receptor complex deduced from cdna sequences: evolution of the

The EMBO Journal vol.5 no. 8 pp. 1799 - 1808, 1986

Primary structure of the T3 -y subunit of the T3/T cell antigenreceptor complex deduced from cDNA sequences: evolution of theT3 -y and 6 subunits

Geoffrey W.Krissansen, Michael J.Owen1, Winston Verbiand Michael J.CrumptonImperial Cancer Research Fund, P.O. Box 123, Lincoln's Inn Fields,London WC2A 3PX, and 1ICRF Tumour Immunology Unit, Department ofZoology, University College, Gower Street, London WC1E 6BT, UK

Communicated by M.J.Crumpton

cDNA clones, whose fusion proteins were recognised by ananti-(T3 -y chain) serum, were isolated from a Xgtll expressionlibrary prepared from the human T leukaemia cell line J6.The clones encoded a unique sequence related to that of theT3 6 chain, and hybridised to two mRNA transcripts of 0.8and 3.5 kb in size, whose expression was restricted to T lym-phocytes. The 182 amino acid sequence deduced from thecDNA revealed a typical signal peptide, a predominantlyhydrophilic 89 amino acid residue domain with two N-glyco-sylation sites, a hydrophobic domain with a centrally locatedglutamic acid residue and a 44-residue domain with at leastone potential serine phosphorylation site for protein kinaseC. Given this arrangement the T3 -y polypeptide most prob-ably has a transmembrane orientation with the N-terminaldomain exposed on the cell surface. The amino acid and nu-cleotide sequences showed marked homology with those ofthe T3 6 chain, suggesting that the respective genes arose byduplication about 200 million years ago. The intracellular andmembrane-proximal half of the extracellular domains wereespecially well conserved.Key words: T3 -y subunit/cDNA sequence/evolution/gene dupli-cation/primary structure

IntroductionAntigen recognition and response by T lymphocytes is mediatedby a complex comprising the antigen receptor per se (Ti), associ-ated non-covalently with the T3 antigen. The Ti molecule is madeup of two immunoglobulin-like transmembrane a and chains,and the T3 antigen of three (or possibly four) invariant polypep-tides (namely -y, 6, e and t) (Borst et al., 1984; Hood et al.,1985; Meuer et al., 1984; Owen and Collins, 1985; Tsoukas etal., 1984). One approach to defining the role of the T3 antigenin mediating the response to antigen-receptor interaction dependsupon the availability of clones encoding the T3 polypeptides. Suchclones also provide important information on the structures ofthe polypeptides. cDNA clones for the T3 6 and e chains haverecently been isolated using two different approaches (Van denElsen et al., 1984; Gold et al., 1986). The 6 chain of 21 000Mr has an extracellular domain of 79 amino acid residues withtwo N-linked oligosaccharides, one transmembrane segment andan intracellular domain of 44 residues (Van den Elsen et al.,1984), whereas the e chain of 19 000 Mr comprises two hydro-philic domains of 104 and 81 amino acid residues with no poten-tial sites for N-linked glycosylation. Both the 6 and e chains havean acidic amino acid residue within the hydrophobic transmem-brane region that conceivably interacts with an appropriately

IRL Press Limited, Oxford, England

placed basic residue in the transmembrane segments of the Tia and ( chains, thereby stabilising the Ti/T3 complex on the cellsurface. Most importantly, the 6 and e polypeptides share noapparent homology with each other or with the Ti a and ( chains(Terhorst et al., 1986). Expression of mRNAs encoding the 6and e chains is restricted to T cells including the most pheno-typically immature human T cell leukaemia cell lines (Terhorstet al., 1986).The -y chain has, in the past, often been regarded as a minor

component of the T3 antigen (Borst et al., 1982) and has proveddifficult to study. As judged from analyses of T3 immunopre-cipitates, its proportion relative to 6 and e appeared to vary, sug-gesting that it is more weakly associated with the otherpolypeptides than 6 and c (Borst et al., 1982, 1983a,b). Also,when separated from surface 1251-labelled T lymphocytes, itresolved poorly on SDS-PAGE as a weak diffuse band of about26 000 Mr and the mature form could not be detected after bio-synthetic labelling with radioactive amino acids (Kanellopouloset al., 1983). The -y chain may, however, be of particular func-tional importance. Thus, chemical cross-linking studies have indi-cated that T3/Ti interaction is mediated principally between theTi , and T3 y polypeptides (Brenner et al., 1985). Furthermore,activation of protein kinase C by phorbol esters induces a down-regulation of the cell surface T3/Ti complex and a concomitantprominent phosphorylation of the T3 y chain, suggesting an inter-dependence of these two events (Cantrell et al., 1985).

This paper describes the isolation and characterization ofcDNAclones that encode the human T3 y polypeptide. The deducedamino acid sequence revealed an 89 amino acid residue predomi-nantly hydrophilic domain containing two potential sites for N-glycosylation, one typical transmembrane segment and a 44-resi-due hydrophilic domain. Comparison of the nucleotide and de-duced amino acid sequences of the -y and 6 polypeptides revealedmarked homology, suggesting that the respective genes arose byduplication about 2 x 108 years ago. The C-terminal domain ofthe 'y polypeptide contained one unique potential site for phos-phorylation on serine by protein kinase C, as well as two otherserine residues which are shared by the corresponding segmentof the 6 polypeptide.

Results and DiscussionIsolation of cDNA encoding the T3 ey polypeptideA cDNA clone encoding the T3 'y polypeptide was identifiedby screening a Xgtl 1 expression library, constructed from thepoly(A)+ mRNA of the human T leukaemia cell line J6, witha combination of polyclonal antisera raised in rats against theindividual purified T3 polypeptides. The properties of the anti-sera, none of which reacted with lysates of B lymphocytes, aresummarized in Figure 1B. By Western blotting, the anti-(y chain)serum recognized bands of 19 000, 23 000 and 25 000-27 000Mr in the microsome fractions of HPB-ALL (Figure IB, lane3) and J6 (not shown), whereas a pool of antiserum from indi-vidual rats immunized with either the 6 or e polypeptide visualizeda broad band of about 20 000 Mr (Figure iB, lane 1). Desig-

1799

G.W.Krissansen et al.

