Conformational characterization of disulfide bonds: A tool for protein classification

Transcription

1 Conformational characterization of disulfide bonds: A tool for protein classification José Rui Ferreira Marques, Rute R. Da Fonseca, Brett Drury, André Melo To cite this version: José Rui Ferreira Marques, Rute R. Da Fonseca, Brett Drury, André Melo. Conformational characterization of disulfide bonds: A tool for protein classification. Journal of Theoretical Biology, Elsevier, 2010, 267 (3), pp.388. < /j.jtbi >. <hal > HAL Id: hal Submitted on 20 Oct 2011 HAL is a multi-disciplinary open access archive for the deposit and dissemination of scientific research documents, whether they are published or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers. L archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.

2 Author s Accepted Manuscript Conformational characterization of disulfide bonds: A tool for protein classification José Rui Ferreira Marques, Rute R. da Fonseca, Brett Drury, André Melo PII: S (10) DOI: doi: /j.jtbi Reference: YJTBI 6152 To appear in: Journal of Theoretical Biology Received date: 16 June 2010 Revised date: 29 August 2010 Accepted date: 8 September 2010 Cite this article as: José Rui Ferreira Marques, Rute R. da Fonseca, Brett Drury and André Melo, Conformational characterization of disulfide bonds: A tool for protein classification, Journal of Theoretical Biology, doi: /j.jtbi This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting galley proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

3 Conformational characterization of disulfide bonds: a tool for protein classification JoséRuiFerreiraMarques 1 (zerui.marques@fc.up.pt) RuteR.daFonseca 2 (rute.r.da.fonseca@gmail.com) BrettDrury 3 (brett.drury@gmail.com) AndréMelo 1,# (asmelo@fc.up.pt) 1 REQUIMTE/DepartamentodeQuímicaeBioquímica,FaculdadedeCiênciasda UniversidadedoPorto,RuadoCampoAlegre,687, Porto,Portugal 2 CIMAR/CIIMAR,CentroInterdisciplinardeInvestigaçãoMarinhaeAmbiental, UniversidadedoPorto,RuadosBragas,177, Porto,Portugal 3 LIAADINESC,RuadeCeuta,118,6º, Porto,Portugal # Correspondingauthor( address:asmelo@fc.up.pt,Tel: ,Fax:

4 Abstract Background Throughoutevolution,mutationsinparticularregionsofsomeproteinstructureshave resulted in extra covalent bonds that increase the overall robustness of the fold: disulfidebonds.thetwostrategicallyplacedcysteinescanalsohaveamoredirectrole inproteinfunction,eitherbyassistingthiolordisulfideexchange,orthroughallosteric effects.inthiswork,weverifiedhowthestructuralsimilaritiesbetweendisulfidescan reflect functional and evolutionary relationships between different proteins. We analyzedtheconformationalpatternsofthedisulfidebondsinasetofdisulfiderich proteins that included twelve SCOP superfamilies: thioredoxinlike and eleven superfamiliescontainingsmalldisulfiderichproteins(sdp). Results Thetwentyconformationsconsideredinthepresentstudywerecharacterizedbyboth structuralandenergeticparameters.thecorrespondingfrequenciespresentdiverse patternsforthedifferentsuperfamilies.theleaststrainedconformationsaremore abundantforthesdpsuperfamilies,whilethe catalytic +/RHookisdominantforthe thioredoxinlikesuperfamily.the allosteric RHSapleismoderatelyabundantforBBI, Crispandthioredoxinlikesuperfamiliesandlessfrequentfortheremaining superfamilies.usingahierarchicalclusteringanalysiswefoundthatthetwelve superfamiliesweregroupedinbiologicallysignificantclusters. Conclusions In this work, we carried out an extensive statistical analysis of the conformational motifsforthedisulfidebondspresentinasetofdisulfiderichproteins.weshowthat

5 the conformational patterns observed in disulfide bonds are sufficient to group proteinsthatsharebothfunctionalandstructuralpatternsandcanthereforebeused asacriterionforproteinclassification. Keywords:Disulfidebond,conformer,clusteranalysis,proteinclassification.

