Molecular simulations of the ribosome and associated translation factors

(1)

Molecular simulations of the ribosome and associated translation factors

Lars V Bock, Michal H Kola´ ! r and Helmut Grubmu¨ller

Theribosomeisamacromolecularcomplexwhichis responsibleforproteinsynthesisinalllivingcellsaccordingto theirtranscribedgeneticinformation.UsingX-ray

crystallographyand,morerecently,cryo-electronmicroscopy (cryo-EM),thestructureoftheribosomewasresolvedatatomic resolutioninmanyfunctionalandconformationalstates.

Moleculardynamicssimulationshaveaddedinformationon dynamicsandenergeticstotheavailablestructuralinformation, therebyhavebridgedthegaptothekineticsobtainedfrom single-moleculeandbulkexperiments.Here,wereviewrecent computationalstudiesthatbroughtnotableinsightsinto ribosomalstructureandfunction.

Address

DepartmentofTheoreticalandComputationalBiophysics,AmFaßberg 11,Go¨ttingen,Germany

Correspondingauthor:Grubmu¨ller,Helmut(hgrubmu@gwdg.de)

CurrentOpinioninStructuralBiology2018,49:27–35

ThisreviewcomesfromathemedissueonTheoryandsimulation EditedbyKrestenLindorff-LarsenandRobertBest

https://doi.org/10.1016/j.sbi.2017.11.003

0959-440X/ã2017TheAuthors.PublishedbyElsevierLtd.Thisisan openaccessarticleundertheCCBY-NC-NDlicense(http://creative- commons.org/licenses/by-nc-nd/4.0/).

Introduction

RibosomesarelargeRNA-proteincomplexeswhichsyn- thesizeproteinsinaprocesscalledtranslation[1].Trans- lation proceeds in a multi-step cycle and involves a messenger RNA (mRNA), transfer RNAs (tRNAs) and anumberofproteins(translationfactors)(Figure1).The investigation of ribosomesnotonly helpstounderstand protein synthesis and its regulation, but also offers a tremendouspotentialformedicinalapplications.Indeed, the ribosome is one of the main antibacterial drug targets, and also themainactor in theproblem of drug resistance [2].

AdvancesinX-raycrystallographyandcryo-EMprovided a wealth of structural information about ribosomes at atomic detail[4,5].However,theaccessible information onribosome dynamics,which isessentialtounderstand ribosomefunction(Figure1),islimitedduetoprecondi- tions of the experimentaltechniques. Specifically, high

structuralresolutioninX-raycrystallographyisobtained only for conformationally homogeneous crystals, and high-resolution cryo-EM requires many images of the complexinthesamechemicalandconformationalstate.

Moleculardynamics(MD)simulationshaveevolvedinto a powerful technique that complements the structural information. Based on first principles physics laws, it allows performing in silico experiments, for example, a removalofachemicallockinthestalledribosome[6^!!], that are inaccessible by other means. Obtaining free energies and transition rates from simulations allows a directcomparisontokineticexperiments,whichiscrucial forvalidationofthesimulationsandtheproposedmolec- ular mechanisms. Further, MD simulations biased by cryo-EM density maps have been successfully used as atoolforhigh-resolutionstructuredetermination(see[7]

andreferencestherein).Finally,andpossiblymostimpor- tantly, MD simulations have enabled us to move from mere correlations to an understanding of causes and effects.

TheMDfieldunderwentarespectableprogresssincethe firstall-atomMDsimulationof theentireribosome[8].

However, MDsimulations of theribosome still remain challenging for two main reasons: the large size of the ribosome and thewiderange oftime scalesrelevant to itsfunction.Tothisaim,thewholearsenalofsimulation methodsisused,rangingfromcoarse-grainedMDsimu- lations (cgMD) of the entire ribosome [9,10^!,54^!!], through structure-based MD (sbMD) [11^!,12^!], explicit-solvent all-atom MD simulations (aaMD) of entireribosome[6^!!,8,13–15,16^!,17]oritsreduced/cutout model [18–28], to a quantum mechanical/molecular mechanicalsetup(QM/MM)explicitlytreatingthequan- tum character of electrons [29,30^!!,31]. The more detailed the description of the system, the higher is thecomputationaleffort,whichleadstoshorteraccessible timescales.Duetothelimitedextentofthisreview,we recommendRefs.[32,33]foramoreindepthdiscussionof thesefundamentaltopics.

Atleastthreereviewshavebeenpublished[34–36]with similarscopetothisOpinion.Here,wefollowupprevious reviewsbySanbonmatsu[34]andA˚qvistetal.[35]from 2012.Ourreviewisorganizedaccordingtotheribosomal translation cycle. We also cover two topics which are closelyrelatedtotranslationandwheresimulationshave proven useful, namely the action of ribosome-binding antibiotics and cotranslational folding. Due to limited

(2)

space,wecanonlydiscussafractionofthemanyarticles publishedrecently.

Initiation

TranslationisinitializedontheAUGstartcodononthe mRNA.Ineukaryotes,thiscodonisrecognizedbyapre- initiationcomplex of the small subunit, initiator tRNA and several initiation factors (IFs), which scans the mRNA[37].Notmuchsimulation workhasfocusedon initiation so far. One exception is a work of Lind and A˚qvistwho investigatedtheroleofIFsoncodonrecog- nition[27].UsingaaMDtheycalculatedrelativebinding freeenergiesforsingle-pointmutationsofthestartcodon inthepresenceandabsenceoftwoIFs.Thesimulations suggestthatthepresenceoftheIFsonthepre-initiation complexincreasestheenergeticpenaltyforbindingnon- cognatecodonsand,thereby,thefidelityofcognatestart codonrecognitionisenhanced.