A

68-

45U-34 -

25 -

12 5- fi

1 2 3 4

B

4m am e*"d _ - _..0

__r-__+>~.

14 -

1

. t

I I

. I S

1 2 3 4 5 6 7 8 9 10 11

Fig. 1. Specificity of antisera against the T3 polypeptides. (A) SDS -PAGE analyses of purified T3 polypeptides used for immunization: lane 2, e; lane 3, 6;lane 4, y stained with silver. Numbers on the side refer to relative masses (Mr) in kd of the mol. wt markers: lane 1, bovine serum albumin (68 000 Mr)ovalbumin (45 000 Mr), glyceraldehyde-3-phosphate dehydrogenase (34 000 Md), immunoglobulin light chain (25 000 Md, cytochrome C (12 500 Md).(B) Western blotting analysis of antisera. Microsomes (2 x 107 cell equivalents/lane) or Endo-F-treated T3 immunoprecipitates (6 x 107 cell equivalents/lane)were separated by SDS-PAGE, transferred to nitrocellulose and blotted. HPB-ALL microsomes, (lanes 1,3); MOLT-4 microsomes, (lanes 2,4);immunoprecipitates prepared from HPB-ALL (lanes 5,7), MOLT-4 (lanes 6,8), BRI 8 (lanes 9,10) and HSB-2 cells (lane 11). Lanes 1,2,5,6 and 9, pooledanti-(E chain) and anti-(6 chain) sera; lanes 3,4,7,8,10 and 11, anti-(+y chain) serum. Numbers on the side refer to relative masses (Md in kd of the pre-stained mol. wt markers (Bethesda Research Laboratories Inc., MD); bovine serum albumin (68 000 Mr), ovalbumin (45 000 Md), a-chymotrypsinogen(26 000 M), (3-lactoglobulin (18 000 Mr), lysozyme (14 000 Mr).

nation of the above bands was based upon the results of Endo-Ftreatment, which leads to reduction in the sizes of the 'y and 6polypeptides to 16 000 and 14 000 Mr, respectively and no de-tectable change in the e chain (19 000 Mr) (Borst et al., 1983a;

1800

Kanellopoulos et al., 1983). The anti-(Qy chain) serum gave a

strong band of 16 000 Mr together with bands of 19 000 and25 000 Mr when blotted against Endo-F-treated purified T3 anti-gen prepared from HPB-ALL and MOLT-4 cells (Figure lB,

a

I Ae

68-

45-

26 -

18-

-68

-45

- 26

-18

. 14IPW

-6 .-AMW .

0.f

Structure of the -y chain of human T cell T3 antigen

Xbal

600

EcoRI

800 bP

I pJ6T36-1 1

I -pJ6T36-2 1

i PJ6T36-3.0

-20Met glu gln gly lys gly leu alaAM G A C-AG GG AG G0C CIM G0T 61

-1 +1val leu ile leu ala ile ile leu leu gln gly thr leu ala gln ser ile lys gly asn his leu wlGMC C AI CMI GCr AW ATT CM CTrr AA GGr ACT GOC C AW AMA GG AAC CAC GMT 130

EX

+10 +20lys val tyr asp tyr gln glu asp gly ser va leu leu thr cys asp ala glu ala lys an ile tkhAAG GIM TAT GAC TAT CM GA OAT GGr TG GM CIrr CG ACT GA GM GM MOCAAA AT AT ACA 199

+40 +50trp p*e lys asp gly lys net ile gly phe leu thr glu asp lys lys lys trp asn le gly ser asnOG Tmr AMA GOT 0GG AAG m Am G0c Tm CrA ACr GAA GAT AAA AAA AMA G AAT CIGG Nmr AAT 268

+60 *

ala lys asp po arg gly net tyr gln cys lys gly ser gln asn lys ser lys pro leu gln al tyrGOC AAG GAC Ocr cGA aGO AmG TAT CAG AA GG IA CA C AAG 'CAMA C CIC CA GIG TAT 337

+80 +90 +100tyr arg met cys gln asn cys ile glu leu an ala ala thr ile ser gly phe leu phe ala glu ileTC AGA AM WT CAG MC TOC ATT GA CTA AMT WA GOC ACC ATA IT GtEC mI CI TIT GCr GA MC 406

Tm+110 +120

val ser ile phe val leu ala val gly val tyr phe ile ala gly gln asp gly val arg gln ser arg

GrC MC ATT TIrC CrcGC GrOT GG0G Gr mC ATT Gcr aM CGAGOT Gm MT CGC CAG MAG 475

Cyt

+130 +140ala ser asp lys gln thr ]eu leu pro asn asp gln leu tyr gln pro Ieu lye asp arg glu asp asp

GCT Ti GAC AM CA ACT CIG TM C MCAT GAC CAG CC mC GMG C Cc AAG GAT CG GAC GAT QC 544

+150 +160gln tyr ser his leu gln gly asn gln leu arg arg asn ***CMG TAC AiGC C Crr CA a C Cl'TIG AGG MAT AGGACrCMGAC7CAACcAGr IM=rCcCC 621

T _A9Clqq>TCCDUAlCrAFIC=E 712

803

ACTACIAV 831

Fig. 2. Nucleotide and deduced amino acid sequence of the T3 -y chain. (A) Restriction endonuclease map. Cleavage sites are indicated by closed circles andthe extent of sequencing is indicated by arrows. The restriction enzymes used were EcoRI, AccI, XbaI and Sau3A. Untranslated and coding regions are

designated - and Li, respectively. (B) Nucleotide and deduced amino acid sequences of the pJ6T3-y cDNA clones. The numbers in the right margin shownucleotide positions, whereas numbers immediately above the amino acid sequence indicate amino acid residue positions. The initiation methionine has beenassigned to position -22, even though the open reading frame extends to the 5' end of the sequence. Two potential sites for N-linked glycosylation areindicated by asterisks. The putative signal peptide (SP), extracellular (EX), transmembrane (Tm) and cytoplasmic (Cyt) domains are designated by horizontalarrows below the sequence. Potential polyadenylation signals are overlined.

1801

A EcoR I

KLT

0

Sau3A Sau3AI

AccI Sau3AI I

200 400

B


a I me* 00 00 *0 *0*

a sr^n

0* 0 0 0 0 0-22 M E Q G K G L A V L I L A I I L L Q G T L A QS I K G N

00 00000000 00 0 00 00000- 0000 000 0000---** *

-21 M E H S T F L S G L V L A T L L S O V S P F

* 00 0 0 0 .H L V K V Y D Y QE D G S V L L T C D A E A K N I T W F K

AAGhTAcCWA1UI1#AITAr1NAC ATATA AAGGAK I P I E E L E D R V F V N C N T S IT W V

0* 0 0* 00 to..0 0D G K M I G F L T E D K K K W N L G S N A K D P R G M YQ C

GA TV L I TRDLGKR I LDPRG IYRCE G T V G T L L S D I T R L D L G K R I L D P R G I Y R C

0K G S Q NAA GG00 0C0r

* 0K S K P L

AAGrCA0 AA*0 000.. 0 *0 00*Q V Y Y R M C Q N C I E L N A A TC0= W _X X C

00000 0000 0000000 0000 000 0 000 00 00000000-*

N G T D I Y K D K E S T V Q V H Y R M C Q S C V E L D P A T

0 0*0 0 0 0. 0I S G F L F A E I V S I F V L A V G V Y F I A G OD G V R O

V A G I I V T D V I A T L L L A L G V F C F A G H E T G R L

* 0 0 0 00 000 0000 0 0 00 00S R A S D K T L L P N D 0 L Y 0 P L K D R E D D Q Y S H L

S G A A D T Q A L L R N D Q V Y Q P L R D R D D A Q Y S H L

00 0o G N Q L R R N

G G N W A R N K

AATAAATATA7W0CI'1'lTCAl

31

90

121 (6)

162 (1)

208(35)

22B(23)

298(65)

315(52)

379(92)

405(82)

469(122)

495(112)

559(152)

585(142)

649(160)

675(150)

