Translational Modifications in the Folding/Misfolding Behaviour of the Third PDZ

13 Pages·2014·2.55 MB·English

The Impact of Extra-Domain Structures and Post-
Translational Modifications in the Folding/Misfolding
Behaviour of the Third PDZ Domain of MAGUK Neuronal
Protein PSD-95
Javier Murciano-Calles1., Marta Marin-Argany2., Eva S. Cobos1, Sandra Villegas2*, Jose C. Martinez1*
1DepartamentodeQu´ımicaF´ısicaeInstitutodeBiotecnolog´ıa,FacultaddeCiencias,UniversidaddeGranada,Granada,Spain,2DepartamentdeBioqu´ımicaiBiolog´ıa
Molecular,FacultatdeBiocie`ncies,UniversitatAuto´nomadeBarcelona,Barcelona,Spain
Abstract
The modulation of binding affinities and specificities by post-translational modifications located out from the binding
pocket of the third PDZ domain of PSD-95 (PDZ3) has been reported recently. It is achieved through an intra-domain
electrostaticnetworkinvolvingsomechargedresiduesintheb2–b3loop(wereasuccinimidemodificationoccurs),thea3
helix(anextra-structuralelementthatlinksthePDZ3domainwiththefollowingSH3domaininPSD-95,andcontainsthe
phosphorylationtargetTyr397),andtheligandpeptide.Here,wehaveinvestigatedthemainstructuralandthermodynamic
aspects that these structural elements and their related post-translational modifications display in the folding/misfolding
pathwayofPDZ3bymeansofsite-directedmutagenesiscombinedwithcalorimetryandspectroscopy.Wehavefoundthat,
althoughalltheassayedmutationsgenerateproteinsmorepronetoaggregationthanthewild-typePDZ3,thosedirectly
affecting the a3 helix, like the E401R substitution or the truncation of the whole a3 helix, increase the population of the
DSC-detected intermediate state and the misfolding kinetics, by organizing the supramacromolecular structures at the
expenseofthetwob-sheetspresentinthePDZ3fold.However,thosemutationsaffectingtheb2–b3loop,includedintothe
prone-to-aggregationregioncomposedbyasingleb-sheetcomprisingb2tob4chains,stabilizethetrimericintermediate
previouslyshowninthewild-typePDZ3andslow-downaggregation,alsomakingitpartlyreversible.Theseresultsstrongly
suggestthatthea3helixprotectstosomeextentthePDZ3domaincorefrommisfolding.Thismightwellconstitutethefirst
examplewhereanextra-element,intendedtolinkthePDZ3domaintothefollowingSH3inPSD-95andinothermembers
oftheMAGUKfamily,notonlyregulatesthebindingabilitiesofthisdomainbutitalsoprotectsPDZ3frommisfoldingand
aggregation. Theinfluenceofthe post-translational modifications inthisregulatory mechanismis alsodiscussed.
Citation:Murciano-CallesJ,Marin-ArganyM,CobosES,VillegasS,MartinezJC(2014)TheImpactofExtra-DomainStructuresandPost-TranslationalModifications
intheFolding/MisfoldingBehaviouroftheThirdPDZDomainofMAGUKNeuronalProteinPSD-95.PLoSONE9(5):e98124.doi:10.1371/journal.pone.0098124
Editor:OlivierKocher,HarvardMedicalSchool,UnitedStatesofAmerica
ReceivedJanuary31,2014;AcceptedApril28,2014;PublishedMay20,2014
Copyright:(cid:2)2014Murciano-Callesetal.Thisisanopen-accessarticledistributedunderthetermsoftheCreativeCommonsAttributionLicense,whichpermits
unrestricteduse,distribution,andreproductioninanymedium,providedtheoriginalauthorandsourcearecredited.
Funding:ThisresearchwassupportedbygrantsCVI-05915,fromtheAndalusianRegionalGovernment;BIO2009-13261-C02andBIO2012-39922-C02,fromthe
SpanishMinistryofScienceandEducationandFEDER;PI13-01330fromInstitutodeSaludCarlosIIIandSGR09-0761fromtheGeneralitatdeCatalunya.J.M-C.
receivedapostdoctoralcontractfromtheSpanishMinistryofScienceandEducation.M.M-A.wassupportedbyaPIF(UAB)fellowship.Thefundershadnorolein
studydesign,datacollectionandanalysis,decisiontopublish,orpreparationofthemanuscript.
CompetingInterests:Theauthorshavedeclaredthatnocompetinginterestsexist.
*E-mail:[email protected](SV);[email protected](JCM)
.Theseauthorscontributedequallytothiswork.
Introduction like signal transduction, cell adhesion or molecular trafficking,
whetheratneuronalsynapsesintheparticularcaseofPSD-95,or
Based on high-throughput experimental and computational at tight junctions, cell growing and division in the case of other
approaches it has been described that interactomes organize in members of this family [7]. From a dynamic point of view, hub
hubsandsuper-hubswhereahighamountofdifferentmetabolic proteinsusuallydisplayaconsiderablestructuraldisorder,showing
routesjoin.Themosttypicalexampleofsuchhubproteinsisthe highlyvariableregionsthatdevelopavarietyofchangesrelatedto
membrane-associatedguanilatekinase(MAGUK)family,towhich the multi-modular arrangement [8]. These regions have been
PSD-95 (post-synaptic density-95 protein) belongs, constituted by identified as the loops and turns connecting secondary structures
three PDZ (postsynaptic density protein-95/discs large/zonula withinthedomainsand, also,theshortsequences connectingthe
occludens-1), one SH3 and one guanilate kinase (GK) domains owndomains[9,10].