Decoding

Alargefraction ofsimulationworkhasfocussedonthe decodingstepprecedingpeptideelongation,presumably because many important details of decoding do not involvelarge-scaleconformationalrearrangements.Dur- ing elongation, aminoacyl-tRNAs are delivered to the ribosome in theform of a ternary complex:the tRNA,

atranslationalGTPase(inbacteria:EF-TuorSelB),anda GTPmolecule.The tRNA decodes theinformationon the mRNA by forming hydrogen bonds (H-bonds) betweencodonandanticodonnucleobases.Remarkably, thefree-energydifferencebetweencorrect(cognate)and incorrect(near-cognate, non-cognate)basepairingalone doesnotexplaintheveryhighfidelityofdecoding[38].

Rather,highfidelityisachievedbyatwo-stepdecoding process:initialselectionleadingtoGTPaseactivationand proofreading. In addition to thefree-energy difference, kineticeffectscontributetothediscrimination.TheGTP hydrolysis rate is increased and tRNA rejection rate is decreasedbytherecognitionofthecorrectcodon [38].

Free-energy aaMD simulations of the decoding region wereusedtoinvestigatethediscriminationbetweennear- cognate and cognate base pairs [22]. Small-subunit nucleotidesA1492andA1493adoptaflipped-outconfor- mationinthepresenceofatRNAand,inthisconforma- tion,interactwiththecodon-anticodonmini-helix.Inthe simulations,flippedoutnucleotidesA1492,A1493along withG530werefoundtoshieldthecodon–anticodonbase pairsfromsolvent.Thisshieldingpreventsinteractionsof near-cognatebasepairswiththesolvent,therebyincreas- ingthefree-energydifferencebetweennear-cognateand cognatebasepairsandthusthediscrimination[22].The

Figure1

BACTERIAL RIBOSOME BACTERIAL

TRANSLATION peptidyl transferase center

exit tunnel

RECYCLING INITIATION

ELONGATION

subunits peptide

formationbond

decoding

hybrid state formation

GTP-dependent translocation assembly

30S

50S L1 stalk IF3

IF1

IF2 fM-tRNA

mRNA

RRF EF-G

RF3

RF1/2 ^EF-G

aa-tRNA EF-Tu

5 nm

EF-Tu

TERMINATION

release proteinnew P-site tRNA

decoding center pre-accommodated tRNA L1 stalk

E-site tRNA large subunit

small subunit

Current Opinion in Structural Biology

StructureofthebacterialribosomeincomplexwithEF-Tu(PDB5AFI[3]).Schemeofthebacterialtranslationcycleasreviewedin[4].30S:small subunit;50S:largesubunit;IF1,IF2,IF3:initiationfactors;fM-tRNA:N-formylmethioninetRNA;aa-tRNA:aminoacyltRNA;EF-Tu,EF-G:elongation factors;RF1,RF2,RF3:releasefactors;RRF:ribosomerecyclingfactor;greentrace:nascentprotein.Thequestionmarkstandsforastopcodon recognition.

(3)

ofnear-cognatecomplexeswhichimplythatthedecoding centerenforcesabase-pairgeometryofmismatchedbase pairsclosetothatofacanonicalbasepair[39].Thisleads to areductionof possible H-bondsrelativeto arelaxed conformation.

Umbrella sampling aaMD of the in-flipping and out- flippingofA1492/A1493inasimulationsystemconsisting of thedecoding center suggested a more active role of these nucleotides[24].Inthesimulations,aflippedout conformation was seen to be more favourable for cognate than for near-cognate base pairs, which would increase the discrimination by stabilizing the cognate codon–anticodon helix.

Aminoglycosidesareaclassofantibioticsthatbindtothe decodingcenterandlocknucleotidesA1492/A1493inthe flipped-out conformation. In this way aminoglycosides promotetheaccommodationofnear-cognate,thuswrong, tRNAs.Paneckaetal.carriedoutaaMDofthedecoding center with bound aminoglycoside paromomycin and resistance-inducing mutations in protein uS12 [21]. An increased flipping rate relative to the wild type was observed,suggestingthatthemutationrestoresthefunc- tionof thesenucleotides.

Inarecentcryo-EMstudy,intermediatestructuresalong the pathway of initial selection leading to the GTPase activationof SelBwereresolvedathighresolution[26].

BothtRNAandSelBundergosubstantialconformational changesduringtheprocesswhichcouldpresentakinetic barrier controlling GTPase activation. To address the questioniftheseconformationalchangesbythemselves arerate-limitingtotheprocess,aaMDofthefreeternary complex in solution was performed. The simulations werestarted fromtheribosome-boundcryo-EMconfor- mations. The tRNA and SelB rapidly interconverted between the different conformations, which allowed theconstructionoftheconformationalfree-energylandscape. The landscape indicates thatthe intrinsic large- scale conformational changes of the tRNA and SelB duringthedeliverytotheribosomearenotrate-limiting to theprocess.

IntheGTPase-activatedstate,theGTPbindingdomain ofEF-TuorSelBdocksontothesarcin-ricinloopofthe large subunit. Wallin et al. used aaMD of the GTP binding site of EF-Tu in the activated state on the ribosometostudytheGTPhydrolysis[20].Free-energy perturbationsimulationssuggestedthatHis81ofEF-Tu cannot act as a general base in the reaction, as was previouslyproposed, butrather stabilizes awatermole- cule involved in the reaction. Further, the sarcin-ricin loop seemsto promotetheGTPaseactivatedconforma- tionofaconservedtripeptidemotifPGHwhichcontains His81.TheconformationofthePGHandMg²⁺positions

high-resolution X-ray structures (see references within [30^!!]). aaMD in combination with empirical valence bond (EVB) method suggested that His81 is doubly protonated which promotes a proton transfer from the stabilizedwatermolecule totheGTPg-phosphate,fol- lowedbyanucleophilicattackofthehydroxideion[30^!!].

