T cell receptor V beta gene bias in rheumatoid arthritis

T cell receptor V beta gene bias in rheumatoid

arthritis.

R N Jenkins, … , K Meek, P E Lipsky

J Clin Invest.

1993;

92(6)

:2688-2701.

https://doi.org/10.1172/JCI116886

.

Polymerase chain reaction (PCR) technology was employed to examine peripheral blood

and synovial T cells in patients with rheumatoid arthritis (RA) for biased utilization of T cell

receptor (TCR) variable region (V) genes. Oligonucleotide primers specific for individual

TCR V beta gene families were used to amplify TCR gene products in a semiquantitative

assay of their relative utilization in unselected T cell populations. Mean V beta expression in

24 RA peripheral blood samples was very similar to that in a panel of 15 normal subjects,

except for a slight decrease in V beta 13.2 expression. V beta utilization in 8 RA synovial

tissue samples and 13 synovial fluid samples was compared to simultaneously obtained

blood samples. Although heterogeneous patterns of skewed V beta utilization were

observed, several significant trends emerged. By a number of approaches to data analysis,

a statistically significant increase in expression of V beta 6 and V beta 15 in synovial T cells

was documented. In addition, increased synovial expression of V beta 14 was found, but

only in the synovial fluid samples. Reduced expression of V beta 1, V beta 4, V beta 5.1, V

beta 10, V beta 16, and V beta 19 was also observed in synovial T cells. These results

indicate that biased V beta gene utilization in different peripheral compartments of RA […]

Research Article

Find the latest version:

T

Cell Receptor VB

Gene Bias in Rheumatoid Arthritis

Robert N. Jenkins, Afzal Nikaein, Alice Zimmermann, Katheryn Meek, and PeterE.Lipsky

The Harold C. SimmonsArthritis Research Center andDepartment ofInternalMedicine, University ofTexas SouthwesternMedical School, Dallas, Texas75235;andDepartment of TransplantImmunology, Baylor UniversityMedicalCenter, Dallas, Texas 75246

Abstract

Polymerasechainreaction(PCR) technologywasemployedto

examineperipheralblood andsynovialT cells inpatientswith rheumatoid arthritis(RA)for biased utilization of T cell recep-tor(TCR)variableregion (V)genes.Oligonucleotide primers specificfor individual TCR

V,6

gene familieswereusedto

am-plifyTCRgene products ina semiquantitative assay of their relative utilizationinunselected T cellpopulations.MeanVB expressionin 24 RAperipheralbloodsampleswasvery similar to thatin a panel of 15 normal subjects, except for a slight decrease in

V,813.2

expression.

VjO

utilizationin8RAsynovial tissuesamples and 13synovialfluidsampleswascomparedto simultaneously obtained blood samples. Although heteroge-neouspatternsofskewedV,@utilizationwereobserved,several significanttrendsemerged. Byanumberofapproachestodata analysis, a statistically

significant

increase in expression of

V/i6

and

V#I5

insynovialTcells was documented. Inaddition, increased synovialexpressionof V/i14 wasfound, butonly in thesynovial fluid samples. Reduced expressionof

V#1, Vj64,

V/i5.1, V/i10, V,816,

andV/i19wasalso observed insynovial T cells. These results indicate thatbiased

V,8

geneutilizationin different peripheralcompartments ofRApatients canbe ob-served inunselectedTcellpopulations,and areconsistentwith theconclusionthatpopulationsofTcellsexpressingthese V/i gene products may be involved in thepathogenesisof the dis-ease.(J. Clin.Invest. 1993.92:2688-2701.)Key words: poly-merasechain reaction * rheumatoid arthritis-synovialT

cells-T cellreceptor * T cellrepertoire

Introduction

Rheumatoidarthritis(RA)isanidiopathic disease character-ized by chronicinflammation of synovium, local destruction ofcartilageandbone,andsystemic illness.Severallines of evi-denceimplicateTcellswhich respondtoantigenic stimuli via theira/f T cell receptor

(TCR)'

in thedisease process ( 1 ). The

majority

ofRApatientsexpress productsofa setofalleles

Addressreprintrequests to Dr. Robert N. Jenkins, Division of

Rheu-matic Diseases,Departmentof Internal Medicine, University of Texas

SouthwesternMedical Center, 5323 Harry Hines Blvd., Dallas, TX

75235-8884.

Receivedfor publication 5March1993 and in revisedform 20May 1993.

1.Abbreviations used in this paper: C or V, constant or variable region (gene); DMARD, disease-modifyingantirheumaticdrug; NSAID,

non-steroidalantiinflammatorydrug; RF, rheumatoid factor; RT, reverse

transcriptase.

at the B 1 locus ofthe HLA-DR region, HLA-DRB 1*0401

(HLA-DR4, Dw4), *0404/*0408 (DR4, Dw14), *0405

(DR4, Dwl5),*0101(HLA-DR1), and*1402(HLA-DRwl4,

Dw16), that areassociated withincreased risk for acquiring the disease (2-7). These classIIMHCmolecules share a sequence of amino acids in the thirdhypervariable region (residues 70-74)ofthe

_HLA-DRfl

chain, indicatingthatthis epitopemay confer disease susceptibility (8). Because the major rolesof class IImolecules are tobindandpresentantigenicpeptides to CD4+ T cells or to exert aninfluenceonthymic selection ofthe CD4+ T cellrepertoire,thesefindingsarecompatiblewith the conclusionthat theintroduction ofanunidentifiedpathogenic antigen intothesynovialmembraneofanimmunogenetically susceptiblehosttriggersa CD4+ T cell-mediated autoimmune responseresulting in the manifestations of RA.

Anumberof studieshaveestablished the

ability

of

antigen

or autoantigen plusMHC combinations toselect for T cells expressing

particular

TCRgenes (9). There is also evidence that thevariableregion (V)gene

expression

profile

of

periph-eralTcells can be influencedbythe MHC molecules

encoun-tered

during thymic

education (10-13). In

addition,

it has

been postulated that

superantigens,

characterized in part by theirabilitytostimulateTcells

utilizing particular

V/I genes, mayplayaroleinthedevelopment of

autoimmunity including

RA(14, 15). Thus, thereareseveral factorsthatcouldlead to thedevelopment of biased TCR

expression

in RA.

Initial studies sought evidence of biased TCRexpression in RAbylooking forthe presenceofoligoclonalTcells.Evidence of

oligoclonality

manifested by the presence of rearranged bands detectedby Southern

blotting

wasrarely found inTcells directly isolated from

synovial

fluid samples (16, 17). More-over, Tcell clonesestablished in culture conditions

designed

to allowoutgrowth of all cells rarely exhibited identical

rearrange-ments (18, 19). However, some studieswere able to detect

oligoclonality

of

synovial

Tcells selectedforthe

ability

togrow in IL-2alone(20, 21).Evenwhenoligoclonalitywasclaimed, however,there was no consensus onthespecific dominant

rear-rangement in differentpatient samples (22, 23). Taken

to-gether, these studies indicate that there is minimal

oligoclonal-ityintherheumatoidsynovium, althoughselection for IL-2-responsiveTcells mayinduce outgrowth ofalimitednumber of clones.A

major

limitation ofthese analyses, however, is that examination of cells for identical rearrangement patterns mightnotdetectthepredominance ofanantigen-or superanti-gen-reactive T cell population utilizing particular Va or VB genefamilies.

Severalstudieshaveexamined RAsynovial fluidor syno-vialtissuefor restricted expression of TCR Vaor VB. Some haveinterpreted theirresults toindicatethat the numberof

_Vfl,

but notVa,genefamilies expressedinsynovialTcellsis lim-ited (24), whereas others have reported that,atleast inpatients with earlydisease, alimitednumberofVa, but not

V/I,

gene familiesareexpressed(25). Although variablenumbersofVa

andVB productswerealso detected in otherstudiesof

synovial

J.Clin.Invest.

©TheAmericanSocietyfor Clinical Investigation,Inc.

fluidorsynovial tissuesamples (14,26),two recentanalysesof synovial fluid Tcellsdetected all of the V:families inevery patient(27, 28). Thus, no consensusconcerningthepotential

biasedexpression of V: or Va in RA synoviumhas emerged

fromthesestudies,mostof which examinedsmall numbersof patients. Some studies included comparisons ofVgene expres-sioninrheumatoid peripheralblood andsynovialTcells. Anal-yses of increased synovial T cell expression ofindividual V genefamilies have generallydemonstratedheterogeneous pat-terns(27-30). However, Paliard etal. (14) reported that, in nine of nine RA patients, expression ofVl14 was greater in synovial fluidTcells thanintheperipheralblood.Anincrease in

Vf14

expressioninsynovial fluidTcells incomparisonto peripheralblood wasfoundin one otherstudy (29),but notin twoothers(27, 28).Therefore,thequestionofwhetherthereis biased representationoractivation ofTcellsexpressing particu-lar Va orVf3productsin rheumatoid synovium remains unre-solved.