739

699

AA 831

Fig. 3. Comparison of the nucleotide and deduced amino acid sequences of the -y and 6 chains. The -y and 6 chain sequences, apart from the 5' and 3'untranslated regions, were aligned so as to maximise the homology. Identical nucleotides and amino acids are indicated by asterisks and dots, respectively.The numbers in the right margin show the nucleotide and amino acid (bracketed) positions. The vertical arrows define the boundaries of the segments chosenfor comparisons of nucleotide divergence (see Materials and methods).

lanes 7,8). The apparent recognition of the E chain by the anti-(-y chain) serum does not appear to reflect contamination of thepurified -y polypeptide used for immunization (Figure IA, lane4), and thus may indicate a shared epitope. In contrast to theabove results, the pooled anti-(6, E chain) sera revealed strongbands of 14 000, 19 000 and 25 000 Mr (Figure 1B). The25 000 Mr band shown in lanes 5-8 was also detected by blot-

ting 'dummy' preparations of T3 antigen prepared from B celllines using an identical procedure to that employed for purifyingT3 (Figure 1B, lanes 9, 10). It most likely represents traceamounts of immunoglobulin light chain released from the UCH-TI (monoclonal antibody)-Sepharose used in the purification ofT3. It was concluded that the anti-(-y chain) serum recognizedthe 'y and E chains and the anti-(S,e) serum the 6 and E chains.

1802

T3-6

T3-Y

T3-6

T3-Y

T3-6

T3-Y

T3-6

T3-'Y

T3-6

T3-Y

T3-6

T3-Y

T3-6

T3-Y

T3-6

T3-Y

T3-6

T3-Y

T3-Y

.


Given these specificities clones encoding the T3 -y chain can beidentified on the basis that they react with the former, but notthe latter antiserum.Two stable clones which satisfied the above criteria were ob-

tained by screening - 2 x 106 transformants of the J6 cDNAlibrary. One clone only yielded a cDNA insert after digestionwith EcoRI. Its cDNA (-0.6 kb) was subcloned into plasmidpUC13 to give the clone pJ6T3'y-1. Determination of thenucleotide sequence of the insert, after subcloning into Ml3mpl8,revealed a unique sequence with good homology to that of theT3 6 chain. Alignment with the deduced coding sequence of theT3 6 chain indicated that pJ6T3-y-1 contained the entire transmem-brane and cytoplasmic coding regions, but lacked the N terminus.Approximately 106 transformants were re-screened using theEcoRI -AccI fragment (Figure 2A) of pJ6T3-y- 1. Two newclones (pJ6T3'y-2 and pJ6T3-y-3) with inserts of -0.78 and0.76 kb respectively, were isolated.Nucleotide and amino acid sequencesThe sequencing strategy together with the nucleotide and deducedamino acid sequences of clones pJ6T37y-1, -2 and -3 are givenin Figure 2A,B. A hydropathy plot (Kyte and Doolittle, 1982)of the deduced amino acid sequence revealed two predominantlyhydrophobic segments, namely amino acid residues -22 to -1and 90-116 (data not shown). Although there is a continuousopen reading frame from nucleotide position 1, several argumentsare in favour of ATG at position 38-40 acting as the initiationsite for translation. Most importantly, it is followed by a predomi-nantly hydrophobic segment characteristic of a signal peptide.The initiation codon is also flanked by the consensus sequence,PurXXAUGPur, for functional initiation codons in eukaryoticmRNAs (Kozak, 1983). Furthermore, the polypeptide definedby this initiation codon has an Mr of -18 000 (excluding thesignal peptide), which is similar to that revealed by SDS-PAGEanalyses of Endo-F-treated T3 -y chain (Figure iB). The N-terminal sequence (residues -22 to 89), preceding a hydrophobicsegment (residues 90-116), incorporates two potential sites forN-linked glycosylation (residues 30 and 70, Figure 2B) in agree-ment with the results of Endo-F digestion (Krissansen et al., inpreparation). This suggests that the N terminus of the moleculeis extracellular and that the encoded polypeptide has a transmem-brane orientation with a predominantly hydrophilic intracellulardomain.Homology with the T3 6 chainComparison of the amino acid sequence deduced from the cDNAclones (Figure 2B) with those for all the proteins in the NationalBiomedical Research Foundation database, using the search pro-gram of Wilbur and Lipman (1983), revealed no significant hom-ologies apart from with T3 6 chain. Figure 3 shows the nucleotideand corresponding amino acid sequences of the T3 -y and 6 chaincDNA clones. A model summarizing the principal structural fea-tures of the T3 ey and 6 chains is shown in Figure 4. It can beseen that the amino acid sequence of the Py chain shows a strikingoverall homology with the predicted sequence of the T3 6 chain(Van den Elsen et al., 1984). The N terminus (residues 1-57)shared the least homology with the corresponding region of the6 chain (34%); in particular, it was necessary to insert nineresidues immediately following the signal peptide to optimizealignment. In contrast, the membrane proximal region (residues58-89) showed considerable homology (63%), especially theshort sequence (residues 76-89) immediately preceding theputative transmembrane segment (71 % conservation at the aminoacid level including the two cysteines at 82 and 85). The remain-

ing two cysteines of the extracellular domain (residues 24 and65) are also conserved as are the sequences flanking cysteine 65.These observations suggest that the cysteines may have an im-portant influence on the folding of the extracellular domain. Con-sistent with this interpretation are the results of fragmentationstudies of the 6 chain which indicated that some of the five half-cysteines are connected by intra-chain disulphide bridges (Vanden Elsen et al., 1984). A comparison of the secondary struc-tures of the extracellular domains using the algorithm of Chouand Fasman (1978) revealed, however, marked differences. Thus,the structure of thePy chain extracellular domain was complexwith no predominance of any particular secondary structure,whereas the 6 chain domain showed stretches of either (3-turnsor,B-strands with the latter being predominant in the segmentbetween the two cysteines at positions 16 and 72. Although thelatter secondary structure would fit in with the presence of animmunoglobulin fold (Amzel and Poljak, 1979), the invariantresidues characteristic of the immunoglobulin domain were ab-sent. The relative positions of the N-linked glycosylation sitesadjacent to the two most N-terminal cysteines is maintained inboth chains, although they are separated from the cysteines byfour to five residues in thery chain.The two predominantly hydrophobic segments of the 7y chain,

namely the signal peptide and the putative transmembrane region,shared little amino acid homology with the corresponding seg-ments of the 6 chain (44 and 37%, respectively). This observationis consistent in the case of the transmembrane region with thegeneral requirement for conservation of hydrophobicity ratherthan amino acid identity. The transmembrane domain containeda centrally placed acidic residue in common with the other T3polypeptides. However, the -y chain possessed glutamic acid atthis position (100) rather than aspartic acid in the 6 and c chains(Van den Elsen et al., 1984; Gold et al., 1986).The putative intracellular domains are of identical size and are

highly conserved in amino acid sequence (57% identity with afurther 18% conservative substitutions). Interestingly, there isa stretch of charged residues in the centre of the intracellulardomains of both the 'y (residues 142-147) and 6 chains.

