inter-helical electrostatic interaction and rotameric ......structure was replaced using...

2015Vol. 1 No. 1: 2

iMedPub Journalshttp://www.imedpub.com

Research Article

DOI: 10.21767/2470-6973.100002

Chemical informaticsISSN 2470-6973

1© Under License of Creative Commons Attribution 3.0 License | This article is available in: www.cheminformatics.imedpub.com/archive.php

Olaposi I Omotuyi1,2 andHiroshi Ueda1,2

1 DepartmentofPharmacologyandTherapeuticInnovation,GraduateSchoolofBiomedicalSciences,NagasakiUniversity,Japan

2 CenterforDrugDiscoveryandTherapeuticInnovation,NagasakiUniversity,Japan

Corresponding author: HiroshiUeda

[email protected]

DepartmentofPharmacologyandTherapeuticInnovation,GraduateSchoolofBiomedicalSciences,NagasakiUniversity,Japan

Tel: +81958192421

Citation: OmotuyiOI,UedaH.Inter-helicalElectrostaticInteractionandRotamericSignaturesinActivatedSphingosine-1-PhosphateReceptor1.ChemInform.2015,1:1.

IntroductionSphingosine-1-phosphate receptor 1 (S1PR, also referred toas endothelial differentiation gene 1 (S1PR)) is one of the 5evolutionarily related G-protein-coupled receptors (GPCRs),which confer bioactivity on sphingosine 1-phosphate (S1P)[1]. Dysfunctional S1P/S1PR signaling has been implicated inexperimental autoimmune encephalomyelitis (EAE) and otherautoimmune neuro-inflammation phenotypes [2], thus lendingstrongevidence for theapproval (FDA-approved)offingolimod(FTY720)asanimmuno-modulatingdrugforthemanagementofrelapsing-remittingmultiplesclerosisviaS1PRmechanism[3].

Toward deeper understanding of ligand-S1PR interaction, twox-raystructures (PDB IDs:3V2W(3.35Å)and3V2Y(2.80Å))ofS1PRincomplexwithanantagonist(ML056)havebeendepositedinto theprotein data bank repository [4]. These structures did

provide the starting structure for the first detailed moleculardynamics simulation study of ligand binding and receptoractivationmechanisms[5,6].

A major highlight from the crystal structure of S1PR is thatinteractionbetween aromatic residues lining the ligand pocketand S1PR-active ligands drives ligand selectivity, activationand antagonism mechanisms [4]. Further mechanistic insightprovided bymicrosecondmolecular dynamics (MD) simulationstudyshowedthatwhenML056isreplacedwithS1Pwithintheorthostericsite,movementofS1Ptailcarbonatoms iscoupledtoflippingoftryptophan269(TM6.48)whichresonatesthroughproximal phenylalanine 265(TM6.44) thus, forcing a similardihedral change. Ultimately, a transmission switch is initiatedfrom these local events which propagates through asparagine63(1.50), aspartate 91(2.50), serine 304(7.46) and asparagine307(7.49)causinginfluxofwatermoleculesandlargemovementintransmembranehelices(thecytoplasmicregion)[5,6].

Inter-helical Electrostatic Interaction and Rotameric Signatures in Activated

Sphingosine-1-Phosphate Receptor 1

AbstractSphingosine-1 phosphate (S1P) receptor 1 (S1PR) is one of the receptorsresponsible for initiating intracellular signal transduction in response toextracellularsignalsencodedinSphingosine-1phosphate(S1P)andotherrelatedagonists. Using microsecond molecular dynamics simulation, rotameric (χ2)changes in the aromatic amino acids lining ligand-binding pocket, inter-helicalelectrostatic interaction, interaction between ligands (S1P and ML056) andselectedextracellularregionsofS1Phavebeenstudied.ThedatapresentedherestronglysuggestedthatS1P-boundS1PRstructurebecameactivebasedonNPxxY-motifrmsdanddissociationofTM3/TM6ioniclockasearlyas300nswhiletheapo- and ML056-bound S1PR were trapped in semi-active and inactive statesrespectively.Tyrosine29evolvedagonist-dependentrotamericdistributionwhiletryptophan117(W3.25),phenylalanine210(F5.47),tryptophan269(W6.48)andphenylalanine 273 (F6.52) exhibited activation-type signatures. Furthermore,whileactivationpromotedTM1/TM4,TM2/TM7andTM4/TM6engagement,priorelectrostaticengagementsinTM3/TM4,TM3/TM6andTM3/TM7weredissolved.N-terminal-heptahelicalbundleinteractionisalsocompromisedininactiveS1PRpossiblydueto reducedengagementofN-terminalresiduesuchaslysine34andlysine 46.Ultimately, S1PR activationby class I agonist followed classicalGPCRactivationparadigm.

