- Email: [email protected]
T8 is homologous to immunoglobulin and T cell receptor variable regions
Cell, Vol. 40, 591-597,
March
1985, Copyright
0092-8674/85/030591-07
0 1985 by MIT
The T Cell Differentiation Antigen Is Homologous to lmmunoglobulin T Cell Receptor Variable Regions Vikas P. Sukhatme,‘t Kurt C. Sizer,’ Amy C. Vollmer: Tim Hunkapiller,* and Jane R. Parries” * Division of Immunology Department of Medicine Stanford University Medical Center Stanford, California 94305 *Division of Biology California Institute of Technology Pasadena, California 91125
Summary Leu-2/T8 is a cell surface glycoprotein expressed by most cytotoxic and suppressor T lymphocytes. Its expression on T cells correlates best with recognition of class I major histocompatibility complex antigens, and it has been postulated to be a receptor for these proteins. We have determined the complete primary structure of Leu-2/T8 from the nucleotide sequence of its cDNA. The protein contains a classical signal peptide, two external domains, a hydrophobic transmembrane region, and a cytoplasmic tail. The N-terminal domain of the protein has striking homology to variable regions of immunoglobulins and the T cell recep tor. The membrane-proximal domain appears to be a hinge-like region similar to that of immunoglobulin heavy chains. The superfamily of immunologically important surface molecules can now be extended to include Leu-2/T8. Introduction The Leu-2IT8 T cell differentiation antigen is a cell surface glycoprotein expressed by distinct subsets of human T lymphocytes. It is the analog and presumed homolog of mouse Lyt-2,3 (Reinherz and Schlossman, 1980; Ledbetter et al., 1981). Traditionally, these differentiation antigens have been considered markers of T cell function: cytotoxic and suppressor cells expressing Leu-2iT8 (Lyt-2,3 in mouse) and helper/inducer cells expressing the alternative antigen Leu-3/T4 (L3T4 in mouse). However, recent data from many laboratories have indicated that expression of Leu-2/T8 or Leu-3/T4 correlates not so much with function as with recognition by T cells of class I (HLAA, B, C) or class II (HLA-DP, DQ, DR) major histocompatibility complex (MHC) antigens, respectively (Engleman et al., 1981; Krensky et al., 1982a, 1982b; Meuer et al., 1982). Leu-2/T8 is thought to be involved in the recognition by cytotoxic T cells of their targets. Monoclonal antibodies directed against Leu-2IT8 inhibit cytotoxicity by most Leu2/T8+ killer T cells (Meuer et al., 1982; Spits et al., 1982; Reinherz et al., 1983). This block is at the step of cont Present Chicago,
address: Department Illinois 60637.
of Medicine,
University
of Chicago,
$02.00/O
Leu=2/T8 and
jugate formation between the cytotoxic cell and the target cell, rather than at the later killing steps (Tsoukas et al., 1982; Landegren et al., 1982; Moretta et al., 1984). It has been suggested that the function of Leu-2/l8 (or Lyt-2,3) is to increase the avidity of, or stabilize the interaction between, the T cell and its target (Meuer et al., 1982; MacDonald et al., 1982; Moretta et al., 1984), perhaps by binding to nonpolymorphic regions of class I MHC molecules (Krensky et al., 1982a; Meuer et al., 1982; Ball and Stastny, 1982; Spits et al., 1982; Biddison et al., 1982; Engleman et al., 1983). On peripheral T cells the Leu-2/T8 protein consists of dimers and higher multimers of a 32-34 kd glycosylated subunit, linked together by disulfide bridges (Ledbetter et al., 1981; Snow et al., 1983). On thymocytes, in addition to homodimers of this polypeptide, a46 kd subunit is disulfide linked to the smaller chain in tetramers and higher multimers (Ledbetter et al., 1981; Snow and Terhorst, 1983). This larger subunit does not bear the T8 determinant, is unrelated to the smaller subunit by peptide mapping, and is presumably not required for the function of the molecule on mature T cells (Snow and Terhorst, 1983). Snow et al. (1984) have recently sequenced 25 N-terminal residues of Leu-2Il8 and found no significant homology to any other proteins. Determination of the complete primary structure of Leu-P/T8 would be of enormous value not only for evolutionary comparisons but also for further studies of the function of the protein in T cell recognition. To this end we have recently isolated cDNA clones that encode Leu-2/T8 (Kavathas et al., 1984). We have now determined the complete amino acid sequence of the Leu-2/T8 protein from the nucleotide sequence of a cDNA clone. We find that the protein has a classical signal peptide, two external domains, a transmembrane region, and an intracytoplasmic tail. One of the external domains has striking sequence homology to variable regions of other immunological recognition molecules. The results suggest a common evolutionary origin of Leu-2/T8, immunoglobulin, and T cell receptor genes. Results Nucleotide Sequence of Leu-2/T8 cDNA The nucleotide sequence of LeuQ/T8 was determined from a single cDNA clone, pLBM, containing a 2 kb insert, although large portions were confirmed on pdditional clones. Figure 1 shows a restriction map of the insert of clone pL2-M, indicating the enzyme sites used for subcloning into Ml3 vectors. The nucleotide sequence of this clone is presented in Figure 2. There are two long open reading frames of similar sizes at the 5’ end of this clone. One of these extends from the beginning of the clone through nucleotide 632. There are no methionines in this frame, so if it represents a translated protein, our clone is missing the initiation codon. The second open reading frame begins at nucleotide 1 and extends through nucleotide 792. This latter frame encodes Leu-2IT8, as deter-
Cell 592
AS
AR
s
R
H
HR
H
H
HR
II
I
I
II
mined by comparison with the N-terminal protein sequence (Snow et al., 1984). The first ATG in this frame is at nucleotide 88. We believe this to be the initiation codon. Accordingly, clone pLPM contains 115 bp of 5’ untranslated region (not all shown), 705 bp of protein coding sequence (235 amino acids), and 1182 bp of 3’untranslated region. Neither this clone nor another extending 85 bp more S’contains a poly(A) tail. The mRNA represented by these clones is approximately 2.5 kb in length (Kavathas et al., 1984). Our clones span 2.1 kb, so we are missing approximately 400 bp. We presume that most of this is 3’ untranslated region and poly(A) tail.
R
3
5 WJT 9
TM c
3’ UT H 100 L?J
Figure
1. Restriction
Map of Leu-P/T8
cDNA
Clone
pLP-M
A map of the 2 kb Eco RI insert of cDNA clone pLP-M is shown. The enzyme sites indicated are those used for subcloning into Ml3 vectors for DNA sequencing. A, Ava I; S, Sau 3A; R, Rsa I; H, Hinf I. The artificial Eco RI sites at the 5’ and 3’ ends are not labeled. The domains of the protein encoded by various segments of the cDNA are demarcated. Open bar: 5’and 3’ untranslated (UT) regions. Hatched: signal peptide (SP). Shaded: external protein domains. Speckled: transmembrane region (TM). Crosshatched: cytoplasmic tail (C).
Lys Ala Ala Glu Gly Leu Asp Thr Gin Arg Phe Ser Gly Lys Arg Leu Gly ASP Th,PAG GCG GCC GAG GGG CTG GAC ACC CAG CGG TTC TCG GGC AAG AGG TTG GGG GAC ICC
Phe TTC
"al Le" Thr Le" Ser GTC CTC ACC CTG AGC
“31 GTG
Pro CCG
“al GTC
Phe TTC
Leu CTG
Pro CCA
Ala GcG
Lys anG
Pro CCC
Thr *CC
Thr ACG
ThrACG
Pro CCA
Ala GCG
Pro CCG
nrg CGA
Pro CCP
Pro CC*
Thr A.cA
Pro CCG
Ala GCG
Pro CCC
Thr ACC
11e ATC
Ala GCG
ser TCG
Gl” C4G
Pro CCC
Leu CTG
ser TCC
Leu CTG
arg CGC
Pro CC4
Gl” GAG
Ala GCG
cys TGC
Al-g CGG
Pro CCA
Ala GCG
Ala GCG
Gly GGG
GlY GGC
Ala GCA
“al GIG
HIS CAC
Thr ACG
.arg AGG
Gly GGG
Le” CTG
ASP GAC
ser Leu Ser Al.3 brg Tyr vai AGC
Figure
CTT
TCG
2. Nucleotide
GCG
AGP
Sequence
TAC
-
GTC
and Deduced
3’UT
Trm 7114
CCCTGTGCAACAGCCACTACATTACTTCAAACTGAGATCCTTCCTTTTGAGGGAGCAAGTCCTTCCCT
Amino
Acid Sequence
of the Insert
of Leu-P/T8
cDNA
Clone
pL2-M
Horizontal arrows denote the beginning of the predicted protein domains. Open circles are placed above the two cysteines in the V-like domain that are conserved in immunoglobulin and T cell receptor V regions. A closed circle is above the potential site for N-linked glycosylation. A horizontal bar is over the 21 amino acid sequence homologous to the mouse IgA hinge. Numbers in the left and right margins, respectively, refer to the first and last nucleotides or amino acids in each line. Abbreviations: L, leader (signal peptide); V, variable region-like domain; H, hinge; TM, transmembrane region; Cy, cytoplasmic tail; 3’UT, 3’ untranslated region.