Protein processingAlthough the exact position at which the signal peptide is cleavedcannot be assigned, the site can be predicted using known cleav-age patterns. Using the computations of Von Heijne (1983) todetermine the site with the highest 'processing probability', cleav-age is postulated to occur after the alanine at position -1. At-tempts to obtain the N-terminal amino acid sequence for the -ychain have consistently failed, suggesting that the et amino groupis blocked. Such a conclusion is consistent with the above assign-ment of the cleavage site, since N-terminal glutaminyl residuesare often blocked naturally by cyclization or if unblocked areparticularly difficult to sequence (Allen, 1981). If this assign-ment is correct, then the extracellular domain of the mature -ychain comprises 89 amino acid residues. In this case, the mature-y chain has an Mr of - 18 200 and contains 10 more residuesthan the 6 chain. This molecular size is anomalously high whencompared with that determined by SDS-PAGE for the Endo-F-treated y chain isolated from surface-labelled T cells (- 16 000Mr; Kanellopoulos et al., 1983). A similar discrepancy has beenfound for the 6 chain, where it was suggested that the discrepan-cy reflects proteolytic cleavage of the cytoplasmic domain (Borstet al., 1984). A similar mechanism could also explain the observ-ed discrepancy for the -y chain. However, no direct evidence fora post-synthetic C-terminal cleavage of the -y chain exists.

1803


SPI 1

-22 1

CHOC--N24 30

CHO Tm P

C-N C-C_E _S-S-ES70 8 S 90 100 1

65 70 62 65 90 116 123 12615

CHO

6CN-I 16

CHO Tm I

CN CC-CD_C-S S

52 7275 60 90 103106 113 140 150

Fig. 4. Comparison of the principal structural features of the T3 -y and 6 chains. Predominantly hydrophobic and hydrophilic stretches are designated *, and-, respectively. SP and Tm refer to signal peptide and transmembrane domains. C, N, E, D and S are the one letter amino acid code for cysteine,asparagine, glutamic acid, aspartic acid and serine, respectively. CHO designates potential attachment site for asparagine-linked glycosylation. P indicatesputative serine phosphorylation sites with the asterisk designating the most probable site for phosphorylation by protein kinase C as predicted from a study ofthe phosphorylation of synthetic substrates (O'Brian et al., 1984). Numbers refer to amino acid positions.

Although a prominent compact band of - 23 000 Mr was reveal-ed in HPB-ALL microsomes by Western blotting with anti+ychain) sera (Figure iB, cf. lane 3 with lane 2) that was coinci-dent with a transient band in biosynthetic pulse-chase experiments(Kanellopoulos et al., 1983), this band gave a 16 000 Mr bandon Endo-F treatment (Figure IB, lane 7). A more likely explana-tion for the observed discrepancy is that it reflects an anomalousbehaviour of the -y polypeptide on SDS -PAGE, a problem whichis particularly associated with low mol. wt polypeptides (Weberet al., 1972).

Recently it has been shown that activators of protein kinaseC, such as phorbol 12,14-dibutyrate and 1-oleoyl-2-acetyl gly-cerol, down-regulate the T3/T cell antigen receptor complex, andcoincidentally markedly phosphorylate the -y chain on serine andto a lesser extent the 6 chain (Cantrell et al., 1985; Davies etal., 1985). Polypeptides containing multiple basic residues fol-lowed by the sequence Ala-Ser can be substrates for phorbol-ester-stimulated phosphorylation by protein kinase C (O'Brianet al., 1984). One possible site in the Py chain that meets these

121 126

requirements is the Arg-Gln-Ser-Arg-Ala-Ser sequence just proxi-mal to the transmembrane segment; interestingly, the serine atposition 126 is an alanine in the 6 chain. However, since the 6chain was weakly phosphorylated one or both of the conservedserines at positions 123 and 150 can probably also act as sub-strates (Figures 3 and 4).

Expression of 'y chain transcriptsIn order to assess the expression of mRNA transcripts encodingthe -y chain, poly(A)+ RNA from normal human peripheralblood T lymphocytes and several cell lines was analysed byNorthern hybridization analysis. In agreement with the expressionof the polypeptide revealed by Western blotting (Figure iB), -y

chain mRNA was detected in peripheral blood T lymphocytesand both mature (J6, HUT-78) and immature (MOLT-4, HSB-2)T leukaemia cell lines, but not in the B lymphoblastoid cell lineBRI 8 (Figure 5). Studies of the regulation of T cell receptorgene expression have shown that transcription of the ,B chain gene

precedes that of the a chain. Thus, the most phenotypically im-mature T-lineage cell lines (HSB-2, MOLT-4) which possess highlevels of ,3 chain transcripts express low or undetectable a chainmRNA (Collins et al., 1985). The expression of transcripts ofthe T3 y and 6 chains appears to be similar to that of the Tichain, in that mRNA sequences were detected in both HSB-2and MOLT4, although at somewhat varying levels (Figure 5).When combined with the finding that the T3 e chain transcriptsare present in the earliest human T cell leukaemias (Furley etal., 1986), these results confirm earlier conclusions that Ca chainexpression occurs relatively late in thymic ontogeny and controls

1804

the surface expression of the T3/T cell antigen receptor complex(Raulet et al., 1985; Collins et al., 1985; Samelson et al., 1985b).

Interestingly, the insert of clone pJ6T3'y-1 hybridized to twotranscripts of -0.8 and 3.5 kb (Figure 5), even though thestringency used to wash the Northern blots was sufficient torender unlikely hybridization to the 6 chain transcript. Further-more, when the EcoRI-AccI fragment of clone pJ6T3-y-1 (Figure2A), corresponding to coding sequence with relatively low (60%)homology to the 6 chain, was used as a probe both sizes oftranscripts were detected with the same relative intensity (datanot shown). When the blot was stripped and re-probed with theinsert of a 6 chain cDNA clone (pPGBC-9; Van den Elsen etal., 1984) the 6 chain transcript was found to run marginallyahead of the 0.8 kb 'y chain mRNA. The larger 'y chain transcriptis unlikely to correspond to an unprocessed precursor RNA sincethe cDNA library was constructed from poly(A) + cytoplasmicRNA prepared after removal of nuclei. Inspection of the y chaincDNA sequence (Figure 2B) reveals that the two -y chain tran-scripts may have arisen by differential usage of polyadenylationsignals. The 3' untranslated region of the cDNA insert of pJ6T37y-1 is 247 bp long and contains three overlapping polyadenylationsignal sites (AATAAA; Proudfoot and Brownlee, 1976). Al-though the 3' end possesses a stretch of nine A's a further 45nucleotides downstream from the polyadenylation signal sites,is seems unlikely that these A's are the start of the polyA tail sincethe distance separating a signal site from the start of polyadenyl-ation is usually restricted to 15-30 bases. Although poly-adenylation signals other than AATAAA can be utilized (e.g.TATAAA, AATACA, ATTAAA), no candidate sequence ispresent between the cluster of AATAAAs and the 3' end ofpJ6T3-y-1. It appears, therefore, that pJ6T3'y-1 arose from tran-scription through the AATAAA cluster and that the stretch ofA's at the 3' end represents genomic sequence. Utilization of a

polyadenylation signal 2.5 kb downstream would result in the3.5 kb transcript detected by Northern hybridization analysis,whereas a 0.8 kb transcript(s) could be generated by polyadenyl-ation commencing at one of the three A's 18 -20 basesdownstream of the AATAAA cluster in pJ6T3y-1. The genera-tion of multiple transcripts by differential usage of polyadenylationsignals is not without precedent. For example, six RNAtranscripts are generated from the murine dihydrofolate reduc-tase gene all of which differ in the length of the 3' untranslatedregion (Farnham and Schimke, 1986).