Thepresent reportof24 RApatientsusedsemiquantitative PCRamplification ofcDNA with

Vf3-specific

primersto test the possibility that T cells in the peripheral blood, synovial fluid,orsynovial tissue exhibit biasedVfl geneutilization. Sev-eralapproachestodata

analysis

indicate biasedV:gene utiliza-tionby T cells inpatients withRA.Althoughthedifferencesare small, several

statistically significant

trends werenoted, with eitherenrichmentor

depletion

of

particular

V:genes noted in thesynoviumorthe

peripheral

blood. These dataindicatethat biased V:gene

family

utilization indifferent

peripheral

com-partmentsof rheumatoid patientcanbeobservedin unselected Tcellpopulationsand areconsistent with the conclusionthat TcellsexpressingtheseparticularTCR geneproductsmay be involvedin thepathogenesis ofRA.

Methods

Subjects.24patients who fulfilled the 1987 American College of Rheu-matology criteria (31 )forRA werestudied. Synovial tissue was ob-tained from 8 patientsundergoingsynovectomyorjoint replacement surgery, andsynovial fluidwasobtainedfrom 11 patients undergoing

therapeutic arthrocentesis. Synovial fluid wasobtained fromtwo of thesepatientson twooccasions foratotalof 13synovialfluid samples. Peripheral blood was obtained from eachof these patients, and from 5 similarpatientsfrom whomanalysisofsynovialTcellswas not

success-ful. Peripheral blood andsynovialtissueorsynovial fluid samples were obtained from 4control patientswith other arthritides. The disease characteristics, age, sex, and medications used by these subjectsatthe time sampleswereobtainedarepresented in Table I. The presenceor

absence ofcirculatingrheumatoid factorswasdetermined by nephe-lometry, exceptfor a few cases in which a latex fixationtest wasused.

Peripheralblood samples from 15subjects who didnothavearthritis servedasnormal controls.

HLAtyping.Peripheral blood lymphocyteswereusedfor serologic

HLAclass IItyping by thelymphocytotoxicitymethod(32). Oligonu-cleotidetypingofHLAclass II geneswasthenperformed accordingto

previously published methods (33, 34).DNAwasextractedfrom

lym-phocytesasdescribed (35), andamplifiedby PCRusing previously describedprimersforDR(3 (33), DQB (36),DP#(36),andDR4

sub-typing. Previouslydescribedoligonucleotideswereusedforprobingdot

blotsof the

_DRf3

(33), DP#(36), DQB (34), and DR4(36) PCR products.

CellpreparationforTCRanalysis.PBMCandsynovialfluid

mono-nuclear cellswereprepared fromheparinizedsamples by density gra-dientcentrifugationoverficoll diatrizoategradients.These cellswere

either used directly for RNA preparation or furtherenriched for T cells

by rosetting with neuraminidase-treated sheep erythrocytesand/or

passage over anylon wool column. Lymphocytes wereprepared as

described(37)from thosesynovialtissuesamplesnotuseddirectlyfor

RNApreparation.Briefly, tissuewasminced with scissors anddigested

withI mg/ml collagenasefor 1-2h at370C,and thenpassedrapidly

over anylonwoolcolumn.

RNAand cDNA preparation. RNAwasdirectly preparedfrom

syno-vialtissuesamples byhomogenizationinguanidinethiocyanatewitha

tissuehomogenizer followedby ultracentrifugationthroughaCsCl

gra-dientasdescribed(38).Cell

preparations

were

homogenized

in

guani-dine thiocyanate by repeatedpassagethrough asmallgauge needle. Total RNA wasextracted from cell homogenates aspreviously de-scribed(39). 1-6 ,ugof RNAwasconvertedtocDNAbyincubation with 8 U avianmyeloblastosisvirusreversetranscriptase(RT)and 20

Ag/ml

oligo-dT primersin thepresenceofactinomycinD(50 ,g/ml)

for 1 hat420C. The cDNAwasheatedto950C for 5 min and then dilutedto afinal volumeequaltothe greater of 25,lper 1 jugofRNA,

or80

yl.

TCR,8-chainPCR. Inpreliminary experiments,each cDNAwas examined, alongsideapositive control,inconstantregion (C)

amplifi-cationstodetermine the relativeamountof TCR-derivedcDNA. Cfl

amplificationswere done with primers 5'Cfl:

GTGTTTGAGCCA-TCAGAA-GCA and

3'C#6:

AAGCCACAGTCTGCTCTACC. For the RNAsamples from patients RA5 throughRA24 (Table I)and from all of the control subjectsa negativecontrol cDNA, prepared

fromRNAwithout added RT, wastestedwithCfl region primersto ensuretherewas nocontaminatingTCRDNA.Productsweresampled

after 20, 25, 30, and 35 cycles todetermine the number ofcycles

through which thespecific product accumulated exponentially (see Fig. 1 A). The subsequentsemiquantitativeassay utilized the 22 VB

family-specific5'primersdesignedby Choietal.(40) and thereverse

primer 3'CB4: ACCCACCAGCTCAGCTCCA in separate

simulta-neousPCRreactions. The number of cycles used(usually 22-30)was designedtostopthereactions while all of theproductswerestill accu-mulating exponentially,asdetermined in thepreliminary experiment.

Thesestepsweretakento ensurethat each PCRproductwas

propor-tionaltotheamountofcorrespondingmRNA.Inaddition,when possi-ble,allof thereagentswerealiquotedfroma mastermixsothat pipet-tingvariabilitydid notproduceartifacts in the PCR. Ineach experi-ment anegativecontrol, withnocDNAadded,was runin parallel for allprimerpairs. PCRamplification ofcDNA samplesfromsubjects RA-1,RA-2, CP-4,andoneof thesetsofsynovialfluidandperipheral

bloodsamplesfrom RA-3wasperformedinindividual tubes ina

reac-tion volume of 50,l.AmplificationofsamplesfromRA-7,NC-I1, and

oneoftwo setsofsynovialfluid mononuclear cellandPBMCsamples

fromRA- 1Iwascarriedoutintriplicatein tubes. PCRamplificationof theremaining 60 cDNA samples wascarriedout intriplicate 20-,ul

reactionsusing96-wellmicrotiterplatesinamodel PTC-100-96

ther-malcontroller(MJ Research,Watertown, MA).Triplicate

amplifica-tionswereperformedtofurther thereproducibilityof theassay.The standarderrorof the threereplicatedeterminationswasusually<10%

of thecorrespondingmean.Thereaction mixture contained 0.2mM dNTPs(20 nMeachofdATP, dCTP, dGTP, anddTTP), 1.5 mM MgCl2,0.3

MuM

oligonucleotideprimers,and20-40U/ml TaqDNA polymerase (PromegaCorp.,Madison, WI).Thetemperaturesused in thePCRwere94°Cfordenaturation, 50°Cforannealing,and72°Cfor

polymerization.Theduration of eachstepwas 1 min.

Determinationof Vf genefamilyutilization.Asampleofeach PCR amplification wastransferred toanylonmembrane(Zeta-ProbeGT,

Bio-RadLaboratories,Richmond, CA) byalkalinetransferusingaslot blotapparatus. Blotswereincubated inhybridization fluid₍₃₈₎

con-taining6x sodium chloride, sodium citrate (SSC), 10% Denhardt's

reagent, 100mg/mldenatured,fragmentedsalmon spermDNA,and

0.5% SDS foratleast 1 hat37°C.Thespecificfl-chainproductswere

thendetectedbyincubation ofthe blotsfor 3-24hat37°Cin

hybridiza-tion fluid containing

[-Y-32P]-labeled

# probe:

ATTCTCCCACAC-CCAAAAGG, derived fromtheCB region. The ,Bprobe

oligonucleo-tide was end-labeled with T4 polynucleooligonucleo-tide kinase using standard

TableI.Subjects Studied

Age/

Donor Diagnosis duration Race/Sex RF Diseasecharacteristics Medications DR DQ Samples

RA 52/1 RA 73/7 RA 56/2 RA 62/25 RA 62/17 RA 37/1 RA 58/12 RA 26/4 RA 57/5 RA 61/19 RA 63/2 RA 26/1 RA 59/2 RA 47/4 RA 21/3 RA 24/1 RA 40/2 RA 57/10 RA 66/10 RA 70/3 RA 57/3 overlap 50/12 RA 42/1 RA 56/12 B/F W/F B/F B/F W/F B/F B/M W/F As/F W/F B/F L/F B/F W/F LA/F LA/F LA/F B/F B/F W/F B/F W/F W/F W/F

OA 44/6 B/M

OA 76/20 W/M JCA 24/20 W/M JCA 27/23 W/F

+ nodular, erosive, class 3 + erosive, class 3

+ erosive, class 3 + erosive,class3 + nodular, erosive, class 3 + class 2

+ nodular, erosive, class 3 + erosive,class3 - erosive, class 3 + erosive, class 3 + erosive, class 3

+ class2

+ erosive, class 2 + nodular,erosive, class 3

+ erosive, class 3 - erosive, class 3 - class2

+ nodular, erosive, class 3

+ erosive, class 3 + nodular, erosive, class 2

+ erosive,class3 + erosive, class 3 - nodular, class 2 + nodular, erosive, class 2

- erosive,class 2 - erosive,class2 30 B/F 31 W/F 41 B/F 53 W/F 36 B/F 29 W/M 36 W/M 37 W/M 28 W/M 31 W/F 30 A/F 33 W/M 25 W/F 32 B/M 32 W/M

Abbreviations used in this table: CyA, cyclosporine A;D-pen,D-penicillamine; HCQ,hydroxychloroquine;JCA,juvenile chronic arthritis; MTX, methotrexate;OA,

osteoarthritis; Pred, prednisone.

