Molecular Basis of T Cell Inactivation by CTLA-4

See allHide authors and affiliations

Science  18 Dec 1998:
Vol. 282, Issue 5397, pp. 2263-2266
DOI: 10.1126/science.282.5397.2263


CTLA-4, a negative regulator of T cell function, was found to associate with the T cell receptor (TCR) complex ζ chain in primary T cells. The association of TCRζ with CTLA-4, reconstituted in 293 transfectants, was enhanced by p56lck-induced tyrosine phosphorylation. Coexpression of the CTLA-4–associated tyrosine phosphatase, SHP-2, resulted in dephosphorylation of TCRζ bound to CTLA-4 and abolished the p56lck-inducible TCRζ–CTLA-4 interaction. Thus, CTLA-4 inhibits TCR signal transduction by binding to TCRζ and inhibiting tyrosine phosphorylation after T cell activation. These findings have broad implications for the negative regulation of T cell function and T cell tolerance.

CTLA-4 is a T cell activation molecule essential for normal homeostasis of T cell reactivity. Engagement and cross-linking of CTLA-4 blocks production of interleukin-2, cell cycle progression, and cell differentiation, whereas in vivo blockade of CTLA-4–B7 interaction enhances autoreactive and tumor-specific T cell activity (1). Although it has been proposed that CTLA-4 affects signals downstream of initial T cell signaling events, several lines of evidence suggest that the negative signaling may occur at the T cell “activation cap” (2). Therefore, we investigated whether engagement of CTLA-4 directly affects proximal events of TCR-induced signaling pathways.

Primary T cells were activated for 2 days with monoclonal antibodies (mAbs) to CD3 and CD28 for optimal CTLA-4 expression and then rested to maximize anti-CD3-mediated signaling events. CTLA-4 cross-linking during restimulation with anti-CD3 mAbs resulted in decreased tyrosine phosphorylation of multiple intracellular proteins migrating between 18 and 40 kD (Fig. 1). Immunoblot analyses demonstrated that the affected proteins migrating at 18 to 23 kD represented the TCRζ chains, whereas the protein migrating at 36 kD was LAT (linker for activation of T cells) (3), an adaptor molecule critical for TCR signaling. In addition, tyrosine phosphorylation of mitogen-activated protein kinases stimulated by the anti-CD3 mAbs was reduced after CTLA-4 cross-linking (4). These data suggest that CTLA-4 can inhibit early TCR signaling events within the TCR complex.

Figure 1

Anti-CD3–induced LAT and TCRζ tyrosine phosphorylation is inhibited by anti-CTLA-4 engagement. Activated T cells (5 ×106) (19) were admixed with 293 cells (2.5 × 106) transiently expressing a membrane-bound single-chain mAb to CD3 in the presence or absence of a membrane-bound single-chain mAb to CTLA-4 (20). The cells were incubated at 37°C for the time indicated and then subjected to lysis in buffer (LB). Immunoprecipitates (IPs) were prepared with mAb FB2 to phosphotyrosine (pTyr), rabbit antiserum to LAT (3), or mAb H146-968 to TCRζ (21, 22). IPs separated on a reducing SDS–12% polyacrylamide gel were transferred to a polyvinylidene difluoride membrane (Millipore, Bedford, MA) and subsequently immunoblotted (IB) with mAb 4G10 to pTyr (Upstate Biotechnology Inc., Lake Placid, NY). Bound proteins were detected by chemiluminescence (Pierce, Rockford, IL). CTLA-4 cross-linking resulted in a consistent but incomplete reduction of tyrosine phosphorylated p23TCRζ and LAT. Results are representative of three independent experiments.

To define the molecular mechanism by which CTLA-4 affected TCR signaling events, we analyzed anti-CTLA-4 immunoprecipitates, prepared from metabolically labeled and activated T cells, by two-dimensional SDS–polyacrylamide gel electrophoresis (PAGE) to identify interacting proteins (Fig. 2A). The bands migrating at 32 kD (nonreduced) and16 kD (reduced) were demonstrated to be TCRζ based on immunoblot analyses with antibodies to TCRζ (Fig. 2B) (5), which suggests that CTLA-4 was associated with TCRζ in these cells. The specificity of CTLA-4–TCRζ binding was confirmed as the 16-kD band was absent in anti-CTLA-4 immunoprecipitates prepared from CTLA-4-deficient T cells (Fig. 2C) and could be specifically blocked by the addition of CTLA-4 immunoglobulin during immunoprecipitation (4, 6).