Leu-2iT8 593
cDNA
Sequence
Figure 3. Comparison Sequences
of the Amino Acid Sequence
of the N-Terminal
Domain
of Leu-2/T8
with lmmunoglobulin
and T Cell Receptor
Variable
Region
The Leu-P/T8 sequence is the first 96 amino acids of the mature protein. Similarities to Leu-2/T8 are denoted by a dot (.); gaps in the alignments are indicated by blank spaces. lmmunoglobulin lambda, kappa, and heavy chain V regions are denoted VI, V,, and Vn. T cell receptor (I and fl chain V regions are denoted V, and V,+ The sequence alignments were derived by comparing all members of a given class of V regions (e.g. kappa, lambda) against one another and Leu-P/T8 simultaneously to achieve the best alignment for a group of sequences, with attention paid to alignment of certain key conserved residues. All of the immunoglobulin V sequences in the Dayhoff bank were utilized, as were all published and some unpublished T cell receptor sequences (L. E. Hood, personal communication). The sequences were from the following sources: human Ha VI-l (Shinoda et al., 1970); human Meg VI-II (Fett and Deutsch, 1974); mouse U61 V, (Vrana et al., 1979); human Len V, (Schneider and Hilschmann, 1975); mouse Vn (Capra and Nisonoff, 1979); rat IR2 Vn (Hellman et al., 1982); mouse TTll V, (Chien et al., 1984); human TY35 V, (Yanagi et al., 1984).
Predicted Amino Acid Sequence of Leu-2lT8 The predicted amino acid sequence of Leu-2IT8 is shown in Figure 2. The protein begins with a 21 amino acid hydrophobic leader sequence, presumably representing a signal peptide that is cleaved off as the protein passes into the endoplasmic reticulum. The indicated start of the mature processed protein is based upon comparison with the published N-terminal protein sequence (Snow et al., 1984). Our N-terminal sequence differs from the latter in only two places. We predict a glutamine at residue 2, while the protein sequence indicates glutamic acid at this position. As the cDNA was sequenced multiple times on both strands, we believe that eitherthis residue became deamidated during the protein purification or there was an error made by reverse transcriptase during the cDNA cloning. Snow et al. (1984) placed a gap of one amino acid between the cysteine at residue 22 and the two leucines at what were considered residues 24 and 25. We predict two amino acids in this region: glutamine at residue 23 and valine at 24. This discrepancy is likely to be a protein sequencing error, since it is at the very end of the determined sequence. However, it is possible that either or both of these discrepancies could be the result of polymorphism. The mature Leu-2/T8 protein consists of 214 amino acids. A 24 amino acid hydrophobic segment begins at position 162. We believe this to represent a transmembrane region. The following 29 amino acid C-terminal segment (residues 186-214) is highly charged (ten basic residues and only one acidic residue) and is presumably an intracytoplasmic tail. Our sequence predicts an ungiycosylated mature protein chain of 23,554 daltons as compared to the described 32,000-34,000 dalton glycosylated subunit size (Ledbetter et al., 1981; Snow and Terhorst, 1983). However, when Snow et al. (1984) treated the isolated protein with trifluoromethane sulfonic acid (TFMS) to remove both N-linked and O-linked carbohydrate, the size was reduced only 1.5-2 kd. The reasons for this discrepancy are unclear, but most likely relate to incomplete cleavage of carbohydrates from the protein with TFMS. Although our sequence predicts one potential site for N-linked glycosylation (Arg-X-Thr or Arg-X-Ser), at position 28, it may not be used, since experiments with endoglycosidase F suggest that there is no N-linked glycosylation (Snow and Terhorst, 1983).
Leu-2/T8 Is Homologous to lmmunoglobulin and T Cell Receptor Variable Regions We searched the Dayhoff protein sequence data bank for homologies of the predicted Leu-2/T8 protein sequence to other proteins. The greatest amount of homology found was to immunoglobulin light chain variable (V) regions, with less but still significant homology to immunoglobulin heavy chain V regions (V,) and V regions of the a and fi chains of the T cell receptor. Comparisons of the Leu-2/T8 protein sequence with examples of immunoglobulin and T cell receptor V region sequences are illustrated in Figure 3. The segment of Leu9/T8 that is homologous to a V region is at the N terminus and extends to amino acid 96. For the V regions the homology to Leu-2/T8 stops just before the D (diversity) segment (if one is present) or the J (joining) segment. With the alignment shown the homology is on the order of 30%-350/o to lambda and kappa V regions, 200/o-22% to VH, and 24% to T cell receptor (I and /I chain V regions. The homology increases substantially (e.g. up to 56% for lambda or 58% for kappa) if one includes conservative amino acid substitutions. As shown in Figure 3, there are seven amino acids that are conserved in all of these sequences. These include the two cysteine residues involved in the classical intrachain disulfide loop of V regions (positions 22 and 94 for Leu-2/T8) and the tryptophan at position 35, which is thought to be important for the proper folding of immunoglobulin domains. Leu-2/T8 contains most of the residues of light chain V regions that are used for association with heavy chains (Poljak et al., 1975) as well as many of the V region contact residues in light chain dimers (Davies et al., 1975). We therefore predict that the two V-like domains in Leu2/T8 dimers are associated with one another (noncovalently), and probably form a single binding site for a presumed ligand. As might be expected, the homology of Leu9IT8 to V regions is predominantly to framework rather than hypervariable regions. The level of homology does not vary substantially across mammalian species. In contrast to the findings for the N-terminal external domain, the 65 amino acid membrane-proximal domain of Leu-2IT8 bears no strong resemblance to constant region domains of immunoglobulins or the T cell receptor and is not homologous to other known proteins. However, a stretch of 21 amino acids in the center of this domain (residues 120-140) is homologous to the hinge region of
Cell 594
Leu-Z/T8
Hinge
Mouse
CHa
Hinge
Figure
4. Homology
PAPRPPTPAPTIASQP * * ** PTPPPPITIkC
of Leu-2/T8
Hinge to Mouse
LSL RP ** *** ** QPSLSLQRP IgA Hinge
Amino acids 120-140 of Leu-2/T8 are shown. Amino acids 220-240 of the mouse a heavy chain (CH,) are below. The entire CH, hinge spans amino acids 218-240. Homologies are denoted by an asterisk (‘), and gaps, by blank spaces. The CH, hinge sequence is from Auffray et al. (1981).
the mouse IgA heavy chain (Figure 4). Although we doubt an evolutionary significance of this homology, evidence discussed below suggests that the domain containing this sequence is a hinge. The remaining protein domains appear unique to Leu9/T8. Structural Predictions We have analyzed the Leu-2/T8 sequence using a formula similar to that of Kyte and Doolittle (1982) for determining the degree of hydrophobicity or hydrophilicity of segments of a protein. Such analyses allow one to predict which portions of a protein are exposed, interior, or embedded in a membrane. A hydrophobicity plot of the Leu-2Il8 sequence is shown in Figure 5. The putative signal peptide (residues -21 to -1) and transmembrane domain (residues 162 to 185) are both demonstrated to be highly hydrophobic, while the presumed intracytoplasmic tail (residues 186 to 214) is very hydrophilic. The V-like N-terminal external domain has alternating stretches of hydrophobic and hydrophilic residues as seen in globular proteins (Kyte and Doolittle, 1982). Furthermore, the plot for this domain is virtually superimposable on those for kappa, lambda, and heavy chain V regions, except for an extra hydrophilic segment in Leu-2/T8 between amino acids 6 and 15 (data not shown). These results suggest a great deal of conservation of structure among these protein domains. We further analyzed the Leu-2Il8 protein sequence according to the parameters of Chou and Fasman (1978) and find that the predicted p-sheet structure of the N-terminal domain is the same as that for light chain V regions. In contrast, the membrane-proximal external domain does not fit into any structure. Therefore this domain, which was shown to contain a sequence homologous to the mouse IgA hinge, is itself predicted to be a hinge region. The Leu-2/T8 Gene Does Not Rearrange in Cells That Express the Protein Variable regions of immunoglobulins and the T cell receptor are not expressed as functional receptor molecules until their genes rearrange during the development of 6 cells or T cells, respectively, and become juxtaposed to either a J or D and J segments just upstream from a constant region gene. Since these proteins are homologous to Leu2/T8, it was important lo determine whether the Leu-2/l8 gene also rearranges in T cells that express this protein. We have therefore compared Southern blots of DNA from human placenta (which does not express Leu-2/T8 protein or mRNA) and from a Leu-2/T8-expressing thymoma cell line, JM, hybridized to the insert of cDNA clone pLBM. As
illustrated in Figure 6, no difference could be detected between the two DNAs with any of the four enzymes used (Barn HI, Eco RI, Apa I, and Eco RV). There is also no difference seen with Hind Ill (data not shown). Since the probe contains the entire coding sequence as well as most of the noncoding portions, we conclude that no major rearrangements are required for expression of this gene. Although we cannot exclude very small rearrangements with either little or no deletion of DNA, immunoglobulin and T cell receptor type rearrangements would have been easily detectable. The data shown in Figure 6 also suggest that Leu9IT8 is encoded by a single gene. The Southern blots show one to three hybridizing bands, depending upon the enzyme. Differential hybridization to 5’ and 3’ probes indicates that when multiple bands are present, they represent portions of a single gene, and all of the hybridizing fragments can be accounted for on a single genomic clone containing a 14 kb insert (data not shown). Of course, it cannot be excluded that the entire Leu-2/T8 gene and large amounts of flanking sequence are repeated in the genome with little or no change. Using low stringency wash conditions (37% 1 x SSC) we have detected one or two extra bands on genomic blots with DNA digested with Barn HI, but not reproducibly with other enzymes. We are currently determining whether these are truly distantly cross-hybridizing genes or aberrant digestion products of the LeuQTTB gene with Barn HI. It is clear, however, that Leu-2TT8 is not a member of a large multigene family. Discussion The structure of Leu-2/T8 is typical of a cell surface protein. It contains a hydrophobic signal peptide, two predicted external protein domains, a transmembrane hydrophobic domain, and an extremely basic intracytoplasmic tail. The protein has a total of nine cysteine residues: three in the N-terminal V-like domain and two each in the membrane-proximal (hinge), the transmembrane, and the cytoplasmic domains. Biochemical data indicate that the three in the V-like domain are not used for interchain disulfide bridging (Snow and Terhorst, 1983). Since our alignment with V regions shows conservation of the cysteines at positions 22 and 94 of Leu9/T8, we predict that these two cysteines are used for an intrachain disulfide bridge as in immunoglobulins and the T cell receptor. The function of the cysteine at position 33 is unknown. Subunit joining could occur between membrane-proximal domains, within the membrane, or within the cytoplasm. Several of these six cysteine residues may be used for interchain bridging, since the molecule often forms large multimers. Perhaps the most interesting finding concerning the structure of Leu-2/T8 is the homology of the N-terminal external domain to variable regions. Many members of the “superfamily” (Jensenius and Williams, 1982; Williams, 1984) of immunologically important surface proteins are much more homologous to constant regions than to variable regions (e.g. class I and class II MHC antigens, B2microglobulin). In contrast to these proteins, Leu-2IT8 is