6 Introduction DisulfidebondsareacommonmotifinNature.Thesestructuralelementshavea significantroleinthethermalstabilityandfunctionofproteins(bhattacharyyaetal., 2004;Creighton,1988;Hogg,2003;Klinketal.,2000;Sardiuetal.,2007).Froman evolutionaryperspective,thesebondsarearelativelyrecentadditiontoprotein structure(brooksandfresco,2002;brooksetal.,2002;jordanetal.,2005;schmidt andhogg,2007)accordingtotherespectivefunctions,thedisulfidebondscanthenbe classifiedasstructural,catalyticorallosteric(schmidtetal.,2006;schmidtandhogg, 2007).Schmidtetal(2006)haveperformedathoroughanalysisofdisulfidespresentin thexraystructuresofthepdbdatabase,andfoundthatbothcatalyticandallosteric disulfidesfellintoparticularstructuralcategories.thetwogroupshadahigher averagepotentialenergy,whichreflectedtheirfunctionalrolethatimpliedeasybond breaking(schmidtetal.,2006). ThedisulfidethreedimensionalstructureishighlyconservedinNatureandhasbeen usedforproteinclustering(cheeketal.,2006;chuangetal.,2003;harrisonand Sternberg,1996;Thanguduetal.,2007).Differentschemeshavebeenintroducedto classifythedisulfideconformers(harrisonandsternberg,1996;hutchinsonand Thornton,1996;OzhoginaandBominaar,2009;Schmidtetal.,2006;Srinivasanetal., 1990)andinthisworkweadoptedtheschemeproposedbySchmidtetal(2006).We analyzedasampleofdisulfidebondsassociatedwithaproteinsetextractedfrom SCOPdatabase(Andreevaetal.,2004;Andreevaetal.,2008;Murzinetal.,1995).The proteinsetincludedelevensuperfamiliesofsmalldisulfiderichproteins(sdp)andthe thioredoxinlikesuperfamily.eachsuperfamilyselectedfortheproteinsethadtofit

7 thefollowingcriteria:(i)containaminimumofthirtydisulfidebonds,(ii)havea minimumoffivepdbstructuresavailable,(iii)havexraystructureswitharesolution higherthan2.5åand(iv)haveonlyuncomplexedstructures.inordertounderstand whetherornotthestructureofthedisulfidesreflectedfunctionalorevolutionary relationshipsbetweenthedifferentproteins,wegroupedthedisulfidefromthe12 superfamiliesindifferentclustersusingahierarchicalclusteringanalysis(hca)anda structuralbaseddistanceprotocol.theresultsdemonstratethattheclusters aggregatesuperfamiliessharebothfunctionalandstructuralpatterns,thereforewe concludethattheuseofdisulfidebondsconformationalpatternsisavalidprotein classificationcriterion. Methodology Theschemeusedinthisworktoclassifythedisulfideconformerswasbasedonfive relevanttorsionangles(figure1).thedisulfidespeciesweretreatedassymmetrical.in thiscontext,onlytwentyconformationalcategorieshadtobeconsidered(table1).for examplethe RHHookconformationalcategorycanbeobtainedbyeither combinationsoftorsionangles(,+,+,,)or(,,+,+,).thisclassificationwasbasedon structuralpatterns(schmidtetal,2006)thatincludedmain,orientationaland peripheralmotifs(table2). [InsertFigure1] [InsertTable1] [InsertTable2]

8 Representativestructuresforthedifferentconformationalcategoriesarepresentedin Tables3to5. [InsertTable3] [InsertTable4] [InsertTable5] TheproteinsetunderstudyischaracterizedinTable6.Wedeterminedthefive relevanttorsionangles( 1, 2, 3, 2 and 1 )foreachdisulfidebond.additionally,the (C C andc C )distancesandthedihedralstrainenergy(dse)werealsoevaluated. [InsertTable6] TheDSEquantitywasexpressed,asafunctionofthefiveabovementionedtorsion angles,bytheempiricalequation(katzandkossiakoff,1986;weineretal.,1984): DSE( kj mol 1 ) 8.37(1 cos(3 )) 8.37 (1 cos(3 ')) 4.18(1 cos(3 )) 4.18(1 cos(3 ')) (1 cos(2 )) 2.51(1 cos(3 )) (1) TheDSEquantityprovidedausefulrankingofthemostfavoreddisulfide conformations.theminimum(2.5kjmol 1 )andthemaximum(84.5kjmol 1 )valuesof DSEcorrespondtothetorsionanglescombinations(60º,60º,±83º,60º,60º)and(0º, 0º,0º,0º,0º),respectively(Schmidtetal,2006).Despiteitssimplicity,thisequation hasbeensuccessfullyappliedforasemiquantitativeevaluationofthestrainenergyin disulfidebonds(schmidtetal.,2006;schmidtandhogg,2007). Representativeconformationsofthedifferenttypesofdisulfidebonds(structural, catalyticorallosteric)areidentifiedintable7.wewillbereferringtobondswiththe conformations+/rhhookas catalytic,andrhstapleas allosteric,becausethese

