New in silico insights into the inhibition of RNAP II by α-amanitin and the protective effect mediated by effective antidotes

(1)

ContentslistsavailableatScienceDirect

Journal

of

Molecular

Graphics

and

Modelling

jo u r n al ho me p ag e :w w w . e l s e v i e r . c o m / l o c a t e / J M G M

New

in

silico

insights

into

the

inhibition

of

RNAP

II

by

␣-amanitin

and

the

protective

effect

mediated

by

effective

antidotes

Juliana

Garcia

a,∗

_,

_Alexandra

_T.P.

_Carvalho

b

_,

_Daniel

_F.A.R.

_Dourado

b

_,

_Paula

_Baptista

c

_,

Maria

de

Lourdes

Bastos

a

_,

_Félix

_Carvalho

a,∗

a_REQUIMTE_Laboratory_of_Toxicology,_Department_of_Biological_Sciences,_Faculty_of_Pharmacy,_University_of_Porto,_Rua_José_Viterbo_Ferreira_n◦_228,_4050-313_Porto,_Portugal

b_Department_of_Cell_and_Molecular_Biology,_{Computational}_and_Systems_Biology,_Uppsala_University,_Biomedical_Center_Box_596,₇₅₁₂₄_Uppsala,_Sweden

c_CIMO/School_of_Agriculture,_Polytechnic_Institute_of_Braganc¸_a,_Campus_de_Santa_Apolónia,_Apartado_1172,_5301-854_Braganc¸_a,_Portugal

a

r

t

i

c

l

e

i

n

f

o

Articlehistory:

Accepted3May2014

Availableonline14May2014

Keywords: ␣-Amanitin Benzylpenicillin Ceftazidime Silybin RNApolymeraseII Triggerloop Bridgehelix

a

b

s

t

r

a

c

t

Poisonous␣-amanitin-containingmushroomsareresponsibleforthemajorcasesoffatalitiesafter mush-roomingestion.␣-AmanitinisknowntoinhibittheRNApolymeraseII(RNAPII),althoughtheunderlying mechanismsarenotfullyunderstood.Benzylpenicillin,ceftazidimeandsilybinhavebeenthemost fre-quentlyuseddrugsinthemanagementof␣-amanitinpoisoning,mostlybasedonempiricalrationale. ThepresentstudyprovidesaninsilicoinsightintotheinhibitionofRNAPIIby␣-amanitinandalsoon theinteractionoftheantidotesontheactivesiteofthisenzyme.

Dockingandmoleculardynamics(MD)simulationscombinedwithmolecularmechanics-generalized Bornsurfaceareamethod(MM-GBSA)werecarriedouttoinvestigatethebindingof␣-amanitinand threeantidotesbenzylpenicillin,ceftazidimeandsilybintoRNAPII.

Ourresultsrevealthat␣-amanitinshouldaffectsRNAPIItranscriptionbycompromisingtriggerloop (TL)function.Theobserveddirectinteractionsbetween␣-amanitinandTLresiduesLeu1081,Asn1082, Thr1083,His1085andGly1088alterstheelongationprocessandthuscontributetotheinhibitionof RNAPII.Wealsopresentevidencesthat␣-amanitincaninteractdirectlywiththebridgehelixresidues Gly819,Gly820andGlu822,andindirectlywithHis816andPhe815.Thisdestabilizesthebridgehelix, possiblycausingRNAPIIactivityloss.

Wedemonstratethatbenzylpenicillin,ceftazidimeandsilybinareabletobindtothesamesiteas ␣-amanitin,althoughnotreplicatingtheunique␣-amanitinbindingmode.Theyestablishconsiderablyless intermolecularinteractionsandtheonesexistingareessentialconﬁnetothebridgehelixandadjacent residues.Therefore,thetherapeuticeffectoftheseantidotesdoesnotseemtobedirectlyrelatedwith bindingtoRNAPII.

RNAP II ␣-amanitin binding site can be divided into speciﬁc zones with different proper-ties providing a reliable platform for the structure-based drug design of novel antidotes for ␣-amatoxinpoisoning.Anidealdrugcandidateshouldbe acompetitiveRNAPIIbinderthat inter-acts withArg726, Ile756,Ala759,Gln760 and Gln767,but not with TLand bridgehelixresidues. ©2014ElsevierInc.Allrightsreserved.

1. Introduction

ThepoisonousAmanitamushroomspeciesAmanitaphalloides

(DeathCap),Amanitaverna(WhiteDeadlyAmanita)andAmanita

virosa(DestroyingAngel)areresponsiblefor90–95%ofthe

fatal-∗ Correspondingauthors.Tel.:+351220428597;fax:+351226093390.

E-mailaddresses:jugarcia18@hotmail.com,garciaju1987@gmail.com

(J.Garcia),felixdc@ff.up.pt(F.Carvalho).

ities occurring after mushroomingestion [1]. Toxinscontained

in these species include bicyclic octapeptides and consist of

nine deﬁned members: ␣-amanitin, ␤-amanitin, ␥-amanitin,

␧-amanitin,amanin,amaninamide,amanullin,amanullinicacid,and

proamanullin[2].Fromthese,␣-amanitinaccountsforabout40%

oftheamatoxins[2].