[1,2,3,4,5,6].Fromafunctionalpointofview,suchahubproteins The most abundant domain type in hub proteins is the PDZ,
donotusuallydisplayanyenzymaticactivityandarearrangedas whichusuallyrecognizesC-terminaltailsofadiversityofproteins,
multi-domainproteins,inwhichthedomainsareconformationally allowing these multi-modular proteins to act as scaffolds for
independent, and are interconnected by relatively short amino functionally-related proteins. Nevertheless, this does not seem to
acidsequences.Thismodulararrangementconfershubproteinsa be the only way of acting of PDZ domains, since a few recent
high conformational plasticity, essential in multifaceted processes studiesdescribesomeconformationalaspectsthatcanbeprofited
PLOSONE | www.plosone.org 1 May2014 | Volume 9 | Issue 5 | e98124
ModulationofPSD95-PDZ3Folding/MisfoldingbyPTMsandExtra-a3Helix
by nature to develop new activities or regulate the existing ones. undergo another post-translational modification consisting in the
Forexample,thesecondPDZdomainofZO(Zonnulaoccludens) cyclization of its side-chain after nucleophilic attack of the main-
proteins self-associates through swapping of the b2–b3 hairpin, chain NH group to the CO atoms of the residue (Figure 1) [20].
driving this phenomena to the polymerization of claudins, the This post-translational modification was found when the tridi-
main eventduring cellular tight junctionsformation [11,12,13]. mensionalstructureofPDZ3domainwassolvedbyX-rayathigh
Structuralmodellingofthemulti-modularorganizationofPSD- resolution. Thesuccinimideringparticipatesinthisintra-domain
95[14]showsthatthewholeproteindevelopsaleveloffunctional electrostatic network, and is stabilized by contacts in the crystal
regulation throughout a high conformational plasticity mainly structure, which suggests that it is organized prior to phase
achievedbytheinter-domainsequences.Thisstructuralflexibility changes; however, it was unstable in solution, disappearing when
allows, for example, the burying of some of the binding sites, sucha PDZ3-crystals were dissolved inbuffer,showinga half-life
domain swapping phenomena [11], or tuned changes in the of approximately 1 hour. This behaviour is commonly observed
specificity/promiscuity ofbinding.It hasbeen proposedtheterm for succinimides found in other proteins, being described as
‘‘supertertiary structure’’ to account for the multiplicity of transient modifications of Asp and Asn residues, able to affect
conformations and states that might coexist and interchange in some protein properties [20]. In fact, the negative impact of the
these multi-modular proteins under equilibrium conditions [15]. ring upon the binding of ligand peptides by PDZ3 strongly
Supportingmodelling,someexperimentalstudieshaveshownthe suggests such a central role, as well as for the b2–b3 loop in the
ability of the linker connecting PDZ1 and PDZ2 domains to binding propertiesof PDZ3,where Asp332islocated [18].
develop different bindingaffinities andspecificities inthetandem Moreover, and opposite to other PDZ domains, this domain
PDZ1–2thanintheisolateddomains[10,16,17].Onthecontrary, shows aC-terminal extra-helix extension, the a3 helix(Figure 1),
the third PDZ domain acts individually. Moreover, the modula- corresponding totheamino acidiclinkertothe following SH3in
tion of binding affinities and specificities by post-translational PSD-95. At a functional level, this extension generates an up-
modificationslocatedoutfromthePDZ3bindingpocketthrough regulation of the affinity of PDZ3 by C-terminal protein ligands
an intra-domain electrostatic network has been reported recently [18]; also, it mediates the interactions among PDZ3 and the
byus[18].Thus,ithasbeenreportedthatresidueTyr397,located following SH3 domain in PSD-95 [7]. Isothermal titration
atsucha3helix(Figure1),canbephosphorylatedbySrctyrosine calorimetry (ITC), nuclear magnetic resonance (NMR) and
kinase, and that the resulting phosphorylated-PDZ3 displays a molecular dynamics (MD) suggest that the salt-bridge between
lower affinity to the CRIPT peptide than PDZ3 [19] or to the Glu334oftheb2–b3loopandArg399ofthea3helixisthemain
KKETAV hexapeptide [18]. In addition, residue Asp332 can responsible for the described regulatory mechanism by such an
Figure1.StructuraldetailsofPSD95-PDZ3domain.Leftpanel:1.4A˚ X-raystructureofPSD95-PDZ3domain(PDBcode:3K82)andtheamino
acidssequenceshowingthedistributionofsecondarystructures.SidechainsofAspresiduesarecolouredinpurpleandthoseofGluresiduesin
green.Rightpanel:adetailofthesalt-bridgesbetweenthea3helix(darkblue)andresiduesoftheb2–b3loop(red).Aligandpeptidecomplexed
withPDZ3(PDBcode:1BE9)isshowninorange.Usingthesamecolourcode,thedistributionofsecondarystructuresalongthesequenceisshown
below.Theuppercentralinsetshowssuccinimide-ringformationatAsp332observedintheX-raystructure(PDBcode:3K82).
doi:10.1371/journal.pone.0098124.g001
PLOSONE | www.plosone.org 2 May2014 | Volume 9 | Issue 5 | e98124
ModulationofPSD95-PDZ3Folding/MisfoldingbyPTMsandExtra-a3Helix
extra-helix,whichstronglyinfluencestheinteractionofapositively Materials and Methods
charged residue of the ligand peptide, usually Lys, with residues
Glu331, Asp332andGlu334 located at the b2–b3 loop[18]. TheplasmidsencodingthePDZ3mutantswerederivedfroma
wild-type PDZ3 plasmid (including residues 302–403 of the full
Previous thermodynamic studies regarding PDZ3 conforma-
PSD-95protein)usingtheQuikChangeSite-DirectedMutagenesis
tionalfeatureshaverevealedthatthewholedomainincludinga3
Kit(Agilent).TheD10ct-PDZ3construct(residues302–393ofthe
helix unfolds under equilibrium conditions through an interme-
full PSD-95 protein) was obtained by PCR amplification. All
diatestatethatmaydrivetothereversiblearrangementoffibrillar
proteins were overexpressed in Escherichia coli BL21/DE3 and
and annular supramacromolecular structures after incubation at
purified by standard protocols [18]. DSC, FTIR, DLS (dynamic
temperatureswhereitispopulated(around60–70uC)[21].These
light scattering) and fluorescence kinetics were carried outas it is
studies confirm the high conformational plasticity of the PDZ3
referredinpreviousarticles.Proteinsampleswerepreparedalsoas
domain.Wehavealsoinvestigatedthemainmolecularaspectsof
described elsewhere[21,22].
theintermediatestatebytransmissionelectronmicroscopy(TEM)
and Fourier transform infrared spectroscopy (FTIR), and found
Results
that one of the two native b-sheets of PDZ3 can reorganize into
the intermediate state to give rise to the fibril b-arrangement.
