalityoftheribosomeswithalteredASD,especiallyinrelationtotheelongationrate,wouldbebeneficial. 3)Subsection“Gene-specificrolesofSDmotifsunderstress”.ThereadersneedabetterexplanationwhytheauthorsswitchedfromΔlogTEtoΔlogRPKMmetricswhentheymovetotheexperimentsinthestressedcells. 4)Theinfluenceofthecompetitionbetweenwtandmutant30Ssubunitsforthetranslationstartsitesontheconclusionsdrawnfromribosomeprofilingshouldbediscussed. Reviewer#3: Theauthorsintroducemutated16SribosomalRNAsintoE.colistrains,alteringtheiranti-ShineDalgarnosequences,toinvestigatehowtheseperturbationsaffecttranslationratesacrosstheE.coligenome.Todothis,theycarryoutribosomeprofilingexperimentstomeasuregenome-wideribosomedensitiesonaSD-modifiedstrains,includingduringexponential,stationary,andcoldshockgrowthphases.Overall,theyfindthatchangingthelast9nucleotidesofthe16SrRNAhasasignificanteffectonthetranscriptomes'translationrates. Overall,thecollectedmeasurementsareinterestingandpotentiallyuseful.However,theanalysissuffersfromaterriblyincompleteknowledgeofwhatcontrolsamRNA'stranslationrate.Thestatisticsappliedaretailoredfora1-factorproblem,wheninfact,therearemanyfactorsthatcontroltranslationrate.Therearealsoinconsistenciesanderrorsintheauthors'calculationsthatshouldbecorrected.Theauthors'conclusionsarenotwellsupportedbytheiranalysis.Themanuscriptrequiressignificantworkforittoproductivelyaddtoourknowledgeofwhatcontrolstranslationrateinbacteria. 1)TheauthorfocusesprimarilyontheimportanceofthesequencecolloquiallyknownastheShine-DalgarnoincontrollingamRNA'stranslationinitiationrate.Theauthorswritethat"InitiationratesvarydependingonhowwellanmRNArecruits30Ssubunitstothestartcodon,andinbacteria,theworkingmodelisthatthisisaccomplishedprimarilybyShine-Dalgarno(SD)motifs."Thisisincorrect.ThecurrentworkingmodelisthatamRNA'stranslationinitiationrateiscontrolledbyatleastfiveimportantmolecularinteractions,onlyoneisthehybridizationbetweenthelast9nucleotidesofthe16SrRNAandthemRNA.Theyinclude: a)thehybridizationbetweenthelast9nucleotidesofthe16SrRNAandthemRNA;b)theunfoldingofmRNAstructuresthatoverlapwiththeribosomefootprint;c)thedifferencesinspacing(physicaldistance)betweenthe16SrRNAbindingsiteandthestartcodon;d)thestandbysite'saccessibility,asdeterminedbythelengthofavailablesingle-strandedRNA;e)thestartcodonanditshybridizationtothetRNA. ThefreeenergyneededtounfoldinhibitorymRNAstructuresisalsoaffectedbythedynamicsofRNAfolding(RNAfoldingkinetics)aswellastherateofribosomebinding. Iftheauthorsbetterunderstoodhowtranslationratewascontrolled,theycouldmoreproductivelyusetheirmeasurementstopushtherealstate-of-the-artforward.Theircurrentconclusionsarealreadysubsumedwithinthestate-of-the-art(i.e.,nothingnew). 2)Anydiscussionof"whichtranslationrateinteractionismostimportant"or"whichtranslationrateinteractionisresponsibleforXwhereastheinteractionYonlyfine-tunesZ"isnotproductiveandcaneasilybecontradictedbyselectingarealcounterexample.Overall,itisthebindingfreeenergyofthe30SribosometothemRNAthatdeterminesitstranslationinitiationrate.EachoftheseinteractionscontributesfreeenergytothisprocessandthemagnitudeofthecontributedfreeenergiescanberoughlyequalacrossaselectionofrealmRNAexamples.ThereareunstructuredmRNAswherethereislittlepenaltyforunfoldinginhibitorymRNAstructures.TherearehighlystructuredmRNAsthathaveconsensusSDssequences.TherearemRNAsthathaveconsensusSDsequencesfarawayfromthestartcodon.AllofthesemRNAscouldhavethesametranslationrate.Whichinteractionismostimportant?That'snottherightquestiontoask,becauseit'smeaningless. 3)Themanuscript'smaintopicistheShine-Dalgarnosequence,buttheauthorsshouldbemadeawarethatatleastthelast9nucleotidesofthe16SrRNAcancontactthemRNAandhybridizetoit.InE.coli,theanti-ShineDalgarnosequenceis5'-ACCUCCUUA-3'andthe"consensus"Shine-Dalgarnosequenceistherefore5'-TAAGGAGGT-3'.Themanuscripttextandtheauthors'calculationsshouldreflectthis. 