Speciﬁc properties characterize these toxins. They are

ther-mostableandarenotremovedbyboilinganddiscardingwateror

byanyformofcookingordrying,andbelongtothemostviolent

http://dx.doi.org/10.1016/j.jmgm.2014.05.002

(2)

J.Garciaetal./JournalofMolecularGraphicsandModelling51(2014)120–127 121

Fig.1. StructureofRNAPII/␣-amanitincomplex.Triggerloop,activesite,DNAand

␣-amanitinarecoloredred,blue,magentaandyellow,respectively.Activesiteis

representedbytheAsp481,Asp483andAsp485catalyticresidues.(For

interpre-tationofthereferencestocolorinthisﬁgurelegend,thereaderisreferredtothe

webversionofthearticle.)

poisons from higher fungi: only one medium-size

amatoxin-containing specimen contains from 10 to 12mg amatoxins, a

potential lethal dose for human adults (lethal dose: LD50 of

0.1–0.5mg/kgbodyweight)[3].

AfteringestionpoisonousAmanitamushrooms,amatoxinsare

readilyabsorbedandaccumulateintheliveruponuptakeviaan

organicanion-transportingoctapeptide(OATP)locatedinthe

sinu-soidalmembraneofhepatocytes[4],whichisthesametransport

systemforbiliaryacids.Themainmolecularmechanismof

toxic-ityistheirstronginhibitionofRNAPII[5]whichisresponsiblefor

thetranscriptionoftheprecursorofmessengerRNA.Thisinturn

causesinhibitionofDNAandproteinsynthesisprocessesandleads

tocelldeath[6].

RNAPIIconsistsofa 10-subunit coreenzyme and a

periph-eralheterodimerofsubunitsRpb4andRpb7(Rpb4/7subcomplex).

The core enzyme comprises subunits Rpb1, Rpb2, Rpb3, Rpb5,

Rpb6,Rpb8, Rpb9,Rpb10,Rpb11andRpb12[7].The activesite

islocatedin theinterfacebetweenthecoresubunitsRpb1and

Rpb2.Thecatalyticprocessinvolvesthenucleotideadditioncycle

(NAC)andbeginswithbindingofanucleosidetriphosphate(NTP)

substratetotheelongationcomplex(EC).Catalyticadditionofthe

nucleotidetothegrowing3-endoftheRNAleadstoformation

ofapyrophosphateion.Pyrophosphatereleaseleadstothe

pre-translocation state,inwhich thenewly added,3-terminal RNA

nucleotideremainsinthesubstratesite.Finally,translocationof

DNAandRNAresultsinthepost-translocationstate,inwhichthe

substratesiteisfreeforbindingthenextNTP,andtheNACcanbe

repeated[8].Thisprocessinvolveastructuralelementnamed

trig-gerloop(TL),whichmakesbothdirectandindirectcontactwithall

featuresofthenucleotideinthepolymeraseactivecenterandthe

bridgehelixwhichguidesthetemplateDNAstrandintheactive

siteandpositionstheDNA-RNAhybridrelativetotheactivesite

[9].TLhasbeenhypothesizedtohaveseveralconformations,but

twoofthem,the“open”(Fig.1)and“closed”conformationswere

observedinX-raystructures[9,10].IntheopenTLstructurethe

substrateentersontheenzyme,andthentheTLrotates(seearrow

inFig.1)totheactivesite(closedconformation)[9].Theﬂexibility

ofthetriggerlooponthecatalysisisdirectlyinﬂuencedbythe

con-formationofthebridgehelix[11].Recentlyacrystalstructureof

␣-amanitinwithyeastRNAPIIwaspublishedrevealingseveralkey

molecularinteractionsthatmaycontributetoinhibitionofRNAPII

[10].Multiplerelevantinteractionsbetween␣-amanitinwithRNAP

IIarelocatedinthebridgehelix(previouslydeﬁned“cleft”region)

andinadjacentpartofRpb1(previouslydeﬁnedfunnelregion).

Basedonthis␣-amanitinmayblocktranslocationbyinteracting

withbridgehelixandpreventingtheconformationalchangeofthe

TLandconsequentlytranscriptionalelongation(Fig.1)[12].

No speciﬁc antidote for intoxications with amatoxins is

available. Based onpre-clinical ﬁndings several antidoteshave

beenusedin amatoxinpoisonings,includinghormones(insulin,

growthhormone,glucagon),steroids,vitaminC,vitaminE,

cime-tidine, thioctic acid, antibiotics (benzylpenicillin, ceftazidime),

N-acetylcysteine,and silybin. Fromthese,only benzylpenicillin,

ceftazidime,N-acetylcysteineandsilybinbeenreportedsome

suc-cessinthepharmacologicmanagement ofamatoxinpoisonings,

though withvaryingextent[1].Furthermore,theprecise

phar-macologicalmechanism ofactionfor thesedrugs remainstobe

elucidated.