Rationale Behind Mutational Analysis of PSD95-PDZ3
Chemical-shiftNMRanalysisrevealedthattheb-sheetcomprised
In this study, we have conceived a mutational approach to
by b2 to b4 strands seems to be the responsible for such a
understand the impact of extra-elements and post-translational
reorganization. This region is organized as a flexible b-sheet and
modificationsinPSD95-PDZ3.Thus,theextra-elementevaluated
has been predicted by different algorithms to be prone to
wasthea3helixlocatedattheC-terminusoftheoriginaryPDZ3
b-aggregation [22].
construct (residues 302–403 of PSD-95 protein). To study its
Furthermore, the intermediate state does not populate at pH
contribution to the folding/misfolding of the domain we have
conditionslowerthan3.5,wherePDZ3unfoldsunderanapparent
truncateditfromPDZ3originarysequence(D10ct-PDZ3,residues
two-state regime. This titration behaviour might be attributed to
302–393).MutantsE334Q/LandE401Rweredesignedtoaffect
theprotonationequilibriumsofsomeGluand/orAspresidues,the
the main contacts between the a3 helix and the PDZ3 core.
onlyoneswhosepK valuesarewithintherangeofpH 3–4[23].
a Finally,mutationD332Pwasdesignedtoemulatethesuccinimide
This evidence strongly pointed to some of the above mentioned
ringformationatresidueAsp332previouslydescribedbyus[20].
Gluresidues,concretelytoGlu334andGlu401,sincethesearethe
Another previous study [18] has shown that Asp332 is solvent
main responsible for the packing of the extra a3-helix to the exposedattheb2–b3loop,notcontactingwithanyresiduesofthe
domain through salt-bridges with residues Lys355 and Arg399,
ligand or of the domain; the cyclation of this position drives to a
respectively [18]. The question still unknown is to which extent
strong modification of the conformational dynamics of the loop,
these electrostatic interactions may also determine the folding/ beingthiseffecttheresponsibleofanetdropinthebindingaffinity
misfolding behaviour of PDZ3, since both salt-bridges would be ofshortlinearpeptides.Theseeffectscanbeconsideredessentially
influenced when Glu residues protonate below pH 3, being the thesamewhenaProorasuccinimidearepresent,despitethering
packing of theextra-helix tothewhole PDZ3altered. occurs at different positions of the backbone. Thus, within this
Basedonthesefactsandintheabsenceofinformationaboutthe context, we designed the mutant D332G to generate the most
energeticandstructuraloriginsofsuchconformationalaspects,we opposite effect to Pro from a conformational and dynamic
have analyzed here, by differential scanning calorimetry (DSC), perspective, which was confirmed by titration calorimetry and
FTIR and other spectroscopic techniques, the folding/misfolding molecular dynamics [18].
behaviour of a truncated version of PDZ3, D10ct-PDZ3, where
ten C-terminal residues organizing the extra-a3 helix have been DSC Thermal Unfolding of Truncated D10ct-PDZ3 and
removed. The D10ct-PDZ3 variant, a priori displaying a typical
E334Q/L, E401R, D332P/G Point Mutations
PDZ fold,iswellfolded althoughpresents abit lessstability than
In previous work, we carried out DSC experiments in 50mM
the original PDZ3. Nevertheless, our previous evidence reveals potassiumphosphatepH 7.5ataheatingrateof1.5K?min21and
that the unfolding behaviour appreciably changes with respect to
asetofproteinconcentrationsrangingfrom40to727mM.Both,
PDZ3,beingafour-statemodelnecessarytoproperlydescribethe
PDZ3[21]andD10ct-PDZ3[24]calorimetrictracesshowedtwo
DSC unfolding traces [24]. To go in deep into these differences,
well-separated endotherms that moved opposite with protein
wehaveperformedamutationalanalysisofthemainsalt-bridges
concentration. In a three-state model this shifting indicates a
between the domain and the extra-a3 helix. We have found that
higher association stoichiometry for the intermediate state with
the changes observed upon unfolding are much more influenced
respecttoboth,thenativeandtheunfoldedones.Weobserveda
bythelong-rangeinteractionLys355-Glu401ratherthanGlu334-
noticeabledifferencerelatedtotheshapeofDSCprofiles,sincein
Arg399.However,ourpreviousanalysisshowedthattheopposite
D10ct-PDZ3 both transitions were sharper than in PDZ3.
istrueinthecaseofbindingtoshortpeptides,wherethelatterwas
Additionally, reversibility decreased noticeably in D10ct-PDZ3,
the almost unique responsible for the influence of the extra-a3
beinglesserthan30%.Asaconsequence,thethree-stateunfolding
helix in the regulation of such interactions. Since a phosphory-
model proposed to analyze PDZ3 traces (N O 1/n I O U) did
n
lationtarget,Tyr397,occursatthea3helix,ouranalysismayalso
notproperlyconverge,andwedecidedtoincludethepossibilityof
serve to understand the influence of phosphorylation in the
having a monomeric intermediate state under equilibrium,
conformationalpropertiesofPDZ3.Finally,wehaveanalyzedthe
together with its own association process (N O I O 1/n I O
n
relevance on folding/misfolding of the succinimide ring resulting U)[24].Theanalysisofthepopulationsdistributionshowedthat,
from circularization of Asp332 through mutations D332G and at the highest protein concentration assayed, the associated
D332P, the latter taken as an experimental approach that intermediate, I , populated almost 100% at the temperature
n
reasonably emulates the succinimide arrangement upon binding interval50–90uCinthecaseofD10ct-PDZ3,whereasinthePDZ3
of short peptides to PDZ3, as previously demonstrated by example it never reached more than 90% and at the narrower
calorimetry andmolecular dynamics simulations[18]. range of 60–80uC.
