Ligation of CD28 or CTLA-4 with some biologicals can activate T cells due to an unexpected superagonist or inverse agonist activity, respectively. The risk of such an outcome limits the therapeutic development of these reagents. Thus, identifying the molecular determinants of superagonist/inverse agonist properties for biologicals targeting costimulatory/inhibitory receptors has not only fundamental value but also important therapeutic implications. In this study, we show that ligation of CTLA-4 with either soluble B7.1 Ig (but not B7.2 Ig) or with a recombinant bispecific in-tandem single chain Fv known as 24:26 induces TCR-independent, T cell activation. Such an inverse agonist activity requires CD28 expression and high CTLA-4 expression and is not seen when CTLA-4 is ligated by membrane-bound B7.1 or B7.2. At the molecular level, the inverse agonist activity of B7.1 Ig or 24:26 correlates with their ability to induce the formation of unique dimer-based, CTLA-4 oligomers on the T cell surface and involves CTLA-4 signaling through its cytoplasmic domain. Our results provide a potential mechanism to explain and to predict inverse agonist activity for CTLA-4 ligands.

Receptors CD28 and CTLA-4 are members of a family of receptors, derived from an ancestral common gene, that respectively costimulate or inhibit T cells in the context of TCR-dependent signaling (1). Despite their opposite functions and under unique circumstances, ligation of CD28 or CTLA-4 with biologicals or with recombinant Ab fragments can, by itself, induce T cell activation in a TCR-independent manner. For CD28, this has been shown with a subset of Abs known as superagonist Abs (2, 3, 4). For CTLA-4, in contrast, historical data had claimed the ability of some Abs to have inverse agonist properties (5, 6, 7, 8), but this has only recently been demonstrated using a bispecific, in-tandem single-chain Fv (ScFv)3 fragment against CTLA-4 known as 24:26 (9).

The pharmacological targeting of CD28 and CTLA-4 as an immunomodulatory strategy to down-regulate autoimmune responses and allograft rejection has generated a lot of interest (10, 11). However, the risk of inducing massive T cell activation, as recently documented in a phase I clinical trial with the CD28 superagonist Ab TGN1412 (12), limits the pharmacological development of these reagents. It would therefore be useful to have the capacity to predict whether a given biological targeting CD28 or CTLA-4 has superagonist or inverse agonist properties. To address this need, we embarked in a study to identify molecular features associated with the inverse agonist properties of CTLA-4 ligands. In this study, we show that the bispecific ScFv 24:26 and, surprisingly, soluble B7.1 Ig act as inverse agonists of CTLA-4 and induce IL-2 production in the absence of concomitant ligation of the TCR. Such an inverse agonist activity is not seen for B7.2 Ig or when B7.1 and B7.2 molecules are membrane bound. Furthermore, the inverse agonist properties of these reagents require CD28 expression and high surface expression of CTLA-4 and correlate with the formation of a unique dimer-based, CTLA-4 oligomer that signals through the cytoplasmic tail of CTLA-4. Based on our findings, we propose that the ability of a CTLA-4-targeting biological to induce the formation of dimer-based oligomers correlates with its inverse agonist properties. Assessing such capacity may minimize the risk and cost of further therapeutic development.

Jurkat E6.1 cells were obtained from American Type Culture Collection. The doxycycline-inducible, CTLA-4 stably transfected Jurkat T cell panel used for these studies has been extensively described (9, 13, 14, 15, 16, 17). The experiments reported here were done with wild-type CTLA-4 and with Y165F CTLA-4, a fully functional molecule that is expressed at high levels on the cell surface or with the indicated mutants. The NNCY CTLA-4 mutant was generated using a QuickChange site-directed mutagenesis kit (Stratagene) by adding the Y→F mutation at residue 165 to the previously described NNC CTLA-4 mutant (15). The B lymphoblastoid cell line LG-2 was provided by Dr. E. Long (National Institute of Allergy and Infectious Disease, National Institutes of Health, Bethesda, MD). Cells were cultured in RPMI 1640 medium supplemented with 10% FCS.

The mouse mAbs 11 and 24, the ScFv molecules 24, 26, and 24:26 against human CTLA-4, mouse anti-human CTLA-4–26 F(ab′)2 and B7.2 Ig were obtained from Wyeth Research (9, 13, 14, 15, 16, 17). The following commercially available Abs were used in these studies: a mouse mAb against CD45 (Transduction Laboratories and BD Biosciences), a goat polyclonal antiserum against actin (Santa Cruz Biotechnology), an anti-human CD45 F(ab′)2 (Caltag Laboratories), and an anti-human CD28 Ab (BD Biosciences). A recombinant human B7.1/Fc chimera (B7.1 Ig) was purchased from R&D Systems. PE-labeled anti-B7.2 and PE-labeled IgG1 were purchased from eBioscience. PE-labeled anti-B7.1 was purchased from BD Biosciences. Staphylococcal enterotoxin E (SEE) was purchased from Toxin Technology. Disuccinimidyl suberate (DSS) was purchased from Pierce.

Doxycycline-induced Jurkat E6.1 T cell transfectants (0.1 × 106/group) were cultured with or without 24:26 or a mixture of 24 ScFv and 26 ScFv or SEE and LG-2 cells at the concentrations indicated and plated in triplicate in 96-well plates for 72 or 48 h, respectively, at 37°C (9). IL-2 in supernatants was measured by ELISA (BD Biosciences).