*Synovialfluidandperipheralbloodwereobtainedontwoseparate occasions from this patient.

*ND,_notdetermined.

methodology (38). Blotswerewashed inasolution of 6XSSC and 0.5% SDS for20minat50°C. Hybridizedprobewasquantified with

theRadioanalytic ImagingSystem (AMBIS SystemsInc.,SanDiego).

Backgroundcountsfromareasofslot blots containingnodetectable bands(watercontrols)weresubtracted with theAMBISsoftware.

Ac-cordingly,theamountof hybridizedprobeisexpectedtobe linearly relatedtotheamountof PCRproduct.The specificityof these

amplifi-cationswasconfirmedbyseparating samples ofproductsby electropho-resisonagarosegels, transferringtonylonmembranes, and probingas

described above.Inpreliminary experimentsitwasdetermined that all

Vfl products detectableon slot blots producedasingle band of the

correctsizeontheagarosegelblots.

Thedataareexpressedasthemeanof thetriplicate samples(where

applicable)of eachVB amplification divided by thesumofthe22

V#

amplifications.TheproportionofeachVB PCRproducttothesumof

all theV3 products will,bydesign,reflecttheproportion oftotalTCR

mRNAthatencodesaparticular VBgenefamily. ControlCfl

amplifi-cationswereperformedineachexperiment,butwerenotusedto

nor-malizetheindividualV,3 databecausethesumof theindividualVfl/

Cf ratios was not consistent from one cDNA to another.

Conse-NSAID NSAID, Gold NSAID NSAID, D-Pen NSAID,MTX NSAID,HCQ NSAID, Gold NSAID,MTX NSAID MTX, Pred NSAID,PRED, MTX NSAID NSAID,MTX NSAID, MTX, HCQ, Pred NSAID NSAID NSAID NSAID, Gold NSAID, Gold NSAID, PRED, MTX NSAID NSAID NSAID,HCQ NSAID, CyA NSAID NSAID NSAID, MTX Pred, NSAID RApatients RA-I RA-2 RA-3 RA-4 RA-5 RA-6 RA-7 RA-8 RA-9 RA-l0 RA-1I RA-12 RA-13 RA-14 RA-15 RA-16 RA-17 RA-18 RA-19 RA-20 RA-21 RA-22 RA-23 RA-24 Arthritic controls CP-1 CP-2 CP-3 CP-4 Normal controls NC-I NC-2 NC-3 NC-4 NC-5 NC-6 NC-7 NC-8 NC-9 NC-I0 NC- lI

NC-12 NC-13 NC-14 NC-15

2, 6, w52 4,11, w52, w53 3,11,w52 4, 8,w52,w53

1,4,w53 4, 11, w52, w53

4, 13,w52,w53 3,7,w52, w53 2, 12,w52 7, 8,w52, w53 3, 13, w52

4, 7,w53 2, 13,w52 3,4,w52, w53 2, 4,w53 7, 10, w53 1, 8,w52 4,6,w52,w53 3, 11, w52, w53 3, 4,w52,w53

ND$ 6, 7,w52, w53

3, 7,w52, w53

1,4,w53

ND ND ND

1,w53

ND

12, 13,w52 8, 10,w52 11,10,w52 6?, 11,w52,w53

7,-7, 11,w52, w53 1,4,w53

4,w6,w52,w53

7,9,w52,w53

2,4,w53 3, 4,w52, w53

8,-,w52 3, 11, w52 3, 4,w52,w53

quently,differences in an individual Vf3/Cf3ratio from one cDNA to anothermightreflect this inconsistency rather than true skewing. In contrast, whenthe data were normalizedwith the sum of the 22 VJ3 amplifications repeat experiments performed on the same cDNA pro-duced nearlyidentical V/3 expression profiles (see Results). This

con-sistency allowedmeaningful comparisons of different cDNAs. The abilityof this semiquantitative assay to detect selective expan-sion or activation ofTcellsbearing a particularVf3gene was demon-strated inpreliminary experiments and in experiments published else-where. PeripheralTcellsstimulated with monoclonal antibodies di-rectedagainstspecificV,3proteins in the presence of IL-2 produced at least a twofold increase in the amount of the appropriateV# PCR product even though there was no detectableincrease in the number of cells in culture (data not shown), demonstrating that increased

amounts ofmRNA resulting from activation ofVA3-specificT cells could be detected.Inaddition, proliferation driven by

_Vf3-specific

su-perantigens resulted inamarkedincrease of the appropriateV13PCR product (41 ).Finally,addition of small numbers of monoclonal T cell populations (5%) to Tcellsderived from the peripheral bloodof a normal donor waseasily detected (data not shown). These preliminary resultsconfirmed that thesemiquantitativePCR could detect changes in both theactivation state and number ofTcells expressingspecific VB gene productsin a heterogeneous population.

Statisticalanalysis. Vf3 expression in the peripheral blood and syn-ovium ofRApatients and in the peripheral blood of control subjects was not normallydistributed for everyVj3gene family. Therefore,

Wil-coxon'sranksum test wasapplied to comparisons ofVflexpression in

RApatients and control groups, and Wilcoxon's signed rank test was appliedtocomparisons of Vj3 expression in simultaneously obtained peripheral bloodandsynovial samples. Student's t test was applied to comparisons of V: expression in peripheral blood and synovium in eachexperimentperformed with triplicate PCR amplifications. Re-gression analysis was used to examine the overall similarity ofV3 ex-pression profilesindifferent experiments. The coefficient

ofdetermina-tion(r2) is reported, withr2= 1.0indicating perfect correlation, andr2 =0indicatingnocorrelation.

Results

Patient population. Samples were obtained from 24 RA pa-tients,4patients with other arthritides, and 15 normal donors. The clinical characteristics of these subjects are presented in Table I. A large majority ofthe RA patients had erosive synovi-tis and all of them werefunctionalclass II or III (42). Rheuma-toid factor(RF) was detected in the serum of 20 RA patients. A majority ( 15/24) of the RA patients was taking a disease-mo-difying antirheumatic drug (DMARD) at the time the samples wereobtained.Ofthe 23 patients on whom serologic HLA-typ-ing was available, 12 expressed either HLA-DR1or HLA-DR4

orboth. 5 ofthe 15 normal subjects tested positive for either or both ofthesemarkers. Of the 8 HLA-DR4+ patients who were DNAtyped,7expressed at least one of the susceptibility con-ferring alleles DRB1*0401, *0404, or *0408. For clarity of pre-sentation, serologic typingof HLA-A,B,andC, and DNA

anal-ysisof DP,DQ,and DR4 subtypes, are not shown.

RAperipheral blood.Fig. 1 illustrates the protocol used in theseexperiments for the analysis ofVflgenefamily expression in unselected T cell populations. The preliminary analysis of theperipheralblood cDNA of one subject indicated that TCR CJ3productaccumulatedexponentially for 30 PCR cycles (Fig.

1 A). Because theC3 primer pair amplifies all TCR 3

chain-derived cDNA, the individual

_V0

PCR products amplified

from this cDNA would also accumulate exponentially for at least 30 cycles. The subsequent experiment utilized 28 PCR cycles for thetriplicate amplificationof this cDNA with the 22

_I

(n

z D

0

Ca

m

A

NUMBER OF CYCLES

B

1.2,3

>- 4.5.1,5.2-3

-j

E 6.1-3.7,8

< 9.10,11

LL

LU 12.13.1,13.2-3

z

LU _14,15,16 CD

17.18.19

20

qf

C

204W4

19 _ _

18 4 _

17 - 4W

16 A

15 - _W _

14 _ 4W 4W 13.2-3

>13.1 _ 4W

E 12 _{aW _ _}

CX11 - 4W_ _ LL

LU 10 -_ _

-z

98 4W d _ 6.1-3

5.2-3 W

4 4:-4W

*_I

_ 4W

3

AW-~-2_41_

-j

LL

LUI z LU

(3

>

D

20 &\&

19

18

17

15

14

13.2-3

13.1.

12 11 10

9

B 7

6.1-3

5.2-3

5.1. 4 3

2

1

, ..

'.~~~

HYIDZE COUNT6°)( SE

HYBRIDIZEDCOUNTS(xlO3:kSEM)

Figure1. Determination ofTcell

Vflgenefamily utilization in PBMC

from patientRA-23.(A) PBMC

cDNA wasamplifiedwith 5'CBand

3'CQ6for 35PCRcycles. Samples were removedafter 20, 25, 30,and 35cycles, transferredto aslotblot,

andhybridized with radiolabeled

fprobe.

