Why do proteases mess up with antigen presentation by re-shuffling antigen sequences?

Why

do

proteases

mess

up

with

antigen

presentation

by

re-shuffling

antigen

sequences?

Juliane

Liepe

,

Huib

Ovaa

and

Michele

Mishto

ThesequenceofalargenumberofMHC-presentedepitopesis

notpresentassuchintheoriginalantigenbecauseithasbeen

re-shuffledbytheproteasomeorotherproteases.Whydo

proteasesthrowaspannerintheworksofourmodelofantigen

taggingandimmunerecognition?Wedescribeinthisreview

whatweknowabouttheimmunologicalrelevanceof

post-translationallysplicedepitopesandwhyproteasesseemto

haveasecond(dark)personality,whichiskeentocreatenew

peptidebonds.

Addresses

1_{Max-Planck-InstituteforBiophysicalChemistry,37077Go¨ttingen,} Germany

2_{DepartmentofChemicalImmunology,LeidenUniversityMedical} Center,NL-2333ZALeiden,TheNetherlands

3_{CentreforInflammationBiologyandCancerImmunology(CIBCI)&} PeterGorerDepartmentofImmunobiology,King’sCollegeLondon, SE11ULLondon,UnitedKingdom

Correspondingauthor:Mishto,Michele([email protected])

CurrentOpinioninImmunology2018,52:81–86

ThisreviewcomesfromathemedissueonAntigenprocessing

EditedbyGennaroDeLibero

ForacompleteoverviewseetheIssue andtheEditorial

https://doi.org/10.1016/j.coi.2018.04.016

0952-7915/ã2018TheAuthors.PublishedbyElsevierLtd.Thisisan openaccessarticleundertheCCBY-NC-NDlicense( http://creative-commons.org/licenses/by-nc-nd/4.0/).

Introduction

Epitopes can have their sequence re-shuffled by

pro-teases, post-translationally modified, trimmed and

bended onto MHC class I (MHC-I) molecules.

Trans-formationscanbesodisguisingthatantigensmighthave trouble evenrecognizing themselves due to these

non-canonical peptides. Nonetheless, the immune system

seemstobeabletoselectivelyidentifythem non-canoni-calepitopesandusethemforpatrollingthestatusofthe cell[1,2].

A growing number of studies about non-canonical

epi-topes has in part whipped out what we learned from

textbooks about antigen presentation. For instance, intrinsiccharacteristics of non-canonical epitopes, espe-cially of those derived from peptide splicing,force the boundariesofourconceptualizationoftheimmunological self [3]. For example, a pre-requisite for streamlined

CD8+Tcellspatrollingbyrecognizingantigenicspliced peptides is that their generation is tightly regulated. Indeed, ifanarbitrarypeptidefragment wereligatedto another fragment we would likely have dramatic

pro-blems during thymocyte selection in the thymus due

to animmensevarietyof splicedpeptidespresentedby corticalandmedullarythymicepithelialcells(cTECsand mTECs, respectively) and othermedullary professional antigenpresentingcells(APCs).Accordingtothethymic selectionmodels[4],onlyahandfulofthymocyteswould survive the negative selection with such an immense antigenic peptides’ variety presented by professional APCs. In agreement with the pre-requisite for stream-linedpatrollingbyCD8+Tcells,thereisagrowingbody of evidence that peptide splicing—and in particular proteasome-catalyzed peptide splicing (PCPS)—is not arandomprocess,andonlyaminorportionofthe theo-retical splicedpeptide isgenerated and presentedto T cells.Whatthesedrivingforcesare,andimplicationsthey can have on the immune response is still to be fully understood.

