CD3ϵ and the B-cell antigen receptor associated Ig-β (B29)

Immunology Letters, 44 (1995) 97-103 Elsevier Science B.V. IMLET 2281

Interplay between the human TCR/ CD3 E and the B-cell antigen receptor associated Ig-p (B29) Brigitte

Miiller

* ,‘, Laura

Cooper

and Cox Terhorst

Division of Immunology, Beth Israel Hospital, Haruard Medical School, Boston, MA 02115. USA (Received

14 October 1994; accepted

18 October 1994)

Key words: Antigen receptor (AgR) complex; T-/B-cell

1. Summary T and B lymphocytes are characterized by the surface expression of highly variable antigen receptors called the T-cell (TCR) and B-cell (BCR) receptor. In both B and T cells, binding of antigen to their respective surface receptors results in transmembrane signaling which leads either to programmed cell death or to proliferatin and differentiation. The human and murine TCR consist of the highly variable TCR cy and p or y and S chains recognizing antigen and the non-covalently associated invariable CD3 complex (y, S, E) and 5. The antigen-recognizing surface membrane-bound immunoglobulin (Ig) molecules on the surface of human and murine B cells are non-covalently associated with a heterodimeric protein complex of Ig-(Y (MB-l) and Ig-p (B29). Like the CD3 complex associated with TCR the Ig-associated proteins are predicted to regulate the assembly and transport of the Ig complex to the cell surface and to couple membrane-bound Ig to intracellular signal transduction pathways. To gain more insight into structure/function relationships between CD3 proteins and Ig-cr and Ig-p, we transiently co-transfected pairs of expression vectors encoding either TCR/CD3 chains on the one hand and Ig-a or Ig-p on the other into Cos cells. Thus we found a very strong interaction between CD3e and Ig-p mediated by the extracellular domains. Experiments in which we could stain Jurkat T cells with soluble Ig-p but not Ig-a! protein indicated the recognition of CD3e by Ig-/3 even in the context of the whole TCR/CD3 complex.

* Corresponding author: Brigitte Miller, Division of Immunology, Beth Israel Hospital, Harvard Medical School, 330 Brookline Avenue, Boston, MA 02115, USA. 1Present address: DeutschesRheumaForschungsZentrum, Nordufer 20, D-13353 Berlin, Germany. SSDf 0165-2478(94)00199-S

interaction

The staining of Jurkat cells by soluble Ig-j3 protein could specifically be inhibited by blocking the extracellular domains of either CD3e or Ig-p with antibodies. These data hint at a possible function for CD3e and Ig-p in constituting a novel pair of surface molecules involved in T-/B-cell interactions.

2. Introduction The T-cell receptor (TCR) on the surface of human T lymphocytes consists of antigen-specific receptor chains ((w and /l or y and S) which are associated with a complex of invariant chains called CD3y, -6, -E and -C [l-3]. In analogy, membrane-bound immunoglobulins (Ig) on the surface of B lymphocytes are non-covalently associated with a heterodimer of glycoproteins containing Ig-a and Ig-fi (also called MB-l and B29) [4-71. The TCR and CD3 chains as well as Ig, Ig-(Y and Ig-/3, are type-1 membrane proteins consisting of an extracellular domain (EC) with 1 or more Ig-like domains, 1 transmembrane region (TM) harboring a charged amino acid and a cytoplasmic tail (CT). For the TCR/CD3 complex it has been shown that the positive charges within the TM domains of TCR CY and p chains stabilize the interaction with CD3 chains and 5 which in turn have negative charges in their TM regions. These TM residues may be utilized during the intracellular assembly and transport of the oligomeric complexes [8,9]. We previously cloned and sequenced full-length CDNAs encoding the human Ig-cu and Ig-/3 proteins [lo]. Aligning the amino acid sequences of the CD3 proteins and Ig-cu and Ig-p showed structural similarities which suggest conserved functions for both the Ig- and TCR associated proteins. They all share an EC region with 1 97