Nethybridized counts are

depictedonthey-axis. (BandC) PCRamplifications ofH20controls

containingnocDNA(B)andof PBMCcDNA(C)wereperformed for 28 cycles.Amplification ofa

control cDNA,prepared fromRNA

without the addition of RT, withCo

primers producednodetectable product after 28 cycles(datanot

shown). (B)Areaof the slotblot

containing samples oftheH20 con-trolsforthe PCRexperiment

de-picted in C. (C) PBMCcDNA was

amplifiedintriplicate usingthe 5'

primers depictedon theleft. Each slotrepresentsaseparate PCR

reac-tion.The slot blotwashybridized

with radiolabeled

fprobe,

washed,

andanalyzed withtheAMBIS

sys-tem.Theimage is presentedinlog

modeforclarity. (D)Nethybridizedcounts(mean±SEM)weredetermined foreachtriplicateandarepresentedonthex-axis.Theareaofthe

slotblotdepictedin Bwasusedasthesample background.

VB gene family specific 5' primers. The hybridized slot blots of thisexperiment are shown in Fig. 1, B and C. Quantifying of the nethybridizedcounts,depictedinFig. 1 D,indicatedalow degree of well-to-well variability in the triplicate PCR amplifi-cations. The normalized data from this experiment are de-picted in Fig.2. Each cDNA in this report wassubjectedto a similaranalysis.

CirculatingTcells from theblood of 24 RA patients with active disease were examined for biased TCR expression in comparisonwithTCRexpressionintheperipheral blood of15 normal donors. Although two blood samples were available from 2 RA subjects, only one sample from each was used in this analysis. Restricted VB gene family utilization was not ob-served in RApatientsnor innormal donors. For example, allof the 22Vfl productsweredetectedatlevelsexceeding 1%of the total in the bloodof patient RA-23 (Fig. 1 C, andFig.2). At least 16 ofthe 22 VB products represented 1% or moreofthe totalin everyblood sample;the meanforallofthe blood

sam-ples was 20 V3 products > 1% (data not shown). Every Vp

family comprised 1% or more of thetotal inthe peripheral blood of90% or moreof the subjectstested, except for V3IO and

V#l

1whichwereexpressedtothis levelin56%and 51%, respectively. The patternsofTCRexpression were heteroge-neousinboth RApatientsand controls. The data summarized in Table IIindicatethat the meanexpressionofeach

Vfl

gene

family

was notmarkedly different inRApatientsand normal donors. However, astatistically significantdecreasein Vf 13.2 expression in RA patientswas observed. Comparison ofthe

peripheral bloodTcell Vf expressioninthe 20 RF+ tothose of the normal donorsproduced results very similar to those ob-tainedby analyzingall of the RApatients (datanotshown).

Toaddress thepossibilitythat the RApatientswho express

HLA-DR 1 or HLA-DR4could represent a distinctsubset,

pe-ripheral bloodVflexpressionbythesepatientswascompared tothatof normaldonorsexpressingeither ofthese markers and the results are summarized in Table II. As in theprevious com-parisons

Vfl

13.2 expression was lower in RA patients. V38 expression was also significantly lower in the patients than in thecontrols. In addition,RA patientsinthisgroupexpressed significantly higher levelsofV,32and Vf35.1 thandid this

nor-mal donor group.

The potential effect of disease duration was explored by comparing peripheral bloodVf3expression insubjects having RAfor2yr or less, and thosehavingRAformorethan 2 yr, to eachotherand to the normalsubjects. There were no statisti-cally significant differencesbetween the group of 16 RA pa-tients with longer disease duration and normal donors, nor betweenthisgroupofpatientsand the groupof 8 with shorter disease duration. However, VB13.2expression (3.0±0.5%)was lower in thepatients withRAfor2yr or less thaninthe normal

donors (Table II) (P < 0.004), whereas V,320 expression

(4.1±0.5%) was higher in this group of RA patients (P

<0.039).

Similarly, the

possibility

that

peripheral

bloodVf expres-siondifferedinpatients taking nonsteroidalantiinflammatory drugs (NSAIDs) aloneandpatients takingDMARDs was

ex-TableI1. Vf3 Gene FamilyUtilizationinPeripheralBloodofRApatientsandNormalDonors

Donorsexpressing HLA-DRlor

All donors HLA-DR4

Normal donors RApatients Normal donors RApatients VOgenefamily (n=15) (n=24) (n=5) (n= 12)

Vo1

0.040±0.006 0.044±0.004 0.033±0.006 0.041±0.003

V#2 0.075±0.013 0.091±0.007 0.052±0.009 0.095±0.010t

Vfl3

0.053±0.010 0.060±0.009 0.085±0.018 0.080±0.015

V#4

0.075±0.010 0.066±0.005 0.066±0.007 0.071±0.009

V#5.1 0.043±0.006 0.047±0.004 0.029±0.004 0.053±0.006t

V#5.2-3 0.034±0.006 0.035±0.004 0.027±0.008 0.038±0.004

V#6.1-3

0.102±0.007 0.106±0.008 0.093±0.006 0.089±0.005

V#7 0.096±0.007 0.093±0.007 0.099±0.010 0.088±0.012

V#8

0.086±0.007 0.078±0.006 0.101±0.010 0.070±0.007t

V#9

0.044±0.005 0.049±0.004 0.054±0.006 0.049±0.006

V010

0.014±0.003 0.013±0.001 0.013±0.004 0.011±0.002

Voll

1 0.012±0.002 0.013±0.002 0.015±0.004 0.013±0.004

V#12 0.029±0.005 0.028±0.003 0.035±0.007 0.030±0.005

VO13.1 0.057±0.006 0.051±0.004 0.053±0.004 0.048±0.007

VO13.2 0.053±0.004 0.038±0.004* 0.053±0.005 0.029±0.004*

V#14 0.034±0.003 0.027±0.003 0.033±0.004 0.025±0.003

V,315 0.023±0.003 0.027±0.003 0.021±0.004 0.025±0.004

V#16

0.020±0.002 0.018±0.002 0.020±0.002 0.018±0.002

V#17 0.043±0.006 0.032±0.004 0.042±0.009 0.037±0.005

V#18

0.020±0.003 0.023±0.002 0.026±0.005 0.026±0.004

V019

0.021±0.003 0.027±0.006 0.020±0.005 0.032±0.011

V#20

0.028±0.004 0.035±0.005 0.030±0.004 0.033±0.005

Datarepresentmean±SEof the value

_VO/2;V/

determined for eachcase.

13.2

>. 13

20-

\\\\\\\mH

19 7

18 _I

17 \ \\

16-sM

,,,

15- ' 14- _{\ \}

as\

-3

-,._I.

I...--E

12-a LL ₁₁

c

10-a) g

OL

>

8I-7I

6.1-3

-5.2-3

-5.1.

-4

3-s

2- _

I--Em~-

Synovial tissue

IZ

1PBMC

*

-2p

F-\\ \ F-\\171

\\ Z B* * \\\MH

+-,,\',\\ l|

\ \ \ \ 71lA

+

EE!N\9

F-0.00 0.02 0.04 0.060.080.100.120.140.16 0.18 0.200.220.24

V

_±/EV#

iSE

Figure2. ComparisonofTcell

VB

genefamily utilization insynovial

tissue and PBMCfrom patient RA-23. RNAprepareddirectly from synovialtissueandwascomparedtothatisolated fromblood drawn

at thetimeof surgery. Significantly different(P<0.05) levelsofV:

gene familyexpressionarenoted(*).

amined by comparing thesegroupsto each other and tothe

normal donors. There were no statistically significant differ-ences in the peripheral blood expression ofany Vf3 families

between the group of 15 patients taking DMARDs and the

normal donors. Incomparison to normal donors (TableII),

the group of 9 patients taking NSAIDs alone exhibited reduced

expression of Vl3.2 (3.4±0.5%, P < 0.009) and

Vf14

(2.2±0.4,P< 0.026).

V:

1

expressionwasslightly higherinthe

patients takingNSAIDs alonethan in the patients also taking

DMARDs(5.4±0.7%vs. 3.8±0.3%, P<0.026).

Taken together,theseresultssuggestthat theentiregroupof

RA patientsexhibited differencesofaminor degree in V: gene

expression intheir peripheral blood incomparisontothe entire

control group of normal donors, butgreater differenceswere

observed if the comparison was limited to groupswhich

ex-pressed similarMHC classII

alleles. Moreover, these analyses

suggested thatperipheral bloodVi expression in thegroupof

patients who had RA for 2 yr or less, and in those taking

NSAIDsalone,differed from thatinnormal donorstoagreater

degree thandid thegroupwith longerduration diseaseorthose

taking DMARDs aswell.Peripheral blood wasalso obtained

from 4 arthritispatientswithdiagnoses other thanRAand the

Vi expression in thesesamples wascompared to that of the

normal donors. There were no statistically significant

differ-ences in themeanexpressionofanyV/ genefamilybetween

thesetwocontrolgroups(datanotshown).