Surprisingly, the different T leukaemic cell lines used the twotranscripts to different extents. Thus, peripheral blood T cellsappeared to express the 3.5 kb transcript only, whereas the leu-kaemic cell lines J6, HUT-78 and MOLT-4 expressed both tran-scripts. In contrast, the immature leukaemic cell line HSB-2

T3- 6

T3-8SP

-21

s%a

150 *o0

Structure of the y chain of human T cell T3 antigen

4b A..

gamma

3-5kb-.-_

0 8kb--

0

a a

4.

Is

a

delta

08kb-b-e.g0 a

Fig. 5. Expression of T3 -y and 6 chain transcripts. Labelled insert (-6 x 108 c.p.m./gg) from pJ6T3-y-l was hybridized to a blot (upper panel)containing 2 ytg of poly(A)+ RNA from peripheral blood T cells (HPBL)and 5 jig of poly(A)+ RNAs from the T leukaemia cell lines HUT-78,MOLT-4, HSB-2 and Jurkat, and the B lymphoblastoid cell line BRI 8. Theblot was then stripped and re-hybridized (lower panel) with labelled insert(- 8 x 108 c.p.m.//g) from the 6 chain plasmid pPGBC-9 (Van den Elsenet al., 1984). Different exposure times are shown in order to emphasize thedifferential expression of the two classes of -y chain transcript. The exposuretimes were: (upper panel) MOLT-4, HSB-2 and BRI 8, 60 h; HPBL, 24 h;HUT-78 and Jurkat, 16 h; (lower panel) HPBL, HSB-2, MOLT-4 and BRI8, 16 h; Jurkat and HUT-78, 6 h. Sizes were determined by comparisonwith the relative mobilities of ribosomal 28S and 18S subunits and of the Tcell antigen receptor a and (3 chain transcripts. Autoradiographs wereoverlaid to compare sizes of the 6 and -y chain transcripts.

expressed the 0.8 kb transcript predominantly. Both transcriptsare, therefore, most probably functional since peripheral bloodT lymphocytes and HSB-2 both contain the polypeptide, in the

Table I. Sequence divergence between T3 -y and 6 gene sequences

Coding regions Silent sites Replacement sitesNo. % No. %

Signal peptide 11/17 65 19/49 39Extracellular region A 14/27.3 51 47/98.3 48Extracellular region B 12/18.5 65 20/77.5 26Transmembrane region 6/22.2 27 21/58.8 36Cytoplasmic region 13/29 45 28/103 27

Non-coding regions N divergenceNo. %

5' non-coding region 16/39 413' non-coding region 44/88 50

Sequence divergence was estimated as described by Lomedico et al. (1979).The data for the T3 6 cDNA is from Van den Elsen et al. (1984); and thecomparions are based on the sequences described in Figure 4. Divergence(N) was calculated directly with each gap treated as a single site and as onesubstitution regardless of length. 5' and 3' non-coding regions were re-aligned to maximise homology. In particular, it was noted that the 5'sequence 1-16 of the y chain showed marked homology to the a chainsequence 29-45 and that the -y chain 3' sequence 723-748 showed markedhomology to the 6 chain sequence 630-651. Silent-site divergence forexamination of the evolutionary histories of the -y and 6 chains was obtainedfrom the overall divergence of the extracellular A (residues 1-57),extracellular B (residues 58-89), transmembrane, cytoplasmic and 3' non-coding regions. Replacement-site divergence was obtained from the overalldivergence of the coding regions. The signal peptide was excluded fromcomparisons because a complete sequence was not available for the murine6 chain signal peptide.

latter case intracellularly (Figure IB, lane 11). It is unclear whatregulates the expression of these two transcripts or whether theyare spliced identically.Evolution of the -y and 6 genesSequence divergence between the Py and 6 genes was quantitatedin separate segments of the coding sequence [namely signal pep-tide, extraceilular A and B (Figure 3), transmembrane and cyto-plasmic regions] and in the 5' and 3' non-coding regions. In thenon-coding regions divergence (N) was calculated directly, butin the former segments both silent-site (S) and replacement-site(R) divergence were calculated (Lomedico et al., 1979). Ssubstitution occurs at a high and approximately constant rate ir-respective of gene function for at least 85-100 million yearsof evolution (Perler et al., 1980), whereas R substitution varieswith the specific function of a protein sequence (Miyata et al.,1980; Lomedico et al., 1979). The results are shown in TableI. The small number of R substitutions for the extracellular Band cytoplasmic domains (26 and 27%, respectively) comparedwith the other domains suggests that amino acid replacementwithin these regions is fairly severely constrained. In contrast,the least constrained extracellular A domain has an almost 2-foldhigher level of R divergence (48%). In most genes the rate ofS substitution exceeds that of R (Miyata et al., 1980). This isnot true, however, for the IgXV genes (Miyata et al., 1980) anddoes not appear to hold for the transmembrane domain of the'y chain (Table I). The significance of these exceptions is not clear.S and R divergence can be used as an evolutionary clock to

estimate gene divergence. Estimates derived from S divergencebecome, however, increasingly error-prone after 85 million yearsof evolution. In contrast, R divergence is constant for at least500 million years and is characteristic for each polypeptide (Jef-freys, 1981). Using the 6 chain as a standard and assuming thatthe selective pressures on the 6 and -y chains are similar, it should

1805


be possible to obtain a more accurate duplication time for the,y and 6 genes from R substitutions. The 6 chain shows 22% Rsubstitution occurring since human/mouse divergence giving aunit evolutionary period (UEP) of 3.4. The human -y chain shows59% (corrected for multiple substitutions) R substitution whencompared with the6 chain, suggesting that 'y, 6 gene duplicationoccurred - 200 million years ago. Interestingly, the T cell recep-tor CO genes also diverged at least 100 millionyears ago (Tunna-cliffe et al., 1985). While such estimates are error-prone it is,nevertheless, clear that the ry, 6 genes duplicated beforehuman/mouse divergence. The recent identification of the murine6 gene and the description of the murine T3 antigen (Van denElsen et al., 1985; Samelson et al., 1985a) endorse thisassignment.

