journal homepage:www.elsevier.com/locate/febsopenbio
Structural and thermodynamic insight into phenylalanine hydroxylase from the hu- man pathogen Legionella pneumophila 夽
Hanna-Kirsti S. Leiros
a,*, Marte Innselset Flydal
b, Aurora Martinez
baThe Norwegian Structural Biology Centre (NorStruct), Department of Chemistry, University of Tromsø, N-9037 Tromsø, Norway
bDepartment of Biomedicine, University of Bergen, Jonas Lies vei 91, 5009 Bergen, Norway
a rt i c l e i n f o
Article history:
Received 25 June 2013
Received in revised form 12 August 2013 Accepted 12 August 2013
Keywords:
Pyomelanin synthesis Thermostability Substrate specificity Aggregation
Salt induced aggregation Pathogen
a b s t r a c t
PhenylalaninehydroxylasefromLegionellapneumophila(lpPAH)hasamajorfunctionalroleinthe synthesisofthepigmentpyomelanin,whichisapotentialvirulencefactor.Wepresentherethecrystal structureoflpPAH,whichisadimericenzymethatshowshighthermostability,withamidpointdenat- urationtemperatureof79◦C,andlowsubstrateaffinity.Thestructurerevealedadimerizationmotif thatincludesionicinteractionsandahydrophobiccore,composedofbothβ-structureandaC-terminal region,withthespecificresidues(P255,P256,Y257andF258)interactingwiththesameresiduesfrom theadjacentsubunitwithinthedimer.Thisuniquedimerizationinterface,togetherwithanumberof aromaticclusters,appearstocontributetothehighthermalstabilityoflpPAH.Thecrystalstructure alsoexplainstheincreasedaggregationoftheenzymeinthepresenceofsalt.Moreover,thelowaffinity forsubstrateL-Phecouldbeexplainedfromthreeconsecutiveglycineresidues(G181,182,183)located atthesubstrate-bindingsite.ThisisthefirststructureofadimericbacterialPAHandprovidesaframe- workforinterpretingthemolecularandkineticpropertiesoflpPAHandforfurtherinvestigatingthe regulationoftheenzyme.
C 2013TheAuthors.PublishedbyElsevierB.V.onbehalfofFederationofEuropeanBiochemical Societies.Allrightsreserved.
1. Introduction
Thegram-negativebacteriumLegionellapneumophilabelongsto theγ-proteobacteria. Innature,L.pneumophilaisan inhabitantof warm freshwater habitats where its multiplication is mainly re- strictedtointracellularnichesinsideamoebalhostsandafterinfec- tionitcontinuesthisintracellularlife-stylewithinthehumanhostby multiplyinginsidemacrophages[1].Humansareadead-endhostfor thispathogen,butitstillcausesoutbreaksofLegionnaires’disease, apotentiallyfatalformofpneumonia,whenitmultipliesinwarm, stagnantwaterthatarespreadinaerosolsthroughhuman-madein- stallationssuchasfountainsandshowers[2–4].Insuchaqueousen- vironmentsL.pneumophilagrowswellattemperaturesintherange 20–48◦C,butitstolerance tohighertemperaturescanleadtore- growthafterheat-disinfection[5].Thenumberofreportedcasesof Legionnaires’disease hasincreasedinthelastdecade[6],making eradication ofthe bacteriumfrominfection sourcesan important task.Thus,thereisaneedforresearchonpossibletargets,notablyon putativevirulencefactors.
夽This is an open-access article distributed under the terms of the Creative Com- mons Attribution-NonCommercial-No Derivative Works License, which permits non- commercial use, distribution, and reproduction in any medium, provided the original author and source are credited.
* Corresponding author. Tel.: + 47 77 64 57 06.
E-mail address: [email protected] (H.-K.S. Leiros).
L.pneumophilaisoneofseveralknownpathogenicbacteriawhose genomesencodeaphenylalaninehydroxylase(PAH).PAHistheen- zymethatcatalyzesthehydroxylationofl-phenylalanine(l-Phe)tol- tyrosine(l-Tyr)usingnon-hemeFe(II)andtetrahydrobiopterin(BH4) ascofactors,anddioxygenasadditionalsubstrate[7].Inmammals, PAHhasamajorcatabolicroleanditsdysfunctionisassociatedto deleterioushyperphenylalaninemia andphenylketonuria, whereas aswehaverecentlyshown[8]L.pneumophilaPAH(lpPAH)hasan importantroleinthegrowthofthebacteriuminmediadeficientin l-Tyrandinthesynthesisofabrownishpigmentcalledpyomelanin.