Inthecurrentstudywereportthemodeofinteractionof

␣-amanitinand three antidotes(benzylpenicillin, ceftazidimeand

silybin) with RNAP II, using docking methods and molecular

dynamics simulations. To secure signiﬁcant sampling, we

per-formed3moleculardynamic simulations(ina totalof34ns)in

explicitwater of each one of theRNAP II/␣-amanitin/antidotes

complexes.Inordertoprovideanewinsightintotheinhibition

mechanismofRNAPIIby␣-amanitin,andthepossibletherapeutic

mechanismofactionofbenzylpenicillin,ceftazidimeandsilybinin

amatoxinpoisoning,weusedbindingenergydecompositionbased

onthemolecularmechanicsgeneralizedBornsurfacearea

(MM-GBSA)approach.

2. Materialsandmethods

2.1. Moleculardocking

MolecularDockingof severalcompoundsishelpfulin

eluci-datingkeyfeaturesofligand–receptorinteractions.Thismethod

allowsustoexploretheinteractionoftheantidoteswithRNAPII

andpredicttheirpreferredorientation,whichwouldforma

com-plexwithoverallminimumenergy.ThecrystalstructureofRNAP

IIcomplexedwith_␣-amanitin(ProteinDataBankentry3CQZand

2VUM)wasusedtoobtainthestartingstructuresforthemolecular

docking[8,11]andonlysubunitsRpb1andRpb2wereused.

TheoptimizedRpb1andRpb2subunitsweredockedwith

␣-amanitin, benzylpenicillin,silybin andceftazidime. Thedocking

procedurewasmadewiththeAutoDock4program[13,14].This

automateddockingprogramusesagridbasedmethodforenergy

calculationoftheﬂexibleligandincomplexwitharigidprotein.

Theprogramperformsseveralrunsineachdockingexperiment.

Eachrunprovidesonepredictedbindingmode.TheLamarckian

geneticalgorithm(LGA)wasusedinalldockingcalculations.The

48×44×44gridpointsalongthex,yandzaxeswascenteredonthe

␣-amanitinwhichwaslargeenoughtocoverallpossiblerotations

ofthetoxinandantidotes.Thedistancebetweentwoconnecting

gridpointswas0.375 ´˚A.Thedockingprocesswasperformedin250

LGAruns.Thepopulationwas150,themaximumnumberof

gener-ationswas27,000andthemaximumnumberofenergyevaluations

was2.5_×106.AftercompleteexecutionofAutoDockten

confor-mations ofligandin complexwiththereceptor wereobtained,

whichwereﬁnallyrankedonthebasisofbindingenergy[15].The

resultingconformationswerevisualizedintheVisualMolecular

Dynamics[16].

Afteranalysisofall thesolutionsobtained, thebestdocking

solutionswerechosen asstartingstructuresfor thesubsequent

(3)

Fig.2.Chemicalstructures:(a)␣-amanitin;(b)benzylpenicillin;(c)ceftazidime;and(d)silybin.

2.2. Optimizationofantidotesand˛-amanitin

Thestructuresof␣-amanitin,benzylpenicillin,ceftazidimeand

silybin (Fig. 2) wereconstructed and optimized in Gaussian at

theHF/6-31G*leveloftheory.For eachantidote,weperformed

twooptimizations:oneinvacuumandanotherinthecondensed

phase.ThepartialchargeswerecalculatedresortingtotheRESP

method.

2.3. Moleculardynamicsimulations

TheenzymewasﬁrstneutralizedbyaddingNa+_ions_and

sol-vatedinacubicboxofTIP3watermolecules,suchthattherewere

atleast10.0 ´˚A ofwaterbetweenthesurfaceoftheproteinandthe

edgeofthesimulationbox.Theinitialgeometryoptimizationof

theenzymewasminimizedintwostages.Intheﬁrststageonlythe

hydrogenandwateratomswereminimized;inthesecondstage

theentiresystemwasminimized.

TheparametersofthechosenmodelswerevalidatedwithMD

simulations in explicit solvent.The MDs were performedwith

ff99SBforceﬁeldandthegeneralizedamberforceﬁeld(Gaff)[17].

Aninitialminimizationwasperformedfollowedbyan

equilibra-tionof500ps.TheequilibrationwasperformedinaNVTensemble

using Langevin dynamics with small restraints on the protein

(100kcalmo1−1).Foreachofthefoursystemsaninitial

produc-tionsimulationof10nswasperformedfollowedbytworandom

initialvelocitiesreplicaruns,totaling34nspersubstrate.This

rep-resentedasubstantialcomputationaleffort,sinceeachsystemis

composedby≈44,000atoms containingprotein,DNA andRNA.

Temperaturewasmaintainedat300KintheNPTensembleusing

Langevindynamicswithacollisionfrequencyof1.0ps−1.Thetime

stepwassetto2fs.Thetrajectoriesweresavedevery500steps

foranalysis.Constantpressureperiodicboundarywasusedwith

anaveragepressureof1atm.Isotropicpositionscalingwasused

tomaintainthepressurewitha relaxationtime of2ps.SHAKE

constraintswereappliedtoallbondsinvolvinghydrogen.The

par-ticlemeshEwald(PME)methodwasusedtocalculateelectrostatic

interactionswithacutoffdistanceof8.0 ˚A.

2.4. Calculationofthebindingenergy

MM-GBSAwasappliedtocomputethebindingenergybetween

theproteinandeachligandandtodecomposetheinteraction

ener-giesonaperresidue basisbyconsideringmolecularmechanics

energiesandsolvationenergies[18].