Interestingly over the short evolutionary period since mice andhumans diverged it is clear that the extracellular domain of the6 chain has diverged (58.2% amino acid homology) more thanthe intracellular domain (77.2% amino acid homology). Sinceall regions of the human 'y gene are not equally divergent in silentsites, it is possible that divergence with the more highly conservedtransmembrane and intracellular domains has been corrected bygene conversion at least 50-100 million years ago.Interaction of T3 with TiThe T leukaemia cell line MOLT-4, which does not expressdetectable T3 antigen of the cell surface, contains significant levelsintracellularly (Link et al., 1985). As judged by Western blot-ting with antisera against the -y, 6 and E chains (Figure IB, lanes6,8), MOLT-4 contains all three T3 polypeptides, although the-y chain was detected only as an immature form of 23 000 Mr(lane 4). MOLT-4 also expresses high levels of Ti-f chain tran-scripts, but Ti-a chain transcripts are almost undetectable (Col-lins et al., 1985). These results suggest the Ti-a chain is requiredfor assembly of the T3/Ti complex and for subsequent process-ing of the T3-,y chain to its mature 25 000-27 000 Mr form.Such regulation is reminiscent of the processing of HLA-A and-B antigens, in which association with 2-microglobulin is necess-ary for the conversion of the oligosaccharide unit of the heavychains to the mature form (Owen et al., 1980).The results of chemical cross-linking studies have indicated

that the predominant association between the T3 antigen and theT cell antigen receptor is between the T3--y and Ti-,B chains (Bren-ner et al., 1985). This association is conceivably strengthenedby a salt bridge between the transmembrane glutamic acid residueof the 'y chain and a corresponding basic residue in the transmem-brane region of the Ti-f chain, as has been suggested previouslyfor the 6 chain by Van den Elsen et al. (1985). Neutralizationof a charged amino acid residue within the transmembrane seg-ment may also be an essential requirement for efficient transpor-tation to the cell surface, as is suggested by recent studies onoligonucleotide-directed mutagenesis of the transmembrane do-main of vesicular stomatitis virus glycoprotein (Adams and Rose,1985).The stoichiometry of the subunits in the Ti/T3 complex is not

known. Since the ey and 6 polypeptides almost certainly arosefrom a duplicated gene it is possible to speculate that in somemolecular forms of the T cell antigen receptor the -y chain maybe able to replace the 6 chain and vice versa. However, it appearsmore likely that since separation over 200 million years ago thesechains have adopted separate functions, a suggestion which issupported by their markedly different degrees of phosphorylation(Cantrell et al., 1985). Also as T3, precipitated with monoclonalantibodies which appear to recognize the 6 chain only, contains

all the -y, 6 and e polypeptides (Borst et al., 1983b), it seemsthat both -y and 6 chains are integral parts of the same macro-molecule.

Materials and methodsCells and antibodiesCell lines were cultured in RPMI 1640 medium containing penicillin (100 U/mil),streptomycin (50 pg/ml) and 5% (v/v) fetal calf serum (Gibco Ltd., Paisley, UK).Peripheral blood T cells (80% T3 antigen-positive, 70% T8 antigen-positive) wereobtained by culturing phytohaemagglutinin-stimulated human peripheral bloodlymphocytes for 10 days in the presence of interleukin-2. Tonsils removed bysurgery were collected from hospitals in the Greater London area and stored at-70°C until used.The mouse monoclonal antibody UCH-TI (Beverley and Callard, 1981) was

purified from either a 40% saturated ammonium sulphate precipitate of a hybridomacell culture supernatant, or from ascitic fluid of the hybridoma grown in miceusing protein A-Sepharose CL-4B (Pharmacia Fine Chemicals) (Ey et al., 1978).It was coupled to Sepharose CL-4B as described by Kumel et al. (1979).

Purification of T3 polypeptidesThe preparation of the T3 -y, 6 and e polypeptides is described elsewhere (Kris-sansen et al., in preparation). Briefly, microsomes prepared from human tonsils(10-20 g wet wt from -40 tonsils) were solubilized in 1% Nonidet-P40. Thesoluble material was pre-cleared with mouse IgG-Sepharose CL-4B prior to ab-sorption with 10 ml of UCH-TI-Sepharose CL4B. The matrix was washed rapid-ly on a sintered glass filter with 500 mi of 10 mM Tris-HCl, pH 7.4, 1% NP-40,M NaCl followed by 500 ml of 0.15 M NaCl, 0.5% Na deoxycholate, pH

8.2. Adsorbed protein was eluted with 50 mM diethylamine, pH 11.5, 0.5% Nadeoxycholate and dialysed against 0.5% Na deoxycholate, pH 8.0. Followinglyophilization, the detergent was extracted with absolute ethanol and the proteincollected by centrifugation. The pellet was dissolved in SDS sample buffer andthe T3 polypeptides isolated by preparative SDS -PAGE using modifications ofthe procedure described by Kelly et al. (1983). Analysis of the purified polypep-tides is shown in Figure 1A.Anti-(T3 polypeptide) seraSprague-Dawley rats were primed with 0.8 ml containing 25-50 pg of oneor other of the purified T3 polypeptides emulsified in Freund's complete adju-vant and were boosted on three occasions at twice weekly intervals with 10-25 pugof polypeptide emulsified in Freund's incomplete adjuvant. All injections weregiven s.c. at multiple sites.

Immunoprecipitation and digestion with Endo-FMicrosomal preparations of HPB-ALL, MOLT-4, HSB-2 and BRI 8 cells (108)were solubilized for 1 hat 0°C in 1 ml of 10 mM Tris-HCI buffer pH 7.4,0.15 MNaCl containing 1% NP-40, 20 mM iodoacetamide, 2 mM phenylmethylsulph-onyl fluoride (PMSF) and 1 mg/mi bovine serum albumin (BSA). Debris wasremoved by centrifuging in a microfuge for 10 min. Supernatants were pre-clearedtwice by incubating for 1 h at 4°C with 100 pl of a 10% (v/v) suspension offixed Staphylococcus aureus Cowan 1 strain organisms followed by centrifuga-tion and were then incubated for 3 h at 4°C with UCH-TI-Sepharose (50 plof beads containing 70 pg of UCH-TI antibody). The beads were washed with1 ml of the above lysis buffer supplemented to 1 M NaCI and then with 1 miof 0.1% NP-40 in 10 mM Tris-HCI, pH7.4. Boundprotein was eluted by boilingfor 5 min in 50 pl of 100 mM Tris-HCI, pH 7.4, containing 1% SDS and thebeads removed by centrifuging. 03-mercaptoethanol was added to 1% and the solu-tion diluted with 450 pl of 100 mM Na phosphate buffer, pH 6.1, containing50 mM EDTA, 1% NP-40 and 1% /3-mercaptoethanol, prior to adding 1.6 unitsof Endo-F and 1 mM PMSF. The mixture was incubated at 37°C for a totalof 36 h with two further additions of Endo-F and PMSF. Protein was recoveredby precipitation with 10% trichloracetic acid, washed with acetone and analysedby SDS-PAGE.

Electrophoresis and Western blottingThe purified T3 polypeptides were analysed by SDS-PAGE according to Laemmli(1970), using 15% slab gels. The gels were stained using the silver method ofMorrissey (1981). For Western blotting, polypeptides were electrophoreticallytransferred at 4°C from SDS gels to nitrocellulose (Schleicher and Shuell) at 300 Vfor 2 h in a Tris glycine-methanol buffer (Towbin et al., 1979). Filters wereblocked overnight at 4°C with 2% (w/v) BSA, 5% (w/v) Marvel (Cadburys),0.05% Tween 20 and 0.02% Na azide in PBS. Transferred polypeptides wererevealed by incubating filter strips overnight at 4°C with antiserum diluted 1:100in PBS containing 2% (w/v) BSA and 0.02% Na azide. After washing three timesin PBS, 0.05% Tween 20, the strips were incubated for 1.5 h with a 1:100 di-lution of biotinylated rabbit anti-rat IgG (Vector Laboratories Inc.) and washed

1806


as above. The strips were finally incubated for 1 h with avidin-biotinylated horse-

radish peroxidase conjugate (Vector Laboratories) and after washing bound anti-

body was visualized with diaminobenzidine (0.5 mg/ml) and hydrogen peroxide(0.06%) in 10 mM Tris-HCI buffer, pH 7.4.