LikesomeoftheotherbacteriathatencodeaPAH,L.pneumophila producesandexcretespyomelaninwhenl-Pheorl-Tyrisavailable [9].Pyomelaninisproducedviathecatabolismofl-Phe/l-Tyrwhen an intermediateofthepathway, homogentisate,accumulatesand subsequentlyautooxidisesandpolymerises.InL. pneumophila,the pyomelaninhasferricreductaseproperties,scavengingandreducing extracellulariron[10].Ironisanessentialmetalforbothhumanhost andbacterialpathogens,amongotherascofactorformanyenzymes, includingPAH.Ironisindeedcriticalforintracellularinfectionby L.pneumophila[11]andproductionofpyomelaninisoneofseveral strategiesforironassimilation[10].Molecularandenzymekinetic characterization hasshownthat lpPAHiswellsuitedforcatalysis athumanbodytemperature,revealingamaximumactivityaround 40◦Candveryhighspecificactivitycomparedtoothereukaryoteand
2211-5463/ $36.00 c2013 The Authors. Published by Elsevier B.V. on behalf of Federation of European Biochemical Societies. All rights reserved.
http://dx.doi.org/10.1016/j.fob.2013.08.006
prokaryotePAHs[8,12–16].TheenzymealsopresentedhighKmval- uesforsubstrateandcofactorBH4(735±50μMand125±25μM, respectively)[8],andthelowaffinityforl-Phehasbeeninterpreted asaplausibleregulatorymechanismtopreservethresholdamounts ofthesubstrate[7].SimilarlytoallotherPAHs,theactivityoflpPAH istotallydependentuponanon-hemeferrousion,but– incontrastto allotherstudiedPAHswhichareiron-boundtetramers(eukaryotes) ormonomers(bacteria)[7,15,17]– lpPAHisisolatedasaniron-free (apo)homodimerwhenexpressedrecombinantly[8].Theestimated hydrodynamicdiameterofapo-lpPAHdimer(8.1 ± 0.1nm)andits surprisinglyhighthermalstability(midpointdenaturationtempera- ture(Tm)=79±0.5◦C)isonlyslightlyincreasedbyFeincorporation [8].
ThehighthermalstabilityoflpPAHmightbeimportantforpreser- vationoftheenzymeattemperatureswhereL.pneumophilasurvives inadormant,non-replicativestate[18]and,basedonacrucialroleof theenzymeinthesynthesisofpyomelanin,wesetofftocharacterize thestructure–function–stabilityrelationshipsindimericlpPAH.The recombinantenzymewascrystallizedanditsstructurehasbeenan- alyzedandcomparedwiththatofPAHsfrommesophilicChromobac- teriumviolaceum(cvPAH;optimumtemperatureforgrowth,25◦C) andpsychrophilicColwelliapsychrerythraea34H(cpPAH;optimum temperatureforgrowth,10◦C)bacteria,aswellaswiththehuman enzyme(hPAH;optimumtemperature,37◦C).Thestructuralfeatures identifyauniquedimerizationmotifandaidtoexplaintheparticular structuraldeterminantsforthermaladaptationoflpPAH,aswellas itslowaffinityforl-PheandBH4[8].
2. Resultsanddiscussion
2.1. Refinementandoverallstructure
Thecrystallizationof lpPAHwaschallengingsincethe crystals grewasplateswithsomedisorderfoundinthefinalmodels,and duetothelowsymmetryspacegroupP21 adecentrotation range ofX-raydata wasneededto obtainmorethat 90%completeness.
Thebestcrystaldiffracted to 2.5 ˚A(Table1) andin theobserved electrondensitymapsthedimericlpPAHstructureisclearlydefined, inparticularforchainsAandB.Thus,thecurrentmodelgivesgood insightintothelpPAHstructureasdescribedbelow.
ThefinallpPAHmodelrefinedtoanR-factorof27.6%andanR-free valueof30.2%(Table1),whichareslightlyhighbutcanbeexplained bythedisorderfoundforchainsCandD.ThefinallpPAHmodelis mostcompleteforchainsAandB,includingresiduesVal8-Asp259 (totally252residuesforbothchains),withlowestmeanB-factors, whereaschainsC(232residues)andD(237residues)havehigher meanB-valuesandmanydisordered residuesleftout ofthefinal model(Table1,Fig.1a).Theobserveddifferencesbetweenthechains canbeexplainedfromthesymmetrycontactsthatchainsAandB makewithone another,whereaschainsCandDare facingwater channelswithlesssymmetry-relatedprotein–proteininteractions.
IntheactivesiteoflpPAHextradifferenceelectrondensitywasob- servedandinterpretedasthreepolyethyleneglycol(PEG;C4H10O3) molecules(chainsAandB)(Fig.2a).
Thecatalyticdomainofothernon-hemeiron-andBH4-dependent aromaticaminoacidhydroxylases (AAAHs)hasa mixedα/βfold, alsofoundforthereportedlpPAHstructure.Therootmeansquare deviation(RMSD;www.ebi.ac.uk/msd-srv/ssm/)oflpPAHcompared tootherstructuresare:1.2 ˚AtocpPAH (bothPDBs2v27,2v28for 252CA-atoms),1.3 ˚Ato cvPAH(bothPDB1LTZ/1LTUfor233/232 CA-atoms)and1.5 ˚AtohPAH-BH4(PDB1J8Ufor193CA-atoms).
2.2. Dimerizationandintersubunitinteractions
InagreementwiththedimericnatureoflpPAHinsolution[8]there aretwodimersintheasymmetricunit,formedbyeitherchainsAand
Table 1
X-ray data collection and refinement statistics for lpPAH.