4)Theauthorsaremis-usingtheribosomeprofilingmeasurementsintheiranalysis.Ribosomeprofilingmeasurementsdonotdirectlymeasuretranslationrates.TheymeasuremRNA-boundribosomedensities.AmRNA'sribosomedensitywilldependonbothitstranslationinitiationrateANDitstranslationelongationrate.Specifically,insteady-stateconditions,theribosomedensitywillbetheratiobetweenthesetwoquantities(initiationrateoverelongationrate).Intheinitialapplicationsofribosomeprofiling,researchersassumedthatallmRNAshavethesametranslationelongationrateinordertoconcludethatribosomedensitymeasurementswereproportionaltotranslationinitiationrates.Thisisnottrue.CodingsequencesinmRNAshaveverydifferenttranslationelongationrates,duetodifferencesinsynonymouscodonusage.UnlesseachmRNAs'translationelongationratesarepredictedordirectlymeasured,ribosomedensitymeasurementscannotbeusedtoinfertheirtranslationinitiationrates.Therefore,whentheauthorswrite"Inpioneeringribosomeprofilingstudiesinbacteria,theparadoxicalobservationwasmadethatthereislittleornocorrelationbetweenthetranslationalefficiencyofageneandthestrengthofitsSDmotif(calculatedusingthermodynamicalgorithmsforRNApairing),ashadbeenanticipatedbasedontheSDmodel."thereisnoactualparadox.TheribosomeprofilingmeasurementswerenotusedcorrectlytotesthowmRNAsequencescontroltranslationrate. 5)Gettingtotheauthors'mainconclusions,theywritethat"ThesedataindicatethattheASDmutantribosomestranslategeneswithweakSDmotifsbetterthangeneswithstrongSDmotifs,exactlytheoppositeofwhatwild-typeribosomesareexpectedtodo."Thisstatementisconfusinggiventherealconclusionoftheauthors,thatallotherfactorsbeingequala"strong"SDmotifdoesresultinhighertranslationthana"weak"SDmotif.It'sonlybecauseofotherconfoundingfactorsthattheinitialanalysisdidnotyieldapositivecorrelation.Anincorrectanalysis(excludingconfoundingvariables)cannotleadtoacorrectconclusion. 6)Figure2Cshowsaveryinterestingandproductiveresult,thatthedifferenceintranslationefficiencybetweenthewild-type"C"ribosomesandtheA-ribosomescorrelatestosomedegreewiththehybridizationfreeenergybetweenthemRNAand(aportionof)theanti-SDsequence.Thisisaproductiveapproachtowardseliminatingkeyconfoundingvariablesbecause,inprinciple,thestrengthsofthefourotherinteractionsthatcontroltranslationinitiationrateshouldnotchangewhenthe16SrRNAaSDsequencesarechanged.However,it'snotapparentinthemanuscripttext,buttheauthorsareusingthemodified16SrRNAsto“eliminatetheSD-aSDinteractionasacontributiontothemRNA'stranslationrate”.SowhentheysubtractthecontributionfromthemodifiedA-ribosome'stranslationratesfromtheC-ribosome'stranslationrate,theyareobservingmoredirectlythecontributionfromtheSD-aSDinteraction.Themanuscripttextshouldmoreclearlyexplainthisexperimentaldesign.Thisisacreativeandvalidwayofusingribosomeprofilingmeasurements. 7)However,thehybridizationfreeenergycalculationscouldbeimproved.First,asmentionedpreviously,thewild-typeaSDsequenceinE.coliisACCUCCUUA.Second,thehybridizationfreeenergycalculationwasonlyperformedontheregionfrom15to6nucleotidesupstreamofthestartcodon,buttheaSDsequencecanhybridizeatotherlocations.Third,thehybridizationbetweenthemRNAandaSDcanaccommodate1or2-nucleotidebulgesorinternalloops. 8)ThemeasurementsincoldshockaregreatlyconfoundedbythehigherexpressionlevelsofRNAchaperonesthatareunfoldingmRNAstructures“atspecificmRNAs”wheretheRNAchaperonesrecognizebindingmotifs.TheconclusionhereshouldbethatRNAchaperonesbindspecificmRNAs,unfoldtheirinhibitorymRNAstructures,andincreasetheirtranslationratesduringcoldshock.ThisisallindependentoftheShine-Dalgarnosequence.Thisprocessalsodoesnotdependonmanyotheruninterestingfactors. 9)TheuseofORF-wideGINIvaluesisoddbecauseit'sgenerallyonlytheregionsurroundingthestartcodonthataffectsitstranslationinitiationrate,andnotthestructureoftheentireORF(whichthiscoefficientisquantifying).Also,usingtheSHAPEreactivityaroundastartcodonasaproxyforRNAstructureisabitmisleadingasribosomesactivelyunfoldRNAstructuresduringtranslationinitiation.AhighlystructuredmRNAwithaconsensusSDsequencewillhaveahighSHAPEreactivity(i.e.,lowRNAstructure)becausetheribosomescanrapidlybindtothemRNAandunfoldthemRNAstructure.SHAPEreactivityismeasuringtheeffectofrapidribosomebindingandnotthecauseofit.RapidribosomebindingcanalsobefacilitatedbyslowRNArefoldingkinetics,called"RibosomeDrafting"intheliterature. 