PLOSONE | www.plosone.org 3 May2014 | Volume 9 | Issue 5 | e98124
ModulationofPSD95-PDZ3Folding/MisfoldingbyPTMsandExtra-a3Helix
calorimetric traces showed two well-separated endotherms that
moved opposite with protein concentration, similarly to the
described above (Figure 4). Comparing the relative shapes for a
defined protein concentration, we roughly found more similarity
with the previously published PDZ3 traces [21] than for the
D10ct-PDZ3 ones [24], even showing a similar reversibility
(50–60%) to the former. Consequently, we have fitted them to
the three-state model used for the wild-type PDZ3, with the
exception of mutant E401R. For every mutant, we assumed as
common the thermodynamic parameters corresponding to the
association equilibrium of the intermediate as well as the heat-
capacityfunctionsfortheN,I andUstates.Weachievedanice
n
convergence, despite some of the transitions were not properly
reproducedwithrespecttotheirpositionsinthetemperaturescale.
Theexplanationwouldbethattheassociationstoichiometryn=3
derived from fittings just reflects an average of a more or less
disperse distribution of misfolded species, clearly influenced by
protein concentration. In any case, the traces can be well
reproducedbytherespectivemodelwhenlessrestrictiveindividual
fittingsarecarriedout(datanotshown).Theenthalpiesandmid-
point temperatures of the other equilibriums were obtained
separately for every DSC trace, being averaged later. Thus, we
proceeded in thesame way than previously in the case of PDZ3.
Figure 2. DSC thermal unfolding profiles of D10ct-PDZ3 as a
functionofpHconditions.Proteinconcentrationwas1.3mg?mL21 The whole set of thermodynamic parameters is collected in
under 50mM buffer, either acetic/acetateat pH4.0 or glycine/HClat Table 2. In the case of mutation E401R we did not achieve any
other pH values. Experimental data are represented by symbols reasonable convergence using the three-state approach, so we
whereassolidlinesthrougharethebest fittingresultsobtainedfrom opted for the four-state one as for D10ct-PDZ3 considering the
respectiveDSCmodelsasdescribedinthetext. samerestrictions.Thisalternativewassuccessful,asitcanbeseen
doi:10.1371/journal.pone.0098124.g002
inFigure 4.
Since a change in the conformational equilibrium of PDZ3
FTIR Secondary Structure Analysis of PDZ3-mutants
belowpH 3exists[23],wehavestudiedherethepHbehaviourof
Native State
theD10ct-PDZ3domainbyDSC.Wecarriedoutexperimentsat
BanddeconvolutionoftheamideI’spectraofPDZ3previously
1.3 mg?mL21 protein concentration under different pH condi-
acquiredinnativeconditions(25uC,50mMpotassiumphosphate
tions; within the range 2.5–3.5 we used Glycine/HCl and at
pH 7.5)generatedsixmain-bands,centeredat1680,1670,1660,
pH 4.0 acetic/acetate buffer. The behaviour is qualitatively
1650,1640and1630cm21,aswellasthree-minorbandsat1692,
similar to PDZ3, since at pH values 4.0–7.5 traces are biphasic 1620and1606cm21[22].Sincethesamecomponentshavebeen
(Figure2)andatmoreacidicconditionsbothtransitionsapproach
reportedforD10ct-PDZ3,weconcludedthatD10ct-PDZ3keepsa
toasingletransitionbut,differently,wecouldnotproperlyfitany
typical PDZ fold [24]. Nevertheless, some rearrangement exists.
of them to the two-state model as done in PDZ3 [23]. In fact,
Strikingly,deletionofthea3heliximpliesonly2%decreaseonthe
fittings in Figure 2 were performed using whether the four-state a-helix band (1650cm21) (Table S1). Besides FTIR is a low-
model described above for pH 4.0, or the three-state model
resolution technique, one plausible explanation could be that the
previouslydescribedforPDZ3fortracesatlowerpHvalues.Inall
a3 helix packs against the flexible b-sheet (strands b2, b3, b4)
of the analyses we considered a stoichiometry of n=3 for the
which,inturn,packsagainsttheshorta1helix.Thus,thedeletion
oligomers, as previously [21]. The thermodynamic parameters
couldresultonabetteraccommodationoftheflexibleb-sheetand
derivedfromtheanalysisarecollectedinTable1.Thepopulations on a stronger packing of a1 helix. In fact, the flexible b-sheet
distribution (Figure 3) shows that the folding intermediate component, centered at 1640cm21, diminished for D10ct-PDZ3
destabilizes when pH drops, as it occurs in the PDZ3 case, since with respect to PDZ3, whereas the component for the stable
itsrelativepopulationdecreasesasawholefrom100%atpH 4.0 b-sheet(attributedtotheb1andb5strandsinPDZ3),centeredat
to no more than 20% at pH 2.5, being 0% below pH 3.0 in the 1630cm21, was clearly increased (22% for D10ct-PDZ3 versus
case of PDZ3. In addition, we observe a net influence of the 14% for WT) (Figure 5 and Table S1). This concurs with an
reversibilityofthemisfoldingpathwayontheglobalreversibilityof increased packing of the b-sheet arranged by b2, b3 and b4
the unfolding process, having a more reversible unfolding when strands surely by improving the hydrogen bonding network.
theintermediateislesspopulated.Thus,reversibilityrangesfrom Therefore,thisincreasedpackingoftheb-sheetsmightwellbethe
80% at pH values 2.0–3.5 to 30% at pH 4.0–7.5. These reason that, despite the decrease of the cooperative unit after
populationanalysesclearlypointtotheconclusionthattruncation truncation of a3 helix (roughly a 10%), the stability of D10ct-
of the extra-a3 helix drives to a net advantage of the associated PDZ3 does not drop dramatically with respect to PDZ3 [24]
intermediatestate,sinceitpopulatesinawidertemperaturerange (Table 2).