Theenergydecompositionwasperformedforgas-phase

ener-gies,desolvation freeenergiescalculated byGBmodel[19] and

nonpolarcontributionstodesolvationusingthelinear

combina-tionsofpairwiseoverlaps(LCPO)method[20].

Conformationalentropywasnotconsideredbecauseouraim

wastoidentifyimportant interactionsbetween the␣-amanitin

andtheantidoteswithRNAPIIresidues,ratherthantoobtainvery

accurateabsolutevaluesforthebindingfreeenergy.

3. Resultsanddiscussion

The determination of the crystal structures with bound

␣-amanitinitshowedthatthistoxinbindingsitewasquitefarfrom

theRNAPIIactivesite.Inhibitioncouldonlybeexplainedif

␣-amanitinbindingcouldleadtosomeconformationalchangethat

wouldaffecttheactivesite.Themysterywaspartiallysolvedwith

thedeterminationofthecomplexRNAPII/DNA.RNA/substrate[9].

InthiscomplextheTLisintheoppositedirectionandinteracting

withthesubstrate.TheTListhusahighlyﬂexiblestructuralmotif

withintheenzymecapableoflargeconformationalchangesateach

catalyticcycle.Superpositionofthepre-catalyticcomplexwiththe

complexwith␣-amanitinshowsthatthelargesizeofthetoxin

couldpreventthemovementoftheTL(Fig.3).

Consequently, a descriptionof the full TL movement is still

missing.Many TLconformations arestill notdescribed andthe

crystalstructuresonlygiveinitialandﬁnalsnapshotsforthis

move-ment.Togaindetailedinsightaboutthemostimportantresidues

forthe␣-amanitin/antidotesdynamicalinteractionwithRNAPII

weperformedandanalyzedMDsimulationsondockedcomplexes

ofRNAPII/_␣-amanitinandwiththeantidotes.Moreover,we

per-formeddetailedhydrogenbondsanalyses(Table1),measuredthe

(4)

Fig.3.Interactionof␣-amanitinwithRNAPII,demonstratedinsilico.Superposition

ofthelowestRMSDfortheaveragestructureofthesimulation(gray),pre-catalytic

complex(magenta,pdbcode2E2H),complexwith␣-amanitin.The␣-amanitinisin

licoricerepresentationandthesubstrateisinyellowlicoricerepresentation).(For

interpretationofthereferencestocolorinthisﬁgurelegend,thereaderisreferred

tothewebversionofthearticle.)

toxin/antidotes.Finally,weperformedenergydecomposition.The

proteinperresidueenergyvaluesgivesusafastandrelatively

reli-ableestimationofthecontributionofeachresiduetothebinding

[21].Ourgoalwasprovideanewinsightintotheinhibition

mech-anismofRNAPIIby␣-amanitinbyidentifyingthecriticalresidues

forRNAPIIbindingandsubsequentlyunderstandinghowantidotes

interactwithRNAPIIand iftheycanbindtothesameposition

withoutinhibitingtheenzyme.

3.1. Identiﬁcationofcriticalresiduesfor˛-amanitinbinding

Superpositionoftheaveragesimulationstructureandthe

crys-talstructureisshownin Fig.4.Themostnoticeabledifferences

betweenthecrystalstructure andoursimulation structurerely

inthedisplacementofTL(Fig.4),whichmaybeattributedtothe

highﬂexibilityandintrinsicmobilityoftheTL.Intheour

simula-tionstructure,Gly1088residueinteractswith␣-amanitinoxygen

(O33),whileinthecrystalstructurethisresidueisfarawayfrom

the␣-amanitin.Thedemonstratedaccuracyofthedocked

com-plexvalidatesourprotocolandsupportstheresultsobtainedfor

theantidotes,forwhichthereisnocrystalstructure(describedin

Section3.2).

Inordertomoreeasilyand accuratelygrasptheinteractions

between the protein and ␣-amanitin we performed an energy

decompositionanalysisofthesimulations.Weresortedtothe

MM-GBSAmethod.Thecalculatedbindingenergiesofeachcomplexcan

beseeninTable2.Evennotconsideringtheentropy,computational

studies using MM-GBSA calculations ondifferent complexes of

protein-inhibitorsshowedgoodcorrelationswithrespectto

exper-imentaldata[21].Individualenergydecompositionofallresidues

inthecomplexwasalsocalculatedinordertoqualitativelyﬁnd

thekeyresiduesthatplayamoreimportantroleinthe␣-amanitin

binding(Figs.5aand6a).Valuesareexpressedasmean±SDofthe

3replicates.Fig.5adepictstherelativepositionoftheinhibitor

andimportantresiduesinthebindingcomplexbyusingthe

low-estroot-mean-squaredeviationRMSDstructureinrespecttothe

averageof the simulation. Thetoxin ␣-amanitin interacts with

residuesArg726,Ile756,Ala759,Gln760,Gln767,Gln768,Ser769,

Gly819,Gly820,Glu822,Leu1081,Asn1082,Thr1083,His1085and

Gly1088(Figs.5aand6a).TheguanidiniumgroupofArg726forms

cation–␲interactionswith␣-amanitinphenylgroup,which

corre-spondstoenergyof−1.71±1.51kcalmo1−1.Thebindingenergyof

residueIle756is−2.53±0.83kcalmol−1,agreeingwiththeCH–␲

interaction of Ile756 alkyl group with␣-amanitin phenylring.