Keywords: Sphingosine-1 phosphate (S1P) receptor; Rotameric signatures;Electrostaticinteraction

Received: June09,2015; Accepted: July13,2015;Published: July16,2016

2015Vol. 1 No. 1: 2

2 This article is available in: www.cheminformatics.imedpub.com/archive.php


In this study, we re-investigated the dynamics of Apo-S1P1,ML056-S1P1 (antagonist-bound) and S1P-S1P1 (agonist-bound)at 1.5 μs (doubling the previously reported production phasetimeof700ns)[5]inordertoprovidefurtherinsightintoS1P1activation mechanism with a focus on identifying aromaticdihedral signatures present in the N-terminal region whichrelaysligand-dependent(andindependent)actionsthroughthearomaticresiduesliningtheorthostericpocketofS1PR,andhowactivation of the receptor re-dictates electrostatic interactionbetweentransmembrane(TM)helices.

Materials and MethodsStarting StructuresThecrystalstructureofS1Preceptorresolvedat2.80Å(PDBID:3V2Y)wasretrievedfromtheproteindatabank(PDB)topreparethe starting co-ordinates. Themissing intracellular loop (ICL)-II(residues 149-155) and ICL-III (residues 232-244) in the crystalstructurewasreplacedusingloop-modellingsuitesinMODELLER(version 9v13) and ROSETTA as previously documented [5]. Inbothcases,thelowestDOPE-scoredloopwasselectedfromthepool of 5000 loop conformations generated using MODELLERsoftware,waspipedintotheROSETTAforrefinementasreported.Tomakethestartingapo-S1PRcoordinate,C-terminusresiduesbeyondhelixH8regionwasremoved.TogenerateML056-S1PRcoordinate,apo-S1PRandthex-raystructure(3V2Y)coordinateswere superimposedusing alignmodule in PyMOL (ThePyMOLMolecular Graphics System, Version 1.5.0.4 Schrodinger, LLC.),followed by the addition of ML056 coordinate from the x-raystructuretoapo-S1PRcoordinate.TogeneratestartingS1P-S1PRcoordinate,2DcoordinateofS1P(C18-Sphingosine1-phosphate,CID5283560)obtainedfromPUBCHEMdatabase[7]convertedintoalow-energy3DcoordinateusingLowModeMDconformationalsearch module in Molecular Operating Environment (MOE,Molecular Operating Environment (MOE), 2013.08; ChemicalComputing Group Inc., 1010 Sherbooke St. West, Suite #910,Montreal,QC,Canada,H3A2R7,2015).The3D-coordinatewasaligned with ML056 (reference, fromML056-S1PR) coordinateusingLigAlign[8]plugininPyMOL,thenewS1Pcoordinatewasaddedtoapo-S1PR.

Biosystems setupOrientation(alongthemembranenormal)ofthebiosystemswasperformed using the PPM server (opm.phar.umich.edu/server.php)[9].Eachoftheorientedbiosystemwasinsertedintoapre-equilibrated 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine(POPC, 68 lipids per leaflet) bilayer using CHARMM-GUIwebserver(www.charmm-gui.org)[10].Ligand(S1PandML056)parametization was performed using ParamChem service(https://cgenff.paramchem.org) as implemented on CHARMM-GUIwebserver. The biosystemswere solvated in TIP3P explicitwatermodelandneutralizedwithNa=/CL-(0.15M).

Molecular dynamics (MD) simulationAll molecular dynamics simulation (GROMACS, ver. 4.6) [11]was performed using CHARMM36 force field [12]. Duringequilibration,thebiosystemsweresubjectedtoconstantpressure

andtemperature(NPT;310K,1bar)conditionsasimplementedin Berendsen temperature and pressure coupling algorithms[13] as implemented inGROMACS. Van derWaals interactionswere estimated at 10 Å, long-range electrostatic interactionswere computed using particle mesh Ewald (PME) summationscheme [14] while equation of atomic motion was integratedusing the leap-frog algorithm [15] at 2 fs time step for a totaltimeof100nswithpositionalrestraints imposedontheheavyatomsinalldirections.Togettwo(n=2)startingcoordinatesfortheproductionphase,aninitial1nsunrestrainedsimulationwasperformed on each biosystem and the co-ordinates at 500 psand800pswereharvestedfromeachbiosystemforlong(1.5μs)productionphasesimulation.

Data AnalysisFordataanalysis,inbuiltanalysistoolkitinGROMACS[11]wasutilized.Inthisstudy,g_mindistmodule(forinteratomicminimumdistance analysis),g_distmodule (for estimating the center ofmassdistancebetweentwogroups),g_angle(forestimatingthechi2 (χ2) dihedral angles) andg_sham (data set for 3D energylandscape plots) were used. Graphs were generated usingMATHEMATICA (Wolfram Research, Inc, ver. 10.0, Champaign,Illinois,2014)statisticalsoftware(3Dplots)andgraphpadprism(GraphPadSoftware,SanDiegoCaliforniaUSA,www.graphpad.com)wasusedforgeneratinglineplotsandbarcharts.