Ig-like domain, a TM region with a negative charge located towards the EC domain and a CT displaying the AgR tail motif [ll]. In transient transfection experiments, some of the TCR/CD3 chains were shown to form subcomplexes [12,13]. Based on that observation, we asked whether the structural homologies found between CD3 proteins and Ig-a and Ig-/3 would allow the formation of mixed subcomplexes. To our surprise we found a strong and specific interaction between CD3e and Ig-p which was not mediated by the TM but by the EC domains. We are now investigating whether both proteins also interact when expressed on the surface of T and B cells, respectively. They would thus constitute a novel pair of surface molecules involved in T-/B-cell interactions taking place during T cell-mediated help in a humoral immune response.

3. Materials and Methods 3.1. Constructs Human Ig-p was cloned into the expression vector pcDL-SR (Y296 [14] as were the constructs encoding only the Ig-/3 EC domain (Ig-P-EC), the Ig-p-5 fusion protein and CD3e EC domain. The EC domains were cloned as amplified PCR fragments using primers which contained appropriate restriction enzyme cleavage sites, which facilitated subsequent insertion into the vector. To join the Ig-/3-EC domain to the &‘-TM/CT tail we used the BumHI site 5’ of the sequence encoding the TM domain. The Ig-/3-IgGl and Ig-cu-IgGl constructs were made by joining the cDNAs encoding the respective EC domains to genomic sequences encoding human IgGl constant region, using the BamHI site at the 3’-end of the exon encoding the CHl domain (see Fig. 2). The expression vector carrying the IgGl construct is pCDM7 and is described in detail elsewhere [15]. All the constructs used for protein expression in Cos cells were made using standard molecular cloning techniques.

temperature in DMEM containing 10% DMSO. After overnight culture in DMEM supplemented with 5% heat-inactivated FCS, cells were trypsinized, and seeded onto fresh plates. For metabolic labeling, cells were left to grow until 48 h after transfection. For the collection of supernatants, cells were left to grow for 7-10 days. On the 4th day, the first supematant was collected and fresh medium was added back to the cells. Finally, both supernatants were pooled, centrifuged to remove nonadherent cells and debris and stored at 4°C. 3.3. Metabolic

labeling and immunoprecipitations

labeled metabolically with Cells were [ 35S]methionine and [ 35Slcysteine using Trans 35S-label (ICN Radiochemicals, Cleveland, OH). Cells (90% confluent, in a lo-cm dish) were pre-incubated for 1 h at 37°C in methionine- and cysteine-free medium, washed and then labeled for 2-3 h at 37°C in the same medium containing 0.2 mCi/ml 35S. Cells were washed again and lysed at 4°C in immunoprecipitation buffer (50% saturated digitonin, 10 mM triethanolamine, pH 7.8, 0.15 M NaCl, 10 mM iodoacetamide, 1 mM EDTA, 1 mM PMSF and 1 pg/ml of the following peptide protease inhibitors: leupeptine, pepstatin, chymostatin and antipain [2]. Postnuclear supematants were prepared by centrifugation, precleared and immunoprecipitated by antibodies immobilized on Protein A-Sepharose (Pharmacia Fine Chemicals, Piscataway, NJ). Immunoprecipitates were washed extensively in immunoprecipitation buffer and solubilized in non-reducing or reducing sample buffer in preparation for SDS-PAGE. PAGE was performed using 12.5% slab gels [16]. Gels were fixed and incubated with 1 M salicylic acid before autoradiography. 3.4. Purification

of soluble Ig fusion proteins

The fusion proteins formed accumulated to approximately l-4 pg/ml in Cos cell supematants at 7 days post-transfection. They were purified using protein ASepharose affinity chromatography.