and in a higher percentage than the observed when such helix is We have also measured FTIR spectra at 25uC for all PDZ3
present at all ofthepH conditions assayed. mutants(Figure5,panelA).Irregularsecondarystructures(1690–
AspreviouslyachievedforD10ct-PDZ3andPDZ3,wecarried 1660cm21)clearlyincreaseinmutantsD332P/GandE334Q/L
out DSC experiments for every PDZ3 point mutation in 50mM located at the b2–b3 loop, but not in E401R which is present at
potassiumphosphatepH7.5ataheatingrateof1.5 K?min21and the a3 helix. This would imply some unpacking resulting from
a set of protein concentrations ranging from 40 to 727mM. The destabilization of the native state upon mutation of residues
PLOSONE | www.plosone.org 4 May2014 | Volume 9 | Issue 5 | e98124
ModulationofPSD95-PDZ3Folding/MisfoldingbyPTMsandExtra-a3Helix
Table1. Thermodynamic parameters ofthe thermalunfolding ofD10ct-PDZ3 domainand mutants underdifferent acidicpH
conditionsin50 mMbuffer, obtainedfrom the analysis ofDSCexperiments*.
N«I«1/3I3«U
DH (T )
N-I N-I
pH TN-I(6C) (kJ?mol21) TI-In(6C) DHI-In(TI-In)(kJ?mol21) TI-U(6C) DHI-U(TI-U)(kJ?mol21) DGN-U(298)(kJ?mol21)
4.0 5361 250620 10668 255624 9362 190615 4862
N«1/3I3«U
pH TN-D(6C) DHN-D(TN-D)(kJ?mol21) TD-In(6C) DHD-In(TD-In)(kJ?mol21) DGN-U(298)(kJ?mol21)
3.5 56.560.4 23365 73.260.4 210763 1862
3 52.460.6 23665 6763 25569 1662
2.5 45.960.4 19663 5763 23366 1062
2.5a 44.560.4 20063 – – 1062
*Experimentalconditionswere50mMacetic/acetatepH4.0;50mMGlycine/HClpH2.5–3.5.Theerrorintervalswerecalculatedasdescribedinthetext.The
valuesofthermodynamicmagnitudesfortheUOInandIOInequilibriumswereestimatedforaPref=100mM.Theheatcapacityfunctionsobtainedfromthe
fittings,inkJ?K21?mol21were:pH4:CpN=212.8+(0.105*T);CpU=29.3+(0.084*T);CpIn=262.4+(0.233*T);CpI=235.4+(0.158*T).pH3.5:CpN=2.66+(0.056*T);
CpU=12.1+(0.034*T);CpIn=296.6+(0.348*T);pH3:CpN=1.50+(0.059*T);CpU=15.1+(0.026*T);CpIn=211.7+(0.105*T);pH2.5:CpN=10.3+(0.032*T);CpU=18.1+
(0.018*T);CpIn=215.3+(0.114*T).
aDataobtainedfromtwo-stateanalysis.
doi:10.1371/journal.pone.0098124.t001
Asp332andGlu334,whichinfactisdetectedbyDSCexperiments Concerningbandsfortheflexibleb-sheet(identifiedtocomefrom
(Table 2). In contrast, E401R and D10ct-PDZ3 spectra do not strandsb2,b3,b4inPDZ3[22])andthestableandlongb-sheet
show differences inthisregionwhencompared toPDZ3. (from strands b1 and b5) the change is, although less obvious, in
Glu401 is located at the C-terminus of the a3 helix; therefore, the same sense to the previously observed for D10ct-PDZ3 [24],
theintroductionofapositivechargeuponE401Rmutationshould i.e.,thebandat1630cm21increasesattheexpenseofthebandat
allow for an interaction with the last carboxyl of the helix, 1640cm21 (Figure 5, panel A). Thus, an improvement of the
decreasingtheelectricdipoleofthea-helixand,asaconsequence, hydrogen bonding network of the flexible b-sheet occurs upon
stabilizingit.Ontheotherhand,anopposingeffectarises,because removing or diminishing the packing of the extra-helix a3. This
the mutation has precluded a salt-bridge with the b2–b3 loop. impliesareorganizationofthenativeb-sheetsinthenativestateof
Thus, the a-helix band(1650cm21) remains essentially thesame these mutants, which might well be related to the monomeric
whencomparedtoPDZ3(17%and16%respectively;TableS1). species detected by DSC for the intermediate states of
Figure3.PopulationsanalysisofthedifferentconformationalstatesuponthermalunfoldingofD10ct-PDZ3underdifferentpH
conditions.Continuouslinesrepresentthetemperatureevolutionofthenativestateandthedottedonestherespectivefortheunfoldedstate.The
associatedintermediateisrepresentedbyabrokenline,whereasthemonomericintermediateatpH4.0doesitbyadot-dashline.
doi:10.1371/journal.pone.0098124.g003
PLOSONE | www.plosone.org 5 May2014 | Volume 9 | Issue 5 | e98124
ModulationofPSD95-PDZ3Folding/MisfoldingbyPTMsandExtra-a3Helix
mutants(24%and22%forE334LandE334Q,respectively,versus
28%forPDZ3),whereasthebandforthestableandlongb-sheet
(strands b1 and b5) remains more or less the same. Thus,
mutations altering the native interaction of Glu334 with the a3
helixsubsequently unpackthe b2–b3 loop.
Definitely, with respect to mutations involved in the contacts
between the a3 helix and the domain, it appears that bands for
irregularsecondarystructures(1690–1660cm21)aretheonesthat
havebeenincreasedatthecostoftheflexibleb-sheet(1640cm21)
whentheinteractionGlu334-Arg399(betweenb2–b3loopanda3
helix ) is lost, whereas there exists a improvement of the b-sheet
hydrogenbondingnetworkinthecaseofdisruptingtheinteraction
Lys355-Glu401 (contacting strand b4 and a3 helix), which is
similar, although less obvious, to the achieved by truncation of
helixa3.