Results and DiscussionStructural basis for ML056 and S1P actions on S1PRPhysiologically, GPCRs exist in multiple but dynamicconformational states spanning inactive and active spectrum[16]. GPCR-modulatory compounds are known to stabilizespecific conformations or ease the transition betweenconformations[17].Sphingosine-1-phosphatereceptor1(S1PR)isamemberofGPCRfamilyofproteinsandthebiologicaltargetofSphingosine-1-phosphate(S1P)andfingolimod[1,3,4,6,18].InordertofurtherunderstandthemechanismunderlyingS1PRactivation,firstwe investigated theconformationsstabilizedbySphingosine-1-phosphate and ML056 [4] in comparison withthe apo-S1PR biosystem. Whilst the starting structures wereessentially inactive as informed by TM3/TM6 center of massdistance ≈ 0.6 depicting an ionic lock [19] (Fig. 1A, upper barchart (I)), at ≈ 300 ns (Fig. 1A), apo-S1PR and sphingosine-1-phosphate-bound S1PR began to evolve active conformationswith broken TM3/TM6 ionic lock (transmembrane (TM)3/TM6distance>0.6)whichisakeyactivationsignature;[5]thebrokenioniclockwasmaintainedtill1.5μs(Fig.1A,upperbarchart(II)).Indeed,theaveragestructuresofproteinwithinthelast500nsofthesimulationsshowedthatTM3andTM6weredissociatedin apo- and sphingosine-1-phosphate-bound but not ML056-bound S1PR Figure 1B. TM3/TM6 ionic lock dissociation alonedoesnotfullyexplainactivationepisode inGPCRs;anotherkeyactivationsignatureisfoundintheconservedNPxxYmotif,wherewater tunneling into activated GPCR is under investigation [5,

2015Vol. 1 No. 1: 2

3© Under License of Creative Commons Attribution 3.0 License


20,21].ByprojectingthedatacollatedfromTM3/TM6distanceand rootmean square deviationofNPxxYmotif to an inactiveconformationalongthe3-Dfreeenergylandscape(Figure 1C, i, ii & iii),weprovidedfurtherinsightintometastableconformationalstates sampled by S1PR biosystems during the simulation. Theresult showed each set-up sampled different conformationsduring the simulation. In apo-S1PR (Figure 1C, i), sampledmetastableconformationshavebrokenTM3/TM6ioniclockbuttheNPxxYmotifwasessentially inactive (Cα-RMSD toNPxxY≤0.3 nm) whereas, ML056 bound EDG1 (Figure 1C, ii), whichparadoxicallyhaditsTM3/TM6trappedinaninactiveioniclockdidevolveactiveNPxxYmotifconformation(Cα-RMSDtoNPxxY>0.5nm). S1P-boundS1PR (Figure 1C, iii) surprisingly showedadynamicflowofstructuresmovingfrominactiveNPxxYmotifconformationtoactiveconformation (Cα-RMSDtoNPxxY>0.4nm),concomitantlywithbrokenTM3/TM6ioniclockbyremovingenergybarriers(ΔG≈0Kj/mol,seearrow).Clearly,forsuccessfulactivationofS1PR,activeNPxxYmotifconformation (Cα-RMSDtoNPxxY>0.4nm)mustbecoupledwithincreaseinTM3/TM6centerofmassdistanceandneitherofthetwoeventsalonemay

fully activate S1PR. Indeed, in rhodopsin, the conformationalchangesoccurringattheNPxxYmotifallowsTM7toinsertintothe3D-spacepreviouslyoccupiedbyTM6(whenionic lockwasformedwith TM3) thereby preventing the re-formation of theioniclockandultimatelystabilizingtheactivatedrhodopsin[22].Furthermore, some inactive GPCRs conformations have beencrystallizedwithbrokenTM3/TM6 ionic lock [23] similarly, inaseriesofexperimentsonHistamineH4receptor,mutantsthatlackionic lock-formingcapacity (R6.30A)didnotpromoteG-proteinactivation. In similar fashion, ionic lock-promoting mutant(R6.30E)didnotalteritsconstitutiveactivity;takentogether,theauthorsconcludedthatreversibleTM3/TM6ioniclockformationmaynotbeageneralrequirementforallclassAGPCRs[24].Inarecentreview[25],theauthorspointedtheabsenceofioniclockinβ2-AR/carazolol complex, and in subsetsofA2AAR,β1AR, andD3R[23,26].Therefore,datapresentedherefullydemonstratedthat apo- and ML056-bound S1PR sufficiently sampled semi-active (intermediate, broken ionic lock without accompanyingNPxxY motif conformational change) and inactive (activation-typeNPxxYwithintactioniclock,signatureofaninverseagonist)

Figure 1 Activation signatures in S1PR (A) IntracellularTM3 (residues140-145) /TM6 (residue250-256)centerofmassdistancewithtime;upperbarchartsi & iirepresentthebarchartrepresentationofTM3/TM6distance(mean±SEM)withinthefist50nsandlast500nsproductionphasesimulationsrespectively.(B)Representativesnapshots(averageoflast500nssimulation)showingthespatiallocationofTM3relativetoTM6intheaveragestructures(cartoonhelix)overthelast500ns(sphereonTM3representtheconservedR3.50,whilethesphereonTM6representS6.30).(C, i-iii)Freeenergysurfaceofthestructuressampledduringthesimulationsapo-S1PRandthoseincomplexwithantagonist(ML056)andagonist(S1P).Unlessotherwisestated,graphsrepresentthemeanplotoftwoindependentsimulations;blue=apo-S1PR,red=ML056-S1PRandgreen=S1P-S1PRbiosystems.