Figure 2

(A) CTLA-4 coprecipitation of metabolically labeled proteins. Activated T cells (20 × 106 cells) (23) were labeled with35S-methionine and 35S-cysteine (Amersham, Arlington Heights, IL). CTLA-4 was immunoprecipitated from cell lysates with mAb UC10-4F10 to CTLA-4 (15). Immune complexes were subjected to two-dimensional SDS-PAGE analysis with 10% acrylamide gels in both the first (nonreduced) and second (reduced) dimensions (5). Arrows indicate proteins coprecipitated with UC10-4F10. (B and C) Identification of p16 TCRζ by immunoblotting. Activated T cells (100 × 106 cells per condition) (23) were lysed and subjected to immunoprecipitation with mAb UC10-4F10 to CTLA-4 or mAb H146-968 to TCRζ. (B) IPs run on two-dimensional SDS–polyacrylamide gel (10% nonreduced, 12% reduced) were subjected to immunoblotting with a rabbit antiserum to TCRζ (Ab 387) (24). The bound proteins were visualized by chemiluminescence. Upper bands in the anti-ζ IP (right) represented higher phosphorylated forms of TCRζ and TCRη. (C) Anti-CTLA-4 IPs prepared from CTLA-4 wild-type (WT) or knockout (KO) mice were separated by one-dimensional reducing SDS-PAGE and immunoblotted with rabbit antiserum to ζ (Ab 387). Ab(L) represents antibody light chain. Independent experiments with similar results were performed five times.

The interaction of CTLA-4 with TCRζ in activated T cells is likely to be complex and may require additional T cell–specific proteins. Therefore, we examined the CTLA-4–TCRζ association in a non–T cell transfection system. Human embryonic kidney epithelial (293) cells were transiently transfected with a plasmid containing murine TCRζ and one encoding murine CTLA-4. The 16-kD TCRζ chain was coprecipitated with CTLA-4 (Fig. 3A). The TCRζ association was specific as mAb to CTLA-4 did not precipitate the 16-kD protein from cells transfected with vector alone or with a truncated form of CTLA-4 lacking the cytoplasmic tail. The importance of the CTLA-4 tail in the TCRζ interaction was confirmed by coexpressing a construct encoding a concatamer of three cytoplasmic tails of CTLA-4 fused to a glutathione S-transferase [GST-(CTLA-4)3] with TCRζ (Fig. 3B). To ensure that the interaction of TCRζ with CTLA-4 occurred at the cell membrane, we examined CTLA-4–TCRζ association in 293 cells transfected with a cell surface–expressed chimeric Tac–ζ construct (extracellular and transmembrane domains of Tac fused to the cytoplasmic domain of TCRζ) (7). A significant amount of Tac–ζ, migrating between 43 and 60 kD, was coprecipitated with CTLA-4 in these transfectants (Fig. 3C). Thus, the association of CTLA-4 with TCRζ can occur at the membrane and does not depend on other T cell–specific proteins.