By calculatingArrhenius plotsfrom MD simulations at multiple temperatures A˚qvist et al. identified a large entropic contributiontothehydrolysisreaction[40].

After GTPhydrolysis,theGTPase dissociates fromthe tRNAwhichallowsthetRNAtomoveintotothepepti- dyl transferase center (PTC) on the large subunit. A tRNA accommodationcorridor was identifiedusing tar- getedaaMDoftheentireribosome[8].aaMDandsbMD oftRNAaccommodationsuggestedthatthereisanaddi- tional intermediate state between the pre-accommodation(A/T)andthefullyaccommodatedstates[8,11^!].Itis characterizedbythetRNAelbowintheaccommodated stateandtheCCA-tailnotyetinthePTC.RecentsbMD suggests that EF-Tu sterically reduces the range of accessible tRNA conformations,specifically in the A/T state [11^!]. Therefore, the presence of EF-Tu is pre- dictedtodestabilizetheA/Tstateandtherebyenhance therateof accommodation.

Peptide bondformation

At the core of ribosomal translation is the catalysis of peptidebondformation[1].Thecurrentreactionmodels pointtoasubstrateassistedmechanism.EarlyaaMDwith EVB calculations suggested that the ribosome reduces thesolventreorganizationenergybyprovidingastableH- bond network, which wouldenhance the peptidebond formation rates[41].

MDsimulationsidentifiedpositionsofseveralwatermole- cules and H-bonds critical for the reaction which were subsequentlyconfirmedbyX-raystructures[42,43].QM/

MM simulationsaswellas high-levelquantum chemical calculations indicated that the transition state forms an eightmemberedringwhichincludesawatermoleculeand that theC–Obondcleavage takesplaceafterC–N bond formation [29,44].These studies reproducedthe experi- mentally observed catalytic effect. A recent QM/MM study additionally proposed the presence of a Mg²⁺ ion inthesurroundingofthePTCandhasshownthatinclud- ingtheionimprovedagreementofthecalculatedwiththe measuredcatalyticeffect,underscoringtheimportanceof ionsin computationalstudiesoftheribosome[31].

SincethePTCisburiedwithinthelargesubunit,during translationthenascentchain(NC)exitsthrougha100A˚

longtunnel(Figure2a). Theexittunnelplaysanactive role in protein synthesis. Certain peptide sequences specifically interact with tunnel walls and induce ribosome stalling[45].Further,theexit tunnelis abinding

(4)

site for a clinically important class of antibiotics [2].

SeveralMDstudieshaveaddressedtheseissues.

Whensynthesizingproteins containingprolinestretches (i.e.severalprolinesinarow),ribosomesbecomestalled.

Stalling is alleviated bya specialized elongation factor, EF-Pinbacteria[46,47].Recently,cryo-EMstructuresof aribosomestalledbyaprolinestretchwithandwithout EF-P were resolved [28]. In aaMD simulations of the PTCregion, EF-P was observed to stabilize theP-site tRNAin aconformationcompatible withpeptide bond formation, while in the absence of EF-P, the tRNA movedawayfromtheA-site tRNA[28].

Acommunicationpathwaybetweenthetunnelwallsand thePTCwasidentifiedinantibiotic-dependentribosome stalling. Biochemical experiments complemented by aaMD of theentire Escherichia coli ribosome suggested [23]thatamacrolideantibioticerythromycin(ERY)allo- sterically alters the properties of the PTC without any directcontactwiththeNC (Figure 2).The simulations captureda dramaticreorientation of U2585and A2602, both more than 8A˚ away from the nearest ERY atom.

Besides this, ERY binding induced a conformational change of its neighbor A2062, a nucleotide claimed to serveasanascent-chainsensor[48].

TheroleofA2062wasfurtherhighlightedbyaaMDofa cubicreducedmodelofthePTCanditssurroundingsin

the presence/absence of ERY [25]. Over 20 unbiased trajectories, each 200–360ns, suggested that allosteric signal transmission occurs via formation of a stem of stacked large-subunit rRNA nucleobases and that, further, thestemformation isinitiated by theA2062 con- formationalchange.

ERY-induced stalling of an ErmB leading peptide (ErmBL) [6^!!] was studied using aaMD of the entire E.coliribosomewithErmBLpeptideinthepresenceand absenceof ERY, complementingcryo-EMexperiments which cannot resolve unstalled ribosome-ErmBL com- plexes (without ERY). The simulations suggested that the ERY induced conformational changes in the PTC resultin ashift ofaLys11ontheA-sitetRNA,thereby preventing peptide bond formation. The simulations predictedamutationoftheLys11toenhancethestalling abilityofErmBL,whichwassubsequentlyconfirmedby biochemicalexperiments[6^!!].