At the same time, Ala759 alkyl group also forms CH–␲

inter-actionswith␣-amanitinphenylgroup(−0.67±0.04kcalmol−1).

Theside-chainnitrogenatomofGln760formsahydrogenbond

with _␣-amanitin O4 (Table 1), leading to a favorable binding

energyof −1.40±0.89kcalmol−1.Thustheindoleportionof

␣-amanitininsertsin the hydrophobicpocketcreatedby Arg726,

Ile756, Ala759 and Gln760 (Fig. 5a). The side-chain oxygen of

Gln767 forms a hydrogen bond with ␣-amanitin N30, which

Table1

Hydrogenbondsformedbetweenthetoxin/antidotesandRNApolymeraseIIa_.

Toxin/Antidote Donor AcceptorH Acceptor Distanceb_{( ´˚A)} _Angleb₍◦₎ _%_per_time_frame

Gln767:OE1 ␣-amanitin:H77 ␣-amanitin:N30 1.81 166.53 60

␣-amanitin:O4 Gln768:HE2 Gln768:NE2 2.87 157.06 80

␣-amanitin ␣-amanitin:O4 Gln760:HE2 Gln760:NE2 2.11 148.55 26

␣-amanitin:O59 Ser769:H Ser769:N 2.04 157.88 88

Glu822:OE1 ␣-amanitin:1H11 ␣-amanitin:O63 2.55 172.43 60

Benzylpenicillin Benzilpenicillin:O3 Hie816:HE2 Hie816:NE2 2.51 134.78 48

Benzilpenicillin:O1 Gln760:1HE2 Gln760:NE2 1.96 167.45 56

Silybin Silybin:O56 Gln760:1HE2 Gln760:NE2 2.17 142.68 75

Silybin:O4 Gly823:H Gly823:N 2.08 150.30 47

a_Hydrogen_bonds_were_analyzed_in_the_average_structures_from_MD_simulation.

b_The_geometric_criterion_for_the_formation_of_H-bonds_is_common_with_an_{acceptor-donor}_distance_less_than_{3.5 ´˚A and}_the_{donor-H-acceptor}_angle_larger_than_120.

Table2

Bindingenergycalculationbetweenthethreeantidotesand␣-amanitinwithRpb1andRpb2subunits(allenergiesareinkcalmol−1_).

Complex Gele Gvdw GInt GGas GGBSUR GGB GGBsol GGBele Gtot

Rpb1Rpb2/␣-amanitin −60.97 −60.91 0.00 −121.88 −8.16 108.70 100.53 47.73 −21.34 Rpb1Rpb2/benzylpenicillin −19.78 −31.34 0.00 −51.12 −4.27 41.53 37.26 21.75 −13.87 Rpb1Rpb2/silybin −21.00 −42.58 0.00 −63.62 −5.87 47.96 42.09 26.95 −21.53 Rpb1Rpb2/ceftazidime −59.96 −38.20 0.00 −98.16 −5.31 89.94 84.63 29.98 −13.53 Gele:electrostaticenergy.Gvdw:vanderwaalsenergy;Gint:internalenergy;GGas:totalgasphaseenergy(sumGele,Gvdw,Gint);GGBSUR:nonpolarcontributionto

solvation;GGB:theelectrostaticcontributiontothesolvationfreeenergy;GGBSOL:sumofnonpolarandpolarcontributionstosolvation;GGBELE:sumoftheelectrostatic

(5)

Fig.4.SuperpositionofthelowestRMSDfortheaveragestructureofRNAPII(cyan)with␣-amanitin(green),crystalstructureofRNAPII(magenta,pdbcode3CQZ)in complexwith␣-amanitin(yellow).

Fig.5.Geometriesofkeyresidues,whichproducesomefavorableinteractionswithRNAPII,areplottedinthecomplexesaccordingtotheaveragestructurefromthe MDtrajectory.(a)RNAPII/␣-amanitincomplex;(b)RNAPII/benzylpenicillincomplex;(c)RNAPII/ceftazidimecomplex;and(d)RNAPII/silybincomplex.Thedashedlines representhydrogenbondsbetweenthe␣-amanitin/antidotesandRNAPII.

(6)

Fig.6. ␣-Amanitin/antidotes-residuesinteractionspectrumof(a)RNAPII/␣-amanitincomplex;(b)RNAPII/benzylpenicillincomplex;(c)RNAPII/ceftazidimecomplexand (d)RNAPII/silybincomplex.Theresidueswithinteractionenergythan−0.5kcalmol−1arelabeled.Valuesareexpressedasmean±s.d.ofthe3replicates.

correspondstointeractionenergyof−2.70±1.60kcalmol−1.The nitrogen atomof Gln768 alsoforms a hydrogen bond with ␣-amanitinO4(Table1),whichisthestrongestinteractionamongall

residues(−2.82±0.88kcalmol−1).The␣-amanitin/Ser769

interac-tionenergyis−2.18±0.80kcalmol−1,whichisduetoahydrogen

bondbetweenSer769nitrogenatomand␣-amanitinO59(Table1).