The MHC-I antigen presentation pathway is on the

contrary welldescribed (Figure1).CD8+Tcellactivity is strongly regulated by which epitopes are presented

onto MHC-I complexes, that is, the MHC-I

immuno-peptidome.AlterationsoftheMHC-Iimmunopeptidome affectthecytotoxicCD8+Tcellresponseagainstviruses andtheefficacyofanti-cancerimmunotherapies[5,6,7]. The immunopeptidome isinfluencedbyseveral factors (seeFigure1)includingantigenavailabilityand charac-teristics [8,9], proteasome processing [1], transport into theendoplasmicreticulum(ER)andloadingtothe pep-tide loading complex (PLC) [10], trimming by ER amino peptidases (ERAPs) [7,11], as well as affinity to differentMHC-Iallotypes’clefts[12].

Part of theMHC-I immunopeptidome canderive from

non-canonical reading frames [13], polymorphic or mutated sequences [2,14,15], non-coding sequences andDRiPs [16,17],orpost-translationallymodified pep-tides [18,19]. The predominant non-canonical peptide populationseemstobe,however,representedbyspliced peptides. Indeed,in theMHC-Iimmunopeptidomesof humanEBV-immortalizedBcellsandprimaryfibroblasts around20–30%ofpeptidesareproducedbyPCPS[20]. Although the average number of molecules of spliced peptidesboundtoMHC-Icomplexesissmallerthanthat of non-spliced peptides [20], spliced epitopes can be

comparable to non-spliced epitopes [21]. For example,

MHC-I-bound spliced epitopes have been found to

prime a specific CD8+ T cell response during Listeria

monocytogenesinfection[22].Furthermore,aspecific acti-vationof CD8+ Tcellstoward splicedepitopesderived from tumor-associated antigens is detectable in mela-noma patients [21] and has led to a regression of the tumormassinamelanomapatientandaleukemiamurine model [23,24]. As a consequence, spliced peptides are interestingnovelcandidatesfor anti-viralvaccine devel-opment[25]andanti-cancerimmunotherapies[1,26].

The

molecular

base

of

the

double

life

of

proteasome

(and

other

proteases?)

in

permanent

balance

between

cleavages

and

ligations

How does it come, however, that the proteasome and possiblyotherproteasesinhumancellsseemtobreakand buildagainpeptidessoefficiently?Wealreadyknowthat PCPSefficiencyispreservedalongevolution[27]andthat therearefactorsthatpromotePCPS.Forinstance,invitro assayssuggestthattheproteasome favorstheligationof somepeptidemotifs [27,28],and wheretheproteasome

preferstocleave,itdoesnotoftensplice[27].However,to understandwhytheproteasomecatalyzessooftenPCPS, weneed someinformationabout itsstructure. The pro-teasome core particle is a barrel-shaped multi-subunit complex. In its internalcavity, it can accumulateup to 200–300smallpeptides[29],or2–3proteins[30,31].Ithas three pairs of catalytic subunits (b1, b2, and b5). The activesite’s threoninenucleophilesfacetheproteasome internalchannel andare surrounded bythe non-primed andprimedsubstrate-bindingsites(Figure2a).The sub-strate degradation rate is driven by the proteolytic-site activity as well as by the peptide transport along the internalchannel[29].The catalyticsubunit substitution, whichisthehallmarkdifferentiatingproteasomeisoforms, affects—atleast at aquantitative level—cleavage-site preferencesandthesubstratedegradationrate[29,32,33]. ItalsoimpactsPCPS,although,likely,onlyina substrate-specificmanner[21,27,34].In cells,PCPScan occur via eithertranspeptidationorcondensation[21,35],although the frequency of the latter mechanismstill needsto be assessed.Ininvitroassays,PCPScanefficientlyoccurby splicingfragmentsderivedfromthesamemolecule—cis PCPS—andfromdifferentmolecules—transPCPS—

Figure1

preferentially from which antigens?