Rheumatoid synovium. The possibility of skewing of

V:

expression in rheumatoid synovium compared to peripheral

blood was examined in 21 samples from 19 patients. Anumber of statistical analyses were carried out to validate the biases observed (see below). Initially, we sought evidence ofbiased VJ3 expression by comparison of the mean utilization ofthe group of synovial samples to that of the group ofcorresponding blood samples. In addition, each case was examined for a 50% difference between the synovial and blood T cells in thelevelof expression of each V13 family. Finally, those cases in which triplicate amplifications were carried out were examined for statistically significant differences in

V:

expression regardless of the magnitude of the disparity. Comparison of groupmeans has the potential advantage of revealing significant trends ap-parent in the entire group that could have been of modest mag-nitude in individual cases. On the other hand, if there was marked heterogeneity within the group, biologically meaning-ful differences in individual cases might be obscured. Tabula-tion of the number of individual comparisons in which a bias was either statistically significant or exceeded 50% was per-formed to address this possibility. In order to examine the possi-bility that biased

V:

utilization may be limited tocertain sub-sets of RA patients, these analyses were performed on: (a) the entire set of samples; (b) those in which either synovial tissue or synovial fluid was analyzed; (c) samples from RF+patients;

(d) samples from subjects known to express HLA-DR1 or

DR4; (e) samples from patients with RA for 2 yr or less, or for

more than 2 yr;

(f)

samples from patients takingDMARDs;

and (g) samples from patients taking NSAIDs alone.

The Vf expression of T cells in synovial tissue at thetimeof joint replacement or synovectomy was determined ineight RA patients, six of whom were seropositive forRF. Histologic ex-amination was performed on four of these synovialspecimens and confirmed the presence of an active inflammatory infil-trate in each case. Vfl expression in simultaneously obtained peripheral blood T cells was used for comparison. Anexample of this type of experiment is shown in Fig. 2. Results of the comparison of the expression of each

V,#

family in theeight synovial tissue samples to that of the correspondingperipheral blood samples are presented in Table

III.

There were statisti-cally significant increases in

V,36.1-3

andV/15 expression in synovial tissue as compared to peripheral blood, and a signifi-cant decrease in synovial tissue expression ofV/Il andV,35. 1.

Moreover, mean expression of

Vp

6 and Vi 15 in RA

synovial

tissue was significantly greater, and Vi5. 1 expression

signifi-cantly lower, than in normal donor peripheral blood (P

<0.034).

Synovial fluid samples obtained at the time oftherapeutic

arthrocentesis were also examined and compared to

simulta-neously obtained peripheral blood T cells in 13 RA cases (syno-vial fluid and blood were obtained from two patients on two

occasions). 12 of the samples were obtained from RF+

pa-tients. This analysis (Table

III)

revealed that V/i6.1-3 and

Vl

14 expression in the 13 synovial fluid T cell samples was significantly higher than that in the peripheral blood. On the other hand, synovial fluid expression of Vf4,

V,1

6, and

_V#ll

9 was significantly lower than in the peripheral bloodsamples. In

addition, mean expression of

VB

4,

Vl

6, and V/319 in RA

synovial fluid was significantly lower than in normal donor peripheral blood (P < 0.0 14).

As was the case with peripheral blood, restricted

_Vl#

gene family expression was not observed in RA synovium. Levels of

T Cell Receptor

V,

3 Gene Bias inRheumatoid Arthritis 2693

"I\ \ \ \ \ \ \ \ \ \\ \ \ -.,.,.,\ \ \\ \ \ \\ \

I

TableIII. Vf3GeneFamily Utilizationin RASynovium andSimultaneouslyObtained

Peripheral

BloodSamples

Synovialtissue samples Synovialfluidsamples Allsynovial samples HLA-DR-1/4 samples

(n=8) (n= 13) (n=21) (n=9)

V#gene

family Blood Tissue Blood Fluid Blood Synovium Blood Synovium

0.045±0.008 0.028±0.004* 0.084±0.012 0.114±0.022 0.044±0.014 0.023±0.003 0.065±0.005 0.041±0.009 0.040±0.003 0.022±0.006* 0.031±0.004 0.032±0.007 0.100±0.008 0.154±0.016* 0.108±0.012 0.118±0.013 0.094±0.011 0.074±0.011 0.056±0.008 0.060±0.012 0.012±0.002 0.007±0.002 0.014±0.003 0.009±0.001 0.026±0.003 0.020±0.004 0.047±0.005 0.062±0.008 0.047±0.007 0.037±0.006 0.033±0.009 0.029±0.005 0.018±0.002 0.037±0.005* 0.019±0.004 0.011±0.002 0.037±0.007 0.030±0.005 0.022±0.002 0.026±0.005 0.023±0.004 0.026±0.007 0.038±0.007 0.040±0.007

0.045±0.004 0.038±0.006 0.097±0.009 0.107±0.010 0.058±0.012 0.065±0.015 0.075±0.008 0.040±0.005§ 0.048±0.006 0.034±0.005 0.034±0.007 0.030±0.004 0.096±0.008 0.132±0.016* 0.089±0.01 0.087±0.008 0.082±0.008 0.094±0.015 0.046±0.005 0.050±0.006 0.015±0.002 0.012±0.002 0.014±0.003 0.013±0.003 0.029±0.004 0.029±0.007 0.054±0.007 0.054±0.007 0.036±0.005 0.031±0.004 0.024±0.003 0.031±0.004* 0.033±0.005 0.037±0.004 0.016±0.002 0.012±0.001* 0.030±0.004 0.033±0.007 0.020±0.002 0.021±0.002 0.024±0.004 0.011±0.002* 0.035±0.005 0.041±0.009

0.045±.004 0.034±.004* 0.044±.003 0.029±.004*

0.092±.007 0.109±.010 0.096±.012 0.100±.016 0.052±.009 0.049±.010 0.064±.018 0.057±.020 0.071±.006 0.040±.005§ 0.080±.010 0.042±.005$ 0.045±.004 0.030±.004* 0.051±.006 0.024±.003*

0.033±.005 0.031±.004 0.034±.004 0.028±.004 0.098±.006 0.140±.012§ 0.089±.007 0.138±.022*

0.096±.008 0.099±.008 0.098±.015 0.098±.015 0.086±.007 0.086±.010 0.076±.009 0.091±.020

0.050±.004 0.054±.006 0.049±.008 0.049±.006 0.014±.001 0.010±.001* 0.011±.002 0.007±.002* 0.014±.002 0.011±.002 0.014±.005 0.013±.004 0.028±.003 0.025±.004 0.026±.006 0.024±.005 0.051±.004 0.057±.006 0.046±.008 0.056±.009 0.040±.004 0.033±.004 0.036±.007 0.034±.007 0.028±.004 0.030±.003* 0.022±.004 0.028±.006$ 0.027±.004 0.037±.003* 0.024±.005 0.040±.003* 0.017±.002 0.012±.001* 0.017±.003 0.012±.002 0.033±.004 0.032±.005 0.038±.007 0.039±.009 0.021±.001 0.023±.002 0.022±.002 0.025±.005 0.024±.003 0.0 17±.003* 0.026±.006 0.018±.007

0.036±.004 0.041±.006 0.037±.006 0.050±.012

Datarepresentmean±SE of the value

_Vfl/2V(B

determinedfor eachcase.

Wilcoxon's signedranktest, *P<_0.05, P<0.01,orIP<0.001incomparisontocorrespondingbloodsamples.

expression ofatleast 17 of the 22 V,3 productswere > 1%in

everysynovial sample; themeanfor thegroupwas20V3

prod-uctsrepresenting 1% or more of the total. Every V,3 family

representedatleast 1%of the total in the synovium of 90%or

moreofthe RA patients,exceptVB10,Vj I1,V"132,Vj3 16, and VfB19 which were expressed atthis level in 38%, 43%, 81%, 67%, and 53%, respectively. Acomparison of theVf3 gene

ex-pression in all of the 21 synovial samples and the correspond-ing blood samples is presented in Table III. VB6. 1-3and

_Vo1

5

were significantly overexpressed in rheumatoid synovium

when all 21 sampleswereconsidered together (P<0.0005 and

P<0.05, respectively). Although synovial T cell expression of

Vfl14 expression was statistically significantly increased (P

<0.02), the difference in themeansofthe synovial and

periph-eralblood sampleswasextremely small (3.0%vs. 2.8%). The

synovial T cell expression of

_V#31,

V,34, V35. 1, V#I 0, Vj3 16, andV, 19wassignificantlylower than expression inthe

corre-spondingRAperipheral blood samples (P<0.05).In

compari-sontothe peripheral blood of normal donors, themean

expres-sionofVj36. 1-3 and VB 15 in the synovial sampleswas

signifi-cantlygreater(P<0.015 and P<0.003, respectively),and the

mean expression ofVf,4andVB16wassignificantlylower (P

.0.001).

Similarresultswereobtained when the comparisonwas

lim-itedtothe 18 RF+cases(datanotshown). Ifthe analysiswas

limitedtotheninecaseswhere thesampleswereobtained from

donors knowntobe HLA-DRl orDR4+,increased synovialT

cell V#6. 1-3,V 134, andV 135expression,and decreased

syno-vial T cell V#1, V34, V,35.1, andV,3I0expressionwereagain

noted(Table III). Moreover, in the synovium of thisgroupof

patients themean

_V#31

5 expressionwassignificantly higher (P

<0.01), and themean Vf4andV,3I0expressionwas signifi-cantly lower (P<0.0 14), than that in the peripheral blood of

normal donors knowntoexpressthese HLA antigens.