Hydrophobicinteractionsmaybethemainforcebetweenbridge

helixresiduesGly819,Gly820and␣-amanitin,whichcorrespondto

energiesof−0.61±0.08and−0.53±0.00kcalmol−1,respectively.

Theinteraction energyof thebridgehelix residue Glu822with

␣-amanitinis−1.90±0.97kcalmol−1,correspondingtoa

hydro-genbondbetweentheGlu822sidechainoxygenand␣-amanitin

hydroxyprolineOH (Table1).The alkyls groupof Leu1081 and

Asn1082interactwith␣-amanitinalkylsbyhydrophobic

interac-tions,whichcorrespondtoenergiesof−1.14and−2.75kcalmol−1,

respectively.TheinteractionenergyofThr1083with_␣-amanitinis

−2.01kcalmol−1,mostlyattributedtodipole–dipoleinteractions.

TLresiduesHis1085andGly1088formdipole–dipoleinteractions

with␣-amanitinhydroxyprolineandproduceinteractionenergies

of−0.53±0.12and−0.62kcalmol−1,respectively,whichis

sup-portingbyexperimentaldata[22].

According to Figs. 5a and 6a and the above analysis, three

valuable ﬁndings can beobserved: (1) Hydrogen bonds, CH–␲

and hydrophobic interactions drivethe bindings of ␣-amanitin

toRNAPII.(2)Ourresultsindicateinteractionswithbridgehelix

residues Gly819, Gly820 and Glu822. Moreover, indirect

con-tacts tothebridgehelix werealso observed.␣-Amanitin binds

to residue Gln768, which in turn binds to His816 and Phe815

andforms a hydrogenbond withSer769.Ser769interacts with

thebridgehelixresidueGly819.(3)␣-AmanitininteractswithTL

residuesLeu1081,Asn1082,Thr1083,His1085andGly1088.These

interactionscanpreventTLmovementandhencecontributingto

inhibittranscription.

3.2. BindingmodepredictionsofantidotestoRNAPII

Inordertoprovideanewinsightintothemechanismofaction

ofbenzylpenicillin,ceftazidimeandsilybininamatoxinpoisoning

weanalyzedtheantidotesbindingtoRNAPII.

According to Figs. 5b and 6b, several residues are involved

in theRNAP II/benzylpenicillin binding. The binding energy of

benzylpenicillin to Ile756 is −1.77±0.49kcalmol−1, agreeing

withCH–␲interactions betweenIle alkyls and benzylpenicillin

phenyl ring (Fig. 5b). Gln760 nitrogen atom binds to

ben-zylpenicillinO1byahydrogenbond,correspondingtoenergyof

−1.63±0.45kcalmol−1 (Table1).Atsametime, otherhydrogen

bond between His816 nitrogen and benzylpenicillin O3

con-tributeswith−0.69±0.14kcalmol−1(Table1).Val765alkylgroup

andbenzylpenicillinalkylgroupgeneratedispersiveinteractions

(−1.03±0.14kcalmol−1).Dipole–dipoleinteractionsareobserved

between Gly820 and benzylpenicillin (−1.54±0.00kcalmol−1).

Leu824alkylsgroupsinteractwithbenzylpenicillinalkylsgroups

by hydrophobic interactions (−0.79±0.23kcalmol−1). Finally,

Gln2218 and Pro2220 form dipole–dipole interactions with

benzylpenicillin carboxyl group, corresponding to energies of

−1.65±0.45 and −0.79±0.20kcalmol−1,respectively. Based on

the above analysis, two valuable ﬁndings can be described:

(1) Residues Ile756, Gln760and Gly820 are common sites for

(7)

benzylpenicillin interacts with bridge helix residues His816,

Gly820andLeu824.

In the case of theRNAP II/ceftazidime complex the

follow-ing residues contribute for binding energy (Figs. 5c and 6c).

Ile756alkylgroup and ceftazidime dihydrothiazine ring

gener-ateCH–␲interactions (Fig.5c), correspondingtoaninteraction

energyof−4.28±1.70kcalmol−1.TheinteractionenergyofAsn757

is−0.80±0.10kcalmol−1,whichcanbeattributedtodipole–dipole

interactionswithceftazidimecarboxylicacidgroup.Dipole–dipole

interactions between Val765 and ceftazidime propylcarboxy

moiety are responsible for energy of −1.06±0.51kcalmol−1.

At same time, identical interactions can be seen between

His816 imidazole ring and ceftazidime propylcarboxy moiety

(−1.04±0.18kcalmol−1). Hydrophobicand dipole–dipole

inter-actions are the main forces between Gln2218 and ceftazidime

propylcarboxy moiety (−1.30±1.11kcalmol−1). At same time

Ser2219formsdipole–dipoleinteractionswiththesamemoiety,

whichcorrespondtointeractionenergyof−1.09±0.38kcalmol−1.

