• Aucun résultat trouvé

Evaluation of Reduced Point Charge Models of Proteins through Molecular Dynamics Simulations: Application to the Vps27 UIM-1 – Ubiquitin Complex

N/A
N/A
Protected

Academic year: 2021

Partager "Evaluation of Reduced Point Charge Models of Proteins through Molecular Dynamics Simulations: Application to the Vps27 UIM-1 – Ubiquitin Complex"

Copied!
22
0
0

Texte intégral

(1)

RESEARCH OUTPUTS / RÉSULTATS DE RECHERCHE

Author(s) - Auteur(s) :

Publication date - Date de publication :

Permanent link - Permalien :

Rights / License - Licence de droit d’auteur :

Institutional Repository - Research Portal

Dépôt Institutionnel - Portail de la Recherche

researchportal.unamur.be

University of Namur

Evaluation of Reduced Point Charge Models of Proteins through Molecular Dynamics

Simulations: Application to the Vps27 UIM-1 – Ubiquitin Complex

Leherte, Laurence; Vercauteren, Daniel

Published in:

Journal of Molecular Graphics and Modelling

Publication date:

2014

Document Version

Peer reviewed version

Link to publication

Citation for pulished version (HARVARD):

Leherte, L & Vercauteren, D 2014, 'Evaluation of Reduced Point Charge Models of Proteins through Molecular

Dynamics Simulations: Application to the Vps27 UIM-1 – Ubiquitin Complex', Journal of Molecular Graphics and

Modelling, vol. 47, pp. 44-61.

General rights

Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights. • Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain

• You may freely distribute the URL identifying the publication in the public portal ?

Take down policy

If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.

(2)

Journal: JMG

Please e-mail or fax your responses and any corrections to:

E-mail:

corrections.esch@elsevier.thomsondigital.com

Article Number: 6354

Fax: +353 6170 9272

Dear Author,

Please check your proof carefully and mark all corrections at the appropriate place in the proof (e.g., by using on-screen

annotation in the PDF file) or compile them in a separate list. Note: if you opt to annotate the file with software other than

Adobe Reader then please also highlight the appropriate place in the PDF file. To ensure fast publication of your paper please

return your corrections within 48 hours.

For correction or revision of any artwork, please consult

http://www.elsevier.com/artworkinstructions

.

Any queries or remarks that have arisen during the processing of your manuscript are listed below and highlighted by flags in

the proof. Click on the ‘

Q

’ link to go to the location in the proof.

Location in

Query / Remark:

click on the Q link to go

article

Please insert your reply or correction at the corresponding line in the proof

Q1

Please confirm that given names and surnames have been identified correctly.

Q2

Figs. 2, 5 to 11 will appear in black and white in print and in color on the web. Based on this, the

respective figure captions have been updated. Please check, and correct if necessary.

Please check this box or indicate your approval if

you have no corrections to make to the PDF file

(3)

JournalofMolecularGraphicsandModellingxxx(2013)xxx–xxx

ContentslistsavailableatScienceDirect

Journal

of

Molecular

Graphics

and

Modelling

jo u r n al h om ep a g 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

Graphical

Abstract

JournalofMolecularGraphicsandModellingxxx (2013)xxx–xxx

Evaluationofreducedpointchargemodelsofproteinsthrough MolecularDynamicssimulations:ApplicationtotheVps27 UIM-1–Ubiquitincomplex

(4)

ContentslistsavailableatScienceDirect

Journal

of

Molecular

Graphics

and

Modelling

jo u r n al h om ep a g 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

Highlights

JournalofMolecularGraphicsandModellingxxx (2013)xxx–xxx

Evaluationofreducedpointchargemodelsofproteinsthrough

MolecularDynamicssimulations:ApplicationtotheVps27UIM-1–Ubiquitincomplex

LaurenceLeherte∗,DanielP.Vercauteren

•ReducedpointchargemodelsprovidestableMolecularDynamicstrajectoriesoftheproteincomplex.

•H-bondnetworksareprogressivelymodifiedasthereductiondegreeincreases.

•Themodelsallowtoprobelocalpotentialhyper-surfaceminimathataresimilartoall-atomones.

•Themodelsallowtosampleproteinconformationsmorerapidlythantheall-atomcaseduetoaloweringofenergybarriers.

(5)

JournalofMolecularGraphicsandModellingxxx(2013)xxx–xxx

ContentslistsavailableatScienceDirect

Journal

of

Molecular

Graphics

and

Modelling

jo u r n al hom epa g 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

Evaluation

of

reduced

point

charge

models

of

proteins

through

Molecular

Dynamics

simulations:

Application

to

the

Vps27

UIM-1–Ubiquitin

complex

1

2

3

Laurence

Leherte

,

Daniel

P.

Vercauteren

Q1 4

UnitédeChimiePhysiqueThéoriqueetStructurale,UniversityofNamur,RuedeBruxelles61,B-5000Namur,Belgium

5 6

a

r

t

i

c

l

e

i

n

f

o

7 8 Articlehistory: 9 Accepted31October2013 10 Availableonlinexxx 11 12 Keywords: 13

Molecularelectrostaticpotential

14

Electrondensity

15

Smoothingofmolecularfields

16

Criticalpoints

17

Pointchargemodel

18 Protein 19 Ubiquitincomplex 20

a

b

s

t

r

a

c

t

Reducedpointchargemodelsofaminoacidsaredesigned,(i)fromlocalextremapositionsincharge densitydistributionfunctionsbuiltfromthePoissonequationappliedtosmoothedmolecular electro-staticpotential(MEP)functions,and(ii)fromlocalmaximapositionsinpromolecularelectrondensity distributionfunctions.Correspondingchargevaluesarefittedversusall-atomAmber99MEPs.Toeasily generatereducedpointchargemodelsforproteinstructures,librariesofaminoacidtemplatesarebuilt. TheprogramGROMACSisusedtogeneratestableMolecular Dynamics trajectoriesofanUbiquitin-ligand complex(PDB:1Q0W),undervariousimplementationschemes,solvation,andtemperatureconditions. Pointchargesthatarenotlocatedonatomsareconsideredasvirtualsiteswithanulmassandradius. Theresultsillustratehowtheintra-andinter-molecularH-bondinteractionsareaffectedbythedegree ofreductionofthepointchargemodelsandgivedirectionsfortheirimplementation;aspecialattention totheatomsselectedtolocatethevirtualsitesandtotheCoulomb-14interactionsisneeded.Results obtainedatvarioustemperaturessuggestthattheuseofreducedpointchargemodelsallowstoprobe localpotentialhyper-surfaceminimathataresimilartotheall-atomones,butarecharacterizedbylower energybarriers.Itenablesallowstogeneratevariousconformationsoftheproteincomplexmorerapidly thantheall-atompointchargerepresentation.

©2013ElsevierInc.Allrightsreserved.

21

1. Introduction

22

Numerous modelshave already been proposedin literature 23

regarding the coarse-graining of biomolecules and their corre-24

spondinginteractionpotentials.Severalreferenceswerealready 25

givenin[1].Reviews[2–6]aswellassoftware[7,8]areregularly

26

publishedonthesubject.

27

Inrecentpublications,wedescribedreducedpointcharge

mod-28

els [9,10] built from critical point (CP) analyses of smoothed

29

molecularproperties,andtheirapplicationstoMolecularDynamics

30

(MD)simulationsofproteins[1].Thesesimulationswereachieved

31

invacuumusingtheprogrampackageTINKER[11]whereinpoint

32

chargeswereconsideredasmassesattachedtotheprotein

struc-33

turethroughharmonicbonds.Themassthatwasassociatedwith

34

thechargeswassettoavalueofm=2inordertolimitthemass

35

increaseoftheproteinstructureandtoallowatimestepvalueof

36

1fs,alowervalueofmimplyingtoostrongadecreaseofthetime

37

steptogetstableMDtrajectories.Allothertermsoftheselected

38

∗ Correspondingauthor.Tel.:+3281724560;fax:+3281725466.

E-mail address:laurence.leherte@unamur.be(L.Leherte).

forcefield(FF),Amber99[12],werecalculatedattheall-atomlevel, 39

i.e.,asintheoriginalFFversion. 40

Inthepresentwork,thedesignofreducedpointchargemodels 41

isnotintendedtoleadtoacoarse-grainedmodelperse.Itispart 42

ofamoreglobalprojectregardingtheanalysisoflowresolution 43

molecularpropertiessuchaselectrondensity(ED)andmolecular 44

electrostaticpotential(MEP).FromtheveryfirstMDapplicationsof 45

areducedpointchargemodeltoproteinstructures[1],itwasfound 46

thatthesecondarystructureoftheproteinswasonlypartlylost 47

duringthesimulationswhiletheoverallthree-dimensional(3D) 48

foldremainedstable.Additionally,adecreaseofaboutafactor2of 49

thecalculationtimewasobservedforthesimulationsinvacuum. 50

Asa perspectivetotheworkreportedin[1],itwasexpectedto 51

adoptotherimplementationapproachestogetridofthenon-zero 52

massassignedtothecharges.Thisperspectiveledtothepresent 53

paper. 54

Inthepresentwork,twomolecularpropertiesleadingtodif- 55

ferent reduced point charge models areconsidered. First, asin 56

[1,10],alimitednumberofpointchargesisobtainedthroughthe 57

searchforthemaximaandminimaofasmoothedversionofthe 58

chargedensity(CD)generatedbytheatomicchargesdefinedin 59

Amber99(orAmber99SB)FF[12].Second,thepointchargesare 60

obtainedthroughasearchofthemaximaofthefullpromolecular 61

1093-3263/$–seefrontmatter©2013ElsevierInc.Allrightsreserved.