2015Vol. 1 No. 1: 2



states respectively while S1P-S1PR demonstrated active-statesignatures.Semi-activeconformationsampledbyapo-S1PRmayprovide a plausible explanation for the basal activity (agonist-indepedent activation) reported in S1PR [27] just as the dataprovidedfurtherinsightintothestructuralbasisforML056andS1Pantagonisticandagonisticactionsrespectively.

Rotameric signatures in aromatic amino acids lining S1P1 orthosteric pocketS1PRactivationsignaturesevokedbyS1PcanonlybeexplainedbyinteractionbetweenresiduesliningtheorthostericpocketandS1P.Undoubtedly,discretelylocalizedligand-residueinteractionsmustbesystematicallyordered intorobusttopologicalchangesin the heptahelical bundle, a key driver in G-protein couplingandactivationprocess[1,22]. Inthisstudy,adetailedstudyofinteractionbetweenS1PR-activeligands(agonistandantagonist)andthearomaticresiduesproximaltoML056intheoriginalcrystalstructure(PDBID3V2Y)[4]hasbeeninvestigatedintermsofsidechaindihedral(χ2)changesFigure 2A. InS1P-boundbiosystem,N-terminal tyrosine29(TYR-29,occurringat3.5Å fromML056phosphate-head group, inset, (Figure 2B, i) assumed a gauche(-)dihedral(0≤χ2≤120)whileinotherbiosystems(ML056-S1PRandapo-S1PR),transdihedralangle120≤χ2<180;−180≤χ2<−120,(Figure 2B, i)waspreferentially sampled.Although in previousstudy, a post-equilibration stable interaction was observedbetweenthisresidueandtheligands investigated[5],however,therotamericeffectsoftheligandsweredistinct.

Transmembrane-2tyrosine98(Y2.57)locatedat4.7Åfromthehead group ofML056 inset, (Figure 2B, ii) sampled gauche (-)dihedral (-120≤χ2<0) independent of biosystem set up but athigherpopulationinagonistboundS1PRcomparedwiththeotherbiosystems (Figure 2B, ii). The restraint on the dihedral spacesampledbytyrosine98inML056bound-S1PRmaybeexplainedbythepresenceofhydrogenbondbetweenthecarbonylgroupofML056andthephenolicgroupoftheresidueandinapo-state,formationofahydrogenbondwithserine304(7.46)[5].

Dihedralchangesintyrosine110(ECL-Iresiduelocatedat7.4Å from thehead groupofML056, inset, (Figure 2B, iii) providedthe first ligand-dependent dihedral event during observed inthis study; here, agonist and antagonist-bound biosystemspreferentially sampled gauche (-) dihedral and gauche (+)dihedral. When this data is interpreted within the context ofprevious findings that ECL-1 occludes the orthosteric site [4],tyrosine 110may therefore represent one of the hypotheticalligand-sensors,whichdrivesstructuralcompactnessofECL-1andtheN-terminal.

Based on the observation that S1P-bound S1PR evolvedactivation-type molecular and structural signatures [5] andthe apo-S1PR is in an intermediate state, then activation-type(distinct patternobservable in agonist- and apo-S1PR) dihedralchangewasassumedtobeobservedintryptophan117(W3.25,located5.0ÅfromtheheadgroupofML056,inset,Figure 2B,iv)wheresampledtrans(120≤χ2<180)dihedralspacewassampledinapo-S1PRandagonist-boundS1PRbiosystems(Figure 2B, iv),phenylalanine 125 (F3.33, located 3.5 Å from the tail carbonatoms of ML056, inset, Figure 2B, v) also sampled activation-

type gauche (-) and gauche (+) dihedral space (Figure 2B, v),phenylalanine 210 (5.47, located 4.5 Å from the tail carbonatomsofML056, inset,Figure 2B, vi), tryptophan269 (W6.48,located3.7ÅfromthetailcarbonatomsofML056,inset,Figure 2B,vii),andphenylalanine273(F6.52,located4.5ÅfromthetailcarbonatomsofML056,inset,Figure 2B,viii)(Figure 2B, vi, vii & viii).First,dihedralchangesinphenylalanine125,210,and273may have beenmediated by hydrophobic interaction betweenthe aromatic nuclei of the residues and the hydrophilic tail ofS1P [5],whichmaybe,enhanced inaromatic-moiety richclassII S1PR agonists [28] which do not depend on the traditionalarginine120(3.28)/glutamate121(3.29)forreceptoractivationasclassIagonists[4,5,27,28].Secondlyandmoreimportantly,the observed rotameric distribution pattern of tryptophan 269in agonist bound biosystem as presented here is similar to χ2angleassociatedwithtransmissionswitchduringS1PRactivationreportedinpreviousstudy[5]andotherclass-AGPCRs[5,22,29,30].