Leu-2/T8 595
cDNA
Sequence
P
‘: Figure
5. Hydrophobicity
Plot of Leu-2/T8
Protein
1
2
k
P
Sequence
Data are plotted according to Kyte and Doolittle (1982). The plotted values are averages over groups of seven contiguous residues. Positive values indicate hydrophobic, and negative values indicate hydrophilic, regionsof the protein. The amino acid sequence is listed above the plot and numbered below. The domains of the protein are labeled as in the legend to Figure 2. 1
thought to be a receptor molecule, and so it is not surprising that it would borrow or share a structure that the immune system has used to such advantage in antigen recognition. It should be stressed, however, that there is no evidence for variability of this “V-like” domain. Two other members of the superfamily (besides immunoglobulins and T cell receptors) are thought to be related to variable regions: the mouse T cell marker Thy-l (Williams and Gagnon, 1982) and the poly(lg) receptor (Mostov et al., 1984). The homology of Leu9IR3 to these molecules is less striking, as is their homology to immunoglobulin V regions. Our sequence analyses indicate that Leu-2/T8 is about equidistant between its two closest relatives: kappa and lambda V regions. If one were to construct a genealogical tree of the immunoglobulin-related superfamily, LeuQTT8 would split off after the divergence of heavy and light chains, and presumably just before the split between kappa and lambda. Chromosomal mapping studies of Leu-2/T8 are intriguing in this regard. Lyt-2,3 has been known for many years to be tightly linked to the kappa locus on mouse chromosome 6 (Itakura et al., 1972; Gottlieb, 1974; Claflin et al., 1978; Gibson et al., 1978). We have recently shown that in the human system the Leu-2/T8 gene also maps extremely close to the kappa locus on chromosome 2 (Sukhatme et al., 1985). This close linkage between the genes encoding kappa and Leu-2/T8, which is maintained across at least two species, lends credence to the idea of a common evolutionary origin of these proteins. Although the N-terminal domain of Leu-2/T8 is remarkably similar to V regions, it differs from them in several important aspects. First, the gene encoding Leu-PfT8 shows no evidence of rearrangement in cells that express the protein. Second, the Leu-2/T8 gene has no close relatives in the human genome, in contrast to most, though not all, V genes. Finally, the nucleotide sequence encoding Leu2/T8 is strikingly distinct from the sequences encoding V regions in two aspects. As is true for most protein-coding sequences (Fickett, 1982), the G+C content of the region encoding the V-like domain (or the entire coding region) of Leu-2/T8 is high (66%). In contrast, the G+C content of immunoglobulin V regions is much closer to random (T. Hunkapiller and L. E. Hood, unpublished results). Furthermore, we have analyzed the Leu-2/T8 cDNA se-
2
3
23.1-
4
23.1-
23.1-m
6
5
7
-
1% c-
-
m
8
9
I)
9.4-
*
6.6-
.
9.4-
9.4-
6.6-
.
6.6-a@
-
44--o
c
4.4-
4.4-
2.3-
2.3-
2.0-
2.0-
2.32.0-
Figure
6. Southern
Blot Analysis
of Leu9/T8
Gene
Genomic DNA (8 rg) from a human thymoma cell line (JM) that expresses Leu-P/T8 (lanes 1, 3. 4, 6, 8) and from human placenta, which does not express Leu-2/T8 (lanes 2, 5, 7, 9) were digested with Barn HI (lanes 1, and 2) Hind III (lane 3) Eco RI (lanes 4 and 5) Apa I (lanes 6 and 7). or Eco RV (lanes 8 and 9) and electrophoresed on a 0.8% agarose gel. Southern blots of the gels were hybridized to the insert of cDNA clone pLP-M labeled with 32P by nick-translation. The migration distances of Hind Ill fragments of A phage DNA are indicated in the left margins in kilobase pairs.
quence with an algorithm similar to that of Fickett (1982) which can distinguish protein coding sequences from noncoding DNA. This test for coding sequences is based upon the nonrandom use of codons, resulting in repetition of particular nucleotides with a periodicity of 3 in coding, but not noncoding, DNA. The sequence encoding Leu2/T8 is clearly predicted as protein-coding, as is that of the V-like domain by itself. In contrast, immunoglobulin V region sequences (but not constant region sequences) are an exception to the rule in that they are most often predicted as noncoding (Fickett, 1982; T. Hunkapiller and L. E. Hood, unpublished results). These results imply that there is less constraint on the evolution of V genes than
Cell 596
on other genes and that, despite its homology to V regions, Leu9/T8 is under much stronger selective pressure to maintain its sequence. Experimental
Pmcedures
Isolation of cDNA Clone pL2-M The inserts of two Leu-2/T8 cDNA clones isolated previously by subtractive hybridization (Kavathas et al., 1984) were labeled with J2P by nick translation and used as probes to rescreen a previously described AgtlO cDNA library. This library was constructed from RNA of L cells that had been transfected with total human genomic DNA and selected for expression and amplification of the Leu-PiT8 gene (Kavathas et al., 1984). Positive clones were screened for insert size. The longest insert was 2 kb. This clone was grown up, and the Eco RI insert was isolated and subcloned into pBR322. The resulting plasmid is referred to as cDNA clone pLBM. DNA Sequencing The Eco RI insert of cDNA clone pLP-M was digested with Rsa I and inserted into EGO RI plus Sma I digested Ml3 vectors (Messing et al., 1981) mpl0 and mpll, or into Sma I digested mpll. The insert was also digested with Hinf I, the Soverhangs filled-in with the Klenow fragment of DNA polymerase I (Bethesda Research Laboratories), and the fragments subcloned into Sma I digested mpll. The largest Hinf I digestion product (the 5’ 1.1 kb Eco RI-Hinf I fragment) was isolated from the pLP-M insert. digested with Ava I and Sau 3A, filled-in with Klenow polymerase, and inserted into Sma I digested mpll. Finally, the entire Eco RI inserts of clone pL2-M and four independent, overlapping cDNA clones were cloned into the Eco RI site of mpll. Except for the first 48 bp of 5’ untranslated sequence, which were only sequenced on one strand, the nucleotide sequences of all these subclones were determined on both strands multiple times by the dideoxynucleoside triphosphate chain termination method of Sanger et al. (1977). The sequence shown was all derived from clone pLP-M, although the partial sequences of other independent cDNA clones agreed entirely except for a single base pair in the 3’untranslated region (a run of six As beginning at nucleotide 1895 in pLP-M and seven As in clone pL2-4). The Ml3 vectors were obtained from Amersham. Southern Blots and Hybridization Genomic DNA was isolated from the T cell lymphoma line JM, by the citric acid procedure (Hieter et al., 1981). Human placental DNA was a generous gift from Dr. Philip Leder (Harvard Medical School). DNA was digested with restriction enzymes (New England Biolabs) for 3 hr and electrophoresed on 0.8% agarose gels. Gels were blotted onto nitrocellulose by the procedure of Southern (1975). Blots were hybridized for 16 hr in 40% formamide, 20 mM Tris-HCI (pH 7.6), 4x SSC. lx Denhardt’s solution, 0.1% SDS, and 30 pglml denatured herring sperm DNA, to the insert from clone pL2-M labeled with 32P by nick translation to a specific activity of 200-400 cpm per picogram. The filters were washed three times at room temperature for 20 min each with 2x SSC, 0.05% SDS, and twice at 52% for 1 hr each with 0.1 x SSC, 0.05% SDS. Exposures with XAR-5 film (Kodak) were overnight at -70% with intensifying screens. Acknowledgments We thank Dr. Philip Leder for the human placental DNA, Dr. Rose Zamoyska for critical reading of the manuscript, Dr. Leroy Hood for helpful discussions and for use of his computer facilities, and Diane O’Neill for preparation of the manuscript. This work was supported by NIH grant Al 19512 to J. R. P V. P S. was a National Institutes of Health postdoctoral fellow. J. R. P. is the recipient of a John A. and George L. Hartford faculty fellowship award. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. Received
December
18, 1984; revised
January
10, 1985
References Auffray, C., Nageotte, R., Sikorav, J.-L., Heidmann, O., and Rougeon, F. (1981). Mouse immunoglobulin A: nucleotide sequence of the structural gene for the a heavy chain derived from cloned cDNAs. Gene 13, 365-374. Ball, E. J., and Stastny, I? (1982). Cell-mediated cyiotoxicity against HLA-D region products expressed in monocytes and B lymphocytes. IV. Characterization of effector cells using monoclonal antibodies against human T cell subsets. Immunogenetics 16, 157-169. Biddison, W. E., Rao, P E., Talle, M. A., Goldstein, G., and Shaw, S. (1982). Possible involvement of the T4 molecule in T cell recognition of class II HLA antigens: evidence from studies of cytotoxic T lymphocytes specific for SB antigens. J. Exp. Med. 156, 1065-1083. Capra, J. D., and Nisonoff, A. (1979). Structural studieson induced antibodies with defined idiotypic specificities. VII. The complete amino acid sequence of the heavy chain variable region of anti-p-azophenylarsonate antibodies from AN mice bearing a cross-reactive idiotype. J. Immunol. 123. 279-284. Chien, Y., Becker, D. M., Lindsten. T., Okamura, M., Cohen, D. I., and Davis, M. M. (1984). A third type of murine T-cell receptor gene. Nature 312, 31-35. Chou, P Y., and Fasman. G. D. (1978). Empirical conformation. Ann. Rev. Biochem. 47, 251-276.