Construction of cDNA libraryCytoplasmic RNA was prepared from human T leukaemic J6 cells (a subline

of Jurkat selected by Dr P.Beverley for high T3 antigen expression) as described

by Favaloro et al. (1980). After two rounds of selection on oligo(dT) cellulose,a cDNA library was constructed from 10 itg of poly(A)+RNA (H.Kataoka, M.

Collins, M.J.Owen and T.Lindahl, unpublished). First-strand cDNA was syn-thesized using AMV-reverse transcriptase and double-stranded (ds) cDNA was

prepared from the DNA hybrid using RNase H and the DNA polymerase I holo-

enzyme, as described by Gubler and Hoffman (1983). EcoRI linkers were ligatedonto blunt-ended DNA and > 500 nucleotides long ds DNA selected using Seph-arose 6B column chromatography was ligated into Xgtl 1 arms which had been

treated with bacterial alkaline phosphatase. After packaging, a total library of

2 x 106 recombinant phage was amplified in the host strain Y1088 prior to screen-

ing (Young and Davis, 1983).

Screening of cDNA libraryThe library was plated on E. coli strain Y1090, as described previously (Youngand Davis, 1985). Plates (20 x 20 cm; each containing approximately 300 000

plaques) were incubated for 3-4 h at 42°C, covered with a nitrocellulose filter

previously soaked in 10 mM isopropyl thio-3-D-galactopyranoside containing50 Agg/ml ampicillin and the fusion protein induced for 12 hat 37°C. Filters were

washed briefly in PBS, 0.05% Tween 20 and blocked overnight at 4°C in PBS

containing 3% (w/v) BSA, 1 jig/ml soybean trypsin inhibitor and 0.02% Na

azide. Rat antisera used for screening were pre-adsorbed by incubating with a

Xgtl 1 lysogen (Y1090) lysate bound to Sepharose CL-4B. Nitrocellulose filters

were incubated overnight at 4°C in heat-sealed plastic bags with a 1:160 dilutionof antiserum in PBS containing 0.1% BSA and 0.02% Na azide. Subsequent stepsto identify immunoreactive clones were as described above for Western blotting.Positive plaques were picked and purified by three further rounds of screening.

In order to isolate a full length clone, the Xgtl 1 library was plated on the E.

coli strain Y1088 and plaque lifts were made onto biodyne filters (Pall Ultrafme

Filtration Corp., NY) as described by Maniatis et al. (1982). Filters (20 x 20 cm;each containing 100 000 plaques) were screened according to the standard pro-cedure (Maniatis et al., 1982) with the EcoRI-AccI fragment of pJ6T3-y-1 (Figure2A) labelled by random hexanucleotide priming (Feinberg and Vogelstein, 1983)to a specific activity of 1.5 x 109 c.p.m./ g. The pre-hybridization and hybridiza-tion were performed at 42°C in solutions containing 50% (v/v) formamide and

5 x SSC; 10% (w/v) dextran sulphate and 1-4 x 106 c.p.m./ml of probe were

incorporated during hybridization. Filters were washed as described by the

manufacturers and given two final washes in 0.25 x SSC, 0.1% SDS at 500C

for 15 min. Positive plaques were detected by autoradiography at -700C usingKodak X-AR5 X-ray film with Dupont intensifying screens (Cronex Lightning-Plus) and were purified by two additional rounds of screening.

DNA sequencingThe insert from cDNA clone pJ6T3'y-1 was subcloned into the EcoRI site of

pUC13, then into the EcoRI site of the M13 vector mpl8 prior to sequencing.Purified insert was also digested with AccI and XbaI before subcloning into AccIl

EcoRI- or XbaI/EcoRI-digested mpl8, respectively. The pJ6T3y-2, -3 cDNA

clones were initially identified by subcloning directly into the EcoRI site of mpl8and sequencing. Subsequently their inserts were amplified in pU13, digested with

Sau3A and subcloned into BamnHI/EcoRl-digested mpl8. Single-stnded templateswere prepared, analysed by agarose gel electrophoresis and sequenced using the

dideoxy chain termination procedure (Sanger et al., 1977). The entire sequenceof the y polypeptide chain was determined on both strands.

Northern blottingPoly(A)+ RNA, prepared as described by Favaloro et al. (1980), was denatured,electrophoresed on 1.1% agarose gels and transferred to nitrocellulose filters

(Thomas, 1980). Filters were hybridized wtih the insert of clone pJ6T3-y-1 or

its EcoRl-AccI fragment (Figure 2A), or with a probe for the T3 6 chain

(pPGBC-9; donated by Dr C.Terhorst; Van den Elsen et al., 1984). Purified

inserts (specific activity - 109 c.p.m./4ug) were labelled by random priming with

a mixed sequence hexadeoxynucleotide (P.L. - Pharmacia) and filling in usingthe Klenow enzyme and [a-32P]dCTP (Feinberg and Vogelstein, 1983). After

hybridization the filter was washed at a stringency of 0.1 x SSC at 500C for

30 min.

Computer analysisParameters used in the search of the National Biomedical Research Foundation

database were k-tuple length = 3, window size = 20, gap penalty = 2. The

Kyte and Doolittle (1982) hydropathy profile was plotted using an average win-

dow length of 6.

AcknowledgementsWe express our gratitude to Dr C.Terhorst for the 6chain probe pPGBC-9, M.Crompton for the Xgtl 1, lysogen-Sepharose beads, M.Hexham for computingthe hydropathy profiles, Drs W.Bodmer, J.Trowsdale and J.Williams, for helpfuldiscussions, Kim Richardson for typing the manuscript, and the hospitals in theGreater London area for providing the human tonsils.

ReferencesAdams,G.A. and Rose,J.K. (1985) Mol. Cell Biol., 5, 1442-1228.Allen,G. (1981) In Work,T.S. and Burdon,R.H. (eds), Laboratory Techniques

in Biochemistry and Molecular Biology. Vol. 9, Elsevier Biomedical, Amster-dam, pp. 31-36, pp. 184-185.

Amzel,L.M. and Poljak,R.J. (1979) Annu. Rev. Biochem., 48, 961-997.Beverley,P.C.L. and Callard,R.E. (1981) Eur. J. Immunol., 11, 329-334.Borst,J., Prendiville,M-A. and Terhorst,C. (1982) J. Immunol., 128, 1560-1565.Borst,J., Prendiville,M-A. and Terhorst,C. (1983a) Eur. J. Immunol., 13, 576-

580.Borst,J., Alexander,S., Elder,J. and Terhorst,C. (1983b) J. Biol. Chem., 258,

5135-5141.Borst,J., Coligan,J.E., Oettgen,H., Pessano,S., Malin,R. and Terhorst,C. (1984)

Nature, 312, 455 -458.Brenner,M.B., Trowbridge,I.S. and Strominger,J.L. (1985) Cell, 40, 183-190.Cantrell,D.A., Davies,A.A. and Crumpton,M.J. (1985) Proc. Natl. Acad. Sci.