proteasome

antigen fragments: spliced and non-spliced

amino peptidases

TAP MHC-I-calreticulin

Tapasin ERp57

ERAP

PLC

Golgi

cytosol

antigen

preferentially from: long,

polar, acidic, disordered, short half-line, DRiPs antigens

CD8+ T cells

extracellular space secretory

vesicles

ER lumen

Current Opinion in Immunology

(Figure2b)[27,28,36].Accordingtothetranspeptidation model[35],theproteasome’scatalyticN-terminal threo-ninenucleophilebreaksthepeptidebondoftheresidue

(PSP-P1) of the protein—thereby forming an

acyl-enzymeintermediatewiththeN-terminalsplice-reactant, coupledtothereleaseoftheinterveningsequence—and, instead ofcatalyzingthe canonicalpeptide hydrolysis,it catalyzestheligationbetweenthePSP-P1residueofthe N-terminalsplice-reactantwiththeresiduePSP-P10ofthe C-terminalsplice-reactant(Figure2b). Proteasome-medi-ated transpeptidation can alsoresult in isopeptide bond formation when a lysine side chain reacts with an acyl enzyme intermediate. Although it has also been shown thatsuchpeptidescaninduceanimmuneresponse,this processhowever remainstobeobservedinvivo[37].

Proteasome-generated sliced epitopes are however not the only examples of spliced epitopes.Indeed, Delong andcolleagues[38]identifiedsomehybridinsulin pep-tides(HIPs), derivedfromtheligation of afragmentof proinsulinwithpeptidesoriginatingfromotherantigens present inthegranules ofthepancreaticb cells.These

trans spliced epitopes are presented by MHC-II com-plexes,therebytriggeringaspecificresponseinCD4+T cells in type 1 diabetic patients [38,39]. In general, extracellular antigenscan beinternalized, processed by

proteolysis in the lysosome, bind the MHC class II

molecules,andthencirculatetothecellsurfaceandback

to the lysosome [40].Although thelysosomal proteases also rely on(thio)ester intermediates, theenzyme cata-lyzing theproductionof theHIPs isstillunknown.

Whydoproteases(frequently)behaveasligases,too?In principle, any proteasethatusesa nucleophileto pro-motehydrolysisthroughanesterintermediatecan cat-alyzetranspeptidation.Henceanyproteasecouldplaya roleintheformationofsplicedpeptides.Althoughitis notunderstoodwhytheproteasomeinparticularseems tocatalyzethisprocesssoefficiently,wecanspeculate on the reasons. Transpeptidation efficiency depends highly on three factors: firstly, high concentration of theaminenucleophilemustbepresentinordertofavor theformation ofanovelpeptide-bondoverhydrolysis; secondly,theesterneedstohaveasufficientlifetimein ordertoallowthereactionandpeptidebondformation over hydrolysis; thirdly, the active site in which this esterintermediateisformedmustbesufficiently acces-sible for nucleophilesto react.The proteasome struc-turecanfavorallthesethreeconditions,asit’saclosed barrel that could have a high local concentration of peptide products, and use substrate-binding sites in proximitytotheproteasome’scatalyticN-terminal thre-oninenucleophile.Anotherresultofpeptides confine-ment in the proteasome barrel could be the fact that

transPCPSseemstooccurlessfrequentlythancisPCPS [27,28,36].

16nm

gates

10nm

N

N N

N

C C

C C

antechamber

(59nm3₎ chamber_(84nm3₎ antechamber_(59nm3₎

PSP-P1

PSP-P1

cleavages by Thr1

spliced peptides

trans PCPS

PSP-P1 PSP-P1′

PSP-P1′

PSP-P1′

normal cis PCPS reverse cis PCPS intervening sequences

(a) (b)