Individual analysis of rheumatoid synovium and periph-eralblood in the (a) 9 samples from patients with RA for 2yr orless;(b) 12 samples from patients with disease formorethan

2yr;(c) 8 samples from patients taking NSAIDs alone; and (d)

13 samples from patients also takingDMARDs,each revealed biased expressionofV,3 genefamiliesverysimilartothatseen

in the entiregroupof RA patients. Restricted expression of Vf3

genefamilieswasnotobserved inanyof thesegroupsof

syno-vial T cell samples.

Bias in individualcases. Itwasevident that all individual

comparisons of synovial and peripheral blood T cells didnot exhibit thesamepattern ofbiasedVgeneexpression. To

ex-plore the possibility that biased utilization ofsomeVB families

by synovial T cells occurs more frequentlythan anticipated

from the data in TableIII,weexamined all individual

compari-sonsof synovium and blood for differences in expression of

50% or more,reasoning that suchadifference is of sufficient

magnitudetobe ofpotential biologic importance.In the

exam-ple shown in Fig. 2, synovial tissue expression of VB5.2-3,

Vfl6.

1-3,

_V#31

5,and Vf l9wasatleast50%greaterthan thatof

peripheral blood in this patient (RA-23). On the other hand, Vf32, V,35.1, VB9, V,3I0, Vf I1, V 12, and

_V#316

expression

wasatleast 50%greaterin peripheral blood.

The number ofcasesinwhich expression ofaparticular

Vfl

Vo1

V,32 V#3 V#4 V35.1

Vf35.2-3 Vi36.1-3 Vfl7

VB8

V#9

VolO

Vo1

1

V#12

Vf313.1

Vf13.2 Vf314

V#15

V#16

V#317

Vf318 Vfli19

gene family differed by 50% or more in synovialtissue and peripheralblood isshown in Table IV. There are at least two

examples of biased expression of every V: gene family.

In-creased synovial tissue expression of V315 andVfB6.1-3 was

frequently seen, whereas V134, V,35.1, V/310,

V#12,

V3ll3.2,

and VB16 expression was frequently greater in theperipheral blood. Biased expression of V315 and Vf35. 1 in the synovial tissue wasfound in all three of the donors knownto express

DR 1 or DR4 or both.

Synovialfluidandperipheralbloodsampleswereobtained ontwooccasions9 mo apart from subject RA- 11. These two comparisons ofsynovial fluid and peripheral blood are de-picted in Fig. 3. At bothtimepoints it is clear that T cell V,3 expression isnotidenticalinthesynovialfluid andperipheral blood.Regression analysiswasused todetermine the degree of overall similarity of V,3 expression.Thecoefficient of determi-nation

(r2)

for peripheral bloodandsynovialTcell V,3 expres-sionwas0.78atthe first time point and0.84 at thesecond time point. This degree of similaritywasrelativelyhigh in compari-son to thatobservedinsimilar experiments; the mean r2 for the 21 comparisons was 0.57 with 95% confidence intervals of 0.47-0.67.ItisalsoevidentfromFig. 3 that peripheral blood (and synovial fluid)TcellVflexpressionwasnot identical at bothtimepoints.Betweentimepointsin RA- 11 the degree of similarity

(r2)

for peripheral blood Vj3 expression was 0.66, andforsynovial fluidV3expressionr2 =0.70. The variability of

Vfl

expressionintheperipheralblood of RA- 11 with time

TableIV. Skewed Vf Gene Family Expression in RA Synovial

Tissueand Fluid

Synovialtissue Synovialfluid All RF+

cases cases HLA-DR-1/4+

(n = 8) (n= 13) comparisons (n=8) VB gene

family Blood Tissue Blood Fluid Blood Synovium

V#1 2* 0 6 2 5 0

VB2 1 3 0 2 1 2

V#3 4 1 2 2 3 2

Vf34

5 1 9 0 5 0

V#5.1 6 1 7 1 6 0

V#5.2-3 2 2 4 2 1 1

Vfl6.1-3

0 4 0 2 0 4

V#7

1 2 2 2 1 2

Vfl8

3 1 1 2 0 2

Vf39

2 1 1 3 0 1

V#10

5 1 7 2 4 0

V#1

1 3 0 5 3 3 0

V,312 4 1 4 4 2 2

V#13.1 0 3 2 2 1 2

V 13.2 4 0 3 1 1 1

V#14 1 0 0 2 0 2

V,#15 0 6 3 6 0 3

Vf#16

4 0 5 0 2 0

Vf317 2 1 4 3 3 1

Vf318 1 3 3 1 0 0

Vj319

3 2 8 1 3 1

V,#20

2 1 1 2 0 1

*_{Number ofcasesin whichexpressionwas50% greater in indicated}

compartment.

was considerably less than that seen between this patient and the other RA patients where the mean r2 = 0.45 (95% confi-dence interval of 0.38-0.52). In control experiments, the

r2

for repeat analysis of the same cDNA was 0.93 for the first RA- 1I

PBMC cDNA, and 0.91 for the second RA- 11 PBMC cDNA.

These results indicate that differences observed in comparisons of different cDNAs are unlikely to be artifactual.

Expression of10 V3 families differed by 50% or more in the first comparison of peripheral blood and synovial fluid T cells in RA- 1 1, whereas 7V: families differed to this degree in the second comparison (Fig. 3). Decreased synovial T cell expres-sion of Vf34 and

_Vf3

19 was seen on both occasions, consistent with the overall similarity of both peripheral blood and syno-vial fluid Vf3 expression at the two timepoints. Samples were also obtained from subject RA-3 on two occasions 3 mo apart (data not shown). The coefficient of determination for syno-vial fluid and peripheral blood expression was 0.76 at the first time point and 0.63 at the second time point. The variability of peripheral blood VJ3 expression in this patient with time(r2

= 0.77) was less than that seen between individuals (see

above). Synovial fluid T cellV: expression exhibited a similar degree of variability with time(r2 = 0.68). A 50% increase in synovial fluid expression of Vfl7 and a 50% decrease in syno-vial fluid expression of V3 12 were observed on both occasions.

A 50% bias was observed in the expression of four other

V:

gene families in only the first comparison, and six others in only the second comparison.

A summary of the 13 synovial fluid analyses is presented in Table IV. Heterogeneous skewing continued to be noted as there are at least two examples of biased expression for eachVA gene family. Similar to the synovial tissue comparisons, in-creased synovial fluid expression of V315and greater

periph-eral blood expression of V34, Vf5.1,

VBI0,

and

_V#

1 6 was

frequently observed. In addition, there were several cases where synovial fluid expression of Vf 1 and V3 19 was substan-tially less than that in the peripheral blood. There were only

two cases in whom peripheral blood expression ofVB14 was

50% less than in the synovial fluid. Only one of the six RF+, HLA-DR 1 or DR4 patients exhibited a 50% or greater decrease in peripheral blood expression of

V,14

in comparison to syno-vial fluid. Nevertheless, peripheral blood expression ofVB14 was modestly butsignificantlylower than that of synovial fluid in this group of six patients (2.5% vs. 3.4%, P = 0.031 ), as it was for the entire group ofsynovialfluid samples (Table

III).

Taking the synovial tissue and synovial fluid samples

to-gethersynovialand peripheral blood TCRVpexpression was

comparedin 21 cases, 18 of which were from seropositive pa-tients. Increasedexpression of

V,1

5, V36. 1-3, Vf2, V3 12, and VB13.1 was observed most frequently in synovial T cells,

whereas increased expression of VB4, VB5.1, VB10,

_V#l9,

V#

1 6 and

_V#3I

was seen more frequentlyin peripheral blood (Table IV). When the samples were analyzed for larger differ-encesbetween synovialand peripheral blood expression ofindi-vidual V3 gene families, biasescontinued to be noted. Thus, when twofold or greater differences were required, 9 of 18 RF+ patients exhibited increasesin peripheral blood expression of

V,34

(no cases of opposite bias),whereas8 RF+ patients mani-fested greaterexpressionof

_V#

15 in synovial T cells (one case of opposite bias).

Acomparisonofsynovialand peripheral blood T cell utili-zation of

Vg

families wasperformedin eight seropositive RA patientswhoexpressed

HLA-DRl

orHLA-DR4(Table IV).

0)

V-(D

1mr

*

*;

T

o

_aO

M b W n * n N -

_o0

M r n n _ n₎ N

CN______V- V- V- r- I . T- V _{I I}

m

CI)

L

*

I

* 0)

N C)

F0

i

h*

*_

It

O o 0 * n N*N _ 0o 0 r1,_{n n _'} * _N

c _

N _- _ - - -

-0i - - - I

_ Z~~~~~~DtO

V-~

~ ~ '-1

At!WoDj

eueo

JA

co

.4

0

0

CD

0 .0

o

0

6

CD

O

C;

0

Co

0

6

to

CD

0.

6

0

C14

O

0

0 I0

6

l

.0

Li

.A

-H

Cu

> 03

0 0 0

Cu

0Cu CJ

~ 0

*E

_ _Cu >

_ Q

o o

* 0

e o

i

61

fi

4

Synovial T cell expression of

Vfl1

5 was at least 50% greater than peripheral blood in five of these patients, whereas in-creasedperipheralbloodexpression ofV: 1,Vf34,

Vf35.