3.2. Cos cell transfections 3.5. FITC staining of Jurkat cells and antibodies Monkey Cos cells were transfected transiently with the corresponding expression vectors as follows. Cos cells at 60% confluence on a lo-cm dish were transfected by 4-6 h incubation at 37°C with 2 pg vector DNA dissolved in 5 ml of DMEM containing 10% NuSerum (Collaborative Research, New Bedford, MA), 400 pg/ ml DEAE dextran (Pharmacia Fine Chemicals, Piscataway, NJ) and 100 PM chloroquine. Washed cells were then shocked by 2-min incubation at room 98

lo6 cells were incubated with 100 ~1 of DMEM containing 0.01% NaN, and either 5 pg/ml monoclonal antibody (mAb) or 30 ,ug/ml purified Ig-fusion protein for 30 min at room temperature. Subsequently they were stained with either FITC-labeled rabbit antimouse or FITC-labeled goat anti-human antibodies (Becton Dickinson, Mountain View, CA> at the recommended dilution.

For the inhibition experiments, Jurkat cells were incubated with 100 ~1 of DMEM containing 0.01% NaN, and 20 pg/ml SP34 (the specifities of which are described in [17]) or lC7.2 (mAb raised against the N-terminal 11 amino acids of human CD3e) prior to staining. Likewise, 3 pg of soluble Ig-p-Ig were incubated with 2 pug of CB3-1 (mAb directed against the EC domain of Ig-/3) [18].

200 97

69

46

4. Results

; C% B

4.1. Interaction Ig-p and CD3e

between

the extracellular

domains

of

Because of the structural homologies between the invariant chains associated with BCR and TCR, we were interested whether we would be able to find partial complexes formed between TCR/CD3 proteins and Ig-(Y or Ig-p. We therefore transiently transfected Cos cells with pairs of cDNAs encoding as 1 partner human Ig-a or Ig-/3 and a TCR, CD3 or 5 chain as the other partner. After metabolic labeling and lysis under mild detergent conditions, the protein complexes were immunoprecipitated with antibodies against TCR or CD3 proteins. The precipitates were subjected to SDS-PAGE to look for proteins coprecipitating with TCR or CD3 proteins and thereby indicating complex formation. We hardly detected any coprecipitation of Ig-a with TCR/CD3 proteins. We did find very little coprecipitates of Ig-p with TCRa and -p. And we also found only a little coprecipitate of Ig-p with CD3-y and CD36. All of the above we interpreted as weak associations. We could not detect any interaction between Ig-/3 and -5 (data not shown). However, we did find a very strong association between CD3 E and Ig-p which turned out to be mediated via the EC domains. Fig. 1 outlines these results. Each lane - except for the controls - corresponds to an individual experiment consisting of a Cos cell transfection, a metabolic labeling and a consecutive immunoprecipitation with SP34 which is an antibody directed against the EC domain of CD3e. The cDNA encoding human CD3e was transfected into Cos cells either alone or in combination with cDNAs encoding human Ig-/3 (whole molecule), the EC domain of Ig-p only or of Ig-P-[-chimera consisting of Ig-P-EC fused to the l-TM/CT domains. Further transfections included a cDNA encoding the CD~E-EC domain in combination with either the cDNA for Ig-/3 (whole molecule) or Ig-P-EC domain only. In each experiment the Ig-p protein was coprecipitated with CD3e. To assess the affinities of the different associations, we compared the amount of Ig-p protein which was coprecipitated. We therefore pasted together differ-

30

Fig. 1. Coprecipitations of Ig-p proteins with CD3<. Cos cells were transiently transfected with cDNAs encoding human CD3e either alone or in combination with the cDNAs encoding the constructs indicated above. Postnuclear supernatants of metabolically labeled Cos cells were then immunoprecipitated with an antibody directed against the EC domain of CD3t. Subsequently the precipitates were subjected to SDS-PAGE to look for coprecipitated proteins indicative of complex formation. Precipitates of the TCR from Jurkat lysates and the BCR from Ramos lysates were run as controls. They show the positions of the proteins in the gel, as indicated by the arrows. The positions of the molecular weight markers are indicated on the left.

