Related Commentary, page 869

Research article

Antibodies against citrullinated proteins enhance tissue injury in experimental autoimmune arthritis
Kristine A. Kuhn,1 Liudmila Kulik,2 Beren Tomooka,3,4 Kristin J. Braschler,2 William P. Arend,2 William H. Robinson,3,4 and V. Michael Holers1,2
3Division 1Department

of Immunology and 2Department of Medicine, University of Colorado Health Sciences Center, Denver, Colorado, USA. of Immunology and Rheumatology, Department of Medicine, Stanford University School of Medicine, Stanford, California, USA. 4Geriatric Research, Education and Clinical Center, Palo Alto Veterans Affairs Health Care System, Palo Alto, California, USA.

Antibodies against citrullinated proteins are specific and predictive markers for rheumatoid arthritis although the pathologic relevance of these antibodies remains unclear. To investigate the significance of these autoantibodies, collagen-induced arthritis (CIA) in mice was used to establish an animal model of antibody reactivity to citrullinated proteins. DBA/1J mice were immunized with bovine type II collagen (CII) at days 0 and 21, and serum was collected every 7 days for analysis. Antibodies against both CII and cyclic citrullinated peptide, one such citrullinated antigen, appeared early after immunization, before joint swelling was observed. Further, these antibodies demonstrated specific binding to citrullinated filaggrin in rat esophagus by indirect immunofluorescence and citrullinated fibrinogen by Western blot. To evaluate the role of immune responses to citrullinated proteins in CIA, mice were tolerized with a citrulline-containing peptide, followed by antigen challenge with CII. Tolerized mice demonstrated significantly reduced disease severity and incidence compared with controls. We also identified novel murine monoclonal antibodies specific to citrullinated fibrinogen that enhanced arthritis when coadministered with a submaximal dose of anti-CII antibodies and bound targets within the inflamed synovium of mice with CIA. These results demonstrate that antibodies against citrullinated proteins are centrally involved in the pathogenesis of autoimmune arthritis.
Introduction Thediagnosisandclassificationofrheumatoidarthritis(RA)is basedonclinicalandradiologicalfindingswiththesupportofa positivetestforrheumatoidfactor(RF)(1),anautoantibodyto theconstantregionofIgGimmunoglobulin.WhileRFisfairly sensitiveforRA,detectedinupto80%ofpatients,thisautoantibodylacksspecificity.Serafrompatientswithotherautoimmune disordersorinfectiousdiseasesaswellasupto15%ofthehealthy, elderlypopulationdemonstratethepresenceofRF(2).Otherless sensitivebutmorespecificautoantibodies,suchastheanti-perinuclearfactorandanti-keratinantibody(AKA),alsohavebeen describedinpatientswithRA(25).Importantly,uponfurther investigation,theanti-perinuclearfactorandAKAwereshownto bindaspecificprotein,filaggrin,foundwithinkeratohyalingranulesandcornifiedepithelium(6,7). Filaggrinisproducedinthelatestagesofterminaldifferentiationofepithelialcellsduringkeratinization.Theprecursorprofilaggrinisproducedandstoredingranulesuntilterminaldifferentiation.Atthispoint,profilaggrinisproteolyticallycleavedand dephosphorylated,andcertainarginineresiduesaredeiminated byspecificpeptidylargininedeiminases(PADs)toformcitrulline residues(811).Whenarginineresidueswithinpeptidesequences derivedfromfilaggrinweredeiminatedinvitro,thesepeptides
Nonstandard abbreviations used:ABTS,2,2-azinobis(3-ethylbenzthiazoline-6sulfonicacid);AKA,anti-keratinantibody;APF,anti-perinuclearfactor;CCP,cyclic citrullinatedpeptide;CIA,collagen-inducedarthritis;CII,bovinetypeIIcollagen;LKP, linearlysinepeptide;LRP,linearargininepeptide;LXP,linearcitrullinatedpeptide; PAD,peptidylargininedeiminase;RF,rheumatoidfactor. Conflict of interest:Theauthorshavedeclaredthatnoconflictofinterestexists. Citation for this article:J. Clin. Invest.116:961973(2006).doi:10.1172/JCI25422.

wererecognizedbyRAsera(12,13).Usingthemoststronglyreactivepeptide,designatedcycliccitrullinatedpeptide(CCP),an ELISAthatmeasuredantibodiesagainstCCPdemonstrated68% sensitivityand98%specificityforRA(14).Furtherstudiesindicatedthatanti-CCPantibodiesmaybepresentformanyyearsbefore theonsetofclinicallyapparentdisease(15,16). Filaggrinisnotexpressedinthesynovium.However,withinthe inflamedsynoviumofpatientswithRA,onepotentiallyrelevant antigen is citrullinated fibrin. Antibodies from patients with RAbindfibrinonWesternblotanalysisofrheumatoidsynovial extracts.Inaddition,antibodiesfrompatientswithRApreferentiallybindcitrullinatedfibrinascomparedwithnoncitrullinated fibrin(17).Othercitrullinatedproteins,suchasvimentin(18,19), arealsorecognizedbyserafrompatientswithRAinacitrulline-specificmanner.Thus,theseantibodiesarespecifictotheposttranslationalcitrullinationandrecognizethisaminoacidinthecontextof flankingaminoacidsequencesinatleast3proteins.Despitethese clinicalassociations,theimmunopathologiceventsthatleadtothe developmentofautoantibodiesagainstcitrullinatedproteinsand whethertheseautoantibodiesarepathogenicremainunknown. Collagen-inducedarthritis(CIA)isananimalmodelofinflammatoryarthritisestablishedbyimmunizingmicewithbovine typeIIcollagen(CII)emulsifiedinincompleteFreundsadjuvant containingMycobacterium tuberculosis (20).BecausetheimmunologicandhistopathologiclesionsinCIAresemblethoseobserved inRA(20),CIAisonemodelinwhichthepathogenesisofRAcan bestudied.Theinitialprocessesofdiseasepathogenesisrequire TcellssinceTcelldeficientmicearenotsusceptibletoCIA(21, 22).Inaddition,highlyefficientpresentationofthecollagen antigenrequirestheI-AqMHCclassIIallele(23,24).Tcellacti961

The Journal of Clinical Investigation http://www.jci.org Volume116 Number4 April2006

research article

Figure 1
Anti-CCP antibodies appear prior to visible arthritis and in parallel with the appearance of anti-collagen antibodies. (A) Six- to 8-week-old DBA/1J mice were untreated or were immunized at day 0 with CII in CFA or CFA alone and then received booster immunizations 21 days later (black arrows). Sera were collected every 7 days and analyzed for antibody levels. Mice were also observed and scored for development of clinical arthritis. Visible signs of arthritis were measured by adding the number of swollen digits on all 4 paws of each mouse. (B) IgG and (D) isotype-specific antibodies against CII were measured by ELISA as described in Methods. (C) IgG and (E) isotype-specific anti-CCP antibodies were measured by ELISA as described in Methods. IgG3 anti-CCP antibodies were below the limits of detection by ELISA and thus are not shown. The data for each mouse within a group are averaged for that group and represented SEM. For each group, n = 12. Statistical significance was determined using a 1-way ANOVA. *P < 0.05; **P < 0.01; ***P < 0.001 in CIA mice compared with CFA-immunized and unimmunized mice. (F) Sera from Rag1/ mice (n = 5) were tested and compared with unimmunized DBA/1J (normal mouse serum (NMS); n = 7 for IgM and n = 23 for IgG). Statistical significance was determined using an unpaired 2-tailed Students t test. #P = 0.0004 for IgM anti-CCP antibodies in NMS versus Rag1/ mice.

vationleadstoaninflammatoryTh1responsebothsystemically andwithinthejoints(2527).ChronicinflammationinCIAis largelyduetocomplement-fixinganti-CIIantibodieswhoseproductionisdrivenbyautoreactiveTcells.Anti-CIIantibodiesin CIAinitiatejointdamagebybindingcollageninarticularcartilageandactivatingcomplement. BecausetherelevanceofautoantibodiesreactivewithcitrullinatedantigenstothepathophysiologyofRAremainsunknown andavalidanimalmodelinwhichtostudytheseantibodiescould providesubstantialinsightsintotheseissues,weinvestigatedthe developmentandpathogenicpotentialoftheseantibodiesin murineCIA.Wedemonstratehereinthepresenceofanti-citrullinatedproteinantibodiesinCIAandshowthattheyplayakey roleinthemaximaldevelopmentofinflammatoryarthritis.In addition,autoantibodiesagainstcitrullinatedproteinscangreatly amplifytissueinjuryinthepresenceofasubmaximalimmuneand inflammatoryinsult. Results Antibodies specific to citrullinated antigens develop in CIA in parallel with anti-CII antibodies and before visible arthritis.Todetermineifantibodies againstcitrullinatedantigensarepresentinCIA,weimmunized6- to8-week-oldmaleDBA/1JmicewithCIIemulsifiedinCFAorwith CFAaloneondays0and21.Beginningat25daysaftertheinitial immunizationwithCIIinCFA,micedevelopedclinicalevidenceof arthritis(Figure1A).Serumwascollectedfrommiceevery7days followingthefirstimmunizationandwasanalyzedbyELISAfor
962

thedevelopmentofautoantibodiesagainstCIIandCCP(Figure1, BandC).Anti-CIIantibodiesbegantodeveloparoundday7after thefirstimmunizationandsteadilyincreasedthroughoutthe study.Similarly,anti-CCPreactivityappearedat7daysafterthe firstimmunizationandplateauedatday21.Inbothcases,theautoantibodiesappearedpriortoanysignsofclinicaldisease. TofurthercharacterizethedevelopmentofCCPreactivityin CIA,weanalyzedtheisotypesofCII-andCCP-reactiveantibodies.StrongIgMandIgG2aanti-CIIantibodiesdevelopedafter immunizationalongwithlesspronouncedIgG1,IgG2b,andIgG3 responses(Figure1D),consistentwithpreviousreports(28).AntiCCPantibodiesoftheIgM,IgG1,IgG2a,andIgG2bisotypesdevelopedafterimmunizationwithCII(Figure1E)whileIgG3wasnot detected(datanotshown).Intriguingly,anti-CCPIgMantibodies werereadilydetectableinpreimmunesera,suggestingthepresence ofnaturalantibodiesinmicetowardacitrullinatedantigen,and increasedsignificantlyafterimmunization(P<0.05atday14and thereafter).ToensurethattherelativelyhighIgManti-CCPtiterin naivemicewasduetoantibodyandnottononspecificserumreactivitydetectedbyELISA,wetestedserumfromRag1/mice,which aregeneticallyunabletogenerateantibodies.Rag1/micedemonstratedsignificantlyloweranti-CCPIgMreactivitycompared withunimmunizedDBA/1Jmice(Figure1F).Nodifferenceinthe levelofanti-CCPIgGwasdetectedbetweenRag1/seraandsera fromunimmunizedDBA/1Jmice,indicatingthatapproximately 7U/mlisthelowerlimitofdetectionforanalysisofmousesera withtheCCPELISA(Figure1F).