Figure 3

Analysis of CTLA-4-TCRζ association in 293 cell transfectants. (A) Coimmunoprecipitation of CTLA-4 with TCRζ. 293 cells were transiently transfected with cDNA constructs encoding murine CTLA-4 WT, cytoplasmic domain-deficient (tailless) (12) CTLA-4, or CTLA-4 tyrosine double mutant (Y201F/Y218F) (25) in the absence or presence of WT p56lck and murine TCRζ (25). Cells were incubated for 40 hours and lysates were prepared. CTLA-4 was precipitated from an equal amount of protein [700 μg per IP (left); 300 μg per IP (right)]. IPs were analyzed by reducing SDS-PAGE and immunoblotted with the antibodies listed. The expression vectors used were pCDNA3 (CTLA-4 and TCRζ, 2.5 μg each) and pEF (p56lck, 1.25 μg). (B) GST–(CTLA-4)3 fusion protein binds TCRζ. 293 cells were transiently transfected with cDNA encoding GST–(CTLA-4)3(12) in the presence or absence of p56lck and TCRζ. GST proteins were precipitated from the lysates (200 μg per sample) with glutathione Sepharose beads (12.5 μl, Pharmacia) and analyzed by electrophoresis and immunoblotting with either rabbit antiserum to TCRζ (Ab 387), or GST mAb (Santa Cruz, Santa Cruz, CA). GST and GST–(CTLA-4)3 proteins migrated at 25 and 40 kD, respectively. (C) CTLA-4 coprecipitates Tac–ζ in 293 transfectants. 293 cells were transiently transfected with cDNAs for Tac–ζ chimera (7) in the absence or presence of WT CTLA-4. CTLA-4 IPs were subjected to SDS-PAGE and immunoblotted with Ab 387. Chimeric Tac–ζ, migrated as a doublet when detected with Ab 387. Ab(H) represents antibody heavy chain. (D) Kinase-defective lck (KA) failed to enhance TCRζ association. 293 cells were transiently transfected with cDNAs for WT CTLA-4, murine TCRζ, and either WT or kinase-defective p56lck(26) and analyzed as described above.

Previous studies have shown that p56lck can regulate phosphorylation of both CTLA-4 and TCRζ (8). Therefore, we examined the effect of p56lck on CTLA-4–TCRζ association. Cotransfection of p56lckresulted in increased CTLA-4–TCRζ association, especially with the higher, phosphorylated form (p18) of TCRζ (Fig. 3A). p56lck also enhanced TCRζ binding to the GST-(CTLA-4)3 fusion protein (Fig. 3B). A kinase-defective mutant p56lck (K273A, KA) was not able to enhance CTLA-4–TCRζ binding and failed to recruit p18 TCRζ to CTLA-4 (Fig. 3D). These results suggest that the tyrosine kinase activity of p56lck was required for enhanced association of CTLA-4 with TCRζ. Two tyrosines (Y201 and Y218) located in the cytoplasmic domain of CTLA-4 are substrates for src family tyrosine kinases (8). Thus, the role of these tyrosines in mediating TCRζ association was examined in cells expressing a tyrosine-deficient [Tyr201 to Phe/Tyr218 to Phe (Y201F/Y218F)] CTLA-4 double mutant. The mutant CTLA-4 bound to TCRζ to a similar extent as wild-type CTLA-4 (Fig. 3A), which suggests that CTLA-4 tyrosine phosphorylation was not required for TCRζ binding. Therefore, the enhancement of TCRζ–CTLA-4 association by p56lck likely depends on tyrosine phosphorylation of TCRζ, although p56lck-mediated phosphorylation of other unidentified molecules may be involved.

CTLA-4 can also bind to a tyrosine phosphatase, SHP-2 (9). Thus, we explored the possibility that CTLA-4 formed complexes with SHP-2 and TCRζ in activated T cells, which accounts for the lack of p18 TCRζ in CTLA-4 immunoprecipitates (Fig. 4A). SHP-2 coprecipitated with antibodies to both TCRζ and CTLA-4. Reciprocal immunoprecipitation of SHP-2 and subsequent immunoblotting with mAbs to TCRζ or CTLA-4 as well as transfection studies (Fig. 4B) confirmed the existence of a multimolecular complex of CTLA-4–SHP-2–TCRζ. Anti-CTLA-4 immunoprecipitates prepared from cells transfected with CTLA-4 and TCRζ contained small but detectable amounts of endogenous SHP-2 present in 293 cells that were significantly enhanced by cotransfecting p56lck (Fig. 4B). Overexpression of SHP-2 increased the amount of SHP-2 coprecipitated with CTLA-4. However, SHP-2 overexpression eliminated the binding of p18 TCRζ to CTLA-4. The failure to observe p18 TCRζ binding to CTLA-4 in SHP-2–overexpressing cells was not due to alterations in protein expression, as similar amounts of CTLA-4 (4) and TCRζ were detected. Likewise, SHP-2 overexpression did not alter the overall amounts of the p18 form of TCRζ within the cell, as equal amounts of p18 TCRζ were present in SHP-2–transfected and control cells. Similar results were obtained with the cells expressing the Tac–ζ chimera (4).