The resistance to a ketolide antibiotic telithromycin (TEL)wasaddressedbyMacKerell’sgroup [18].aaMD ofasphericalreducedmodelwerecombinedwithgrand canonical Monte Carlo to enhance sampling of water positions in the exit tunnel. The study compared the wild-type complex to three drug-resistant variants:

A2058G mutant as well as mono-methylated and dimethylated A2058. Based on geometric analyses of H-bonds and other contacts it was rationalized why

Figure2

(a)

uL29 uL23

uL23

uL22 180°

uL22 uL22

uL4

uL4 uL4

A2062

A2058

ERY Lys11

A2062

U2585

A2602 NC

P-site tRNA A-site tRNA ERY

tunnel exit

P-site

tRNA A-site

tRNA

P-site tRNA vestibule

constriction

PTC

uL24

(b)

(a)Schemeoftheribosomeexittunnelwithseveralproteinshighlighted.NC,nascentchain;ERY,erythromycin;PTC,peptidyltransferasecenter.

(b)Contextoftheerythromycin(ERY,ingreen)binding;figurebasedonPDB:5JTE[6^!].Severallargesubunitnucleotidesarehighlightedinbold red.TwoproteinsuL4anduL22formaconstrictionsite.Thenascentpeptideisshownastransparentsurface.

(5)

tion but remains susceptible to methyl-mediated resistance.The keyplayer seemstobeanH-bondbetween TEL:2⁰-OHandA2058thatismaintainedaftermutation to G2058,but disruptedbymethylations.

Cotranslational proteinfolding

Theexittunnelcanaccommodate30–60AAs,depending onthelevelofNCcompaction.Therateoftranslationof about4–22 AApersecondin bacteria [49] providesthe NC with sufficient time to explore its conformational space and to startfoldingwhenstill boundto theribosome-tRNA complex. A number of simulation studies have tackledcotranslational folding(see Ref.[50] for a dedicated review). Particular interest liesin thetrigger factor(TF),thefirstchaperoneencounteredbytheNC.

FreeTFinsolutionexhibitsamonomer–dimerequilib- rium.ItwasstudiedbyseveralgroupsusingaaMD[51–

53].TheTF’sN-terminalandheaddomainscanassociate makingacompacttertiarystructure.Theconclusionthat TF is very flexible is supported across all the studies, however thestability ofthecompact structure isforce- fielddependentandremainselusive.Itwaspointedout though that the compact structure prevents TF from binding to the ribosome, as supported by an elastic networkmodelofthelargesubunitandTFcomplex[53].

TheassociationofTFandE.coliribosomewasstudiedby cryo-EMandcgMDofareducedribosomemodel[10^!].

Several 1.2-ms long trajectories showed interdomain motions: whiletheN-terminaldomain remainedbound to the ribosome, the head domain fluctuated between boundandunboundstates.Themotionsweresimilarto those identified in solution simulations [51,52]. In the ribosome-bound simulations, longer NC made the TF morerigidascomparedtotheshorterNCconstruct.The authors speculated that the loss of TF flexibility may facilitateTFunbindingfromtheribosome,whilekeep- ingTFandNC stilltogether.

UsingNMRandcgMD,Deckertetal.studiedtheroleof TF in synthesis of the disordered peptide a-synuclein (aSyn) [54^!!]. Decent agreement between NMR and cgMDwasobserved.The authorssuggestedweakasso- ciationofaSynandribosomesurfaceandconcludedthat there might be a specific affinity on the surface for aromatic residues. About 50 AAs are needed from the PTCto initiateinteractionswithTF.

NMRwasusedtostudycotranslationalfoldingofapairof immunoglobulin-like domains FNL5 and FLN6 [17].

aaMDrestrainedbychemicalshiftsgeneratedanensem- ble of protein conformations on the ribosome. The N- terminal FNL5 domainwas shownto adoptnative-like fold only after emerged well beyond the tunnel. The FNL6, whichwaslocatedclosertotheC-terminusthan

interactedwith theribosomalsurface.

Acombinedsimulation-experimentalstudyshowedthata protein can fold already inside the tunnel vestibule (Figure 2a)[9].Azinc-fingerdomain,probedbycgMD, cryo-EM, and biochemicalexperiments onstalled ribosomes, folds in the tunnel between uL22 and uL23 proteins. Thefoldingwas observedincgMDatphysio- logical (310K) as wellas at cryo-EM(140K) temperatures, although the structural agreement with the cryo-EM modelwasratherpoor.

tRNA translocation

After peptidebond formationtheNCisattachedtothe tRNAresidingintheAsite,thetRNAinthePsiteisleft deacylatedandtheEsiteisunoccupied(Figure3a).The two tRNAsthentranslocate tothePandEsites,either spontaneouslyonslowtimescales,orveryrapidlyinthe presence of thetranslational GTPase EF-G.Transloca- tion of tRNAs is accompanied by large-scale collective motions of the ribosome: relative rotation of ribosomal subunitsand L1-stalkmotion.The L1stalk,which isa flexiblepartofthelargesubunit,isincontactandmoves alongwiththetRNAfromthePtotheEsite(Figure3b).

Whitford et al. extractedeffective diffusion coefficients for small subunit head and body rotations and tRNA displacementfromanaaMDoftheclassicalpre-translo- cationstate[13].Thediffusioncoefficientstogetherwith experimentalratesofthesemotionsprovideupper-bound estimatesof free-energybarriers.

Toobtaindynamicsandenergeticsofintermediatestates of spontaneous translocation, X-ray structures were refinedagainstcryo-EMreconstructions,therebyobtain- ing 13 near-atomic resolution structures [14]. From aaMDsoftheseintermediatestates,order-of-magnitude transition rates between states were estimated for motions of the L1 stalk and the tRNAs as well as for intersubunit rotations(Figure 3c). Theserates revealed rapiddynamicsoftheL1stalkandintersubunitrotations on sub-microsecond timescales, whereas the tRNA motionswereseentoberate-limitingformosttransitions.