Finally, mutation of residue Asp332 causes the highest content
of irregular secondary structures (1690–1660cm21) among
mutants studied in this work, being 45% and 42% for D332G
andD332Prespectively,versus34%forPDZ3(panelAinFigure5
andTableS1).Thisconcursaftermutationofaresiduelocatedin
the exposed-to-solvent b2–b3 loop, specially increasing flexibility
when the introduced residue is a Gly. The succinimide ring
formed by this residue in PDZ3, that may drive to a decreased
flexibilityoftheb2–b3loop[20],cannotbespontaneouslyformed
in these mutants, but is mimicked by the Pro mutation. A scarce
reduction of the a-helix component has occurred upon mutation
of residue Asp332 (13% for both D332G and D332P versus 16%
for PDZ3) (Table S1). This could be due to a decrease of the a3
helix content. The band for the flexible b-sheet has slightly
decreased in both, 20% and 23% for D332G and D332P,
respectively,versus28%forPDZ3(panelAinFigure5andTable
S1). However, the stable and long b-sheet (strands b1 and b5)
remains the same. This is because in these cases, bands for
irregularsecondarystructures(1690–1660cm21)aretheonesthat
have increased.
In summary, the increase in irregular structures achieved by
Asp332 might be a consequence of the probable contribution of
suchresiduetotheinteractionsbetweenthea3helixandtheb2–
b3 loop, in the same way than the near residue Glu334, both
pointing towards Arg399 of a3 helix in the X-ray structure
(Figure 1). Thus, the disruption of the interactions between these
twoPDZ3regionswilldrivetoanadditionaldestabilizationofthe
prone-to-aggregation motif of PDZ3, comprised by strands b2 to
b4[22],whichinfactdecreasesthenative-statestabilityasseenby
DSCexperiments (Table2).
Misfolding and Aggregation of D10ct-PDZ3 and Point
Mutations
To definitely confirm whether the intermediate state of D10ct-
Figure 4. Thermal unfolding profiles of PSD95-PDZ3 mutants
monitored by DSC as a function of protein concentration. PDZ3,themostextremechangeinthePDZ3sequence analyzed
Experimental conditions were 50mM buffer, either acetic/acetate at inthiswork,mayevolvetosupramacromolecularstructures,asit
pH4.0 or glycine/HCl at the other pH values. Experimental data are happens in the PDZ3 case, we incubated a 727mM sample at
representedbygraysymbolswhereassolidlinesthrougharethebest 60uC for a long-time period. We followed such an evolution by
fittingobtainedfromrespectiveDSCmodelsasdescribedinthetext.
DLS,whichshowedanincreaseofthehydrodynamicradiuswith
doi:10.1371/journal.pone.0098124.g004
the incubation time (Figure S1). In agreement, size exclusion
chromatographyoftheseincubatedD10ct-PDZ3samplesat60uC
D10ct-PDZ3 and E401R. In the rest of PDZ3 constructs and
inaSuperdex-75column(GEhealthcare)showedasingleelution
mutants such amonomeric stoichiometryhas not beendetected.
peak at the exclusion volume of the column ($70kDa); this was
MutationofresidueGlu334causesahighercontentofirregular
observed even with non-incubated samples when heated at 60uC
secondarystructures(1690–1660cm21;39%and40%forE334L
justinsidethecolumn(datanotshown).TEManalysisshowedthe
and E334Q, respectively) than that in PDZ3 (34%; panel A in
appearanceoffibrilsafter1-dayincubation.Evenat37uC,sample
Figure5andTableS1).Thea-helixcomponentremainsthesame.
incubationdrove tofibrils aftera long-time incubation,around 1
The band for the flexible b-sheet has decreased a bit in both
month (Figure 6).
PLOSONE | www.plosone.org 6 May2014 | Volume 9 | Issue 5 | e98124
ModulationofPSD95-PDZ3Folding/MisfoldingbyPTMsandExtra-a3Helix
PLOSONE | www.plosone.org 7 May2014 | Volume 9 | Issue 5 | e98124
ModulationofPSD95-PDZ3Folding/MisfoldingbyPTMsandExtra-a3Helix
Figure5.FTIRstructuralanalysisofPSD95-PDZ3mutants.BarsrepresentationofthedifferentcomponentsofthedeconvolutedFTIRspectra
acquiredin50mMpotassiumphosphatepH7.5.(A)at25uCforPDZ3,D10ct-PDZ3,andsingle-pointmutants.Uponincubationat60uCforseveral
time-periodsforPDZ3(B),D10ct-PDZ3(C),E401R(D),D332G(E),D332P(F),E334L(G),E334Q(H).TheresultsforPDZ3havebeenpublishedpreviously
[22]andareshownhereforcomparativepurposes.
doi:10.1371/journal.pone.0098124.g005
These results bear out noticeable differences with respect to misfolding route, we have incubated at 60uC (where the
PDZ3andpointmutations,wherea varietyofoligomericspecies intermediate species are maximally populated) various D10ct-
thatroughlycorrelatewiththeDLSexperimentswereobservedby PDZ3 samples at 727mM and pH 3.0. Inspection by TEM
chromatography [21]. Particles size increased in a more moder- (Figure 6) showedthat after 4 hoursincubation proto-fibrils were
ated way than in D10ct-PDZ3, having only curly fibrils after 1 present, becoming longer after 1 day; however, it is necessary to
month incubation at 37uC. Accordingly, TEM analysis only wait until 1 week to observe abundant fibrils, which were
revealed roughly similar structures to the shown in Figure 6 at a straighter and longer than the ones obtained at pH 7.5
time not shorter than one week [21]. Another interesting feature (Figure 6). We also performed ThT and ANS fluorescence
revealedfromTEMandDLSanalysesistheirreversiblecharacter emission measurements at pH3.0 with these protein samples
of D10ct-PDZ3 fibrils, whereas in the case of PDZ3 and point (Figure S3). ANS fluorescence did not show any lag-phase but,
mutationsweobserveapartialreversibilityofthesearrangements, opposite to what seen at pH 7.5, ThT fluorescence did. Thus, in
whichcouldreturntotheN-stateaftercoolingdownto20uCand/ fullagreementwithTEM,wehavesomeintermediatespopulating
or dilutionof thePDZ3solution [21,22]. from the very beginning but nucleation is necessary prior to the
Therefore,wecanconcludethatthea3helixseemstoprotectto elongation phase that gives rise tostraight andlong protofibrils.