9 twotypesofbondswerefoundtobeintimatelyassociatedwiththoseconformational categories(schmidtetal.,2006). Acomputerprogram,designatedbyDisulph,wasdevelopedtoperformthe calculations.thedisulfidebondspropensitypr A,forasuperfamilyAwithnp A PDB structures,wascalculatedas, np A Pr 100 nssk nresk, (2) k 1 A 1 np A wherenss k andnres k wererespectivelythenumberofdisulfidebondsandthenumber ofcodedresiduesinthepdbstructurek.thisquantityevaluatesthefrequencyofthe disulfidebondswithinasuperfamily.itiscalculatedastheaveragefrequency associatedwithacorrespondentsampleofpdbstructures. Thefrequenciesassociatedwithalltheconformationalcategories,definedinTable1, werethenevaluatedforeachsuperfamilyandforthesample.thesequantitieswere 2 usedtobuildasquareeuclideandistancesmatrix,whoseelements( ( A, B) ) weredefinedas: d Euclidean d Euclidian ( A, B) ( freq( i, A) freq( i, B)) ;A=1,...,12andB=1,...,12 i1 (3) Inequation(3),freq(i,A)andfreq(i,B)arerespectivelythefrequencyofconformational categoryiinthesuperfamiliesaandb.thesquareeuclideandistancesmatrixdefines ametricforevaluatingthesimilaritiesbetweenobjectsinndimensionalspacesand thereforecanbeusedinclusteranalysis.

10 Inordertorepresentthismatrix,weadoptedtheintuitiveformalismintroducedby (Xieetal.,2000).Thecoordinatesoftheoriginalobjects(thetwelvesuperfamiles) wereprojectedinthe3dcartesianspacebyminimizingthesquaredeviationcost functionsd: 12 A1 2 d ( A, B) d Euclidian ( A, B) 2 SD, (4) A1B1 whered(a,b)wasthedistancebetweentheprojectionsthesuperfamiliesaandbin the3dcartesianspace.weusedthenewtonmethodtocarryouttheiterative minimizationprocess.theprocedureassociatedwithequation(4)wasintroducedfor visualizinglargechemicaldatabases(xieetal.,2000).theminimizationofthis equationprovidedanappropriaterepresentationoftheoriginalhighspaceofthe chemicaldescriptorsinalowdimensionalspace(2dor3d). ThesquareEuclideandistancesmatrixwasthenusedforaHCAprocedure(Johnson andwichern,2007),whichprovidedaclassificationofthesuperfamiliesindifferent clusters.weevaluatedtheconsistencyofthehcapartitioning,bytheevaluationof thesquareeuclideandistancesmatrixintheclusterspace.theelementsofthismatrix wereallthemeansquaredistancesbetweenaclusterc i with n C i superfamiliesanda 2 clusterc j with n superfamilies( MSd Euclidean ( Ci, C j ) )andwithinaclusterc i C j ( MSd 2 Euclidean ( C i ) ): n n Ci C j 2 2 MSd Euclidian ( Ci, C j ) (1 ( nc n i C )) j deuclidian ( A, B) (5) A1B1

11 nc i A1 2 2 MSd Euclidian ( Ci ) (2 ( nc n i C )) i deuclidian ( A, B) (6) A1B1 ThismatrixwasdefinedaccordingtothemeanlinkagecriterionwithintheHCA procedure(johnsonandwichern,2007).thedissimilaritybetweentwoclustersc i and C j increasedwiththeincreasingofthecorrespondentnondiagonalelement 2 ( MSd Euclidean ( Ci, C j ) ).Ontheotherhand,thesimilaritywithinaclusterC i increases withthedecreasingofthecorrespondentdiagonalelement( MSd 2 Euclidean ( C i ) ). InthisworkweusedtheHCAdivisivemethodwhichpartitionedsuccessivelyaninitial setwithnobjectsintofinerclusters.thecorrespondentalgorithmwasthefollowing: (i) (ii) Assignthenobjectstoasinglecluster. Computeadistancematrixintheclusterspaceusinganappropriatemetric. As was mentioned above, we adopted a square Euclidean metric in this work. (iii) (iv) Findtheleastsimilarobjectsandseparatethemindifferentclusters. Repeat steps (ii) and (iii) until the diagonal elements of this matrix being significantlysmallerthanthenondiagonalones. Results Thecharacterizationofthedisulfideconformationalcategoriesfoundinoursampleis presentedintable7.the LHSpiralisthemostfrequentlyobservedcategory(28.9%) andhasthelowestdse(11.5kjmol 1 ).Additionally,sixleaststrainedcategories(