Perhaps,themostinterestingresultyetintheactivation-typeχ2dihedralserieswasobservedintheproline(proline308,P7.50)residue of NPxxY motif. In most GPCRs [5, 22, 29], rotamericchangesintryptophan269(transmissionswitch)occurmutuallyinclusiveof largestructuralrearrangementinTM7mediatedbytheprolinecomponentofNPxxYmotif.Inthisstudy,proline308in agonist- and antagonist-bound S1PR preferentially sampledgauche (+) χ2 (-120≤χ2<0) dihedral space while trapped ingauche (-) χ2 (0≤χ2<120) dihedral space in semi-active S1PR(apo-state); as observed in the 3D-free energy surface plot.Thisdata indicated thatNPxxYkinkmightnotbecoupledwiththe movement of tyrosine 311 (307-NPxxY-311) (Y7.53) intothe heptahelical bundle (required formaintaining GPCRs in anactivestate).Similarly,thedatalendavaluablecredencetotheproposition that GPCRs conformationmay not be explained intwostate(ON/OFFforactivatedandinactivatestates),rather,amore acceptablemulti-state “molecular rheostats-like” state incontinuumconformationdynamicsseparatedinspaceandtimebyenergyhasbeenproposed[22,25].

Energetics of transmembrane helix interaction and ligand-S1PR bindingUltimately, the local events at GPCR transmembrane helicessumuptocauselargechangesintheoverallreceptortopologytherebydrivingactivationandinaction[4,5,20,22,25,29].Sinceearlier in this study, we have provided some evidence, whichsuggestedthatapo-andagonistboundS1PRshowedactivation-type structural arrangement and dihedral signatures, next, wesought to elucidate the energy of interaction between specificTMhelicesinthedistinctconformationalstatesidentifiedinthisstudy (active (agonist bound), semi-active (intermediate, apo-state) and inactive (antagonist bound) states). Data presentedheresuggestedthatduringS1PRactivation(semi-activationandfull-activation), electrostatic interaction is essential betweenTM1(residue44-70)andTM4(residue170-177);suchinteractionseemslost in inactivatedS1PR(Figure 3A, i). Insimilar fashion,the intensity of TM2 (residue 83-102) / TM7 (residue 293-312) interaction showed correlation with receptor activation.Interesting, starting at ≈ 300 ns, the three biosystems began

2015Vol. 1 No. 1: 2



to evolve a distinct pattern (Fig. 3A, ii) as fully active (agonistbound)S1PRinteractionenergy≈-270Kj/moltilltheendofthesimulation,the interactionenergyof thesemiactivebiosystemwas tightly maintained at -200 Kj/mol while the antagonistboundbiosystemevolvedthe least intense interactionwiththeaverageenergyvalueof-110Kj/mol (Figure 3A, iii).Additionally,a weak interaction between TM4/TM6 (residue 252-278) mayalso contribute to S1PR activation (Figure 3A, iii). In contrast,repulsioninTM3/TM4(-75vs.-120Kj/molforactiveandinactiverespectively,Figure 3A,iv),TM3(residue117-144)/TM6(-40vs. -100Kj/molforactiveandinactiverespectively,Figure 3A,v)andTM3/TM7(10vs.-12Kj/molforactiveandinactiverespectively,Figure 3A,vi)mayberequiredforS1PRactivation.WhereasTM3/TM6repulsionisthemostcommonlydocumentedstructuralforactivationinalmostallknownGPCRs[4,5,20,22,25,29,30,31],this study reports that electrostatic interaction was also lostbetween TM3/TM4and TM3/TM7during S1PR activation. It isinteresting to note that disruption in salt-bridge (electrostaticinteraction)betweenTM3andTM7reportedlycauseconstitutiveactivationinagreementwithourfindings[31,32]andthatlarge

repulsion between these helices are counter-balanced withattractions in other TM helices as observed n TM2/TM7 andTM4/TM6.

ThecontributionoftransmembraneinteractiontoS1PRactivationispresentedintheschematicdiagram (Figure 3B)andtheaveragestructureofthethreebiosystemswithinthelast500nsshowedalargedisplacementbetweenTM3andTM4(Figure 3C).Intermsof binding affinity, sphingosine-1-phosphate showed two-foldhigher energy of interactionwith S1PR comparedwithML056(-375.16vs.-156.13Kj/mol)andthedifferenceinaffinitymaybetracedtoitshigherinteractionwithlysine46(K1.33),lysine111(ECL-II), arginine120 (R3.28), lysine200 (K5.37)and lysine285(ECL-III)(Figure 3D).