predictions
of protein
Claflin, J. L., Taylor, B. A., Cherry, M., and Cubberly, M. (1978). Linkage in mice of genes controlling an immunoglobulin kappa-chain marker and the surface alloantigen Ly-3 on T lymphocytes. Immunogenetics 6, 379-387. Davies, D. R., Padlan, E. A., and Segal, D. M. (1975). Threedimensional structure of immunoglobulins. Ann. Rev. Biochem. 44, 639-667. Engleman, E. G., Benike, C. J., Grumet, F. C., and Evans, R. L. (1981). Activation of human T lymphocyte subsets: helper and suppressorlcytotoxic T cells recognize and respond to distinct histocompatibility antigens. J. Immunol. 127, 2124-2129. Engleman, E. G., Benike, C. J., Metzler, C., Gatenby, F? A., and Evans, R. L. (1983). Blocking of human T lymphocyte functions by anti-Leu-2 and anti-Leu-3 antibodies. J. Immunol. 130, 2623-2628. Fett, J. W., and Deutsch, H. F. (1974). Primary chain. Biochemistry 13, 4102-4114. Fickett, J. W. (1982). Recognition of protein quences. Nucl. Acids Res. 10, 5303-5318.
structure
coding
of the Meg I
regions
in DNA se-
Gibson, D. M., Taylor, B. A., and Cherry, M. (1978). Evidence for close linkage of a mouse light chain marker with the Ly-2,3 locus. J. Immunol. 121, 1585-1590. Gottlieb, f? (1974). Genetic correlation of a mouse light chain variable region marker with a thymocyte surface antigen. J. Exp. Med. 140. 1432-1437. Hellman, L., Petterson, U., Engstrom, A., Karlsson, T., and Bennich, H. (1982). Structure and evolution of the heavy chain from rat immunoglobulin E. Nucl. Acids Res. 10, 6041-6049. Hieter, i? A., Hollis, G. F.. Korsmeyer, S. J., Waldmann, T. A., and Leder, P (1981). Clustered arrangement of immunoglobulin A constant region genes in man. Nature 294, 536-540. Itakura, K., Hutton, J. J., Boyse, E. A., and Old, L. J. (1972). Genetic linkage relationships of loci specifying differentiation alloantigens in the mouse. Transplantation 13, 239-243. Jensenius, J. C., and Williams, A. F. (1982). The T lymphocyte receptor-paradigm lost. Nature 300, 583-588.
antigen
Kavathas, P., Sukhatme, V. I?. Herzenberg, L. A., and Parnes, J. R. (1984). Isolation of the gene coding for the human T lymphocyte antigen Leu-2 (T8) by gene transfer and cDNA subtraction. Proc. Natl. Acad. Sci. USA 81, 7688-7692. Krensky, A. M., Clayberger, C., Reiss, C. S., Strominger, J. L.. and Burakoff, S. J. (1982a). Specificity of OKT4 cytotoxic T lymphocyte clones. J. Immunol. 129, 2001-2003. Krensky,
A. M., Reiss, C. S., Mier, J. W., Strominger,
J. L., and Burakoff,
Leu-P/T8 597
cDNA
Sequence
S. J. (1982b). Long-term human cytolytic T-cell HLA-DR6 antigen are OKT4’. Proc. Nan. Acad. 2369. Kyte, J., and Doolittle, hydropathic character
lines allospecific for Sci. USA 79, 2365-
R. F. (1982). A simple method for displaying of a protein. J. Mol. Biol. 157, 105-132.
the
Landegren, U., Ramstedt, U., Axberg, I., Ullberg, M., Jondal, M., and Wigzell, H. (1982). Selective inhibition of human T cell cytotoxicity at levels of target recognition or initiation of lysis by monoclonal OKT3 and Leu-2a antibodies. J. Exp. Med. 755, 1579-1584. Ledbetter, J. A., Evans, R. L., Lipinski. M., Cunningham-Rundles, C., Good, R. A., and Herzenberg, L. A. (1981). Evolutionary conservation of surface molecules that distinguish T lymphocyte helper/inducer and cytotoxic/suppressor subpopulations in mouse and man. J. Exp. Med. 153, 310-323. MacDonald, H. R., Glasebrook, A. L., Bron, C., Kelso, A., and Cerottini, J. C. (1982). Clonal heterogeneity in the functional requirement for Lyt2/3 molecules on cytolytic T lymphocytes (CTL): possible implications for the affinity of CTL antigen receptors. Immunol. Rev. 68, 89-115. Messing, J.. Crea, R., and Seeburg. P H. (1981). A system DNA sequencing. Nucl. Acids Res. 9, 309-321.
for shotgun
Meuer, S. C., Hussey, R. E., Hodgdon, J. C., Hercend, T., Schlossman. S. F., and Reinherz, E. L. (1982). Surface structures involved in target recognition by human cytotoxic T lymphocytes. Science 278, 471-473. Moretta. A., Pantaleo, G., Mingari, M. C., Moretta, J. C. (1984). Clonal heterogeneity in the requirement molecules in human cytolytic T lymphocyte function. 921-934.
L., and Cerottini. for T3, T4, and T8 J. Exp. Med. 759,
Mostov, K. E., Friedlander, M., and Blobel, G. (1984). The receptor for transepithelial transport of IgA and IgM contains multiple immunoglobulin-like domains. Nature 308, 37-43. Poljak, R. J., Amzel, L. M., Chen, B. L., Phizackerley, R. P, and Saul, F. (1975). Structural basis for the association of heavy and light chains and the relation of subgroups to the conformation of the active site of immunoglobulins. Immunogenetics 2, 393-394. Reinherz, E. L., and Schlossman, function of human T lymphocytes.
S. F. (1980). The differentiation Cell 79, 821-827.
and
Reinherz. E. L., Meuer, S. C., and Schlossman, S. F. (1983). The human T cell receptor: analysis with cytotoxic T cell clones. Immunological Rev. 74, 83-112. Sanger, F., Nicklen, S., and Coulson, A. R. (1977). DNA sequencing with chain-terminating inhibitors. Proc. Natl. Acad. Sci. USA 74, 5463-5467. Schneider, M., and Hilschmann, N. (1975). The primary structure of a monoclonic immunoglobulin-L-chain of subgroup IV of the kappa type (Bence-Jones protein Len). Hoppe-Seyleh 2. Physiol. Chem. 356, 507-557. Shinoda, T., Titani, K., and Putman, of human A chains. II. Chymotryptic Ha. J. Biol. Chem. 245, 4475-4487.
F. W. (1970). Amino acid sequence peptides and sequence of protein
Snow, P, Spits, H., deVries, J., and Terhorst, C. (1983). Comparison of target antigens of monoclonal reagents OKT5, OKT8 and Leu-2a, which inhibit effector function of human cytotoxic T lymphocytes. Hybridoma 2, 187-199. Snow, P M., and Terhorst, C. (1983). The T8 antigen is a multimeric complex of two distinct subunits on human thymocytes but consists of homomultimeric forms on peripheral blood T lymphocytes. J. Biol. Chem. 258, 14.675-14,681. Snow, P M.. Keizer, G., Coligan. J. E., and Terhorst, C. (1984). Purification and N-terminal amino acid sequence of the human T cell surface antigen T8. J. Immunol. 733, 2058-2066. Southern, fragments
E. M. (1975). Detection of specific separated by gel electrophoresis.
sequences among DNA J. Mol. Biol. 98, 503-517.
Spits, H.. ljssel, H., Terhorst, C., and deVries, J. E. (1982). Establishment of human T lymphocyte clones highly cytotoxic for an EBVtransformed B cell line in serum-free medium: isolation of clones that differ in phenotype and specificity. J. Immunol. 728, 95-99. Sukhatme,
V. P., Vollmer,
A. C., Erikson,
J.. Isobe,
M., Croce,
C., and
Parnes. J. R. (1985). The gene for the human T cell differentiation antigen Leu-P/T8 is closely linked to the x light chain locus on chromosome 2. J. Exp. Med. 767, 429-434. Tsoukas, C. D., Carson, D. A., Fong, S., and Vaughan, J. H. (1982). Molecular interactions in human T cell-mediated cytotoxicity to EBV. II. Monoclonal antibody OKT3 inhibits a post-killer target recognition/adhesion step. J. Immunol. 729, 1421-1425. Vrana, M., Rudikoff. a hapteninhibitable Williams, A. F. (1984). Nature 308, 12-13.
S., and Potter, M. (1979). The structural basis x-chain idiotype. J. Immunol. 722, 1905-1910. The immunoglobulin
superfamily
takes
of
shape.
Williams, A. F., and Gagnon, J. (1982). Neuronal cell Thy-l glycoprotein: homology with immunoglobulin. Science 276, 696-703. Yanagi, Y., Yoshikai, Y., Leggett, K., Clark, S., Aleksander, I., and Mak, T. (1984). A human T cell-specific cDNA clone encodes a protein having extensive homology to immunoglobulin chains. Nature 308, 145-149.