Further,calculatedmolecularforcesrevealedthattheL1 stalkisactivelypullingthetRNAfromPtoEsiteasone of themainmechanismsacceleratingbarriercrossing.

UsingaaMD,itwasshownthattheintersubunitcontact network adaptsto differentintersubunitrotation angles resulting in a relatively constant intersubunit binding enthalpy[16^!].Arotation-independent affinitybetween the subunit is a prerequisite for rapid rotation since differentaffinitieswouldleadtobarriershinderingrota- tion.Inadditiontothecentralcontactsclosetotheaxisof rotation, peripheral contacts were found to be strong, despitetheirlargerelativeshift.Thissteadycontribution

(6)

isachievedbyexchangingcontactpartnersinthecourse ofrotation.

Recently, sbMDsimulations were applied to studythe relation between small subunit head rotation and the mRNA–tRNA translocation [12^!]. The authors chose thenon-rotatedstateoftheribosomeasanenergymini- mumaswellastwoadjacentbindingsitesforeachofthe twotRNAs.Themodel wasabletoreproduceaknown intermediate tRNA binding conformations (ap/P–pe/E) and predictedan intermediatestate with a tilted head, whichhasnotbeenobservedinexperimentsyet.Inthis proposed intermediate, the PE loop, which sterically separatesthePandEbindingsitesonthesmallsubunit, isdisplaced,possiblyallowingtheanticodonstemloopof thetRNAtomovefromPtoEsite.Afterremovalofsteric interactionsoftRNAwithsmallsubunitproteinuS13and the PE loop, the tilted head state was not populated indicatingthattiltingresults fromtheseinteractions.

Inanearlierstudy,Ishidaetal.appliedacryo-EMfitting MDapproachto studyreversetranslocationinthepres- enceofEF-G[15].Thesimulationswerestartedfroma post-translocation structure and then driven towards a pre-translocationstatebymaximisingthecorrelationtoa pre-translocationcryo-EMmap.Afree-energylandscape wasestimatedusingumbrellasimulations.Unexpectedly, they observed a clockwise rotation in the simulations, whengoingfromthepoststatetotheintermediatestate, whilefrom cryo-EMstructures of intermediate statesa

counter-clockwise rotation was expected. The authors speculatedthatthemovementoftheheadduringreverse translocationmight bedifferent fromthat in whatthey call ordinary translocation. However, since the simulations describe an equilibrium process, there can be no difference due to directionality. The tilted head state describedbyNguyenetal.[12^!]wasnotobservedinthe simulationsbyIshidaetal.Thedifferencesbetweenthe observed free-energy landscapes underscorethe severe sampling problem which is a major challenge for all simulationsofsuchlargemolecularcomplexes.

Termination

Thetranslationcycleterminatesinaseriesofstepsafter anmRNAstopcodonispresentedintheAsite(Figure1).

First,areleasefactor(RF)recognizesastop codon and hydrolyzesthepeptidyl-tRNA bondinthePTC.Inthe laststepof termination,thelargeand smallsubunits of theribosomeseparate.Thesubunitsarerecycledinthe nexttranslationround.

Inmitochondria,non-standardstopcodonshaveevolved [55].Invertebrates,itisstillunclearwhatfactorsrecog- nizestopcodonsandhowmanystopcodonsactuallyexist.

Lindetal.carriedoutaaMDofareducedsphericalmodel and calculated relative bindingfree energies of several codonswhenboundtoavarietyofRFs[19].Theirresults suggestthatneitherofthetwomitochondrialRFsmtRF1 and mtRF1a, homologsto the bacterial RF1,is able to read non-standard AGA nor AGG stop codons. The

Figure3 (a)

50S

30S

30S head

30S body (b)

(c)

90°

post4 pre1

pre1 pre2

pre2 pre3

pre3 pre4

pre4

transition rates >1/(25 ns) 1/µs <1/µs pre5

pre5 post1

post1 post2

post2 post3

post3 post4

post4 post4

90° head

tilting

head rotation

body rotation

E P A E P A

L1-stalk motion

tRNA^fMetmotion tRNA^Val

motion

tRNA^Val motion tRNA^fMet

motion L1-stalk motion

(a)Pre-translocationstructureoftheribosomewithtRNAsinAandPsites(green,purple).(b)MotionsaccompanyingtRNAtranslocation.(c) Transitionratesfordifferentmotionsbetweendifferentstatesestimatedfromthesimulations.FigureadaptedfromRef.[14].

(7)

couldberesponsibleforthepeptidereleaseinmitochon- driaassupportedbylaterbiochemicalexperiments[56].

The hydrolysisof peptidyl-tRNA bond allowsreleasing the newly synthesized proteinfrom the ribosome.The atomisticdetailsofthehydrolysismechanismarenotyet fullyunderstood.Kazemietal.testedseveralmechanisms bymeansofdensityfunctionaltheoryonareducedmodel ofthebacterialPTC(224atoms)[57^!].Theyconcluded thatabase-catalyzedmechanism,whichinvolvesadepro- tonation of the P-site tRNA A76 2⁰-OH group, is the only one consistent with the experimental activation energies, kinetic solvent isotope effect values, and pH dependence.

Summary and outlook

Anincreasingnumberoftheoryandcomputationresearch groups haveacceptedthechallengesposed byamacro- molecularsystemofthehugesizeofaribosome,yielding awealthofmechanisticinsightsintothisamazingmolec- ular machine. This advance has been fueled by the

‘resolution revolution’ of the cryo-EM field, and the obtained high-resolution structures of many conforma- tionalstates.Mostofthestrikingresultsinthisfieldhave not been obtained by isolated simulation work, but in closecollaborationswithotherfields,mostnotablycryo- EM,spectroscopy,andbiochemistry.