12 LHSpiral,+/RHSpiral,+/LHSpiral,RHSpiral,+RHSpiraland/+RHHook)areclearly prevalent(63.1%)relativetotheremainderofthemoststrainedcategories(36.9%). Therepresentativeconformationsforcatalytic(+/RHHook)andallosteric(RHStaple) disulfidebondshavemoderatedsevalues.wefoundthed(c C ')distancestobe morerelevantfordisulfideconformationalspecificitiesthanthed(c C ')distances (Table7).Thed(C C ')distanceswerequiteinsensitivetothenatureof conformationalcategories(variesfrom3.3to4.0),whilethed(c C ')distances hadasignificantvariationovertheseries(from4.4to6.0).forinstance,in agreementwithschmidtetal(2006),the RHStapleconformationwascharacterized bysignificantlowerd(c C ')distancesthantheotherconformationalcategories. [InsertTable7] [InsertTable8] [InsertFigure2] Thefrequenciesforthedifferentconformationalcategories,calculatedforeach superfamily,arepresentedintable8andfigure2.fromthisfigure,itisevidentthat thioredoxinlikeandsdpsuperfamiliesexhibitverydistinctconformationalpatterns. TheleaststrainedconformationsaresignificantlyabundantinSDPsuperfamilies presentsignificantabundances(from43.4%to86.5%),butoccurataverylow frequencyinthioredoxinlikesuperfamily(13.8%).thisisobviousforthemoststable conformation( LHSpiral)forwhichtheSDPsuperfamiliespresentfrequenciesatleast fourtimeslargerthanthethioredoxinlikefrequency(from12.1%to43.8%against 3.1%;Table8andFigure2).Mostofthedisulfidebondsofthioredoxinlike superfamily(50.8%)areassociatedwiththe catalytic +/RHHookconformation,

13 whereasthisisrelativelyrare(from0.0%to7.7%)forthesdpsuperfamilies(table8 andfigure2).ontheotherhand,the allosteric RHSapleismoderatelyabundantfor BBI(24.2%),Crisp(24.1%)andthioredoxinlike(16.9%)superfamiliesandscarce (from0.0%to5.7%)fortheremaindersuperfamilies. [InsertFigure3] Furtherinsightintohowthestructuralsimilaritiesbetweendisulfidescanreflect relationshipsbetweendifferentproteinswasobtainedwithahcaprocedure,whose dendrogram(murtagh,1984)ispresentedinfigure3.the3dcartesianprojectionof therespectivesquareeuclideandistancesmatrixisrepresentedinfigure4together withthesixclustersidentifiedbythisanalysis.fourclustersreflectthemainstructural andfunctionalmotifsidentifiedinthesample: o Cluster1includesthecatalyticproteinsofthioredoxinlikesuperfamily, withthelowestdisulfidepropensitiesandadominant/secondary structure; o Cluster4includesmostofthemetabolicsuperfamilies(CystineKnot, EGFLamininandPlantlectins),withadominantsecondarystructure; o Cluster5includesmostofthetoxin/defensesuperfamilies(Defensin like,omegatoxins,smallsnaketoxinsandscorpionsliketoxins),with moderatetohighdisulfidepropensitiesandadominantsecondary structure; o Cluster2includestheplantproteaseinhibitorsofBBIsuperfamily,with highdisulfidepropensitiesandadominantsecondarystructure. [InsertFigure4]