LG-2 cells (1 × 106/group) were washed and stained with PE-labeled anti-B7.1 or anti-B7.2 or isotype- matched control. Samples were then washed in PBS and analyzed by flow cytometry (FlowJo; Tree Star).

Doxycycline-induced, CTLA-4-expressing Jurkat cells (2 × 106/group) were stimulated with or without 24:26, B7.1 Ig, B7.2 Ig, anti-CTLA-4 mAb 24, anti-CTLA-4–26 F(ab′)2, anti-CD45 F(ab′)2 or LG-2 cells pretreated with or without SEE for the indicated times at 37°C. The cells were further treated with or without DSS (2 mM in DMSO) for 30 min at room temperature. Cells were washed and subsequently lysed in standard lysis buffer containing Triton X-100 (1%) at 4°C for 30 min. Cell lysates were collected and an equal volume of lysis buffer was added to the remaining pellet fraction. The pellet fractions were resuspended using a Vibra Cell sonicator (Sonics and Materials). Protein samples were prepared in reducing (containing 2-ME) sample buffer as previously described (15). Protein samples were resolved by SDS-PAGE and analyzed by Western blotting using a digital image analyzer (Alpha Innotech).

Unpaired Student’s t tests were performed using GraphPad Prism software. A difference between groups was considered significant when p ≤ 0.05.

As previously reported, ligation of CTLA-4 with B7 molecules on the surface on an APC results in the inhibition of superantigen-induced T cell responses (9, 13, 14, 15, 16, 17). To corroborate this inhibitory function of CTLA-4, we have extensively used an experimental system based on the response of doxycycline-inducible, wild-type or mutant CTLA-4 stably transfected Jurkat T cells to SEE and APC. In the absence of doxycycline in the culture medium, these T cells mount a very robust IL-2 response to SEE:APC. When doxycycline is added to the medium, these T cells express significant amounts of CTLA-4 in the cell surface and, under these conditions, the amount of IL-2 produced in response to SEE:APC is significantly abrogated (Fig. 1 A).

FIGURE 1.

CTLA-4 can function as an inhibitor and as an activator of T cells depending on the ligand that it engages. A, Y165F CTLA-4 Jurkat T cells were treated with or without doxycycline and further cultured with LG-2 cells in the presence or absence of SEE (0.1 ng/ml) for 48 h. B, Y165F CTLA-4 Jurkat T cells were stimulated with 24:26 or 24 ScFv and 26 ScFv for 48 h at 37°C. C, Doxycycline-induced Jurkat cells transfected with Y165F CTLA-4 (control CTLA-4), M10 CTLA-4 (lacking the 24 binding site), M11 CTLA-4 (lacking the 26 binding site), or M10/M11 CTLA-4 (lacking both 24 and 26 binding sites) were stimulated with 24:26 at the concentrations indicated for 48 h at 37°C. D, Jurkat cells stably transfected with M10, M11, or M10/M11 CTLA-4 mutants were treated with or without doxycycline and cultured with LG-2 cells in the presence or absence of SEE (1.0 ng/ml) as in A. IL-2 production from all experiments was measured from the harvested supernatants by ELISA. **, p < 0.01; ***, p < 0.001.

FIGURE 1.

CTLA-4 can function as an inhibitor and as an activator of T cells depending on the ligand that it engages. A, Y165F CTLA-4 Jurkat T cells were treated with or without doxycycline and further cultured with LG-2 cells in the presence or absence of SEE (0.1 ng/ml) for 48 h. B, Y165F CTLA-4 Jurkat T cells were stimulated with 24:26 or 24 ScFv and 26 ScFv for 48 h at 37°C. C, Doxycycline-induced Jurkat cells transfected with Y165F CTLA-4 (control CTLA-4), M10 CTLA-4 (lacking the 24 binding site), M11 CTLA-4 (lacking the 26 binding site), or M10/M11 CTLA-4 (lacking both 24 and 26 binding sites) were stimulated with 24:26 at the concentrations indicated for 48 h at 37°C. D, Jurkat cells stably transfected with M10, M11, or M10/M11 CTLA-4 mutants were treated with or without doxycycline and cultured with LG-2 cells in the presence or absence of SEE (1.0 ng/ml) as in A. IL-2 production from all experiments was measured from the harvested supernatants by ELISA. **, p < 0.01; ***, p < 0.001.

Close modal

In marked contrast to the above-mentioned results and as we have previously shown in primary human T cells and T cell lines, ligation of CTLA-4 with a bispecific, in-tandem ScFv known as 24:26, by itself, activated T cells as illustrated by the dose-dependent production of IL-2 of CTLA-4-expressing Jurkat T cells (Fig. 1 B). Such a response was not observed without CTLA-4 expression and increased as the surface expression of CTLA-4 increased, as shown with the fully functional Y165F CTLA-4 variant that is expressed at high levels on the cell surface (9). In addition, the 24:26-induced IL-2 response was not observed when using unlinked 24 and 26 ScFvs. Therefore, from a pharmacological point of view, 24:26 acts as an inverse agonist of CTLA-4 because it converts this receptor from an inhibitor to an activator of T cells.