ent exposures of the same gel to obtain similar intensities for the CD3e bands in the different lanes. As Fig. 1 shows, the coprecipitations worked best using whole molecules of Ig-/3 and CD3e. To find out whether the interaction between Ig-/3 and CD3e is mediated via the TM domains, as was shown for subcomplexes formed by CD3 proteins, or whether the EC domains are involved we used the Ig-p-5 and EC domain constructs. The Ig-/3--l construct encodes a fusion protein of the Ig-P-EC joined to the l-TM/ CT domains whereas the EC domain constructs encode soluble forms of the EC domains of Ig-fi and CD~E, only. As Fig. 1 shows, the Ig-p-5 construct was coprecipitated with the whole molecule of CD3e whereas the native l-chain was not (data not shown). Thus, complex formation seems to be independent of the TM regions. The coprecipitation also worked using the EC domains only, indicating that the association is indeed mediated via the EC domains. Less Ig-p protein was coprecipi99

NR

tated when the whole molecule of CD3e was used to coprecipitate the EC domain of Ig-p. And the least coprecipitation resulted from the usage of the CD3e EC domain and the whole molecule of Ig-/3. The control lanes show immunoprecipitates of lysates of metabolically labeled Jurkat and Ramos cells with antibodies directed against CD3e and human Ig, respectively. They indicate the positions of TCR/ CD3 and Ig/Ig(~/1g-/3 proteins in the gel.

“3 T Q 25 -bb -

4.3. Staining of Jurkat cells with soluble Z&I-Zg Fig. 3a shows that Jurkat cells stained positive when incubated with purified Ig-p-Ig and a FITC-labeled secondary antibody directed against human IgG. The staining of CD3e with the mAb SP34 served as a positive control. The incubation with Ig-a-Ig and the corresponding secondary antibody resulted in a very minor staining. To determine whether the staining we obtained with Ig-/?-Ig was specific we undertook the experiments shown in Fig. 3b. Here, we tried to inhibit the staining. We therefore incubated Jurkat cells prior to an incubation with Ig-p-Ig with the antibodies SP34 and IC7.2, which are directed against the EC domain of CD3e. SP34 can be used to stain T cells positive for CD3e (lower panel of Fig. 3a,b) whereas IC7.2 can only be used to immunoprecipitate CD3e protein but does not stain T cells (presumably it does not recognize CD36 in the context of the whole TCR/CD3 complex) (see Fig. 3b, lower panel). Along the same lines we incubated purified Ig+Ig with the antibody CB3-1 prior to an incubation with Jurkat cells. CB3-1 recog100

c3)m TT c2 tl --b&

zoo9669-

4.2. Construction and isolation of soluble Zg-p-Zg and Zg-a-Zg fusion proteins To investigate whether the EC domain of Ig-/3 can also interact with the CD3e protein in the context of the whole TCR/CD3 complex on the surface of T cells we set out to stain the Jurkat T cell line with a soluble Ig-/3 fusion protein. A soluble Ig-a protein served as a control throughout the following experiments. The lower part of Fig. 2 shows the domain structures of our soluble fusion proteins. The EC domains of Ig-cu and Ig-p were joined to the human IgGl CHl, hinge (H), CH2 and CH3 constant domains. The resulting proteins were Ig-a-Ig and Ig-p-Ig, respectively. The upper part of Fig. 2 shows an SDS-PAGE of immunoprecipitates of the supernatants of metabolically labeled Cos cells transfected with the constructs encoding either Ig-cy-Ig or Ig-p-Ig. The soluble fusion proteins still contain 2 intact cysteins in their respective hinge regions so that they behave like Ig molecules which are secreted as homodimers (as shown in the non-reduced precipitate).