1,V 10, and Vj 16 was seen infive, six,eight,six,andfourcases, respec-tively.

In total, 19comparisons of synovial and peripheral blood T cells were done using triplicate amplifications ofcDNAwith eachprimer pair.Statisticalanalysiscouldthereforebedirectly appliedtothesecomparisons. Examinationof the data in this way hastheadvantageof detecting biased V: gene expression that may bestatistically significantbut notof sufficient

magni-tude to meet the 50% cutoffused in the analyses presented

above. Forexample, in thecomparisonshown inFig. 2, syno-vial tissue T cell and peripheral blood T cell expression ofVfB2, Vfl5.1, V,36.1-3, V,132, Vf313.1, and V 16 were significantly different(P <0.05). Thedifference in

_V#2

expression, how-ever, was< 50%.Similarly,in both oftheexperiments shown inFig. 3 there were three examples of significant bias in which themagnitude ofthedifferencewas< 50%. TableVpresents the numberofcaseswhereexpression ofindividualV3families wassignificantly increased in one compartment or the other.

Vf2, V36.1-3,

and

_V#

1 5 were the three families most fre-quentlyincreased in synovialTcells, whereas there was a large number ofcases ofincreased peripheral blood expression of

V 1,

V34, Vf5.

1,

V/,

IO and

V/

l 9. Similar trends were seen in

the group of 17 RF+ cases (data not shown), and theeight

comparisons from RF+, HLA-DRI/4+ cases (Table V).

Table V. StatisticallySignificantlyBiased

_Vfl

GeneFamily

Expression in synoviumvs.PeripheralBloodofRAPatients

All comparisons All RF+HLA-DR-I/4+

(n = 19) comparisons (n=8)

VBgene

family Blood Synovium Blood Synovium

VB1 7* 2 5 0

V/2

3 7 1 2

V33

5 3 3 2

V34 12 1 5 0

VB5.1 11 1 6 0

V/i5.2-3

4 3 1 1

V06.1-3

0 10 0 4

VB7 2 3 1 2

V#8 1 3 0 2

VB9 1 2 0 1

VB10 7 1 4 0

V/Bll 5 2 3 0

V#12

5 4 2 2

VB13.1 3 3 1 2

V#313.2

4 3 1 1

V#14

1 4 0 2

VB15 1 7 0 3

Vf16 5 0 2 0

V#17 6 1 3 1

V#18

2 1 0 0

VB19 9 1 3 1

VB20 3 1 0 1

*_{Numberofcomparisonsdone withtriplicatePCRamplificationsin}

whichexpressionwassignificantlygreater in the indicated

compart-ment(P<0.05).

Ofnote, V/6.1-3 wassignificantly overexpressed in 10 of 19synovia (Table V), although this overexpression was > 50% in only 6 of 21 cases (Table IV). Similarly, there were only 2 of

13 cases where V/14 expression was 50% less in peripheral

bloodthan synovial fluid (Table IV), but of the 12 synovial fluidcomparisons done in triplicate, 4 exhibited a statistically significantdecrease inperipheralblood V/i14expression(data

notshown).

Discussion

The TCRrepertoire of peripheral T cells is determined in part

by thegeneticelementswhichencode the TCR, byself-MHC

antigen-directedpositive and negative selection during thymic education, and presumably by theantigens or superantigens theindividualhas encountered. As RA is a chronic inflamma-torydisease withagenetic susceptibility linked to MHC class II genes, any orallofthesedeterminants could produce a bias in theperipheral bloodorsynovialTCR repertoire. The present studyprovides the strongest evidenceto date thatbiased V/ geneutilization canbe observed in unselected T cell popula-tions from RApatients.

There wassubstantial subject-to-subject variability in pe-ripheral bloodV/ geneexpression both in normal donors and

RA patients. Nevertheless, the mean expression of each V/I genefamilywasnotmarkedlydifferentin RA patients, normal donors,or asmall groupof subjects withotherarthritides.The meanexpression of

V/i1

3.2 wasslightly lower in RA patients, however. Smalldifferencesin the expression of otherV/igene familieswerealso observed in both the subsetofpatients with RAfor2 yr orless and in the subset of patients taking NSAIDs

alone. However, alongitudinal study ofa larger numberof

patients would be needed to determine whether TCR bias in peripheralblood T cellsisblunted by passageof timeor

treat-mentwithDMARDs.

Itispossiblethat the subtlebiasin theVgeneexpression of peripheralTcellsinRAresultsfromencounterswith

superan-tigens,

chronic stimulation byantigen, or from otherfactors knownto influenceVgeneexpression, such asdifferencesin theMHCantigensexpressed by the normal donors and the RA patients (11).Toaddress thepossibilitythatexpression of spe-cific HLA molecules may influence V/ gene utilization we

comparedperipheralbloodV/ expression byRApatientswho expressed the disease susceptibility conferring MHC class II

molecules HLA-DR1, or HLA-DR4, to thatof normal donors

expressing either ofthese markers.Thisanalysisrevealed

sev-eralsignificant differences between these groups.

V/Il

3.2 and

V/8

expressionwas lowerin the patients than in the controls, whereas

V/2

and

V/i5.

1 expression washigher. The observa-tionthatagreaterdegree ofbiaswasevidentinacomparisonin which the two groups shared some MHC alleles suggests that

eventsrelated to thediseaseprocess

itself,

such as

antigen

or

superantigen stimulation,altertheexpression ofTCRV/

prod-uctsin theperipheralbloodofRApatients.

Paliardetal. ( 14) observed that the expressionof V/I14 in theperipheralbloodofsixof nine RApatientswas undetect-able orsubstantially diminished incomparison to

peripheral

bloodof normal donors and arthritic controls. Inthecurrent

study,theV/ 14PCRproductwasdetectable in everyRA pa-tientand normal donorexamined, representing0.9%or more

ofthe total. The mean

expression

in RA

patients

waslower

than that in thecontrolgroup

(2.7%

vs

3.4%),

but the

encedid not achievestatistical significance(P = 0.07). In the

current groupofRA patientsknown to expressHLA-DR1 or

DR4 (as did all of those studied by Paliard et al.) the mean peripheralbloodexpression ofV 14was lower than thatofthe control donors whoexpressed these HLA antigens (TableIII), but again the difference was not statistically significant (P

=0.16).Differencesin the studypopulationsmayexplainthe somewhatdisparate results obtainedin these two studies. In addition, there were a numberof methodological differences thatmighthaveinfluencedthe results,including stimulation of thecellspriorto RNApreparation intheaforementionedstudy aswellas major differences inthe PCRtechnique, meansof normalization ofthe data, and data analysis. Inpreliminary experiments we have attempted to determine the impact of thesevariablesonthe data generatedin eachstudy,but have

been unable to document that these methodological

differ-ences explain the disparate results. Thus, for example, we

found that inclusion of an additionalpairofoligonucleotides (such as Caprimers)toprovideaninternalamplification con-troldid notenhancethereproducibilityof the

_V#

amplifica-tions.Inaddition,although wefoundthatexpression ofa num-berof theV3 familieswassignificantlyalteredbystimulation ofthe cellswith anti-CD3 antibodiesand IL-2beforeanalysis, this didnotdecreaseexpression of Vf14bybloodTcells.Since mostofthe datawasgeneratedfrom triplicatePCR amplifica-tionsand thestandarderrorof the mean wasquitesmall, the current datapersuasivelyargue thatV 14isnotsubstantially decreasedintheblood ofpatients withRA.One other study of 3 RApatientsand 2 normal donors came to asimilar conclu-sion(27).

A numberofprevious studies haveanalyzedTCRVgene

expression in RA usinga variety of techniques, butdid not produce a clear picture of thedegreeof restricted Vgene ex-pressionin RAsynoviumorperipheralblood(Table VI). Stud-iesexaminingRAperipheralblood andsynovium for biasedV geneexpressionhave also notproducedconsistent results,

per-haps because manyoftheseexaminedonly small numbersof

patients. Two studies describedgreatersynovial fluid

expres-sion ofV 14 (14,29).Inthe currentstudyVf314expressionin synovial fluid was significantly greater than in the simulta-neously obtained peripheral blood samples. This bias is consis-tent with the results of Paliard et al.( 14). However, the magni-tude of the difference was much less marked in the current study than in theprior study and was not observed in synovial tissue. It is noteworthy that

V#314

expression has been observed to be enriched in CD8+ T cells (43-45) which are usually increased in RAsynovialfluid(46-48). Thus,it ispossiblethat differences in

Vf3

14expression are at least partially related to alterations in the proportion of CD8+ cells found in peripheral blood andsynovialfluid.