X-ray statistics lpPAH
PDB entry 4BPT
Beamline Bessy, BL14.1
Space group P 2 1
Unit cell a = 90.32 ˚A
b = 60.12 ˚A c = 124.04 ˚A β= 94.07 ◦
Resolution ( ˚A) 25–2.5
(highest bin) (2.64–2.5)
Wavelength ( ˚A) 0.91841
No. of unique reflections 41 892 (6 267)
Multiplicity 2.3 (2.3)
Completeness (%) 90.9 (94.0)
Mean ( < I >/<σI> ) 8.6 (2.2)
R -sym (%) a 8.4 (39.4)
Wilson B -factor ( ˚A 2) 41.8 Refinement
Resolution ( ˚A) 10–2.5
R -factor (all reflections) (%) a 27.64
R -free (%) a 30.20
No. of atoms 8156
No. of water molecules 162 No. of other molecules 6 PEG
No of residues chain A / B / C / D 252 / 252 / 232 / 237 R.m.s.d. bond lengths ( ˚A) 0.018
R.m.s.d. bond angles ( ◦) 2.08 Average B -factor ( ˚A 2)
All atoms 47.2
Protein (chain A / B / C / D) 29.8 / 32.2 / 67.8 / 63.0 PEG / Water molecules 26.2 / 47.4 Ramachandran plot:
Most favored regions (%) 91.1 Additionally allowed regions (%) 6.4 Disallowed regions (%) 2.5
aR sym = ( h i| I i( h ) − I ( h ) |) / ( h II ( h )), where I i( h ) is the i th measurement of reflection h and < I ( h ) > is the weighted mean of all measurements of h .
DorchainsBandC(Fig.1a).Bothdimericinterfaceshavesimilar sizeandthesameresiduesareinvolved,thusonlytheADinterface will bedescribed.Here there arestrong ionicinteractions(<4 ˚A) involvingAsp198A-Arg247D,Asp228A-Lys172D,Arg247A-Asp198D and Lys172A-Asp228D.There are twoadditionalhydrogen bonds fromAtoDandDtoAinvolvingArg202OtoLeu204Nandone longer(3.6 ˚A)polarinteractionfromIle199OtoArg212NH1(Fig.
1b).Totallythereare8/7hydrogenbonds,4/3ionpairs<4 ˚Aand additional3/3ionpairs4–6 ˚AfortheAD/BCdimers(Table2).
Upondimerformation21,100 ˚A2,i.e.about30%oftheaccessible surfacearea(ASA)foreachmonomer,isburied(Table2).Comparedto otherfunctionaldimers,thisburiedASAise.g.similartothatinisoc- itratedehydrogenasefromthepsychrophilicbacteriumDesulfotalea psychrophila[19]andlargerthaninalkalinephosphatasefromthe antarcticbacteriumTAB5andinthreeotheralkalinephosphatases [20].Significantly,thelpPAHdimerinterfaceisverydifferentfrom thatofdimerictruncatedhPAH(residues103–428)[21],sinceinthe lattertheC-terminalresidues(411–424inhPAH)formtwoβ-strands involvedinthedimerization(Fig.1c).Also,whileinhPAH(residues 103–428)thedimerinterfaceisatthe(left)sideofthecatalyticdo- main(Fig.1c),thelpPAHdimerisformedatthetopofchainA,with differentresiduesinvolvedindimerization.InlpPAH,hydrogenbonds (Arg202OtoLeu204N)bindoneβ-strand(residues201–204)from chainAtothesamestrandfromchainD,andtheseresidues,together withion-pairformingresidues172,198,247and228,arelocatedto- wardsthemiddleoftheproteinchain.Thisdiffersfromthetruncated hPAHdimerwheretheinteractionsonlyinvolveC-terminalresidues (Fig.1c)[21].Furthermore,theburiedASAof519 ˚A2(3.7%ofASAper monomer)measuredinthehPAHdimerinterface,ismuchsmaller
Fig. 1. Overall structure of lpPAH. (a) The two lpPAH dimers (chains A–D) in the asymmetric unit. (b) Ionic interactions and (d) stacking interactions at the dimer interface of chains A (red) and D (sand). (c) Superimposition of hPAH truncated dimer (subunits in yellow and cyan; PDB 1PAH [ 21 ]) onto the lpPAH dimer (chains A and D in red and sand, respectively). (e) A large aromatic cluster, with the unique lpPAH residues Phe211, Phe236 and Phe244. (f) Residues 163–168 and 225–229 (motifs in green) predicted to be prone to β-aggregation by the TANGO-algorithm, surrounded by ion pairs and aromatic clusters [ 38 ] in one lpPAH dimer. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
Fig. 2. The lpPAH active site. (a) The active site of apo-lpPAH with three polyethylene glycol (PEG) molecules and one water molecule (WAT, in red). (b) lpPAH with modeled BH 2(green), and (c) with substrate analogue THA, BH 2and Fe 2+(green) modeled from hPAH-BH 4-THA (PDB 1MMK ). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
thaninthelpPAHdimer.