Basedontheaboveanalysis,two importantconclusionscanbe

obtained:(1)ResidueIle756andiscommonsite for␣-amanitin

andceftazidime.(2)Ourresultsindicatethatceftazidimeinteract

withbridgehelixresidueHis816.

Finally,forRNAPII/silybinbindingthekeyresiduesareseven

(Figs. 5d and 6d). The binding energy of silybin to Ile756 is

−1.58±0.27kcalmol−1, corresponding to a CH–␲ interaction

between Ile756 alkyl and silybin phenyl group. The

interac-tion energy of Asn757 with silybin phenyl ring has energy of

−1.06±0.81kcalmol−1, which mostly results of NH–␲

interac-tions. Gln760side-chainnitrogen forms a hydrogenbond with

silybin O56 (−1.50±0.79kcalmol−1) (Table 1). Dipole–dipole

interactions between residues Ser769, Val770, Gly819, Gly820

and Arg821 with silybin result in the energies contributions

of −1.25±0.14, −1.01±0.10, 0.95±0.29, −1.40±1.17 and

−0.95±0.50kcalmol−1,respectively.Silybinphenylringcontacts

withGlu822alkylgrouptogenerateahydrophobicCH–_␲

inter-action (−1.38±0.60kcalmol−1). The Gly823 forms a hydrogen

bondwithsilybinO4(−1.49±0.89kcalmol−1)(Table1).The␲–␲

contactsanddipole–dipoleinteractionsbetweenresidueHis1085

and silybin diphenol ring result in an energy contribution of

−1.50±0.07kcalmol−1.Fromtheseresults,wecanconclude:(1)

ResiduesIle756,Gln760,Gly819,Gly820,Glu822andHis1085are

commonsitesfor␣-amanitinandsilybin.(2)Ourresultsindicate

thatsilybininteractwithbridgehelixresiduesGly819,Gly820and

Glu822andTLresidueHis1085.

4. Conclusions

Inthisstudy,dockingand MDsimulationcoupledwith

MM-GBSAenergydecompositionhavebeencarriedouttoclarifythe

inhibitionmechanismofRNAPIIby␣-amanitinandtoprovidea

newinsightintotheplausiblemechanismofactionofthree

anti-dotesusedinamatoxinpoisoning.

WehypothesizethatTLresiduesLeu1081,Asn1082,Thr1083,

His1085andGly1088,andbridgehelixresiduesGly819,Gly820

andGlu822contributeforthehighbindingafﬁnityof␣-amanitin

withRNAP II. Our data clearly reinforces thehypothesis of an

importantroleofthebridgehelix[23]andTLintheelongation

process and are consistent withthe existence of a network of

functionalinteractionsbetweenthebridgehelixandTLthat

con-trolfundamentalparametersofRNAsynthesis.Ourdatasuggest

that␣-amanitininterfereswithbridgehelixmovementduringthe

translocationandwiththemovementoftheTL,whichclosesover

theactivesiteduringthenucleotideincorporation(Fig.1).

More-over,accordingtoKaplanetal.(2008),weshowthattheinteraction

of␣-amanitinwithHis1085contributesfortheinhibitionofthe

enzyme.Thisissupportedbytheﬁndingsthatasubstitutionof

ala-nineorphenylalanineatposition1085speciﬁcallyrendersRNAPII

highlyresistantto␣-amanitin[22].Also,thereareevidencesthat

residue1085mayplayacriticalroleinthecatalyticmechanism

[24].Taken togetherourresultsareingoodagreementwithall

literaturedata.

IntheanalysisoftheantidoteswefocusedontheTL,bridge

helixandotheradditionalresiduesthatareinvolvedin␣-amanitin

binding.It seems thatbenzylpenicillin, ceftazidimeand silybin,

althoughbindingtothesameRNAPIIbindingsite,cannotreplicate

␣-amanitinbindingmode.Theantidotesestablishconsiderablyless

intermolecularinteractionsandtheonesexistingareessential

con-ﬁnetothebridgehelixandadjacentresidues.Thetherapeuticeffect

ofthestudiedantidoteson␣-amatoxinpoisoningseemsnottobe

directlyrelatedwithbindingtoRNAPII.

ThesestructuralinsightsaboutthemolecularaspectsofRNAP

IIinhibitioncanprovideareliableplatformforthestructure-based

drugdesignagainst␣-amatoxinpoisoning.Wesuggestthatanideal

drugshouldbeacompetitiveRNAPIIbinderabletostrongly

inter-actwithArg726,Ile756,Ala759,Gln760andGln767,butnotwith

bridgehelixandTLresidues.

Acknowledgements

TheauthorsgratefullyacknowledgetheFoundationforthe

Sci-enceandTechnology(FCT,Portugal)forﬁnancialsupportandalso

thankFCTforPhDgrantSFRH/BD/74979/2010.Weacknowledge

QtrexclusterandSNIC-UPPMAXforCPUtimeallocation.

References

[1]F.Enjalbert,S.Rapior,J.Nouguier-Soule,S.Guillon,N.Amouroux,C.Cabot, Treatmentofamatoxinpoisoning:20-yearretrospectiveanalysis,J.Toxicol. Clin.Toxicol.40(2002)715–757.

[2]J.Vetter,Toxinsofamanitaphalloides,Toxicon36(1998)13–24.