10)ThedatainFigure6justsaysthatA-ribosomescaninitiatetranslationrateatotherstartcodonsbecausetheynowhavemorenegativebindingfreeenergiestothosestartcodons,comparedtotheannotatedones.TheauthorscouldperformhybridizationcalculationsusingtheA-ribosome'saSDsequencetoinvestigatewhetherthese"newstartcodons"haveanearby"SD"sequencethatiscomplementarytotheA-ribosome'saSD.Thatwouldbeinteresting. [Editors’note:furtherrevisionsweresuggestedpriortoacceptance,asdescribedbelow.] Thankyouforresubmittingyourworkentitled"TranslationalinitiationinE.colioccursatthecorrectsitesgenome-wideintheabsenceofmRNA-rRNAbase-pairing"forfurtherconsiderationbyeLife.YourrevisedarticlehasbeenevaluatedbyJamesManleyastheSeniorEditor,andthreereviewers,includingJoeWadeastheReviewingEditorandReviewer#1. Thereviewersandeditorsagreethattherevisedmanuscriptisgreatlyimproved,andwearepleasedtoprovisionallyacceptthemanuscript.Weaskthatyoumakeafewsmallchangesinresponsetothereviewers'comments.First,reviewer2hastwominorconcernsthatareeasilyaddressed.Second,basedonreviewer3'scomments,theconclusionsregardingtheimportanceofA-richsequencesshouldbesoftenedalittle.Thereviewers'commentsarelistedbelow: Reviewer#1: Theauthorshavedoneanexcellentjobimprovingthemanuscript.Removingthedataoncold-shockhasimprovedthefocusandreadability.Moreover,thenewanalysesinFigures3and4makeamorecompellingcasethatShine-Dalgarnosequencesareneithernecessarynorsufficientforstartsiteselection. Reviewer#2: ThestreamlinedpaperofSaitoetal.readsmuchbetterthantheoriginalversionanddeliversaclearandimpactfulmessage. Ibelieveitcanbepublishedafterauthorsaddresstworemainingissues: Theauthorsreferto"thenumberofelongatingribosomespermRNAasaproxyofinitiationrates".Thisisincorrect:therewouldbetwiceasmanyribosomesonanmRNAthatistwiceaslongasanotherone,evenifthosetwowouldhavethesameinitiationrate.ThecorrectmetricsisnotthenumberofribosomespermRNAbuttheribosomedensity(theirnumbernormalizedbymRNAlength).ThisdoesnotaffectconclusionsofthepaperbecauseauthorsnormalizeRiboSeqreadsbyRNASeqreads.Yet,Iwouldtrytoavoidthisconfusion. Theauthorswrite:"Interestingly,incomparinginternalAUGcodonsthatsupportinitiationinourribosomeprofilingdatatothosethatdonot,wefoundthatA'sareenrichedbothupstreamanddownstreamofinitiationsites(Figure5A).….Thisresultsfromendogenousinitiationsites…".However,Figure5Adoesnotdealwiththeinternalinitiationsites,butwiththeannotatedsiteslackingSD. Reviewer#3: Withtherevisions,theauthorshavegreatlyimprovedthemanuscript'sintroductorydescriptionoftranslationandtheoverallanalysisoftheirdataset,leadingtoamorelaser-focusedandwell-supportedsetofconclusions.Theseresultsprovideanexcellentandclarifyingviewofthesequencedeterminantsandinteractionsthatcontroltranslationinitiationratewithinnatural(highlyevolved)mRNAsbycleanlyseparatingtheroleoftheSD:aSDinteractionfromotherfactors,includingthepresence/absenceofinhibitorymRNAstructures. TheIntroductionprovidesamorecomprehensivedescriptionoftheseveralsequencedeterminantsandfactorsthatcontroltranslationinitiationrate,whichisessentialtowardsunderstandingtheauthors'excellentdataset.Theanalysisclearlyexplainshowtheirdatasetprovidescomparativemeasurementswithvs.withouttheSD:aSDinteractionandhowthosemeasurementsquantifyitseffectontranslationrate.Startcodonselectionisapropertyofallthefactorsthatcontroltranslationinitiationrate,andisalsolikelyapropertyofribosome-ribosomedynamicsalongthemRNA. https://doi.org/10.7554/eLife.55002.sa1 [Editors’note:theauthorsresubmittedarevisedversionofthepaperforconsideration.Whatfollowsistheauthors’responsetothefirstroundofreview.] Reviewer#1: ThispaperdescribesaninnovativeapproachtoprobetheimportanceofShine-Dalgarno(S-D)sequencesintranslationinitiationinEscherichiacoli.[…]IalsohaveconcernsabouttheinterpretationofFigures6and7thatimpacttheoverallconclusions. -Thepresentationofthecoldshockdataisconfusing.TheoverallconclusionisthatS-D-dependenceislostatalmostallgenesduringcoldshock.However,asmallsubsetofgenesappearstodependstronglyontheS-D.Thedistinctionbetweentheeffectonthemajorityofgenesandtheeffectonasmallsubsetshouldbeexplainedmoreclearly.Thesimplestinterpretationofthesedataisthatmoststartcodonsarehighlystructuredduringcoldshock,butthosethataredonotrelyontheirS-Ds.Thismodelislargelyconsistentwithpreviouswork. Wehaveremovedthefiguresdealingwithcoldshockandotherstresses.Weagreethattheyarelargelyconsistentwithpreviouswork.Theconfusionraisedbythecoldshockstoryappearstohavetakenawayfromthemainstory. -IdisagreewiththeinterpretationofFigure6.Thedatashowthatforthealteredribosomes,annotatedstartcodonsareusedfarmoreefficientlythanthecollectionofallotherATGsequenceswithinORFs.However,therearemanymoreATGswithinORFsthanannotatedstartcodons,andeveniftranslationreliesheavilyonS-Dsequences,youwouldexpectthatmostATGswithinORFswouldnotbeselectedbyalternativeribosomesbecauseonlyasmallsubsetwillhaveappropriateS-Dsequences,andmanymaybeweaklyexpressed.Myinterpretationofthesedataisthatalternativeribosomesdouseannotatedstartcodons,butthereisnowaytotellhowselectivelytheydothis.Amoreappropriatecomparisonwouldbeof(i)annotatedstartcodonsto(ii)ATGswithinORFswheretheATGisassociatedwithasequencethatispredictedtofunctionasagoodS-Dforthealternativeribosome. WehaveaddedanalysestothenewFigure4andFigure4—figuresupplement1showinginitiationatinternalAUGcodonspredictedtohavehighaffinityforthemutantASDsequences.ThesedatashowthatinitiationoccurswithallfourribosometypesregardlessoftheSDstrengthorspecificity,butthatinitiationismostefficientwhentheSDandASDarecomplementary. -AnotherconcernIhavewithFigure6isthatpresumablysome,andperhapsmanyoftheannotatedstartcodonswillhavegoodSDmatchesforthealternativeribosomes.Figure2Csuggeststhatthenumberwithgoodmatcheswillbefairlyhigh.IstheribosomedensityatannotatedstartcodonssimplyduetothesubsetofstartcodonsthathavereasonableSDmatchestothealteredASD?AnotherwaytothinkaboutthisistoaskwhetherthestartcodonscontributingtothesignalinFigure6CarethesamestartcodonsthatcontributetothesignalinSupplementaryFigure5A-B. WeaddedanalysestothenewFigure3showingthatinitiationoccurswithallthreemutantribosomesatannotatedstartsitesthathavenoaffinityforthemutantASDsequences. -Figure7EshowstheimportanceofanA-richsequenceinthecontextofstartcodonslackingagoodS-D.SimilartoFigure6,thesedatahighlightthecontributionofnon-SDsequencestotranslationinitiation,buttheydonotprovideanyinformationabouttherelativeimportanceofthedifferentsequenceelements. Wehaveremovedclaimsabouttherelativeimportanceofdifferentsequenceelements. Reviewer#2: Critique: Thisisaninteresting,intriguingandimportantstudy.Theresultsareniceandcleanandtheimplicationsareimportantforunravelingthefundamentalmechanismoftranslationinitiationinbacteria.Althoughthepaperisgenerallywellwritten,itwashardattimestofollowtheauthorslogicandIstronglyencouragetheauthorstotrytoclarifythemessage,whichoftenwashardtoextract. 1)Hereareseveralexamples: -Abstract:Thestatement"Werevealagenome-widecorrelationbetweentheSDstrengthandtranslationalefficiency"isfollowedby"thisglobalcorrelationislostandasubsetofgenes[…]becomes[dependent]onSDmotifsfortranslation".Thisishardtodigest. -Figure4Clegend("thestrengthoftheSDmotifsdetermineswhetherwild-typeorASDmutantsarerecruitedtomessages")issupposedtocontrastFigure4Flegend("theunstructuredSDmotifscanrecruitwildtyperibosomesmoreeffectivelythantheyrecruitASDmutants").However,theysoundnearlyidenticalandthus,donotaccuratelycommunicatethepointtheauthorsapparentlyaretryingtomake. -"geneswithstrongSDmotifaretranslatedbetterbyribosomeswithcanonicalASD":betterincomparisonwiththeASD-mutantribosomesorbetterincomparisonwiththegeneswithweakSD? Weremovedthesectiononstressconditionsthatwashardtofollowandtakingawayfromthemainpointofthemanuscript. 2)Aleksashinetal.,2019,haveshownthatalteringASDin16SrRNAcompromisesrRNAmaturation.Althoughthepresenceofunprocessedsequencesatthe5'and3'endoftheASD-mutant16SrRNAwouldnotlikelychangethegeneralconclusionsofthepaper,hypotheticallyitcouldaffectthefunctionalityandelongationrateofthemutantribosomes.Iamwonderingwhetherauthorshavecheckedhowwelltheirmutant16SrRNAsareprocessed.Irrespectively,IbelieveamoredetaileddiscussionofthegeneralfunctionalityoftheribosomeswithalteredASD,especiallyinrelationtotheelongationrate,wouldbebeneficial. RNA-seqanalysesofrRNA(priortonucleasetreatment)isnowshowninFigure1—figuresupplement1anddiscussedearlyintheResultssection(subsection“SelectiveprofilingofribosomeswithmutantASDsequences”). 3)Subsection“Gene-specificrolesofSDmotifsunderstress”.ThereadersneedabetterexplanationwhytheauthorsswitchedfromΔlogTEtoΔlogRPKMmetricswhentheymovetotheexperimentsinthestressedcells. Thissectionwasremoved. 