Ligand binding causes heptahelical bundle packing with extracellular loops and the N-terminal region. In the crystal structure (PDB ID: 3V2Y), N-terminal and theextracellular loop (ECL) (ECL-I is green, ECL-II is blue, ECL-III is

Figure 2 Rotameric signatures in aromatic amino acids lining S1P1 orthosteric pocket (A) Surface representation(green)of thedistributionof thearomaticaminoacids liningML056bindingpocket in theoriginal crystalstructure(PDBID:3V2Y).(B)χ2dihedralangledistributionoftyrosine29(i)tyrosine98(ii)tyrosine101(iii)tryptophan117(iv)phenylalanine125(v)phenylalanine210(vi)tryptophan269(vii)andphenylalanine273(viii)whichrepresentkeyaromaticresidues inS1PRorthosteric site. (ix) χ2dihedraldistributionofNPxxYmotif-proline308.Unlessotherwisestated,graphsrepresentthemeanplotoftwoindependentsimulations.Theinsetsrepresentthepositionofeacharomaticaminoacidfromtheligandintheoriginalx-raystructure(thedistanceofseparationisgiveninÅ).Blue=apo-S1PR,red=ML056-S1PRandgreen=S1P-S1PRbiosystems.

2015Vol. 1 No. 1: 2



Apo-S1PRS1PR-ANTS1PR-AGO

TM3 TM

4

TM3 TM

4

TM3 TM

4

Apo-S1PR S1PR-ANT S1PR-AGO

A

0.0 0.5 1.0 1.5-60

-50

-40

-30

-20

Time ( μs)

TM1-

4 (K

j/m

ol)

0.0 0.5 1.0 1.5-300

-250

-200

-150

-100

Time ( μs)

TM2-

7 (K

j/m

ol)

0.0 0.5 1.0 1.5-100

-80

-60

-40

-20

0

Time ( μs)

TM3-

6 (K

j/m

ol)

0.0 0.5 1.0 1.5-30

-20

-10

0

10

Time ( μs)

TM3-

7 (K

j/m

ol)

0.0 0.5 1.0 1.520

25

30

35

Time ( μs)

TM4-

6 (K

j/m

ol)

0.0 0.5 1.0 1.5-150

-100

-50

Time ( μs)

TM3-

4 (K

j/m

ol)

i ii

iii iv

v vi

B

C

20 40 60 80 100 120 140 160 180 200 220 240 260 280 300 320

-40

-30

-20

-10

Lig

and

-S1P

R1

inte

ract

ion

ener

gy

(Kj/

mol

)

S1PR1 residues

D120

K111K46

K200K285

-375.16 (9.09)-156.13 (6.08)

S1PR-AGOS1PR-ANT

R120

D

Figure 3 Energetics of transmembrane helix interaction and ligand-S1PR binding (A) Time-dependent changes inelectrostaticinteractionbetween(i)TM1/TM4(ii)TM2/TM7(iii)TM4/TM6(iv)TM3/TM4(v)TM3/TM6(vi)TM3/TM7(B) Schematic representation of electrostatic interactions in S1PR helices in active and inactivestates.(C)RepresentativesnapshotsofaverageS1PRstructureofthethreebiosystemsoverthelast500ns.(D)DecomposedfreeenergyprofileshowingtherelativecontributionofselectedS1PRaminoacidstoML056and S1Pbinding, estimated free energyof S1PR-ligandbinding is shownasmean (SEM).Unless otherwisestated,graphsrepresentthemeanplotoftwoindependentsimulations.Blue=apo-S1PR,red=ML056-S1PRandgreen=S1P-S1PRbiosystems.

cyanandN-terminalisyellow)regionsobservablypackedcloselywith the TM helices (not shown) while trapping the ligand(ML056,pinksphere)(Figure 4A) intotheorthostericsite[4,5].To understand how the two ligands differentially interact with

theN-terminalregion,theleastdistanceseparatingeachoftheligands from tyrosine 29 and lysine 34 were calculated. Bothligandswerelocatedatameandistanceof0.2nmfromtyrosine29 (Figure 4B, i)throughoutthesimulation;thisisconsistentwith

2015Vol. 1 No. 1: 2



thefindingsofYuanetal.,[5]thatbothML056andS1Pformedstable hydrogen bond interaction with tyrosine 98. ComparedwithML056,sphingosine-1-phosphateshowedhigherresidencewith lysine34atmeandistanceof0.2nm(ML056=0.7nm,Fig.4B,ii)asreportedpreviously[4,5].ECL-Iresidueserine105maydictatetightpackingwithML056and sphingosine-1-phosphateasbothligandswerespacedat0.2and0.3nmfromtheresidue(Figure 4B, iii). Similar observation was made for ECL-II asrepresentative residue showed 0.4 nm (mean distance fromvaline 194, ECL-II) separation from the ligands (Figure 4B, iv). Furthermore,thenet(valuesforstructuresgeneratedalongthetrajectoriesvaluecalculatedforcrystalstructure,PDBID:3V2Y)center ofmass distance of separation between theN-terminaland the heptahelical bundle (with the removal of extracellularloopresiduesfromthecalculation)wasalsocalculated.Clearly,the data indicated that N-terminal (N1-40) was tightly packedwith theheptahelicalbundle inactivebutnot in inactiveS1PR(Figure 4B, v).Finally,itwasobservedthatbothligandsdidnotexhibitanyobservabledifferenceintermsoftheirdistancefromtheactivesiteresidues(arginine120&glutamate121)[4](Figure 4B, vi & vii).