R

46-

30-

kDa _

1

Y

Y

Y

Ig-P-EC

CH2

CH3 ]

CH2

CH3 1 tg-a-lg

DPE

l!J-ws

YYYYYY 1

Ig-n-EC OPE

Fig. 2. Immunoprecipitates of soluble Ig-p -1g and Ig-a -1g proteins. Cos cells were transiently transfected with the cDNAs encoding Ig-p-Ig and Ig-a-Ig. After labeling the cells metabolically the supematants were collected and the soluble fusion proteins were immunoprecipitated with antibodies directed against the human IgG Fc region. The precipitates were then run on an SDS-PAGE under non-reducing (NR) and reducing (R) conditions. The lower part outlines the domain structures of Ig-/3-Ig and Ig-a-Ig. The EC domains were fused to the C-terminal 3 amino acids (DPE) of the human IgCl CHl, the hinge (H), CH2 and CH3 domains. The glycosilation sites of the EC domains are indicated by Y.

nizes the EC domain of Ig-p and thus we assume it also binds to soluble Ig-p. However, it does not stain Jurkat T cells (see Fig. 3b, lower panel). The upper panel of Fig. 3b shows that a pre-incubation of Jurkat cells with SP34 almost completely inhibited a subsequent staining with Ig+Ig whereas the incubation with IC7.2 hardly had any effect. The staining was also drastically reduced by a pre-incubation of Ig-/3-Ig with CB3-1. Thus we could show, that the staining of Jurkat cells with Ig-p-Ig was specific in the sense that it could be inhibited by masking either Ig-/? or CD3e.

5. Discussion The proliferation and differentiation of resting and activated B cells requires contact-dependent interaction

Log

SP34

Fluorescence

IC7.2

Log

Fluorescence

CB3-1

.

Fig. 3. FACS analysis of the staining of Jurkat cells with soluble Ig-p-Ig. a: results of the staining with soluble Ig-p-Ig, the mAb SP34 directed against CD3r, and Ig-a -1g. b: results of the inhibition of the staining by pre-incubation of the Jurkat cells with mAbs SP34 and IC7.2 and pre-incubation of Ig-p-Ig with mAb CB3-1, respectively. In comparison the staining of Jurkat cells with mAbs SP34, IC7.2 and CB3-1 is shown (for further explanation see text).

with T, cells. This interaction is regulated by cytokines and by non-specific physical interactions with T, cells mediated by cell adhesion molecules. Involvement of those adhesion molecules results in costimularory signals which are essential but in itself do not induce a response [19-221. So far, a number of pairs of surface adhesion molecules have been implicated in B-/T-cell interactions. Among them CD40/gp39 and B7-1 (also called CDSO)/ CD28 or B7-l/ CTLA-4. Of those T-cell surface markers, only CD28 can be found on resting T cells. Gp39 as well as Cl-LA-4 are expressed on activated T cells. Thus, interactions via CD40/gp39 and B7-l/ CTIA-4 ensure a response to activated T cells which results in proliferation and class switch for the B cell. Of the B-cell markers, B7-1 is expressed on activated cells and through the interaction with CD28 a T cell will respond to an activated B cell. This results in proliferation and cytokine production of the T cell. Our results show that TCR-associated CD3e and Ig-p strongly associate with each other when transiently co-expressed in Cos cells. This association is mediated by the EC domains as illustrated in Fig. 1. The CD3e chain not only coprecipitates the native Ig-/3 protein but also a protein comprising the Ig-/3 EC domain alone. Also, an Ig-@-l-chimera containing the Ig-p EC and the l-TM/CT domains is coprecipitated whereas the native t chain is not (data not shown). Strongest evi-