4)Theinfluenceofthecompetitionbetweenwtandmutant30Ssubunitsforthetranslationstartsitesontheconclusionsdrawnfromribosomeprofilingshouldbediscussed. ThispossibilitywasaddedtotheDiscussion. Reviewer#3: 1)TheauthorfocusesprimarilyontheimportanceofthesequencecolloquiallyknownastheShine-DalgarnoincontrollingamRNA'stranslationinitiationrate.Theauthorswritethat"InitiationratesvarydependingonhowwellanmRNArecruits30Ssubunitstothestartcodon,andinbacteria,theworkingmodelisthatthisisaccomplishedprimarilybyShine-Dalgarno(SD)motifs."Thisisincorrect.[…]Theircurrentconclusionsarealreadysubsumedwithinthestate-of-the-art(i.e.,nothingnew). WeaddedamoredetaileddescriptionofthefactorsthataffectinitiationratestotheIntroduction,includingthepointslistedabove. 2)Anydiscussionof"whichtranslationrateinteractionismostimportant"or"whichtranslationrateinteractionisresponsibleforXwhereastheinteractionYonlyfine-tunesZ"isnotproductiveandcaneasilybecontradictedbyselectingarealcounterexample.Overall,itisthebindingfreeenergyofthe30SribosometothemRNAthatdeterminesitstranslationinitiationrate.EachoftheseinteractionscontributesfreeenergytothisprocessandthemagnitudeofthecontributedfreeenergiescanberoughlyequalacrossaselectionofrealmRNAexamples.ThereareunstructuredmRNAswherethereislittlepenaltyforunfoldinginhibitorymRNAstructures.TherearehighlystructuredmRNAsthathaveconsensusSDssequences.TherearemRNAsthathaveconsensusSDsequencesfarawayfromthestartcodon.AllofthesemRNAscouldhavethesametranslationrate.Whichinteractionismostimportant?That'snottherightquestiontoask,becauseit'smeaningless. Wehaveremovedlanguagethatfocusesontherelativecontributionoftheindividualfactorsthataffecttranslationalinitiation.Weagreethatouranalysesdonotallowustodeterminetheirrelativecontributions. 3)Themanuscript'smaintopicistheShine-Dalgarnosequence,buttheauthorsshouldbemadeawarethatatleastthelast9nucleotidesofthe16SrRNAcancontactthemRNAandhybridizetoit.InE.coli,theanti-ShineDalgarnosequenceis5'-ACCUCCUUA-3'andthe"consensus"Shine-Dalgarnosequenceistherefore5'-TAAGGAGGT-3'.Themanuscripttextandtheauthors'calculationsshouldreflectthis. WerevisedtheIntroductiontoexplicitlystatethatupto9bpcanform.Drawingonpreviouswork,werefertothe“consensus”asGGAGGbecauseitistheG’sthatareoverrepresentedupstreamofstartcodons(seethedataforE.coliinFigure5—figuresupplement1). 4)Theauthorsaremis-usingtheribosomeprofilingmeasurementsintheiranalysis.Ribosomeprofilingmeasurementsdonotdirectlymeasuretranslationrates.TheymeasuremRNA-boundribosomedensities.AmRNA'sribosomedensitywilldependonbothitstranslationinitiationrateANDitstranslationelongationrate.Specifically,insteady-stateconditions,theribosomedensitywillbetheratiobetweenthesetwoquantities(initiationrateoverelongationrate).Intheinitialapplicationsofribosomeprofiling,researchersassumedthatallmRNAshavethesametranslationelongationrateinordertoconcludethatribosomedensitymeasurementswereproportionaltotranslationinitiationrates.Thisisnottrue.CodingsequencesinmRNAshaveverydifferenttranslationelongationrates,duetodifferencesinsynonymouscodonusage.UnlesseachmRNAs'translationelongationratesarepredictedordirectlymeasured,ribosomedensitymeasurementscannotbeusedtoinfertheirtranslationinitiationrates.Therefore,whentheauthorswrite"Inpioneeringribosomeprofilingstudiesinbacteria,theparadoxicalobservationwasmadethatthereislittleornocorrelationbetweenthetranslationalefficiencyofageneandthestrengthofitsSDmotif(calculatedusingthermodynamicalgorithmsforRNApairing),ashadbeenanticipatedbasedontheSDmodel."thereisnoactualparadox.TheribosomeprofilingmeasurementswerenotusedcorrectlytotesthowmRNAsequencescontroltranslationrate. WerevisedtheIntroductiontoexplicitlystatethatupto9bpcanform.Drawingonpreviouswork,werefertothe“consensus”asGGAGGbecauseitistheG’sthatareoverrepresentedupstreamofstartcodons(seethedataforE.coliinFigure5—figuresupplement1). 5)Gettingtotheauthors'mainconclusions,theywritethat"ThesedataindicatethattheASDmutantribosomestranslategeneswithweakSDmotifsbetterthangeneswithstrongSDmotifs,exactlytheoppositeofwhatwild-typeribosomesareexpectedtodo."Thisstatementisconfusinggiventherealconclusionoftheauthors,thatallotherfactorsbeingequala"strong"SDmotifdoesresultinhighertranslationthana"weak"SDmotif.It'sonlybecauseofotherconfoundingfactorsthattheinitialanalysisdidnotyieldapositivecorrelation.Anincorrectanalysis(excludingconfoundingvariables)cannotleadtoacorrectconclusion. Thesentence,“ThesedataindicatethattheASDmutantribosomestranslategeneswithweakSDmotifsbetterthangeneswithstrongSDmotifs”describestheobservationsinFigure2B,theresultofalltheotherfactorsexceptforSD-ASDpairing.Weremovedthephrase“exactlytheoppositeofwhatwild-typeribosomesareexpectedtodo”thatseemstohavecausedtheconfusion. 