March
1985, Copyright
0092-8674/85/030591-07
0 1985 by MIT
The T Cell Differentiation Antigen Is Homologous to lmmunoglobulin T Cell Receptor Variable Regions Vikas P. Sukhatme,‘t Kurt C. Sizer,’ Amy C. Vollmer: Tim Hunkapiller,* and Jane R. Parries” * Division of Immunology Department of Medicine Stanford University Medical Center Stanford, California 94305 *Division of Biology California Institute of Technology Pasadena, California 91125
Summary Leu-2/T8 is a cell surface glycoprotein expressed by most cytotoxic and suppressor T lymphocytes. Its expression on T cells correlates best with recognition of class I major histocompatibility complex antigens, and it has been postulated to be a receptor for these proteins. We have determined the complete primary structure of Leu-2/T8 from the nucleotide sequence of its cDNA. The protein contains a classical signal peptide, two external domains, a hydrophobic transmembrane region, and a cytoplasmic tail. The N-terminal domain of the protein has striking homology to variable regions of immunoglobulins and the T cell recep tor. The membrane-proximal domain appears to be a hinge-like region similar to that of immunoglobulin heavy chains. The superfamily of immunologically important surface molecules can now be extended to include Leu-2/T8. Introduction The Leu-2IT8 T cell differentiation antigen is a cell surface glycoprotein expressed by distinct subsets of human T lymphocytes. It is the analog and presumed homolog of mouse Lyt-2,3 (Reinherz and Schlossman, 1980; Ledbetter et al., 1981). Traditionally, these differentiation antigens have been considered markers of T cell function: cytotoxic and suppressor cells expressing Leu-2iT8 (Lyt-2,3 in mouse) and helper/inducer cells expressing the alternative antigen Leu-3/T4 (L3T4 in mouse). However, recent data from many laboratories have indicated that expression of Leu-2/T8 or Leu-3/T4 correlates not so much with function as with recognition by T cells of class I (HLAA, B, C) or class II (HLA-DP, DQ, DR) major histocompatibility complex (MHC) antigens, respectively (Engleman et al., 1981; Krensky et al., 1982a, 1982b; Meuer et al., 1982). Leu-2/T8 is thought to be involved in the recognition by cytotoxic T cells of their targets. Monoclonal antibodies directed against Leu-2IT8 inhibit cytotoxicity by most Leu2/T8+ killer T cells (Meuer et al., 1982; Spits et al., 1982; Reinherz et al., 1983). This block is at the step of cont Present Chicago,
address: Department Illinois 60637.
of Medicine,
University
of Chicago,
$02.00/O
Leu=2/T8 and
jugate formation between the cytotoxic cell and the target cell, rather than at the later killing steps (Tsoukas et al., 1982; Landegren et al., 1982; Moretta et al., 1984). It has been suggested that the function of Leu-2/l8 (or Lyt-2,3) is to increase the avidity of, or stabilize the interaction between, the T cell and its target (Meuer et al., 1982; MacDonald et al., 1982; Moretta et al., 1984), perhaps by binding to nonpolymorphic regions of class I MHC molecules (Krensky et al., 1982a; Meuer et al., 1982; Ball and Stastny, 1982; Spits et al., 1982; Biddison et al., 1982; Engleman et al., 1983). On peripheral T cells the Leu-2/T8 protein consists of dimers and higher multimers of a 32-34 kd glycosylated subunit, linked together by disulfide bridges (Ledbetter et al., 1981; Snow et al., 1983). On thymocytes, in addition to homodimers of this polypeptide, a46 kd subunit is disulfide linked to the smaller chain in tetramers and higher multimers (Ledbetter et al., 1981; Snow and Terhorst, 1983). This larger subunit does not bear the T8 determinant, is unrelated to the smaller subunit by peptide mapping, and is presumably not required for the function of the molecule on mature T cells (Snow and Terhorst, 1983). Snow et al. (1984) have recently sequenced 25 N-terminal residues of Leu-2Il8 and found no significant homology to any other proteins. Determination of the complete primary structure of Leu-P/T8 would be of enormous value not only for evolutionary comparisons but also for further studies of the function of the protein in T cell recognition. To this end we have recently isolated cDNA clones that encode Leu-2/T8 (Kavathas et al., 1984). We have now determined the complete amino acid sequence of the Leu-2/T8 protein from the nucleotide sequence of a cDNA clone. We find that the protein has a classical signal peptide, two external domains, a transmembrane region, and an intracytoplasmic tail. One of the external domains has striking sequence homology to variable regions of other immunological recognition molecules. The results suggest a common evolutionary origin of Leu-2/T8, immunoglobulin, and T cell receptor genes. Results Nucleotide Sequence of Leu-2/T8 cDNA The nucleotide sequence of LeuQ/T8 was determined from a single cDNA clone, pLBM, containing a 2 kb insert, although large portions were confirmed on pdditional clones. Figure 1 shows a restriction map of the insert of clone pL2-M, indicating the enzyme sites used for subcloning into Ml3 vectors. The nucleotide sequence of this clone is presented in Figure 2. There are two long open reading frames of similar sizes at the 5’ end of this clone. One of these extends from the beginning of the clone through nucleotide 632. There are no methionines in this frame, so if it represents a translated protein, our clone is missing the initiation codon. The second open reading frame begins at nucleotide 1 and extends through nucleotide 792. This latter frame encodes Leu-2IT8, as deter-
Cell 592
AS
AR
s
R
H
HR
H
H
HR
II
I
I
II
mined by comparison with the N-terminal protein sequence (Snow et al., 1984). The first ATG in this frame is at nucleotide 88. We believe this to be the initiation codon. Accordingly, clone pLPM contains 115 bp of 5’ untranslated region (not all shown), 705 bp of protein coding sequence (235 amino acids), and 1182 bp of 3’untranslated region. Neither this clone nor another extending 85 bp more S’contains a poly(A) tail. The mRNA represented by these clones is approximately 2.5 kb in length (Kavathas et al., 1984). Our clones span 2.1 kb, so we are missing approximately 400 bp. We presume that most of this is 3’ untranslated region and poly(A) tail.
R
3
5 WJT 9
TM c
3’ UT H 100 L?J
Figure
1. Restriction
Map of Leu-P/T8
cDNA
Clone
pLP-M
A map of the 2 kb Eco RI insert of cDNA clone pLP-M is shown. The enzyme sites indicated are those used for subcloning into Ml3 vectors for DNA sequencing. A, Ava I; S, Sau 3A; R, Rsa I; H, Hinf I. The artificial Eco RI sites at the 5’ and 3’ ends are not labeled. The domains of the protein encoded by various segments of the cDNA are demarcated. Open bar: 5’and 3’ untranslated (UT) regions. Hatched: signal peptide (SP). Shaded: external protein domains. Speckled: transmembrane region (TM). Crosshatched: cytoplasmic tail (C).
Lys Ala Ala Glu Gly Leu Asp Thr Gin Arg Phe Ser Gly Lys Arg Leu Gly ASP Th,PAG GCG GCC GAG GGG CTG GAC ACC CAG CGG TTC TCG GGC AAG AGG TTG GGG GAC ICC
Phe TTC
"al Le" Thr Le" Ser GTC CTC ACC CTG AGC
“31 GTG
Pro CCG
“al GTC
Phe TTC
Leu CTG
Pro CCA
Ala GcG
Lys anG
Pro CCC
Thr *CC
Thr ACG
ThrACG
Pro CCA
Ala GCG
Pro CCG
nrg CGA
Pro CCP
Pro CC*
Thr A.cA
Pro CCG
Ala GCG
Pro CCC
Thr ACC
11e ATC
Ala GCG
ser TCG
Gl” C4G
Pro CCC
Leu CTG
ser TCC
Leu CTG
arg CGC
Pro CC4
Gl” GAG
Ala GCG
cys TGC
Al-g CGG
Pro CCA
Ala GCG
Ala GCG
Gly GGG
GlY GGC
Ala GCA
“al GIG
HIS CAC
Thr ACG
.arg AGG
Gly GGG
Le” CTG
ASP GAC
ser Leu Ser Al.3 brg Tyr vai AGC
Figure
CTT
TCG
2. Nucleotide
GCG
AGP
Sequence
TAC
-
GTC
and Deduced
3’UT
Trm 7114
CCCTGTGCAACAGCCACTACATTACTTCAAACTGAGATCCTTCCTTTTGAGGGAGCAAGTCCTTCCCT
Amino
Acid Sequence
of the Insert
of Leu-P/T8
cDNA
Clone
pL2-M
Horizontal arrows denote the beginning of the predicted protein domains. Open circles are placed above the two cysteines in the V-like domain that are conserved in immunoglobulin and T cell receptor V regions. A closed circle is above the potential site for N-linked glycosylation. A horizontal bar is over the 21 amino acid sequence homologous to the mouse IgA hinge. Numbers in the left and right margins, respectively, refer to the first and last nucleotides or amino acids in each line. Abbreviations: L, leader (signal peptide); V, variable region-like domain; H, hinge; TM, transmembrane region; Cy, cytoplasmic tail; 3’UT, 3’ untranslated region.