Inadditiontotheionicinteractions,whichappeartosealthelpPAH dimertogether,theinteractinginterfaceiscomposedoftheβ-strands
Table 2
Structure and dimer interaction analysis of lpPAH. The calculations are done for the residues in the crystal structure only, as indicated.
lpPAH
No. of res. in chain A / B / C / D 252 / 252 / 232 / 237 No. of H-bonds per residue in a
Chain A 0.881
Chain B 0.865
Chain C 0.853
Chain D 0.844
Ion pairs
No. of < 4 ˚A Chain A / B / C / D 13 / 10 / 4 / 8 No. of 4–6 ˚A Chain A / B / C / D 10 / 12 / 5 / 6 No. of ion pairs per residue a
< 4 ˚A Chain A / B / C / D 0.052 / 0.040 / 0.017 / 0.034 Dimeric interface
No. of H-bonds b
Chain A to D 8
Chain B to C 7
No. ion pairs < 4 / 6 ˚A
Chain A to D 4 / 2
Chain B to C 3 / 3
Accessible surface area (ASA) of dimer ( ˚A 2)
Chain A and D 21,150
Chain B and C 21,065
Buried intersubunit surface (% of dimer)
Chain A to D 30.2
Chain B to C 30.4
aIn crystal structure.
bIncluding ionic interactions.
Val201-Arg202-Ile203-Leu204-Phe205andVal223-Tyr224-Phe225- Val226-Ile227fromeachsubunit,andtheadjacenthydrophobicC- terminalcoils(Glu251-Asp259)wherethetwoTyr257residuesestab- lishanimportantstackinginteraction(Fig.1d).Theuniquedimericar- rangementinlpPAH,andnotablythepresenceofthesehydrophobic/ aromaticinteractions,seemstobeanimportantdeterminantofthe highthermalstabilityoftheenzyme[8](seebelow).
2.3. Irondependence
Evenwhenexcessironisaddedduringproteinexpression,recom- binantlpPAHispurifiedfromEscherichiacoliasanapo(metal-free) enzyme(0.07 ± 0.03molFe(II)/molsubunit)[8].Also,whenactiv- ityofthenativeenzymeismeasuredinlysatesofL.pneumophila, additionof100μMFe(II)intheassaygivesseveral-foldhigheractiv- ity[8],indicatingthattheactivesiteoflpPAHissub-saturatedwith catalyticironintracellularly.AcomparisonofLpPAHtocpPAH(PDB 2v27),cvPAH(PDB1LTV)andhPAH-BH4(PDB1J8U),identifiesthe iron-bindingresiduesinlpPAHasHis122,His127andGlu167.Inlp- PAH(chainA)awatermoleculeisfoundatthesimilarpositionas Fe(II)inotherPAHsinthemetal-boundholo-form(Fig.2a).Fromthe observedelectrondensityitisclearlyasolventmoleculeandnota metalion,inagreementwithlpPAHbeingisolatedasanapoenzyme [8].
2.4. TheBH4cofactorandsubstratebindingsites
AvailablestructuresofhPAHcomplexedwithBH4 (orBH2)ei- theraloneortogetherwiththeTHAsubstrateanalogue((hPAH-BH4 (PDB1J8U);hPAH-BH4-THA(PDB1MMK)andcvPAH-BH2(PDB1LTZ)) provideaframeforcomparativefunctionalanalyses.lpPAHshowsa Km(BH4)of125± 25μM,whichindicatesloweraffinityfortheco- factorcomparedto hPAH(Km(BH4)=15–29μM)[22]andcvPAH (Km(BH4)=15–21μM)[16,23].Onestructuralexplanationforthe loweraffinityisthatlpPAHcannotprovidethesamenumberofhydro- genbondsuponcomplexformationwithBH4ascvPAH(fromAsp104) andhPAH(fromSer251)duetoAla88(Fig.2b).ThesidechainofAla88 inlpPAHishydrophobicandcannotmakeahydrogenbondtoO-1’
ofBH2/BH4.Still,lpPAHshouldbeabletomaketwohydrogenbonds frombothIle86OandNasobservedinhPAH-BH4andcvPAH-BH2, asshowninFig.2bwhereBH2ismodeled.lpPAHhasnevertheless higheraffinityforBH4thancpPAH(Km(BH4)=300μM)[15]where Phe88(Ala88inlpPAH)mightstericallyhamperthecofactorbinding, thusexplainingthepoorbindingofBH4tothecold-activecpPAH.
TheenzymekineticsoflpPAHalsoindicatedalowaffinityforits substrate,withaKm(l-Phe)=735 ± 50μM[8]comparedwitha Km(l-Phe)=59 ± 10μMforcvPAH[16]andKm(l-Phe)=60μM fortruncatedhPAH(residues103–428)[24].Westudiedthestruc- tureforpossibleresiduedeterminantsofthelowaffinityforl-Phe in lpPAH. The substrate binding-site is best characterizedin the crystalstructureoftruncateddimerichPAHboundtotwosubstrate analogsl-norleucine(PDB1MMT)and3-(2-thienyl)-l-alanine(THA;
PDB1MMK) [25]. When modelingTHA intolpPAH theconserved Arg107interactswiththecarboxylgroupofl-Phe(Fig.2c).Onthe otherhand,Tyr114inlpPAH,whichisconservedinallPAHs,isnot intherightorientationto contributeinasimilarwayasinhPAH tothehydrophobicbindingsite[25],althoughrotationintoanother side-chainrotamermightimprovethel-Phebindingproperties(Fig.