6)Figure2Cshowsaveryinterestingandproductiveresult,thatthedifferenceintranslationefficiencybetweenthewild-type"C"ribosomesandtheA-ribosomescorrelatestosomedegreewiththehybridizationfreeenergybetweenthemRNAand(aportionof)theanti-SDsequence.Thisisaproductiveapproachtowardseliminatingkeyconfoundingvariablesbecause,inprinciple,thestrengthsofthefourotherinteractionsthatcontroltranslationinitiationrateshouldnotchangewhenthe16SrRNAaSDsequencesarechanged.However,it'snotapparentinthemanuscripttext,buttheauthorsareusingthemodified16SrRNAsto“eliminatetheSD-aSDinteractionasacontributiontothemRNA'stranslationrate”.SowhentheysubtractthecontributionfromthemodifiedA-ribosome'stranslationratesfromtheC-ribosome'stranslationrate,theyareobservingmoredirectlythecontributionfromtheSD-aSDinteraction.Themanuscripttextshouldmoreclearlyexplainthisexperimentaldesign.Thisisacreativeandvalidwayofusingribosomeprofilingmeasurements. WerevisedthelanguageattheendoftheIntroductionandthebeginningoftheResultssectiontobetterexplainourexperimentaldesign.ThefactthatwecanisolatetheeffectsofSD-ASDinteractionsfromtheotherfactorsthatsetinitiationratesexplainswhyweusestatisticsforsinglevariablesandfocusprimarilyontheSDmechanismofinitiation. 7)However,thehybridizationfreeenergycalculationscouldbeimproved.First,asmentionedpreviously,thewild-typeaSDsequenceinE.coliisACCUCCUUA.Second,thehybridizationfreeenergycalculationwasonlyperformedontheregionfrom15to6nucleotidesupstreamofthestartcodon,buttheaSDsequencecanhybridizeatotherlocations.Third,thehybridizationbetweenthemRNAandaSDcanaccommodate1or2-nucleotidebulgesorinternalloops. ThefactthatweseeastrongcorrelationbetweenourcalculatedSDaffinitiesanddifferencesinribosomeoccupancy(WT–mutant)arguesthatthecalculationsarebasicallyreliable.Weseethehighestcorrelationwhenaffinitiesarecalculatedusingthe10ntbetween-15and-6fromtheAUGandweshowthedataforvariouswindowswithdifferentSDdistancesinFigure2—figuresupplement2.Thecalculationsarequiterobusttochangesinparameters:weseelittleornodifferencesinSDROcorrelationsifweuse9ntofASDsequencetocalculateaffinitiesinsteadof7nt,orifweallowtheASDtopairanywherebetween-20to0upstreamofAUG,orifweusetheRBScalculatortogeneratetheΔGvalues. 8)ThemeasurementsincoldshockaregreatlyconfoundedbythehigherexpressionlevelsofRNAchaperonesthatareunfoldingmRNAstructures“atspecificmRNAs”wheretheRNAchaperonesrecognizebindingmotifs.TheconclusionhereshouldbethatRNAchaperonesbindspecificmRNAs,unfoldtheirinhibitorymRNAstructures,andincreasetheirtranslationratesduringcoldshock.ThisisallindependentoftheShine-Dalgarnosequence.Thisprocessalsodoesnotdependonmanyotheruninterestingfactors. WeremovedthesectionontheroleofSDmotifsunderstressbecauseitgeneratedconfusionandornithologicalreferenceswithoutstrengtheningthemainpointofourmanuscript. 9)TheuseofORF-wideGINIvaluesisoddbecauseit'sgenerallyonlytheregionsurroundingthestartcodonthataffectsitstranslationinitiationrate,andnotthestructureoftheentireORF(whichthiscoefficientisquantifying).Also,usingtheSHAPEreactivityaroundastartcodonasaproxyforRNAstructureisabitmisleadingasribosomesactivelyunfoldRNAstructuresduringtranslationinitiation.AhighlystructuredmRNAwithaconsensusSDsequencewillhaveahighSHAPEreactivity(i.e.,lowRNAstructure)becausetheribosomescanrapidlybindtothemRNAandunfoldthemRNAstructure.SHAPEreactivityismeasuringtheeffectofrapidribosomebindingandnotthecauseofit.RapidribosomebindingcanalsobefacilitatedbyslowRNArefoldingkinetics,called"RibosomeDrafting"intheliterature. CarolGrossandcolleaguesshowedthatORF-wideGINIvaluesarehighlycorrelatedwithtranslationalefficiencygenome-wide.ThisistruewhethertheDMSprobingisdoneinvivo(whereribosomescouldaffectstructurebyunwindingtheRNA)ortoalesserextentwithpurifiedRNAinvitro.KevinWeeksalsoshowedthatRNAstructuresarecorrelatedinvivoandinvitrousingtheSHAPEreagent.ThesedataarguethatatleasttosomeextentmRNAstructureisdrivingtranslationrates.ThiswasclarifiedintheResultssectioninthediscussionofFigure5E. 