The inverse agonist activity of 24:26 required the presence of both CTLA-4 epitopes recognized by the 24 ScFv and the 26 ScFv, M10 (58ELT60) and M11 (65SICT68) respectively, because the IL-2 production in response to 24:26 was abolished when the M10 or M11 or both epitopes were mutated (Fig. 1,C). However, mutations at either or both of these epitopes did not abolish the inhibitory function of CTLA-4 in response to its coligation with the TCR (Fig. 1 D) (9). Of interest, we found that the inhibitory capacity of CTLA-4 mutants involving the M10 epitope was enhanced over that for the M11 epitope mutant CTLA-4 (inhibition of 50–75% of the IL-2 response vs 30%, respectively). The molecular basis for such an effect on B7-mediated inhibitory function of CTLA-4 is under investigation.

Because the inverse agonist activity of 24:26 required the expression of both CTLA-4 epitopes recognized by this biological, we hypothesized that such an activity would correlate with the formation of a CTLA-4 oligomer. To test this hypothesis, we examined the presence of this oligomer by looking at the changes in CTLA-4 distribution in soluble and insoluble fractions upon ligation with 24:26 (Fig. 2). Immunoblotting of whole-cell lysates from 24:26 stimulated CTLA-4-expressing Jurkat cells revealed a consistent decrease in the level of CTLA-4 compared with that in nonstimulated T cells (Fig. 2,A). The reduction of CTLA-4 (∼30%) from the whole-cell lysate fraction prompted us to examine its redistribution to the remaining insoluble pellet fraction. We found that CTLA-4 was present in the insoluble fraction of cells stimulated with 24:26 (Fig. 2 A). This correlated with a redistribution of surface CTLA-4 from the membrane to the cytoskeletal fraction (data not shown). Neither decreased levels of CTLA-4 from the soluble lysate fraction or redistribution to the pellet were detected when CTLA-4-expressing T cells were stimulated with anti-CTLA-4 26 F(ab′)2, indicating that any bivalent ligation of CTLA-4 cannot induce this oligomerization. In addition, such a unique oligomerization was not observed for CD45 when this receptor was ligated with anti-CD45 F(ab′)2.

FIGURE 2.

The inverse agonist activity of 24:26 correlates with its ability to induce a unique CTLA-4 oligomer. A, 24:26 ligation of CTLA-4 results in a decrease of CTLA-4 from the soluble fraction of whole-cell lysates and an increase in the insoluble fraction. Y165F CTLA-4 Jurkat T cells were induced with doxycycline and stimulated in the absence or presence of 24:26 (10 μg/ml), anti-CTLA-4–26 F(ab′)2 (10 μg/ml), or anti-CD45 F(ab′)2 for 10 min at 37°C. Soluble cell lysate and insoluble fraction samples were prepared and immunoblotted for CTLA-4, CD45, and actin. Densitometry was used to compare unstimulated vs stimulated CTLA-4 levels in the soluble fraction. NS, Nonstimulated; RDU, relative densitometry units. B, Y165F CTLA-4-expressing cells were stimulated with or without 24:26 for 10 min at 37°C, followed by treatment with the chemical cross-linker DSS. Resuspended pellet fractions were prepared and blotted for CTLA-4.

FIGURE 2.

The inverse agonist activity of 24:26 correlates with its ability to induce a unique CTLA-4 oligomer. A, 24:26 ligation of CTLA-4 results in a decrease of CTLA-4 from the soluble fraction of whole-cell lysates and an increase in the insoluble fraction. Y165F CTLA-4 Jurkat T cells were induced with doxycycline and stimulated in the absence or presence of 24:26 (10 μg/ml), anti-CTLA-4–26 F(ab′)2 (10 μg/ml), or anti-CD45 F(ab′)2 for 10 min at 37°C. Soluble cell lysate and insoluble fraction samples were prepared and immunoblotted for CTLA-4, CD45, and actin. Densitometry was used to compare unstimulated vs stimulated CTLA-4 levels in the soluble fraction. NS, Nonstimulated; RDU, relative densitometry units. B, Y165F CTLA-4-expressing cells were stimulated with or without 24:26 for 10 min at 37°C, followed by treatment with the chemical cross-linker DSS. Resuspended pellet fractions were prepared and blotted for CTLA-4.

Close modal

To directly document the oligomerization of CTLA-4 induced by 24:26, we used chemical cross-linkers. The addition of a cell-impermeable, chemical cross-linker of short arm length before cell lysis revealed the presence of CTLA-4 tetramers and higher order oligomers in the insoluble pellet fractions of 24:26-stimulated T cells (Fig. 2 B). Altogether, this evidence indicates that the CTLA-4 inverse agonist 24:26 induces the formation of a unique CTLA-4 oligomer, whose properties are not simply due to bivalent ligation.

CTLA-4 is expressed as a homodimer that is stabilized by a disulfide bridge at the cysteine residue 122 and by dual N-glycosylation at positions 78 and 110 (17). Given the above-mentioned data, we hypothesized that the unique oligomer induced by the inverse agonist 24:26 would require expression of intact CTLA-4 homodimers. To test this hypothesis, we used T cells expressing a monomeric form of CTLA-4 (known as NNCY CTLA-4) that cannot dimerize but is expressed on the cell surface (Fig. 3,A) (17). We observed that 24:26 did not induce the unique CTLA-4 oligomer when T cells expressed monomeric CTLA-4 (Fig. 3 B).

FIGURE 3.