Inthiscontext,weseeafewavenueswhichhave,inour view, not yet been exploited to their full potential by simulations.First,thenewhigh-resolutioncryo-EMmaps obtainedathighpacedemandimprovedandautomated refinement protocols, where MD may playa vital role.

Second, realistic simulations of single-molecule experi- mentswillenhancethestructuralanddynamicinterpre- tation.And,finally,morereliableprotocolsandmethods foraccuratefree-energycalculationsarekeytoaquanti- tativeandcausalunderstandingof howribosomeswork.

Wethinksuchanunderstandingwillhavetobeinterms of molecular drivingforces,Markov processes, andulti- matelythefulldynamicswithinamultidimensionalfree- energylandscape.Understandingtheribosomeremains, therefore,notonlyacomputationalandmethodological, but alsoaconceptualchallenge.

Acknowledgements

TheworkwassupportedbytheMaxPlanckSocietyandbytheDeutsche Forschungsgemeinschaft(FOR1805).

References and recommendedreading

Papersofparticularinterest,publishedwithintheperiodofreview, havebeenhighlightedas:

! ofspecialinterest

!!ofoutstandinginterest

1. RodninaMV,BeringerM,WintermeyerW:Howribosomesmake peptidebonds.TrendsBiochemSci2007,32:20-26http://dx.doi.

2. WilsonDN:Ribosome-targetingantibioticsandmechanisms ofbacterialresistance.NatRevMicrobiol2014,12:35-48http://

dx.doi.org/10.1038/nrmicro3155arXiv:1011.1669v3http://www.

nature.com/doifinder/10.1038/nrmicro3155.

3. FischerN,NeumannP,KonevegaAL,BockLV,FicnerR, RodninaMV,StarkH:StructureoftheE.coliribosome-EF-Tu complexat<3A˚ resolutionbyCs-correctedcryo-EM.Nature 2015,520:567-570http://dx.doi.org/10.1038/nature14275.

4. SchmeingTM,RamakrishnanV:Whatrecentribosome structureshaverevealedaboutthemechanismoftranslation.

Nature2009,461:1234-1242http://dx.doi.org/10.1038/

nature08403arXiv:NIHMS150003,http://www.nature.com/

doifinder/10.1038/nature08403.

5. FrankJ:Thetranslationelongationcycle–capturingmultiple statesbycryo-electronmicroscopy.PhilosTransRSocLondB BiolSci2017,372In:http://rstb.royalsocietypublishing.org/

content/372/1716/20160180.

6.

!!

ArenzS,BockLV,GrafM,InnisCA,BeckmannR,Grubmu¨llerH, VaianaAC,WilsonDN:Acombinedcryo-EMandmolecular dynamicsapproachrevealsthemechanismofErmBL- mediatedtranslationarrest.NatCommun2016,7:12026http://

dx.doi.org/10.1038/ncomms12026http://www.nature.com/

doifinder/10.1038/ncomms12026..

All-atomMDsimulationsoftheentireribosometogetherwithcryo-EM reveal moleculardetailsofan antibiotic-dependent ribosomalstalling mechanismwhichareconfirmedbytoeprintingexperiments.Thisstudy emphasizesthebenefitsofaninterdisciplinaryapproach.

7. KirmizialtinS,LoerkeJ,BehrmannE,SpahnCMT,

SanbonmatsuKY:Usingmolecularsimulationtomodelhigh- resolutioncryo-EMreconstructions.MethodsEnzymol2015, 558:497-514http://dx.doi.org/10.1016/bs.mie.2015.02.011In:

http://linkinghub.elsevier.com/retrieve/pii/S0076687915001846.

8. SanbonmatsuKY,JosephS,TungC-S:Simulatingmovementof tRNAintotheribosomeduringdecoding.ProcNatlAcadSciU SA2005,102:15854-15859http://dx.doi.org/10.1073/

pnas.0503456102In:http://www.ncbi.nlm.nih.gov/pubmed/

16249344.

9. NilssonOB,HedmanR,MarinoJ,WicklesS,BischoffL, JohanssonM,Mu¨ller-LucksA,TrovatoF,PuglisiJD,O’BrienEP, BeckmannR,vonHeijneG:Cotranslationalproteinfolding insidetheribosomeexittunnel.CellRep2015,12:1533-1540.

10.

!

DeengJ,ChanKY,vanderSluisEO,BerninghausenO,HanW, GumbartJ,SchultenK,BeatrixB,BeckmannR:Dynamic behavioroftriggerfactorontheribosome.JMolBiol2016, 428:3588-3602http://dx.doi.org/10.1016/j.jmb.2016.06.007In:

http://www.sciencedirect.com/science/article/pii/

S0022283616302157.

Cryo-EMstudyiscomplementedbycoarse-grainedMDsimulationsof theribosomewithtriggerfactor.Theyauthorsproposeaway,howthe triggerfactorinteractswiththeemergingnascentchain.

11.

!

NoelJK,WhitfordPC:HowEF-Tucancontributetoefficient proofreadingofaa-tRNAbytheribosome.NatCommun2016, 7:13314.

A unique combination of structure-based MD of a ribosome EF-Tu complexandkineticmodellingsuggestedthatthepresenceofEF-Tu enhancestherateoftRNAaccommodation.

12.

!