2c).Otheradjacentconservedresiduesthatpromotel-Phebinding are Pro118, His122,Trp163 andPhe168where thetwolatterare foundoppositetoPro118.hPAHhasaprolineresiduewherelpPAH hasGln116.Therearethreeconsecutiveglycines(Gly181,Gly182, Gly183)inlpPAHcorrespondingtoGly,AlaandGlyinhPAHandin cvPAH.Theseglycinesarealsofoundincold-activecpPAHwhichhas lowsubstrateaffinity([S0.5](l-Phe)=1.1 ± 0.1mM)[15].Assum- ingthatthecurrentorientationismaintainedinl-Phe-boundlpPAH, thethreeconsecutiveglycines(Gly181,Gly182,Gly183)andTyr114, mightexplaintheobservedrelativelylowaffinityofthisenzymefor l-Phebinding.
2.5. ThermalstabilityoflpPAH
Characterizationofthethermalstabilityofiron-freeapo-lpPAH bothbycirculardichroismanddifferentialscanningcalorimetry(DSC) hasrevealedathermostableenzymewithTm=79±0.5◦C[8],which ismuchhigherthantheTm-valuesmeasuredforPAHfromeukary- otesorothermesophilicbacteriastudiedsofar,clusteringaround 55◦C[9,26,27].Infact,lpPAHshowsahigherTm-valuethantheapo formofPAHfromthethermophileChloroflexusaurantiacus(caPAH;
Tm∼64◦C)[12].Moreover,similarTm-valuesaremeasuredforthe Fe(II)-bound holoenzymesofboth lpPAH(Tm 80◦C)andther- mophiliccaPAHsincethebindingofFe(II)seemstobealargestabi- lizationdeterminantforcaPAH[12]whereasithasalmostnoeffect onthestabilityoflpPAH[8].Thus,itseemsthatspecificstructural determinantsarerelatedtotheintrinsicstabilityofapo-lpPAH,and wenotablyfocusedonspecificclustersofionicresidues,hydrophobic residuesincludingaromaticclusters,anddimerization.
2.6. Polarinteractionsandaminoacidsequencecontent
Ashydrogenbondsandion-paircontentappeartoberelatedto proteinthermostabilityandthermaladaptation,seee.g.[28,29],we analyzedtheseinteractionsinlpPAH(Table2)andcomparedthem withthoseincpPAH,cvPAHandhPAH[15].ThelpPAHaminoacidse- quencehas45%,38%and19%sequenceidentitytocpPAH(Tm=52◦C) [15],cvPAH(Tm =64◦C)[27]andhPAH(Tm=53◦C)[15],respec- tively(Fig.S1;Supplementalinformation).lpPAHhas0.84–0.88hy- drogenbondsperresidue(Table2).Comparatively,cpPAHwasfound tohaveahighernumberofhydrogenbondsperresidue(0.981for holo-cpPAH,PDB2V27,and0.951forapo-cpPAH,PDB2V28)[15].
It ishowever important tokeep inmind that thesenumbersare dependentonresolutionoftheX-raystructureandrefinementre- strains.Furthermore,formanythermophilicenzymes,extendedion- pairclustersareshowntoexplaintheirhighthermostability(seee.g.
Table 3
Residue content analysis. (a) Residues in an aromatic cluster and (b) number of Phe (F), Tyr (Y), Trp (W) and His (H) residues in lpPAH, cpPAH, cvPAH and truncated hPAH.
Residue No. lpPAH cpPAH cvPAH hPAH
(a)
91 F F F F
92 F F F L
100 F F F F
162 F Y Y Y
139 F Y Y Y
143 Y Y Y I
205 F F F L
211 F L M A
236 F I A
244 F L L
(207) (P) (I) (L) (L)
No. of aromatic residues in cluster
10 7 7 4
(b)
lpPAH cpPAH cvPAH hPAH
(103–428) In Gene (no.
aa)
272 267 297 326
No. of F 20 16 17 22
No. of Y 13 12 12 20
No. of W 4 4 7 3
No. of H 4 5 6 10
Total No.
(%) of F, Y, W, H
41 (15.1%) 37 (13.9%) 41 (14.1%) 55 (16.9%)
[28,29]).However,forlpPAHboththetotalnumberandthenumberof ion-pairsperdefinedresiduearelow(0.017–0.052;Table2)andthe largestion-paircluster(<4 ˚A)comprisesonlythreeresidues.Thus, fromthecurrentlpPAHstructurethehydrogenbondsandsalt-pairs donotseemtoexplainthehighthermotoleranceofthisenzyme.
But when analyzing the amino acid sequence, the high Arg/ (Arg + Lys)ratio(0.67/0.39/0.61/0.44inlp/cp/cv/hPAH),thelow numberofGlyresidues(10/13/14/15inlp/cp/cv/hPAH)andthePro content(14/12/14/12inlp/cp/cv/hPAH),mightallcontributetothe thermostabilityinlpPAHcomparedtootherPAHs,asinferredfrom comparativeanalysesofthermophilicproteinsversusmesophilicand psychrophilicorthologsaimingtorevealmechanismsoftemperature adaptation[15,30].Whilearginineresiduesappeartobesuperiorto lysinestowithstandhightemperatures,fewerglycinesandadditional prolinesareassociatedtoareductioninconformationalflexibility [30–33].