The unique oligomer induced by 24:26 requires the expression of CTLA-4 dimers. A, Jurkat T cells stably transfected with monomeric (NNCY) CTLA-4 were induced with doxycycline and the surface expression was measured by flow cytometry (thick line, CTLA-4; shaded profile, isotype-matched Ab). B, Monomeric CTLA-4 does not form an activating oligomer. NNCY CTLA-4 Jurkat T cells were induced with doxycycline and stimulated in the presence or absence of 24:26 (165 μg/ml) for 30 min at 37°C. Soluble and insoluble fractions were prepared and immunoblotted for CTLA-4 and actin. C, 24:26 does not act as an inverse agonist when engaging monomeric CTLA-4. NNCY CTLA-4-expressing Jurkat T cells were stimulated with or without 24:26 at the concentrations indicated for 48 h. IL-2 production was measured by ELISA. ***, p < 0.001.

FIGURE 3.

The unique oligomer induced by 24:26 requires the expression of CTLA-4 dimers. A, Jurkat T cells stably transfected with monomeric (NNCY) CTLA-4 were induced with doxycycline and the surface expression was measured by flow cytometry (thick line, CTLA-4; shaded profile, isotype-matched Ab). B, Monomeric CTLA-4 does not form an activating oligomer. NNCY CTLA-4 Jurkat T cells were induced with doxycycline and stimulated in the presence or absence of 24:26 (165 μg/ml) for 30 min at 37°C. Soluble and insoluble fractions were prepared and immunoblotted for CTLA-4 and actin. C, 24:26 does not act as an inverse agonist when engaging monomeric CTLA-4. NNCY CTLA-4-expressing Jurkat T cells were stimulated with or without 24:26 at the concentrations indicated for 48 h. IL-2 production was measured by ELISA. ***, p < 0.001.

Close modal

To further confirm that the inverse agonist activity of 24:26 correlated with its capacity to induce the formation of a dimer-based CTLA-4 oligomer, we tested whether T cells expressing the monomeric NNCY CTLA-4 mounted an IL-2 response to 24:26. We found that, in contrast to dimeric CTLA-4-expressing T cells (control CTLA-4), 24:26 failed to induce an IL-2 response in monomeric CTLA-4-expressing T cells (Fig. 3 C). Although monomeric CTLA-4 was not functional as an activator, this molecule was still able to inhibit IL-2 production in response to SEE:APC, as we have previously reported (17). Altogether, these data led us to conclude that the inverse agonist activity of 24:26 correlates with the formation of a unique CTLA-4 dimer-based, high-order oligomer.

If the inverse agonist activity of a CTLA-4 ligand correlates with its ability to form a high-order CTLA-4 oligomer, then one should be able to generate inverse agonists of CTLA-4 by using recombinant versions of its natural ligands that force such an oligomerization. Two natural ligands of CTLA-4 have been reported to date: B7.1 and B7.2 (18, 19, 20). They have distinct profiles of expression and biochemical features, most importantly B7.1 has a higher affinity and avidity for CTLA-4 given its dimeric structure (21, 22, 23). Crystallographic data have suggested that ligation of CTLA-4 with B7.1, itself being a homodimer, may induce a lattice-like oligomer (17, 23, 24). With this in mind, we tested whether a soluble B7.1 and B7.2 (B7.1 Ig, B7.2 Ig) had inverse agonist properties on T cells expressing high levels of CTLA-4 on their cell surface.

We examined the ability of B7 molecules to induce the formation of a high-order CTLA-4 oligomer similar to that induced by 24:26. We found that engagement of CTLA-4 with soluble B7.1 Ig resulted in the formation of such an oligomer as detected by redistribution of CTLA-4 to the insoluble fraction (Fig. 4,A). In contrast, soluble B7.2 Ig was not able to induce the formation of such an oligomer (Fig. 4,A). Membrane-bound B7.1 and B7.2 molecules, expressed on LG-2 cells, as shown in Fig. 4,B, were not able to induce such an oligomer either in the presence or absence of superantigen stimulation (Fig. 4 A).

FIGURE 4.

Soluble B7.1 Ig can act as an inverse agonist of CTLA-4. A, Y165F CTLA-4 Jurkat T cells were treated overnight with doxycycline and further stimulated with 24:26 (100 and 10 μg/ml), B7.1 Ig (10 μg/ml), B7.2 Ig (10 μg/ml), and LG-2 cells in the absence or presence of SEE (10 ng/ml). Soluble and insoluble samples were prepared and immunoblotted for CTLA-4. B, LG-2 cells express B7 molecules. The expression of B7.1 and B7.2 on LG-2 cells was measured by flow cytometry (thick line, B7.1 or B7.2; shaded profile, isotype-matched Ab). C, Y165F CTLA-4 Jurkat T cells were treated with or without doxycycline and further stimulated with or without 24:26 (10 μg/ml), B7.1 Ig (10 μg/ml), B7.2 Ig (10 μg/ml), and LG-2 cells alone for 48 h at 37°C. IL-2 production from the harvested supernatants was measured by ELISA. **, p < 0.01; ***, p < 0.001.

FIGURE 4.