NguyenK,WhitfordPC:Stericinteractionsleadtocollective tiltingmotionintheribosomeduringmRNA–tRNA translocation.NatCommun2016,7:10586http://dx.doi.org/

10.1038/ncomms10586In:http://www.scopus.com/inward/

record.url?eid=2-s2.0-84957560646&partnerID=tZOtx3y1.

Structure-based MDsimulations ofspontaneous tRNA translocation, including and excluding certain interactions, indicate that the small subunitheadtiltingisaconsequenceoftRNA–mRNAinteractions.The possibility to switch off interactions computationally helps revealing causality.

13. WhitfordPC,BlanchardSC,CateJHD,SanbonmatsuKY,LeeH:

ConnectingthekineticsandenergylandscapeoftRNA translocationontheribosome.PLoSComputBiol2013,9:

e1003003http://dx.doi.org/10.1371/journal.pcbi.1003003.

(8)

14. BockLV,BlauC,Schro¨derGF,DavydovII,FischerN,StarkH, RodninaMV,VaianaAC,Grubmu¨llerH:Energybarriersand drivingforcesintRNAtranslocationthroughtheribosome.Nat StructMolBiol2013,20:1390-1396http://dx.doi.org/10.1038/

nsmb.2690.

15. IshidaH,MatsumotoA:Free-energylandscapeofreversetRNA translocationthroughtheribosomeanalyzedbyelectron microscopydensitymapsandmoleculardynamics simulations.PLOSONE2014,9:e101951http://dx.doi.org/

10.1371/journal.pone.0101951In:http://journals.plos.org/

plosone/article?id=10.1371/journal.pone.0101951.

16.

!

BockLV,BlauC,VaianaAC,Grubmu¨llerH:Dynamiccontact networkbetweenribosomalsubunitsenablesrapidlarge- scalerotationduringspontaneoustranslocation.NucleicAcids Res2015,43:6747-6760http://dx.doi.org/10.1093/nar/gkv649 http://www.ncbi.nlm.nih.gov/pubmed/26109353http://www.

pubmedcentral.nih.gov/articlerender.fcgi?artid=PMC4538834.

All-atomMDofentireribosomeisusedtosimulate13conformational states.Combining thedynamics obtainedforthedifferentstates,the studysuggestsintersubunitinteractionshaveevolvedtoenable rapid large-scalerotation.

17. CabritaLD,CassaignauAM,LaunayHM,WaudbyCA, WlodarskiT,CamilloniC,KaryadiM-E,RobertsonAL,WangX, WentinkAS,GoodsellLS,WoodheadCA,VendruscoloM, DobsonCM,ChristodoulouJ:Astructuralensembleofa ribosome-nascentchaincomplexduringcotranslational proteinfolding.NatStructMolBiol2016,23:278-285.

18. SmallMC,LopesP,AndradeRB,MacKerellAD:Impactof ribosomalmodificationonthebindingoftheantibiotic telithromycinusingacombinedgrandcanonicalMonteCarlo/

moleculardynamicssimulationapproach.PLoSComputBiol 2013,9:e1003113http://dx.doi.org/10.1371/journal.

pcbi.1003113.

19. LindC,SundJ,A˚qvistJ:Codon-readingspecificitiesof mitochondrialreleasefactorsandtranslationterminationat non-standardstopcodons.NatCommun2013,4:2940http://dx.

doi.org/10.1038/ncomms3940In:http://www.ncbi.nlm.nih.gov/

pubmed/24352605.

20. WallinG,KamerlinSCL,A˚qvistJ:EnergeticsofactivationofGTP hydrolysisontheribosome.NatCommun2013,4:1733http://

dx.doi.org/10.1038/ncomms2741http://www.nature.com/

doifinder/10.1038/ncomms2741.

21. PaneckaJ,MuraC,TrylskaJ:Interplayofthebacterial ribosomalA-site,S12proteinmutationsandparomomycin binding:amoleculardynamicsstudy.PLOSONE2014,9:

e111811http://dx.doi.org/10.1371/journal.pone.0111811http://

dx.plos.org/10.1371/journal.pone.0111811.

22. SatpatiP,SundJ,A˚qvistJ:Structure-basedenergeticsof mRNAdecodingontheribosome.Biochemistry2014,53:1714- 1722http://dx.doi.org/10.1021/bi5000355http://pubs.acs.org/

doi/abs/10.1021/bi5000355.

23. SothiselvamS,LiuB,HanW,RamuH,KlepackiD,AtkinsonGC, BrauerA,RemmM,TensonT,SchultenK,Va´zquez-LaslopN, MankinAS:Macrolideantibioticsallostericallypredisposethe ribosomefortranslationarrest.ProcNatlAcadSciUSA2014, 111:9804-9809http://dx.doi.org/10.1073/pnas.1403586111 http://www.ncbi.nlm.nih.gov/pubmed/24961372http://www.

pubmedcentral.nih.gov/articlerender.fcgi?artid=PMC4103360.

24. ZengX,ChughJ,Casiano-NegroniA,Al-HashimiHM,BrooksCL:

FlippingoftheribosomalA-siteadeninesprovidesabasisfor tRNAselection.JMolBiol2014,426:3201-3213http://dx.doi.org/

10.1016/j.jmb.2014.04.029In:http://linkinghub.elsevier.com/

retrieve/pii/S0022283614002307.

25. MakarovGI,GolovinAV,SumbatyanNV,BogdanovAA:

Moleculardynamicsinvestigationofamechanismofallosteric signaltransmissioninribosomes.BiochemMosc2015, 80:1047-1056http://dx.doi.org/10.1134/S0006297915080106 http://link.springer.com/10.1134/S0006297915080106.