Immune responses to local introduction of anantigen re-sult in infiltrationby antigen-specificT cells andby large num-bers ofnonspecific T cells. Thus, one can assume that T cells responding to pathogenic peptides in rheumatoid synovium represent a smallminority of infiltratingTcells. Thelarge pro-portion ofnonspecificTcells mayproduce variabilityin theV: expressionprofileof T cellpopulationswith time as was noted inthesamples obtained fromthe same

patient

at two

points

in time. These considerations may alsopartly explainthe differ-encebetween the presentstudyand those of others. Examina-tion of alargenumber ofpatientswithasemiquantitative ap-proach comparing synovial and peripheral blood T cell Vf expression in this study allowed the identification of biased TCRexpression despitethe presenceofnonspecificTcells ex-pressingother TCRVf3 geneproducts. Manyoftheprevious reports studiedan insufficientnumberofpatients for consis-tentbiasesto emerge, ordid notperform comparisonsto pe-ripheralbloodrendering conclusions regarding TCR bias some-whatspeculative. Moreover, the current study was designed to providean accuratedetermination ofV:expressionby T cells as they are foundin situ within the synovium or peripheral bloodofRApatients. Therefore, stimulation ofthe cellsprior to

analysis,

aswasdone in otherstudies, wasavoided. More-over,sincethereisnoconclusiveevidencethat one or another ofthe

specific

Tcellsubsets present in thesynovium is responsi-blefor diseasepathogenesis,noeffortwasmade toisolate

indi-Table VI.SummaryofStudiesofVGene Bias in RA

SynovialTcellexpression

Reference n Sourceof material Restricted Biased*

24 3 SFandPB

Vf7

in 2

Va: no

14 9 SFandPB

_Vf3:

no

Vf14

in 9

29 5 SFandPB _Vfl:no

V#14

in 3,V#18in 2

58 5 IL-2R+STTcells VB3 in 3,

Vf39

in 3,

VfB14

in 4,

Vf17

in 4

30 2 SF Va: no Va14,Va15,VaFR1

25 9 tissue Vie: no

Va:heterogeneousnumber detected

26 10 tissue

V13:

heterogeneousnumber detected

Va:heterogeneous number detected

27 3 SF and PB

Vj3:

no

Vj2

in 2

28 8 SF and PB Va: no

VaIO

in 3,

Val5

in 3,

Val8

in 3

Vfl:no

Vfl4

in 3,

Vf35

in 3,VB13in 3

n,numberof patients studied. Abbreviationsused in this table: IL-2R, IL-2 receptor; PB, peripheral blood; SF, synovial fluid;ST,synovial tissue.

vidual subpopulations before analysis. Finally, the PCR ap-proach weused toanalyze Vf gene expressionis sensitive to changes in the activation status of T cellsaswellastheir

num-ber because recently activated T cells containing increased

amountsof RNA result inanincreasedPCR productfromthe

appropriateVf/Cf primerpair. Theseconsiderationssupport the conclusion that the biases reported in the current study which have not been reported in previous studies, and are rather small in magnitude, mayreflect disease related

differ-ences inTCR

_V#

geneexpressioninbloodversussynovium.

In the current report, the possibility of skewingofV: ex-pression in rheumatoid synovium compared to peripheral bloodwasexaminedin 21 samplesfrom 19patients. Whereas wefoundskewedexpression ofeach VAfamily in at least two comparisons, the current experimental results provide evi-dence forconsistently increased expression of Vfl6andV 15 in rheumatoid synovium.Expression of

Vf36.

1-3inthe 8synovial tissue samples, in the 13 synovial fluid samples, and in the

group including all 21 synovial samples was significantly

higherthan inthecorrespondingperipheralbloodsamples (Ta-bleIII).Furthermore,in10of 19 comparisonsdonewith tripli-cate amplifications synovial

V#6.1-3

expression was signifi-cantlyhigherthanin peripheralblood (TableV), although the magnitude of increased expression was sometimes less than 50%. The

expression

ofVf15 in thesynovial tissue samples,in the groupof21synovial samples, and in the groupof synovial samples from HLA-DR1/DR4+ patients was significantly greater than that in the simultaneously obtained peripheral blood samples (Table III). The bias toward synovial T cell expression of Vf 15wasalsoapparent in

synovial

fluid samples inwhich6of13 casesexhibitedatleasta50% greater expres-sion of V315 (Table IV). Thus, increased

Vf6.1-3

and in-creasedV315expression inrheumatoidsynoviumwere estab-lished byanumberof

analytic

methods.

Evidence of biasedV:expression in rheumatoid synovium wasalsorevealed bydecreased expressionofcertain

VI3

fami-lies.

Vfl1,

VB4, VB5.1,

VflI0,

VB16, andVf l9 expression by synovialTcellswas

significantly

lowerthan in the correspond-ingbloodsamples.Ithasbeenreported that

V35.

1

expression

wasskewedtoward the CD4+ subset(43, 45, 49).Thissubset bias maycontribute totherelativedecrease in VB5.1 expres-sioninsynovial

fluid,

butis probablylessofafactorin

synovial

tissuewhere the relative

proportion

ofCD4 and CD8

single

positive

cellsis usually the sameas in peripheral blood (37, 50). Evidence of similar bias in the

expression

of

Vfl, Vj4,

Vf

I0,

V#136,

or

_VO3l9

by CD4+ vs. CD8+ cellshas notbeen reported and therefore alterations in

expression

ofthese are

likelyunrelatedtobiases ofsubsetdistribution in various

com-partments.

The analyses we performed did not suggest that the

pres-enceof RF,theexpression ofHLA-DR 1 orDR4, theduration ofdisease,orthemedicationsusedhad asubstantialimpacton thelikelihoodofdetecting biased

Vf3

expression in rheumatoid

synovial

T cells. Thus, the subset of synovial samples from seropositive patients, from patients known to express

HLA-DR1 or DR4, from patients with RA for 2 yr or less, from

patients with RA for more than 2 yr, from those medicated with NSAIDs alone, and from those also being treated with DMARDs allexhibitedsimilarly biasedTCR VBexpression in

comparison

to the corresponding peripheral blood samples.

For example, all of the biases presentin the

HLA-DRl

/4+

synovia

were also present in the

analysis

ofall 21

synovial

samples (Tables III and IV). However, greater numbers ofsero-negative RA patients, and longitudinal study of several pa-tients, would be requiredtodetermine definitively the impact of RF, disease duration, ormedicationson V:expression by synovial T cells.

Thehigh level of expression of the V36 family in the syno-vial and peripheral blood T cells of RA patients, and in the peripheralblood of the control subjects is likely to be related to

the size of the V,36 family which contains at least nine

members. Howeverit isevident from the data in Tables II and

IIIthat levels of expression are notstrictly dependent on the numberof genes in a genefamily, and, therefore, arelikelyto

be influenced by thymic selection and antigen stimulation.

Withinrheumatoidsynovium, skewedexpression ofV:gene

families among Tcells might result from altered migration into and retention in thesynovium, or from selective proliferation oractivation (orinactivation) of cells utilizing these V gene families.

Superantigenstimulation, characterized by Vf specific ac-tivation of T cells that is often followed by depletion or inacti-vation of cells bearing the appropriateVJ3 (51-54), could ac-countforselectiveenrichment or depletion ofselected V: fami-lies in RA. Although several superantigens interact with one or moreof the relevant

Vfl

genefamilies (55), no known superan-tigen couldexplain allofthe observedbiasesin theperipheral

blood of RA patients who express HLA-DR1 or DR4.

Simi-larly, the current data do notimplicate a knownsuperantigen

as being responsible for simultaneous expansionof V36 and

Vj 15 bearing T cells in rheumatoid synovium. V36 is stimu-lated by staphylococcalenterotoxin (SE) A (R.Jenkins and D.

Karp, unpublished results) and SEE (55), whereas V315 is

stimulated by SEB, SEC2 (56), and streptococcal pyrogenic

exotoxin (SPE)A(57).However, noneofthesesuperantigens interactswithboth V,36and Vf 1 5 bearing T cells.Finally,no

known superantigen could explain the observed depletionof

severalV:families in rheumatoidsynovium. Nevertheless, it remains possible that the actions of as yet unidentified

superan-tigens account for the biases observed in RA synovium and

peripheral blood.

The present studyof blood samplesfrom 24 RA patients, 4 control patients, and 15 normal donors, and of synovial tissue from 8 RA patients and 13 RA synovial fluid samples, repre-sentsthe mostextensiveexamination ofVf gene utilization in RA to date. The results clearly indicate that the number ofV3 familiesexpressed byRAperipheralblood andsynovialTcells is notsignificantlyrestricted. More importantly, several lines of evidence indicate biasedVp geneutilization indifferent periph-eral compartments of rheumatoidpatients can be observedin unselectedTcellpopulations. Theseanalyses revealed several statistically significanttrends,with either enrichmentor deple-tion ofparticular Vp genes noted in thesynovium. Increased expression ofVB6 and

_V#315

wasevidentin amajority of pa-tients.Inaddition,Vf4, V35.1,V310,Vfl16,and

Vfl1

9

expres-sion wasdecreased in rheumatoid synovium in amajority of

thecomparisons. Becausethe PCR assay is sensitivetochanges in activation status, alterations ineither the number ofcells expressing these Vpfamilies or the activationstateof these cells could explainthedifferences observed in the

Vp-specific

PCR productsfrom RA T cells. Therefore, the dataare consistent with the postulate that antigen or superantigen driven pro-cesses maycause

Vfl-specific

enrichmentor

depletion

ofTcells in RA.