dence for the interaction through the EC domains is provided by the coprecipitation of the Ig-/3 EC with the CD3e EC domain. All of the minor interactions we saw between Ig-a! and TCR/CD3 chains and between Ig-p and TCR or CD3y and -6 we interpreted as unspecific and possibly due to some generic interaction between proteins containing Ig domains. For the TCR/CD3 proteins it was shown that the formation of subcomplexes is mediated by an interaction between a positively charged amino acid located in the TM domains of TCRa or -p chains and a negatively charged amino acid in the TM domains of CD3 chains or 5 [12,13]. Our finding that CD3e and Ig-p associate via their EC domains suggests a novel idea which implies the possibility of an interaction even if CD3e is expressed on the surface of a T- and Ig-p on the surface of a B cell. Thus, their interaction would make them adhesion molecules possibly playing a role in facilitating and/or enhancing the association between T and B cells. To investigate the CD3 E/ Ig-/3 interaction on a cellular level, we stained whole Jurkat T cells with a soluble Ig-p protein which we isolated as a Ig-P-IgGl fusion protein (see Figs. 2 and 3a). The Jurkat cells stained brightly with Ig-p-Ig yet were negative for Ig-a-Ig. To further check for the specificity of these stainings, we performed inhibition-of-staining experiments by blocking CD3e on the T cell with an antibody directed against the EC domain (shown in Fig. 3b). Along the same lines we also incubated soluble Ig-fi with an antibody against Ig-P-EC domain prior to the staining of Jurkat cells. The blocking of CD3e EC domain by SP34, which can also be used to stain T cells, was almost complete. There was no blocking using IC7.2, which was expected since it does not stain T cells either, yet it served as a control for any mouse antibody added to the system. Blocking the Ig-p EC domain did also very effectively but not completely inhibit the staining of Jurkat cells, which we attribute to a molar excess of soluble Ig-/3 protein over the CB3-1 antibody. Thus we can say that the staining of Jurkat cells with soluble Ig-/3 protein is specific and can be inhibited by masking either the CD3e- or the Ig-P-EC domains. In summary, our staining data together with the interaction shown between the CD3e and Ig-p EC hint at a novel pair of surface molecules involved in adhesion. But, is there at all room or need for a novel pair of adhesion molecules like CD3~/1g-P? They both are members of the surface receptor complexes which recognize antigen. The TCR/ CD3 complex itself mediates the antigen-specific interaction between T and B cells and, thus, any additional function for Ig-p and CD3e is bound to be minor. Secondly, their expression is inde101

STEP 1

TC 5

STEP 2 gp

39iCD40

CD28,CTLA-4/B7-1 CTLA-4/B7-2 CD2/LFA-3 LFA-l/ICAM-1 Fig. 4. Model for T-/B-cell interactions. Step 1: antigen-specific interaction between a CD4+ T cell and an antigen-presenting B cell which is mediated via the antigen. Step 2: antigen-independent interaction which is mediated via the BCR and TCR complexes, respectively. Additional surface molecules which act in pairs to enhance cell-cell interactions are listed. For further discussion see text.

pendent of the respective activation stage which means that they are constitutively expressed on T and B lymphocytes once a certain developmental stage is passed. Thus they can hardly add to any fine-tuned regulation. Yet it is possible to envision another scenario, which is illustrated in Fig. 4. In a first step T and B cells interact via the TCR/CD3/ CD4 and Ag/ MHC complexes, respectively. Once this antigen-specific association is made, the more unspecific interaction between CD3e and Ig-p prolongs and/or enhances T-/B-cell interactions to allow for optimal help. It is also conceivable that the unspecific CD3e/Ig-/3 interaction takes place as a first contact which is strengthened as soon as TCR/CD3/CD4 have bound the Ag/MHC complex. Without the antigen-specific interaction both cells would dissociate again. This notion of a novel function for CD3e and Ig-p gains support by the conservation found between the amino acid sequences of human and mouse EC which is 60% for Ig-/3 and 53% for CD3e

ml. To gain further insight into a possible involvement of CD3e and Ig-/3 in T-/B-cell interactions or costimulation, we are planning binding experiments like those done to show the interaction between CD4 and MHC II proteins [23]. Similarly, we will express an isolated Ig-/3 or CD3e protein on the surface of a fibroblast line and then test for the binding of whole T and B cells, respectively. Thus we hope to obtain a model to study 102