some extent the PDZ domain from misfolding, since the Finally,wemeasuredthefluorescenceofallthesesamplesafter
precursory intermediate state is clearly lowly populated being coolingdownto25uCwiththegoalofcheckingtheirreversibility
the following misfolding route kinetically slow and partially of D10ct-PDZ3 fibrils. In agreement with our previous evidence,
reversible whena3 helixisattached tothedomain. fluorescenceemissionremainedalmostunaltered,evenafterthree
To understand the fibrils organization mechanism in D10ct- days at room temperature (Figure S4), which confirms the
PDZ3wemeasuredfluorescenceemissionbyThT(thioflavineT) irreversible nature of D10ct-PDZ3 fibrils. In the case of point
and ANS (8-aniline sulphonic acid). Under the same buffer mutations, reversibility was similar to the previously reported for
conditionsusedinthepreviousexperiments,wepreparedsamples PDZ3 [21].
at 40 and 727mM protein concentration, including saturating
concentrations of ThT (12.5 mM) or ANS (20 mM). The progres- Structural Analysis of PDZ3-mutants Intermediate State
sive heating of these samples from 20 to 60uC revealed a slight and Misfolding
increase of fluorescence at 40uC for the higher protein concen- TheconformationalchangesthatundergoPDZ3uponincuba-
trationandabove50uCforthelowerone(FigureS2,leftpanels), tionat60uChavebeenalreadypublished[22].Inshort(Figure5,
in full agreement with the populations analysis formerly done for panel B), the component for the flexible b-sheet dramatically
D10ct-PDZ3 [24]. Such increase is due to the binding of both decreased while concomitantly the component for b-aggregates
chemicals to newly solvent-exposed hydrophobic pockets in the increased. The more structured part of the native anti-parallel
domain (ANS) and the simultaneous appearing of b-aggregates b-sheet and the a-helix components also scarcely decreased,
(ThT). We also observed a parallel increase in the fluorescence although the differences were too small to be unequivocally
intensity withprotein concentration. determined.
Afurtherincubationofthesesamplesat60uCforseveralhours The conformational changes in D10ct-PDZ3 upon incubation
(Figure S2, right panels) showed that fluorescence emission at 60uC have been studied here, after 5 min of equilibration and
dramatically increased at the very beginning of the incubation, after incubation for 4, 8, and 16 days (Figure 5, panel C). After
which reveals the lack of a lag-phase for the association of 5 min at 60uC D10ct-PDZ3 showed an increase in the region
monomers in the In state, as it happens in PDZ3 and point corresponding to loops/turns, and this increase became progres-
mutations. Nevertheless, the maximum intensity is reached at sivelygreateruponincubation;i.e.,at25uCthesebandscomprised
roughly30mininthecaseofThTand350minforANS,whereas 33%ofthespectrumareawhereasafter16daysthevaluereached
in the PDZ3 analysis these maximums where found at 100min 52% (Table S1). Although some decrease of the a-helix
and400minrespectively[21].Thefollowingdropoffluorescence component occurred just upon 5min incubation, the value
isduetotheincreaseinviscosityofthesolution,whichgenerates increased after 8 days but did not recover the initial value
anincreaseoflightdispersionphenomenaasaconsequenceofthe (Figure 5, panel C). For PDZ3, 4 days of incubation were
appearance ofthepoorlysolublesupramacromolecularstructures necessarytodetectsuchadecreaseandtherecoveryafter16days
[25,26,27]. When the incubation was done with a protein of incubation was completed (Figure 5, panel B). Also, just by
concentration of 40mM the maximums are difficult to be incubating for 5min the component for the flexible b-sheet
evaluated since the curves do not drop appreciably, because droppedinD10ct-PDZ3andstayedmoreorlessthesameuntil16
aggregates are in a clearly lower amount than at 727mM. In all days (Figure 5, panel C). This was also the case for PDZ3 but,
cases, fibrillation kinetics in the presence of both ThT and ANS again, the drop occurred after 4 days (Figure 5, panel B). In
(FigureS2,rightpanels)lackthelagphasefeaturingamyloidfibril contrast to PDZ3, which maintained the band for the stable
formation and behave as it has been described for curved fibrils, b-sheet more or less similar during 16 days, in D10ct-PDZ3 this
i.e. those worm-likefibrils formedby b2-microglobulin [28]. band clearly decreased by just incubating for 5 min. This means
ThepopulationdistributionofthedifferentD10ct-PDZ3species thatbothnativeb-sheets,theflexibleandthestableones,arelost
(Figure 3) also shows that the folding intermediate destabilizes infavouroftheb-aggregatecomponentinD10ct-PDZ3.However,
whenpHdrops,asinthePDZ3case,sinceitsrelativepopulation theb-aggregatecomponentismaximalatday4,reaching29%of
decreasesasawholefrom100%atpH 4.0tonomorethan20% the area of the spectrum, decreasing thereafter in favour of the
at pH 2.5. To understand how this decrease may affect the above mentioned increase of the loop/turn component. After 16
PLOSONE | www.plosone.org 8 May2014 | Volume 9 | Issue 5 | e98124
ModulationofPSD95-PDZ3Folding/MisfoldingbyPTMsandExtra-a3Helix
pH7.5obtainedfromtheanalysisofDSC 21?DG(343)(kJmol)U-In 26255 26247 26255 26243 264110 2211??DDG(343)(kJmol)G(298)(kJmol)I-InN-U 2662752810 2662465120 100M.Theobtainedheatcapacityfunctionsm22T);C=113.0(0.389*T);C=12.8D332P:++pInpN2=16.4(0.029*T);C=85.5(0.309*T);D10ct-++UpIn2284*T);C=34.6(0.138*T).+pI