USA, 82, 8158-8162.Chou,P.Y. and Fasman,G.D. (1978) Annu. Rev. Biochem., 47, 251-276.Collins,M.K.L., Tanigawa,G., Kissonerghis,A-M., Ritter,M., Price,K.M., Tone-

gawa,S. and Owen,M.J. (1985) Proc. Natl. Acad. Sci. USA, 82, 4503-4507.Davies,A.A., Cantrell,D.A. and Cnrmpton,M.J. (1985) Biosci. Rep., 5, 867-876.Ey,P.L., Prowse,S.J. and Jenkin,C.R. (1978) Immunochemistry, 15,429-436.Farnham,P.J. and Schimke,R.T. (1986) Mol. Cell Biol., 6, 365-371.Favaloro,J., Treisman,R. and Kamen,R. (1980) Methods Enzymol., 65, 718-720.Feinberg,A.P. and Vogelstein,B. (1983) Anal. Biochem., 132, 6-13.Furley,A.J., Weilbaecher,K., Dhaliwal,H.S., Ford,A.M., Chan,L.C., Molgaard,

H.V., Toyanaga,B., Mak,T., Van den Elsen,P., Gold,D., Terhorst,C. andGreaves,M.F. (1986) Cell, in press.

Gold,D.P., Puck,J.M., Pettey,C.L., Cho,M., Coligan,J., Woody,J.N. andTerhorst,C. (1986) Nature, 321, 431-434.

Gubler,U. and Hoffman,B.J. (1983) Gene, 25, 263-269.Hood,L., Kronenberg,M. and Hunkapiller,T. (1985) Cell, 40, 225-229.Jeffreys,A.J. (1981) In Williamson,R. (ed.), Genetic Engineering. Vol. 2,

Academic Press, London, pp. 1-48.Kanellopoulos,J.M., Wigglesworth,N.M., Owen,M.J. and Crumpton,M.J. (1983)EMBOJ., 2, 1807-1814.

Kelly,C., Totty,N.F., Waterfield,M.D. and Crumpton,M.J. (1983) Biochem. Int.,6, 535-544.

Kozak,M. (1983) Microbiol. Rev., 47, 1-45.Kumel,G., Daus,H. and Mauch,H. (1979) J. Chromatogr., 172, 221-226.Kyte,J. and Doolittle,R.F. (1982) J. Mol. Biol., 157, 105-132.Laemmli,U.K. (1970) Nature, 227, 680-685.Link,M.P., Stewart,S.J., Warnke,R.A. and Levy,R. (1985) J. Clin. Invest., 76,248-253.

Lomedico,P., Rosenthal,N., Efstratiadis,A., Gilbert,W., Kolodner,R. andTizard,R. (1979) Cell, 18, 545-558.

Maniatis,T., Fritsch,E.F. and Sambrook,J. (1982) Molecular Cloning. A Labora-tory Manual. Cold Spring Harbor Laboratory Press, NY.

Meuer,S.C., Acuto,O., Hercend,T., Schlossman,S.F. and Reinherz,E.L. (1984)Annu. Rev. Immunol., 2, 23-50.

Miyata,T., Yasunaga,T. and Nishida,T. (1980) Proc. Natl. Acad. Sci. USA, 77,7328-7332.

Morrissey,J.H. (1981) Anal. Biochem., 117, 307-310.O'Brian,C.A., Lawrence,D.S., Kaiser,E.T. and Weinstein,I.B. (1984) Biochem.

Biophys. Res. Commun., 124, 26-302.Owen,M.J. and Collins,K.L. (1985) Immunol. Lett., 9, 175-183.Owen,M.J., Kissonerghis,A-M. and Lodish,H.F. (1980) J. Biol. Chem., 255,9678-9684.

Perler,F., Efstratiadis,A., Lomedico,P., Gilbert,W., Kolodner,R. and Dodgson,J.(1980) Cell, 20, 555-566.

Proudfoot,N.J. and Brownlee,G.G. (1976) Nature, 263, 211-214.Raulet,D.H., Garman,R.D., Saito,H. and Tonegawa,S. (1985) Nature, 314, 103-

107.Samelson,L.E., Harford,J.B. and Klausner,R.D. (1985a) Cell, 43, 223-231.Samelson,L.E., Lindsten,T., Fowlkes,B.J., Van den Elsen,P., Terhorst,C., Davis,M.M., Germain,R.N. and Schwartz,R.H. (1985b) Nature, 315, 765-768.

Sanger,F., Nicklen,S. and Coulson,A.R. (1977) Proc. Natl. Acad. Sci. USA,74, 5463-5467.

Terhorst,C., Georgopoulos,K., Gold,D., Oettgen,H., Pettey,C., Ucker,D. and

1807


Van den Elsen,P. (1986) In Feldman,M. and McMichael,A. (eds), Regulationof Immune Gene Expression. Humana Press, New York, in press.

Thomas,P.S. (1980) Proc. Natl. Acad. Sci. USA, 77, 5201-5206.Towbin,H.,. Staehelin,T. and Gordon,J. (1979) Proc. Natl. Acad. Sci. USA, 76,4350-4354.

Tsoukas,C.D., Valentine,M., Lotz,M., Vaughan,J.H. and Carson,D.A. (1984)Immunol. Today, 5, 311-313.

Tunnacliffe,A., Kefford,R., Milstein,C., Forster,A. and Rabbitts,T.H. (1985)Proc. Natl. Acad. Sci. USA, 82, 5068-5072.

Van den Elsen,P., Shepley,B-A., Borst,J., Coligan,J.E., Markham,A.F., Orkin,S.and Terhorst,C. (1984) Nature, 312, 413 -418.

Van den Elsen,P., Shepley,B-A., Cho,M. and Terhorst,C. (1985) Nature, 314,542-544.

Von Heijne,G. (1983) Eur. J. Biochem., 133, 17-21.Weber,K., Pringle,J.R. and Osborn,M. (1972) Methods Enzymol., 26, 3-27.Wilbur,W.J. and Lipman,D.J. (1983) Proc. Natl. Acad. Sci. USA, 80, 726-730.Young,R.A. and Davis,R.W. (1983) Science, 222, 778-782.Young,R.A. and Davis,R.W. (1985) In Setlow,J. and Hollaender,A. (eds), Genetic

Engineering. Vol. 7, Plenum Press, New York, in press.

Received on 21 May 1986

Note added in proofThe T3 -y chain purified from human tonsil (Figure IA, lane 4) was cleavedbetween residues 58 and 59 (asp-pro) with formic acid, and the unseparatedpolypeptide fragments were sequenced on an Applied Biosystems machine, model470A. A single contiguous sequence of 22 amino acid residues (excluding theasparagine N-linked glycosylation site at cycle 12) was obtained which agreedwith the protein sequence deduced from the T3 -y cDNA (Figure 2B) (Krissansenet al., in preparation).

1808

receptor complex deduced from cdna sequences: evolution of the

Documents