Characterization of the primary structure of T cell receptor beta chains in cells infiltrating the salivary gland in the sicca syndrome of HIV 1 infection Evidence of antigen driven clonal selection suggested by restricted combinations of V beta J beta g

Characterization of the primary structure of T

cell receptor beta chains in cells infiltrating the

salivary gland in the sicca syndrome of HIV-1

infection. Evidence of antigen-driven clonal

selection suggested by restricted combinations

of V beta J beta gene segment usage and

shared somatically encoded amino acid

residues.

E Dwyer, … , S Itescu, R Winchester

J Clin Invest. 1993;92(1):495-502. https://doi.org/10.1172/JCI116593.

Infection with HIV-1 occasionally results in a sicca syndrome, termed the diffuse infiltrative lymphocytosis syndrome, characterized by infiltration of the salivary glands with a

predominance of CD8 T cells. This response is strongly associated with certain MHC class I and class II alleles. To define the salivary gland T cell receptor (TCR) repertoire, the primary structure of the TCR beta-chains was determined using in situ cDNA synthesis followed by the "anchored" polymerase chain reaction. The sequences of 59 beta-chains from five individuals with diffuse infiltrative lymphocytosis syndrome shared structural features

suggesting antigenic clonal selection. Certain combinations of V beta J beta gene segments were selectively overrepresented in the repertoire sample, demonstrating a common

restricted usage of certain V beta and J beta gene segments. The beta-chains derived from these overrepresented V beta J beta combinations revealed a preference for specific amino acids at position 97 in the third complementarity-determining region, a residue postulated to contact peptide antigen. Moreover, the nucleotides encoding this position were not germline in origin. TCR beta-chains in nonoverrepresented V beta J beta combinations did not

exhibit preferential usage of selected somatically encoded residues. The pattern of TCR Research Article

Find the latest version:

http://jci.me/116593/pdf

Characterization of the Primary Structure of T Cell Receptor

_#

Chains in Cells

Infiltrating the Salivary Gland in the Sicca

Syndrome

of HIV-1 Infection

Evidence ofAntigen-driven Clonal Selection Suggested by Restricted Combinations of

Vft

Jft

Gene

_Segment

_Usage

and Shared

_Somatically

Encoded Amino Acid Residues Edward Dwyer,Silviu Itescu, and Robert Winchester

DepartmentofPediatrics, Columbia University, College ofPhysicians and Surgeons, New York 10032

Abstract

Infection with HIV-1 occasionally results in a sicca syndrome, termed thediffuseinfiltrative lymphocytosis syndrome, char-acterizedbyinfiltration of the salivary glands witha predomi-nanceofCD8Tcells.Thisresponseis strongly associated with certainMHC classIandclassII alleles.To definethesalivary gland T cell receptor(TCR) repertoire,theprimarystructure of theTCR

#-chains

wasdeterminedusinginsitu cDNA synthe-sis followedby the"anchored"polymerasechain reaction.The sequencesof 59

#-chains

fromfive individuals with diffuse in-filtrative lymphocytosis syndrome shared structural features suggesting antigenic clonal selection. Certain combinations of V13

_JO3

genesegments were selectively overrepresented in the repertoiresample,demonstratinga commonrestricted usage of certain

Vft

andJ(3gene segments.The,8-chains derived from theseoverrepresented

_Vft

J(3combinationsrevealed a prefer-enceforspecificamino acidsatposition 97 inthethird comple-mentarity-determining region,a residuepostulated to contact peptide antigen. Moreover,thenucleotidesencodingthis posi-tionwere notgermlineinorigin. TCR(-chainsin nonoverrepre-sentedV,@ J(Bcombinations didnotexhibit preferential usage of selectedsomaticallyencodedresidues.The pattern of TCR (3-chainsexpressed in the salivary gland of a control person with primarySjogren'ssyndrome was considerably more heteroge-neousanddifferentfrom that found indiffuseinfiltrative lym-phocytosissyndrome. (J.Clin.Invest.1993.92:495-502.) Key words: T cellrepertoire*Sjogren'ssyndrome-diffuse

infiltra-tivelymphocytosissyndrome-immunoglobulin* autoimmunity

Introduction

Anunusualform ofsicca syndrome termedthe diffuse infiltra-tive lymphocytosissyndrome

(DILS),'

is a distinctive but

in-frequenthostresponse to infection with HIV. DILS is charac-terized byinfiltrationofpredominantly CD8 T cells into lacri-mal and salivary glands and is frequently accompanied by a

Address correspondence to Edward Dwyer, M.D., Department of

Pedi-atrics,Columbia University, College of Physicians and Surgeons, 630 W. 168th Street, New York, NY 10032.

Received forpublication 12 August 1992 and in revised form 12 January 1993.

1.Abbreviationsused in this paper: CDR3, third

complementarity-de-terminingregion; DILS, diffuse infiltrative lymphocytes; OR, odds ra-tio; TCR,Tcell receptor.

significant peripheral CD8 T cell lymphocytosisas well as a maintenence ofrelatively normal levels ofcirculating CD4 cells (1). The majority ofperipheral CD8 and CD4 T cells exhibit a mature memory phenotype characterized by the abundantexpressionof(31 and (2integrins,without a detect-able increase in cells bearingmarkers ofrecentactivation such asHLA-DR moleculesorIL-2 receptors(2).HIV- 1 antigens areidentifiableonlyin monocytelineagecellsatsites of tissue infiltration and the virus is culturable from bloodmonocytes (3). Susceptibilitytodeveloping DILS is stronglyassociated withthe DRB1*1102 and DRB1*1301 alleles of HLA-DR5 and DR6, respectively, while HLA-B35 is significantly de-creased infrequency (4).These featurescontrastwith themore

common idiopathic autoimmune disorder of primary Sjogren's syndromewhich is characterizedbyinfiltration ofthe

salivaryglands withapredominanceofCD4 T cells and

suscep-tibilityis associated with HLA-DR2orDR3 (5).

TheabilityofTcellstorespondtoforeign antigenssuchas

HIV- I isdetermined by the clonallydistributed a(3 T cell

re-ceptor(TCR)which isamembrane-bound 88-kD heterodimer thatrecognizes foreign peptides presented byself MHC mole-cules(6).Theunique recognitionstructureofeach clonal TCR isdetermined by the combinatorialpairingof distinct V,D, andJ gene segmentswhichtogetherconstitute each individual molecule. Inthecaseof the TCR

#

chain,thepotential reper-toire iscomposedof 24V: families,2 D(3 segments, and 13

Jo3

segments (7). Although several studies of the (3-chain TCR repertoire in humans have demonstratedanonrandom distri-bution ofVf3 familyexpression in both normalsand various diseasestates(8-14),arecentinvestigation specifically focus-ingontheprimarystructureoftheentire(-chain variable

re-gion revealed a nonbiased pairing of individual V(3 families withspecific

J(3

segments(15).Inadditiontotheuseof these multiple germline elements, structuraldiversity ofthe(3chain

isgreatlyenhancedby exonucleic"nibbling" andNregion di-versification during junctional pairing of individualV, D,and J segments which results in random deletion or addition of individual nucleotides in generating nongermline encoded amino acids (6); thisregionof the molecule whichexclusively emphasizessomaticdiversityis referredto asthethird comple-mentarity-determining region(CDR3), and it is this regionof

the TCR(-chainwhichis postulatedto directly interact with foreign peptides presented by self MHC molecules ( 16, 17).In supportof thismodelofmolecularinteractionare several stud-ies which havedemonstrated invitrothatmurine T cell clones responsivetoacytochrome c-derived peptide express TCR(3 chains that show not only restricted usage ofspecific V(3J(3 combinations, but perhaps more importantly, also exhibita preferentialexpression of particularnongermlinesomatically encoded amino acids in the CDR3 region, most notably at amino acidposition 97 (18-20) which is five residuescarboxyl J.Clin. Invest.

©TheAmerican Society for Clinical Investigation, Inc.

to theinvariantcysteine at position 92. Further emphasizing the importance of determining complete primary structure data inobtaining evidenceof antigen-specific immune respon-siveness is therecentfindingthat single nonconservative sub-stitutionsatspecificamino acid positions in the cytochrome c peptideare capable oftransferring responsivenessto alterna-tive clones which express TCR 13 chains featuring different

VflJfl

combinations as well as different non-conservative amino acid substitutions at position 97 of the CDR3 re-gion( 18).

This study analyzes the structure of the TCR

f1-chain

to gain insight intothebiologicbasis ofthe infiltration into tissues by T cellsin the DILS response to HIV-1 infection. Several alternative possibilities exist for mechanisms underlying the accumulation of lymphocytes, each of which have different predictions interms of the character of the T cell repertoire as determinedby thestructure ofthe TCR13chains. In view ofthe increased frequencyof B cell lymphoma in DILS (4), the lym-phocytic accumulation in tissues could represent a parallel preneoplastic expansion ofone or a few T cell clones which would berecognizedbymarked TCR13chain homogeneity at thelevelof nucleotidesequence. In contrast, a process driven by nominal antigen should exhibitsome evidence of selection atthelevelofprimary proteinstructure bydemonstrating pref-erentialusageofV13andJ13elements, and especially, character-istic features ofthe CDR3region.In the instance of superanti-gendrive, which has beensuggested by tworeportsofHIV+

patients (12, 21), onewould expect preferential deletion or

enhancement ofcertain V13 families while allowingrandom structural features in the CDR3 region and the lack ofany

preferentialusageofspecific

_Jo

elements. Since CD4 lympho-cytesandmonocyteswhichfunctionasreservoirs ofHIV-1are systemically disseminated, thisantigen orsuperantigen drive would be expected in both the involved tissue and elsewhere in thebody. Finally,theprocessoftissue accumulation could be unrelated to any antigen drive or neoplastic event, perhaps

reflecting secondarytissuerecruitment ofbloodlymphocytes

that have undergoneactivation as aresult oflocallysecreted

cytokines,and inthiscase,no

commonality

of

primary

struc-turewould be evident inacomparisonofTCR13 chains from differentindividuals withDILS.

Currenttechniques assessingTCR structural features use eithera limitedset ofmonoclonal antibodies to measure V13

frequencies,orusepotentiallymoreinformative molecular

bio-logic techniqueswhichmaybedividedintotwo

general

catego-ries. In one,illustratedby construction ofaformalcDNA li-brary, therepertoireisdetermined by

sequencing

individual 1

chain clones to define CDR3

organization

and obtain a fre-quency distribution ofV13andJ13usageby

enumerating

their occurrence. Intheother, primers

specific

for each oftheV1 segmentsareusedin

quantitative

PCR methods which is

rela-tivelysimpleandhas theimportant advantagethat all known

V1

genesegmentscanbe measured(22). However, indetermi-natevariabilities intheefficiencyof

particular primers

intro-duces uncertainties into the

quantitation

process and the methodispresently incapableof

determining

V13J13 combina-torial frequencies or

evaluating

CDR3 structure. This report usesanovelapproach which entailsaninitial stage of in situ cDNAsynthesis performedon afrozen

_{section, thereby}

elimi-natinganybias introducedbyinvitro

manipulation

or differ-ential

priming,

and is followed

by

amplification

of cDNA

_by

the anchoredpolymerasechain reaction

(23)

which has been

demonstratedtononartifactuallyamplify all expressedV13 seg-ments ( 15, 24); theprimarystructure is obtained through nu-cleotide sequencing. Thisapproach addresses the role of anti-genspecificitybyinfiltratingTcellsin salivarygland tissue by

concentratingonthepresence of preferentialV13J1 combina-tionsaswellastheamino acidstructureofthe third

comple-mentarity-determining region in TCR

13-chains

isolated di-rectly from tissue.

Methods

Patients. Each individual with DILS met the diagnostic criteria of at least 6 mo of salivary gland enlargement, HIV-I positivity confirmed by isolation of the virus and demonstration of specific antibodies by ELISA and Western blotting, elevated CD8 T cell levels in the periph-eralblood, and at least two foci oflymphocyticinfiltration with a

pre-dominanceof CD8lineageTcellsdemonstrated in lip biopsy speci-mens. Theindividual with primary Sjogren's syndrome had symptom-atic xerostomia, morethantwo lymphocytic foci on minor salivary glandbiopsy, and anti-Ro/ SS-A autoantibodies. All tissue analyzed in this report wasresidual diagnosticlipbiopsymaterial obtained with

informedconsent and under anapproved protocol.

Methods.The method of insitucDNAsynthesiswasaccordingto a modificationof the method of Tecott et al. (25). Briefly, all specimens weresnapfrozen with liquid nitrogen and stored until cut into

_10-'qm

sections.Eachtissue sectionwashybridized for 12-16 h at room tem-perature with 50,lofasolution containing 50% formamide, 4X SSC, anda36-bpprimer(80 ng/ml) complementary to a sequence in the constant region of the TCR d-chain (CCCATTCACCCACCCAGC-TCAGCTCCACGTGGTCGGG). After hybridization, sections were washed in2xSSC at room temperature for 10 min for three washes, with a final wash in.5xSSC at 40'C for 2 h. cDNA was synthesized by adding to each tissue sectionavolumeof 50_Alwhich contained 10 U

AMVreversetranscriptase (Promega Corp., Madison, WI; buffer

sup-plied)and 1mMdNTP's, and incubated at420Cfor 1 h. Each section was washed three times in2x SSC at room temperature for 10 min, followed byafinalwashat40'Cfor2 h. Foreach sample, at least five

sectionsfrom eachspecimenwereprocessed simultaneously according

tothis protocol. The cDNAwaselutedfrom each tissue section with 0.15 M NaOH. .25 volume ofI MTris-HCl(pH=6.0) was added to eachspecimenfollowed by ethanol precipitation. The cDNAwas

sub-sequently amplified byamodification of the anchored PCR procedure of Lohet al. (23). Specifically, the first strand cDNA productwas

G-tailed with terminaldeoxynucleotidyltransferase (Promega Corp.;

buffersupplied) using 1 mM dGTPfor 30 min at 37°C, and then

amplified usinga.1-MM34-merC-tailedprimerwhich also hasa restric-tion site forNotI (TTGACGCGGCCGCTACTAGCCCCCCCCCCC-CCCC),a.8-MuMprimer containing onlytheNotI site (TTGACGCGG-CCGCTACTAGC),anda 1-AMthird primer complementarytothe

constantregion, 5' of theoriginal36-mer (CCTGTGGCCAGGCAC-ACCAG).The PCRprotocol usingaHybaid (Hookand Tucker

In-strumants, UnitedKingdom)thermalcycler consisted ofdenaturation

at94°C for 1 min,annealingat53°Cfor 2 min, andsynthesisat72°C

for 3 min. The productwassize-selectedusingalowmeltingagarosegel

and200-400-bp fragmentswerereamplifiedfor 20cycles (94°Cfor 1

min, 55°C for 2 min, and72°C for 3 min) usingaconstantregion

primer 5' of thepreviouslyusedconstantregionprimer (GTGTGG-CCTTTTGGGTGTGG)while the otherprimerwasthe 20-mer NotI primer. Theproductwasagainsize-selectedonanagarosegel,digested

withNotI,andligated intoBluescript(Stratagene Inc., LaJolla, CA)

which had been digested withNotIandSmaI.The

_{f3-galactosidase}

sys-tem wasemployedtoselect white colonies whichwerethen screenedby

restrictiondigestionwith SstI andBgl II,whichselectivelycut aBglII

restriction site present in the,B-chain constantregion just 5' of the last-used primer. All clones with inserts revealedby BglII and SstI digestionweresequenced usinga_{DNA sequencer}(373A; Applied

Bio-systems Inc., FosterCity, CA) usingthe Taq dye deoxy terminator

sequencing kit (AppliedBiosystemsInc.) inaccordance with the manu-facturer's protocol.

Statistics. Statistical analysis used Chi-square analysis or Fisher's exact test for whichanodds ratioisgiven for each calculation. In each instance the null hypothesis for V(3 or J(3 frequency calculations was that therewas anequal likelihood ofthe presence ofeach gene element. Nomenclature. TheV(3family designation is according to Wilson et al.(26) and Robinson (27); the(-chainamino acid position number is according to Kabatet al. (28). The length of the CDR3 regions is arbitrarily definedasthe numberof amino acid residues between the invariantcysteine atposition 92 and the invariant phenylalanine at position 108 from which is subtracted the length of the shortest CDR3 region togive the absolute number of varying residues.

Results

Thenucleotide sequence andinferred primary aminoacid se-quence of 62 TCR 3-chainswereobtained fromlipbiopsy

speci-mensof five patients with HIV-associatedDILS. Their length ranged from 100 to 250bp andreflected extension from the constant region primerto well intothe V(3segment. Each (3 chain had adifferent G-tailed terminus, indicatingthat they arosefrom separateprimingevents. The number ofA-chains from eachindividual rangefrom 10 to 17. 59ofthe Vf3 seg-mentssequenced could beunambiguously assignedto oneof the 24 knownV: families and, with few exceptions, have se-quenceidentity with known V: alleles; theremaining 3 V(3 sequences were tooshorttounequivocally assigntoaspecific

V:family, but were includedin theanalysis ofJ(3 usage and CDR3structure. ThepatternsofV(3andJ(3segment usage as well as the organization ofthestructureoftheCDR3 region wasdeterminedbyenumeration ofthe (3chainsequences. For referencepurposes and to controlforthepossibility of intrinsic salivary gland specificity,55TCR (-chains derived from the lip biopsy ofanindividual withprimary Sjcgren's syndrome are also included in theanalysis.

V(3Frequency distribution. Fig. 1 shows the frequency of

V( segment usage inthe TCR of T cells infiltrating salivary tissue ofeach ofthefive samples from DILS as determined from the sequence obtained for each isolate. 15 of the 24 known V(3 familiesare collectively expressed in this sample. Althoughrelativelyheterogeneous, thefrequency distribution of V(3segments for thegroupshowsa distributionbiased to-wardsthe useof certainV(3familiesandisnon-Gaussian(x2

= 63.65,P=0.00001). Six V(3 families, V(32, 5,6, 8, 12,and 13,areexpressedat alevelgreaterthan one standarddeviation above themedian(median= 2.0, SD= _{2.6) and collectively} constitute 64% of the total V(3 families expressed. With the

exception ofpatient3 whereV(35 was used in four of the 10

(3-chainsisolatedfromthatindividual, there isnot adominant usageofonespecific V(3 family bytheentiregroup orbyanyof the other fourpatients.8ofthe 24known V(3 familieswere not

encountered in any ofthe (3 chain sequences from the five

samples.

Fig. 2 compares the frequency ofV(3 usage in the 59 (3 chainsof the DILS group with 53 TCR(-chainsobtained from the same anatomic site from one individual with primary Sjogren's syndrome. That20ofthe 24 V(3 familiesare repre-sented intheSjogren'ssyndrome sample indicates the V(3 reper-toirediversityin thissinigleperson exceedsthat found in the entire DILS group. Further emphasizing differences in V( usageof thesetwodiseasesinvolvingaccumulationof

lympho-cytesinthesameorganis the observation thatthereare seven

R _PATIENT5

E 5

a U 4

E

N 3

C

Y

2

2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

VP3

FAMILIES

Figure 1. Frequency of TCRV(3families inDILS. 15V(3families are expressed in 59 TCR(-chainsexpressed in the salivary glands of five individuals with DILS (median=2.0, SD=2.6). The distribution is nonrandom(x2=63.65, P=0.00001)and exhibits sixV(3families (V(32,5, 6, 8, 12, and 13) which are expressed at a level greater than

onestandard deviation above the median which collectively account for 64% of the totalV(3distribution.

V(3 families, V(34, 9, 10, 11, 15, 16, and 23, expressed inthe Sjogren's syndrome sample thatarenotfoundinthesampleof DILStissue.TheTCRrepertoire observedin the

Sjogren's

syn-drome sample contains only three V(3 families, V(35, 6, 13,

expressedatalevelgreaterthanonestandard deviationabove the median (median = _2.0, SD = 2.4), andthese constitute 45%ofthetotalV( segmentexpression.

J( Frequency distribution. The frequency of J(3 segment expression for persons with DILS is shown in Fig. 3. All J(3

18

16

14

UDILS

E 12 _ElSs

R 10

C 8

E

N 6

T

4

2

0

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

VP FAMILIES

Figure 2. Comparative distribution of TCR

V(3

families in the salivary glands ofDILS(n=59) andSjbgren'ssyndrome (n=53).One in-dividual with Sjogren's syndrome expresses 20V(3families which represents greater diversity than the collective

V(3

repertoire of five individuals with DILS; this includes seven

V(3

families

(V(34,

9, 10,

segmentsanalyzedareidenticaltopreviouslyreportedJ13 seg-ment sequences. Ofthe 13 known

_J#

segments, 10 are repre-sented inthe 1-chain repertoire ofthe T cellsinthe salivary glandsin DILS.The overalldistributionis moderately heteroge-neous(median=4.0, SD=4.5)in that there is not a singular dominance ofoneJ13segment;however,thereisapreferential skewing of

_Jfl

segment usage,with four ofthe total of 13 J13 segments,J31.l, 2.1, 2.5, and 2.7, foundtobe present at a level greaterthanone standard deviation above the median and col-lectivelyaccountfor 70% ofthe expressed

_J#

segments in glan-dulartissue

(X2

= 48.2, P = 0.000003). Three J13 segments,

J#1.3,

J132.4, and

J#2.6,

are notrepresented in anyofthe 62 TCR 3-chainsanalyzed.

Fig. 4 shows thefrequency distributionofJ13segment ex-pressionfrom the individual with Sjogren'ssyndrome as com-paredtothe that amongthefivepersonswith DILS.There is considerable overlap between thetwo groupswithregard to the preferentialuseof certainJ13segments, e.g., J 1. 1,J132.1,and

J132.7;however, selective differences areevidentasshown by thesignificantly increased frequencyof

J#2.5

usagein DILS (P = _{0.017, odds ratio}= 9.72) andatrend of enhancedusageof

J12.3

inSjogren's syndrome which approaches statistical signif-icance (P=0.086).

D13 Frequency distribution. Analysis of D13 segment usage

revealsthat there isnot asignificant difference in theusageof either ofthetwoD13segments,with

D13l

1. found in 321-chains

and D12.1 in 30 1-chains. One 13-chain incorporated both

D13 1.1 andD12.1 and 5 13-chain transcripts didnotexhibitan identifiableD13segment. Inaddition,within each D13segment group,therewasnot ademonstrablepreferencefor theuseof

oneparticulartranslationalreadingframe for either D131.1 or

D12.1 (datanotshown).

Inferred length of the CDR3 region. Fig. 5 illustratesthe length of the CDR3regionsfor the1-chainsof DILSaswellas

primary

Sjogren's

syndrome,arbitrarilydefinedasthe number

F

R 11

E Q E

E e

N

C

Y 4

n PATIENT1 * PATIENT 2 5 PATIENT 3

ES PATIENT4 5 PATIENT 5

1 5 16 21 22 2.3 24 2.5 26 2.7

Jo SEGMENTS

Figure 3. Frequency of TCR

J(3

segments in DILS. 10 of the 13J#

segmentsareexpressedin62 TCR(-chainsin thesalivaryglandsof five individuals with DILS(median=4.0,SD=4.5)Thedistribution

is nonrandom

(X2

=48.2,P=0.000003)with fourJ/3segments

(Jj

l.l, 2.1,2.5,and2.7)expressedatafrequencygreater thanone

standard deviation above the median whichcollectivelyaccountfor 70%of thetotal distribution.

P

E

15-R

10

E

N T

5-0I

1.1 1.2 1.3 1.4 1.5 1.6 2.1 2.2 2.3 2.4 2.5 2.6 2.7

J13 SEGMENTS

Figure 4.Comparative distribution of TCRJ13segmentexpression in the salivary glands of DILS (n = 62) and Sjogren's syndrome (n

=55). Thereis considerable similarity in the skewness of

_J#

segment usage inDILS and Sjogren's syndrome, with the only significant

dif-ference beingthepreferential usage of_J#2.5by the DILS patients (P

=0.017).

ofamino acid residues between the invariantcysteine at posi-tion 92 and the invariant phenylalanineatposition 108 from which is subtractedthe lengthofthe shortest CDR3 region, 9 residues, toderivethe numberof inserted residues.The total numberofamino acids in the CDR3 region ofthe DILS 1-chains is significantlylonger (mean = 12.4, SD = 1.72)and shows less variability as compared to the Sjogren's patient (mean = 11.7, SD= 1.93,P=0.04).TheCDR3 region ofthe

DILS

1-chains

showmore size restriction inthat 73% are 11,

12,or 13amino acids in length,or 2, 3, or 4 inserted residues, whereas the CDR3regions oftheindividualwithSjogren's syn-dromedisplaymorevariabilityinsize witharandom distribu-tion from 9 to 14amino acids in length,orfromzero tofive inserted residues.

Associationof V: with J1. Fig. 6 depictstheassociation that

wasfound inagiven 13 chain between thepresenceofaspecific

V13

family and theoccurrence ofaparticular

J13

segment in eachofthe59 TCR

1-chains

fromDILStissue.Inadditionto theapparently background random combinatorialpairing of

some V1 families with individual J13 segments, there is evi-dence foranon-random selectiveusageofparticular V13J13 com-binations. Forexample, the combination ofV135J132.1 is

en-countered in fourseparate1-chainsderived from three differ-entindividuals (P=0.009,OR= 13.14),

accounting

for7%of

all

1-chains

as

against

an expected

frequency

of 1.8%. The

V16J12.7

combination is represented intherepertoireof four differentpersons(P= _0.026,OR =_7.33), and the combina-tion of

V#1

3J3

1.1 is expressed in three separate 13-chains

(P

= 0.022,OR= 12.00).Foreach of the V13segments

expressed

in these threeparticular

V13J13

combinations,the associatedJ13

constitutesgreater than 50%ofthetotalJ13segmentsassociated withthatparticular V1 family; V1 1

7J12.

1 alsofulfills this crite-rion butonlyapproachesstatistical

significance

foraV13J13 com-bination (P = _0.08). The TCR 1-chains

containing

the

V1J1

combinations

V135J12.

1,

V136J132.7,

V:13

3J1

1.1, and

V13

1

7J132.

1, aredesignated "overrepresented" 13-chains.

Withtheexceptionoftwo

13-chains

whichrepresent exam-ples of

V135J12.

1combinations,the

repertoire

of

Sjogren's

30-25

p

E

20+

R

E

s-E

N

10-T

-

5-I

I

1.1

*

DILS

0

Ss

I

ii

1.2

Ji

S E G M E N T S

1.4

1.5

1.6

2.1

2.2

2.3

2.5

8 9 10 11 12 13 14 15 16 17 18 19 (0) (1 ) (2) (3) (4) (5) (6) (7) (8) (9)

AMINO ACID LENGTH

Figure 5.Amino acid length of the CDR3regioninindividual TCR

f-chains in DILS(n=₆₂₎and_Sjogren'ssyndrome (n= _55).The

totallengthof the CDR3regionis definedasthe number ofamino

acids between the invariantcysteineatposition92 and theinvariant

phenylalanineatposition 108;the numberof inserted aminoacids,

shown inparentheses, is determinedbysubtractingthe number of amino acids in the shortest CDR3region (e.g., 9residues)fromthe

total numberof amino acidsineachCDR3region.The total number of amino acids in the CDR3regionof the TCRdchains of the

pa-tients with DILS (mean= 12.4,SD=1.72)issignificantlygreaterand

shows lessvariabilitythan that ofSjogren's syndrome(mean= 11.7,

SD= 1.93,P=0.04). 73% of CDR3regionsinDILS exhibittwo to

fourinserted residues while the CDR3regionsof theSjogren's

syn-dromepatientdemonstrate randomvariation between 0 and 5 in-sertedresidues.

drome does notinclude anyexamples of expressed chains which encode the overrepresented VfJf combinationspresent

in DILS (data not shown). Rather it demonstrates entirely

different overrepresented VflJf combinations which include Vf5Jf2.7 (P= 0.042, OR = 5.7),Vf6Jf1.I (P=0.011, OR

= 10.75), and VflI5Jf2.3 (P=0.04, OR= 18.8).

Association of particular VflJf combinations with CDR3 structure.Inviewoftheobservationthat fl-chains with

particu-larVflJf combinations were found in salivary glands of

pa-tients withDILSatlevelsgreaterthan expected by chance, the

CDR3regions offl-chains composed of these overrepresented combinations werecompared tothose where no preferential

VflJf combinationswereobserved. The primarystructuresof

theCDR3regions of selected f-chainspresentin these

overrep-resentedVfJfl combinationsareshown in TableI.Specifically

with regardtoamino acid position 97, itcanbeseenthat there

isastrongrelationship betweentheuseof specificVflJfl

combi-nationsandtheidentity of the amino acidatposition97. For

example,inthecombinationVf1 3J 1.1 fl-chains (TableIA),

serine ispresent atposition 97intwoof thethreeisolates(P=

0.01,OR = 54)and, as indicated by Fig. 7which showsthe totalfrequency ofindividual amino acidsat position97,this accountsfortwoofthe four serine residuesusedatthisposition

in theentire sampleof62 f-chains. Similarly, glutamine

ac-countsfortwoofthe four residuesat position97 indchains

2.7

VP

FAMILIES

1 2 3 5 6 7 8 12 13 14 17 19 20 21 22

1 2 2 1 13

1 1

1

1

1

1

1 1 1 1 1 1 1

1 1 1 1 1 1

_ 1

1

1 1 X 4 1 _ _ 2 _ _

1 1 2 _ 2

1 14 1 12 2

Figure 6. Coexpressionof VA families withspecificJflsegmentsin individual TCR a chains in the salivary glands of DILS. There is a preferential associationof certainVflfamilies withspecificJ.: seg-ments as demonstratedby thepairings ofVfl5J,32.1 (P=0.009, OR

= 13.14), Vfl6Jfl2.7 (P = 0.026, OR = 7.33), and V,313J1l.l (P

=0.022,OR= 12.00).

with the combination

_Vf5JO2.1

(Table I B, P = 0.01, OR

= 54), and in thecaseof the Vf6Jf2.7 combination(TableI C),aspartic acid is presentat position 97 intwo of the four :-chains(P=0.01, OR= 54), againaccountingfortwoof the fouroccurrencesof this amino acidatthisposition. A further exampleof this association is found in the Vl17familywhere

twoof the three isolatesarepairedwith

J02.

1. Table I shows thatbothof theseVfl17Jf2. 1 chainsareencoded with a glu-tamic acidatposition97(P=0.006).Inaddition, emphasiz-ingtheimportance ofsimilarityinresidues foundatposition 97 is the thepresence of identical amino acidsadjacentto posi-tion 97 intwoofthe four

V#Jf3

categories,e.g., serine residues

atposition 98 in the f chain with the Vf6Jf32.7 combination andtyrosineresiduesatposition96in theV:1

_3JDO

1. 1 combina-tion. Moreover, it should be noted that with theexceptionof VlI7Jf2.1, whereboth ofthe f-chainsarederived from the

sameindividual,alltheother

_VOfJfl

pairingsarecontributed by different individuals.

The significance of the relationship between Vj3Jfl and amino acidresidue atposition 97 is further enhanced bythe observation that, in most of the ,8-chains, the amino acids found at and aroundposition 97are encodedby somatically addednucleotides,rather than theindividualVandJ segment

germline elements used elsewhere in the chain. The somati-cally added nucleotidesare depicted in lowercase letters. As canbeseenin TableIdifferentcodons encode thesameamino acidatthesepositions, suggestingthatfactorsoperatingatthe leveloftheexpressedproteinrather thanatthe levelofDNA

areinvolvedinthisnonrandomrepresentation of aminoacid residues.The53f-chainsexpressed in thesalivary tissue ofthe

oneindividual withprimary

Sjogren's

syndromedonotexhibit

asignificantassociationbetween anyVflJf combinationanda

C; ~~~~~~~~~~~~0

U

0~~~~~~~~~~~~~~~

0

r- r- -4 r-

r--w~~

~

~

a.

~~~

a.

~~~

~

~

00 _H~ HE 0 0 0

H HH H H-1

H E- cs coi coa N * * f4

oo~~~~~~~~~~~~~~~-_EHE -4;E-4; E S:oX H v Ev X HEv~~~~~~~zX X Co >4 44 9x -: m

X C) C) C; Q ~~~~~~~~~~~~~~~~~~~~~U~~E- ~ bo bo

U~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~C

.t H H H H H H H H _X

bo~~~~~~~~~~~~~~~~4

H H 0D C 0D 0 0D 0D

S O O Q.UUcy Q U Q U

0

¢) 0 0 0 0 0

.~~~~~~~~~0

U a~~~~Un O¢

ON Q b U

>~~~~

0J

0D

0

00

"'DX

00 0

U HH UOUc H

¢

Q _{H 0 0}

cn O cD cD

0u

U

00 ¢ O ¢ ¢ X U _4)

.X oy

_m0~~~~~~~~~b

U) ~~~~~U(n CU

_Ud

_U E¢aZ

_Q

n _UU U) ¢o

_O

U ¢

_a

A3

₀

0

k u C)u CU U U S

z U cU4 C 4 co c co '0

o4 CTJ > Clj co (n CD U U4cO o-4E

00 H 004 C.) HCcoEU

UU¢~ ¢co¢00 0

I-, 0~~~~~~~~~~~~~~~~~ 0 c 0 C

Soo U U U U U U E- p

EnU

. En-n 0 E < 4En 4-a E U

Q U~~OH U 0 U0 0~~'

t:a;; r .0

r--qdr)r-4 Lr) V-4 r

c - 0 0 0

Q- 00. .

k <

~1

14

12

F

R 10

E

Q 8

U

N*

C*

2

G A L V F Y S T 0 N E D K R H P C WM

AMINO ACID

Figure 7. Distribution of amino acid residues used at position 97 in

individualTCRf-chainsin DILS. *Amino acids represented in

pref-erential

V#Jf3

pairings.(A, Ala, C, Cys, D, Asp, E, Glu, F, Phe, G, Gly, H, His,I, Ile, K, Lys, L, Leu, M, Met,N, Asn,P, Pro,Q, Gln, R, Arg,S. Ser,

T,

Thr, V, Val, W, Trp, Y, Tyr).

Discussion

Thecentral finding of this study, usingdirect enumeration of theprimarystructural elements of each isolatedTCR

f3

chain, isthat asubpopulationofTcellsinfiltrating theminorsalivary glands ofHIV + individuals with DILS express a restricted fi-chain TCRrepertoire.Thisrepertoireexhibits structural

char-acteristicsconsistent with what would be predictediftheclonal selection of theseTcells occurred throughan antigen driven

mechanism.Thefindings thatsupport anantigen driven selec-tioneventinclude: theselectiveoverrepresentation of

particu-lar

VflJf3

combinations, thepresenceofidentical amino acid

residuesatposition97infichains included in thesesameVflJ,3

combinationsevenifthe chainsarederived fromdifferent indi-viduals, andthat theamino acidsatthis positionare specified by different codons whichare notspecifiedby the germline.As a counterpoise tothese nonrandom eventsis the observation that there isnoevidenceof bias in theuse of theDflelements whichaddsfurther supporttothe considerable evidence in fa-vorof selective usageofthe otherTCR gene elements. That these fl-chain structural features are specific to the immune responseofDILSissupported by thefindingthat threeofthe fouroverrepresented VflJfl combinations ofthe DILS reper-toire arenotrepresented inanyofthe:-chains ofthe person with

Sjogren's

syndrome.More generally, a recent study of 250 TCR

f-chains

from fivehealthyindividuals, includingapair of identical twins, didnotdemonstrateapreferential association betweenanyparticularVflfamily and Jfgenesegment in indi-vidual TCR f chains ( 15 ). Theseobservations argue against the presenceof

biasing

structuralfeatures,eitherofthe genome orof thematureproteinproduct, as beingresponsiblefor the selective

VflJfl

combinationspresent in DILS. Taken together withthefact that DILS is adistinctivehost response to HIV- I infection whichoccurs in the setting ofparticularMHC class I and II alleles, the present observationssuggest the possibility that theunique featuresof the response of DILS are mediated bytherecognition propertiesof these TCR

fi

chains for certain HIV-encoded peptides.

Theidentification ofasubset offl-chainscharacterized by overrepresented combinations of

VflJfl

segments allowed an

approachtotheproblem of identifying TCR'sfrom clones that might have specifically respondedto an antigen as opposed to TCR's from other clones thatreflect nonspecific recruitment mechanisms. The fact that analysis of the CDR3 regions of these twocatagories offl-chainsrevealed a highlynonrandom occurrence of particular amino acidsat position 97 supports the rationale of thisapproach. For example, each of theV3Jf chains from the overrepresented groups in DILS (e.g.,

Vf5Jfl2.

1,

V06Jf2.7,

V# I3Jl1.1, and V13I7Jl2.I )

demon-strates aselectivity for specificaminoacidsat position 97 ofthe CDR3 region, which isthe samestructural residueimplicated in direct interaction withthe antigenic peptide in several in vitro and transgenicinvestigations(18-20). The significance of this association isenhanced by theobservationthat in each

ofthe four preferredVfJf classes thecommonamino acid is encoded by nongermline nucleotides. The existenceofseveral structurally distinct overrepresented

VfJ#

categories suggests thepossible existence ofseveralantigenic peptidesresponsible for eliciting the immune response in the salivary glands of DILS.

Moststudies examining either tissue-specificor peripheral

Vf repertoireshavedemonstrated elements of nonrandomness and while the Vfl repertoire of DILS identifiessix of 24 Vfl families(Vf2,5,6,8, 12, and 13)asaccountingfor 64% of the total Vf segmentsisolated,theprincipal findinghere is that the VBrepertoire of each individual with DILS ismoresimilarto

other individuals with DILS than it isto theindividual with

Sjogren's syndrome. For example, the one individual with

Sjogren's syndromeexhibitsgreaterdiversityinexpressing 20 of 24Vflfamilies than thecollectiverepertoireoffive individ-uals with DILS whichonlyexpress 15 Vfl families.Moreover,

sevenof the Vf familiesexpressedin theSjogren's syndrome patient's repertoire are absent from any of the DILS reper-toires. Additionally,it should be noted that the datafromthe individual withSjogren'ssyndromeareingeneralagreement witha recentstudy which found apreferential usageof

V,32

and

_Vol

3 from lip biopsiesin primary Sjogren's syndrome

(29).Inthepresentstudy Vf 13 is themostfrequentlyused Vfl

family, accountingfor 17%ofthe Vflrepertoire;however, Vfl2

comprises only 3.7% ofthe total repertoire. Both DILS and

Sjogren's syndromeexhibitconsiderable overlapin

expressing

skewed

Jf

segment usagewherebothdiseasesemphasize only4 of 13

_Jol

segmentsinaccountingfor 70% of therepertoire.

Whileasuperantigenmechanismhasbeenpostulatedas an

explanationfor distortions in thefrequency ofVf family repre-sentation in AIDS ( 12), it is unlikely that thepresentfindings

of skewedVfl expressionin DILSare aconsequence of expo-sure to anHIV-relatedsuperantigen;suchamechanismfailsto providean explanation for the selectivity of the

Jo

segments and the similarities in other features of CDR3 structurethat werefound inthe present study. Another possibleexplanation for Vflgene segmentrestriction could be selective trafficking

into the salivaryglandofTcellsexpressing particular

VO

fami-lies. Comparative analysis with primary

Sjdgren's

syndrome, used hereprincipallyas amethodologicand biologic control, suggests that the

Vf3

biasin DILSis notsimply the result of salivary glandtropismofaselect subsetofTcellssincethe two diseasesdisplay Vflrepertoires thatare onlypartially overlap-ping.

antigen-specific T cells in pathologic lesions. For example, Bloom andcolleagues demonstrated that only 2% of infiltrat-ingT cellsincutaneous tuberculoid leprosy lesions responded inanantigen-specific manner to Mycobacterium leprae (30), and the relevance ofthese data to the current study is enhanced bythe recentfinding of asubsetofpeptide-specifichuman T cell clones responsiveto M. leprae that alsouseTCR f-chains withinasingleVflfamilyand also encode identical amino acids atposition 97 of the CDR3 region (31 ). Similarly, in studying the antigen-induced animal model of autoimmunity, experi-mentalautoimmune encephalomyelitis,Cohen andcolleagues recently reportedthat only5% ofTcellsderived from patho-logic lesionswere reactive to the stimulating antigen myelin basicprotein (32). The estimation ofan evenhigherfrequency ofantigen-specificTcells in DILS salivary tissuemay be par-tially explained by the findings of severalrecentstudies which have consistently shown that HIV + individuals demonstrate markedlyelevated numbersof HIV-reactiveTcells evenin the peripheral blood (33-35); presumably this ispartly a conse-quence of thesystemic distribution ofHIV- 1 coupledwith the anatomical localization of the infectiontotheimmunesystem itself. Infact,despiteelementsoforganspecificitysuch asthat exhibited in DILS, the disseminated nature ofthe infection predicts that many of the same structurally selective TCR's couldbeisolated from the peripheralblood asthose isolated from the salivary gland, especially in consideration of the uniquely prominent peripheral CD8 lymphocytosis in this HIV + group.

Acknowledgments

Wethank Ms. ModriePayne for excellent assistanceinpreparationof

themanuscript.

This workwasfunded in partbyagrant from the New Yorkchapter oftheArthritis Foundation (E. Dwyer), a Postdoctoral Fellowship

from the Arthritis Foundation (S. Itescu), National Institutes of Health grantsAl19411andAR39626,and AmericanFoundation for AIDS Research grant 500255-12-PG.

References

1. Itescu,S., L. J. Brancato,J. Buxbaum,P. K. Gregersen,C. C.Rizk,S.

Croxson,G.Solomon,and R. Winchester. 1990.Adiffuse infiltrative CD8 lym-phocytosissyndrome in humanimmunodeficiencyvirus(HIV) infection:ahost immune responseassociatedwith HLA-DR5.Ann.Intern.Med. 112:3-10.

2. Itescu,S.,J.Dalton, H. Zhang, and R. Winchester. 1993. Tissueinfiltration

inaCD8lymphocytosissyndromeassociated withHIV- 1infectionhas the pheno-typeofanantigendriven process. J.Clin.Invest.91:2216-2225.

3. Itescu,S.,E.Dwyer, S.Choi,H.Zhang,Y.Huang,A.Friedman-Kien,and R.Winchester. 1992. Thesicca syndromein HIV- 1infection isdependentonthe hostimmune response rather than theinfectingviralstrain.Arthritis Rheum.

35:S210.(Abstr.)

4.Itescu,S.,U.Mathur-Wagh,M. L.Skovron,L.J.Brancato,M.Marmor,A. Zeleniuch-Jacquotte, andR.Winchester. 1992. HLA-B35 isassociated with accel-eratedprogressiontoAIDS. J.AcquiredImmuneDefic. Syndr.5:37-45.

5. Fox, R.I., F. V.Howell,R.C. Bone,andP.Michelson. 1984.Primary Sjogren syndrome: clinical andimmunopathologic features. Semin. Arthritis Rheum. 14:77-105.

6.Matis,L. A. 1990. The molecular basis of T-cellspecificity.Annu.Rev.

Immunol.8:65-82.

7.Kronenberg, M., G.Siu,L. E.Hood,and N.Shastri. 1986. The molecular

geneticsof the T-cellantigenreceptor and T-cellantigen recognition.Annu.Rev. Immunol. 4:529-591.

8.Howell,M.D.,J.P.Diveley,K. A.Lundeen,A.Esty,S.T.Winters,D. J. Carlo, and S. W.Brostoff.1991.Limited T-cell receptor beta-chainheterogeneity

amonginterleukin 2receptor-positivesynovialTcellssuggestsarolefor

superan-tigenin rheumatoid arthritis. Proc.Natl.Acad. Sci. USA. 88:10921-10925. 9. Paliard, X., S. G. West, J. A. Lafferty, J. R. Clements, J. W. Kappler, P. Marrack, and B. L.Kotzin. 1991. Evidence for the effects ofasuperantigenin

rheumatoidarthritis. Science(Wash. DC).253:325-329.

10.Bucht, A.,J.R.Oksenberg, S. Lindblad, A. Gronberg, L. Steinman, and L. Klareskog. 1992. Characterization ofT-cell receptor alpha beta repertoire in syno-vialtissuefromdifferenttemporal phases ofrheumatoid arthritis.Scand.J. Im-munol. 35:159-165.

11.Kotzin,B. L.,S. Karuturi, Y. K. Chou, J. Lafferty, J. M. Forrester, M. Better, G. E. Nedwin, H. Offner, and A. A. Vandenbark. 1991. Preferential T-cell receptor beta-chain variable gene usein myelin basic protein-reactive T-cell clones frompatientswith multiple sclerosis. Proc.Nati.Acad. Sci. USA. 88:9161 -9165.

12.Imberti,L., A. Sottini, A. Bettinardi, M. Puoti, and D. Primi. 1991. Selec-tivedepletion in HIV infection of T cells that bear specific T cell receptor V beta sequences. Science(Wash.DC). 254:860-862.

13.Williams,W. V., Q. Fang, D. Demarco, J. VonFeldt, R. B. Zurier, and D.B. Weiner. 1992.Restricted heterogeneity of T cell receptor transcripts in rheumatoidsynovium.J.Clin.Invest.90:326-333.

14. Davey, M. P., M. M. Meyer, D. D. Munkirs, D. Babcock, M. P. Braun, J. B. Hayden, and A. C. Bakke. 1991. T-cell receptor variable beta genes show differentialexpressionin CD4 and CD8 T cells. Hum. Immunol. 32:194-202.

15.Rosenberg,W. M.C., P. A. H. Moss, and J. I. Bell. 1992. Variation in human T-cell receptorVBand JB repertoire: analysis using anchor polymerase chain reaction. Eur. J. Immunol. 22:541-549.

16.Chothia, C.,D. R. Boswell, and A. M. Lesk. 1988. The outline structure of the T-cellalphabeta receptor. EMBO (Eur. Mol. Biol.Organ)J. 7:3745-3755. 17.Davis,M. M., and P. J. Bjorkman. 1988. T-cell antigen receptor genes and T-cellrecognition.Nature(Lond.). 334:395-402.

18. Jorgensen, J. L., U. Esser, B. F. De StGroth, P. A. Reay, and M. M. Davis. 1992.MappingT-cell receptor-peptide contacts byvariantpeptide immunization ofsingle-chaintransgenics.Nature(Lond.).355:224-230.

19. Engel, I., and S. M. Hedrick. 1988. Site-directed mutations in the VDJ

junctional regionofaTcell receptor beta chain cause changes in antigenic pep-tiderecognition.Cell.54:473-484.

20. Hedrick,S. M., I. Engel, D. L. McElligott, P. J. Fink, M. L. Hsu, D.

Hansburg,and L.A.Matis. 1988.Selection of amino acid sequences in the beta chainofthe T cellantigenreceptor.Science(Wash. DC). 239:1541-1544.

21.Laurence,J.,A.S. Hodtsev, and D. N. Posnett. 1992. Superantigen impli-cated independenceofHIV-1replicationin T-cells on TCRVBeta expression. Nature(Lond.).358:255-259.

22. Choi, Y. W., B. Kotzin,L. Herron, J. Callahan, P. Marrack, and J.

Kappler. 1989. Interactionof Staphylococcusaureustoxin "superantigens" with humanTcells. Proc.NatI. Acad.Sci. USA. 86:8941-8945.

23.Loh,E. Y., J. F.Elliott,S.Cwirla, L. L. Lanier, and M. M. Davis. 1989.

Polymerasechain reaction withsingle-sidedspecificity: analysis of T cell receptor deltachain. Science(Wash.DC).243:217-220.

24.Plaza,A., D. H.Kono, andA.N.Theofilopoulos.1991. New human VB genes andpolymorphicvariants.J.Immunol. 147:4360-4365.

25.Tecott,L.H.,J. D.Barchas,andJ. H.Eberwine.1988. Insitu

transcrip-tion:specific synthesisofcomplementaryDNAin fixed tissue sections. Science

(Wash.DC).240:1661-1664.

26.Wilson,R.K.,E.Lai,P.Concannon,R. K.Barth,and L. E. Hood. 1988.

Structure,organizationandpolymorphismof murine and human T-cell receptor

alphaand betachain gene families.Immunol.Rev.101:149-172.

27.Robinson,M. A.1991. The human T cell receptorbeta-chaingene

com-plex containsatleast 57variablegene segments.Identification ofsixVbeta genes infournewgenefamilies.J.Immunol. 146:4392-4397.

28.Kabat,E.A.,T. T.Wu,H.M.Perry,K. S.Gottesman,and C. Foeller. 1991.U.S.Dept.of Health and HumanServices, NIH, Bethesda,MD.

29.Sumida, T.,F.Yonaha,T.Maeda,E.Tanabe,T.Koike,H.Tomioka,and S.Yoshida. 1992.Tcell receptor repertoireofinfiltratingTcellsin lipsof

Sjogren'ssyndromepatients.J.Clin. Invest. 89:681-685.

30.Modlin,R.L.,J.Melancon-Kaplan,S. M. M.Young,C.Pirmex,H.King,

J.Convit,T.H.Rea,and B. R.Bloom. 1988.Learningfrom lesions:patternsof tissue inflammation inleprosy.Proc.Natl.Acad.Sci. USA. 85:1213-1217.

31.vanSchooten,W.C.A.,J. L.Ko,N.vanderStoep,J. B.A.G.Haanen,L.

Pickering,R.R. P.deVries,andB. R.Bloom. 1992. T-cell receptor B-chain gene usage in the T-cellrecognitionofMycobacterium leprae antigensinone

tubercu-loid leprosypatient.Proc.Natl.Acad.Sci. USA. 89:11244-11248.

32.Mor, F.,andI.R.Cohen. 1992.Tcells in the lesion ofexperimental

autoimmuneencephalomyelitis.J.Clin.Invest.90:2447-2455.

33.Clerici, M.,D. R.Lucey,R.A.Zajac,R. N.Boswell,H. M.Gebel,H.

Takahashi,J.A.Berzofsky,and G. M.Shearer. 1991. Detection ofcytotoxicT

lymphocytesspecificforsynthetic peptidesofgpl60inHIV-seropositive individ-uals. J.Immunol.146:2214-2219.

34.Lieberman, J.,J. A.Fabry,M.C.Kuo,P.Earl,B.Moss,and P. R. Skolnik. 1992.CytotoxicTlymphocytesfromHIV-1 seropositiveindividualsrecognize

immunodominantepitopesinGp 160 andreverse_{transcriptase.} J.Immunol.

148:2738-2747.

35.Hadida,F.,A.Parrot,M. P.Kieny,B.SadatSowti,C.Mayaud,P.Debre,

and B. Autran. 1992.Carboxyl-terminalandcentral regions ofhuman

immuno-deficiencyvirus- 1NEF_{recognized by cytotoxic}Tlymphocytesfromlymphoid

organs. Anin vitrolimitingdilutionanalysis.J.Clin. Invest. 89:53-60.