26. FischerN,NeumannP,BockLV,MaracciC,WangZ,PaleskavaA, KonevegaAL,Schro¨derGF,Grubmu¨llerH,FicnerR,RodninaMV, StarkH:ThepathwaytoGTPaseactivationofelongationfactor SelBontheribosome.Nature2016,540:80-85http://dx.doi.org/

10.1038/nature20560.

27. LindC,A˚qvistJ:Principlesofstartcodonrecognitionin eukaryotictranslationinitiation.NucleicAcidsRes2016, 44:8425-8432http://dx.doi.org/10.1093/nar/gkw534https://

academic.oup.com/nar/article-lookup/doi/10.1093/nar/gkw534.

28. HuterP,ArenzS,BockLV,GrafM,FristerJO,HeuerA,PeilL, StarostaAL,WohlgemuthI,PeskeF,Nova´9cekJ,BerninghausenO, Grubmu¨llerH,TensonT,BeckmannR,RodninaMV,VaianaAC, WilsonDN:Structuralbasisforpolyproline-mediatedribosome stallingandrescuebythetranslationelongationfactorEF-P.

MolCell2017,68:515-527http://dx.doi.org/10.1016/j.

molcel.2017.10.014e6.http://linkinghub.elsevier.com/retrieve/pii/

S109727651730789X.

29. XuJ,ZhangJZH,XiangY:AbinitioQM/MMfreeenergy simulationsofpeptidebondformationintheribosome supportaneight-memberedringreactionmechanism.JAm ChemSoc2012,134:16424-16429http://dx.doi.org/10.1021/

ja3076605http://pubs.acs.org/doi/abs/10.1021/ja3076605.

30.

!!

A˚qvistJ,KamerlinSCL:Theconformationofacatalyticloopis centraltoGTPaseactivityontheribosome.Biochemistry2015, 54:546-556http://dx.doi.org/10.1021/bi501373ghttp://pubs.acs.

org/doi/abs/10.1021/bi501373g..

ExtensiveQM/MMsimulationsofGTPhydrolysisonEF-Tugiveimportant insightsintothereactionmechanisminlinewithkineticexperimentsand, moreover,areabletorationalizetheeffectsofmutations.

31. Swiderek" K,MartiS,Tun˜o´nI,MolinerV,BertranJ:Peptidebond formationmechanismcatalyzedbyribosome.JAmChemSoc 2015,137:12024-12034http://dx.doi.org/10.1021/jacs.5b05916 http://pubs.acs.org/doi/10.1021/jacs.5b05916.

32. VangavetiS,RanganathanSV,ChenAA:AdvancesinRNA moleculardynamics:asimulator’sguidetoRNAforcefields.

WileyInterdiscipRevRNA2017http://dx.doi.org/10.1002/

wrna.1396.http://doi.wiley.com/10.1002/wrna.1396.

33. ponerJ,KreplM,Bana´P,Ku¨hrova´ P,Zgarbova´ M,Jure9ckaP, HavrilaM,OtyepkaM:Howtounderstandatomisticmolecular dynamicssimulationsofRNAandprotein-RNAcomplexes?

WIRESRNA2017,8:e1405http://dx.doi.org/10.1002/wrna.1405 http://doi.wiley.com/10.1002/wrna.1405.

34. SanbonmatsuKY:Computationalstudiesofmolecular machines:theribosome.CurrOpinStructBiol2012,22:168-174 http://dx.doi.org/10.1016/j.sbi.2012.01.008In:http://linkinghub.

elsevier.com/retrieve/pii/S0959440X12000231.

35. A˚qvistJ,LindC,SundJ,WallinG:Bridgingthegapbetween ribosomestructureandbiochemistrybymechanistic computations.CurrOpinStructBiol2012,22:815-823http://dx.

doi.org/10.1016/j.sbi.2012.07.008In:http://www.sciencedirect.

com/science/article/pii/S0959440X12001121.

36. MakarovGI,MakarovaTM,SumbatyanNV,BogdanovAA:

Investigationofribosomesusingmoleculardynamics simulationmethods.BiochemMosc2016,81:1579-1588http://

dx.doi.org/10.1134/S0006297916130010In:http://link.springer.

com/10.1134/S0006297916130010.

37. HinnebuschAG:Thescanningmechanismofeukaryotic translationinitiation.AnnuRevBiochem2014,83:779-812http://

dx.doi.org/10.1146/annurev-biochem-060713-035802http://

www.annualreviews.org/doi/10.1146/annurev-biochem-060713- 035802.

38. RodninaMV,WintermeyerW:Fidelityofaminoacyl-tRNA selectionontheribosome:kineticandstructuralmechanisms.

AnnuRevBiochem2001,70:415-435http://dx.doi.org/10.1146/

annurev.biochem.70.1.415http://www.annualreviews.org/doi/

10.1146/annurev.biochem.70.1.415.

39. DemeshkinaN,JennerL,WesthofE,YusupovM,YusupovaG:A newunderstandingofthedecodingprincipleontheribosome.

Nature2012,484:256-259http://dx.doi.org/10.1038/nature10913 http://www.nature.com/doifinder/10.1038/nature10913.

40. A˚qvistJ,KamerlinSCL:Exceptionallylargeentropy

contributionsenablethehighratesofGTPhydrolysisonthe ribosome.SciRep2015,5:15817http://dx.doi.org/10.1038/

srep15817In:http://www.ncbi.nlm.nih.gov/pubmed/26497916.

41. TrobroS,A˚qvistJ:Mechanismofpeptidebondsynthesisonthe ribosome.ProcNatlAcadSciUSA2005,102:12395-12400