Soluble B7.1 Ig can act as an inverse agonist of CTLA-4. A, Y165F CTLA-4 Jurkat T cells were treated overnight with doxycycline and further stimulated with 24:26 (100 and 10 μg/ml), B7.1 Ig (10 μg/ml), B7.2 Ig (10 μg/ml), and LG-2 cells in the absence or presence of SEE (10 ng/ml). Soluble and insoluble samples were prepared and immunoblotted for CTLA-4. B, LG-2 cells express B7 molecules. The expression of B7.1 and B7.2 on LG-2 cells was measured by flow cytometry (thick line, B7.1 or B7.2; shaded profile, isotype-matched Ab). C, Y165F CTLA-4 Jurkat T cells were treated with or without doxycycline and further stimulated with or without 24:26 (10 μg/ml), B7.1 Ig (10 μg/ml), B7.2 Ig (10 μg/ml), and LG-2 cells alone for 48 h at 37°C. IL-2 production from the harvested supernatants was measured by ELISA. **, p < 0.01; ***, p < 0.001.

Close modal

Based on the ability to induce high-order CTLA-4 oligomerization, we predicted that soluble B7.1 Ig but not B7.2 Ig or membrane-bound B7 molecules would have inverse agonist properties. To corroborate this prediction, we determined the capacity of these molecules to induce T cell activation through CTLA-4. To do this, we cultured Jurkat T cells with or without doxycycline to induce CTLA-4 expression and then stimulated these cells with 24:26 soluble B7.1 Ig and B7.2 Ig or membrane-bound B7 molecules (on LG-2 cells) for 48 h. As shown in Fig. 4 C, IL-2 production was detected upon stimulation with 24:26 and with B7.1 Ig only under conditions of a high level of expression of CTLA-4, while B7.2 Ig and membrane-bound B7 molecules were unable to induce any IL-2 response. The lack of any significant IL-2 response by these T cells without induction of CTLA-4 expression indicated that 24:26 and B7.1 Ig act as inverse agonists of CTLA-4 and not as superagonists of CD28, which is expressed at high levels by these cells. The magnitude of IL-2 production induced by these biologicals likely reflects the differential affinities of these ligands for CTLA-4, being higher for 24:26 than for soluble B7.1 Ig, and this higher than for B7.2 Ig. Altogether, these data show that soluble ligands for CTLA-4 can act as inverse agonists of this receptor and induce T cell activation through an oligomerization-dependent manner.

The inverse agonist response of these biologicals could primarily result from signaling through CTLA-4 or, alternatively, from signaling through another receptor trapped in these oligomers. To distinguish between these two options, we made use of Jurkat T cells expressing doxycycline-inducible CTLA-4 lacking the entire cytoplasmic tail (TLESS CTLA-4). First, we tested whether this biological would induce the unique CTLA-4 oligomer associated with inverse agonist activity. As one would expect based on the preservation of the extracellular domain of the tailless CTLA-4 molecule, we found that TLESS CTLA-4 still oligomerized in response to 24:26, as illustrated by the reduction of TLESS CTLA-4 from the whole-cell lysate fraction of 24:26-stimulated cells and subsequent appearance in the corresponding insoluble fraction (Fig. 5 A). Thus, the oligomerization of CTLA-4 induced by 24:26 is not mediated by interactions through its intracellular domain.

FIGURE 5.

The inverse agonist activity of 24:26 requires the cytoplasmic tail of CTLA-4. A, 24:26 can induce an “activating” oligomer of TLESS CTLA-4. TLESS CTLA-4 Jurkat T cells were stimulated with or without 24:26 for 10 min at 37°C. Soluble and insoluble fractions were prepared and immunoblotted for CTLA-4 and actin. B, 24:26 does not act as an inverse agonist when ligating TLESS CTLA-4. Y165F (control CTLA-4) or TLESS CTLA-4-expressing Jurkat T cells were stimulated with 24:26 for 48 h and IL-2 production was measured by ELISA. ***, p < 0.001.

FIGURE 5.

The inverse agonist activity of 24:26 requires the cytoplasmic tail of CTLA-4. A, 24:26 can induce an “activating” oligomer of TLESS CTLA-4. TLESS CTLA-4 Jurkat T cells were stimulated with or without 24:26 for 10 min at 37°C. Soluble and insoluble fractions were prepared and immunoblotted for CTLA-4 and actin. B, 24:26 does not act as an inverse agonist when ligating TLESS CTLA-4. Y165F (control CTLA-4) or TLESS CTLA-4-expressing Jurkat T cells were stimulated with 24:26 for 48 h and IL-2 production was measured by ELISA. ***, p < 0.001.

Close modal

Next, we tested whether the CTLA-4 oligomer signaled through the cytoplasmic domain of this receptor. This was done by stimulating TLESS CTLA-4-expressing T cells with 24:26 and measuring the IL-2 response. We found that ligation of TLESS CTLA-4 with 24:26 did not induce any IL-2 production (Fig. 5 B). Therefore, we concluded that the inverse agonist activity of 24:26 results from signals relayed through the cytoplasmic tail of CTLA-4 oligomers.