Leu-2iT8 593
cDNA
Sequence
Figure 3. Comparison Sequences
of the Amino Acid Sequence
of the N-Terminal
Domain
of Leu-2/T8
with lmmunoglobulin
and T Cell Receptor
Variable
Region
The Leu-P/T8 sequence is the first 96 amino acids of the mature protein. Similarities to Leu-2/T8 are denoted by a dot (.); gaps in the alignments are indicated by blank spaces. lmmunoglobulin lambda, kappa, and heavy chain V regions are denoted VI, V,, and Vn. T cell receptor (I and fl chain V regions are denoted V, and V,+ The sequence alignments were derived by comparing all members of a given class of V regions (e.g. kappa, lambda) against one another and Leu-P/T8 simultaneously to achieve the best alignment for a group of sequences, with attention paid to alignment of certain key conserved residues. All of the immunoglobulin V sequences in the Dayhoff bank were utilized, as were all published and some unpublished T cell receptor sequences (L. E. Hood, personal communication). The sequences were from the following sources: human Ha VI-l (Shinoda et al., 1970); human Meg VI-II (Fett and Deutsch, 1974); mouse U61 V, (Vrana et al., 1979); human Len V, (Schneider and Hilschmann, 1975); mouse Vn (Capra and Nisonoff, 1979); rat IR2 Vn (Hellman et al., 1982); mouse TTll V, (Chien et al., 1984); human TY35 V, (Yanagi et al., 1984).
Predicted Amino Acid Sequence of Leu-2lT8 The predicted amino acid sequence of Leu-2IT8 is shown in Figure 2. The protein begins with a 21 amino acid hydrophobic leader sequence, presumably representing a signal peptide that is cleaved off as the protein passes into the endoplasmic reticulum. The indicated start of the mature processed protein is based upon comparison with the published N-terminal protein sequence (Snow et al., 1984). Our N-terminal sequence differs from the latter in only two places. We predict a glutamine at residue 2, while the protein sequence indicates glutamic acid at this position. As the cDNA was sequenced multiple times on both strands, we believe that eitherthis residue became deamidated during the protein purification or there was an error made by reverse transcriptase during the cDNA cloning. Snow et al. (1984) placed a gap of one amino acid between the cysteine at residue 22 and the two leucines at what were considered residues 24 and 25. We predict two amino acids in this region: glutamine at residue 23 and valine at 24. This discrepancy is likely to be a protein sequencing error, since it is at the very end of the determined sequence. However, it is possible that either or both of these discrepancies could be the result of polymorphism. The mature Leu-2/T8 protein consists of 214 amino acids. A 24 amino acid hydrophobic segment begins at position 162. We believe this to represent a transmembrane region. The following 29 amino acid C-terminal segment (residues 186-214) is highly charged (ten basic residues and only one acidic residue) and is presumably an intracytoplasmic tail. Our sequence predicts an ungiycosylated mature protein chain of 23,554 daltons as compared to the described 32,000-34,000 dalton glycosylated subunit size (Ledbetter et al., 1981; Snow and Terhorst, 1983). However, when Snow et al. (1984) treated the isolated protein with trifluoromethane sulfonic acid (TFMS) to remove both N-linked and O-linked carbohydrate, the size was reduced only 1.5-2 kd. The reasons for this discrepancy are unclear, but most likely relate to incomplete cleavage of carbohydrates from the protein with TFMS. Although our sequence predicts one potential site for N-linked glycosylation (Arg-X-Thr or Arg-X-Ser), at position 28, it may not be used, since experiments with endoglycosidase F suggest that there is no N-linked glycosylation (Snow and Terhorst, 1983).
Leu-2/T8 Is Homologous to lmmunoglobulin and T Cell Receptor Variable Regions We searched the Dayhoff protein sequence data bank for homologies of the predicted Leu-2/T8 protein sequence to other proteins. The greatest amount of homology found was to immunoglobulin light chain variable (V) regions, with less but still significant homology to immunoglobulin heavy chain V regions (V,) and V regions of the a and fi chains of the T cell receptor. Comparisons of the Leu-2/T8 protein sequence with examples of immunoglobulin and T cell receptor V region sequences are illustrated in Figure 3. The segment of Leu9/T8 that is homologous to a V region is at the N terminus and extends to amino acid 96. For the V regions the homology to Leu-2/T8 stops just before the D (diversity) segment (if one is present) or the J (joining) segment. With the alignment shown the homology is on the order of 30%-350/o to lambda and kappa V regions, 200/o-22% to VH, and 24% to T cell receptor (I and /I chain V regions. The homology increases substantially (e.g. up to 56% for lambda or 58% for kappa) if one includes conservative amino acid substitutions. As shown in Figure 3, there are seven amino acids that are conserved in all of these sequences. These include the two cysteine residues involved in the classical intrachain disulfide loop of V regions (positions 22 and 94 for Leu-2/T8) and the tryptophan at position 35, which is thought to be important for the proper folding of immunoglobulin domains. Leu-2/T8 contains most of the residues of light chain V regions that are used for association with heavy chains (Poljak et al., 1975) as well as many of the V region contact residues in light chain dimers (Davies et al., 1975). We therefore predict that the two V-like domains in Leu2/T8 dimers are associated with one another (noncovalently), and probably form a single binding site for a presumed ligand. As might be expected, the homology of Leu9IT8 to V regions is predominantly to framework rather than hypervariable regions. The level of homology does not vary substantially across mammalian species. In contrast to the findings for the N-terminal external domain, the 65 amino acid membrane-proximal domain of Leu-2IT8 bears no strong resemblance to constant region domains of immunoglobulins or the T cell receptor and is not homologous to other known proteins. However, a stretch of 21 amino acids in the center of this domain (residues 120-140) is homologous to the hinge region of
Cell 594
Leu-Z/T8
Hinge
Mouse
CHa
Hinge
Figure
4. Homology
PAPRPPTPAPTIASQP * * ** PTPPPPITIkC
of Leu-2/T8
Hinge to Mouse
LSL RP ** *** ** QPSLSLQRP IgA Hinge
Amino acids 120-140 of Leu-2/T8 are shown. Amino acids 220-240 of the mouse a heavy chain (CH,) are below. The entire CH, hinge spans amino acids 218-240. Homologies are denoted by an asterisk (‘), and gaps, by blank spaces. The CH, hinge sequence is from Auffray et al. (1981).
the mouse IgA heavy chain (Figure 4). Although we doubt an evolutionary significance of this homology, evidence discussed below suggests that the domain containing this sequence is a hinge. The remaining protein domains appear unique to Leu9/T8. Structural Predictions We have analyzed the Leu-2/T8 sequence using a formula similar to that of Kyte and Doolittle (1982) for determining the degree of hydrophobicity or hydrophilicity of segments of a protein. Such analyses allow one to predict which portions of a protein are exposed, interior, or embedded in a membrane. A hydrophobicity plot of the Leu-2Il8 sequence is shown in Figure 5. The putative signal peptide (residues -21 to -1) and transmembrane domain (residues 162 to 185) are both demonstrated to be highly hydrophobic, while the presumed intracytoplasmic tail (residues 186 to 214) is very hydrophilic. The V-like N-terminal external domain has alternating stretches of hydrophobic and hydrophilic residues as seen in globular proteins (Kyte and Doolittle, 1982). Furthermore, the plot for this domain is virtually superimposable on those for kappa, lambda, and heavy chain V regions, except for an extra hydrophilic segment in Leu-2/T8 between amino acids 6 and 15 (data not shown). These results suggest a great deal of conservation of structure among these protein domains. We further analyzed the Leu-2Il8 protein sequence according to the parameters of Chou and Fasman (1978) and find that the predicted p-sheet structure of the N-terminal domain is the same as that for light chain V regions. In contrast, the membrane-proximal external domain does not fit into any structure. Therefore this domain, which was shown to contain a sequence homologous to the mouse IgA hinge, is itself predicted to be a hinge region. The Leu-2/T8 Gene Does Not Rearrange in Cells That Express the Protein Variable regions of immunoglobulins and the T cell receptor are not expressed as functional receptor molecules until their genes rearrange during the development of 6 cells or T cells, respectively, and become juxtaposed to either a J or D and J segments just upstream from a constant region gene. Since these proteins are homologous to Leu2/T8, it was important lo determine whether the Leu-2/l8 gene also rearranges in T cells that express this protein. We have therefore compared Southern blots of DNA from human placenta (which does not express Leu-2/T8 protein or mRNA) and from a Leu-2/T8-expressing thymoma cell line, JM, hybridized to the insert of cDNA clone pLBM. As
illustrated in Figure 6, no difference could be detected between the two DNAs with any of the four enzymes used (Barn HI, Eco RI, Apa I, and Eco RV). There is also no difference seen with Hind Ill (data not shown). Since the probe contains the entire coding sequence as well as most of the noncoding portions, we conclude that no major rearrangements are required for expression of this gene. Although we cannot exclude very small rearrangements with either little or no deletion of DNA, immunoglobulin and T cell receptor type rearrangements would have been easily detectable. The data shown in Figure 6 also suggest that Leu9IT8 is encoded by a single gene. The Southern blots show one to three hybridizing bands, depending upon the enzyme. Differential hybridization to 5’ and 3’ probes indicates that when multiple bands are present, they represent portions of a single gene, and all of the hybridizing fragments can be accounted for on a single genomic clone containing a 14 kb insert (data not shown). Of course, it cannot be excluded that the entire Leu-2/T8 gene and large amounts of flanking sequence are repeated in the genome with little or no change. Using low stringency wash conditions (37% 1 x SSC) we have detected one or two extra bands on genomic blots with DNA digested with Barn HI, but not reproducibly with other enzymes. We are currently determining whether these are truly distantly cross-hybridizing genes or aberrant digestion products of the LeuQTTB gene with Barn HI. It is clear, however, that Leu-2TT8 is not a member of a large multigene family. Discussion The structure of Leu-2/T8 is typical of a cell surface protein. It contains a hydrophobic signal peptide, two predicted external protein domains, a transmembrane hydrophobic domain, and an extremely basic intracytoplasmic tail. The protein has a total of nine cysteine residues: three in the N-terminal V-like domain and two each in the membrane-proximal (hinge), the transmembrane, and the cytoplasmic domains. Biochemical data indicate that the three in the V-like domain are not used for interchain disulfide bridging (Snow and Terhorst, 1983). Since our alignment with V regions shows conservation of the cysteines at positions 22 and 94 of Leu9/T8, we predict that these two cysteines are used for an intrachain disulfide bridge as in immunoglobulins and the T cell receptor. The function of the cysteine at position 33 is unknown. Subunit joining could occur between membrane-proximal domains, within the membrane, or within the cytoplasm. Several of these six cysteine residues may be used for interchain bridging, since the molecule often forms large multimers. Perhaps the most interesting finding concerning the structure of Leu-2/T8 is the homology of the N-terminal external domain to variable regions. Many members of the “superfamily” (Jensenius and Williams, 1982; Williams, 1984) of immunologically important surface proteins are much more homologous to constant regions than to variable regions (e.g. class I and class II MHC antigens, B2microglobulin). In contrast to these proteins, Leu-2IT8 is