2.7. Aromaticinteractionsandtheuniquedimerizationinterface
OnestrikingfeatureinlpPAHisanextendedaromaticclusterpro- trudingfromtheconservedactivesiteresiduesPhe91,Phe92(Leuin hPAH),Phe100,Phe139(Tyrincp,cv,hPAH)andPhe162.InlpPAHthe clusterisextendedbyresiduesTyr143(Tyrincp,cvPAH),Phe205(Phe incp,cvPAH),Phe211,Phe236andPhe244,andenclosedbyPro207 (Fig.1e).Intotalthereare10/7/7/4aromaticresiduesinlp/cp/cv/ hPAHinthisaromaticclusterfoundinonesinglelpPAHchain(Table 3).
ThereisanotheraromaticclusterinlpPAH(Trp28,Phe32,Phe47, Phe121,Phe125)linedbyPro62andPro129whichissimilarinthe comparedPAHsexcepthPAHwhichhasoneCys(Phe123inlpPAH).
Structuresoftheotheraromaticaminoacidhydroxylasefamilymem- berstryptophanhydroxylase(PDB1MLW,3E2T,3HF6)andtyrosine hydroxylase(PDB1TOH,2TOH,2XSN)alsocontainahighnumber ofaromaticresidues(Phe,Tyr,Trp,His;datanot shown).Forthe comparedPAHstructuresthenumberofphenylalanineresiduesis relativelyhighinlpPAH(20/16/17/22inlp/cp/cv/hPAH),andmany
Table 4
Structure–energetics correlations. The theoretical unfolding changes in heat capacity ( C p) and enthalpy at 60 ◦C ( H 60), and at 79 ◦C (the T m-value for lpPAH) ( H 79), calculated from the changes in apolar and polar accessible surface area ( ASA apand ASA p) based on the crystal structure of dimeric lpPAH for the A and D or B and C chains, respectively, that form two unique dimers.
ASA ap
( ˚A 2)
ASA p
( ˚A 2)
C p
(kcal / K / mol)
H 60
(kcal / mol)
H 79a
(kcal / mol) Chains A
and D
12967.4 8183.0 4.2 147.5 218 b
Chains B
and C 12939.5 8125.9 3.7 145.9 216 b
aThe H 79, calculated from H 60and C pusing the Kirchhoff equation.
bFor comparison, the experimental, calorimetric enthalpy change ( H ), calculated from the DSC scan in the absence of salt ( Fig. 3 a and Flydal et. al. [ 8 ], was 169.9 ±0.2 kcal / mol.
oftheadditionalaromaticresiduesareinthedescribedclusters.For hPAHthereare326residuesinthecharacterizeddimericconstruct (Gly103-Gln428)andthePhecontentislower(6.7%)thaninlpPAH (7.4%)(Table3).Ingeneral,aromaticresiduescanformstackingpi–pi interactionsbothinparallelandperpendicularfashionwhentheir ring centers are closerthan 7 ˚A [34,35].Mutation of one central aromaticresidueinisocitratedehydrogenasefromthethermostable ThermotogaMaritimadecreasedthemeltingtemperatureby3.5◦C [28].Thus forlpPAH thesetwoclustersprobablycontributetoits highthermalstability.
Furthermore,athirdrelevantaromaticinteractionisformedby Tyr257inonesubunitstackingwiththesameresidueintheadjacent subunitinthedimer.ThetwoTyr257residuesareheldincorrect orientationbyPro255 andPro256(unique tolpPAH; Fig.S1)and facethecorrespondingresiduesintheotherchain(Fig.1d).Phe258, whichfollowsPro255,Pro256andTyr257,isalsospecifictolpPAH andcontributestothearomatic/hydrophobiccharacterofthemost C-terminalpartofthedimerizationarea.Thisaromaticclusteristhus essentialtotheformationofthedimerizationinterfaceintheenzyme fromL.pneumophila.Thisinterfaceisunique amongthearomatic aminoacidhydroxylasesreportedtodate,andcoversalargerarea andimpliesamuchlargernumberofintersubunitinteractionsthan thedimerizationmotifinthemammalianenzymes(Fig.1candsee above).Furthermore,lpPAHistheonlyreporteddimeric bacterial PAH,whichstronglypointstooligomerizationasamechanismfor increasingthethermalstabilityinthebacterialscaffold,assuggested forotherproteins(forareviewsee[31]).
2.8. Effectofsaltonthethermaldenaturation
InordertoobtainthermodynamicinsightsonPAHfromL.pneu- mophila,andonthestructure-energeticscorrelations,weinvestigated thethermaldenaturationoftheenzymebothintheabsenceandthe presenceofsalt.Aswehavepreviouslyshownbydifferentialscanning calorimetry(DSC),theunfoldingoflpPAHatpH7.0intheabsenceof NaClprovidesanendothermictransitionwithTm=79± 0.5◦Cwith acalorimetricenthalpychange(H)=169.9±0.2kcal/mol([8]and Fig.3a).ThetheoreticalpredictionofHattheTm(H79)obtained byenergycalculationsusingthecrystallographicstructureprovidesa higherenthalpychange(H79=216/218kcal/molforchainsAD/BC;
Table4),inagreementwiththeirreversiblethermaldenaturationof lpPAHandapartiallystructureddenaturedstate.Actually,CDanaly- sisat100◦Chaveshownanincreasednegativeellipticityat216nm [8],indicatingthattheremainingstructureinthedenaturedstatein- cludesβ-structure,whichmightresultfromintersubunitformation ofcross-βinteractions[36].