14 Theremaindertwoclustersreflectdivergencesfromthementionedmotifs: o Cluster3includesCrispsuperfamilyandisadivergencefromcluster5. Thisclusterincludestoxin/defenseproteinswithlowdisulfide propensitiesandadominantsecondarystructure. o Cluster6includesBPTIlikeandKringlelikesuperfamilies.Thisclusteris theleastwellcharacterizedandincludesproteinswithsmalldisulfide propensitiesanddifferentbiologicalfunctions.theelementsofthis clustersharemorediffusepropertiesas(i)theyareconstrainedbythree disulphidebondswiththesamedisulfidetopology(16,24and35)and (ii)theyareassociatedwiththeregulationofsimilarbiologicalprocesses (bindingmediation,proteolyticactivity,bloodclotting,etc.). WerepresenttheEuclideandistancesmatrixfortheclusterspaceinTable9.Fromthe analysisofthistable,wecanverifythatthemeansquaredistancesbetweenthe clustersaresignificantlargerthanwithintheclusters.theseresultsstronglyindicate thatthehcapartitioningisconsistent. Conclusions [InsertTable9] Inthiswork,wecarriedoutanextensivestatisticalanalysisoftheconformational motifsforthedisulfidebondsfoundinsetofdisulfiderichproteinsfromtwelvescop superfamilies.

15 Thefrequenciesofthetwentyconformationalcategoriesprovidedanearspectral representationofthe12dimensionhyperspaceunderstudy.thegeneraltrends observedinthissamplewerequiteconsistentwiththeresultsobtainedbyother authors(schmidtetal.,2006;schmidtandhogg,2007)forthreedifferentproteinsets. Wecalculatedtherootmeansquaredeviationsbetweenourandthepreviously obtainedfrequencies.thethreevaluesobtainedwerealllowerthan2.6%. TheHCApartitioningofthedatausingasquareEuclideandistancesmatrixresultedin a number of clusters, the majority of which aggregates superfamilies sharing both functional and structural patterns. The only exception is cluster 6, whose elements presentedmorediffuseconnections.wethereforesuggesttheuseofdisulfidebonds conformational patterns as a criterion in SDP classification, as well as to recognize main divergences between SDP and other disulfiderich superfamilies. However, the generalized application of this methodology for protein classification has to be subjectedtofurtherinvestigation. Acknowledgements: WethanktheFundaçãoparaaCiênciaeaTecnologia(FCT)foradoctoralscholarship granted to José Rui Ferreira Marques. Rute R. da Fonseca was funded by FCT (SFRH/BPD/26769/2006). We thank the Universidade do Porto for an electric wheelchairandatrackerpro(acomputerinputdevicethattakestheplaceofamouse forpeoplewithnohandmovement)grantedtojoséruiferreiramarques.

16

17 References Andreeva,A.,Howorth,D.,Brenner,S.E.,Hubbard,T.J.P.,Chothia,C.,Murzin,A.G., 2004.SCOPdatabasein2004:refinementsintegratestructureandsequence familydata.nucleicacidsresearch32,d226d229. Andreeva,A.,Howorth,D.,Chandonia,J.M.,Brenner,S.E.,Hubbard,T.J.P.,Chothia, C.,Murzin,A.G.,2008.DatagrowthanditsimpactontheSCOPdatabase:new developments.nucleicacidsresearch36,d419d425. Bhattacharyya,R.,Pal,D.,Chakrabarti,P.,2004.Disulfidebonds,theirstereospecific environmentandconservationinproteinstructures.proteinengineering DesignandSelection17, Brooks,D.J.,Fresco,J.R.,2002.Increasedfrequencyofcysteine,tyrosine,and phenylalanineresiduessincethelastuniversalancestor.molecular&cellular Proteomics1, Brooks,D.J.,Fresco,J.R.,Lesk,A.M.,Singh,M.,2002.Evolutionofaminoacid frequenciesinproteinsoverdeeptime:inferredorderofintroductionofamino acidsintothegeneticcode.molecularbiologyandevolution19, Cheek,S.,Krishna,S.S.,Grishin,N.V.,2006.Structuralclassificationofsmall,disulfide richproteindomains.journalofmolecularbiology359, Chuang,C.C.,Chen,C.Y.,Yang,J.M.,Lyu,P.C.,Hwang,J.K.,2003.Relationship betweenproteinstructuresanddisulfidebondingpatterns.proteinsstructure FunctionandGenetics53,15. Creighton,T.E.,1988.Disulfidebondsandproteinstability.Bioessays8,5763.

18 Harrison,P.M.andSternberg,M.J.E.,1996.Thedisulphidebetacross:Fromcystine geometryandclusteringtoclassificationofsmalldisulphiderichproteinfolds. JournalofMolecularBiology264, Hogg,P.J.,2003.Disulfidebondsasswitchesforproteinfunction.Trendsin BiochemicalSciences28, Hutchinson,E.G.andThornton,J.M.,1996.PROMOTIFAprogramtoidentifyand analyzestructuralmotifsinproteins.proteinscience5, Johnson,R.A.,andWichern,D.W.,2007.AppliedMultivariateStatisticalAnalysis. PrenticeHall,NewJersey. Jordan,I.K.,Kondrashov,F.A.,Adzhubei,I.A.,Wolf,Y.I.,Koonin,E.V.,Kondrashov,A. S.,Sunyaev,S.,2005.Auniversaltrendofaminoacidgainandlossinprotein evolution.nature435, Katz,B.A.andKossiakoff,A.,1986.Thecrystallographicallydeterminedstructuresof atypicalstraineddisulfidesengineeredintosubtilisin.journalofbiological Chemistry261, Klink,T.A.,Woycechowsky,K.J.,Taylor,K.M.,Raines,R.T.,2000.Contributionof disulfidebondstotheconformationalstabilityandcatalyticactivityof ribonucleasea.europeanjournalofbiochemistry267, Murtagh,F.,1984.CountingdendrogramsAsurvey.DiscreteAppliedMathematics7, Murzin,A.G.,Brenner,S.E.,Hubbard,T.,Chothia,C.,1995.SCOPAstructural classificationofproteinsdatabasefortheinvestigationofsequencesand structures.journalofmolecularbiology247,

19 Ozhogina,O.A.andBominaar,E.L.,2009.Characterizationofthekringlefoldand identificationofaubiquitousnewclassofdisulfiderotamers.journalof StructuralBiology168, Sardiu,M.E.,Cheung,M.S.,Yu,Y.K.,2007.Cysteinecysteinecontactpreferenceleads totargetfocusinginproteinfolding.biophysicaljournal93, Schmidt,B.,Ho,L.,Hogg,P.J.,2006.Allostericdisulfidebonds.Biochemistry45, Schmidt,B.andHogg,P.J.,2007.SearchforallostericdisulfidebondsinNMR structures.bmcstructuralbiology7,49. Srinivasan,N.,Sowdhamini,R.,Ramakrishnan,C.,Balaram,P.,1990.Conformationsof disulfidebridgesinproteins.internationaljournalofpeptideandprotein Research36, Thangudu,R.R.,Sharma,P.,Srinivasan,N.,Offmann,B.,2007.Analycys:Adatabasefor conservationandconformationofdisulphidebondsinhomologousprotein domains.proteinsstructurefunctionandbioinformatics67, Weiner,S.J.,Kollman,P.A.,Case,D.A.,Singh,U.C.,Ghio,C.,Alagona,G.,Profeta,S., Weiner,P.,1984.Anewforcefieldformolecularmechanicalsimulationof nucleicacidsandproteins.journaloftheamericanchemicalsociety106, Xie,D.X.,Tropsha,A.,Schlick,T.,2000.Anefficientprojectionprotocolforchemical databases:singularvaluedecompositioncombinedwithtruncatednewton minimization.journalofchemicalinformationandcomputersciences40,

20

21 Figurescaptions Figure1:Graphicalrepresentationofthefivetorsionanglesusedtoclassifythe disulphideconformers. Figure2:Frequenciesforthedisulfideconformationalcategories. Figure3:Dendrogramforthehierarchicalclusteringanalysis.Thefollowingnotation wasadopted:(1)crisp,(2)cystineknot,(3)defensinlike,(4)egflaminin,(5)omega toxins,(6)plantlectins,(7)smallsnaketoxins,(8)scorpionliketoxins,(9)bbi,(10) BPTIlike,(11)Kringlelikeand(12)Thioredoxinlike. Figure4:.Projected3DCartesianrepresentationofthesquareEuclideandistances matrixandclustersobtainedbythehierarchicalclusteringanalysis.thefollowing notationwasadopted:(1)crisp,(2)cystineknot,(3)defensinlike,(4)egflaminin,(5) Omegatoxins,(6)Plantlectins,(7)Smallsnaketoxins,(8)Scorpionliketoxins,(9)BBI, (10)BPTIlike,(11)Kringlelikeand(12)Thioredoxinlike.

22 Table1.Classificationofdisulphidebondsinconformationalcategories(Schmidtetal, 2006). Disulphidecategory # ' ' -LHSpiral -RHHook + + +/-RHSpiral /-LHSpiral + -RHSpiral /-RHHook RHSpiral LHHook + -/+RHHook RHStaple + +/-LHHook + + -/+LHHook + + +/-LHStaple LHStaple + + +LHSpiral + + +LHHook RHHook /-RHStaple + + +LHStaple RHStaple # LH:Lefthandedoriented;RH:Righthandedoriented;:Negative valuefortherespectivetorsionangle;+:positivevalueforthe respectivetorsionangle.

23 Table 2. Characteristic conformational motifs used for disulphide classification Main motifs Spiral Staple Hook Orientational motifs LH - RH + Peripheral motifs / /+ - +

24 Table 3. Representative structures for the spiral conformational categories. -LHSpiral -RHSpiral +LHSpiral +RHSpiral +/-LHSpiral +/-RHSpiral

25 Table 4. Representative structures for the staple conformational categories. -LHStaple -RHStaple +LHStaple +RHStaple +/-LHStaple +/-RHStaplel

26 Table 5. Representative structures for the hook conformational categories. -LHHook -RHHook +LHHook +RHHook +/-LHHook +/-RHHook -/+LHHook -/+RHHook 24

27 Table 6. Characterization of the protein set under study. The sample used in the statistical analyses is considered to include all the disulphide bonds identified in this protein set. Superfamily Dominant secondary structure No. of PDB Propensity # structures No. of disulphide bonds Function Crisp 5.3 % 6 54 Toxins/defense Cystine-Knot 3.7 % Metabolic Defensin-like 7.4 % Toxins/defense EGF-Laminin 6.4 % Metabolic Omega toxins 8.9 % Toxins/defense Plant lectins 9.9 % Metabolic Small snake toxins 6.5 % Toxins/defense Scorpion-like toxins 7.9 % Toxins/defense BBI (Bowman Birk Inhibitors) 9.6 % 5 33 Protease inhibition BPTI-like 5.1 % Protease inhibition Kringle-like 3.7 % Metabolic Thioredoxin-like 0.8 % Isomerase catalysis # Calculated by equation 2. 25

28 Table 7. Average parameters for the disulphide bonds conformational categories in the sample under study. Representative conformations for structural (-LHSpiral), catalytic (+/-RHHook) and allosteric (-RHStaple) disulphide bonds are represented in bold. Conformational category Frequency DSE/kJ mol -1 d(c -C ')/ d(c -C ')/ -LHSpiral 28.9% RHHook 9.9% /-RHSpiral 8.6% /-LHSpiral 7.9% RHSpiral 7.0% /-RHHook 6.1% RHSpiral 6.0% LHHook 5.2% /+RHHook 4.7% RHStaple 4.0% /-LHHook 2.2% /+LHHook 1.9% /-LHStaple 1.6% LHStaple 1.5% LHSpiral 1.4% LHHook 1.2% RHHook 0.7% /-RHStaple 0.6% LHStaple 0.4% RHStaple 0.1% Least strained # 63.1% Most strained 36.9% # The six conformational categories with the smallest DSE have a grey background. 26

29 Table 8. frequencies for the different conformational categories. Superfamily Categorie Sample -LHSpiral 35.2% 43.8% 26.4% 46.1% 13.8% 29.0% 27.3% 28.3% 12.1% 28.6% 32.7% 3.1% 28.9% -RHHook 0.0% 5.4% 18.9% 8.7% 20.7% 24.0% 6.7% 9.3% 0.0% 4.8% 3.8% 10.8% 9.9% +/-RHSpiral 7.4% 14.3% 7.5% 5.2% 8.0% 20.0% 3.8% 12.1% 3.0% 4.8% 3.8% 1.5% 8.6% +/-LHSpiral 11.1% 4.5% 3.8% 9.6% 2.3% 22.0% 4.3% 2.8% 0.0% 19.0% 30.8% 6.2% 7.9% -RHSpiral 14.8% 9.8% 3.8% 2.6% 3.4% 0.0% 17.2% 3.2% 21.2% 4.8% 1.9% 1.5% 7.0% +/-RHHook 0.0% 3.6% 3.8% 3.5% 2.3% 0.0% 2.9% 7.7% 0.0% 2.4% 0.0% 50.8% 6.1% +RHSpiral 0.0% 3.6% 1.9% 3.5% 4.6% 4.0% 12.9% 9.7% 0.0% 0.0% 1.9% 1.5% 6.0% -LHHook 0.0% 1.8% 7.5% 6.1% 5.7% 1.0% 5.3% 10.1% 6.1% 2.4% 3.8% 1.5% 5.2% -/+RHHook 5.6% 0.0% 0.0% 1.7% 11.5% 0.0% 2.9% 4.9% 12.1% 23.8% 15.4% 0.0% 4.7% -RHStaple 24.1% 0.0% 5.7% 1.7% 4.6% 0.0% 2.4% 0.4% 24.2% 0.0% 0.0% 16.9% 4.0% +/-LHHook 0.0% 0.0% 9.4% 1.7% 5.7% 0.0% 2.9% 2.0% 0.0% 0.0% 1.9% 3.1% 2.2% -/+LHHook 0.0% 1.8% 5.7% 0.0% 3.4% 0.0% 3.8% 2.4% 0.0% 0.0% 0.0% 0.0% 1.9% +/-LHStaple 1.9% 0.9% 0.0% 0.0% 10.3% 0.0% 1.4% 0.0% 6.1% 0.0% 1.9% 3.1% 1.6% -LHStaple 0.0% 0.9% 3.8% 0.9% 2.3% 0.0% 1.0% 1.6% 6.1% 9.5% 0.0% 0.0% 1.5% +LHSpiral 0.0% 8.9% 0.0% 0.0% 1.1% 0.0% 1.0% 1.2% 0.0% 0.0% 0.0% 0.0% 1.4% +LHHook 0.0% 0.0% 1.9% 2.6% 0.0% 0.0% 2.4% 1.2% 6.1% 0.0% 0.0% 0.0% 1.2% +RHHook 0.0% 0.9% 0.0% 1.7% 0.0% 0.0% 0.5% 1.2% 0.0% 0.0% 1.9% 0.0% 0.7% +/-RHStaple 0.0% 0.0% 0.0% 0.9% 0.0% 0.0% 1.4% 0.8% 3.0% 0.0% 0.0% 0.0% 0.6% +LHStaple 0.0% 0.0% 0.0% 2.6% 0.0% 0.0% 0.0% 0.8% 0.0% 0.0% 0.0% 0.0% 0.4% +RHStaple 0.0% 0.0% 0.0% 0.9% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 0.1% 27

30 Table 9. Square Euclidian distances matrix for the cluster space. Cluster % 34.89% 40.29% 45.79% 32.03% 44.03% % 0.00% 8.77% 25.38% 13.42% 21.20% % 8.77% 0.00% 12.00% 11.24% 12.72% % 25.38% 12.00% 6.18% 19.99% 24.61% % 13.42% 11.24% 19.99% 3.98% 19.42% % 21.20% 12.72% 24.61% 19.42% 1.83% 28

31 4. Figure N C 2 ' ' C N C S S C Figure 1

32 4. Figure Figure 2 Frequency 55% 50% 45% 40% 35% 30% 25% 20% 15% 10% 5% 0% Crisp Cystine-Knot Defensin-like EGF-Laminin Omega toxins Plant lectins Small snake toxins Scorpion-like toxins BBI BPT-like Kringle-like Thioredoxin like 15% -LHSpiral -RHHook +/-RHSpiral +/-LHSpiral -RHSpiral +/-RHHook (catalytic) +RHSpiral -LHHook -/+RHHook -RHStaple (allosteric) 10% 5% 0% +RHStaple +LHStaple +/-RHStaple +RHHook +LHHook +LHSpiral -LHStaple +/-LHStaple -/+LHHook +/-LHHook Disulphide conformational categories

33 4. Figure 12 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, , 2, 3, 4, 5, 6, 7, 8, 10, , 3, 4, 5, 6, 7, 8, 10, , 4, 6, 10, 11 3, 5, 7, , 11 2, 4, 6 3, 5, 7, 8 Cluster 1 Cluster 2 Cluster 3 Cluster 6 Cluster 4 Cluster 5 Step 1 Step 2 Step 3 Step 4 Step 5

34 4. Figure 9 Cluster 2 12 Cluster Cluster Cluster Cluster 6 6 Cluster 4