ConclusionThe high-resolution S1PR structure resolved and deposited inproteindatabankrepository[4]haschangedourunderstandingof lipid-type ligand recognition, binding, and activationof Edg family ofGPCRs forever (as S1PR is the first familyof this class). Following the success in crystallography,

S1PR activation details were provided using microsecondmolecular dynamics simulation experiments by Yuan etal [5, 20]. The major achievement of these studies wasitemizing theprecisecontributionofkeymoleculareventsin the build up to S1PR activation, and “putting paid to”thedoubtsastowhetherintra-helicalwatermoleculesarenecessary for GPCR activation. Here, we have explored afewareasnotaddressedinthestudysuchas:exploringthedihedral plasticity in aromatic residues lining the ligand-binding pocket. Interestingly, these residues displayedvery distinct rotameric patterns in S1P-, ML056- and apoS1PR biosystems. Because these aromatic residues arecontributed by N-terminal, extracellular loops, and theheptahelical bundle, ligand-specific and activation typerotameric signatures have been suggested. This studyalso identified that fully active, intermediate and inactiveconformationsofS1PRareaccompaniedbydistinguishableinter-helical electrostatic interaction. While activationpromotedTM1/TM4,TM2/TM7andTM4/TM6engagement,priorelectrostaticengagementsinTM3/TM4,TM3/TM6andTM3/TM7 were dissolved. Ultimately, S1PR activation byclassIagonistfollowedclassicalGPCRactivationparadigm.

Declaration of interestThis work was supported by Platform for Drug Discovery,Informatics, and Structural Life Science from the Ministry ofEducation,Culture, Sports, Science andTechnology, Japan. Theauthorsdeclarenoconflictofinterest.

2015Vol. 1 No. 1: 2



Figure 4 Ligand interaction with extracellular loops and the N-terminal region and role in heptahelical bundle packing (A) RepresentationoftheN-terminal(yellowsurface),extracellularloop-I(greensurface),extracellularloop-II(bluesurface),extracellularloop-III(cyansurface)andtheligand(pinkspheres).Themeanminimumdistancebetweentheligandandtyrosine29(B)(i) lysine34(ii)serine105(iii)valine194(iv)netdistancebetweentheN-terminal and theheptahelical bundle in thedirectionof themembranenormal (v)meanminimumdistancebetweenthe ligandandarginine120 (vi)andglutamate121 (vii)Unlessotherwisestated,graphsrepresentthemeanplotoftwoindependentsimulations.Blue=apo-S1PR,red=ML056-S1PRandgreen=S1P-S1PRbiosystems.

2015Vol. 1 No. 1: 2



References1 Rosen H, Gonzalez-Cabrera PJ, Sanna MG, Brown S (2009)

Sphingosine 1-phosphate receptor signaling. Annual review ofbiochemistry78:743-768.

2 GarrisCS,WuL,AcharyaS,AracA,BlahoVA,etal.(2013)Defectivesphingosine 1-phosphate receptor 1 (S1P1) phosphorylationexacerbates TH17-mediated autoimmune neuroinflammation.Natureimmunology14:1166-1172.

3 Brinkmann V (2009) FTY720 (fingolimod) in Multiple Sclerosis:therapeuticeffectsintheimmuneandthecentralnervoussystem.Britishjournalofpharmacology158:1173-1182.

4 Hanson MA, Roth CB, Jo E, Griffith MT, Scott FL (2012) CrystalstructureofalipidGprotein-coupledreceptor.Science335:851-855.

5 YuanS,WuR,LatekD,TrzaskowskiB,FilipekS(2013)LipidreceptorS1P(1) activation scheme concluded from microsecond all-atommolecular dynamics simulations. PLOS computational biology 9:e1003261.

6 Hurst DP, Schmeisser M, Reggio PH (2013) Endogenous lipidactivatedGprotein-coupledreceptors:emergingstructuralfeaturesfromcrystallographyandmoleculardynamicssimulations.Chemistryandphysicsoflipids169:46-56.

7 Wang Y, Xiao J, Suzek TO, Zhang J, Wang J, Bryant SH (2009)PubChem:apublicinformationsystemforanalyzingbioactivitiesofsmallmolecules.Nucleicacidsresearch37:W623-633.

8 HeifetsA,LilienRH(2010)LigAlign:flexibleligand-basedactivesitealignmentandanalysis. Journalofmoleculargraphics&modelling29:93-101.

9 LomizeMA,PogozhevaID,JooH,MosbergHI,LomizeAL(2012)OPMdatabaseandPPMwebserver:resourcesforpositioningofproteinsinmembranes.NucleicAcidsRes40:D370-376.

10 MillerBT,SinghRP,Klauda JB,HodoscekM,BrooksBR,WoodcockHL3rd(2008)CHARMMing:anew,flexiblewebportalforCHARMM.Journalofchemicalinformationandmodeling48:1920-1929.

11 PronkS,PallS,SchulzR,LarssonP,BjelkmarP,etal.(2013)GROMACS4.5: a high-throughput and highly parallel open sourcemolecularsimulationtoolkit.Bioinformatics29:845-854.