(6)

EDofthemolecularstructureand,asinthefirstapproach,charge

62

valuesareassignedtothosemaximausingaleast-squarecharge

63

fittingprocedure.Thislastmolecularpropertyiseasilycalculated

64

usingtheso-calledPromolecularAtomShellApproximation(PASA)

65

formalismthatwasdevelopedbyAmatandCarbó-Dorca[13,14].

66

Thosetwoapproachesledtovariousimplementationswhichare

67

discussedin thepresentpaper. Theprogram GROMACS[15,16]

68

wasselectedduetoitsimplementationtowardsshortercalculation

69

timesandthepossibilitytodefinevirtualsites,i.e.,particleswith

70

nulmassandradiusthatarecoupledtothemolecularstructure

71

throughgeometricalrules.

72

Theaimofthepaperistogodeeperinthemodellingof

pro-73

teinstructuresanddynamicsusingreducedpointchargemodels,

74

withanapplicationtoaparticularproteincomplex.Suchasystem

75

wasselectedtoprovideinformationontheusefulnessofthe

mod-76

elstosimulateshortpeptidesandproteinsaswellastheirmutual

77

interactions.Theeffectofthepointchargedistributionsonboth

78

theintra-andinter-molecularinteractionsinvarioussimulation

79

conditions,suchastemperatureandsolvation,isalsoinvestigated.

80

Atthisstageofourresearchwork,onlytheelectrostaticpartof

81

theFFismodified.Energetic,structural,anddynamicalproperties

82

arecalculatedandcomparedtotheall-atomones,whichiseasily

83

achievedasnoconversionstageisrequiredbetweenthereduced

84

andall-atommodels.However,keepingasignificantnumberof

all-85

atomcontributionstotheFFlimits,forthemoment,thepossible

86

gainincalculationtime.

87

Inthenextsection,wedetailhowthemodelsandtheir

imple-88

mentationwere designed. Then, MDsimulation resultsfor the

89

proteincomplexinvolvingtheUbiquitinInteractingMotifUIM-1

90

ofproteinVps27andUbiquitin(PDBaccesscode1Q0W)with

vari-91

ouspointchargemodels,andunderdifferentsimulationconditions

92

(temperature,solvationstate),areanalysedanddiscussed.

93

2. Backgroundtheory 94

Inthissection,wepresentthemathematicalformalismthatwas

95

usedtodesignamolecularreducedpointchargerepresentationand

96

itscorrespondingchargevalues.Asalltheseaspectswerealready

97

detailed before[9,10], we only providea short overview. First,

98

thesmoothingalgorithmisbrieflydescribed.Then,theapproach

99

appliedtolocatethepointchargesispresented,aswellasthe

pro-100

ceduretoassignchargevalues.Finally,theautomationprocedure

101

thatisimplementedtorapidlydeterminethepointchargelocations

102

foranyproteinstructureisexplained.

103

2.1. Smoothingofamolecularproperty

104

Inthepresentapproachtogeneratesmoothed3Dfunctions,a

105

smoothedCDorEDmapisalowerresolutionversionthatisdirectly

106

expressedasthesolutionofthediffusionequationaccordingtothe

107

formalismpresentedbyKostrowickietal.[17].Fromtheformalism

108

givenin[1],thesmoothedanalyticalCDdistributionfunctiona,s(r)

109

thatisobtainedfromanatomicchargeqaandthePoissonequation

110 isexpressedas: 111 a,s(r)= qa (4s)3/2e−r 2/4s (1) 112

wherea,s,andr,standfortheatomindex,thesmoothingfactor

113

(inBohr2,1Bohr=0.52918×10−10m),andthedistanceversusthe 114

atomposition,respectively.ThefullpromolecularEDiscalculated

115 as: 116 a,s(r)= 3



i=1

a,i where a,i=˛a,ie−ˇa,ir 2 (2) 117 with 118 ˛a,i=Zawa,i



2ςa,i 



3

2 1 (1+8ςa,is)3/2 and ˇa,i= 2ςa,i (1+8ςa,is) 119 (3) 120

where Za, wa,i and a,i, are theatomic number of atoma, and 121

thetwofittedparameters,respectively.Unsmoothedfunctionsare 122

obtainedbyimposings=0Bohr2. 123

2.2. Searchforcriticalpoints 124

AnalgorithminitiallydescribedbyLeungetal.[18]wasimple- 125

mented tofollow the trajectories of CPs,more specifically,the 126

maximaand/orminimainaCDorEDfunction,asafunctionofthe 127

degreeofsmoothing.Asalreadyreportedbefore[9],weadapted 128

theirideato3Dmolecularpropertyfunctions,f,suchas: 129

rf(s)=rf(s−s)+

ff(s)(s)· (4) 130

whererstandsforthelocationvectorofapointina3Dfunction, 131

suchasamolecularscalarfield,and/f(s)isthesteplength. 132

Thevariousstepsoftheresultingmerging/clusteringalgorithm 133

areasfollows.First,atscales=0,eachatomofamolecularstructure 134

isconsideredeitherasalocalmaximum(peak)orminimum(pit) 135

ofthescalarfieldf.Allatomsareconsequentlytakenasthestarting 136

pointsofthemergingprocedure.Second,assincreasesfrom0toa 137

givenmaximalvaluesmax,eachpointmovescontinuouslyalonga 138

gradientpathtoreachalocationinthe3Dspacewheref(s)=0.On 139

apracticalpointofview,thisconsistsinfollowingthetrajectoryof 140

thepeaksandpitsonthemolecularpropertysurfacecalculated 141

atsaccordingtoEq.(4).Thetrajectorysearch isstoppedwhen 142 |f(s)|islowerorequaltoalimitvalue,gradlim.Onceallpeak/pit 143

locationsarefound,closepointsaremergediftheirinter-distance 144

islowerthantheinitialvalueof1/2.Theprocedureisrepeated 145

foreachselectedvalueofs.Iftheinitialvalueistoosmallto 146

allowconvergencetowardsalocalmaximumorminimumwithin 147

thegivennumberofiterations,itsvalueisdoubled(ascalingfactor 148

thatisarbitrarilyselected)andtheprocedureisrepeateduntilfinal 149

convergence. 150

2.3. Chargecalculation 151

TostayconsistentwiththeanalyticalexpressionoftheAmber99 152

FF,onlypoint charge values areassigned toeach ofthe CPsof 153

a3Dmolecularpropertyfield.Inrecentliterature,onealsofinds 154

coarse-grainedelectrostaticenergytermswhichalsoinvolvedipo- 155

lar terms, such asin thework of Spiga et al. [19]. The charge 156

fittingprogramQFIT[20]wasusedasdetailedin[9].AllMEPgrids 157

werebuiltusing theAmber99[12] atomiccharges whichwere 158

assignedusingthesoftwarePDB2PQR[21,22],withagridstepof 159

0.5 ˚A.FittingswereachievedbyconsideringMEPgridpointslocated 160

between1.4and2.0timesthevanderWaals(vdW)radiusofthe 161

atoms.Thesetwolimitingdistancevalueswereselectedafterthe 162

so-calledMerz-Singh-Kollmanscheme[23].Sidechainsandmain 163

chainsoftheaminoacids(AA)weretreatedseparately,asdiscussed 164

in[9]. 165

Inallfittings, thetotal electricchargeand themagnitudeof 166

themoleculardipolemomentwereconstrainedtobeequaltothe 167

correspondingall-atomAmber99values.Alldipolemomentcom- 168

ponentswerecalculatedwiththeoriginoftheatomcoordinates 169

(7)

L.Leherte,D.P.Vercauteren/JournalofMolecularGraphicsandModellingxxx(2013)xxx–xxx 3 3. Designofaminoacidreducedpointchargemodels

171

ReducedpointchargerepresentationsofeachofthetwentyAAs

172

wereobtainedbyconsideringtheAAsinspecificconformational

173

states.ExceptforGlyandAla,mostrecurrentrotamerswere

gen-174

eratedbyconsideringtheangularconstraintsgiveninTable2of[9].

175

AllAAswereconsideredaselectricallyneutralexceptforArg+,His+,

176

Lys+,Asp−,andGlu−.Histidinewasalsomodelledinitsneutral

177

protonatedstatesHisandHis␧.

178

3.1. CD-basedtemplates

179

FromextendedpentadecapeptidechainsGly7-AA-Gly7

gener-180

atedusing SMMP05[24,25] onlythecentral AAwas keptwith

181

mainchainatoms(C␣ C O)AA(N H)AA+1.Then,thedesignofthe 182

AA point charge templateswas achievedin four stages, as

fol-183

lows.First,isolatedAAstructureswereassigned Amber99atom

184

chargesusingPDB2PQR[21,22].Sidechainextremawerelocated

185

usingourmerging/clusteringalgorithmappliedtotheCD

distri-186

butionfunctionssmoothedats= 1.7Bohr2,with

init=10−4Bohr2

187

andgradlim=10−6e−Bohr−2.Thiswascarriedoutseparatelyforthe

188

positivelyandnegativelychargedatoms.Second,thecharge

val-189

uesoftheresultingpeaksandpitstogetherwerefittedversusthe

190

all-atomMEPgeneratedfromthesidechain atomsonly.In this

191

procedure,severalrotamerdescriptionswereconsidered

accord-192

ingtotheiroccurrenceprobability(seeTable2of[9]).Third,the

193

mainchainpointchargeswerelocatedinaccordancewiththemotif

194

foundforGly8inanextendedGly15strand[9]and,fourth,asecond

195

chargefittingprocedure,nowcarriedoutversustheMEPcalculated

196

usingalltheAAatoms,wasachievedtodeterminethecharge

val-197

uesofthetwomainchainpointchargeswhilepreservingtheside

198

chainpointchargevaluesfirstobtained.