Microarray analysis of genes regulated by CTLA-4 ligation suggests that this receptor inactivates T cells by inhibiting CD28-dependent gene transcription (25). Based on this observation, we hypothesized that, by analogy to its inhibitory function, the inverse agonist response upon ligation of CTLA-4 involved cross-talk between signaling from this receptor and CD28 signaling. Thus, we tested whether the agonist and inverse agonist responses through CTLA-4 required concomitant expression of CD28 by using CD28+ and CD28 variants of CTLA-4-expressing T cells. We found that IL-2 production was inhibited by CTLA-4 in CD28+ T cells. However, we observed that ligation of CTLA-4 in CD28 T cells did not inhibit the IL-2 response to SEE:APC (Fig. 6,A) despite the much reduced level of T cell activation of cells due to the lack of CD28 costimulation. The requirement for CD28 on CTLA-4-mediated inhibition of T cell activation was also applicable to the inverse agonist response through CTLA-4. As shown in Fig. 6 B, the inverse agonist activity of 24:26 was only observed in T cells expressing high levels of CTLA-4 and CD28. Based on these data, we concluded that CD28 expression is required for the function of CTLA-4 both as an inhibitor and as an activator of T cells.

FIGURE 6.

The inhibitory and activating functions of CTLA-4 require the expression of CD28. A, CD28 expression is required for CTLA-4-dependent T cell inactivation. CD28 and CD28+ Jurkat T cells were treated overnight with or without doxycyline (1 μg/ml) to induce the expression of Y165F CTLA-4. T cells were cultured with LG-2 cells (1:5 ratio) and increasing concentrations of SEE for 48 h. IL-2 production was measured by ELISA. B, CD28 expression is required for the inverse agonist activity of 24:26 on CTLA-4. CD28 and CD28+ Jurkat T cells were induced with doxycycline to express Y165F CTLA-4. T cells were cultured in the presence of increasing concentrations of 24:26 for 48 h, and IL-2 production was measured as in A. **, p < 0.01; ***, p < 0.001.

FIGURE 6.

The inhibitory and activating functions of CTLA-4 require the expression of CD28. A, CD28 expression is required for CTLA-4-dependent T cell inactivation. CD28 and CD28+ Jurkat T cells were treated overnight with or without doxycyline (1 μg/ml) to induce the expression of Y165F CTLA-4. T cells were cultured with LG-2 cells (1:5 ratio) and increasing concentrations of SEE for 48 h. IL-2 production was measured by ELISA. B, CD28 expression is required for the inverse agonist activity of 24:26 on CTLA-4. CD28 and CD28+ Jurkat T cells were induced with doxycycline to express Y165F CTLA-4. T cells were cultured in the presence of increasing concentrations of 24:26 for 48 h, and IL-2 production was measured as in A. **, p < 0.01; ***, p < 0.001.

Close modal

The inhibitory function of CTLA-4 on T cell responses is well established in vitro and in vivo (1). However, early experimental data generated in vivo using Abs against CTLA-4 had suggested that, under certain conditions, this receptor can enhance T cell responses (5, 6). The complexity of this system made it difficult to interpret this result. More recently, though, we have been able to demonstrate that CTLA-4 can also function as an activator of T cells. This occurred when CTLA-4, expressed at high levels on the plasma membrane, was ligated with a recombinant, bispecific, Ab fragment called 24:26 (9). The mechanism for this unexpected inverse agonist activity was unknown. In this study, we show that the inverse agonist activity of not only 24:26 but also other soluble CTLA-4 ligands and biologicals correlates with their ability to induce a unique oligomerization state that relays activating signals through the cytoplasmic tail of CTLA-4. This ligand-induced architecture likely provides stability for the interaction of CTLA-4 with key signaling molecules that result in activating signaling and ultimately T cell activation.

The inverse agonist activity through CTLA-4 is not restricted to 24:26. We show here that soluble B7.1 Ig, but not B7.2 Ig, can act as an inverse agonist of CTLA-4. B7 molecules expressed on the surface of an APC are also unable to act as inverse agonists, suggesting a differential ability of soluble and membrane-bound B7.1 and B7.2 to engage CTLA-4, which may correlate with different binding affinities and avidity and different functional outcomes in vivo (1). The architectural constraint of molecules in a lipid bilayer may also provide restrictions in their ability to engage other receptors. Interestingly, a soluble form of B7.1 and B7.2 has been observed (26, 27) and elevated levels of soluble B7.1 and B7.2 are detected in patients with autoimmune diseases such as systemic lupus erythematosus and rheumatoid arthritis (28, 29, 30). Of interest, although we have detected the inverse agonist activity of 24:26 in primary T cells consistently, the inverse agonist activity of B7.1 Ig may be dependent on the ability of primary T cells to up-regulate CTLA-4 at a very high density, a scenario that may occur only in a local microenvironment. Therefore, soluble B7.1 may act in vivo as an inverse agonist of CTLA-4, leading to the activation of potentially autoreactive T cells.

It is important to note that the redistribution of CTLA-4 to the insoluble fraction only represents the ability of a biological to induce the formation of a unique CTLA-4 oligomer, but it is not a quantitative indicator of the ability of that biological to induce T cell activation. As stated above, the magnitude of IL-2 production likely reflects the differential affinities of the ligands for CTLA-4. Because 24:26 is composed of ScFv fragments of two specific Abs against CTLA-4, it likely has a higher affinity for CTLA-4 than soluble B7.1 Ig. The higher affinity may translate into a longer interaction with CTLA-4 allowing for sustained signaling and therefore higher levels of IL-2 production.