Current Opinion in Immunology

Molecularbasefortheunexpectedhighfrequencyofpeptidesplicing.(a)Thehuman20Scoreparticleoftheproteasomeisshownbasedonthe structuregeneratedby[45].ThechainsB,C,H,I,J,Q,R,S,YandZarehiddenfromthestructureinordertoseetheinnerproteasomecavities withthecentralchamberanditstwoantechambers.The_aand_bsubunitsarecoloredingreyandblue,respectively.Asanexampleofacatalytic subunit,heb2subunitisshowninpinkwithitsactivesitethreonineinred.(b)Proteasome-generatedsplicedpeptidescanbeformedby:firstly,

cisPCPS,whenthetwosplice-reactantsderivefromthesamepolypeptidemolecule;theligationcanoccurinnormalorder,thatis,followingthe orientationfromN-terminustoC-terminusoftheparentalprotein(normalcisPCPS),orinthereverseorder(reversecisPCPS);secondly,trans

Theoretical

impact

of

PCPS

in

recognizing

the

immunological

self

One majorfeatureof peptidesplicing isthe theoretical increaseofthenumberofsequencesthatcanbederived fromtheantigenpoolandbeallocatedintheMHC-Iand MHC-II clefts. This enlargement could have implica-tionsin therecognitionof theimmunological selfbyT cells.Indeed,itcouldincreasetheriskofmimicry,which isthephenomenonwherebytwoepitopeshavesequence similaritiesandarerecognizedbythesameTcellclone [1].Inparticular,wenameas‘zwitterpeptide’apeptide thatcanbederivedfromthehumanself-proteomeaswell as fromapathogenproteome (Figure 3).In2012,Calis

et al. [41] investigated the sequence overlap between humanself-peptidesandalargesetofnon-self-peptides derivedfromvirusesandbacteriainthecontextofCD8+ Tcellrecognition. Theyfound thatless than 1%of all theoretical possible 9-mernon-spliced peptidesderived frompathogenshaveasequenceidenticaltothe theoret-icalhumannon-splicedpeptides,thatis,arezwitter pep-tides. If the zwitter antigenic peptides were presented similarlybymTECsandotherprofessionalAPCsinthe medullary thymus and by dendritic cells (DCs) in the periphery,wewouldexpecttheabsence,attheperiphery, ofCD8+T-cellclonesrecognizing,withhighaffinity,the

zwitter peptides presented by DCs, because they have beeneliminatedduringthethymicnegativeselection[4]. Thisphenomenoncouldinpartexplaintheoccurrenceof holesintheTcellrepertoireandintheirabilitytotackle

viralinfections [41].Onthecontrary,if thezwitter anti-genic peptides were efficiently presented by DCs and otherAPCsintheperipherybutnotbymTECsandother

professional APCs in the thymic medulla, we would

expectattheperipherythepresenceofpotentially auto-reactiveCD8+T-cellclones,whichcouldbeprimedand activatedbyDCsandotherAPCsinlymphnodesduring thepathogeninfectionandafterwardsattackhumancells andparticipateto inanautoimmuneresponse.

Inmultiplesclerosis,forinstance,myelin-reactiveCD8+ T cell are theorized to mediate the cytotoxic activity againsttheoligodendrocytesleadingtothecharacteristic de-myelinationandplaqueformation.Furthermore, asso-ciationsbetweenmultiplesclerosis,someMHC-Ivariants (e.g.HLA-B*07)andEpstein-Barrvirus(EBV)infection havebeenreported,andithasbeenhypothesizedthatan EBVinfection couldtriggertheprimingofautoreactive CD8+Tcellclonesthroughmimicry[42].Usingasimilar approachasCalisetal.[41],wecancomparetheoverlapof theoretical9merpeptides(eithersplicedornon-spliced)

derived from 24 human myelin proteins (MBP, MAG,

MOG,PLPandisoforms)andfrom9EBVantigens(i.e.