199

Allmainchainpointcharges,observedtobelocatedveryclose

200

totheCandOatoms,weresetexactlyonthoseatoms[1]

(Supple-201

mentaryInformationSI1).AllAAbearsidechainchargesexcept

202

Ala,Gly,Ile,Leu,andVal.AAmodelsaregiveninSI2andwere

203

discussedwithdetailsin[1].Inthepresentwork,anextra

proto-204

nationstateforHiswasgenerated,i.e.,His+.Itischaracterizedby

205

thehighestnumberofpointcharges,i.e.,6,versustheother

histi-206

dineresidues,i.e.,4and5.AsmostofthesepointchargesofHis+

207

areclosetoHatoms,theyweresettobelocatedexactlyonthese

208

atomstofacilitatetheimplementationofthepointchargemodel.

209

Fortheendresidues,achargeof+0.9288or−0.9288e−isseton

210

theNandOXTatoms,respectively[9,10].Inthefurtherpartsof

211

thispaper,themodelwillbereferredtoasmodelmCD.

212

Asecondpointchargedescriptionwasderivedfromthemodel

213

describedabove.Inthissecondmodel,tofullyfacilitatethe

imple-214

mentationoftheAAmodelsinGROMACS,mostofthepointcharges

215

weresetexactlyonatomsoftheresidues,andachargefitting

algo-216

rithmwasagainapplied.ResultsarepresentedinSI3.Thisimplies

217

thatonlythreeAAresidues,His+,Phe,andTrp,haveapointcharge

218

thatisnotlocatedonanatomoftheirstructure.Inthatmodel,one

219

obtainsendchargevaluesof+0.7705and−0.7705fortheendN

220

andOXTatoms.Inthefurtherpartsofthispaper,thesecondmodel

221

willbereferredtoasmodelmCDa.

222

AlastmodelbasedonthepointchargedistributionmCDwas

223

consideredsimilarlytotheapproachadoptedin[1]withthe

pro-224

grampackageTINKER[11].Amassm=2wasassignedtoeachpoint

225

chargeandharmonicconstraintswereappliedtobondsandangles

226

presentedinSI2.ThemodelwillbereferredtoasmodelmCDh.

227

3.2. PASA-basedtemplates

228

CPsearchesofthePASA EDdistributionfunctionswere

car-229

riedouttogeneratestillcoarserchargedescriptionsfortheAAs.

230

Indeed,withtheCD distributionfunctionsdepictedabove,it is

231

not possible toobtain less than two main chain point charges 232

perresidue,i.e.,onenegativeandonepositivechargeassociated 233

withtheOand Catoms, respectively.Withintheframeworkof 234

thePASA,theEDdepends onlyontheatomic numberZa ofthe 235

atoms,notontheircharge(Eqs.(2)and(3)).Theuseofthemerg- 236

ing/clustering algorithm wascarried out withinit=10−4Bohr2 237

andgradlim=10−5e−Bohr−2.Thislimit isanorderofmagnitude 238

greaterthanintheCDcaseastoofineagradlimthresholdincreases 239

thepossibilitytomisstherecognitionofduplicateCPsduringthe 240

search. A smoothing degreeof s=1.4Bohr2 wasconsidered. An 241 exceptionoccursforTrp forwhichoneobservestwosidechain 242

CPsatsbeloworequalto1.05Bohr2.Weselectedthatvalueto 243 betterdifferentiatethatresiduefromtheotheraromaticresidues 244

characterizedbyonlyoneCP.AseachAAmainchaininvolvesonly 245

oneCP,itschargevaluewasdirectlysetasthesumoverthecor- 246

respondingatomiccharges.Then,forAAsinvolvingmorethanone 247

sidechainCP,achargefittingprocedurewasappliedasfortheCD- 248

basedmodels.Forendresidues,oneobservednomainchainCPon 249

theterminalNH3+group.ThechargeofthemainchainCPofthe 250

N-terminalresidueisthusincrementedby+1,whilethemainchain 251

chargeoftheC-terminalresidueisincrementedby−1.Individual 252

AApointchargerepresentationsaregiveninFig.1.Regardingthe 253

sidechains,mostoftheresiduesarecharacterisedbyonlyoneCP. 254

Theirlocationismostlydeterminedbytheatomswiththehighest 255

atomicnumber,i.e.,S,O,N,andC.TheHatomsdonotsignificantly 256

affecttheEDdistributionfunctionsandthismakesallHisresidues 257

lookingalike.Inthefurtherpartsofthispaper,themodelwillbe 258

referredtoasmodelmPASA.Itsimplementationwithintheprogram 259

GROMACSisdetailedinSI4. 260

3.3. Automatedpointchargegenerationprocedure 261

Thefourpointchargetemplatesdescribedaboveareestablished 262

forisolatedAAstructures.Theirpropertiesarethusindependenton 263

theneighbourhoodoccurringinaparticularproteinofacomplex 264

structure.Thispresentsagreatadvantagewhenthoseproperties 265

are transferable.In a previous work[9], one indeedsuggested, 266

throughagoodapproximationofMEPprofilesofionchannelsand 267

numerousfreeenergycalculations,thattransferabilityoccursfor 268

rigidproteinstructures.Additionally,intheirpaperregardingthe 269

optimalnumberofcoarse-grainedsitesinbiomolecularcomplexes, 270

Sinitskiyetal.concludedthatthetransferabilityofindividualpro- 271

teinpropertiesbetweenunboundandboundstatesissupported 272

bythepossibilitytocoarse-graincomplexpartnersindependently 273

onefromeachother[27]. 274

To study large protein structures, an automation stage was 275

developedtorapidlylocatepointchargesonthestructure.Itisfully 276

basedontheapplicationofasuperimpositionalgorithmofCPtem- 277

platesofeachAAontotheircorrespondingall-atomstructureof 278

theproteinunderstudy.WeusedtheprogramQUATFIT[28,29] 279

to,first,superimposealimitedsetofatomsfromthetemplateon 280

thestudiedstructure,andthenusetheresultingtransformation 281

matrixtogeneratethecorrespondingpointchargecoordinates.The 282

templatesobtainedfromCDdistributionfunctionswerealready 283

madeavailableinTable3of[9].TheHis+templatenewlystudied 284

inthepresentworkisprovidedinSI5.Additionally,thetemplates 285

obtainedfromtheanalysisofthePASAEDdistributionfunctions 286

arereportedinSI6.TheGROMACStopologyfile,whereinpoint 287

chargesaredefinedasvirtualsites,isfurthergeneratedthroughan 288

in-houseprogramthatoutputsgeometricalparametersasreported 289

inSI1,3,and4,forthemCD,mCDa,andmPASAmodels,respectively. 290

WhenaGROMACSvirtualsiteisgenerated,anumberofatoms, 291

twoorthree,areselectedtodetermineitslocationinspace.The 292

choiceofthereferenceatomsforeachpointchargeisnotunique. 293

Wehavemostofthetimeconsideredtheclosestatomsbyexcluding 294

(8)

Fig.1.Pointchargemodelforthe20AAresiduesasestablishedats=1.4Bohr2fromthehierarchicalmerging/clusteringalgorithmappliedtotheall-atomPASAEDfunction.

PointchargesarenumberedasinSupportingInformationSI4.FiguresweregeneratedusingOpenDX[26].

rotationsaroundtheC Obond.Othermodelsmightobviouslybe

296

tested,forexampleforneutralHisresidues,thatarenotusedin

297

thepresentstudies,especiallytodeterminetheirinfluenceon

H-298

bondinteractions.DuringaMDsimulation,theforcesactingonthe

299

virtualsitesareredistributedamongtheirreferenceatoms.Force

300

redistributionwaspartlylimitedinmodelmCDabylocatingmost

301

ofthepointchargesonatoms.

302

Asalreadymentioned,pointchargesareconsideredasvirtual

303

sitesthat actonly throughCoulombinteractions. Anadditional

304

implementationstrategywasconsideredwherepointchargesare

305

seen as masses interactingwith the protein structure through

306

restrainedharmonicbonds.Theadvantageofsuchanapproachlies 307

inthefactthattheelectrostaticforcesactingonthechargesarenot 308

redistributedamongtheatoms,butthemodelisbiasedbyartificial 309

massesaddedtothesystem. 310

4. ApplicationtotheMDoftheVps27UIM-1–Ubiquitin 311

complex 312

Vps27UIM-1,ashort␣-helicalstructuremadeof24AAresidues 313

(numbered255–278in thePDB),isknowntointeractwiththe 314

(9)

L.Leherte,D.P.Vercauteren/JournalofMolecularGraphicsandModellingxxx(2013)xxx–xxx 5

Table1

DescriptionofthepointchargemodelsusedfortheAmber99SB-basedMDsimulationsoftheVps27UMI-1–Ubiquitincomplex.