Leu-2/T8 595
cDNA
Sequence
P
‘: Figure
5. Hydrophobicity
Plot of Leu-2/T8
Protein
1
2
k
P
Sequence
Data are plotted according to Kyte and Doolittle (1982). The plotted values are averages over groups of seven contiguous residues. Positive values indicate hydrophobic, and negative values indicate hydrophilic, regionsof the protein. The amino acid sequence is listed above the plot and numbered below. The domains of the protein are labeled as in the legend to Figure 2. 1
thought to be a receptor molecule, and so it is not surprising that it would borrow or share a structure that the immune system has used to such advantage in antigen recognition. It should be stressed, however, that there is no evidence for variability of this “V-like” domain. Two other members of the superfamily (besides immunoglobulins and T cell receptors) are thought to be related to variable regions: the mouse T cell marker Thy-l (Williams and Gagnon, 1982) and the poly(lg) receptor (Mostov et al., 1984). The homology of Leu9IR3 to these molecules is less striking, as is their homology to immunoglobulin V regions. Our sequence analyses indicate that Leu-2/T8 is about equidistant between its two closest relatives: kappa and lambda V regions. If one were to construct a genealogical tree of the immunoglobulin-related superfamily, LeuQTT8 would split off after the divergence of heavy and light chains, and presumably just before the split between kappa and lambda. Chromosomal mapping studies of Leu-2/T8 are intriguing in this regard. Lyt-2,3 has been known for many years to be tightly linked to the kappa locus on mouse chromosome 6 (Itakura et al., 1972; Gottlieb, 1974; Claflin et al., 1978; Gibson et al., 1978). We have recently shown that in the human system the Leu-2/T8 gene also maps extremely close to the kappa locus on chromosome 2 (Sukhatme et al., 1985). This close linkage between the genes encoding kappa and Leu-2/T8, which is maintained across at least two species, lends credence to the idea of a common evolutionary origin of these proteins. Although the N-terminal domain of Leu-2/T8 is remarkably similar to V regions, it differs from them in several important aspects. First, the gene encoding Leu-PfT8 shows no evidence of rearrangement in cells that express the protein. Second, the Leu-2/T8 gene has no close relatives in the human genome, in contrast to most, though not all, V genes. Finally, the nucleotide sequence encoding Leu2/T8 is strikingly distinct from the sequences encoding V regions in two aspects. As is true for most protein-coding sequences (Fickett, 1982), the G+C content of the region encoding the V-like domain (or the entire coding region) of Leu-2/T8 is high (66%). In contrast, the G+C content of immunoglobulin V regions is much closer to random (T. Hunkapiller and L. E. Hood, unpublished results). Furthermore, we have analyzed the Leu-2/T8 cDNA se-
2
3
23.1-
4
23.1-
23.1-m
6
5
7
-
1% c-
-
m
8
9
I)
9.4-
*
6.6-
.
9.4-
9.4-
6.6-
.
6.6-a@
-
44--o
c
4.4-
4.4-
2.3-
2.3-
2.0-
2.0-
2.32.0-
Figure
6. Southern
Blot Analysis
of Leu9/T8
Gene
Genomic DNA (8 rg) from a human thymoma cell line (JM) that expresses Leu-P/T8 (lanes 1, 3. 4, 6, 8) and from human placenta, which does not express Leu-2/T8 (lanes 2, 5, 7, 9) were digested with Barn HI (lanes 1, and 2) Hind III (lane 3) Eco RI (lanes 4 and 5) Apa I (lanes 6 and 7). or Eco RV (lanes 8 and 9) and electrophoresed on a 0.8% agarose gel. Southern blots of the gels were hybridized to the insert of cDNA clone pLP-M labeled with 32P by nick-translation. The migration distances of Hind Ill fragments of A phage DNA are indicated in the left margins in kilobase pairs.
quence with an algorithm similar to that of Fickett (1982) which can distinguish protein coding sequences from noncoding DNA. This test for coding sequences is based upon the nonrandom use of codons, resulting in repetition of particular nucleotides with a periodicity of 3 in coding, but not noncoding, DNA. The sequence encoding Leu2/T8 is clearly predicted as protein-coding, as is that of the V-like domain by itself. In contrast, immunoglobulin V region sequences (but not constant region sequences) are an exception to the rule in that they are most often predicted as noncoding (Fickett, 1982; T. Hunkapiller and L. E. Hood, unpublished results). These results imply that there is less constraint on the evolution of V genes than
Cell 596
on other genes and that, despite its homology to V regions, Leu9/T8 is under much stronger selective pressure to maintain its sequence. Experimental
Pmcedures
Isolation of cDNA Clone pL2-M The inserts of two Leu-2/T8 cDNA clones isolated previously by subtractive hybridization (Kavathas et al., 1984) were labeled with J2P by nick translation and used as probes to rescreen a previously described AgtlO cDNA library. This library was constructed from RNA of L cells that had been transfected with total human genomic DNA and selected for expression and amplification of the Leu-PiT8 gene (Kavathas et al., 1984). Positive clones were screened for insert size. The longest insert was 2 kb. This clone was grown up, and the Eco RI insert was isolated and subcloned into pBR322. The resulting plasmid is referred to as cDNA clone pLBM. DNA Sequencing The Eco RI insert of cDNA clone pLP-M was digested with Rsa I and inserted into EGO RI plus Sma I digested Ml3 vectors (Messing et al., 1981) mpl0 and mpll, or into Sma I digested mpll. The insert was also digested with Hinf I, the Soverhangs filled-in with the Klenow fragment of DNA polymerase I (Bethesda Research Laboratories), and the fragments subcloned into Sma I digested mpll. The largest Hinf I digestion product (the 5’ 1.1 kb Eco RI-Hinf I fragment) was isolated from the pLP-M insert. digested with Ava I and Sau 3A, filled-in with Klenow polymerase, and inserted into Sma I digested mpll. Finally, the entire Eco RI inserts of clone pL2-M and four independent, overlapping cDNA clones were cloned into the Eco RI site of mpll. Except for the first 48 bp of 5’ untranslated sequence, which were only sequenced on one strand, the nucleotide sequences of all these subclones were determined on both strands multiple times by the dideoxynucleoside triphosphate chain termination method of Sanger et al. (1977). The sequence shown was all derived from clone pLP-M, although the partial sequences of other independent cDNA clones agreed entirely except for a single base pair in the 3’untranslated region (a run of six As beginning at nucleotide 1895 in pLP-M and seven As in clone pL2-4). The Ml3 vectors were obtained from Amersham. Southern Blots and Hybridization Genomic DNA was isolated from the T cell lymphoma line JM, by the citric acid procedure (Hieter et al., 1981). Human placental DNA was a generous gift from Dr. Philip Leder (Harvard Medical School). DNA was digested with restriction enzymes (New England Biolabs) for 3 hr and electrophoresed on 0.8% agarose gels. Gels were blotted onto nitrocellulose by the procedure of Southern (1975). Blots were hybridized for 16 hr in 40% formamide, 20 mM Tris-HCI (pH 7.6), 4x SSC. lx Denhardt’s solution, 0.1% SDS, and 30 pglml denatured herring sperm DNA, to the insert from clone pL2-M labeled with 32P by nick translation to a specific activity of 200-400 cpm per picogram. The filters were washed three times at room temperature for 20 min each with 2x SSC, 0.05% SDS, and twice at 52% for 1 hr each with 0.1 x SSC, 0.05% SDS. Exposures with XAR-5 film (Kodak) were overnight at -70% with intensifying screens. Acknowledgments We thank Dr. Philip Leder for the human placental DNA, Dr. Rose Zamoyska for critical reading of the manuscript, Dr. Leroy Hood for helpful discussions and for use of his computer facilities, and Diane O’Neill for preparation of the manuscript. This work was supported by NIH grant Al 19512 to J. R. P V. P S. was a National Institutes of Health postdoctoral fellow. J. R. P. is the recipient of a John A. and George L. Hartford faculty fellowship award. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. Received
December
18, 1984; revised
January
10, 1985
References Auffray, C., Nageotte, R., Sikorav, J.-L., Heidmann, O., and Rougeon, F. (1981). Mouse immunoglobulin A: nucleotide sequence of the structural gene for the a heavy chain derived from cloned cDNAs. Gene 13, 365-374. Ball, E. J., and Stastny, I? (1982). Cell-mediated cyiotoxicity against HLA-D region products expressed in monocytes and B lymphocytes. IV. Characterization of effector cells using monoclonal antibodies against human T cell subsets. Immunogenetics 16, 157-169. Biddison, W. E., Rao, P E., Talle, M. A., Goldstein, G., and Shaw, S. (1982). Possible involvement of the T4 molecule in T cell recognition of class II HLA antigens: evidence from studies of cytotoxic T lymphocytes specific for SB antigens. J. Exp. Med. 156, 1065-1083. Capra, J. D., and Nisonoff, A. (1979). Structural studieson induced antibodies with defined idiotypic specificities. VII. The complete amino acid sequence of the heavy chain variable region of anti-p-azophenylarsonate antibodies from AN mice bearing a cross-reactive idiotype. J. Immunol. 123. 279-284. Chien, Y., Becker, D. M., Lindsten. T., Okamura, M., Cohen, D. I., and Davis, M. M. (1984). A third type of murine T-cell receptor gene. Nature 312, 31-35. Chou, P Y., and Fasman. G. D. (1978). Empirical conformation. Ann. Rev. Biochem. 47, 251-276.