LMP1, LMP2, BMLF1, BMRF1, BZLF1, BRLF1,

BNRF1,BLLF1,EBNA3).All27peptidestheoretically common to myelin and EBVantigens are spliced pep-tides, since there are no identical non-spliced peptide sequencesbetweenthesetwosetsofantigens.Amongthe 27theoreticalzwitterpeptides,13peptidesarepredicted

Figure3

Pathogen: EBV Host: Human

EBV latent membrane protein 1 (LMP1) Myelin oligodendrocyte glycoprotein (MOG)

zwitter spliced peptide

IC₅₀ (HLA-B*07:02) = 58 nM

Current Opinion in Immunology

accountsfor onlythetheoretical presenceor absenceof apeptidein theimmunopeptidomeof APCs.To better estimatetherealprevalenceofzwitterantigenicpeptides andtheirrecognitionbyCD8+Tcells,weshouldconsider the TCR degeneracy, theaffinity/avidity of TCRs and MHC-I-peptides,andthedynamicsofthedifferentsteps of theMHC-Iantigenpresentation(Figure1)including the,onlypartiallydescribed,drivingforcesofPCPS.This preliminaryinsilicoresult,however,confirmsthatPCPS could play a particularly relevant role in the central tolerance, the occurrence of large holes in the T cell repertoire, and the autoimmune response mediated by CD8+Tcells.

Concluding

remarks

Thesurprisingevidencesreportedinthelastyears,which

suggestthatMHC-I(andinpartMHC-II)

immunopep-tidomes are populated by spliced peptides, need to be confirmedbyapplyingdifferentapproachesbefore under-standing the magnitude of their immunological rele-vance. However, the implications of peptide splicing couldexceedtheedgesof antigenpresentation.If pep-tidesplicingwereacommonreactionforotherproteases rather than proteasome, we could speculate that post-translationally spliced peptides (and why not spliced proteins?) could be involved in other aspects of the immune response and cell metabolism, as it has been proved for proteasome-processed non-spliced peptides and proteins [43,44]. If this hypothesis were correct, peptide splicing could be a further regulatory layer in thelifeofacellandanorganism.

Conflict

of

interest

statement

Acknowledgements

WethankDarijaMuharemagicforproofreadingthemanuscript.Workin theOvaalabissupportedbytheInstituteforChemicalImmunology,a gravitationprogrammefinancedtheNetherlandsFoundationforScientific Research(N.W.O.).

References

and

recommended

reading

Papersofparticularinterest,publishedwithintheperiodofreview, havebeenhighlightedas:

ofspecialinterest ofoutstandinginterest

1. MishtoM,LiepeJ:Post-translationalpeptidesplicingandTcell responses.TrendsImmunol2017,38:904-915.

2. GranadosDP,LaumontCM,ThibaultP,PerreaultC:Thenatureof selfforTcells-asystems-levelperspective.CurrOpinImmunol 2015,34:1-8.

3. GrignolioA,MishtoM,FariaAM,GaragnaniP,FranceschiC, TieriP:Towardsaliquidself:howtime,geography,andlife experiencesreshapethebiologicalidentity.FrontImmunol 2014,5:153.

5.

TranPasettoE,RobbinsA,ZhengPF,Z,RayLuS,YC,GrohPrickettEMetTD,al.:GartnerT-celltransferJJ,JiaL,therapy targetingmutantKRASincancer.NEnglJMed2016,

375:2255-2262.

Here,theauthorsshowthatthetreatmentofmetastaticcolorectalcancer byadoptiveTcelltherapytargetingtherecurrentdrivermutationKRAS G12Dleadstothemetastasiseradication.

6. TenzerS,WeeE,BurgevinA,Stewart-JonesG,FriisL, LamberthK,ChangCH,HarndahlM,WeimershausM,GerstoftJ etal.:AntigenprocessinginfluencesHIV-specificcytotoxicT lymphocyteimmunodominance.NatImmunol2009,

10:636-646.

7. TextorA,SchmidtK,KloetzelPM,WeissbrichB,PerezC,CharoJ, AndersK,SidneyJ,SetteA,SchumacherTNetal.:Preventing tumorescapebytargetingapost-proteasomaltrimming independentepitope.JExpMed2016,213:2333-2348.