No.ofwater molecules

Totalno.ofpointcharges associatedwithbothcomplex partners(Vps27UIM-1/Ubiquitin)

No.ofnon-atomic pointcharges

Boxsize(nm)(from finalsnapshot) All-atom 10,553 394/1227 0 6.915 All-atom-2 13,269 394/1227 0 7.442 mCD 10,542 96/286 112 6.909 mCDa 10,551 96/286 3 6.899 mCDh 10,542 96/286 112 6.916 mPASA 10,551 36/102 136 6.919

hydrophilicresidues,encompassingahydrophobicmotifthat con-316

tainsresiduesLeu262,Ile263,Ala266,Ile267,Leu269,andLeu271, 317

thatinteractwithresiduesLeu8,Ile44,Val70,andAla56ofUbiquitin 318

[30,32].AccordingtoSwansonetal.[30],electrostaticinteractions

319

areexpectedtooccurbetweennegativelychargedGlu−residues

320

ofVps27UMI-1(Glu−257,Glu−259,Glu−260,andGlu−261)and

321

Arg+residuesofUbiquitin(Arg+42,Arg+72,andArg+74),aswell

322

as between Glu−273 and His+68, while H-bonds are observed

323

betweenAla266andHis+68,Ser270andAla46andGly47,aswell

324

asbetweenGly47andHis+68[33].

325

The study of such a protein-protein system using MD

326

approachesisnotnew[33–35]andtheapplicationspresentedin

327

thispaperareintended totest andassess,versustheirall-atom

328

counterpart,thepoint chargereducedmodelsdevelopedabove.

329

Ourreferenceworksarethusthedataavailableinliteratureaswell

330

asourownall-atomMDsimulationresults.

331

Molecularsimulationconditionswerekeptascloseaspossible

332

ofthoseproposedbyShowalterand Brüschweilerintheirwork

333

abouttheAmber99SBFF[36].MDtrajectoriesofthesystemwere

334

runusingtheGROMACS4.5.5programpackage[15,16]withthe

335

Amber99SBFF[37]underparticlemeshEwaldperiodicboundary

336

conditions.Long-rangedispersioncorrectionstoenergyand

pres-337

surewereapplied.Theinitialconfigurationswereretrievedfrom

338

theProteinDataBase(PDB:1Q0W)andsolvated,ifrequired,using

339

TIP4P-Ewwatermolecules[38]soasproteinatomslieatleastat

340

1.2nmfromthecubicboxwalls.TheVps27UMI-1andUbiquitin

341

partnersinvolveeach394and1227atoms,respectively.To

can-342

celthenetchargeofstructure1Q0W,twoNa+ionswereadded

343

tothesystemusingtheiongeneratortoolofGromacs.As

speci-344

fiedin[30],theHisresidueofUbiquitinisfullyprotonated(His+

345

state).Thesystemswerefirstoptimizedandthenheatedto50K

346

througha10pscanonical(NVT)MD,withatimestepof2fsand

347

LINCSconstraintsactingonbondsinvolvingHatoms.The

trajec-348

torywasfollowedbytwosuccessive20psheatingstages,at150

349

and300K,underthesameconditions.Next,eachsystemwas

equi-350

libratedduring 50ps in theNPT ensembleto relaxthesolvent

351

molecules.Finally,a20nsMDsimulationwasperformedintheNPT

352

ensemble,forsolvatedsystems.Invacuum,onlytheNVTensemble

353

wasused.The‘V-Rescale’and‘Parrinello-Rahman’algorithmswere

354

selectedtoperformNVTandNPTsimulations,respectively.Incase

355

ofobviouslackofequilibration,anextraproductionrunof20ns

356

wasperformed.WhenconsideringmodelmCDh, theconstraints

357

actingonthebondsinvolvingHatomshadtoberemovedandthe

358

timestepwassetequalto1fs,thusleadingtotwicethenumber

359

ofMDiterationsasintheall-atom,mCD,andmCDasimulations.

360

Snapshotsweresavedevery2ps,i.e.,twicethevalueconsidered

361

byShowalterandBrüschweiler[36];thatchoicedidnot

signifi-362

cantlyaltertheresults.Adescriptionofthesystemsunderstudyis

363

presentedinTable1.Thetotalnumberofpointchargestobe

con-364

sideredfortheproteincomplexisreducedbyafactorof4.2and11.7

365

fortheCD-andPASA-basedmodels,respectively.Dependingupon

366

theimplementation,thenumberofnon-atomicchargesislargely

367

variable.Forinstance,thereareonlythreeofsuchpointcharges

368

inmodelmCDa,whichoriginatefromtheHis+andPheresidues

369

ofUbiquitin.Thereare,ineachsimulation,approximately10,500 370

watermoleculesthatarenotcoarse-grained.Alargersolvationbox 371

wasusedforthesystemnamedAll-atom-2whichcorrespondstoa 372

highlydifferentstructureofthecomplex.Thisparticularcasewill 373

bedescribedlaterinthepaper,inSection4.4. 374

It is clear that a real gain in simulation time will be pos- 375

sibleif coarse-graining occursat thesolvent level. Areview of 376

coarse-grainedwatermodelscan,forexample,befoundin[39,40]. 377

Workingwithanimplicitsolventrepresentationisanothereffi- 378

cientwaytolargelyreducethecalculationtimeasdiscussedin 379

[41,42].Asarecentexample,letusmentiontheapproachadopted 380

inthecoarse-grainedFFPRIMObyKaretal.[43].Atthepresent 381

stageofourwork,noinformationisavailableregardingtheradius 382

valuetobeassignedtothenon-atomicpointcharges.Theapproach 383

weemployedearliertocalculatefreeenergyofsolvationusingthe 384

programAPBS[44],i.e.,toassignanulradiustothepointcharges 385

[10],appearednottobeeffectivewithGROMACS.Thus,interfacing 386

ourreducedpointchargemodelswithcoarse-grainedorimplicit 387

solventrepresentationsis a perspectivetobring tothepresent 388

work. 389

4.1. Molecularelectrostatic maps 390

MEPmapsaredisplayedinFig.2forthefouroptimizedcom- 391

plexes,i.e.,thestartingpointsoftheMDsimulations,described 392

using the all-atom, mCD, mCDa, and mPASA models. Solvent 393

molecules and ions werenot included in thecalculations. It is 394

noticedthatthecontoursdisplayedatFig.2,i.e.,anegativeMEP 395

contourforVps27UIM-1(−0.2e−Bohr−1)andapositiveMEPcon- 396

tourforUbiquitin(+0.1e−Bohr−1),showsimilar3Dshapesforeach 397

ofthefourmodels.Similarbehaviourswerealreadyputforwardin 398

thestudyofrigidsystemssuchasionchannels[9].Atshortrange, 399

e.g.,in thecaseof intramolecularinteractions, thepoint charge 400

models are expected toaffect thedynamical behaviour of the 401

moleculesasdetailedfurtherinthepaper.Topreliminaryillustrate 402

thatassumption,MEPmapsaredisplayedfortwoindividualAAs, 403

Glu−andHis+(Fig.3).Inthecaseofglutamate,oneobserveshigher 404

MEPvaluesatthetwonegativesidechainchargesformodelmCD 405

and,obviously,theabsenceofsuchaseparationformodelmPASA 406

whichinvolvesonlyonesidechainnegativecharge.Theall-atom 407

representationofthemainchainistheonlyonetoletappeartwo 408

negativelychargedsites.Onemaythusexpectstructuralchanges, 409

especiallyinsecondarystructureelements.Forprotonatedhisti- 410

dine,thepositiveMEPareaslooksimilar,exceptforthemainchain 411

andthemPASAmodel.Onethusassumesthatachangeinthepoint 412

chargemodelwillaffect,atleast,theformationofH-bondsduringa 413

MDsimulation.AsseenlaterduringtheanalysisoftheMDtrajecto- 414

ries,changesinthesecondarystructureelementswillbeobserved, 415

aswellasinthefoldofthecomplexinsomecases.Asatentative 416

toeliminatetheeffectofUbiquitinstructuralchangesonVps27 417

UIM-1,separateMDsimulationswerecarriedoutbyrestraining 418

Ubiquitintoitscrystalstructure.SuchtrialsdidnotpreventVps27 419

UIM-1toalter its shape and theseMDsimulations willnot be 420

(10)

Fig.2.MEPcontoursofUbiquitin(red:+0.1e−Bohr−1)andVps27UIM-1(blue:−0.2eBohr−1)intheirinitialoptimizedconfiguration,asobtainedusingtheAll-atom,

mCD,mCDa,andmPASAmodels.Ubiquitinanditsligandaredisplayedusinglightblueandblackspheres,respectively.FiguresweregeneratedusingOpenDX[26].(Foran

interpretationofthereferencestocolourintheartwork,thereaderisreferredtothewebversionofthearticle).

Q2

Fig.3. MEPcontoursof(top)Glu−273(−0.4to−0.1e−Bohr−1)and(bottom)His+68(0.1to0.5eBohr−1)asobtainedusingtheAll-atom,mCD,mCDa,andmPASAmodels.