ThermodynamicparametersofthethermalunfoldingofthePSD95-PDZ3domainandmutantsin50mMpotassiumphosphateTable2.experiments*. N1/3IU««3 66222111???DDDT(C)H(T)(kJmol)G(298)(kJmol)T(C)H(T)(kJmol)N-UN-UN-UN-UU-InU-InU-In a66662670.40.53352039679.21.213020PDZ3 66662670.11.53604033888.62.014540D332G 66662671.62.03602533789.92.016030D332P 66662667.41.330010(1–3)25496.01.514520E334L 66662670.11.53702532689.01.514520E334Q UNI1/3I«««3 2222211111666?????DDDDDHHH(T)(kJmol)G(298)(kJmol)(T)(kJmol)G(298)(kJmol)(T)(kJmol)(C)(C)(C)TTTN-IN-IN-II-UI-UI-UI-InI-InN-II-UI-In b66666662651.63.02502028878.53.021510176124.6413515D10ct-PDZ3 66666662664.43.531530571087.57.0300253512127.7411520E401R *Theerrorintervalswerecalculatedasdescribedinthetext.ThevaluesofthermodynamicmagnitudesfortheUInandIInequilibriumswereestimatedforaP=OOref2211?222?molwere:C=9.23(0.095*T);C=2.91(0.064*T);C=99.0(0.347*T);C=9.02(0.093*T);C=11.42(0.045*fromthefittings,inkJKPDZ3:D332G:+++++pNpUpInpNpU22222(0.103*T);C=6.69(0.058*T);C=105.6(0.364*T);C=24.4(0.143*T);C=10.7(0.104*T);C=38.4(0.170*T);C=7.5(0.085*T);CE334L:E334Q:++++++pUpInpNpUpInpNp222222=9.7(0.096*T);C=19.1(0.117*T);C=30.4(0.131*T);C=31.9(0.146*T);C=13.6(0.107*T);C=34.4(0.144*T);C=79.1(0.CE401R:PDZ3:+++++++pNpUpInpIpNpUpInaDatatakenfrom[21].bDatatakenfrom[24].doi:10.1371/journal.pone.0098124.t002
PLOSONE | www.plosone.org 9 May2014 | Volume 9 | Issue 5 | e98124
ModulationofPSD95-PDZ3Folding/MisfoldingbyPTMsandExtra-a3Helix
Figure6.TEManalysisofPDZ3-mutantssupramacromolecularstructuresdevelopedafterdifferentincubationperiods.Upperrowof
pictures:A727mMsolutionofD10ct-PDZ3in50mMpotassiumphosphatepH7.5wasincubatedfordifferenttimeintervals,fromlefttoright,1day,
3daysand5days;whereasthelatterpicturewastakenfromasampleincubatedat37uCduring1month.Thesecond-rowpanelsshowaTEM
analysisofthesupramacromolecularstructuresdevelopedbyD10ct-PDZ3solutionsat727mMin50mMglycine/HClbufferpH3.0afterdifferent
incubationperiods.Thehorizontalblackbarcorrespondswithalengthof200nm.
doi:10.1371/journal.pone.0098124.g006
days, the b-aggregate component comprises 17% of the spectra days of incubation and maintaining or so its value at the end.
(Table S1). Concerning to the main secondary structures reorganizing to
Therefore,inthecaseofPDZ3,thea3helixdiminishestosome generate the aggregation band (1620cm21) both the flexible
extenttheaggregationtendencybymaintainingtheintegrityofthe (1640cm21)andthestableb-sheet(1630cm21)areinvolvedinall
nativeb-sheetformedbychainsb1andb5.InthecaseofD10ct- of themutants,but speciallyinthose ina3 helix.
PDZ3, apart from b5 chain becoming the C-terminal of the Insummary,albeitsomedifferences,themainchangeforallthe
molecule, there is a reorganization of the native b-sheet which spectraistheearlieranddramaticincreaseofthe1620cm21band
propitiateschainsb1andb5tobecomeprecursorsofaggregation (corresponding to an ordered b-aggregate), mainly for D10ct-
togetherwithchainsb2tob4.Althoughtheaggregationprocessis PDZ3 and E401R (up to 29% and 26%, respectively, versus 12%
faster in the D10ct-PDZ3 variant, the b-aggregate component for PDZ3), achieved just after 5 min of incubation. In concor-
comprises the same percentage of the spectra after 16 days dance, this increase should be related with the conformational
incubationasinthecaseofthecompletePDZ3(17%versus19%). reorganizationofthewholenativeb-sheetthatshoulddrivetothe
These differences would be indicative of the presence of the monomeric intermediate. In general, mutants located at the a3
monomericintermediatestateinD10ct-PDZ3,whichimmediately helix (D10ct-PDZ3 and E401R) are more prone to aggregation
appears upon heating at 60uC. However, at the end the trimeric than PDZ3, in full agreement with TEM and DLS evidence,
intermediate aggregates more or less in the same way for PDZ3 althoughafter16daysthe1620cm21componentisthesamethan
than for D10ct-PDZ3. thatforPDZ3(,18%),probablybecausethepredominanceofthe
The conformational changes that undergo the set of PDZ3 trimeric intermediate. Nevertheless, mutants D332G/P and
mutantsuponincubationat60uChavebeenalsostudiedbyFTIR E334L/Q, both into the prone-to-aggregation region of PDZ3,
after 5 min of equilibration, as well as after incubation for 4, 8, showagreateraggregationtendencythanPDZ3justuponinitial
and 16 days at this temperature (Figure 5). As in the case for incubation and maintain the b-aggregate band at around 25%.
D10ct-PDZ3, in the E401R mutation the aggregation band ThisconcurswiththeabsenceofthemonomericstateintheDSC
(1620cm21) immediately appeared upon heating at 60uC and traces fitting. In consequence, itcan beclaimed that although all
further decreased (Figure 5, panel D), this decrease being also thesemutationsgenerateproteinsmorepronetoaggregationthan
concomitant with an increase in the region corresponding to the wild-type PDZ3, those directly affecting the a3 helix stabilize
loops/turns. However, mutations involving the b2–b3 loop the monomeric intermediate whereas those affecting the b2–b3
behaved more or less as PDZ3 (Figure 5, panels E-H), that is, loopstabilize thetrimeric intermediate.
progressivelyincreasedtheareaoftheaggregationbanduptofour
PLOSONE | www.plosone.org 10 May2014 | Volume 9 | Issue 5 | e98124