10)ThedatainFigure6justsaysthatA-ribosomescaninitiatetranslationrateatotherstartcodonsbecausetheynowhavemorenegativebindingfreeenergiestothosestartcodons,comparedtotheannotatedones.TheauthorscouldperformhybridizationcalculationsusingtheA-ribosome'saSDsequencetoinvestigatewhetherthese"newstartcodons"haveanearby"SD"sequencethatiscomplementarytotheA-ribosome'saSD.Thatwouldbeinteresting. ThenewFigures3and4nowincludeanalysesofsetsofinitiationsiteswithvariousaffinitiesfortheWTormutantASDsequencesshowingmoreclearlythattranslationoccursatstartcodonsevenwithoutstrongSD-ASDpairing. [Editors’note:whatfollowsistheauthors’responsetothesecondroundofreview.] Reviewer#2: ThestreamlinedpaperofSaitoetal.readsmuchbetterthantheoriginalversionanddeliversaclearandimpactfulmessage. Ibelieveitcanbepublishedafterauthorsaddresstworemainingissues: Theauthorsreferto"thenumberofelongatingribosomespermRNAasaproxyofinitiationrates".Thisisincorrect:therewouldbetwiceasmanyribosomesonanmRNAthatistwiceaslongasanotherone,evenifthosetwowouldhavethesameinitiationrate.ThecorrectmetricsisnotthenumberofribosomespermRNAbuttheribosomedensity(theirnumbernormalizedbymRNAlength).ThisdoesnotaffectconclusionsofthepaperbecauseauthorsnormalizeRiboSeqreadsbyRNASeqreads.Yet,Iwouldtrytoavoidthisconfusion. Thelanguagewaschangedto“ribosomedensity”insteadof“thenumberofribosomes.” Theauthorswrite:"Interestingly,incomparinginternalAUGcodonsthatsupportinitiationinourribosomeprofilingdatatothosethatdonot,wefoundthatA'sareenrichedbothupstreamanddownstreamofinitiationsites(Figure5A).….Thisresultsfromendogenousinitiationsites…".However,Figure5Adoesnotdealwiththeinternalinitiationsites,butwiththeannotatedsiteslackingSD. Thanksforcatchingthismistake;theDiscussionwasupdatedtoreflectthis. https://doi.org/10.7554/eLife.55002.sa2 Authordetails KazukiSaito DepartmentofMolecularBiologyandGenetics,JohnsHopkinsUniversitySchoolofMedicine,Baltimore,UnitedStates Contribution Conceptualization,Formalanalysis,Investigation,Methodology,Writing-originaldraft Competinginterests Nocompetinginterestsdeclared RachelGreen DepartmentofMolecularBiologyandGenetics,JohnsHopkinsUniversitySchoolofMedicine,Baltimore,UnitedStates HowardHughesMedicalInstitute,JohnsHopkinsUniversitySchoolofMedicine,Baltimore,UnitedStates Contribution Conceptualization,Fundingacquisition,Writing-reviewandediting Competinginterests Reviewingeditor,eLife "ThisORCIDiDidentifiestheauthorofthisarticle:" 0000-0001-9337-2003 AllenRBuskirk DepartmentofMolecularBiologyandGenetics,JohnsHopkinsUniversitySchoolofMedicine,Baltimore,UnitedStates Contribution Conceptualization,Formalanalysis,Supervision,Fundingacquisition,Writing-reviewandediting Forcorrespondence [email protected] Competinginterests Nocompetinginterestsdeclared "ThisORCIDiDidentifiestheauthorofthisarticle:" 0000-0003-2720-6896 AllenRBuskirk RachelGreen KazukiSaito Thefundershadnoroleinstudydesign,datacollectionandinterpretation,orthedecisiontosubmittheworkforpublication. TheauthorsthankDanielGoldman,ColinWu,andBorisZinshteynforcriticalreadingofthemanuscript,aswellasDavidMohrattheGeneticsResourcesCoreFacility,JohnsHopkinsInstituteofGeneticMedicine,forsequencingassistance.ThisstudywasfundedbyaJSPSfellowship(KS),NIHgrantGM110113(ARB),andHHMI(RG). JamesLManley,ColumbiaUniversity,UnitedStates JosephTWade,WadsworthCenter,NewYorkStateDepartmentofHealth,UnitedStates JosephTWade,WadsworthCenter,NewYorkStateDepartmentofHealth,UnitedStates AlexanderMankin,UniversityofIllinois,Chicago,UnitedStates Received:January9,2020 Accepted:February14,2020 AcceptedManuscriptpublished:February17,2020(version1) VersionofRecordpublished:February26,2020(version2) ©2020,Saitoetal.ThisarticleisdistributedunderthetermsoftheCreativeCommonsAttributionLicense,whichpermitsunrestricteduseandredistributionprovidedthattheoriginalauthorandsourcearecredited. 