(11)

L.Leherte,D.P.Vercauteren/JournalofMolecularGraphicsandModellingxxx(2013)xxx–xxx 7

Fig.4.SecondarystructureofthecomplexVps27UIM-1–Ubiquitinobservedduringthelast20nsAmber99SB-basedMDtrajectoriesinwater(left)andinvacuum(right) at300K,asobtainedusingtheAll-atom,All-atom-2,mCD,mCDa,mCDh,andmPASAmodels.Vps27UIM-1andUbiquitininvolvethefirst24andlast76aminoacidresidues, respectively.Secondarystructureelementsarecolour-codedasfollows:Coil(white),␣-helix(blue),␲-helix(purple),310helix(grey),␤-sheet(red),␤-bridge(black),bend

(green),turn(yellow),chainseparation(lightgrey).

flexiblesystems,exceptforconstraintsappliedtobondsinvolving

422

Hatomsasmentionedabove.

423

4.2. All-atomMDtrajectories

424

AfirstanalysisoftheMDtrajectoriesobtainedforthemodel

425

namedAll-atomdealtwiththesecondarystructureofthewhole

426

complexandshowedthatitisslightlymoreconservedinwater

427

(Fig.4left)thaninvacuum(Fig.4right),duetothestabilizing

con-428

tributionofwaterontheproteinstructure.Invacuum,allsecondary

429

structureelementslike␣-helicesand␤-strands,except␤-strands

430

2–7and12–16ofUbiquitin,areshorter.Inbothcases,thehelical 431

structureoftheUIM-1unitisloosen,toalargerextendinvacuum 432

(Figs.4and5). 433

Astudyof the3D foldoftheproteincomplex wasachieved 434

toespeciallydeterminethebindingofVps27UIM-1toUbiquitin. 435

Distancemapsbetweentheatomsofthetwopartnerswerebuilt 436

byconsideringtheminimaldistancesbetweentheAAatomsof 437

thetwopartnersand arereportedin Fig.6.Thedistances were 438

calculatedoverthelast10nsoftheMDtrajectoriestominimize 439

theeffectofa slowsecondary structure relaxationprocessthat 440

(12)

Fig.5.FinalsnapshotsoftheproteincomplexVp27UIM-1(red)–Ubiquitin(blue)obtainedfromthelast20nsAmber99SB-basedMDtrajectoriesinwater(left)andin vacuum(right)at300K,asgeneratedusingtheAll-atom,All-atom-2,mCD,mCDa,mCDh,andmPASAmodels.Structuralelementsarecolour-codedasfollows:␤-strandsof UbiquitininteractingcloselywithVps27UIM-1(yellow),Arg+72 and Arg+74 (green), Glu−257andGlu−259toGlu−261(white),H-bondsbetweenthetwopartners(purple spheres).FiguresweregeneratedusingVMD[45].(Foraninterpretationofthereferencestocolourintheartwork,thereaderisreferredtothewebversionofthearticle).

mCD,mCDa,and mCDh.Inthemapgeneratedbytheanalysisof

442

thesolvatedAll-atomMDtrajectory,oneclearlydistinguishesthree

443

regionsextendedalongtheVps27UIM-1chain.Thisextensionis

444

duetothespatialalignmentofVps27UIM-1withanumberof

␤-445

strandsofUbiquitin.Thefirstregioncorrespondstothecontacts

446

occurringbetweensegment259to272ofVps27UIM-1and

␤-447

strand4–10ofUbiquitin,whilethesecondandthirdregionsaredue

448

tocontactswith␤-strands40–45and48–49and␤-strand66–72, 449

respectively.Theshortestdistances,below0.3nm,appearinthese

450

twolastareas.Thereisalastregionofinterest,generatedbythe

451

contactsbetweenGlu−257andGlu−259ofVps27UIM-1andArg+

452

residueslocatedattheC-terminalsegmentofUbiquitin.Invacuum,

453

thesefourregionsareenhancedduetothecloserlocationofVps27

454

UIM-1versusUbiquitin.Theyarelargerastheynowinvolvethe

455

C-terminalresiduesofVps27UIM-1withemphasizedelectrostatic

456

“contacts”betweentheN-terminalGlu−residues(257,259–261)

457

ofUIM-1andN-terminalArg+residues(72and74)ofUbiquitin.

458

Theanalysisof selectedenergytermsaveragedover thelast

459

10nsoftheMDtrajectoriesprovidedtheresultsreportedinTable2.

460

Theabsolutevalueoftheinteractionenergybetweenthetwo

part-461

nersofthecomplex,|E12|, occurswitha ratioof4.6 and 13.6% 462

versusthetotalsolutepotentialenergy(E1+E2),inwaterandin 463

vacuum,respectively.WhenCoulombinteractionsareconcerned,

464

thecorresponding ratio in water, 2.2%, also increases toreach

465

a value of 7.8% in vacuum. The higher relative contribution of

466

|E12|and|Cb12|supportsthehighercompactnessofthecomplex 467

asjustdiscussed fromdistancemaps.Intra-moleculartotal and

468

Coulombpotential energies of Vps27 UIM-1,i.e., |E1| and |Cb1|

469

terms,areratherconstant.Theycontributeto25.9and24.7%,and

470

26.2and23.6%,respectivelyinwaterandinvacuum(Table2).The 471

vacuum-inducedcompactnessinvolvesadecreaseinthemobil- 472

ityoftheUIMatomsversusUbiquitin asillustratedbytheRoot 473

MeanSquareFluctuations(RMSF)oftheVps27UIM-1atomswith 474

Table2

EnergyratioscalculatedfromvaluesobtainedfromAmber99SB-basedMD trajec-tories(reportedinSI7)fortheVps27UIM-1Ubiquitinsystem.Indices‘1’and ‘2’,standforVps27UIM-1andUbiquitin,respectively.EandCbstandfortotal (Coulomb+Lennard–Jones)andCoulombpotentialenergy,respectively.

E12 E1+E2 Cb1+Cb2Cb12 E1+E2E1 Cb1+Cb2Cb1 Solvated All-atom −4.6 −2.2 25.9 24.7 All-atom-2 −2.1 −0.6 26.8 25.1 mCD −2.2 −1.0 21.1 20.5 mCDa −3.7 −2.0 24.2 22.8 mCDh 0.001 −363.6 18.5 53.4 mPASA 34.6 34.8 12.9 29.9 mCD(277K) −2.3 −1.4 22.4 21.2 mCD(250K) −2.1 −1.0 22.6 21.3 mCD(150K) −2.4 −0.9 21.5 20.9 Vacuum All-atom −13.6 −7.8 26.2 23.6 All-atom-2 −11.4 −6.6 24.7 22.7 mCD −9.6 −5.9 19.3 18.0 mCDa −11.4 −7.0 24.2 22.1 mCDh −5.8 −620.4 18.1 −3.0 mPASA 24.0 23.3 21.3 27.9 mCD(277K) −6.4 −3.4 18.5 17.2 mCD(250K) −8.4 −4.8 19.5 18.3 mCD(150K) −7.2 −4.0 18.5 17.2

(13)

L.Leherte,D.P.Vercauteren/JournalofMolecularGraphicsandModellingxxx(2013)xxx–xxx 9

Fig.6.DistancemapsoftheproteincomplexVps27p–Ubiquitinestablishedduringthelast10nsoftheAmber99SB-basedMDtrajectoriesinwater(left)andinvacuum (right)at300K,asobtainedusingtheAll-atom,All-atom-2,mCD,mCDa,mCDh,andmPASAmodels.Scaleiscolouredusingadistanceincrementof0.1nm.(Foraninterpretation ofthereferencestocolourintheartwork,thereaderisreferredtothewebversionofthearticle).

All

-a

tom

All

-atom

-2

mCD

mCDa

mCDh

mPASA

0.0 0.4 0.8 1.2 0 200 400 600 800 1000 1200 1400 1600 RMSF (nm) Atom number 0.0 0.4 0.8 1.2 0 200 400 600 800 1000 1200 1400 1600 RMSF (nm) Atom number 0.0 0.4 0.8 1.2 0 200 400 600 800 1000 1200 1400 1600 RMSF (nm) Atom number 0.0 0.4 0.8 1.2 0 200 400 600 800 1000 1200 1400 1600 RMSF (nm) Atom number 0.0 0.4 0.8 1.2 0 200 400 600 800 1000 1200 1400 1600 RMSF (nm) Atom number 0.0 0.4 0.8 1.2 0 200 400 600 800 1000 1200 1400 1600 RMSF (nm) Atom number

Fig.7. RMSFoftheVps27UIM-1andUbiquitinatoms(atoms1–394and395–1621,respectively)inwater(black)andinvacuum(red)calculatedfromthelast10nsofthe Amber99SB-basedMDtrajectoriesat300K,asobtainedusingtheAll-atom,All-atom-2,mCD,mCDa,mCDh,andmPASAmodels.(Foraninterpretationofthereferencesto colourintheartwork,thereaderisreferredtothewebversionofthearticle).

(14)

Table3

AveragenumbersofH-bondsoccurringbetweenvariouscomponentsoftheVps27UIM-1–Ubiquitinsystem,asobtainedfromtheanalysisofthelast10nsofthe Amber99SB-basedMDtrajectoriesat300K.