8. PearsonH,DaoudaT,GranadosDP,DuretteC,BonneilE, CourcellesM,RodenbrockA,LaverdureJP,CoteC,MaderSetal.:

MHCclassI-associatedpeptidesderivefromselective regionsofthehumangenome.JClinInvest2016,

126:4690-4701.

9. HoofI,vanBaarleD,HildebrandWH,KesmirC:Proteome samplingbytheHLAclassIantigenprocessingpathway.PLoS ComputBiol2012,8:e1002517.

10.

BleesTrowitzschA,JanulieneS,MoellerD,HofmannA,TampeT,R:KollerStructureN,SchmidtoftheC,human MHC-Ipeptide-loadingcomplex.Nature2017.

Here,theauthorsdeterminethestructureofthehumannativeMHC-I peptide-loadingcomplexbyelectroncryo-microscopy,therebyproviding insightsinthedynamicsoftheMHC-I-peptideassembly.

11. ChenH,LiL,WeimershausM,EvnouchidouI,vanEndertP, BouvierM:ERAP1-ERAP2dimerstrimMHCI-boundprecursor peptides;implicationsforunderstandingpeptideediting.Sci Rep2016,6:28902.

12. VaughanK,XuX,CaronE,PetersB,SetteA:Decipheringthe MHC-associatedpeptidome:areviewofnaturallyprocessed liganddata.ExpertRevProteomics2017,14:729-736.

13. LaumontCM,DaoudaT,LaverdureJP,BonneilE, Caron-LizotteO,HardyMP,GranadosDP,DuretteC,LemieuxS, ThibaultPetal.:GlobalproteogenomicanalysisofhumanMHC classI-associatedpeptidesderivedfromnon-canonical readingframes.NatCommun2016,7:10238.

14. Bassani-SternbergM,Pletscher-FrankildS,JensenLJ,MannM:

MassspectrometryofhumanleukocyteantigenclassI peptidomesrevealsstrongeffectsofproteinabundanceand turnoveronantigenpresentation.MolCellProteomics2015,

14:658-673.

15. Bassani-SternbergM,BraunleinE,KlarR,EngleitnerT,SinitcynP, AudehmS,StraubM,WeberJ,Slotta-HuspeninaJ,SpechtK etal.:Directidentificationofclinicallyrelevantneoepitopes presentedonnativehumanmelanomatissuebymass spectrometry.NatCommun2016,7:13404.

16. ApcherS,MillotG,DaskalogianniC,ScherlA,ManouryB, FahraeusR:Translationofpre-splicedRNAsinthenuclear compartmentgeneratespeptidesfortheMHCclassI pathway.ProcNatlAcadSciUSA2013,110:17951-17956.

17. KrachtMJ,vanLummelM,NikolicT,JoostenAM,LabanS,van derSlikAR,vanVeelenPA,CarlottiF,deKoningEJ,HoebenRC etal.:Autoimmunityagainstadefectiveribosomalinsulingene productintype1diabetes.NatMed2017,23:501-507.

18. AlpizarA,MarinoF,Ramos-FernandezA,LombardiaM,JekoA, PazosF,ParadelaA,SantiagoC,HeckAJ,MarcillaM:A molecularbasisforthepresentationofphosphorylated peptidesbyHLA-Bantigens.MolCellProteomics2017,

16:181-193.

breakingimmunetoleranceinautoimmunedisease? J Autoimmun2017,84:12-20.

20.

LiepeKloetzelJ,PM,MarinoStumpfF,SidneyMP,HeckJ,JekoAJ,A,MishtoBuntingM:ADE,largeSettefractionA, of HLAclassIligandsareproteasome-generatedspliced peptides.Science2016,354:354-358.

Inthisstudy,theauthorsidentifyforthefirsttimethelargepopulationof splicedpeptidesintheMHC-Iimmunopeptidomeofnon-tumoralcells.In thosecells,splicedpeptidesrepresentaround30%ofthepeptidevariety, onefourthoftheiramountandexclusivelypresentaroundonethirdofthe antigens.