The Journal of Clinical Investigation http://www.jci.org Volume116 Number4 April2006

research article
Figure 2
A population of antibodies in CIA is specific for the citrullinated proteins filaggrin and fibrinogen. (A) Sera from an RA patient; (B) CIA, day 35 after immunization; (C) adjuvant control; or (D) an untreated mouse were applied to tissue sections of rat esophagus. Arrows indicate areas of reactivity on epithelium. (E) Mouse sera from arthritic mice on day 35 after immunization were preabsorbed with citrullinated peptide (LXP) or (F) control peptide (LRP). These were then applied to tissue sections of rat esophagus. Serum antibody binding was detected with FITC-conjugated goat anti-mouse IgG. Magnification, 40. (G) Western blot of unmodified (un) and citrullinated (cit) fibrinogen probed with RA patient or mouse sera. In second and third lanes, proteins are stained with Coomassie to demonstrate relatively equal transfer for Western blot analysis.

Oneadditionalmeansbywhichcitrullinatedproteinreactivity isdetectedinpatientswithRAisindirectimmunofluorescenceon ratesophagealepithelium.IthasbeenshownthatAKA,anti-filaggrin,andanti-CCPantibodiesinRApatientserareactwithcitrullinatedfilaggrinintheratesophagealepithelium(7,12).Toassess thisantibodyreactivityinCIA,weappliedserafromarthriticand controlmicetosectionsofratesophagus.Serafromarthriticmice onday35aftertheinitialimmunizationdemonstratedastaining patternsimilartothatseenwithRAserawhileserafromadjuvant controlandunimmunizedmicedidnotdemonstratesimilarreactivity(Figure2,AD). Observed CCP reactivity in CIA is specific to the citrulline modification. ThekeyfeatureofreactivityinserumfrompatientswithRAtothese antigensisitsmarkedspecificitytowardcitrullinatedepitopesrather thantowardtheunmodifiedproteinthatlackscitrullineresidues (12,14).ThespecificityofantibodybindinginCIAserawasevaluatedusingpeptideblockingofantibodyreactivity.Serumobtained onday35aftertheinitialimmunizationwithCIIinCFAwasdiluted inPBSorPBScontaining100g/mllinearcitrullinatedpeptide (LXP)or100g/mlofthecontrol,linearargininepeptide(LRP),and appliedtoratesophagealtissuesections.TheLXPinhibitedreactivityinbothRAsera(datanotshown)andCIAserawhereasthecontrolLRPpeptidedidnot(Figure2,EandF).Theseresultsstrongly suggestthatthereactivityinmicewithCIAtofilaggrininthistissue isspecifictocitrullinatedfilaggrin,asseeninpatientswithRA. Citrullinatedfibrinhasbeenidentifiedasatargetantigenwithin theinflamedRAsynovium,andantibodiesspecifictocitrullinated

fibrinarereadilydetectedintheseraofRApatients(17).Therefore, asasecondmeasureforcitrullinespecificityoftheantibodiesin CIA,wetestedseraforreactivityagainstcitrullinatedfibrinogen. PooledserafromindividualswithRA,frommiceimmunizedwith CIIinCFAorwithCFAonly,orfromnormalmouseserumwere studiedbyWesternblotanalysisusingunmodifiedandcitrullinatedfibrinogen.TheonlyreactivityobservedwasagainstcitrullinatedfibrinogeninserafrompatientswithRAandmicewithCIA (Figure2G),providingadditionalsupportforthespecificityofthe autoantibodiesinCIAforcitrullinatedantigens. Some lupus models demonstrate anti-CCP reactivity but without citrulline specificity.Anotherstrikingfeatureofreactivitytocitrullinatedantigensisthespecificityofthisautoantibodyresponsefor patientswithRA.Otherformsofarthritisandautoimmunediseaseslacktheseantibodies(14).Toevaluatethespecificityofantibodiesagainstcitrulline-modifiedtargetsforCIA,wemeasured CCPreactivityin2murinemodelsoflupus:SNF1andMRL/lpr. Inagreementwithotherstudies(29),MRL/lprmiceat4months ofage(n=5)producedhightitersofanti-CCPantibodies(SupplementalFigure1,AandB;supplementalmaterialavailableonline withthisarticle;doi:10.1172/JCI25422DS1).Anti-CCPantibodieswerenotobservedin4-month-oldSNF1mice(n=4).Indirect immunofluorescenceindicatedthepresenceofMRL/lprserum reactivityintheepitheliumofratesophagus;however,incontrast tomicewithCIA,bothLRPandLXPinhibitedthisserumbinding(SupplementalFigure1C).Thisobservationsuggestedthat theanti-CCPreactivityobservedinMRL/lprmicedidnotexhibit specificityforcitrullinatedfilaggrin,thusdifferingfromtheautoantibodypatternobservedinCIA.Inaddition,aWesternblot ofcitrullinatedandunmodifiedfibrinogenprobedwithpooled MRL/lprserafailedtodemonstrateserumreactivitywitheither formoffibrinogen(datanotshown),furthersupportingadifferencebetweenthespecificityofanti-CCPreactivityinmicewith CIAascomparedwithMRL/lprmice. Tolerance to a citrulline-containing epitope partially protects mice from CIA. ToleranceofmicetopathogenicepitopesofCIIhasbeen showntoprotectmicefromdevelopingarthritiswhensubsequentlychallengedwiththisantigen(30,31).Therefore,ifcitrulline-containingepitopescontributedtothediseaseprocess,wereasoned thattolerancetosuchepitopeswouldreducediseaseseverityand incidence.Totestthishypothesis,0.3mgofCII,ovalbumin,LXP, orlinearlysinepeptide(LKP)wasadministeredintravenouslyfor3 daysto6-week-oldmaleDBA/1Jmicetoinducetolerancetothese antigens.LXPisa19aminoacidpeptidederivedfromhuman filaggrinthatcontainsasinglecitrullineresidueand4arginines
963

The Journal of Clinical Investigation http://www.jci.org Volume116 Number4 April2006

research article

Figure 3
Mice tolerized with a citrullinated peptide are significantly protected from CIA. (A and B) Six- to eight-week-old male DBA/1J mice were tolerized to OVA (n = 19), CII (n = 16), LXP (n = 16), or LKP (n = 12). After challenging these mice with immunization of CII in CFA, disease severity (A) and incidence (B) were measured. (C) Paws from mice in each tolerance group were analyzed for histologic parameters of arthritis as described in Methods. (D) Anti-CII antibodies were measured by ELISA, and (E) antibodies against citrullinated fibrinogen were determined by Western blot. All data are shown as group averages SEM. Statistical significance was determined by 1-way ANOVA. *P < 0.05; **P < 0.01 in comparison with the OVA and LKP groups. (F) Sera from tolerized mice were applied to synovial antigen arrays to characterize the autoantibody response. The relative reactivities for the labeled antigens in a pool of sera from 9 mice in each LXP- and LKP-tolerized group are shown, as is the clustering of mice based on the relationships of antibody reactivity between the mice.

(12).Sinceanyarginineresiduemaybecomedeiminatedinvivo byendogenousPADs,alysine-substitutedpeptide(LKP),which maintainstheoverallchargeofthepeptidewhileeliminatingthe possibilityofbecomingdeiminatedinvivo,waschosenasacontrolfortheseexperiments. Threedaysfollowingthelastdoseoftolerogen,micewerechallengedondays0and21withCIIinCFAtoinduceCIA.Asshown previously(30),diseaseinmicetolerizedwithovalbuminwasnot affectedwhilemicetolerizedwithCIIfailedtodeveloparthritis (Figure3,AandB).CIAdevelopmentinmicetolerizedwiththe controlLKPpeptidewasalsounaffectedwhereastheseverityof diseaseinLXP-tolerizedmicewassignificantlyreduced,byapproximately60%,comparedwiththeovalbuminandLKP-tolerized groups(Figure3A).Further,theincidenceofCIAinLXP-tolerized micewasalsoreduced(Figure3B).Histologicevaluationofthe jointsfromthesemiceconfirmedtheclinicalphenotype,demonstratingsignificantlyreducedinflammation,pannusformation, andcartilageandbonedestructionintheLXP-tolerizedmicecomparedwithovalbumin-andLKP-tolerizedcontrols(Figure3C).As expected,inductionoftolerancewithCIIledtosignificantreductionsinanti-CIIantibodylevels(~74%IgG1and~96%IgG2a;
964

Figure3D).However,anti-CIIantibodylevelswereunaffectedin theothergroups.WesternblotanalysisrevealedthatantibodiesagainstcitrullinatedfibrinogenwerereducedintheCII-and LXP-tolerizedgroupsbutunaffectedintheovalbuminandLKP controlgroups(Figure3E).Inthesegroupsofmice,antibodyreactivityagainstadditionalunmodifiedandcitrullinatedantigens wasmeasuredusingsynovialantigenarrays(19)(pleasenotethat peptidesequencesusedinFigures3and4areinSupplemental Table1).MicetolerizedwithLXPcomparedwithLKPandOVA exhibitedadecreaseinepitopespreadingtobothcitrullinatedand nativeepitopes(Figure3Fanddatanotshown).Wealsoobserved reducedepitopespreadinginCII-tolerizedmicecomparedwith OVAandLKP-tolerizedgroups(datanotshown).Fromthesedata, weconcludethatimmuneresponsestocitrullinatedproteinsare importanttothedevelopmentofinflammatoryarthritis. Monoclonal antibodies specific to citrullinated fibrinogen enhance arthritis.Asafurthertestofthepathogenicpotentialofantibodiesagainst citrullinatedproteins,wedeterminedwhetherantibodiesagainst theseautoantigenswouldenhancedisease.Toaccomplishthis, wefirstidentifiedmonoclonalantibodiesspecifictocitrullinated fibrinogen.Wethentransferredthemaloneandincombination

The Journal of Clinical Investigation http://www.jci.org Volume116 Number4 April2006

research article

Figure 4
Antibodies specific to citrullinated fibrinogen substantially enhance submaximal arthritis. (A) Purified D513, 1042, or 1618 monoclonal antibody was applied to a Western blot of unmodified and citrullinated fibrinogen. To compete for antigen binding on the blot, soluble unmodified or citrullinated fibrinogen was added to D513 prior to applying the antibody to the Western blot. (B) The relative reactivities of the monoclonal antibodies for unmodified and citrullinated fibrinogen were determined by applying increasing amounts of biotinylated unmodified and citrullinated fibrinogen (fibr) and CII to ELISA plates coated with monoclonal antibody. The amount of protein binding was detected as described in Methods and is displayed as OD. (C) Additional antigens recognized by D513, 1042, and 1618 were detected using synovial antigen arrays. The relative reactivity for the antigens recognized is shown. (D) Arthrogen, a cocktail of 4 monoclonal antibodies against CII, was transferred intravenously into 6- to 8-week-old male DBA/1J mice at increasing doses followed by 50 g LPS intraperitoneally 3 days later in order to determine which dose would establish submaximal disease. n = 3 per dose group. (E) D513 (n = 9) or control monoclonal antibody (n = 6) were administered alone or in combination with the submaximal dose of Arthrogen (1 mg). (F and G) Either 1 mg or 2 mg of monoclonal antibody 1042 or 1618 (n = 6) or 2 mg HB5 control monoclonal antibody (n = 6) was combined with 1 mg of Arthrogen and transferred into mice as performed previously with D513. Data shown are the average disease score per group SEM. Statistical significance was determined using 1-way ANOVA. *P < 0.05; **P < 0.01 in comparison with submaximal Arthrogen alone.
The Journal of Clinical Investigation http://www.jci.org Volume116 Number4 April2006 965

research article

Figure 5
D513, 1042, and 1618 monoclonal antibodies bind antigens within the inflamed synovium of mice with CIA. (A) Immunohistochemistry was performed using 4 g/ml D513, 1042, and 1618. Antibody binding is indicated by DAB precipitate (brown). Tissue was counterstained with methyl green. A mix of anti-CII monoclonal antibodies and secondary reagents without primary antibody served as negative controls for staining the synovium. Representative serial sections are shown. Magnification, 100. Arrows point to synovium, and asterisks indicate cartilage. (B) Skin from the site of immunization, draining inguinal LN, and synovium were collected from unimmunized mice, mice immunized with CFA, or mice with CIA at different time points. Western blots of the protein isolated from these tissues were performed with a polyclonal antibody that recognizes the amino acid citrulline or (C) with the monoclonal antibodies 1042, 1618, and D513. Representative blots are shown. Results of Western blots probed with antibodies 1618 and D513 were essentially identical to 1042 and, thus, are not shown.

withacocktailofanti-CIIantibodiesintomice.Todevelopan antibodyspecifictoacitrullinatedprotein,andnotingthepresenceofnaturalIgMantibodieswiththisspecificity(Figure1F),we screenedapreviouslycreatedpanelofIgM-producinghybridomas establishedfromnaiveC57BL/6splenocytesforantibodyspecificitytocitrullinatedfibrinogen.Oneclone,D513,demonstrated suchspecificreactivitybyWesternblotanalysis(Figure4A).To confirmthattheD513epitopewaspresentinthenativeconformationofcitrullinatedfibrinogen,wedeterminedwhetherthe reactivityofD513ontheWesternblotcouldbeblockedbypreincubatingtheantibodywithsolubleunmodifiedfibrinogenor

966

citrullinatedfibrinogen.Wefoundthatonlytheadditionofcitrullinatedfibrinogentotheantibodycouldblockreactivityonthe Westernblot(Figure4A).AsademonstrationoftherelativeaffinityofD513forsolubleunmodifiedandcitrullinatedfibrinogen andacontrolprotein,CII,0.5gmonoclonalantibodywasbound toImmulon2HBplates,andincreasingamountsofbiotinylated proteinswereaddedtotheboundantibody.D513hadhighaffinityfornativecitrullinatedfibrinogenbutdidnotbindeither unmodifiedfibrinogenorCII(Figure4B).Finally,weidentified additionalantigensforD513byapplyingthemonoclonalantibodytothesynovialantigenarray(SupplementalTable1).D513

The Journal of Clinical Investigation http://www.jci.org Volume116 Number4 April2006

research article
Figure 6
Antibodies against citrullinated fibrinogen can overcome the protective effect of tolerance to citrullinated antigens. Mice were tolerized with LXP, challenged twice with CII in CFA, and then administered either 2 mg of equal amounts of 1042 and 1618 or 2 mg of the control HB5 antibody on days 22 and 28 after the initial challenge with CII in CFA. (A) Arthritis scores were measured between days 21 and 35, and (B) histologic evaluation was performed on paws collected on day 35. Data are shown as group averages SEM. Statistical significance was determined by 1-way ANOVA. *P < 0.05; **P < 0.01 in comparison with the LXP plus HB5 control group. Mice tolerized with LKP, challenged with CII in CFA, and given monoclonal antibodies 1042/1618 or HB5 served as additional controls. No difference was detected between these groups and the LXP plus 1042/1618 group (data not shown).

boundseveralcitrullinatedfibrinogenpeptidesaswellascitrullinatedformsofvimentin,clusterin,andkeratin(Figure4C).In additionalexperiments,D513didnotreactwithtypicaltargetsof cross-reactivenaturalantibodies,includingdouble-strandedDNA, single-strandedDNA,andphospholipids(datanotshown).These resultsintotodemonstratethatD513specificallybindscitrullinatedepitopesderivedfromfibrinogenandotherantigens. TomeasurethepathogenicpotentialofD513,wetransferred 2mgofpurifiedantibodyintravenouslyinto6-to8-week-old DBA/1Jmice,followedby50gLPS3dayslater.Noneofthe micedevelopedanyvisiblesignsofarthritis(Figure4E).Wethen askedwhetheranantibodytocitrullinatedfibrinogencould contributetotissueinjurybyenhancingsubmaximaldisease establishedbyacocktailofmonoclonalantibodiesagainstCII (Arthrogen).Todetermineasubmaximaldoseofanti-CIIantibodies,increasingdosesofArthrogenwereadministeredto6-to 8-week-oldmaleDBA/1Jmicefollowedby50gLPS3dayslater (Figure4D).OnemilligramofArthrogeninducedverymilddiseaseandwasthereforechosentouseincombinationwith1mg and2mgD513.Theadditionofboth1mgand2mgD513to 1mgArthrogensignificantlyincreasedtheseverityofarthritisby morethan2.5fold(Figure4E).TodemonstratethattheenhancementofdiseasewasduetoD513andnottoanoverallincreasein totalIgadministered,wealsotransferred2mgofamonoclonal IgMspecifictotrinitrophenol-KLH(TNP)incombinationwith 1mgArthrogen.Thistreatmentdidnotresultinasignificant enhancementofdisease(Figure4E). Finally,weidentified2IgG2amonoclonalantibodies,1042and 1618,fromafusionofsplenocytesandLNcells,respectively,from DBA/1JmicewithCIAthatwerespecifictocitrullinatedfibrinogen(Figure4,AandB)aswellasforcitrullinatedpeptidesderived fromfilaggrinandvimentin(Figure4C).Todemonstratethatthe citrullinatedfibrinogen-specificIgGantibodiesthatappearin CIAareindeedpathogenic,wetransferred1mgor2mgofpurified1042or1618incombinationwith1mgArthrogenintomice asdescribed.Bothmonoclonalantibodiessignificantlyacceleratedandenhanceddisease(Figure4,FandG).Twomilligrams

ofacontrolIgG2amonoclonalantibody,HB5,whichspecifically recognizeshumancomplementreceptor2andnotthemurine homolog(32,33),didnotaccelerateorenhancediseasewhen combinedwith1mgArthrogen(Figure4,FandG).Theseresults furthersupporttheconclusionthatantibodiesspecifictocitrullinatedproteinsarepathogenic. D513, 1042, and 1618 monoclonal antibodies bind target antigens in inflamed synovium.Citrulline-modifiedfibrinogenhasbeen demonstratedtobepresentintheinflamedsynoviumofmice withCIA(34).Todetermineiftheenhancementofarthritisby thetransferofmonoclonalantibodiesspecifictocitrullinated fibrinogenisduetotheirbindingthisantigeninthesynovium, weperformedimmunohistochemistryonjointtissuesections fromunimmunizedmice,miceimmunizedwithCFA,andmice thatdevelopedCIAfollowingimmunizationwithCIIinCFA.All 3monoclonalantibodiesboundantigenswithinthepannusof micewithCIAbutdidnotbindtissuefromunimmunizedor adjuvantcontrolmice(Figure5A).Anti-CIImonoclonalantibodies(Arthrogen)wereusedforpositivecontrolstainingofcartilage,whereCIIislocatedwithinthejoint,andnegativecontrol stainingofsynovium.OurresultsareconsistentwiththisexpectedpatternofstainingforCII. Additionally, we isolated proteins from the skin, draining inguinalLNs,andsynoviumofunimmunizedmice,adjuvant controlmice,andmicewithCIAondays7and35afterimmunization.Equalproteinamountsofcellularlysateswereloaded into10%Bis-Trisacrylamidegels(SupplementalFigure2),and Western blot analysis of this protein was performed using a polyclonalanti-citrullineantibodyandouranti-citrullinated proteinmonoclonalantibodies.Totalcitrullinatedproteins,as indicatedbytheblotprobedwiththeanti-citrullinepolyclonal antibody,mildlyincreasedinthesynoviumofmiceadministered CFAbyday7afterimmunization,butbyday35,totalcitrullinatedproteinwassimilartosynoviumfromunimmunizedcontrol mice.Anincreaseintotalcitrullinatedproteinwasnotobserved insynoviumfrommicewithCIAonday7afterimmunization; however,byday35,therewasareadilyapparentincreaseincitrullinatedprotein(Figure5B).InsynoviaofmicewithCIA,bandsat approximately6070kDaandatapproximately105kDagreatly increasedinintensityfromday0today35.Nodifferencesin skinordrainingLNswereobserved.Citrullinatedantigensat approximately70kDaandapproximately105kDadetectedby ourmonoclonalantibody1042werealsoincreasedinthesynoviumbyday35inmicewithCIA(Figure5C).Amodestincreasein citrullinatedantigendetectedby1042wasalsoobservedinthe
967

The Journal of Clinical Investigation http://www.jci.org Volume116 Number4 April2006

research article
synoviumfrommicegivenCFAbyday35,butthisincreasewas notasgreatasseeninsynoviumfrommicewithCIAonday35. Mostintriguingly,though,thegreatestincreaseincitrullinated antigensatapproximately70kDaandapproximately105kDa detectedby1042wasinthedrainingLNsonday7afterimmunizationwithCIIinCFA.WesternblotanalyseswiththemonoclonalantibodiesD513and1618wereequaltothatwith1042(data notshown).Thesedatasuggestthatcitrullinatedantigensare presentedinthedrainingLNsshortlyafterimmunizationwith CIIinCFAandcouldhelptodrivethedevelopmentofantibodies againstthesecitrullinatedantigens.Thisadditionalcitrullinated antigen-antibodyinteractionwithinthejointisthelikelymechanismforamplificationofdisease. Administration of antibodies against citrullinated fibrinogen can overcome protective tolerance induced by LXP.OurdatasuggestthattolerizationofmicewithLXPprotectsthemfromthefulldevelopment ofCIAbypreventingepitopespreadingtoincludeantibodyreactivitytowardcitrullinatedantigens(Figure3,EandF).Therefore, todetermineifantibodiesagainstcitrullinatedantigenscould overcometheprotectionprovidedbytolerancetotheseantigens, wetolerizedmicetoLXPandchallengedthemwithCIIinCFA asbefore,thenadministered2mgofequalamountsof1042and 1618monoclonalantibodiesor2mgofHB5controlmonoclonal antibodyondays22and28aftertheinitialimmunizationwith CIIinCFA.Treatmentoftolerizedmicewith1042and1618ledto thedevelopmentoffulldiseasewhilemicetreatedwiththecontrol HB5monoclonalantibodyremainedprotectedfromfullarthritis,measuredbothclinicallyandhistologically(Figure6).Wealso usedascontrolsmicetolerizedwithLKP,challengedwithCIIin CFA,andgiveneitherthe1042/1618monoclonalantibodycocktailortheHB5monoclonalantibodycontrol.Wedidnotobserve significantdifferencesindiseaseorhistologicscoresinthese2 groupscomparedwiththeLXP-tolerizedgroupthatreceivedthe 1042/1618monoclonalantibodies(datanotshown).Insum,we confirmthattheprotectiveeffectoftolerancewithLXPisindeed adecreaseinthedevelopmentofantibodiesagainstcitrullinated antigens,asreplacementofantibodiesagainstcitrullinatedantigensusingthe1042/1618monoclonalantibodiesspecifically resultsintheredevelopmentofseverearthritis. Discussion Inthisreport,wedemonstratetheimportanceofcitrullinated proteinsandantibodiesagainstsuchselfantigensforthedevelopmentofinflammatoryarthritis.Using3methods,antibodiesagainstcitrullinatedantigensweredetectedinCIAandwere showntobespecifictoacitrulline-containingprotein.Induction oftolerancetoacitrullinatedpeptideledtoasignificantreductioninsusceptibilitytoCIAuponchallengewithCII,indicating thatimmuneresponsestowardcitrullinatedantigensareinvolved indiseasepathogenesis.Furthermore,monoclonalantibodies specifictocitrullinatedfibrinogensubstantiallyenhancedarthritiswhencoadministeredwithanti-CIImonoclonalantibodies andreversedtheprotectionfrominflammatoryinjuryfollowing inductionoftolerancewithacitrullinatedpeptide.Webelieve thattheseresultsintotodefinitivelydemonstratethatantibodies againstcitrullinatedproteinsareimportantcontributingfactors inthedevelopmentofinflammatoryarthritis. AntibodiesintheserumofpatientswithRAhavebeenshown to specifically target citrullinated filaggrin, fibrinogen, and vimentin. These antibodies have been detected, using indi968

rectimmunofluorescence,ontheepitheliumofratesophagus, whichcontainscitrullinatedfilaggrin(7,12);byWesternblotof theunmodifiedandcitrullinatedformsoffibrinogen(17)and vimentin(18);andbyanELISAwithpeptides(CCP)basedon filaggrinsequencesthatcontaincitrullineresidues(14).Using eachofthesemethods,wewereabletodemonstratethesame citrulline-specificantigenreactivityinCIA.Ourdatavalidate thismurinemodelofinflammatoryarthritisforfurtherinvestigationofthedevelopmentandpathogenicityofantibodies againstcitrullinatedproteinsinRA. Asafirstmeasureofthepathogenicpotentialofcitrullinated antigens,weestablishedtolerancetotheseantigensinmicefollowedbychallengewithCII.WechosetheLXPpeptideasacitrullinatedtolerogenbecausethispeptidecontainedasinglecitrulline residuesurroundedbyaflankingaminoacidsequencedescribed inapopulationofRApatientstoconferantibodyreactivity(12, 13).Additionally,fromourindirectimmunofluorescencestudies,LXPinhibitedCIAserumreactivitywiththeepitheliumof ratesophagus,suggestingthatthispeptidecontainedanepitope thatwascentraltothecitrulline-specificimmuneresponse.By tolerizingmicetoLXP,micewerepartiallyprotectedfromCIA uponchallengewithCIIinCFA.Thelikelycauseofthisprotectionwasareductionofantibodiesagainstcitrullinatedproteins. However,wecannotexcludetoleranceofcitrullinated-peptide specificTcells,whichmayalsocontributetothesuppressed arthritisthatwasobserved.Nevertheless,administrationofantibodiesagainstcitrullinatedproteinsovercametoleranceand restoredseverediseaseinmice,indicatingthattheproduction ofantibodiesagainstcitrullinatedproteinsisthekeypathogenic pathwayleadingtomaximalarthritis. Todirectlydemonstratethepathogenicityofantibodiesspecific tocitrullinatedproteins,wedevelopedmonoclonalantibodiesspecifictocitrullinatedfibrinogen(D513,1042,and1618)andtransferredthemintomicealoneandincombinationwithantibodies againstCII.Thesemonoclonalantibodiesalonecouldnotinduce arthritisatadoseof2mgpermouse;however,thisresultwasnot surprising.Onereasonisthattoinducearthritisbytransferof monoclonalantibodiesagainstCII,acombinationofatleast3differentantibodies,eachrecognizingdistinctlydifferentepitopes, isrequired(35).Inaddition,inanothermodelofautoimmune arthritis,thetransferofantibodiesspecifictoglucose-6-phosphate isomerasefromK/BxNmicealsohasbeenshowntorequireaminimumof3monoclonalantibodiesagainstthetargetedantigenfor visiblyapparentdiseasetobeestablishedinmice(36).Therefore, weusedthearthrogenicanti-CIImonoclonalantibodycocktailto establishminimaldiseaseinmice,witharthritisbecomingsubstantiallymoreseverewiththeadditionofD513,1042,or1618. Itislikelythatthesemonoclonalantibodiesrecognizedcitrullinatedproteinsinsynovium,thepresenceofwhichhasbeenpreviouslydescribedinCIA(34).Citrullinatedproteinsinthesynovia werelikelyinduced bytheanti-CIImonoclonalantibodycocktail. OurstudiesbyimmunohistochemistryofD513,1042,and1618 ininflamedsynoviumaswellasWesternblotsofsynoviumfrom micewithCIAprobedwithantibodiesagainstcitrullineandcitrullinatedproteinsareconsistentwiththisconclusion(Figure5). Anti-CCPreactivityhasalsobeendemonstratedinothermurine modelsofautoimmunity,namelyinMRL/lprmice(29,37).Since MRL/lprmicemaydevelopspontaneousarthritis(38),itisconceivablethatanti-CCPantibodieswouldbepresentinthesemiceas partofthepathogenicprocessesleadingtoautoimmunearthritis.

The Journal of Clinical Investigation http://www.jci.org Volume116 Number4 April2006

research article
Weconfirmedthepresenceofanti-CCPantibodiesinMRL/lpr miceandfurthercharacterizedtheseantibodies.However,while theanti-CCPantibodiespresentinMRL/lprmiceconsistedofthe sameisotypesasseeninCIA,anti-CCPantibodiesinMRL/lprmice lackedspecificityforacitrullinatedantigen.Therefore,ourresults areconsistentwiththoseofothers(37)whohaveconcludedthat anti-CCPreactivityinMRL/lprmicerepresentsadifferentimmunologicprocessthanthatpresentinRA. Ourresultsareincontrastwiththoseofothers(34)whereantibodiesagainstCCPandcitrullinatedfibrinogenwerenotdetected inmicewithCIA.Severalpossibilitiesmayexplainthesediffering observations.Themiceinourstudymayhavedevelopedamore robustimmuneresponseaftertheinitialimmunizationduetodifferencesintheenvironmentofanimalhousingfacilitiesorinspecificimmunizationtechniques.TheDBA/1miceinourstudywere alsoobtainedfromadifferentsource,andgeneticdriftmayhave occurredbetweenthe2sources,affectingtheabilityofthemice toproduceantibodiesagainstcitrullinatedproteinsinresponse toCIIimmunization.Inaddition,weuseddifferentsourcesfor secondaryreagentsinourstudies. Citrullinationofproteinshasbeenshowntobreaktolerancein mice.Whenimmunizedwithhumanormurinefibrinogen,which areapproximately70%homologousinaminoacidsequence,mice donotproduceantibodiesagainstthenativeprotein.However,if thefibrinogenistreatedwithPAD,suchthatitbecomescitrullinatedpriortoimmunization,micedeveloparobustantibody responsetothecitrullinatedprotein(39).Furthermore,sincePAD hasnotbeenshowntobeexpressedinthethymus(40),Tcellsreactivetocitrullinatedantigensarenotlikelytobenegativelyselected. Reasonably,deiminationofselfproteinscouldresultinimmune responsestoselfantigensthatwouldotherwisenotdevelop. TheresponsetowardacitrullinatedproteininCIAislikelyto havearisenfromaneventduringwhichproteinwasdeiminated duringorshortlyafterimmunization.Antibodiesagainstboth CCPandCIIoccurredconcurrentlyapproximately7daysfollowingimmunizationwithCIIinCFA.However,adjuvantalone couldnotelicitanantibodyresponse,suggestingthattheCII antigentriggeredanantibodyresponsetocitrullinatedproteins. WewereunabletodetectthepresenceofcitrullineintheCIIused forimmunizationthrougheitheradiacetylmonoximecolorimetricassay(sensitivityof10nMcitrulline;ref.41)orbyaminoacid analysis,whichhasasensitivityof0.02%oftotalCIIaminoacid content(datanotshown).However,typeIIcollagendoescontain acoreepitopesuggestedtoconferCCPreactivitywhenthearginineisdeiminated(aminoacidsequenceGRT,ref.13;aminoacids 148150inCII,aminoacids288290inmurinetypeIIcollagen). Conceivably,manyproteinsbecamedeiminatedduringorafter immunization,whichcouldalsoleadtotheantibodyresponses tocitrullinatedantigens. Twopossibledeiminationeventsmayhaveoccurredafterimmunizationofourmice.EitherthemycobacteriaintheadjuvantcontainedaPAD,orinfiltratingneutrophilsandmacrophagesthat expressPADIV(34,40,42)atthesiteofimmunizationorinthe jointreleasedtheenzymeintotheextracellularmatrixduring apoptosis.TheformerpossibilityisunlikelysincenoPADhasbeen describedinmycobacteria,andthemycobacteriahavebeenheat killedpriortouseintheadjuvant.Thelatterpossibilityismore likely,especiallyinlightofothersobservationswithinthesynoviumofarthriticmice.NeutrophilsexpressingtypeIVPADhave beenreportedtomigrateintotheinflamedsynoviumofmicewith

CIAorstreptococcalwallarthritisconcordantwiththepresenceof citrullinatedfibrinogen(34).Theseauthorshypothesizedthatthe infiltratingneutrophilsreleasedPADintotheextracellularmatrix duringapoptosiswherecalciumconcentrationsweresufficiently hightosustainenzymaticactivity.Asfibrinisacommonconstituentoftheinflamedsynovium,itmayserveasalocalsubstratefor PAD,generatingcitrullinatedantigensinsitu. Inourmodel,onepossiblescenarioisthatlocalinflammation ateitherthesiteofinjectionorwithinthejointensuedshortly afterimmunization.InfiltratingneutrophilsatthissiteunderwentapoptosisandreleasedPAD.Argininewithintheinjected CIIorotherlocalantigenssuchasfibrinmayhavebecomedeiminated,leadingtotheselectionoftheantibodiesagainstcitrullinatedproteinsthatwereobservedinmurineCIA.Weattemptedto demonstrateanincreaseincitrullinatedantigensintissuesafter immunization.Asskincontainsfilaggrin,whichisknowntobe citrullinated,itisdifficulttodemonstrateanincreaseincitrullinatedproteininthistissue.However,wediddetectincreased citrullinatedproteinofapproximately70kDaandapproximately 105kDaindrainingLNsfromthesiteofimmunization,suggestingthatafterimmunization,protein(s)becamecitrullinatedand werepresentedintheLNs.IfmiceweretolerizedtoLXPprior toimmunizationwithCII,theseantibodyresponseswouldbe inhibited,resultingindecreasedarthritis.Thedevelopmentof IgGantibodiesagainstcitrullinatedantigenssuggeststhisisa Tcellmediatedprocess,thoughwehavenotformallyevaluated thisissue.Additionally,weobservedincreasedcitrullineandcitrullinatedproteinofapproximately6070kDaandapproximately 105kDainsynoviumofmiceafterimmunization;however,this increasewasnotobservedonday7afterimmunizationwithCII inCFA,butbyday35,indicatingthatsignificantcitrullination ofjointantigensoccurredafterthedevelopmentofantibodies againstcitrullinatedproteins.Thus,ourdatasuggesttheinitial antigen(s)thatdrivetheautoantibodyresponsetocitrullinated proteinsoriginateoutsidethejoint.Furtherstudiesarenecessary, though,toevaluatethisissue. Regardless of how antibodies against citrullinated proteins developandwhattheirtargetsare,theseantibodiescontributesignificantlytodisease.AftertransferofD513,1042,or1618,these monoclonalantibodieslikelybindtheirantigen,citrullinated fibrinogen,withininflamedsynovium.Thiswouldleadtofurther immunecomplexformation,complementactivation,andneutrophilandmacrophageproductionofproinflammatorymolecules, therebyenhancingtheinflammatoryresponseswithinthejointof miceandalsoenhancingvisiblyapparentdisease. The hypothesis that posttranslational modification of self proteinsbreakstoleranceisnotlimitedtocitrullination.Within theCIAmodel,anotherposttranslationalmodificationhasbeen describedasimportantinbreakingTcelltolerancetoCII.Alysine residueatposition264ofCIImaybeposttranslationallymodifiedthroughhydroxylationandgalactosylation.SuchposttranslationalmodificationviaglycosylationoflysineresiduesinCIIcan occurinthechondrocyte,resultinginabreakintolerancetoCIIin bothhumansandmice(43,44).Treatmentofneonatalmicewith agalactosylatedCII259273 peptideprotectedmicefromthedevelopmentofCIAbetterthantheunmodifiedpeptide(43).Further, micetransgenicforhumanCIIandDRB*0401/humanCD4were immunizedwithhumanCIIandgeneratedstrongerrecallTcell responseswhenrestimulatedwiththegalactosylatedCII259273 peptide(44).Inanotherautoimmunesystem,phosphorylationof
969

The Journal of Clinical Investigation http://www.jci.org Volume116 Number4 April2006

research article
selfproteinsbyserine/threoninekinasesduringapoptosisgeneratesautoantigensrecognizedbyserafrompatientswithsystemic lupuserythematosus(45).Finally,inpatientswithceliacdisease, deamidationofgliadinbytissuetransglutaminasebreaksTcelltolerancetogliadin(46,47),andthesubstrate-enzymecomplexgeneratesadditionalTcellepitopesinthesepatients(48).Hence,thereis increasingdatathatposttranslationalmodificationofselfproteins canplayacriticalroleinthedevelopmentofautoimmunity. Insum,ourstudiesestablishthatanautoantibodyresponse tocitrullinatedproteins,whichisindistinguishablefromthat observedinRA,developsinCIA.Additionally,thisreportisthe first formal demonstration, to our knowledge, that antibodiesagainstcitrullinatedproteinscontributetothepathogenesis ofinflammatoryarthritis.However,manyquestionsremainto beanswered.Forexample,howanimmuneresponsespecificto citrullinatedproteinsdevelopsinCIAaswellasinpatientswith RAisunknown.Additionally,therelevanttargetsoftheseantibodiesinjointsorinperipheraltissuearenotknown.Also,whether selectionofcitrullinatedautoantibodiesemergesfromapreexistingnaturalantibodyrepertoire,asourstudiessuggest,isunclear. Nevertheless,wehaveshownhereinthatCIAisavaluablemodelin whichmanyofthesequestionsmaybeinvestigated.Furthermore, knowingthatantibodiesagainstcitrullinatedproteinsareinvolved inthepathogenesisofRAmayallowforthedevelopmentoftargetedtherapeuticsthatinhibittheseantibodiesortheirdevelopment.Suchtherapeuticapproachesmayreducediseaseseverityin patientswithactivediseaseorperhapsevenpreventtheonsetof clinicallysignificantarthritisinsusceptibleindividualswhoare stillasymptomaticbutexhibitanti-CCPantibodies(15,16). Methods
Animals and immunizations.CIIpurifiedfromcalfarticularcartilage(Elastin ProductsCompanyInc.)wassolubilizedin0.1Maceticacidataconcentrationof4mg/mlandstoredat70C.Forimmunization,CIIwasemulsified inanequalvolumeofincompleteFreundsadjuvant(BDDiagnostics)supplementedwith4mg/mlM. tuberculosisH37RA(BDDiagnostics)tomake completeFreundsadjuvant(CFA).Six-toeight-week-oldDBA/1Jmalemice (JacksonLaboratory)wereimmunizedintradermallywith0.1mlCIIinCFA orwithCFAaloneondays0and21.Fromdays25through35aftertheinitialimmunization,micewerescoreddailybyanindividualblindedtotheir treatmentsforsignsofarthritisinthepawsbasedonthefollowingscale:0, norednessorswelling;1,1digitswollen;2,2digitsswollen;3,3digitsswollen;and4,entirepawswollenwithankylosis.Thescoresforeachof4pawsof amouseweretotaledtogiveafinalscorewithamaximalseverityof16. Toleranceinanimalswasinducedbyadministering300gofprotein (CIIorovalbumin)orpeptide(LXPorLKP;seebelow)inPBSintravenously dailyfor3days.Threedaysfollowingthefinaldoseoftolerogen,micewere immunizedwithCIIinCFAaccordingtothesameprotocolasabove,with day0oftheexperimentcorrespondingtothedayonwhichanimalswere firstimmunizedwithCIIinCFA.Toovercometolerance,micetolerized andchallengedasabovewerethenadministeredeither2mgtotalof1mg 1042plus1mg1618or2mgofthecontrolHB5monoclonalantibody describedbelowondays22and28aftertheinitialCII/CFAchallenge. Passivearthritiswasinducedbyintravenoustransferofacocktailof monoclonalantibodiesagainstCII(Arthrogen-CIA;ChemiconInternational)and/orthemonoclonalantibodiesD513,1042,or1618,which specificallybindcitrulline-modifiedfibrinogen.AnIgMmonoclonalantibodytotrinitrophenol-KLH(anti-TNP;BDBiosciencesPharmingen) and an IgG2a monoclonal antibody specific to human complement receptor2(HB5)wereadministeredasnegativecontrols.Previousstudies
970

havedemonstratedthatHB5doesnotbindmurinecomplementreceptor2(32,33)andthereforedoesnotmodulatethecomplementsystem inmice.Intraperitonealinjectionof50g/mouseofLPSfollowed3days laterafteradministrationofeachantibody.Arthrogenwastitratedto determinethedosethatwouldyieldsubmaximaldiseaseinanimalsfor useincombinationwithanticitrullinatedfibrinogenmonoclonalantibodiesandthecontrolantibody.AllanimalworkwasapprovedandperformedundertheguidelinessetbytheUniversityofColoradoInstitutionalAnimalCareandUseCommittee. Peptides.LXP(SHQESTcitGRSRGRSGRSGS)andLRP(SHQESTRGRSRGRSGRSGS)weresynthesizedasdescribed(12).Anotherpeptide,in whichallarginineresiduesweresubstitutedwithlysine,wasalsosynthesizedandservedasanadditionalcontrol(LKP;SHQESTKGKSKGKSGKSGS).QualitycontrolbymassspectroscopyandHPLCrevealedthe peptidestobeoftheappropriatesequenceandmorethan95%pure. Measurement of antibodies.Serafromeachmousewerecollectedfromthe tailveinontheinitialdayofimmunizationandevery7daysthereafter.On thedayofsacrifice,serawerecollectedbycardiacpuncture.Anti-CCPantibodytitersweredeterminedusingtheDIASTATAnti-CCP2kit(second generationpeptides;Axis-ShieldDiagnosticsLtd.)followingmanufacturersinstructionswiththefollowingmodifications:formostexperiments mouseserawerediluted1:10or1:100insamplediluent,andthesecondaryantibodywassubstitutedwithalkalinephosphataseconjugatedgoat anti-mouseIgM,IgG,IgG1,IgG2a,IgG2b,orIgG3(CALTAGLaboratories) diluted1:1,000inPBS.ThekitstandardwasutilizedforallELISAswithout modificationofthesecondaryantibodytomaintainconsistencybetween allplatesandtodemonstraterelativereactivitybetweenisotypes. Anti-CIIantibodytiterswereassayedbyELISA.Immulon2HB96-well plates(DYNEXTechnologies)werecoatedwith0.5g/wellbovineCII (ChondrexInc.)overnightat4C.Plateswerewashed3timeswithPBS containing0.5%Tween-20and1%BSAandthenblockedwith10%BSAin PBSfor4hoursat4C.Mouseserawerediluted1:500,1:1,000,or1:2,000 inPBSand50l/welladdedinduplicate;antibodieswerethenallowedto bindovernightat4C.Foreachplate,astandardconsistingofpooledsera fromarthriticmicewasapplied.Plateswerewashed3timeswithPBS/0.5% Tween-20.Horseradishperoxidaseconjugated(HRP-conjugated)goat anti-mouseIgM,IgG,IgG1,IgG2a,IgG2b,orIgG3(CALTAGLaboratories)diluted1:1,000inPBSwasaddedat50l/wellandallowedtobindfor 4hoursat4C.Again,plateswerewashed3timeswithPBS/0.5%Tween20.ELISAwasdevelopedfor30minutesusing50l/well,1%H2O2,in2,2azinobis(3-ethylbenzthiazoline-6-sulfonicacid)(ABTS)solutionandread at405nmwitha492-nmreferencefilter. Anti-filaggrin antibody reactivity was determined by indirect immunofluorescence.Undilutedseraorseradiluted1:10inPBSfrom arthritic,adjuvantcontrol,anduntreatedmicewereappliedtoratesophagealsections(ScimedxCorporation)for1houratroomtemperaturewith humidity.Theslideswerethenwashed3timesfor5minutesinPBS.FITCconjugatedgoatanti-mouseIgG(BDBiosciencesPharmingen)wasapplied for1houratroomtemperaturewithhumidity.Theslideswereagainwashed 3timesfor5minutesinPBSandacoverslipplacedontheslide.Reactivity wasvisualizedunderafluorescentmicroscope.Forpeptidecompetitionof serumbindingtotissue,serawerediluted1:10inPBS,with100g/mlLAP inPBS,orwith100g/mlLXPinPBSandincubatedat4Covernightbefore beingappliedtotissuesections. Antibodiesagainstunmodifiedandcitrullinatedfibrinogenwereassessed byWesternblotanalysis.Humanfibrinogen(Sigma-Aldrich)wasdeiminated invitrobyrabbittypeIIPAD(Sigma-Aldrich)atasubstrate/enzymeratio of1mg:2Ufor2hoursat37Cin0.1MTris-HClpH7.5,10mMCaCl2, and5mMDTT.Thereactionwasstoppedbytheadditionof20mMEDTA. Astheunmodifiedfibrinogencontrol,fibrinogenandPADwerecombined

The Journal of Clinical Investigation http://www.jci.org Volume116 Number4 April2006

research article
asdescribed,but20mMEDTAwasaddedimmediatelytopreventenzyme activityonthesubstrate.ForWesternblotting,2.5gofunmodifiedor citrullinatedfibrinogenwasloadedintoeachwellofaNuPAGE10%BisTrisgel(InvitrogenCorp.)andtransferredontoanitrocellulosemembrane (AmershamPharmaciaBiotech).Themembranewasblockedfor1hourat roomtemperaturewith10%drymilkinPBS.PooledserafromRApatients, arthriticmice,orcontrolmicewerediluted1:100inPBSwith1%BSAand incubatedwiththemembraneovernightat4C.Themembranewasthen washedwithPBS/0.5%Tween-203timesfor10minuteseachwash.HRP-conjugatedgoatanti-humanoranti-mouseIgG(CALTAGLaboratories)diluted 1:2,000inPBSwasincubatedwiththemembranefor2hoursatroomtemperature.AfterwashingthemembranewithPBS/0.5%Tween-203timesfor 10minutes,equalvolumesofluminescencereagentandoxidizingreagent (ECLreagent;AmershamPharmaciaBiotech)wereappliedfor1minute.The membranewasthentransferredtoaplasticbagandfilmwasexposed. Histologic analysis of joints and immunohistochemistry.Atthetimeofsacrificeon day35afterimmunizationwithCIIinCFA,pawswereremovedandplaced in10%bufferedformalin.Tissuewasthendecalcifiedin5%formicacidfor3 days,followedbyembeddinginparaffin.Sectionsof8mwerestainedwith toluidineblueandevaluatedbyatrainedobserverblindedtothetreatment group.Eachmousewasassessedahistologicdiseaseactivityscorefrom0to 5determinedbyinflammation,pannusformation,cartilagedamage,bone damage,andanoverallscorebasedon5jointsetsperanimal. Forimmunohistochemistry,5-mparaffin-embeddedsectionswerefixed toSuperfrostPlusmicroscopeslides(FisherScientificInternational).Slides wereheatedto60Cuntilparaffinwasmelted.Tissuewasfixedtotheslide andrehydratedbysoakinginxylenesfor2changes,5minuteseach,followed by2changes,2minuteseach,of100%ethanol,95%ethanol,70%ethanol, andPBS.Bovinetesticularhyaluronidase(10,000U/ml)(Sigma-Aldrich)was appliedtothetissuefor30minutes.Intrinsicperoxidaseactivitywasthen inactivatedbyincubationfor1hourinmethanolwith3%H2O2.Afterwashingtissue2timesfor5minutesinPBS,tissueantigenswereimmunostained usingaMouseonMouse(M.O.M.)tissue-stainingkit(VectorLaboratories). D513,1042,and1610weredilutedto4g/mlforapplicationtotissue,and appropriatesecondaryantibodyat1.5g/mlwassubstitutedwhennecessary.Stainedsectionswereexaminedbylightmicroscopy. Development and purification of monoclonal antibodies.IgMmonoclonal antibodiesweredevelopedfromsplenocytescollectedfromanaive,wildtypeC57BL/6mousethatwerefusedwiththeSP2/0-AG14myeloma celllinebythestandardprotocoltoestablishhybridomas.Tocreate IgGmonoclonalantibodies,splenocytesanddrainingLNcellsfroma mousewithCIAwerefusedwiththeFOX-NYmyelomacellline.SuccessfulfusionswerescreenedbyWesternblotandselectedbasedonspecificreactivitywithcitrullinatedfibrinogen.OnepositiveIgM-producing hybridoma,D513,and2positiveIgG-producinghybridomas,1042and 1618,wereclonedfurtherandstudied. AntibodiesfromexhaustedsupernatantofculturedD513,1042,and 1618cellswereaffinitypurified.SupernatantfromD513wasappliedto acolumnofagarosebeadscoatedwithantibodiesagainsthumanIgM, chain(Sigma-Aldrich),cross-reactivewithmouseIgM.Supernatant from1042and1618wasappliedtoacolumnofproteinGcoatedsepharosebeads(AmershamBiosciences).Boundantibodywaselutedwith 0.1Mglycine,pH2.3,andcollectedinto1.5MTris,pH8.8.ElutedantibodywasdialyzedagainstPBS,pH7.4,for48hoursandconcentrated usingcentrifugalfiltration(CentriconPlus-20;Millipore).AntibodyconcentrationwasdeterminedbytheA280ofthesample,andpuritywasconfirmedbyrunningona10%Bis-Trisgel. Reactivityofthepurifiedmonoclonalantibodywasconfirmedbyapplying10g/mlmonoclonalantibodyinPBStoWesternblotsofunmodified andcitrullinatedfibrinogenasdescribedpreviously.Thesecondaryanti

bodiesusedintheseanalyseswereHRP-conjugatedgoatanti-mouseIgM forD513andHRP-conjugatedgoatanti-mouseIgGfor1042and1618, diluted1:2,000inPBS.Todemonstratespecificityforfibrinogenonthe blot,200g/mlunmodifiedfibrinogenorcitrullinatedfibrinogenwas preincubatedwith10g/mlD513for1houratroomtemperatureprior toapplyingtotheWesternblot. Relativebindingtounmodifiedfibrinogen,citrullinatedfibrinogen, andCIIwasdeterminedbyELISA.WellsofImmulon2HBplates(DYNEX Technologies)werecoatedwith100l/wellof5g/mlmonoclonalantibodiesD513,1042,or1618inPBSovernightat4C.Theplateswere washed3timeswithPBS/0.5%Tween-20andthenblockedwith200l/well 1%BSAinPBSfor1houratroomtemperature.Biotinylatedunmodifiedfibrinogen,citrullinatedfibrinogen,orCIIwasappliedatincreasing concentrationsfrom0pmolto450pmolforD513orfrom0pmolto 90pmolfor1042and1618.Theproteinwasallowedtobindantibody for1houratroomtemperature.Theplateswerewashed5timeswith PBS/0.5%Tween-20.Weadded100l/wellofstreptavidin-HRPdiluted 1:5,000inPBStowellsandallowedittobindfor1houratroomtemperature.Afterwashingtheplates5timeswithPBS/0.5%Tween-20,theplates weredevelopedwithABTSsolutiondescribedaboveandreadat405nm witha492nmreferencefilter. TheIgGsubclassof1042and1618wasdeterminedbyamodifiedsandwichELISA(datanotshown).A96-wellImmulon1Bplate(DYNEXTechnologies)wascoatedwith100l/wellof2g/mlgoatanti-mouseIgG1, IgG2a,IgG2b,orIgG3overnightat4C.Theplatewaswashed3timeswith PBS/0.5%Tween-20followedbyblockingwith200l/well1%BSAinPBS for1houratroomtemperature.Weadded100l/wellof4g/mlbiotinylated1042or1618induplicatetowellscoatedwitheachoftheisotypes. AstandardconsistingofpurifiedmouseIgG1,IgG2a,IgG2b,orIgG3was alsoappliedtowellsinduplicateat100l/wellfor1houratroomtemperature.Theplateswerewashed5timeswithPBS/0.5%Tween-20.Streptavidin-HRPwasdiluted1:5,000inPBSandappliedat100l/welltowellsin which1042or1618wereapplied.Goatanti-mouseIgG(total)wasdiluted 1:2,000andappliedat100l/welltostandardwellsfor1houratroom temperature.Theplateswerewashed7timeswithPBS/0.5%Tween-20.The ELISAwasthendevelopedfor30minutesusing50l/well1%H2O2in ABTSandreadat405nmwitha492nmreferencefilter. Purified1042,1618,andHB5antibodiesandunmodifiedfibrinogen, citrullinatedfibrinogen,andCIIwerebiotinylatedbyadding20mmolexcess sulfo-NHS-LC-biotin(PierceBiotechnology)andincubatingfor2hourson ice.UnboundbiotinwasremovedbydialysisagainstPBSfor48hours. Array analyses of sera and monoclonal antibodies.SynovialantigenarrayscontainingproteinsandoverlappingpeptidesrepresentingautoantigencandidatesinRAandCIAwereproducedaspreviouslydescribed(19).Arrays wereprobedwith1:150dilutionsofseraderivedfrommicewithCIAtolerizedwithLXPorLKP,or0.5gofpurifiedmonoclonalantibody,followed bydetectionofantibodybindingwithCy3-conjugatedanti-mouseIgG/M secondaryantibodyoraCy3-avidinconjugate(JacksonImmunoResearch LaboratoriesInc.).Thesignificanceanalysisofmicroarraysalgorithm(49) wasappliedtoidentifyantigenfeatureswithdifferencesinarrayreactivity betweenLKP-andLXP-treatedmicewithCIAorbetweenduplicatearrays probedwithD513,1042,1618,andsecondaryreagents.Clustersoftware (version2.11)andTreeViewsoftware(version1.60)wereusedtoorderand displaytheresults,respectively(50).Additionaldetailedprotocolshave beenpreviouslypublished(19,51).SupplementalTable1displaysselected antigensandtheirsequencesthatareonthisarray. Collection and Western blot of animal tissues.MicewereimmunizedwithCFA aloneorCIIinCFAasdescribedabove.Tissuewascollectedfrom3mice ofeachgroup:unimmunized;CFA,day7afterimmunization;CFA,day 35;CIA,day7;andCIA,day35.Skinatthesiteofimmunization,draining
971

The Journal of Clinical Investigation http://www.jci.org Volume116 Number4 April2006

research article
inguinalLNs,andsynoviumweredissectedfromthemiceandimmediately frozeninliquidnitrogenandstoredat70Cuntilprocessing.Tissuewas processedbyhomogenizationin1mlRIPAbuffer(1%TritonX100,0.5% deoxycholicacid,150mMNaCL,20mM-glycerophosphate,20mMTris, pH8.0,5mMEGTA,3mMMgCl2,and0.1%SDSinwater)containing1 mMDTTand1tabletproteaseinhibitorcocktail(CompleteMini,EDTA free;RocheAppliedScience).Homogenizedsampleswereplacedonice for15minutesandthencentrifugedat4C,10,000gfor15minutes.The supernatantwascollectedandthepelletdiscarded.Proteinconcentration wasdeterminedbytheBio-RadBSAproteinassay(Bio-Rad).Fromeach sample,20gproteinwasloadedonto10%Bis-Trisgelsandrunat80V untiltheloadingdyefrontranjustoffthegel.Theproteinwasthentransferredtonitrocelluloseat75Vfor1hour.Theblotwasblockedovernight at4Cwith10%milkinPBS.Primaryantibodies,eitheranti-citrulline (BiogenesisLtd.),D513,biotinylated1042,orbiotinylated1618monoclonalantibodies,wereappliedtotheblotataconcentrationof5g/mlfor1 houratroomtemperature.Afterwashingtheblot3timesfor10minutes withPBScontaining0.5%Tween-20,secondaryantibody,HRP-conjugated goatanti-rabbitIgG,oranti-mouseIgMwasdiluted1:2,000inPBS,or HRP-conjugatedstreptavidinwasdiluted1:5,000inPBSandappliedto theblotfor1houratroomtemperature.Theblotwasagainwashedwith
1.Arnett,F.C.,etal.1988.TheAmericanRheumatismAssociation1987revisedcriteriafortheclassificationofrheumatoidarthritis.Arthritis Rheum. 31:315324. 2.Aho,K.,Palusuo,T.,andKurki,P.1994.Marker antibodiesofrheumatoidarthritis:diagnosticand pathogeneticimplications.Semin. Arthritis Rheum. 23:379387. 3.Nienhuis,R.L.,andMandema,E.1964.Anewserum factorinpatientswithrheumatoidarthritis;the antiperinuclearfactor.Ann. Rheum. Dis.23:302305. 4.Young,B.J.,Mallya,R.K.,Leslie,R.D.,Clark,C.J., andHamblin,T.J.1979.Anti-keratinantibodiesin rheumatoidarthritis.Br. Med. J.2:9799. 5.Vincent,C.,etal.1989.Highdiagnosticvaluein rheumatoidarthritisofantibodiestothestratumcorneumofratoesophagusepithelium,socalledantikeratinantibodies.Ann. Rheum. Dis. 48:712722. 6.Hoet,R.M.,Boerbooms,A.M.,Arends,M.,Ruiter, D.J.,andvanVenrooij,W.J.1991.Antiperinuclear factor, a marker autoantibody for rheumatoid arthritis:colocalisationoftheperinuclearfactor andprofilaggrin.Ann. Rheum. Dis.50:611618. 7.Simon,M.,etal.1993.Thecytokeratinfilamentaggregatingproteinfilaggrinisthetargetofthe so-calledantikeratinantibodies,autoantibodiesspecificforrheumatoidarthritis.J. Clin. Invest. 92:13871393. 8.Steinert,P.M.,Cantieri,J.S.,Teller,D.C.,LonsdaleEccles,J.D.,andDale,B.A.1981.Characterization ofaclassofcationicproteinsthatspecificallyinteractwithintermediatefilaments.Proc. Natl. Acad. Sci. U. S. A.78:40974101. 9.Resing, K.A., Dale, B.A., and Walsh, K.A. 1985. Multiplecopiesofphosphorylatedfilaggrininepidermalprofilaggrindemonstratedbyanalysisof trypticpeptides.Biochemistry.24:41674175. 10.Harding,C.R.,andScott,I.R.1983.Histidine-rich proteins(filaggrins):structuralandfunctionalheterogeneityduringepidermaldifferentiation.J. Mol. Biol.170:651673. 11.Watanabe,K.,etal.1988.Combinedbiochemical andimmunochemicalcomparisonofpeptidylargininedeiminasespresentinvarioustissues.Biochim. Biophys. Acta.966:375383. 12.Schellekens,G.A.,etal.1998.Citrullineisanessential constituentofantigenicdeterminantsrecognized byrheumatoidarthritis-specificautoantibodies. J. Clin. Invest.101:273281.
972

PBS/0.5%Tween-203timesfor10minutes.ECLreagent(AmershamPharmaciaBiotech)wasappliedfor1minute.Themembranewasthentransferredtoaplasticbagandfilmwasexposed. Statistics. All statistical analyses were performed using GraphPad Prism version 4 (GraphPad Software). Specific tests performed are notedwithinfigurelegends.