[3]T.Wieland,ProgressinTryptophanandSerotoninResearch,WalterdeGruyter &Co,Berlin,NewYork,1984.

[4]K.Letschert,H.Faulstich,D.Keller,D.Keppler,Molecularcharacterizationand inhibitionofamanitinuptakeintohumanhepatocytes,Toxicol.Sci.91(2006) 140–149.

[5]T.J.Lindell,F.Weinberg,P.W.Morris,R.G.Roeder,W.J.Rutter,Speciﬁcinhibition ofnuclearRNAPIIbyalpha-amanitin,Science170(1970)447–449.

[6]T.Wieland,Thetoxicpeptidesfromamanitamushrooms,Int.J.Pept.Protein Res.22(1983)257–276.

[7]P.Cramer,K.J.Armache,S.Baumli,S.Benkert,F.Brueckner,C.Buchen,etal., Structure ofeukaryoticRNApolymerases,Annu. Rev.Biophys.37(2008) 337–352.

[8]F.Brueckner,J.Ortiz,P.Cramer,AmovieoftheRNApolymerasenucleotide additioncycle,Curr.Opin.Struct.Biol.19(2009)294–299.

[9]D.Wang,D.A.Bushnell,K.D.Westover,C.D.Kaplan,R.D.Kornberg,Structural basisoftranscription:roleofthetriggerloopinsubstratespeciﬁcityand catal-ysis,Cell127(2006)941–954.

[10]D.A.Bushnell,P.Cramer,R.D.Kornberg,Structuralbasisoftranscription: alpha-amanitin-RNAPIIcocrystalat2.8Aresolution,Proc.Natl.Acad.Sci.U.S.A.99 (2002)1218–1222.

[11]L.Tan,S.Wiesler,D.Trzaska,H.Carney,R.Weinzierl,Bridgehelixand trig-gerloopperturbationsgeneratesuperactiveRNApolymerases,J.Biol.7(2008) 1–15.

[12]X.Q.Gong,Y.A.Nedialkov,Z.F.Burton,Alpha-amanitinblockstranslocationby humanRNAPII,J.Biol.Chem.279(2004)27422–27427.

[13]G.M.Morris,D.S.Goodsell,R.S.Halliday,R.Huey,W.E.Hart,R.K.Belew,etal., AutomateddockingusingaLamarckiangeneticalgorithmandanempirical bindingfreeenergyfunction,J.Comput.Chem.19(1998)1639–1662.

[14]G.M.Morris,R.Huey,W.Lindstrom,M.F.Sanner,R.K.Belew,D.S.Goodsell,etal., AutoDock4andAutoDockTools4:automateddockingwithselectivereceptor ﬂexibility,J.Comput.Chem.30(2009)2785–2791.

[15]R.Huey,G.M.Morris,A.J.Olson,D.S.Goodsell,Asemiempiricalfreeenergyforce ﬁeldwithcharge-baseddesolvation,J.Comput.Chem.28(2007)1145–1152.

[16]W.Humphrey,A.Dalke,K.Schulten,VMD:visualmoleculardynamics,J.Mol. Graph.14(1996)33–38.

[17]D.A. Case,T.E.Cheatham,T.Darden,H.Gohlke,R.Luo,K.M. Merz,etal., TheAmberbiomolecularsimulationprograms,J.Comput.Chem.26(2005) 1668–1688.

[18]P.A.Kollman,I.Massova,C.Reyes,B.Kuhn,S.Huo,L.Chong,etal., Calculat-ingstructuresandfreeenergiesofcomplexmolecules:combiningmolecular mechanicsandcontinuummodels,Acc.Chem.Res.33(2000)889–897.

(8)

J.Garciaetal./JournalofMolecularGraphicsandModelling51(2014)120–127 127 [19]A.Onufriev,D.Bashford,D.A.Case,Modiﬁcationofthegeneralizedbornmodel

suitableformacromolecules,J.Phys.Chem.B104(2000)3712–3720.

[20]J. Weiser,P.S.Shenkin, W.C.Still,Approximate atomicsurfaces from lin-ear combinations of pairwise overlaps (LCPO), J. Mol. Biol. 20 (1999) 217–230.

[21]P.D.Lyne,M.L.Lamb,J.C.Saeh,Accuratepredictionoftherelativepotenciesof membersofaseriesofkinaseinhibitorsusingmoleculardockingandMM-GBSA scoring,J.Med.Chem.49(2006)4805–4808.

[22]C.D.Kaplan,K.M.Larsson,R.D.Kornberg,TheRNAPIItriggerloopfunctions insubstrateselectionandisdirectlytargetedbyalpha-amanitin,Mol.Cell30 (2008)547–556.

[23]Y.A.Nedialkov,K.Opron,F.Assaf,I.Artsimovitch,M.L.Kireeva,M.Kashlev, etal.,TheRNApolymerasebridgehelixYFImotifincatalysis,ﬁdelityand translocation,BBA–GeneRegul.Mech.1829(2013)187–198.

[24]A.T.P.Carvalho,P.A.Fernandes,M.J.Ramos,ThecatalyticmechanismofRNAP II,J.Chem.TheoryComput.7(2011)1177–1188.