Figure 4

SHP-2 associates with CTLA-4–TCRζ complexes and regulates binding of TCRζ to CTLA-4. (A) Equal numbers (100 × 106) of activated T cells (23) were lysed and subjected to IP with mAbs to CTLA-4 (UC10-4F10), rabbit antiserum against SHP-2 (27), and mAb to TCRζ (H146). IPs were analyzed by SDS-PAGE and immunoblotted with a rabbit antiserum to SHP-2 (27), goat polyclonal antibodies to CTLA-4 (Q20, Santa Cruz, Santa Cruz, CA), or rabbit antiserum to TCRζ (Ab 387). (B) 293 cells were transiently transfected with either WT CTLA-4 or mutated CTLA-4 (Y201F/Y218F) and TCRζ in the presence or absence of p56lck or SHP-2. CTLA-4 IPs prepared from lysates (400 μg per sample) or whole cell lysates (35 μg per lane) were electrophoresed and immunoblotted with rabbit antibody to SHP-2 (27) or antiserum to TCRζ (Ab 387). Results are representative of three separate experiments.

Similar to wild-type CTLA-4, the tyrosine mutant CTLA-4 could interact with SHP-2 (Fig. 4B) and abolished p18 TCRζ binding to CTLA-4. These results are in contrast to the reported dependence of the in vitro interaction between CTLA-4 and SHP-2 on phosphotyrosines and SH2 domains (9). The difference may be due to the in vivo rather than in vitro analysis and suggests that there may be a phosphotyrosine-dependent as well as -independent association of CTLA-4 and SHP-2. As CTLA-4 does not possess SH2 domains, it is possible that TCRζ binding to CTLA-4 is indirect and may depend on SHP-2 or the phosphotyrosine-dependent non-SH2 domain binding in CTLA-4.

A model can be envisioned to explain the process by which CTLA-4 regulates T cell responses. T cell activation, initiated by TCR ligation and CD28 costimulation, results in the recruitment of p56lck to the TCR cap and phosphorylation of multiple substrates including TCRζ, ZAP-70, and LAT. CTLA-4, newly expressed (1) or preexisting in resting cells (10), is exported to the cell surface and binds to B7 molecules present in antigen presenting cells (11). CTLA-4 membrane localization may be facilitated by colocalization to the site of TCR engagement through direct interaction with TCRζ and p56lck-induced tyrosine phosphorylation of CTLA-4 (12). The interaction of TCRζ and CTLA-4 brings the phosphatase, SHP-2, into the complex where it promotes TCRζ dephosphorylation either directly by acting on TCRζ or indirectly by regulating p56lck kinase activity. In this regard, several studies have demonstrated a relationship between src family kinase activity and SHP-2 function (13). Currently, the molecular nature of TCRζ–CTLA-4 binding is not known. Given the fact that p56lck enhanced TCRζ binding to the Y201F/Y218F CTLA-4 mutant, CTLA-4–TCRζ interaction most likely depends on tyrosine phosphorylation of TCRζ or other molecules that might be critical for their association. Further studies will help to elucidate the molecular basis of CTLA-4–TCRζ binding. Finally, previous studies have shown that the ordered phosphorylation of TCRζ establishes the threshold for T cell activation (14). The ability of CTLA-4 to bind to and dephosphorylate the p23 form of phosphorylated TCRζ may decrease the extent and duration of TCRζ phosphorylation and thereby antagonize TCR signal transduction. Further studies are needed to elucidate the functional importance of CTLA-4 in tolerance and antagonist peptide activity. However, the current studies provide a conceptual framework for developing approaches to regulate T cell function through CTLA-4.

  • * These authors contributed equally to this work.

  • To whom correspondence should be addressed. E-mail: jbluest{at}


View Abstract

Navigate This Article