Whenthethermaldenaturationtakesplaceinthepresenceof 200mMNaCl,macroscopicaggregationis clearlyobservedatthe endofthethermaltransition,whichisinadditionlessendothermic
Fig. 3. Effect of salt on the thermal denaturation and aggregation of lpPAH. Thermal denaturation was monitored by (a) DSC and (b) DLS. lpPAH (30 μM subunit) was heated in 20 mM Na-Hepes, pH 7.0 (black lines) and in the same buffer with 200 mM NaCl (grey lines). The scan rate was 1 K / min in DSC measurements (a) and the average size of the particles ( Z -average) was estimated by scattering intensity measurements monitored every 3 ◦C after an equilibration time of 1 min (b).
andverydistortedbytheaggregationassociatedexothermicprocess, hinderingthedeterminationoftheTmandHatthoseconditions(Fig.
3a).Still,theapparentTmwithsaltislowerthanwithout(Fig.3a).This stimulatingeffectofsaltonaggregationmightbeassociatedtothe specialdimerizationarea,includinghighcontentofβ-structureand intersubunit hydrophobicinteractionssurroundedbyseveralionic pairs(Fig.1f).Hence,thepresenceofsaltcould initiallyfavorthe separation oftheintersubunit ionicbonds andthenreinforcethe attractivehydrophobicinteractionsandenhanceaggregation,asalso inferredinstudiesofaggregation-proneproteinsindifferentsolvents includingsalt-freewater(seee.g.[37]).
Furtherinsightsonthesalteffectontheaggregationpropensity andtypeofthermal-inducedaggregationoflpPAHwereobtainedby thermaldependentdynamiclightscattering(DLS).AsshowninFig.
3b,theonsetofaggregationisretardedwhentheenzymeisheated inbufferwithoutsalt,andthesizeoftheaggregatesare>100-fold larger,consistentwiththeaggregatesinsaltbeingvisibletothenaked eye.
ThespecialarchitectureoflpPAHattheintersubunitregion,with acentralcoreofβ-strandsestablishingintermolecularhydrophobic interactionsandsurroundedbyionicpairs,indicatesthatthecentral areamightbemostpronetoaggregateinaβ-typeofinteraction.We
analyzedtheaggregationpropensityoftheproteinusingtheTANGO algorithm[38],whichidentifiedresidues163–168and225–229as thosehavingthehighestpropensitytodenaturation.Whileresidues 225–229indeedcomprisetheβ-strandattheintersubunitinterface inthedimer(Fig.1f),residues163–168arelocatedonthelastturn ofanα-helixconsecutivetotheformerstrand,andwellorientedto formacross-β-aggregationupon alocalconformationalchangein thehelix.Serpinsandprionproteinsareprototypeofthiskindof aggregation[39],butithasalsobeenfoundinotherproteinsthatdo notformamyloidfibersuponaggregation[36].
2.9. Physiologicalrelevanceofhighthermalstability
Itisnotknownwhetherthermalstabilityextendstootherpro- teins fromL. pnemophila. To ourknowledge, lpPAH isthe onlyL.
pneumophila protein forwhich thedenaturation temperature has beendetermined.Althoughitisnotobviouswhyanenzymefrom L. pneumophila wouldwithstand temperatureswell abovenormal growth-conditions,itisinterestingthatL.pneumophilaisabletoen- teradormant,viablebutnon-culturable(VBNC)stateinresponseto lownutrient-availabilityandotherenvironmentalstressconditions [18].Thesebacteriacanberesuscitateduponenteringapermissive amoebalhostandcontinuebothgrowthandpathogenicity[40,41].It willbeinterestingtoinvestigateifhighthermostabilityisageneral traitofL.pneumophilaproteinsorifitisaspecialpropertyofsome few, includingPAH.Inthislatercaseitistemptingtospeculateif lpPAHcouldbepreservedduringVBNCduetoabeneficialroleupon resuscitation.
2.10. Conclusion
The presented crystalstructure of thermostablephenylalanine hydroxylasefromLegionellapneumophila(lpPAH)showedadimeric structurewiththeuniqueTyr257interactingwiththesameresidue intheadjacentsubunit,stabilizedby theadjacentPro255,Pro256 andPhe258residues.Theuniquedimericinterfaceincludeionpairs andaromaticinteractions,andtwoadditionalaromaticclustersper monomeraretheplausiblestructuraldeterminantsofthethermal stabilityoflpPAH.Inthepresenceofsalt(200mMNaCl),macroscopic proteinaggregationwasobserved,compatiblewithaneffectofsalt bothdestabilizingtheion-pairsatthedimerinterfaceandreinforcing theattractivehydrophobicinteractions.
3. Materialandmethods
3.1. OverexpressionandpurificationoflpPAH
Recombinantwild-type(wt)lpPAHwasoverexpressedinE.coli strainBL21-Codon Plus(DE3)RILasfusion protein(His)6-ZZ-lpPAH witha TobaccoEtchVirus (TEV)-cuttingsitebetweenthe His-tag andZZcarrierasdescribed[8].The(His)6-ZZ-lpPAHfusionprotein wascutwith(His)6-taggedTEVproteaseovernightat4◦Candthe isolatedlpPAHwasobtainedbycollectingtheflow-throughfroma TALONcolumn,followedbybufferexchangeinadesalting(PD-10) columninto20mMHepes,pH7.0and200mMNaCl.
Theprotein concentrationoflpPAHwasmeasured spectropho- tometrically using the extinction coefficient ε280 = 1.20 (mg/ mL)−1cm−1.Thepurifiedproteinwasstoredinliquidnitrogen.
3.2. Crystallization,structuresolutionandrefinement
Crystallization experiments on lpPAH were performed with a Phoenixcrystallizationrobot(ArtRobbinsInstruments),atroomtem- peraturein96-wellformatwithMRCplates.Volumesusedwere60 μlreservoirsolutionperwell,anddropsfrom0.5μlwellsolution plus 0.5μlprotein solutionwith15–21 mg/mlprotein insitting
dropexperiments.Initially480differentin-housemadeconditions werescreenedfollowedby optimizationaroundseveralsuccessful conditions.Still,additionofglycerol,sugars,lowmolecularweight polyethyleneglycol(PEG)andchangeinpHandprecipitantconcen- tration,onlyimprovedthecrystalqualityslightly.Thelargestcrystals grownhadaproteinconcentrationof15–17mg/mlandwithreser- voirsolutionscontaining1.6MNaK2PO4and4%(v/v)PEG400.The crystalswerecryoprotectedin1.6MNaK2PO4and27%etyleneglycol, andthenflashfrozeninliquidnitrogen.
X-raydataforlpPAHwascollectedatBessy,BL14.1at100Kwith awavelengthof0.91841 ˚A,6sexposure,0.25◦oscillationperframe, andintotal103◦ofdatawasusedinthefinaldataset.Thedatawas integratedandscaledwiththeprogramXDS[42]andstructurefactors wereobtainedusingTRUNCATE[43].ThelpPAHstructurewassolved bymolecularreplacement(MR),withahomologymodelmadefrom Colwelliapsychrerythraea34HPAH(cpPAH;PDB2V27;45%sequence identity)[15]assearchmodelwiththeprogramPHASER[44].Three moleculeswerefirstidentifiedfromtheMRsolutionandinspection oftheelectrondensityidentifiedthefourthmoleculethatwasfurther downintheMRsolutionlist.
ThestructurewasrefinedwithREFMAC5[45]andmanuallyrebuilt inWinCoot[46].TranslationLibrationScrew-motion(TLS)refinement wasutilizedduetodisorderinsomeofthechains.Phenixwasalso triedasrefinementprogrambutitdidnotimprovethestatisticsnor theelectrondensitymaps.
InthispaperwehavecomparedourlpPAHstructuretocpPAH (PDB2V27)[15],cvPAH-BH2(PDB1LTZ)[47],hPAH-Fe(II)-BH4(PDB 1J8U) [25],hPAH-BH4-thienylalanine(THA)(PDB1MMK)[48] and hPAH(PDB1PAH)[21].hPAHinthesestructuresisadimerictruncated formincludingresidues103–428.
3.3. Differentialscanningcalorimetry(DSC)
DSCwasperformed usingaMicroCalVP-DSCmicrocalorimeter (GEHealthcare).Asampleof30μMsubunitlpPAHindegassed20mM Na-Hepes,pH7.0,withorwithout200mMNaCl,asindicated,was heatedfrom37to90◦Cusingascanrateof1K/min.Abuffer–buffer referencetracewassubtractedandthedatawasnormalizedwith respecttoconcentrationtoobtainexcessheatcapacity(Cp)asafunc- tionoftemperatureusingtheMicroCal-enabledOrigin7.0.software.
Thesecurveswerethenanalyzedtodeterminethemidpointdenat- urationtemperature(Tm)andthecalorimetricenthalpychange(H) fortheunfoldingtransitions.
3.4. Dynamiclightscattering(DLS)
ThermallyinducedaggregationoflpPAHwasmeasuredbyDLS usingaNanosizerS(MalvernInstruments,Sweden)equippedwitha He-Nelaser(633nm)andafixed173◦backscatteringangle.30μM subunitlpPAHin20mMNa-Hepes,pH7.0,withorwithout200mM NaCl,washeatedfrom37to90◦C.Theaveragesizeoftheparticles (Z-average)wasestimatedbyscatteringintensitymeasurementsand monitoredasafunctionoftemperaturewithmeasurementsevery 3◦Cafteranequilibrationtimeof60s.
3.5. Structure-basedtheoreticalunfoldingenthalpyvaluesandanalysis ofaggregationpropensity
Calculation of the theoretical unfolding heat capacity change (Cp) andenthalpychange at60◦C(H60) wasperformedusing thestructure–energeticsrelationshipsdevelopedbyFreireandco- workers[49],asexplained[26],basedontheapolarandpolarac- cessiblesurfacearea(ASAapandASAp,respectively).ASAapand ASApwerecalculatedbyGetarea[50]usingthecrystalstructureof dimericlpPAH(eitherchainsAandDorchainsBandCwhichform twouniquedimersintheasymmetricunit).