predictions
of protein
Claflin, J. L., Taylor, B. A., Cherry, M., and Cubberly, M. (1978). Linkage in mice of genes controlling an immunoglobulin kappa-chain marker and the surface alloantigen Ly-3 on T lymphocytes. Immunogenetics 6, 379-387. Davies, D. R., Padlan, E. A., and Segal, D. M. (1975). Threedimensional structure of immunoglobulins. Ann. Rev. Biochem. 44, 639-667. Engleman, E. G., Benike, C. J., Grumet, F. C., and Evans, R. L. (1981). Activation of human T lymphocyte subsets: helper and suppressorlcytotoxic T cells recognize and respond to distinct histocompatibility antigens. J. Immunol. 127, 2124-2129. Engleman, E. G., Benike, C. J., Metzler, C., Gatenby, F? A., and Evans, R. L. (1983). Blocking of human T lymphocyte functions by anti-Leu-2 and anti-Leu-3 antibodies. J. Immunol. 130, 2623-2628. Fett, J. W., and Deutsch, H. F. (1974). Primary chain. Biochemistry 13, 4102-4114. Fickett, J. W. (1982). Recognition of protein quences. Nucl. Acids Res. 10, 5303-5318.
structure
coding
of the Meg I
regions
in DNA se-
Gibson, D. M., Taylor, B. A., and Cherry, M. (1978). Evidence for close linkage of a mouse light chain marker with the Ly-2,3 locus. J. Immunol. 121, 1585-1590. Gottlieb, f? (1974). Genetic correlation of a mouse light chain variable region marker with a thymocyte surface antigen. J. Exp. Med. 140. 1432-1437. Hellman, L., Petterson, U., Engstrom, A., Karlsson, T., and Bennich, H. (1982). Structure and evolution of the heavy chain from rat immunoglobulin E. Nucl. Acids Res. 10, 6041-6049. Hieter, i? A., Hollis, G. F.. Korsmeyer, S. J., Waldmann, T. A., and Leder, P (1981). Clustered arrangement of immunoglobulin A constant region genes in man. Nature 294, 536-540. Itakura, K., Hutton, J. J., Boyse, E. A., and Old, L. J. (1972). Genetic linkage relationships of loci specifying differentiation alloantigens in the mouse. Transplantation 13, 239-243. Jensenius, J. C., and Williams, A. F. (1982). The T lymphocyte receptor-paradigm lost. Nature 300, 583-588.
antigen
Kavathas, P., Sukhatme, V. I?. Herzenberg, L. A., and Parnes, J. R. (1984). Isolation of the gene coding for the human T lymphocyte antigen Leu-2 (T8) by gene transfer and cDNA subtraction. Proc. Natl. Acad. Sci. USA 81, 7688-7692. Krensky, A. M., Clayberger, C., Reiss, C. S., Strominger, J. L.. and Burakoff, S. J. (1982a). Specificity of OKT4 cytotoxic T lymphocyte clones. J. Immunol. 129, 2001-2003. Krensky,
A. M., Reiss, C. S., Mier, J. W., Strominger,
J. L., and Burakoff,
Leu-P/T8 597
cDNA
Sequence
S. J. (1982b). Long-term human cytolytic T-cell HLA-DR6 antigen are OKT4’. Proc. Nan. Acad. 2369. Kyte, J., and Doolittle, hydropathic character
lines allospecific for Sci. USA 79, 2365-
R. F. (1982). A simple method for displaying of a protein. J. Mol. Biol. 157, 105-132.
the
Landegren, U., Ramstedt, U., Axberg, I., Ullberg, M., Jondal, M., and Wigzell, H. (1982). Selective inhibition of human T cell cytotoxicity at levels of target recognition or initiation of lysis by monoclonal OKT3 and Leu-2a antibodies. J. Exp. Med. 755, 1579-1584. Ledbetter, J. A., Evans, R. L., Lipinski. M., Cunningham-Rundles, C., Good, R. A., and Herzenberg, L. A. (1981). Evolutionary conservation of surface molecules that distinguish T lymphocyte helper/inducer and cytotoxic/suppressor subpopulations in mouse and man. J. Exp. Med. 153, 310-323. MacDonald, H. R., Glasebrook, A. L., Bron, C., Kelso, A., and Cerottini, J. C. (1982). Clonal heterogeneity in the functional requirement for Lyt2/3 molecules on cytolytic T lymphocytes (CTL): possible implications for the affinity of CTL antigen receptors. Immunol. Rev. 68, 89-115. Messing, J.. Crea, R., and Seeburg. P H. (1981). A system DNA sequencing. Nucl. Acids Res. 9, 309-321.
for shotgun
Meuer, S. C., Hussey, R. E., Hodgdon, J. C., Hercend, T., Schlossman. S. F., and Reinherz, E. L. (1982). Surface structures involved in target recognition by human cytotoxic T lymphocytes. Science 278, 471-473. Moretta. A., Pantaleo, G., Mingari, M. C., Moretta, J. C. (1984). Clonal heterogeneity in the requirement molecules in human cytolytic T lymphocyte function. 921-934.
L., and Cerottini. for T3, T4, and T8 J. Exp. Med. 759,
Mostov, K. E., Friedlander, M., and Blobel, G. (1984). The receptor for transepithelial transport of IgA and IgM contains multiple immunoglobulin-like domains. Nature 308, 37-43. Poljak, R. J., Amzel, L. M., Chen, B. L., Phizackerley, R. P, and Saul, F. (1975). Structural basis for the association of heavy and light chains and the relation of subgroups to the conformation of the active site of immunoglobulins. Immunogenetics 2, 393-394. Reinherz, E. L., and Schlossman, function of human T lymphocytes.
S. F. (1980). The differentiation Cell 79, 821-827.
and
Reinherz. E. L., Meuer, S. C., and Schlossman, S. F. (1983). The human T cell receptor: analysis with cytotoxic T cell clones. Immunological Rev. 74, 83-112. Sanger, F., Nicklen, S., and Coulson, A. R. (1977). DNA sequencing with chain-terminating inhibitors. Proc. Natl. Acad. Sci. USA 74, 5463-5467. Schneider, M., and Hilschmann, N. (1975). The primary structure of a monoclonic immunoglobulin-L-chain of subgroup IV of the kappa type (Bence-Jones protein Len). Hoppe-Seyleh 2. Physiol. Chem. 356, 507-557. Shinoda, T., Titani, K., and Putman, of human A chains. II. Chymotryptic Ha. J. Biol. Chem. 245, 4475-4487.
F. W. (1970). Amino acid sequence peptides and sequence of protein
Snow, P, Spits, H., deVries, J., and Terhorst, C. (1983). Comparison of target antigens of monoclonal reagents OKT5, OKT8 and Leu-2a, which inhibit effector function of human cytotoxic T lymphocytes. Hybridoma 2, 187-199. Snow, P M., and Terhorst, C. (1983). The T8 antigen is a multimeric complex of two distinct subunits on human thymocytes but consists of homomultimeric forms on peripheral blood T lymphocytes. J. Biol. Chem. 258, 14.675-14,681. Snow, P M.. Keizer, G., Coligan. J. E., and Terhorst, C. (1984). Purification and N-terminal amino acid sequence of the human T cell surface antigen T8. J. Immunol. 733, 2058-2066. Southern, fragments
E. M. (1975). Detection of specific separated by gel electrophoresis.
sequences among DNA J. Mol. Biol. 98, 503-517.
Spits, H.. ljssel, H., Terhorst, C., and deVries, J. E. (1982). Establishment of human T lymphocyte clones highly cytotoxic for an EBVtransformed B cell line in serum-free medium: isolation of clones that differ in phenotype and specificity. J. Immunol. 728, 95-99. Sukhatme,
V. P., Vollmer,
A. C., Erikson,
J.. Isobe,
M., Croce,
C., and
Parnes. J. R. (1985). The gene for the human T cell differentiation antigen Leu-P/T8 is closely linked to the x light chain locus on chromosome 2. J. Exp. Med. 767, 429-434. Tsoukas, C. D., Carson, D. A., Fong, S., and Vaughan, J. H. (1982). Molecular interactions in human T cell-mediated cytotoxicity to EBV. II. Monoclonal antibody OKT3 inhibits a post-killer target recognition/adhesion step. J. Immunol. 729, 1421-1425. Vrana, M., Rudikoff. a hapteninhibitable Williams, A. F. (1984). Nature 308, 12-13.
S., and Potter, M. (1979). The structural basis x-chain idiotype. J. Immunol. 722, 1905-1910. The immunoglobulin
superfamily
takes
of
shape.
Williams, A. F., and Gagnon, J. (1982). Neuronal cell Thy-l glycoprotein: homology with immunoglobulin. Science 276, 696-703. Yanagi, Y., Yoshikai, Y., Leggett, K., Clark, S., Aleksander, I., and Mak, T. (1984). A human T cell-specific cDNA clone encodes a protein having extensive homology to immunoglobulin chains. Nature 308, 145-149.