10,670 Pageviews 1,066 Downloads 31 Citations Articlecitationcountgeneratedbypollingthehighestcountacrossthefollowingsources:Crossref,Scopus,PubMedCentral. Atwo-partlistoflinkstodownloadthearticle,orpartsofthearticle,invariousformats. Downloads(linktodownloadthearticleasPDF) ArticlePDF FiguresPDF Opencitations(linkstoopenthecitationsfromthisarticleinvariousonlinereferencemanagerservices) Mendeley Citethisarticle(linkstodownloadthecitationsfromthisarticleinformatscompatiblewithvariousreferencemanagertools) KazukiSaito RachelGreen AllenRBuskirk (2020) TranslationalinitiationinE.colioccursatthecorrectsitesgenome-wideintheabsenceofmRNA-rRNAbase-pairing eLife9:e55002. https://doi.org/10.7554/eLife.55002 DownloadBibTeX Download.RIS Categoriesandtags ResearchArticle ChromosomesandGeneExpression Shine-Dalgarno ribosomeprofiling retapamulin translationalinitiation Researchorganism E.coli Ofinterest Furtherreading Furtherreading HowDNAsequenceaffectsthedynamicsandpositionofRNAPolymeraseII(PolII)duringtranscriptionremainspoorlyunderstood.HereweusednaturallyoccurringgeneticvariationinF1hybridmicetoexplorehowDNAsequencedifferencesaffectthegenome-widedistributionofPolII.WemeasuredthepositionandorientationofPolIIineightorganscollectedfromheterozygousF1hybridmiceusingChRO-seq.Ourdatarevealedastronggeneticbasisfortheprecisecoordinatesoftranscriptioninitiationandpromoterproximalpause,allowingustoredefinemolecularmodelsofcoretranscriptionalprocesses.OurresultsimplicateDNAsequence,includingbothknownandnovelDNAsequencemotifs,askeydeterminantsofthepositionofPolIIinitiationandpause.WereportevidencethatinitiationsiteselectionfollowsastochasticprocesssimilartoBrownianmotionalongtheDNAtemplate.Wefoundwidespreaddifferencesinthepositionoftranscriptiontermination,whichimpacttheprimarystructureandstabilityofmaturemRNA.Finally,wereportevidencethatallelicchangesintranscriptionoftenaffectmRNAandncRNAexpressionacrossbroadgenomicdomains.Collectively,werevealhowDNAsequencesshapecoretranscriptionalprocessesatsinglenucleotideresolutioninmammals. Auxin-inducibledegronsareachemicalgenetictoolfortargetedproteindegradationandarewidelyusedtostudyproteinfunctioninculturedmammaliancells.HerewedevelopCRISPR-engineeredmouselinesthatenablerapidandhighlyspecificdegradationoftaggedendogenousproteinsinvivo.Mostbutnotallcelltypesarecompetentfordegradation.Bycombiningligandtitrationswithgeneticcrossestogenerateanimalswithdifferentalleliccombinations,weshowthatdegradationkineticsdependuponthedoseofthetaggedprotein,ligand,andtheE3ligasesubstratereceptorTIR1.RapiddegradationofcondensinIandcondensinII-twoessentialregulatorsofmitoticchromosomestructure-revealedthatbothcomplexesareindividuallyrequiredforcelldivisioninprecursorlymphocytes,butnotintheirdifferentiatedperipherallymphocytederivatives.Thisgeneralisableapproachprovidesunprecedentedtemporalcontroloverthedoseofendogenousproteinsinmousemodels,withimplicationsforstudyingessentialbiologicalpathwaysandmodellingdrugactivityinmammaliantissues. CRISPRtechnologyhasmadegenerationofgeneknock-outswidelyachievableincells.However,onceinactivated,theirre-activationremainsdifficult,especiallyindiploidcells.Here,wepresentDExCon(Doxycycline-mediatedendogenousgeneExpressionControl),DExogron(DExConcombinedwithauxin-mediatedtargetedproteindegradation),andLUXon(lightresponsiveDExCon)approacheswhichcombineone-stepCRISPR-Cas9-mediatedtargetedknockinoffluorescentproteinswithanadvancedTet-inducibleTRE3GSpromoter.Theseapproachescombineblockadeofactivegeneexpressionwiththeabilitytore-activateexpressionondemand,includingactivationofsilencedgenes.Systematiccontrolcanbeexertedusingdoxycyclineorspatiotemporallybylight,andwedemonstratefunctionalknock-out/rescueinthecloselyrelatedRab11familyofvesicletraffickingregulators.Fluorescentproteinknock-inresultsinbrightsignalscompatiblewithlow-lightlivemicroscopyfrommonoallelicmodification,thepotentialtosimultaneouslyimagedifferentallelesofthesamegene,andbypassestheneedtoworkwithclones.Proteinlevelsareeasilytunabletocorrespondwithendogenousexpressionthroughcellsorting(DExCon),timingoflightillumination(LUXon),orbyexposingcellstodifferentlevelsofauxin(DExogron).Furthermore,ourapproachallowedustoquantifypreviouslyunforeseendifferencesinvesicledynamics,transferrinreceptorrecycling,expressionkinetics,andproteinstabilityamonghighlysimilarendogenousRab11familymembersandtheircolocalizationintripleknock-inovariancancercelllines. Signupforemailalerts Privacynotice Menu Researchcategories Authorguide Reviewerguide About InsideeLife Community Innovation Backtotop