12 HuangJ,MacKerellADJr(2013)CHARMM36all-atomadditiveproteinforce field: validationbasedon comparison toNMRdata. Journal ofcomputationalchemistry34:2135-2145.

13 Eslami H, Mojahedi F, Moghadasi J (2010) Molecular dynamicssimulationwithweakcouplingtoheatandmaterialbaths.TheJournalofchemicalphysics133:084105.

14 DardenT,PereraL,LiL,PedersenL(1999)Newtricksformodelersfromthecrystallographytoolkit: theparticlemeshEwaldalgorithmand itsuseinnucleicacidsimulations.Structure7:R55-R60.

15 CuendetMA,VanGunsterenWF(2007)Onthecalculationofvelocity-dependent properties in molecular dynamics simulations using theleapfrog integration algorithm. The Journal of chemical physics 127:184102.

16 ParkPS,LodowskiDT,PalczewskiK(2008)ActivationofGprotein-coupledreceptors: beyond two-state models and tertiary conformationalchanges.Annualreviewofpharmacologyandtoxicology48:107-141.

17 NygaardR,ZouY,DrorRO,MildorfTJ,ArlowDH,etal(2013)Thedynamicprocessofbeta(2)-adrenergicreceptoractivation.Cell152:532-542.

18 ChibaK,Adachi k (2012)Discoveryoffingolimod, the sphingosine1-phosphatereceptormodulatoranditsapplicationforthetherapyofmultiplesclerosis.Futuremedchem4:771-781.

19 XieXQ,ChowdhuryA (2013)Advances inmethods tocharacterizeligand-inducedioniclockandrotamertogglemolecularswitchinGprotein-coupledreceptors.MethEnzymol520:153-174.

20 YuanS,FilipekS,PalczewskiK,VogelH(2014)ActivationofG-protein-coupled receptors correlates with the formation of a continuousinternalwaterpathway.Natcommun5:4733.

21 FritzeO,FilipekS,KuksaV,PalczewskiK,HofmannKP,etal.(2003)RoleoftheconservedNPxxY(x)5,6Fmotif intherhodopsingroundstateandduringactivation,ProceedingsoftheNationalAcademyofSciencesoftheUnitedStatesofAmerica,100:2290-2295.

22 RosenbaumDM,RasmussenSG,KobilkaBK(2009)ThestructureandfunctionofG-protein-coupledreceptors.Nature459:356-363.

23 MoukhametzianovR,WarneT,EdwardsPC,Serrano-VegaMJ,LeslieAG, ET AL. (2011) Two distinct conformations of helix 6 observedin antagonist-bound structures of a beta1-adrenergic receptor.Proceedings of the National Academy of Sciences of the UnitedStatesofAmerica108:8228-8232.

24 SchneiderEH,SchnellD,StrasserA,DoveS,SeifertR(2012)ImpactoftheDRYmotifandthemissing"ioniclock"onconstitutiveactivityandG-proteincouplingofthehumanhistamineH4receptor.TheJournalofpharmacologyandexperimentaltherapeutics333:382-392.

25 Audet M, Bouvier M (2012) Restructuring G-protein- coupledreceptoractivation.Cell151:14-23.

26 DoreAS,RobertsonN,ErreyJC,NgI,HollensteinK(2011)StructureoftheadenosineA(2A)receptorincomplexwithZM241385andthexanthinesXACandcaffeine.Structure19:1283-1293.

27 WindhRT,LeeMJ,HlaT,AnS,BarrAJ,ManningDR(1999)Differentialcouplingofthesphingosine1-phosphatereceptorsEdg-1,Edg-3,andH218/Edg-5totheG(i),G(q),andG(12)familiesofheterotrimericGproteins.TheJournalofbiologicalchemistry274:27351-27358.

28 Gonzalez-CabreraPJ, Jo E, SannaMG,BrownS, LeafN (2008) Fullpharmacological efficacy of a novel S1P1 agonist that does notrequire S1P-like headgroup interactions. Molecular pharmacology74:1308-1318.

29 TrzaskowskiB,LatekD,YuanS,GhoshdastiderU,DebinskiA(2012)ActionofmolecularswitchesinGPCRs--theoreticalandexperimentalstudies.Currentmedicinalchemistry19:1090-1109.

30 Shi L, Liapakis G, Xu R, Guarnieri F, Ballesteros JA, et al. (2002)Beta2 adrenergic receptor activation. Modulation of the prolinekinkintransmembrane6byarotamertoggleswitch.TheJournalofbiologicalchemistry277:40989-40996.

31 Robinson PR, Cohen GB, Zhukovsky EA, Oprian DD (1992)Constitutivelyactivemutantsofrhodopsin.Neuron9:719-725.

32 NikiforovichGV,MarshallGR (2006) 3Dmodelingof the activatedstatesofconstitutivelyactivemutantsofrhodopsin.Biochemicalandbiophysicalresearchcommunications345:430-437.

inter-helical electrostatic interaction and rotameric ......structure was replaced using...

Documents