Acknowledgments FundingforthesestudieswasprovidedbytheSmythProfessorshipinRheumatology(V.M.Holers),MedicalScientistTraining ProgramgrantT32GM008497(K.A.Kuhn),NIHNHLBIcontractN01HV28183,andtheDepartmentofVeteransAffairs (W.H.Robinson). ReceivedforpublicationApril21,2005,andacceptedinrevised formJanuary3,2006. Addresscorrespondenceto:V.MichaelHolers,DepartmentofMedicine,UniversityofColoradoHealthSciencesCenter,BoxB115,4200 EastNinthAvenue,Denver,Colorado80262,USA.Phone:(303) 315-7952;Fax:(303)315-5540;E-mail:michael.holers@uchsc.edu.

mediatingcollagen-inducedarthritis.Cell Immunol. 172:2128. 25.Mussener, A., Litton, M.J., Lindroos, E., and Klareskog,L.1997.Cytokineproductioninsynovialtissueofmicewithcollagen-inducedarthritis (CIA).Clin. Exp. Immunol.107:485493. 26.Thornton,S.,Duwel,L.E.,Boivin,G.P.,Ma,Y.,and Hirsch,R.1999.Associationofthecourseofcollagen-inducedarthritiswithdistinctpatternsof cytokineandchemokinemessengerRNAexpression. Arthritis Rheum.42:11091118. 27.Mauri,C.,Williams,R.O.,Walmsley,M.,andFeldmann,M.1996.RelationshipbetweenTh1/Th2 cytokinepatternsandthearthritogenicresponse in collagen-induced arthritis. Eur. J. Immunol. 26:15111518. 28.Watson,W.C.,andTownes,A.S.1985.GeneticsusceptibilitytomurinecollagenIIautoimmunearthritis.ProposedrelationshiptotheIgG2autoantibody subclassresponse,complementC5,majorhistocompatibilitycomplex(MHC)andnon-MHCloci. J. Exp. Med.162:18781891. 29.Lopez-Hoyos,M.,etal.2003.DefectsintheregulationofBcellapoptosisarerequiredfortheproduc tionofcitrullinatedpeptideautoantibodiesinmice. Arthritis Rheum.48:23532361. 30.Myers, L.K., Stuart, J.M., Seyer, J.M., and Kang, A.H.1989.IdentificationofanimmunosuppressiveepitopeoftypeIIcollagenthatconfersprotectionagainstcollagen-inducedarthritis.J. Exp. Med. 170:19992010. 31.Myers,L.K.,etal.1993.TcellepitopesoftypeII collagenthatregulatemurinecollagen-induced arthritis.J. Immunol.151:500505. 32.Carel,J.C.,Myones,B.L.,Frazier,B.,andHolers, V.M. 1990. Structural requirements for C3d,g/ Epstein-Barrvirusreceptor(CR2/CD21)ligand binding,internalization,andviralinfection.J. Biol. Chem.265:1229312299. 33.Molina,H.,etal.1991.AnalysisofEpstein-Barr virus-binding sites on complement receptor 2 (CR2/CD21)usinghuman-mousechimerasand peptides. At least two distinct sites are necessaryforligand-receptorinteraction.J. Biol. Chem. 266:1217312179. 34.Vossenaar,E.R.,etal.2003.Citrullinationofsynovial proteins in murine models of rheumatoid arthritis.Arthritis Rheum.48:24892500. 35.Terato,K.,etal.1992.Inductionofarthritiswith monoclonal antibodies to collagen. J. Immunol.

13.Girbal-Neuhauser,E.,etal.1999.Theepitopestargetedbytherheumatoidarthritis-associatedantifilaggrinautoantibodiesareposttranslationallygeneratedonvarioussitesof(pro)filaggrinbydeimination ofarginineresidues.J. Immunol.162:585594. 14.Schellekens,G.A.,etal.2000.Thediagnosticpropertiesofrheumatoidarthritisantibodiesrecognizingacycliccitrullinatedpeptide.Arthritis Rheum. 43:155163. 15.Rantapaa-Dahlqvist, S., et al. 2003. Antibodies againstcycliccitrullinatedpeptideandIgArheumatoidfactorpredictthedevelopmentofrheumatoidarthritis.Arthritis Rheum.48:27412749. 16.Nielen,M.M.,etal.2004.Specificautoantibodies precedethesymptomsofrheumatoidarthritis:a study of serial measurements in blood donors. Arthritis Rheum.50:380386. 17.Masson-Bessiere,C.,etal.2001.Themajorsynovialtargetsoftherheumatoidarthritis-specific antifilaggrinautoantibodiesaredeiminatedforms ofthealpha-andbeta-chainsoffibrin.J. Immunol. 166:41774184. 18.Vossenaar,E.R.,etal.2004.Rheumatoidarthritisspecificanti-Saantibodiestargetcitrullinated vimentin.Arthritis Res. Ther.6:R142R150. 19.Hueber,W.,etal.2005.Antigenmicroarrayprofilingofautoantibodiesinrheumatoidarthritis. Arthritis Rheum.52:26452655. 20.Courtenay,J.S.,Dallman,M.J.,Dayan,A.D.,Martin, A.,andMosedale,B.1980.Immunisationagainst heterologoustypeIIcollageninducesarthritisin mice.Nature.283:666668. 21.Corthay,A.,Johansson,A.,Vestberg,M.,andHolmdahl,R.1999.Collagen-inducedarthritisdevelopmentrequiresalphabetaTcellsbutnotgamma deltaTcells:studieswithTcell-deficient(TCR mutant)mice.Int. Immunol.11:10651073. 22.Ranges,G.E.,Sriram,S.,andCooper,S.M.1985. PreventionoftypeIIcollagen-inducedarthritis byinvivotreatmentwithanti-L3T4.J. Exp. Med. 162:11051110. 23.Wooley,P.H.,Luthra,H.S.,Stuart,J.M.,andDavid, C.S.1981.TypeIIcollagen-inducedarthritisin mice.I.Majorhistocompatibilitycomplex(Iregion) linkage and antibody correlates. J. Exp. Med. 154:688700. 24.Rosloniec,E.F.,Whittington,K.B.,Brand,D.D., Myers,L.K.,andStuart,J.M.1996.Identification ofMHCclassIIandTCRbindingresiduesinthe typeIIcollagenimmunodominantdeterminant

The Journal of Clinical Investigation http://www.jci.org Volume116 Number4 April2006

research article
148:21032108. 36.Maccioni, M., et al. 2002. Arthritogenic monoclonalantibodiesfromK/BxNmice.J. Exp. Med. 195:10711077. 37.Vossenaar,E.R.,etal.2004.Absenceofcitrullinespecificautoantibodiesinanimalmodelsofautoimmunity.Arthritis Rheum.50:23702372. 38.Hang, L., Theofilopoulos, A.N., and Dixon, F.J. 1982.Aspontaneousrheumatoidarthritis-likediseaseinMRL/lmice.J. Exp. Med.155:16901701. 39.Hida,S.,Miura,N.N.,Adachi,Y.,andOhno,N. 2004.Influenceofargininedeiminationonantigenicityoffibrinogen.J. Autoimmun.23:141150. 40.Suzuki,A.,etal.2003.Functionalhaplotypesof PADI4,encodingcitrullinatingenzymepeptidylargininedeiminase4,areassociatedwithrheumatoid arthritis.Nat. Genet.34:395402. 41.Boyde,T.R.,andRahmatullah,M.1980.Optimizationofconditionsforthecolorimetricdeterminationofcitrulline,usingdiacetylmonoxime.Anal. Biochem.107:424431. 42.Vossenaar,E.R.,etal.2004.Expressionandactivity of citrullinating peptidylarginine deiminase enzymes in monocytes and macrophages. Ann. Rheum. Dis.63:373381. 43.Backlund,J.,etal.2002.GlycosylationoftypeII collagenisofmajorimportanceforTcelltolerance andpathologyincollagen-inducedarthritis.Eur. J. Immunol.32:37763784. 44.Backlund,J.,etal.2002.PredominantselectionofT cellsspecificfortheglycosylatedcollagentypeIIepitope(263-270)inhumanizedtransgenicmiceand inrheumatoidarthritis.Proc. Natl. Acad. Sci. U. S. A. 99:99609965. 45.Utz,P.J.,Hottelet,M.,Schur,P.H.,andAnderson, P.1997.Proteinsphosphorylatedduringstressinducedapoptosisarecommontargetsforautoantibodyproductioninpatientswithsystemiclupus erythematosus.J. Exp. Med.185:843854. 46.Arentz-Hansen,H.,etal.2000.TheintestinalT cellresponsetoalpha-gliadininadultceliacdiseaseisfocusedonasingledeamidatedglutamine targetedbytissuetransglutaminase.J. Exp. Med. 191:603612. 47.Anderson,R.P.,Degano,P.,Godkin,A.J.,Jewell, D.P.,andHill,A.V.2000.Invivoantigenchallenge inceliacdiseaseidentifiesasingletransglutaminase-modifiedpeptideasthedominantA-gliadin T-cellepitope.Nat. Med.6:337342. 48.Dieterich,W.,etal.1997.Identificationoftissue transglutaminaseastheautoantigenofceliacdisease. Nat. Med.3:797801. 49.Tusher,V.G.,Tibshirani,R.,andChu,G.2001.Significanceanalysisofmicroarraysappliedtotheionizingradiationresponse.Proc. Natl. Acad. Sci. U. S. A. 98:51165121. 50.Eisen, M.B., Spellman, P.T., Brown, P.O., and Botstein,D.1998.Clusteranalysisanddisplayof genome-wideexpressionpatterns.Proc. Natl. Acad. Sci. U. S. A.95:1486314868. 51.Robinson,W.H.,etal.2002.Antigenarraysformultiplexcharacterizationofautoantibodyresponses. Nat. Med.8:295301.

The Journal of Clinical Investigation http://www.jci.org Volume116 Number4 April2006

973