UIM–Ubiquitin Complex–water UIM–water

Solvated Vacuum Total Mainchain Sidechains N H C O

All-atom 3.6±1.7 14.0±1.4 287.8±8.3 105.2±5.0 182.5±7.3 20.9±2.9 69.1±3.7 84.5±4.6 All-atom-2 0.9±1.0 10.4±1.1 300.1±8.9 119.6±5.1 180.5±7.4 9.3±1.6 95.2±1.3 97.4±4.9 mCD 1.6±1.0 7.7±1.5 330.2±9.5 197.0±6.0 133.2±7.1 23.8±3.9 158.2±4.2 90.0±5.0 mCDa 0.6±0.8 5.7±1.7 333.2±11.4 175.1±7.8 157.9±6.8 23.2±3.8 138.7±5.9 101.6±4.8 mCDh 2.0±1.2 4.7±1.4 300.4±9.3 176.2±6.7 124.2±7.2 20.8±3.7 140.1±6.0 89.6±5.0 mPASA 2.3±1.4 3.2±1.4 116.3±8.2 21.7±4.1 94.6±6.7 6.5±2.6 6.2±2.3 42.2±4.8 mCD(277K) 1.2±0.9 1.0±0.9 328.4±9.2 188.3±5.3 140.1±7.4 24.1±3.6 148.6±3.6 104.8±6.3 mCD(250K) 0.6±0.7 5.5±1.2 308.5±8.5 161.7±8.1 146.8±7.2 18.0±3.3 128.4±6.0 101.6±4.9 mCD(150K) 3.1±0.4 3.2±0.7 264.2±5.4 118.0±2.3 146.2±4.8 9.3±1.6 95.2±1.3 81.1±2.8

time(Fig.7).Oneclearlydistinguishesadrasticdecreaseinthe

475

RMSFvalues associatedwiththeatomsoftheend segmentsof

476

thepeptideUIM-1,i.e., aroundatoms1–100and 300–394.This

477

decreaseinmobilityiscorrelatedtoahigheraveragenumberof

H-478

bondsoccurringbetweenthetwopartnersofthesystem,14.0±1.4,

479

aboutfourtimesthenumberofH-bondsinwater,i.e.,3.6±1.7,as

480

reportedinTable3.H-bondsaredeterminedbasedoncut-off

val-481

uesof30◦ and0.35nmfortheangleHydrogen–Donor–Acceptor

482

andthedistanceDonor–Acceptor,respectively.Inwater,verylow

483

numbersofH-bondscanbeobservedbetweenthetwopartners.

484

Forinstance,thesolvatedcomplexletsappearonlytwoH-bonds

485

att=20ns,which areformedbyN H(Gly47)···OG(Ser270)and

486

NE2-HE2(His+68)···OE2(Glu−273)atoms.ThesetwoH-bondsare

487

characterizedbyapercentageofoccurrenceabove80%duringthe

488

simulationtimeandarethusthemostpersistentonesas

empha-489

sizedinSI8wheretheoccupancyoftheH-bondsformedbetween

490

thetwoproteinpartnersisreported.H-bondoccupancyalongMD

491

trajectorieswascalculatedusingVMD1.9.1[45]withthreshold

dis-492

tanceandanglevaluesof3.5 ˚Aand30◦,respectively.Contrarily,

493

invacuum,upto15 H-bondsappear att=20ns(Fig.5), which

494

mostlyinvolvetheendsegmentsofUIMwithGlu−residuesthat

495

foldtowardstheArg+residuesofUbiquitin.Invacuum,alarger

496

numberofH-bondsappearasasubstitutetotheprotein-solvent

497

H-bondnetworkobservedinthesolvatedstate.

498

ProteinhydrationcanbestudiedthroughtheanalysisofRadial

499

DistributionFunctions(RDF)asplottedinFig.8.Asexplainedinthe

500

paperbyVirtanenetal.[46],afirsthydrationshelloccursbetween

501

0.1and0.2nmfromtheproteinatoms,andisfollowedbya

sec-502

ondsharplymarkedshelljustbelow0.3nm(Fig.3of[46]).Inthe

503

presentwork,RDFbetweenoxygenatomsofwaterandtheprotein

504 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 RDF (O w -p ro te in) r (nm) All-atom All-atom-2 Uncharged protein atoms mCD

mCDa mCDh

mPASA

Fig.8.RDFofwateroxygen–proteinatompairsofthesolvatedsystemVps27UIM

-1–Ubiquitincalculatedfromthelast10nsoftheAmber99SB-basedMDtrajectories at300K,asobtainedusingtheAll-atom,All-atom-2,mCD,mCDa,mCDh,mPASA,and neutralproteinatommodels.(Foraninterpretationofthereferencestocolourin

theartwork,thereaderisreferredtothewebversionofthearticle).

atomsshowaslightfirsthydrationshellatadistanceof0.192nm 505

withacontactdistanceof0.154nm.Thisshellinvolvesaverylim- 506

itednumberofwatermolecules;integrationunderthefirstpeak 507

oftheRDFfunctionleadstoavalueof84H2Omolecules.Asecond 508

shellappearsatabout0.280nm,adistancethatwasactuallyiden- 509

tifiedasthefirsthydrationshellofcrystallineproteinsbyChen 510

etal.[47].ThesecondshellidentifiedbyChenetal.,correspond- 511

ingtowaterinteractionswithproteinnon-polaratomsis,inour 512

simulatedproteinsystem,locatedat0.372nm. 513

BesidesadifferenceintheUIMshapebetweenthesolvatedand 514

vacuumstates(Fig.5),oneadditionallynoticesthatthegyration 515

radiusrGofUbiquitinisaffectedbytheenvironment.Indeed,mean 516

valuesof1.20±0.01and1.15±0.01nmareobtainedinwaterand 517

invacuum,asreportedinTable4whereinaverageswerecalcu- 518

latedoverthelast10nsof thefinal MDtrajectoriesasdriftsin 519

rGwerestillappearingduringthefirst10ns.Theslightstructure 520

contractionobservedinvacuumisduetothelackofinteractions 521

withsurroundingwatermoleculesandcomeswithreducedatom 522

fluctuations(Fig.7). 523

Regardingthesolventitself,acalculationofthemeansquare 524

displacement as a function of time, carried out for molecules 525

locatedwithin0.35nm, andbetween0.35 and 1.2nmfromthe 526

proteinstructure,showsthatallsetsofwatermoleculesbehave 527

asaBrownianfluidwithasimilarself-diffusioncoefficientDof 528

(2.38±0.01)×10−5and(2.40±0.02)×10−5cm2s−1,respectively 529

(Table4).TheveryslightdecreaseinDforwatermoleculesinter- 530

actingcloselywiththesolutedoesnotappeartobesignificant;all 531

Dvaluesstayclosetotheself-diffusioncoefficientofwatercalcu- 532

latedwiththeTIP4P-Ewpotential,i.e.,(2.4±0.06)×10−5cm2s−1 533

[38]. 534

4.3. mCDMDtrajectories 535

AsexplainedlaterwhendiscussingtheshapeoftheUbiquitin 536

partner,asecond20nsMDproductionstagewascarriedoutforthe 537

solvatedcomplexwhenusingthemCDmodel.Allresultsdiscussed 538

belowwereobtainedfromtheanalysisofthatadditionalrun. 539

Thestudyofthesecondarystructureofthecomplexinwaterand 540

invacuumdirectlyshowsanenhancedlossofthesecondarystruc- 541

tureelementsofthesystemversustheAll-atomsimulationresults 542

(Fig.4).Partofthe␤-strandsdisappearsandhelicesarethemotifs 543

thatarethemostperturbedduringthesimulations.However,in 544

vacuum,regularmotifs,especiallythehelixoftheUbiquitinstruc- 545

ture,appeartobeslightlymorepreserved.Indeed,inwater,the 546

electrostaticinteractionoftheproteinwiththesolventmolecules 547

isparticularlymodifiedasillustratedbythenumberofH-bonds 548

occurringbetweenthemainchainandthesidechainsofthesolute 549

andwater(Table3).Surprisingly,alargeraveragenumberofmain 550

chainH-bonds,197.0±6.0versus105.2±5.0,isobtainedformCD 551

despitetheabsenceofchargesonNandHatoms.Itappearstobe 552

(15)

L.Leherte,D.P.Vercauteren/JournalofMolecularGraphicsandModellingxxx(2013)xxx–xxx 11

Table4

MeangyrationradiusrGofUbiquitinandself-diffusioncoefficientDofwatermoleculessolvatingtheVps27UIM-1–Ubiquitinsystemasobtainedfromtheanalysisofthe

last10nsoftheAmber99SB-basedMDtrajectoriesat300K.

rG(nm) D(×10−5cm2s−1)