21. EbsteinF,Textoris-TaubeK,KellerC,GolnikR,VigneronN,Van denEyndeBJ,Schuler-ThurnerB,SchadendorfD,LorenzFK, UckertWetal.:Proteasomesgeneratesplicedepitopesbytwo differentmechanismsandasefficientlyasnon-spliced epitopes.SciRep2016,6:24032.

22.

PlatteelSchalkwijkACM,HH,LiepeCardosoJ,Textoris-TaubeR,KloetzelPM,K,MishtoKellerM,C,HenkleinSijtsA:Multi-P, levelstrategyforidentifyingproteasome-catalyzedspliced epitopestargetedbyCD8+Tcellsduringbacterialinfection. CellRep2017,20:1242-1253.

Here,theauthorsprove,forthefirsttime,thatproteasome-generated spliced epitopes trigger a specificactivation of CD8+ T cellsduring Listeriainfection,therebyaddressingthecytotoxicresponseagainsta bacteriaantigenotherwisenotimmunogenicinthatmousemodel.

23. WarrenEH,VigneronNJ,GavinMA,CouliePG,StroobantV, DaletA,TykodiSS,XuerebSM,MitoJK,RiddellSRetal.:An antigenproducedbysplicingofnoncontiguouspeptidesinthe reverseorder.Science2006,313:1444-1447.

24. DaletA,RobbinsPF,StroobantV,VigneronN,LiYF,El-GamilM, HanadaKI,YangJC,RosenbergSA,VandenEyndeBJ:An antigenicpeptideproducedbyreversesplicinganddouble asparaginedeamidation.ProcNatlAcadSciUSA2011,108: E323-E331.

25. PlatteelACM,LiepeJ,vanEdenW,MishtoM,SijtsA:An unexpectedmajorroleforproteasome-catalyzedpeptide splicingingenerationofTcellepitopes:isthererelevancefor vaccinedevelopment? FrontImmunol2017,8:1441.

26. MeliefCJM,KesslerJH:NovelinsightsintotheHLAclassI immunopeptidomeandT-cellimmunosurveillance.Genome Med2017,9:44.

27. MishtoM,GoedeA,TaubeKT,KellerC,JanekK,HenkleinP, NiewiendaA,KlossA,GohlkeS,DahlmannBetal.:Drivingforces ofproteasome-catalyzedpeptidesplicinginyeastand humans.MolCellProteomics2012,11:1008-1023.

28. BerkersCR,deJongA,SchuurmanKG,LinnemannC,MeiringHD, JanssenL,NeefjesJJ,SchumacherTN,RodenkoB,OvaaH:

Definitionofproteasomalpeptidesplicingrulesfor high-efficiencysplicedpeptidepresentationbyMHCclassI molecules.JImmunol2015,195:4085-4095.

29. LiepeJ,HolzhutterHG,BellavistaE,KloetzelPM,StumpfMP, MishtoM:Quantitativetime-resolvedanalysisreveals intricate,differentialregulationofstandard-and immuno-proteasomes.Elife2015:4.

30. SharonM,WittS,FeldererK,RockelB,BaumeisterW, RobinsonCV:20Sproteasomeshavethepotentialtokeep substratesinstoreforcontinualdegradation.JBiolChem 2006,281:9569-9575.

31. HutschenreiterS,TinazliA,ModelK,TampeR:Two-substrate associationwiththe20Sproteasomeatsingle-moleculelevel. EmboJ2004,23:2488-2497.

32. MishtoM,LiepeJ,Textoris-TaubeK,KellerC,HenkleinP, WeberrussM,DahlmannB,EnenkelC,VoigtA,KuckelkornU etal.:Proteasomeisoformsexhibitonlyquantitative

differencesincleavageandepitopegeneration.EurJImmunol 2014,44:3508-3521.