T-/B-cell interaction and some aspects of the costimulation taking place during humoral immune responses in vivo. A major focus of research in the future will certainly be the interference with costimulation because of its potential usefulness for the treatment of antibody-mediated diseases. Acknowledgements We would like to thank T. Wileman for generously providing the CD3e expression vector and for permanent encouragement. We would also like to thank C. Berek for critically reading the manuscript. And we are very greatful to T. Nakamura and M. Cooper for provid ing the CB-3 antibody. This research project was supported by an NIH grant (IDCRC ~1-31541). References HI Samelson, L.E., Harford, J.B. and Klausner, R.D. (1985) Cell 43, 223.

l21Oettgen,

H.C., Petty, C.L., Maloy, W.L. and Terhorst, C. (1986) Nature 320, 272. [31 Exley, M., Terhorst, C. and Wileman, T. (19911 Seminars Immunol. 3, 28. [41Hombach, J., Leclercq, L., Radbruch, A., Rajewsky, K. and Reth, M. (19881 EMBO J. 7, 3454.

[5] Campbell, KS. and Cambier, J.C. (1990) EMBO J. 9, 441. [61 Van Noesel, C.J.M., Borst, J., De Vries, E.F.R. and van Lier, R.A.W. (1990) Eur. J. Immunol. 20, 2789. [7] Van Nocsel, C.J.M., van Lier, R.A.W., Cordell, J.L., Tse, A.G.D., van Schijndel, G.M.W., de Vries, E.F.R., Mason, D.Y. and Borst, J.B. (1991) J. Immunol. 146, 3881. [8] Wileman, T., Carson, G.R., Shih, F.F., Concino, M.F. and Terhorst, C. (1990) Cell Reg. 1, 907. 191 Bonifacino, J.S., Cosson, P., and Klausner, R.D. (1990) Cell 63, 503. [lo] Mueller, B., Cooper, L., and Terhorst, C. (1992) Eur. J. Immunol. 22, 1621. [ll] Reth, M. (1989) Nature 338, 383. [12] Wileman, T., Carson, G.R., Concino, M., Ahmed, A. and Terhorst, C. (1990) J. Cell Biol. 110, 973. [13] Hall, C., Berkhout, B., Alarcon, B., Sancho, J., Wileman, T. and Terhorst, C. (1991) Int. Immunol. 3, 359.

[14] Takebe, Y., Se&i, M., Fujisawa, J., Hoy, P., Yokota, K., Arai, K., Yoshida, M. and Arai, N. (1988) Mol. Cell. Biol. 8, 466. [15] Aruffo, A., Stamenkovic, M.M., Underhill, C.B. and Seed, B. (1990) Cell 61, 1303. [16] Laemmli, U. (1970) Nature 227, 680. [17] Pessano, S., Oettgen, H.C., Bhan, A.K. and Terhorst, C. (1985) EMBO J. 4, 337. [18] Nakamura, T., Kubagawa, H. and Cooper, M.D. (1992) Proc. Natl. Acad. Sci. USA 89, 8522. [19] Parker, D.C. (1993) Ann. Rev. Immunol. 11, 331. [20] Clark, E.A. and Ledbetter, J.A. (1994) Nature 367, 425. [21] Hathcock, KS., Laszlo, G., Pucillo, C., Linsley, P. and Hodes, R.J. (1994) L. Exp. Med. 180, 631. [22] Noelle, R.J., Ledbetter, J.A. and Aruffo, A. (1992) Immunol. Today 13, 431. [23] Doyle, C. and Strominger, J.L. (1987) Nature 330, 256.

103