Solvated Vacuum Within0.35nmof thesolute Between0.35and 1.20nmfromthe solute All-atom 1.20±0.01 1.15±0.01 2.38±0.01 2.40±0.02 All-atom-2 1.19±0.01 1.12±0.00 2.46±0.08 2.39±0.04 mCD 1.36±0.02 1.15±0.00 2.35±0.05 2.31±0.02 mCDa 1.30±0.01 1.12±0.00 2.20±0.02 2.33±0.02 mCDh 1.28±0.01 1.13±0.00 2.37±0.03 2.36±0.05 mPASA 1.17±0.01 1.14±0.01 2.46±0.08 2.47±0.05 mCD(277K) 1.25±0.01 1.13±0.00 1.21±0.01 1.26±0.04 mCD(250K) 1.20±0.01 1.12±0.00 0.38±0.00 0.43±0.00 mCD(150K) 1.20±0.00 1.12±0.00 ∼10−5 ∼10−5 versustheAll-atomcase,affectstheformationofsuchatypeof

inter-554

action.EvenintheabsenceofchargesontheNandHatomsofthe 555

AAmainchains,theaveragenumberofH-bondsisnotdrastically 556

modified,with23.8±3.9versus20.9±2.9bondsobservedforthe 557

mCDandAll-atommodels,respectively.Contrarily,thenumberof 558

H-bondsinvolvingthesidechainatomsisreducedtoanaverage 559

valueof133.2±7.1versus182.5±7.3formCDandAll-atom mod-560

els.TheconsequenceofthosechangesinthenumberofH-bonds 561

formedwiththesolventisillustratedinFig.8,whereitisclearly

562

seenthattheveryfirsthydrationshellisreducedtoaweak

shoul-563

derintheRDFcurvewithmodelmCD.Nevertheless,astudyofthe

564

distanceandanglevaluesadoptedbytheH-bondsshowsthatthe

565

distributionsformodelmCDaresimilartothosethatarevalidfor

566

theAll-atommodel,i.e.,centredaround0.27nmand10.5◦ (Fig.9

567

top).Itishowevernoticedthat,regardingintra-proteinH-bonds

568

(Fig.9bottom),ifthedistancedistributionisrathersimilartothe

569

all-atomone,thereisasignificantdisplacementofthemaximum

570

oftheangledistributiontowardshigheranglevalues,i.e.,about25◦ 571

versus15◦fortheAll-atommodel.Theselasttrendsareobservedfor 572

boththesolvatedandisolatedsystems. 573

Asvisualizedin Fig.5,both ends ofthe solvatedUIM-1get 574

separatedfromUbiquitinwhilethecentralhydrophobicsegment 575

(Ile263toLeu271)staysattheproximityofthetwostable␤-strands 576

ofUbiquitin,i.e.,segments39–44and67–70.Theconsequenceof 577

suchaconfigurationchangeisillustratedbythedistancemapsdis- 578

playedinFig.6whereacontactareainvolvingthefirstresidues 579

ofUbiquitinfadesawayduetothelossofthetwofirst␤-strand 580

elementsoccurringalongtheUbiquitinchain,i.e.,segments2–7 581

and12–16.Onthecontrary,␤-strands40–45and48–49,aswell 582

as strand 66–72 are strongly preserved but the H-bonds they 583

formarecharacterizedbyoccurrencepercentagevalueslowerthan 584

20%.Nevertheless,theloweraveragenumberofH-bondsformed 585

betweenthetwopartnersadoptasimilarbehaviouraswiththe 586

All-atommodel,inthesensethatitisverylimitedinwaterwhile 587

0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 0.0 0.1 0.2 0.3 0.4 0.5 Dis tance dis tribuon funcon

Hydrogen -Acceptor distance (nm)

All-atom

Uncharged protein atoms mCD mCDa mCDh mPASA 0.00 0.01 0.02 0.03 0.04 0.0 10.0 20.0 30.0 40.0 50.0 Angle dis tribuon funcon

Donor -Hydrogen - Acceptor angle (°)

0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 0.0 0.1 0.2 0.3 0.4 0.5 Dis tance dis tribuon funcon

Hydrogen -Acceptor distance (nm)

0.00 0.01 0.02 0.03 0.04 0.0 10.0 20.0 30.0 40.0 50.0 Angle dis tribuon funcon

Donor -Hydrogen - Acceptor angle (°)

Fig.9.(Left)Distanceand(right)angledistributionfunctionsofthesolvatedsystemVps27UIM-1–Ubiquitincalculatedfromthelast10nsoftheAmber99SB-basedMD trajectoriesat300K,asobtainedusingtheAll-atom,mCD,mCDa,mCDh,mPASA,andneutralproteinatommodels.(Top)Protein-waterH-bonds,(bottom)intra-molecular H-bonds.(Foraninterpretationofthereferencestocolourinthe artwork,thereaderisreferredtothewebversionofthearticle).

(16)

Table5

Totalandinter-proteinenergyvalues(kJmol−1)oftheoptimizedconfigurationsofsystemVps27UIM-1–UbiquitininvacuumusedasstartingpointsforAmber99SB-based

MDsimulations.Subscripts‘1’and‘2’standforVps27UIM-1andUbiquitin,respectively.‘14’denotesinteractionsbetweenatomsseparatedby3chemicalbonds.

All-atom mCD mCDa mCDh mPASA Stretching 334.9 287.4 153.2 193.6 150.9 Bending 619.8 593.2 741.0 557.1 663.5 Torsion 3780.0 3774.0 3845.5 3773.7 3777.6 Improper 43.2 45.1 55.5 47.5 23.9 LJ-14 1823.4 1824.8 1574.0 1785.6 1471.7 Cb-14 17,025.5 20,693.4 16,909.2 20,670.8 – LJ −1651.8 −1631.1 −2473.6 −1774.7 −2773.3 Cb −26,887.7 −28,909.8 −26,792.9 −32,096.4 −2687.9 Cb12 −996.9 −1038.2 −1239.5 −1285.7 −932.7 LJ12 13.1 15.8 −130.0 −3.6 −155.4

itincreasesinvacuum(Table3).Asanexample,thereareno

H-588

bondsoccurringatt=20nsin thesolvatedsystem(Fig.5).The

589

apparentseparationofVps27UIM-1fromUbiquitindoeshowever

590

notcomewithadrasticchangeofthenumberofH-bondsformed

591

betweentheUIMandthesolvent(Table3).Ontheaverage,there

592

isanincreaseofonly5.5H-bonds,from84.5±4.6to90.0±5.0.

593

Indeed,theexpectedlargerincreaseofthenumberofH-bondsdue

594

totheconfigurationchangeiscompensatedbythedecreaseinthe

595

possibilitytoformH-bondsduetothereducedpointchargemodel.

596

Energyvaluesobtainedwiththedifferentpointchargemodels

597

arehardlycomparableonetoeachother.Wethereforechoseto

598

considerenergyratiossuchas|E12|/(E1+E2)and|Cb12|/(Cb1+Cb2)

599

to evaluate the proportion of the protein–protein interaction

600

energyversusintra-molecularenergyvalues(Table2).In water,

601

thecorrespondingenergyratios,2.2and1.0%respectively,indicate

602

alowerrelativeimportanceofthemCDinter-molecularpotential

603

energyversustheAll-atommodel,i.e.,4.6and2.2%,respectively,

604

with, however, a similar ratio [E12/(E1+E2)]/[Cb12/(Cb1+Cb2)].

605

Similartrendsareobserved invacuum.Toallowmore detailed

606

comparisonsbetweenenergycontributionsfromthevariouspoint

607

charge models, a detailed decomposition of the total potential

608

energywas achievedfor theoptimized initialstructures ofthe

609

complexinvacuumtoavoidanysolventcontribution(Table5).

610

Allenergytermsoftheseconformationallyclose3Dstructuresare

611

ofthesameordersofmagnitude,butthebondenergy islower

612

versustheall-atomcontribution,287.4versus334.9kJmol−1,and

613

theCoulombterminvolvingatomsseparatedbythreebonds,

Cb-614

14,ishigheranddestabilizing,20,693.4versus17,025.5kJmol−1.

615

Similarly to the All-atom model, the RMSF function clearly

616

emphasizesthegreatermobilityoftheVps27UMI-1endsversus

617

Ubiquitininwater(Fig.7).Ubiquitinitselfisalsoaffectedbythe

618

changeinthepointchargerepresentation.Thiscanbeshown,for

619

example,byananalysisofitsgyrationradiusrGwhich

progres-620

sivelyincreasesduringthefirst20nsproductionstage.Asalready

621

mentioned,anadditional20nssimulationwasperformedand

con-622

firmeda highervalueof thegyrationradiusrG forUbiquitin in

623

waterthaninvacuum,with1.36±0.02and1.15nm,respectively

624

(Table4).Asactuallyseenfurtherinthepaper,allrGvaluesobtained

625

inwaterarehigherthanthoseinvacuum,regardlessofthepoint

626

chargemodelused.ModelmCDleadstoanincreaseofrGbyabout

627

12%(from1.20to1.36nm),whilethechangeinvacuumis

imper-628

ceptible(1.15nmin bothcases)even ifthesecondary structure

629

elementsareaffected.Itindicatesthatthesolventmayserveasan

630

intermediateinthemodificationoftheproteinstructure.

631

Havingobservedthatthestructuralstabilityofthesystemis

632

modifiedwithmodelmCDversustheAll-atomrepresentation,

addi-633

tionalMDsimulationswereachievedatthreelowertemperatures,

634

i.e.,277, 250, and150K.The twolast temperaturevalues were

635

notselectedtoreflect aphysical stateforwater (theyare both

636

belowthefreezingpointofthesolvent)butwerechosentolocally

637

probethepotentialenergyhyper-surfaceofthesystem.The

anal-638

ysesofthe150Ktrajectoryclearlyshowstableproteinstructures,

639

as illustratedby the time evolution of the secondary structure 640

(Fig.10).Athighertemperaturevalues, a deconstructionofthe 641

secondarystructureelements,particularlythehelices,isobserved, 642

withaslowdownasthetemperaturedecreases.Simultaneously, 643

theRMSFoftheatomsofbothpartnersalsodecreasesand,at150K, 644

donotshowanymaximaatproteinends. 645

At the lowest temperatures, i.e., 250 and 150K, the time- 646

dependencyofrGforUbiquitinshowslittlefluctuationswithboth 647

anaveragevalueof1.20nm(Table4).Suchavalueisstrictlycom- 648

parabletothemeanAll-atomvalue,i.e.,1.20nm,obtainedat300K. 649

Contrarily,a contractionof Ubiquitin invacuum isobserved at 650

temperatureslowerthan300K.Indeed,meanvaluesof1.15and 651

1.12nmareobtainedat300and250K,respectively.Asthesec- 652

ondarystructureispreservedatlowtemperatures,oneassumes 653

thatthepotentialminimumoccurringattheall-atomlevelalso 654

existsforthemCDpointchargemodel.Thiscanalsobededuced 655

fromvacuumsimulationscarriedoutattemperaturesbelow300K. 656

Theanalysisofthesecondarystructureelementsinsuchconditions 657

showsaverystablestructure(Fig.10)withpreservedhelicesand␤- 658

strands,thatarehoweverslightlyshorterthaninthecorresponding 659

All-atomsimulation. 660

TheevolutionofthenumberofH-bondsformedbetweenVps27 661

UIM-1andUbiquitinisdescribedinTable3.Inwateraswellas 662

invacuum,thetrendisnotmonotonic,i.e.,thelowestnumberof 663

H-bonds,i.e.,0.6±0.7and1.0±0.9,isnotobservedatthelowest 664

temperaturebutat250and277K,respectively.Belowandbeyond 665

thosevalues,thenumbersincreaseextremelyfastinvacuumbut 666

smoothlyinwater.Itmightbedue,atlowertemperatures,toa 667

freezingofthestructurefavouringapersistenceoftheH-bonds 668

and,athighertemperatures,toanincreasedprobabilitytoform 669

polarcontactswithdiverseresiduesduetotheincreasedmobility 670

oftheatoms.OneclearlydistinguishesalossinpersistentH-bonds 671

when using modelmCD in water versusthecorresponding All- 672

atomcase. AllH-bondsarenow characterizedbyanoccurrence 673

degreebelow20%.Suchvaluesareincreasedonlywhenworking 674

invacuumand/orbyreducingthetemperature.Indeed,inwater 675

atT=150K,onefindthreetypesofH-bonds,i.e.,Leu73···Glu−259, 676

Gly47···Ser270,His+68···Glu−273.Thetwolastareidenticaltothe 677

All-atomH-bonds(SI8). 678

Studying thebehaviour of model mCDat low temperatures 679

allows tore-evaluate the H-bondratioscalculated from values 680

reportedinTable3.LetusfirstconsiderthatthemCDsystematthe 681

lowtemperatureof150Kprobessimilarconformationsthanthe 682

All-atomsystem.TheproportionofmainchainH-bondsrepresents 683

44.7%ofthetotalnumberofH-bonds.Thatvaluestayshigherthan 684

thecorrespondingAll-atomone,i.e.,36.6%.Thus,oneconcludesthat 685

thechangeinthepointchargemodelindeedleadstoanincreaseof 686

themainchainH-bonds,regardlessofachangeintheconformation. 687

Aspecificstudyoftheintra-molecularH-bondsoccurringinsol- 688

vatedUbiquitinshows thatatemperaturedecrease tendstoan 689

increaseinthenumberofsuchH-bonds,goingfromaveragesof 690

(17)

L.Leherte,D.P.Vercauteren/JournalofMolecularGraphicsandModellingxxx(2013)xxx–xxx 13

Fig.10.SecondarystructureofthecomplexVps27UIM-1–Ubiquitinobservedduringthelast20nsmCDAmber99SB-basedMDtrajectoriesinwater(top)andinvacuum (bottom)atvarioustemperatures.Vps27UIM-1andUbiquitininvolvethefirst24andlast76aminoacidresidues,respectively.Secondarystructureelementsare colour-codedasfollows:coil(white),␣-helix(blue),␲helix(purple),310helix(grey),␤-sheet(red),␤-bridge(black),bend(green),turn(yellow),chainseparation(lightgrey).(For

aninterpretationofthereferencestocolourinthe artwork,thereaderisreferredtothewebversionofthearticle).

and150K,respectively(Table6),i.e.,onecomescloserandcloserto

692

thevalueof48.0±3.2obtainedfortheAll-atommodel.Thetrend

693

isslightlydifferentforVps27UIM-1withaminimumnumberof

694

H-bonds,1.9±1.3,observedat277K.Asinterpretedearlierinthe

695

Table6

Averagenumbersofintra-molecularH-bondsoccurringinVps27UIM-1andin Ubiq-uitinasobtainedfromtheanalysisofthelast10nsoftheAmber99SB-basedMD trajectoriesat300K.

Vps27UIM-1 Ubiquitin

Solvated Vacuum Solvated Vacuum All-atom 16.2±2.8 28.4±2.1 48.0±3.2 81.8±3.7 All-atom-2 12.0±2.4 26.7±2.2 50.1±3.4 81.7±3.8 mCD 4.6±1.5 13.9±2.4 11.5±2.7 31.3±3.5 mCDa 2.5±1.3 13.7±2.2 14.3±3.3 42.6±3.7 mCDh 6.0±2.1 12.0±2.0 16.5±3.1 30.5±3.7 mPASA 3.8±2.0 5.3±1.7 14.5±3.6 19.7±3.3 mCD(277K) 1.9±1.3 13.0±2.3 16.0±2.8 34.9±3.7 mCD(250K) 5.6±1.6 11.8±2.2 23.3±3.1 41.5±3.6 mCD(150K) 11.5±1.4 14.2±1.9 32.9±2.7 39.4±3.2

paper,highernumbersofH-bondsobservedatlowertemperatures 696 mightbeduetoanincreasedpersistenceoftheH-bondswhile,at 697 highertemperatures,toanincreasedmobilityoftheatoms. 698 AstheproteinstructuresmodelledwithmCDreorganizeatroom 699 temperature,bothinwaterandinvacuum,oneconcludesthatthe 700 potentialhyper-surfacecorrespondingtothereducedpointcharge 701 modelis,atleastlocally,characterizedbylowerenergybarriers 702

thanwiththeall-atommodel[6]. 703

Inconclusion,themCDmodelallowstoobtainstableMDtra- 704

jectorieswithoutanyseparationofthecomplexpartners.Ithasa 705

contractingeffectinvacuumthatdoesnotoccurinwater,leading, 706

inthatlastcase,toamoremobilepeptidethanintheall-atomcase. 707

Theoverall3Dfoldingofthecomplexispreservedespeciallyforthe 708

largestandglobularpartner(thismaybepartlyduetothepresence 709

ofall-atomvdWcontributionstotheFF)whilethesecondarystruc- 710

tureissignificantlydismantledforthehelix-shapedVps27UIM-1 711

peptide.Thiseffectiscancelledatlowertemperatures,i.e.,atabout 712

250K,whichimpliesthatthechangeinthepointchargerepresen- 713

tationdoesnotaffectthelocationoftheenergyminimumonthe 714

Figure

Fig. 1. Point charge model for the 20 AA residues as established at s = 1.4 Bohr 2 from the hierarchical merging/clustering algorithm applied to the all-atom PASA ED function.
Fig. 3. MEP contours of (top) Glu−273 (−0.4 to −0.1 e − Bohr −1 ) and (bottom) His+68 (0.1 to 0.5 e − Bohr −1 ) as obtained using the All-atom, mCD, mCDa, and mPASA models.
Fig. 4. Secondary structure of the complex Vps27 UIM-1–Ubiquitin observed during the last 20 ns Amber99SB-based MD trajectories in water (left) and in vacuum (right) at 300 K, as obtained using the All-atom, All-atom-2, mCD, mCDa, mCDh, and mPASA models
Fig. 5. Final snapshots of the protein complex Vp27 UIM-1 (red)–Ubiquitin (blue) obtained from the last 20 ns Amber99SB-based MD trajectories in water (left) and in vacuum (right) at 300 K, as generated using the All-atom, All-atom-2, mCD, mCDa, mCDh, and
+4

Références

Documents relatifs

In a conventional design and manufacturing process, turbine blades are modeled based on reverse engineering or on parametric modeling with Computer Fluids Dynamics (CFD)

Since the frequency comb used for these experiments was already self-referenced using the standard technique of feedback to the optical pump of the gain medium [11], we were able

Dal momento che l'essenza del dibattito viene rivelata anche dalle scritture contabili patrimoniali proprie del sistema bancario, consideriamo contemporaneamente sia

In termini concreti: nei programmi e nella politica la DSC fa più spesso riferimento ai diritti umani, integra più sistematicamente i principi inerenti ai diritti umani, pone in

In ambito strettamente civile, la Svizzera – soprattutto in re- lazione alla sua dimensione demografica – è tra i paesi più generosi, sia dal punto di vista delle do-

Les charges fixes spécifiques sont diminuées des produits spécifiques, on obtient ainsi les charges fixes spécifiques nettes pour le spectacle. Les éléments

Bei erster Gelegenheit tauscht Hans den Klumpen gegen ein Pferd, wird jedoch prompt abgeworfen; tauscht das Pferd gegen eine Kuh, kann aber nicht melken; tauscht die Kuh gegen