33. ArciniegaM,BeckP,LangeOF,GrollM,HuberR:Differential globalstructuralchangesinthecoreparticleofyeastand mouseproteasomeinducedbyligandbinding.ProcNatlAcad SciUSA2014,111:9479-9484.

34. DaletA,StroobantV,VigneronN,VandenEyndeBJ:Differences intheproductionofsplicedantigenicpeptidesbythestandard proteasomeandtheimmunoproteasome.EurJImmunol2011,

41:39-46.

35. VigneronN,StroobantV,ChapiroJ,OomsA,DegiovanniG, MorelS,vanderBruggenP,BoonT,VandenEyndeBJ:An antigenicpeptideproducedbypeptidesplicinginthe proteasome.Science2004,304:587-590.

36. DaletA,VigneronN,StroobantV,HanadaK,VandenEyndeBJ:

Splicingofdistantpeptidefragmentsoccursinthe proteasomebytranspeptidationandproducesthespliced antigenicpeptidederivedfromfibroblastgrowthfactor-5.J Immunol2010,184:3016-3024.

37. BerkersCR,deJongA,SchuurmanKG,LinnemannC, GeenevasenJA,SchumacherTN,RodenkoB,OvaaH:Peptide splicingintheproteasomecreatesanoveltypeofantigenwith anisopeptidelinkage.JImmunol2015,195:4075-4084.

38.

DelongArmstrongT,WilesM,PowellTA,BakerRL,ReisdorphRL,BradleyN,KumarB,BarbourNetG,al.:ReisdorphPathogenicR, CD4Tcellsintype1diabetesrecognizeepitopesformedby peptidefusion.Science2016,351:711-714.

Inthisstudy,theauthorsidentifyforthefirsttimesplicedepitopesthat derivefromthefusionofaportionofinsulinandaportionofotherantigens presentinthepancreaticbcells.Theseso-calledHybridInsulinPeptides trigger the activation of CD4+ T cellsderived from type 1 diabetes patients.

39. BabonJA,DeNicolaME,BlodgettDM,CrevecoeurI,ButtrickTS, MaehrR,BottinoR,NajiA,KaddisJ,ElyamanWetal.:Analysisof self-antigenspecificityofislet-infiltratingTcellsfromhuman donorswithtype1diabetes.NatMed2016,22:1482-1487.

40. RockKL,ReitsE,NeefjesJ:Presentyourself!ByMHCclassI andMHCclassIImolecules.TrendsImmunol2016,37:724-737.

41. CalisJJ,deBoerRJ,KesmirC:DegenerateT-cellrecognitionof peptidesonMHCmoleculescreateslargeholesintheT-cell repertoire.PLoSComputBiol2012,8:e1002412.

42. GeginatJ,ParoniM,PaganiM,GalimbertiD,DeFrancescoR, ScarpiniE,AbrignaniS:Theenigmaticroleofvirusesinmultiple sclerosis:molecularmimicryordisturbedimmune

surveillance? TrendsImmunol2017,38:498-512.

43. DianzaniC,BellavistaE,LiepeJ,VerderioC,MartucciM, SantoroA,ChiocchettiA,GigliottiCL,BoggioE,FerraraBetal.:

Extracellularproteasome-osteopontincircuitregulatescell migrationwithimplicationsinmultiplesclerosis.SciRep2017,

7:43718.

44. Kravtsova-IvantsivY,CiechanoverA:Theubiquitin-proteasome systemandactivationofNF-kappaB:involvementofthe ubiquitinligaseKPC1inp105processingandtumor suppression.MolCellOncol2015,2:e1054552.

45. SchraderJ,HennebergF,MataRA,TittmannK,SchneiderTR, StarkH,BourenkovG,ChariA:Theinhibitionmechanismof human20Sproteasomesenablesnext-generationinhibitor design.Science2016,353:594-598.

46. FleriW,PaulS,DhandaSK,MahajanS,XuX,PetersB,SetteA: