Regulation of the Mature HumanT CellReceptor y Repertoire by Biased V-J Gene RearrangementHitoshi Kohsaka, Pojen P. Chen, Atsuo Taniguchi, William E. R. Ollier, * and Dennis A. CarsonThe Department of Medicine and the Samand Rose Stein Institute for Research on Aging, University of California, San Diego, La Jolla,California 92093; and *Arthritis and Rheumatism Council Epidemiology Research Unit,University of Manchester, Manchester, United Kingdom

Abstract

To delineate how gene rearrangement influences the expressedhuman 'y T cell repertoire, we generated T cell receptor "y(TCRy) V domain-specific cDNAlibraries from the peripherallymphocytes of eight donors and sequenced a total of 232TCR-t gene transcripts. The libraries consisted of both in-frame and out-of-frame rearranged TCRy genes. The in-frameTCRy gene transcripts were used to determine the diversity offunctional T cells, whereas the out-of-frame transcripts, pri-marily derived from a,B T cells, were used to assess the frequen-cies of TCRV-y-Jy rearrangements in progenitor T lympho-cytes. The results showed that both sets of transcripts exhib-ited strikingly restricted V-y-J-y combinations. Only 11 of 40potential V-y-J'y rearrangements were common (2 3% of to-tal). The pattern of gene usage in the functional and nonfunc-tional transcripts was similar and did not differ markedlyamong donors. The only exception was the predominance ofVy9-JP in potentially functional transcripts from seven ofeight individuals. These results show that V-y-Jy rearrange-ment is nonrandom and suggest that the diversity of TCRygenes in the functional 'y T cell repertoire partly depends uponpreferentially rearranged Vy-J'y gene combinations. However,the expansion of Vy9 /V62 T cells in adult peripheral blood canonly be explained by antigenic selection of relatively rare V'y9-JP recombinants. (J. Clin. Invest. 1993. 91:171-178.) Keywords: gene recombination * adult peripheral T cell * anchoredpolymerase chain reaction * biased V-J gene usage * nonfunc-tional transcript

Introduction

T lymphocytes express two mutually exclusive types of antigenreceptor, either af3 or -ye, in association with the CD3 molecu-lar complex. The smaller y subset of T cells has been found inall vertebrate species examined. Most 'y cells do not expressCD4 or CD8 accessory molecules on their surface, and theirrecognition of antigen is not restricted by classical MHCmole-cules (1). Murine 'y T cells have a distinctive ontogeny andtissue distribution. The presence of two classes of murine 'yS T

Address correspondence to Hitoshi Kohsaka, M.D., Department ofMedicine 0663, University of California, San Diego, 9500 GilmanDrive, La Jolla, CA92093-0663.

Received for publication 24 March 1992 and in revised form 13August 1992.

cells, i.e., intraepidermal lymphocytes with restricted V'y/Vbreceptors and secondary lymphoid organ lymphocytes withvariable Vy /VA receptors, suggest that these two types of cellsmight fulfill special roles in the immune system (2). Human yT cells do not have the same tissue tropism. The physiologicalfunctions of murine and human yb T cells are still unclear.

A small set of V, D, and J gene segments encode human yereceptors. This limited diversity makes it possible to assess thefactors that control development and expression of the mature'ye T cell repertoire. By analogy with a13 T cell receptor(TCR),' these could include preferential gene recombination,positive and negative selection by self-antigens in the thymus,and peripheral expansion of T cell clones by antigens. To date,the adult 'ye T cell repertoire has been studied mainly withmonoclonal antibodies. Two dominant subsets of human ySTcells have been recognized in peripheral blood (3). Southernblot analyses of genomic DNAfrom cloned lymphocytes fromeach subset indicate that one class uses V'y9-JP-Cy 1 /VV62-Je 1-Ce 1 to form the TCRheterodimer whereas the other uses V'y2or Vy4-J2-Cy2/ VA I -Je I -CA 1 (4).

Evidence for regulation of TCRy gene rearrangement inearly ontogeny has come from analyses of ye transcripts in thefetal and neonatal lymphoid tissues (5, 6). However, whetherselective rearrangement of TCR-y genes influences the adultexpressed human -ye T cell repertoire is not known. One way todetermine the effect of intrinsic gene rearrangement on theexpressed repertoire is to compare the frequencies of all poten-tial Vy-Jy rearrangements in functional and nonfunctionalTCRVy genes. This is possible because the TCRy cDNA inperipheral lymphocytes includes both functionally and non-functionally rearranged transcripts (7).

We, therefore, used the anchored polymerase chain reac-tion (PCR) method (8) to construct TCRy cDNA librariesfrom the peripheral blood lymphocytes of eight individuals,compared gene frequencies among the in-frame and out-of-frame TCR^y transcripts, and sequenced a total of 232 V-y-J^ygenes. Except for V-y9-JP-C'y 1, both functional and nonfunc-tional genes exhibited a similarly restricted set of Vy-Jy combi-nations. Weconclude that regulation of TCR V^y gene rear-rangement is a significant force in molding the mature func-tional -ye T cell repertoire. However, antigenic selection, as inthe case of V-y9-JP-bearing cells, can potentially expand evenvery rare recombinants.

Methods

Blood samples. PBMCswere obtained from eight donors. Their agesranged from 39 to 65 yr. Mononuclear cells were isolated by density

1. Abbreviations used in this paper: PCR, polymerase chain reaction;TCR, T cell receptor.

Rearrangement of Human T Cell Receptor y Genes 171

J. Clin. Invest.© The American Society for Clinical Investigation, Inc.0021-9738/93/01/0171/08 $2.00Volume 91, January 1993, 171-178

centrifugation and were stored in medium containing 10% DMSOat-80°C until used.

RNA isolation and first-strand cDNA synthesis. Total RNAwasisolated from 5 x 105 PBMCs using RNAzolB (Cinna/Biotecx,Friendswoods, TX), which is based on the acid guanidium thiocya-nate-phenol-chloroform method (9). First-strand cDNAwas synthe-sized from total RNAusing an oligo-dT primer and Superscript reversetranscriptase (Life Technologies, Inc., Gaithersburg, MD). The result-ing cDNA/RNAhybrids were first size selected (0.5-3 kb) on agarosegels and extracted with phenol. The remaining RNAwas hydrolyzed byincubation at 60°C in 0.3 MNaOH.

dG tailing offirst-strand cDNA and anchored PCR. To amplify anentire TCRy V domain cDNA segment regardless of sequence, thefirst-strand cDNAwas poly dG tailed with dGTPand terminal deoxy-transferase. Subsequently free dGTP was removed by two sequentialammonium acetate precipitation steps. The G-tailed first-strandcDNAwas subjected to primary anchored PCRamplification in 100 ,ulof 10 mMTris-HCl (pH 8.3), 50 mMKCI, 2 mMMgCl2, 0.001%gelatin, 200 ,uM dNTPs, 1 U Perfect Match Polymerase Enhancer(Stratagene, Inc., La Jolla, CA), and 2.5 U AmpliTaq DNApolymer-ase (Perkin-Elmer Cetus Instruments, Norwalk, CT). The primersconsisted of a Cy-specific antisense oligonucleotide (Cyb: 5'-CACG-GTCGACTCAAGAAGACAAAGGTATGT)and anchor primers(9:1 mixture of ANprimer: 5'-ATTACGGCGGCCGCGGATCCandANC primer: 5'-ATTACGGCGGCCGCGGATCCCCCCCCCCCC-CC), at a concentration of 1 MM. The ratio of AN1 and AN1Cwas based on the report by Loh et al. (8). The anchor primerscontained sequences that are recognized by restriction enzyme NotI(GCGGCCGC)and BamHI (GGATCC), respectively, whereas theC'yb primer had a sequence recognized by SalI (GTCGAC) at its 5'end. These restriction sites facilitated the subsequent subcloning of thePCRproducts. The amplification included 20 cycles of 95°C for 30 s,42°C for 30 s, and 72°C for 1 min followed by final extension of 7 min.The products were size selected (500-750 bp) on agarose gels and onethird were removed for nested PCR. The reaction mixture was thesame as that of the primary anchored PCRexcept for the primers, 1 4Meach of ANprimer and antisense Cya primer ( 5'-ACGCGTCGACGG-AAGAAAAATAGTGGGCTT,which contained a Sal I recognition se-quence at its 5' end, and primes upstream of the Cyb primer sequence)and the concentration of MgCl2 (1.5 mM). 20 cycles of nested PCRwere carried out at an annealing temperature of 55°C, and the productswere purified by phenol extraction.

Construction of TCRy-specific cDNA libraries. A short fragment ofthe multicloning site of pBluescript II SK+ (Stratagene, Inc.) was re-leased by enzymatic digestion with Sacd and Sal I restriction enzymesand inserted into the polylinker of the M1 3mp19 vector, to produce aM13 vector that includes Sal I and NotI sites. The PCRproducts weredigested with Sal I and NotI to make cohesive ends, and fragments ofthe appropriate size were separated on agarose gels and extracted withphenol. The purified cDNA was ligated with the SalI/NotI-digestedM13 vector and the ligated vector was electrotransformed into SUREEscherichia coli (Stratagene, Inc.).

Screening andfrequency analysis of TCR-y-specific cDNA libraries.M13 plates, with plaques containing amplified TCR-y V domaincDNA, were lifted five times with Hybond-N+ nylon membranes(Amersham Corp., Arlington Heights, IL). Both an internal Cy probe(Cyc: 5'-TAAACAACTTGATGCAGATG,recognizing a region up-stream of the Cya primer-priming sequence) and Vy family-specificprobes (V-yI: 5'-GTCAGAAATCTTCCAACTTGGAAG,VyII: 5'-GACGGCACTGTCAGAAAGGAATCT,VyIII: 5'-TCAGGCTTTG-GAGCACCTGATCT,and VyIV: 5'-CAAAGGCTTAGAATATT-TAT) were labeled with digoxigenin in 20 Ml of 300 pmol oligonucleo-tide, 200 mMpotassium cacodylate, 25 mMTris-HCl (pH 6.6), 1.5mMCoCI2, 0.25 mg/ml bovine serum albumin, 125 ,M digoxigenin-1 l-dUTP (Boehringer Mannheim Biochemicals, Indianapolis, IN), 5,MM dATP, and 50 U terminal deoxytransferase. Five membranes werehybridized with the different probes at 42°C. After washing the mem-

branes, positive plaques were visualized using the Genius system(Boehringer Mannheim Biochemicals). Briefly, the membranes wereblocked with blocking reagent and incubated with antidigoxigenin an-tibody conjugated with alkaline-phosphatase. Then they were soakedin an adamantyl 1,2-dioxetane phosphate (AMPPD; Tropix, Inc., Bed-ford, MA) solution and incubated at 37°C to activate the phosphatase.Positive signals were recorded on x-ray film and enumerated.

Conventional PCR. The ability of the VyIV PCRprimer to amplifyrearranged Vy 11 genes (from the VyIV family) was confirmed by con-ventional PCR. Template DNAwas isolated from PBMCs. Amplifica-tion was carried out using 100 Ml of 10 mMTris-HCl (pH 8.3), 50 mMKCI, 1.5 mMMgCl2, 0.001% gelatin, 200MMdNTPs, 2.5 UAmpliTaqDNApolymerase, and 1 MuMof the VyIV family-specific sense oligonu-cleotide and either the Cy-specific antisense oligonucleotides (Cyb orCyc: 5'-ATCTGCATAAGTTGTTTAprimer without restriction en-zyme site) or a J 1,J2-specific antisense oligonucleotide (J 12:5'-GTGTTGTTCCACTGCCAAAG).The PCR products were visual-ized after separation by agarose gel electrophoresis.

Enrichment ofa(x T cell or -yb T cell populations. PBMCsfrom onedonor were incubated at 2 x 107 cells for 30 min at 4°C with FITC-conjugated mouse anti-TCR6 chain antibody (TCR6 1; T Cell Sciences,Inc., Cambridge, MA) at 0.8 ,g/ml in 0.8 ml of RPMI 1640 mediumsupplemented with 10% bovine serum albumin. Cells were washedtwice and incubated for 30 min at 4°C with 0.3 mgof magnetic beadsconjugated to sheep anti-mouse IgG (DYNAL, Inc., Great Neck,NY). After washing and magnetic removal of the beads, cell-coatedbeads were used directly for RNAextraction of cells enriched in yb Tlymphocytes. The remaining cells were further incubated with 3 mgofthe beads to maximize the removal of yb T cells and bound cells wereremoved by magnetic separation. The final cell suspension was used forRNAextraction of cells enriched in a/3 T lymphocytes. Cytofluorome-tric analysis showed that the enriched afi T cell population contained0.3% yb cells compared with 6% in unfractionated PBMCs.

DNAsequencing. Replica form double-strand DNAand/or single-strand phage DNA were extracted from bacterial cultures contain-ing recombinant M13 and used as sequencing templates. Sequenc-ing was carried out with the dideoxynucleotide chain-terminationmethod (10).

Results

TCR'y V domain gene repertoire of potentially functional andnonfunctional transcripts. Of the 14 reported human Vy genes,8 are potentially functional, whereas 6 are pseudogenes. Theeight potentially functional genes are divided into four fami-lies. The VyI family includes five functional genes (Vy2, V'y3,Vy4, Vy5, Vy8), whereas the three other Vy families haveonly one functional member (Vy9, Vy 10, and V-y 1 1, respec-tively). Four oligonucleotides (V-yl, VyII, VyIlll, and V'yIV)were designed to specify members of each corresponding Vyfamily. Poly G-tailed cDNAs were synthesized from mRNAsof PBMCsobtained from eight donors. Expressed Vy geneswere amplified from the cDNAs using primary anchored PCRand nested PCR. After TCRy-specific cDNAlibraries were con-structed in a modified M13 vector, they were screened withboth a Cy probe and Vy family-specific probes. Gene frequen-cies were calculated by dividing the number of Vy family posi-tive clones by the number of total TCRy clones.

In the first set of experiments, the reproducibility and fidel-ity of the amplification method was validated. cDNAobtainedfrom one donor was subjected to three different PCRamplifi-cations. The calculated frequency data were conserved amongdifferent experiments and varied by < 5%.

In eight separate TCRy gene-specific libraries from eight

172 H. Kohsaka, P. P. Chen, A. Taniguchi, WE. R. Ollier, and D. A. Carson

donors, no clones were positive with the VyIV family-specificprobe. Approximately 10 clones from the set of each Vy familypositive plaques were randomly selected and a total of 232V-y-Jy-C,y genes were sequenced. All genes could be classifiedaccording to the usage of Jy chains. Since the VyI-family in-cludes five functional genes, Vyl subgroup positive clones werefurther classified by the usage of Vy genes. Although theCy-specific primers used for amplification and probing werecomplimentary to both the Cy 1 gene and the Cy2 gene, the Cychain usage was deduced from the known Jy segment configu-rations (11).

The potential functionality of each clone was determinedby examining the sequence at the V-J junction for rearrange-ments maintaining the correct reading frame. Representative

Table I. Representative Frequency Analyses of Individual Donors

Donor A Vyl Vyll (V-y9) V4III (Vy 10)

Frequency in total transcripts 26.3 39.9 33.8Frequency in potentially

functional transcripts 27.0 52.2 20.8Frequency in nonfunctional

transcripts 25.6 28.3 46.1

J C segment usage VYI VYII VYIII(n= 10) (n= 11) (n= 10)

Translational frame In Out In Out In Out

JP I -CY 1 2* 2 2JP-C'yl 6J 1-C71 2§ 1JP2-Cy2J2-C72 3* 311 1 4 1 4

* Vy8(2). t Vy3, Vy5, Vy8. tVy4, Vy5. 11 Vy2(2), Vy3.

Donor B VyI V-rI (V-y9) VyIII (Vy 10)

Frequency in total transcripts 29.0 3.0 68.1Frequency in potentially

functional transcripts 51.0 1.2 47.9Frequency in nonfunctional

transcripts 0 5.4 94.6

J C segment usage VYI V-II VyIiI(n = 8) (n = 9) (n = 10)

Translational frame In Out In Out In Out

JPI-Cyl 2* 2 5JP-C' IlJ1-C'y1 2 1JP2-Cy2JP-Cy2 6* 1 7

* Vy2(2). * Vy2, Vy3, Vy8(4).

results from three donors (A, B, and C) are shown in Table I.Although the TCRy repertoires varied among donors, a simi-larly skewed pattern of Vy-Jy combinations was observed in allcases.

The V-J junctional sequences of amplified TCRy cDNAfrom the same donor are shown in Table II. An extensive num-ber of different sequences at the V-y-Jy junction (N region) wasobserved. Only a single pair of identical sequences was foundamong 30 separate clones. These data show that the PCRam-plification was performed without preferential amplification ofparticular sequences.

Average frequencies of transcripts classified by Vy and Jychain usage. On the basis of the frequency and junctional se-quence data of Vy family genes, the frequency of each Vy-Jy

Donor C VyI V-yII (Vy9) VyITI (V 10)

Frequency in total transcripts 48.8 41.5 9.6Frequency in potentially

functional transcripts 41.0 52.0 7.0Frequency in nonfunctional

transcripts 63.4 20.0 16.6

J C segment usage Vyl V'yII VYIII(n = 10) (n = 9) (n = 10)

Translational frame In Out In Out In Out

JPl-CYl 1* 1 2JP-Cy1 6J l-Cy IJP2-Cy2J2-C,y2 6* 3§ 2 1 5 2

* Vy4. $ Vy2(3), V-y3, V-y4, Vy8; § V-y8(3).

The frequencies of expression of the Vyl, Vyll and VyIII gene fami-lies in the libraries were calculated by enumerating positive plaquesof each Vy family, compared to the total Cy positive plaques. Then,

10 clones from each Vy family were picked up randomly, se-quenced, and were classified according to the usage of Vy and Jy genesegments. For each family, the numbers of in-frame (in) and out-of-frame (out) transcripts are shown. The individual Vyl familygene members are indicated with footnote symbols. The use of Cygene segments was deduced from the known configuration of TCRygenes (1 1).

Rearrangement of Human T Cell Receptor y Genes 173

+ + + + + + + + + + + + + +

C-1 - I.- ,- 1 - C-) - -1 I. - , I.-, C- I--, C-C C- CI- C-) C-) `-) C-

CDCDCD O CD U CD H F, H

EqHE- E-E H vE

000000v¢ ¢ H EHHE-

E-4EH U :

HHHOOH H

EOH0E¢

¢¢¢E-4 E- ¢E

00<<: HHc¢ EA E¢ C

HE-H 0-E-0

UE-4 ¢:

OOH

EH E- H E UOOO000 <<00E<4¢0000DCD CD H O H COE HHOOOHOUq D H<¢OHE<0 00OOCU C DH00 000H;UCD

0 ~ <00¢O OOHH~~~~~C

< <<00~~Eo 0 00000D 00000<0D CD 0HD ¢HHHHOO000000 E)0E0

¢ ¢ ¢ ¢OHHH<<<<<<<000O COO OOHDHDC)COOO)CDOOOOHHHCE00q v0 E0 C0OOC

E-E

E-4¢

<4EqEvE vE E- ¢

E E-4 Ev EvE C

CD4 U C C C E-4

¢ H H H¢¢¢ ¢¢ ¢

¢H ¢H¢ HH¢H¢¢ EOOO¢¢ HC

H H HEvH HqH <qE

~

HO

¢ ¢CD

uCD C

Eq E-4 E-E H Eu C

OOD vHH

00

qHH00C

00CDCDC00CDCHE

HH¢¢C)C00D vE

0000DODC)OOOHHD¢<<<<00EE EO OOOHH000000Eq¢0000<vHEqHHHHHHHv E

H ¢HOOOH H4

EH H¢ HHH H OHE4E- E E E-4

H E

OEOH

C-- E-4E-¢ ¢-- E-4

<< 0H E- E-4 H

<, HOOE4E4H

H E- E4 0O<-4E

¢ 4 E- E-4 E H E4HH H-E4 ¢ E- E-

¢ E-4 E-4 E4 E-4P

H H H H

¢< <00 0

H H HHE H HH00 000OCC DC000000DCC00000000UC CHHHHHEvH HH00000000;D D00000000zC CHHH E HHHHHv00000000CC C CHHHEvHEvH HHH00000000U) D

HHHHHHHH00000000

H E<<<<<<0<C;DHHHHHHHHHHHvEvEvHHHHH

O O 00000t} oC1 'I st oo (1 all (ON (1 all O" a, (ON Os (:CON - - - - - - - - - -

07- 0 m vtt 00ID0WO" -o a-'N'N

174 H. Kohsaka, P. P. Chen, A. Taniguchi, WE. R. Ollier, and D. A. Carson

q EH F4H H FqEH H Ev

Eq H Eq0 0 0.

I00

HE-4FH

HE4 -4

HH

H

0<u¢ ¢ ¢

<: ¢ ¢¢ ¢

uDc

cD u

¢ ¢-40-4

E- E4 E-4uu

¢u u

u

~E

cU

-

z

CU

0.

02

0-

_Cm

CU

CU

U)

0^

U)

0

0

U)

0

C-

CU

U)

O.)

o c-U)

U)

CU

o Q

C-C

0-

~U)

._ _

U)'_

LLCs

L.00

CC.

CC-)

0

0

C--)

CC-)

c-C)

C-,

CC-)

0C.)

-0

Table III. Average Frequencies of Transcripts for Each V-y-J'y Combination from Eight Donors

Vy gene

J-y-Vy combination V-y2 V-y3 V-y4 V-y5 V-y8 VY9 V-y10

Potentially functional transcriptsJPI-CY 1 1.5 0 1.0 0 1.1 0 5.3JP-Cy I 0 0 0 0 0 25.3 0Jl-C'yl 0 7.0 0 0 0 0.4 4.1JP2-Cy2 0 0 0.6 0 0 0 0J2-C-y2 6.6 3. 1 8.2 1.4 9.2 8.9 16.6

Nonfunctional transcriptsJPI-CYl 0 0 0 0 0 0 17.8JP-Cy I 0 0 0 0 0 0.9 0Jl-CYl 0 0 0.9 0.9 0 1.0 13.9JP2-C,y2 0.6 1.1 0 0 0 0 0J2-Cy2 9.7 3.5 4.1 4.6 5.8 13.1 22.1

Average frequencies of the potentially functional transcripts (upper) and nonfunctional transcripts (lower) for each V-y-Jy-Cy combination areshown. The data are from eight individuals, including the donors shown in Table I. The frequency of each Vy gene family was determined byplaque hybridization in all eight libraries; and then individual gene usage and gene functionality was assessed by sequence analysis of 10 clonesfrom each Vy family, in all eight libraries. Finally, the average frequencies for the V-y-Jy combinations were calculated.

combination was determined in the potentially functional andnonfunctional repertoire of each donor. Since the basic patternof Vy-Jy combinations was similar among donors, the averagefrequencies of transcripts with each Vy-J-y combination werecalculated (Table III). In both the in-frame transcripts and theout-of-frame transcripts, a biased increase of particular Vy andJ-y pairs was observed. The J2 gene segment was used in 54%ofthe in-frame transcripts and in 63% of the out-of-frame tran-scripts. The VT I0-JP1, VT I0-J 1, and VT 10-J2 gene combina-tions also occurred frequently in both the in-frame and out-of-frame transcripts. Among the 232 clones sequenced, the JP2gene segment was used only three times (weighed frequency2.3%). In all the donors except one, Vy9-JP gene rearrange-ments were dominant among the in-frame transcripts and wererare among the out-of-frame transcripts.

No detectable transcription of the TCR VT]] gene in adultperipheral lymphocytes. As noted earlier, the VyIV family-spe-cific probe did not hybridize to any clones in the TCRy-specificcDNA libraries. Conventional PCR reactions with a Cyaprimer and a VyIV family-specific primer also failed to am-plify any TCRy gene product. To validate the ability of thechosen VyIV-specific oligonucleotide to prime the VT 11 geneand to verify the presence of a rearranged VT 11 gene in PBMC,genomic DNA from several donors was amplified with theVyIV primer and the J 1 ,J2-specific primer (J 12). In each case,PCRyielded the expected 230-bp product, under conditionswhere no detectable products were amplified in cDNA fromPBMCusing the same VTIV primer (results not shown). Col-lectively, these results indicate that TCRVT 11 transcription isminimal in human lymphocytes that have rearranged TCRV-y I I genes.

The source of nonfunctional transcripts. On the basis of theVTy gene family frequencies in the total transcripts from severaldonors and the sequence data, we estimated that 28-54% of thetranscripts were nonfunctional because of frame shifts or intro-

duced termination codons. To discern the source of the non-functional TCRy gene transcripts, PBMCsfrom one donorwere sorted into ca3 T cell-enriched and y6 T cell-enrichedpopulations. RNAwas extracted from each population or fromtotal lymphocytes and TCRy transcripts were amplified by an-chored PCR. After constructing TCRy libraries, randomlypicked clones were sequenced and examined for their potentialfunctionality. The results showed that unfractionated lympho-cyte population had 73% potentially functional TCRT tran-scripts compared with 27% in the a: T cell-enriched popula-tion and 90% in the yb T cell-enriched population. Taken to-gether, these data imply that most nonfunctional TCRytranscripts in peripheral blood lymphocytes come from a: Tcells.

Discussion

These experiments show that the expressed potentially func-tional TCRTgene repertoire in adult peripheral lymphocytesderives from a limited set of V'y-Jy combinations. Only 11 of40 potential Vy-Jy rearrangements were detected at a fre-quency of > 3%. Unrelated subjects had the same basic patternof Vy-J-y gene use. This restricted usage of particular V-y-Jygene segments could be explained by three possible mecha-nisms. One is the presence of potent and commonantigens inthe periphery, which can stimulate and expand T cells bearingparticular types of TCRy genes. Another is the preferentialrearrangement of Vy segments to Jy segments at the progeni-tor T cell level. The last mechanism is positive or negativeselection in the thymic environment, which is known to affectV: gene usage in peripheral T cells of mice (12-15).

To delineate how the regulation of V-y-Jy gene recombina-tion places limits on the mature yb T cell repertoire, transcriptsof potentially functional and nonfunctional rearranged TCR-y

Rearrangement of Human T Cell Receptor T Genes 175

genes were compared. The frequencies of out-of-frame V'y-Jygene rearrangements cannot be controlled directly by antigenselection in the thymus or the periphery. The results showedthat a similarly restricted set of Vy-Jy-Cy genes was used inboth functional and nonfunctional TCRy gene rearrange-ments. Most of the V'y-J'y combinations that appeared fre-quently in the in-frame transcripts were also found in the out-of-frame transcripts. As an example, the TCR J2 gene,rearranged to several different Vy gene segments, was foundfrequently in both functional and nonfunctional transcripts.

The reputed locations of the V, J, and Cgene segments andtheir frequencies of recombination in both functional and non-functional transcripts are shown in Fig. 1. The prevalence ofV,y 10 and J2 might be explained by positional effects, becausethe TCRVy 10 gene is the most downstream V-y gene (exceptfor Vy 11) and J2 is the most downstream J-y gene. In thisregard, Triebel et al. ( 16) showed that the unexpressed TCRyalleles of eight cloned y T-cell lines had downstream V'Y genesjoined to upstream Jy segments. On the basis of analysis of fourneonatal and three fetal thymocyte clones, Krangel and col-leagues (5) proposed an ordered rearrangement of TCRygenes. McVay et al. (6) also described regulated expression ofTCR,y and TCRbgenes in fetal thymus and gut lymphocytes.However, the TCRy genes studied by these workers had mini-mal N region additions, as has been seen with both TCRandimmunoglobulin genes that rearrange early in ontogeny ( 17-

21 ). In contrast, both the functional and nonfunctional TCRygenes in mature peripheral blood lymphocytes displayed exten-sive junctional diversity (Table II). Although the adult andfetal TCRy repertoires are not equivalent, the transcripts inmature peripheral blood -y5 T cells still derived from a limitedset of V-J combinations. Despite the fact that yb T cell clonesexpressing Vy 10 gene product have not been reported, wefound many functionally rearranged Vy 10 genes in peripheralblood lymphocytes. The discrepancy is probably due to thecurrent lack of specific antibodies against the Vy 10 region.

The TCRVy 11 gene is known to be potentially functional,but it has been unclear if it can express a TCRy gene product.In our experiments, no V'y 11 gene transcription in peripheralblood lymphocytes was detected using either conventionalPCR or anchored PCR, although we could readily amplifyrearranged Vy 11 genes from the same specimens. These resultssuggest that TCRVy 11 genes are rarely transcribed in matureperipheral blood lymphocytes, in accord with results recentlyreported by two other groups (21, 22). McVay et al. (6) usedconventional PCR to demonstrate V111 gene expression infetal tissues. However, the PCRprimer that they used for Vy 11gene amplification (termed GVIV) is very homologous to theabundantly transcribed Vy 10 gene, sharing 20 out of 22 bases,and these workers noted that expression of the Vy 11 gene wasalways accompanied by expression of the Vy 10 gene. Indeed,we found that the cloned Vy 10 gene was readily amplified by

Potentially functional transcripts

~ 25.3%%116.6%

9.2%-

~ 8.9%

-~ l 8.2%-7.0% I6.6%

5.3%vy2 vy3 Vy4 V' V8 Vy9 Vylo Vyl n JP ji cyl JP2 J2 Cy2

4.1%,

4.6%

5.8%X

9.7%

13.9%-17.8%1

L 0- - - -0.9% - --

Non-functional transcripts

Figure 1. Genomic organization of potentially functional Vy, Jy, and C-y genes on chromosome 7 and the recombination frequencies. The eightmost frequent Vy-Jy combinations in potentially functional transcripts and in nonfunctional transcripts are shown. The frequency of Vy9-JPin nonfunctional transcripts is shown for reference.

176 H. Kohsaka, P. P. Chen, A. Taniguchi, W. E. R. Ollier, and D. A. Carson

I

-I

PCRusing the Cy-specific primers with the GVIV primer ofMcVay, but not with the VyIV primer chosen for the presentstudies (data not shown).

Because antigenic stimulation can potentially expand evenrare T cells in a population, the Vy-J-y pairs in the functionalTCRy repertoire could be different, and more or less diverse,than those in the nonfunctional repertoire. However, the twogene repertoires were remarkably similar. The major discrep-ancy was in the V-y9-JP transcripts, that accounted for only0.9% of the out-of-frame transcripts, but 25.3% (on the aver-age) of the in-frame transcripts. This predominance of func-tional Vy9-JP transcripts was observed in seven of eight do-nors. The data indicate that lymphocytes expressing rearrangedV-y9-JP-Cy 1 genes were not preferentially generated during 'y6T cell development but instead were expanded postthymicallyfrom small numbers of precursors. Consistent with this conclu-sion is the occurrence of two identical V'9-JP joining se-quences in randomly chosen transcripts from the same donor(Table II). The monoclonal antibody TiyA recognizes a V-y9-associated epitope and stains the major population of periph-eral y6 T cells in most adults (23). Triebel et al. (24) showedthat all TiyA+ T cell lines transcribe V-y9-JP-Cy 1 rearrangedgenes. Parker and colleagues (25) discovered that Vy9 cells arenot predominant at birth but expand in the first 10 years of life.In this regard, T cells with V'y9/V62 TCRhave been shown toreact with staphylococcal enterotoxin A, with the Daudi andMolt-4 lymphoblastoid cell lines, and with mycobacterial anti-gens (26-29).

Although it is still controversial if progenitor T cells onlyattempt a and : rearrangements when y and 6 rearrangementsfail to produce functional transcripts, TCRy gene rearrange-ments are often found in a13 T cells (30, 31 ). Someaf lympho-blastoid cell lines are known to transcribe TCRy genes (32).The results of our experiments with enriched af3 and hyb T cellpopulations suggest that the nonfunctional TCR-y gene tran-scripts in peripheral blood lymphocytes also derive mainlyfrom a: T cells.

It is possible that some functionally rearranged TCRy tran-scripts come from af T cells and that different T cell clonestranscribe V^y genes at different rates. However, it is very un-likely that these transcripts strongly influenced the overall V'y-Jy frequencies in the entire T cell population. Thus, we foundthat Vy9-JP and Vy2- or Vy4-J2 expression occurred mostfrequently (average of 25.3 and 14.8%; respectively) in the poolof potentially functional transcripts. The prevalence of the tworearrangements agrees with previous enumeration analysescarried out with monoclonal antibodies (3, 4).

In summary, the random assortment of V and J gene seg-ments in the TCR y locus can produce 40 different recombi-nants. Amongthe 232 genes that were analyzed, only 11 combi-nations appeared frequently in both in-frame and out-of-frametranscripts. These data show that biased V7y-J-y rearrangementis an important force in molding the mature TCRy repertoire.(The nomenclature used for human TCRy gene segments isthat of Lefranc et al. [33] and Forster et al. [34].)

Acknowledgments

The authors acknowledge the secretarial assistance of Ms. Nancy Noonin the preparation of this manuscript.

This work was supported in part by grants AR25443 and AR40770

from the National Institutes of Health and by a grant from the Ciba-Geigy Corporation (Suncrest, NJ).

References

1. Brenner, M. B., J. L. Strominger, and M. S. Krangel. 1988. The -yb T cellreceptor. Adv. Immunol. 43:133-192.

2. Allison, J. P., and W. L. Havran. 1991. The immunobiology of T cells withinvariant -y/6 antigen receptors. Annu. Rev. Immunol. 9:679-705.

3. Falini, B., L. Flenghi, S. Pileri, P. Pelicci, M. Fagioli, M. F. Martelli, L.Moretta, and E. Ciccone. 1989. Distribution of T cells bearing different forms ofthe T cell receptor ay/ in normal and pathological human tissues. J. Immunol.143:2480-2488.

4. Fagioli, M., E. Ciccone, C. Bottino, B. Falini, F. Grignani, A. Moretta, L.Moretta, and P. G. Pelicci. 1990. The C-yl-encoded disulphide-linked and theC-y2-encoded nondisulphide-linked forms of the -y/6 heterodimer use different yand a variable regions. Blood. 76:279-284.

5. Krangel, M. S., H. Yssel, C. Brocklehurst, and H. Spits. 1990. A distinctwave of human T cell receptor -yl/ lymphocytes in the early fetal thymus: evi-dence for controlled gene rearrangement and cytokine production. J. Exp. Med.172:847-859.

6. McVay, L. D., S. R. Carding, K. Bottomly, and A. C. Hayday. 1991.Regulated expression and structure of T cell receptor -y/6 transcripts in humanthymic ontogeny. EMBO(Eur. Mol. Bio. Organ.) J. 10:83-91.

7. Yoshikai, Y., B. Toyonaga, Y. Koga, N. Kimura, H. Griesser, and T. W.Mak. 1987. Repertoire of the human T cell y genes: high frequency of nonfunc-tional transcripts in thymus and mature T cells. Eur. J. Immunol. 17:119-126.

8. Loh, E. Y., J. F. Elliott, S. Cwirla, L. L. Lanier, and M. M. Davis. 1989.Polymerase chain reaction with single-sided specificity: analysis of T cell receptora chain. Science (Wash. DC). 243:217-220.

9. Chomczynski, P., and N. Sacchi. 1987. Single-step method of RNAisola-tion by acid guanidinium thiocyanate-phenol-chloroform extraction. Anal. Bio-chem. 162:156-159.

10. Sanger, F. 1981. Determination of nucleotide sequences in DNA. Science(Wash. DC). 214:1205-1210.

11. Quertermous, T., W. M. Strauss, J. J. Van Dongen, and J. G. Seidman.1987. Human T cell gammachain joining regions and T cell development. J.Immunol. 138:2687-2690.

12. Kappler, J. W., N. Roehm, and P. Marrack. 1987. T cell tolerance byclonal elimination in the thymus. Cell. 49:273-280.

13. Bill, J., 0. Kanagawa, D. L. Woodland, and E. Palmer. 1989. The MHCmolecule I-E is necessary but not sufficient for the clonal deletion of Vf1 1 -bear-ing T cells. J. Exp. Med. 169:1405-1419.

14. Sha, W. C., C. A. Nelson, R. D. Newberry, D. M. Kranz, J. H. Russell, andD. Y. Loh. 1988. Positive and negative selection of an antigen receptor on T cellsin transgenic mice. Nature (Lond.). 336:73-76.

15. Liao, N. S., J. Maltzman, and D. H. Raulet. 1989. Positive selectiondetermines T cell receptor V,B14 gene usage by CD8+ T cells. J. Exp. Med.170:135-143.

16. Triebel, F., F. Faure, M. Graziani, S. Jitsukawa, M. P. Lefranc, and T.Hercend. 1988. A unique V-J-C-rearranged gene encodes a gammaprotein ex-pressed on the majority of CD3+ T cell receptor-a / 3-circulating lymphocytes. J.Exp. Med. 167:694-699.

17. Elliott, J. F., E. P. Rock, P. A. Patten, M. M. Davis, and Y. H. Chien. 1988.The adult T cell receptor b-chain is diverse and distinct from that of fetal thymo-cytes. Nature (Lond.). 331:627-631.

18. Feeney, A. J. 1990. Lack of N regions in fetal and neonatal mouse immu-noglobulin V-D-J junctional sequences. J. Exp. Med. 172:1377-1390.

19. Feeney, A. J. 1991. Junctional sequences of fetal T cell receptor d chainshave few N regions. J. Exp. Med. 174:115-124.

20. George, J. F., Jr., and H. W. Schroeder, Jr. 1992. Developmental regula-tion of D , reading frame and junctional diversity in T cell receptor-beta tran-scripts from human thymus. J. Imrnunol. 148:1230-1239.

21. Olive, C., P. A. Gatenby, and S. W. Serjeantson. 1991. Variable gene usageof T cell receptor y and 6 chain transcripts expressed in synovia and peripheralblood of patients with rheumatoid arthritis. Clin. Exp. Immunol. 87:172-177.

22. Doherty, P. J., S. X. Yang, R. M. Laxer, E. D. Silverman, R. Inman, and S.Pan. 1992. Evidence for clonal expression of T cell receptor Vyll + T cells in thesynovial fluid of patients with arthritis. J. Immunol. 149:295-299.

23. Faure, F., S. Jitsukawa, F. Triebel, and T. Hercend. 1988. Characteriza-tion of human peripheral lymphocytes expressing the CD3--y/b complex withanti-receptor monoclonal antibodies. J. Immunol. 141:3357-3360.

24. Triebel, F., M. P. Lefranc, and T. Hercend. 1988. Further evidence for asequentially ordered activation of T cell rearranging gammagenes during T lym-phocyte differentiation. Eur. J. Immunol. 18:789-794.

Rearrangement of Human T Cell Receptor y Genes 177

25. Parker, C. M., V. Groh, H. Band, S. A. Porcelli, C. Morita, M. Fabbi, D.Glass, J. L. Strominger, and M. B. Brenner. 1990. Evidence for extrathymicchanges in the T cell receptor -y/6 repertoire. J. Exp. Med. 171:1597-1612.

26. Fisch, P., M. Malkovsky, S. Kovats, E. Sturm, E. Braakman, B. S. Klein,S. D. Voss, L. W. Morrissey, R. DeMars, W. J. Welch, et al. 1990. Recognition byhuman Vy9/V62 T cells of a GroEL homolog on Daudi Burkitt's lymphomacells. Science (Wash. DC). 250:1269-1273.

27. Sturm, E., E. Braakman, P. Fisch, R. J. Vreugdenhil, P. Sondel, and R. L.Bolhuis. 1990. HumanVy9-Vb2 T cell receptor-gamma delta lymphocytes showspecificity to Daudi Burkitt's lymphoma cells. J. Immunol. 145:3202-3208.

28. De Libero, G., G. Casorati, C. Giachino, C. Carbonara, N. Migone, P.Matzinger, and A. Lanzavecchia. 1991. Selection by two powerful antigens mayaccount for the presence of the major population of human peripheral y/b Tcells. J. Exp. Med. 173:1311-1322.

29. Rust, C. J., F. Verreck, H. Vietor, and F. Koning. 1990. Specific recogni-tion of staphylococcal enterotoxin A by human T cells bearing receptors with theV-y9 region. Nature (Lond.). 346:572-574.

30. Ishida, I., S. Verbeek, M. Bonneville, S. Itohara, A. Berns, and S. Tone-gawa. 1990. T-cell receptor gammadelta and gammatransgenic mice suggest arole of a gammagene silencer in the generation of alpha beta T cells. Proc. Natl.Acad. Sci. USA. 87:3067-3071.

31. Takeshita, S., M. Toda, and H. Yamagishi. 1989. Excision products oftheT cell receptor gene support a progressive rearrangement model of the c/6 locus.EMBO(Eur. Mol. Biol. Organ.) J. 8:3261-3270.

32. Kimura, N., R. P. Du, and T. W. Mak. 1987. Rearrangement and organi-zation of T cell receptor gammachain genes in human leukemic T cell lines. Eur.J. Immunol. 17:1653-1656.

33. Lefranc, M. P., A. Forster, R. Bear, M. A. Stinson, and T. H. Rabbitts.1986. Diversity and rearrangement of the human T cell rearranging gammagenes: nine germ-line variable genes belonging to two subgroups. Cell. 45:237-246.

34. Forster, A., S. Huck, N. Ghanem, M. P. Lefranc, andT. H. Rabbitts. 1987.New subgroups in the human T cell rearranging V gammagene locus. EMBO(Eur. Mol. Biol. Organ.) J. 6:1945-1950.

178 H. Kohsaka, P. P. Chen, A. Taniguchi, W. E. R. Ollier, and D. A. Carson