As implied by the differential effect of soluble B7.1 Ig vs membrane-bound B7.1, the structure of the activating CTLA-4 oligomer is distinctly different from the lattice-like oligomerization that is necessary for CTLA-4-mediated inhibition (17, 23, 24). The inhibitory lattice-like oligomer is regulated in a hierarchical manner by disulfide bonding, N-glycosylation, and B7 engagement and does not require homodimeric CTLA-4 (17). In addition, it can occur under conditions of low levels of expression of CTLA-4. In contrast, the activating oligomer requires the expression of intact CTLA-4 homodimers because monomeric CTLA-4 does not oligomerize under these conditions. Moreover, the inverse agonist activity of CTLA-4 requires high levels of surface expression of this receptor. Based on these observations, we hypothesize that the basic unit of the activating CTLA-4 oligomer is two CTLA-4 homodimers engaged by a single 24:26 molecule or B7.1 Ig molecule.

In addition to oligomerization, CTLA-4-mediated T cell activation requires the cytoplasmic tail of CTLA-4. Removal of the intracellular domain renders CTLA-4 unable to respond to 24:26 stimulation. Recent data have suggested that the bivalency of the extracellular domain of CTLA-4 can lend activating properties to the tail of CD28 (31). However, our data indicate that the activating function of CTLA-4 is not only dependent on its valency and state of oligomerization but also requires the presence of its cytoplasmic tail. Because this domain does not contain any intrinsic enzymatic activity, the necessity of its expression suggests that key signaling molecules likely associate with the tail to induce T cell activation.

The mechanism proposed here to explain the inverse agonist activity of 24:26 or B7.1 Ig parallels the mechanism by which anti-CD28 superagonist Abs induce T cell activation. Crystallographic studies have revealed a flattening of this superagonistic Ab along the cell membrane, allowing for putative CD28 cross-linking (32, 33). Likewise, the location of the CTLA-4 epitopes recognized by 24:26 reside proximal to the T cell membrane (9), potentially providing a stable, high-avidity interaction between 24:26 and multiple CTLA-4 molecules that would mediate similar cross-linking. The presence of higher order oligomers of CTLA-4 in the insoluble fraction following treatment with 24:26 and DSS, a chemical cross-linker joining molecules closer than 11Å, further suggests that 24:26 simultaneously engages multiple CTLA-4 molecules.

The signaling plasticity of CTLA-4 is in line with the structural similarities between CD28 and CTLA-4 and their common ancestral origin (34, 35, 36). It is tempting to speculate that divergent mutations within each gene likely placed architectural constraints on their B7-binding capabilities, while still having the capacity to signal similarly under unique oligomerization conditions. We show here that the inhibitory and activating function of CTLA-4 requires the expression of CD28. This was somewhat surprising since two reports had claimed that CD28 is not required for CTLA-4 function (37, 38). However, more recent data using transgenic mice overexpressing CTLA-4 have demonstrated that, under some circumstances, CD28 is required for CTLA-4 function (39). The lack of inhibition of T cell responses in the absence of CD28 indicates that the delivery of a negative signal by CTLA-4 acts on the CD28 pathway. Similarly, the activation of T cells through CTLA-4 seems to initiate or enhance CD28 signaling.

The biological significance of the data presented here is twofold. On one hand, it reconciles the historical observations of an activating role for CTLA-4 with the well-established inhibitory function of this receptor by providing a molecular explanation for both inhibitory and activating responses through CTLA-4 ligation. On the other hand, the assessment of the capacity of CTLA-4-targeting ligands to induce an “activating” CTLA-4 oligomer allows one to identify those immunomodulatory biologicals with inverse agonist properties, thus throwing a note of caution in their further therapeutic development and, possibly, avoiding the devastating effects recently seen with the administration of soluble CD28 superagonistic Abs.

We thank Drs. Mary Collins (Wyeth Research), David H. Margulies (National Institute of Allergy and Infectious Diseases, National Institutes of Health), Beatriz Carreno (Washington University, St. Louis, MO), Vincent Ling (Dragonfly Therapeutics), James Riley (University of Pennsylvania, Philadelphia, PA), and the members of the Madrenas laboratory for helpful comments and discussions at different stages of this project.

The authors have no financial conflict of interest.

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.

1

This work was supported by grants from the Canadian Institutes of Health Research and the Kidney Foundation of Canada. W.A.T. holds a Canadian Institutes for Health Research Doctoral Award and J.M. holds a Canada Research Chair in Transplantation and Immunobiology.

3

Abbreviations used in this paper: ScFv, single-chain Fv; SEE, staphylococcal enterotoxin E; DSS, disuccinimidyl suberate.

1
Teft, W. A., M. G. Kirchhof, J. Madrenas.
2006
. A molecular perspective of CTLA-4 function.
Annu. Rev. Immunol.
24
:
65
-97.
2
Siefken, R., R. Kurrle, R. Schwinzer.
1997
. CD28-mediated activation of resting human T cells without costimulation of the CD3/TCR complex.
Cell. Immunol.
176
:
59
-65.
3
Luhder, F., Y. Huang, K. M. Dennehy, C. Guntermann, I. Muller, E. Winkler, T. Kerkau, S. Ikemizu, S. J. Davis, T. Hanke, T. Hunig.
2003
. Topological requirements and signaling properties of T cell-activating, anti-CD28 antibody superagonists.
J. Exp. Med.
197
:
955
-966.
4
Tacke, M., G. Hanke, T. Hanke, T. Hunig.
1997
. CD28-mediated induction of proliferation in resting T cells in vitro and in vivo without engagement of the T cell receptor: evidence for functionally distinct forms of CD28.
Eur. J. Immunol.
27
:
239
-247.
5
Linsley, P. S., J. L. Greene, P. Tan, J. Bradshaw, J. A. Ledbetter, C. Anasetti, N. K. Damle.
1992
. Coexpression and functional cooperation of CTLA-4 and CD28 on activated T lymphocytes.
J. Exp. Med.
176
:
1595
-1604.
6
Wu, Y., Y. Guo, A. Huang, P. Zheng, Y. Liu.
1997
. CTLA-4-B7 interaction is sufficient to costimulate T cell clonal expansion.
J. Exp. Med.
185
:
1327
-1335.
7
Kwon, H., H. S. Jun, L. Y. Khil, J. W. Yoon.
2004
. Role of CTLA-4 in the activation of single- and double-positive thymocytes.
J. Immunol.
173
:
6645
-6653.
8
Pandiyan, P., D. Gartner, O. Soezeri, A. Radbruch, K. Schulze-Osthoff, M. C. Brunner-Weinzierl.
2004
. CD152 (CTLA-4) determines the unequal resistance of Th1 and Th2 cells against activation-induced cell death by a mechanism requiring PI3 kinase function.
J. Exp. Med.
199
:
831
-842.
9
Madrenas, J., L. A. Chau, W. A. Teft, P. W. Wu, J. Jussif, M. Kasaian, B. M. Carreno, V. Ling.
2004
. Conversion of CTLA-4 from inhibitor to activator of T cells with a bispecific tandem single-chain Fv ligand.
J. Immunol.
172
:
5948
-5956.
10
Melero, I., S. Hervas-Stubbs, M. Glennie, D. M. Pardoll, L. Chen.
2007
. Immunostimulatory monoclonal antibodies for cancer therapy.
Nat. Rev. Cancer
7
:
95
-106.
11
Bluestone, J. A., E. W. St Clair, L. A. Turka.
2006
. CTLA4Ig: bridging the basic immunology with clinical application.
Immunity
24
:
233
-238.
12
Suntharalingam, G., M. R. Perry, S. Ward, S. J. Brett, A. Castello-Cortes, M. D. Brunner, N. Panoskaltsis.
2006
. Cytokine storm in a phase 1 trial of the anti-CD28 monoclonal antibody TGN1412.
N. Engl. J. Med.
355
:
1018
-1028.
13
Baroja, M. L., D. Luxenberg, T. Chau, V. Ling, C. A. Strathdee, B. M. Carreno, J. Madrenas.
2000
. The inhibitory function of CTLA-4 does not require its tyrosine phosphorylation.
J. Immunol.
164
:
49
-55.
14
Baroja, M. L., L. Vijayakrishnan, E. Bettelli, P. J. Darlington, T. A. Chau, V. Ling, M. Collins, B. M. Carreno, J. Madrenas, V. K. Kuchroo.
2002
. Inhibition of CTLA-4 function by the regulatory subunit of serine/threonine phosphatase 2A.
J. Immunol.
168
:
5070
-5078.
15
Carreno, B. M., F. Bennett, T. A. Chau, V. Ling, D. Luxenberg, J. Jussif, M. L. Baroja, J. Madrenas.
2000
. CTLA-4 (CD152) can inhibit T cell activation by two different mechanisms depending on its level of cell surface expression.
J. Immunol.
165
:
1352
-1356.
16
Darlington, P. J., M. L. Baroja, T. A. Chau, E. Siu, V. Ling, B. M. Carreno, J. Madrenas.
2002
. Surface cytotoxic T lymphocyte-associated antigen 4 partitions within lipid rafts and relocates to the immunological synapse under conditions of inhibition of T cell activation.
J. Exp. Med.
195
:
1337
-1347.
17
Darlington, P. J., M. G. Kirchhof, G. Criado, J. Sondhi, J. Madrenas.
2005
. Hierarchical regulation of CTLA-4 dimer-based lattice formation and its biological relevance for T cell inactivation.
J. Immunol.
175
:
996
-1004.
18
Linsley, P. S., W. Brady, M. Urnes, L. S. Grosmaire, N. K. Damle, J. A. Ledbetter.
1991
. CTLA-4 is a second receptor for the B cell activation antigen B7.
J. Exp. Med.
174
:
561
-569.
19
Freeman, G. J., F. Borriello, R. J. Hodes, H. Reiser, J. G. Gribben, J. W. Ng, J. Kim, J. M. Goldberg, K. Hathcock, G. Laszlo, et al
1993
. Murine B7-2, an alternative CTLA4 counter-receptor that costimulates T cell proliferation and interleukin 2 production.
J. Exp. Med.
178
:
2185
-2192.
20
Freeman, G. J., J. G. Gribben, V. A. Boussiotis, J. W. Ng, V. A. Restivo, Jr, L. A. Lombard, G. S. Gray, L. M. Nadler.
1993
. Cloning of B7-2: a CTLA-4 counter-receptor that costimulates human T cell proliferation.
Science
262
:
909
-911.
21
Hathcock, K. S., G. Laszlo, C. Pucillo, P. Linsley, R. J. Hodes.
1994
. Comparative analysis of B7-1 and B7-2 costimulatory ligands: expression and function.
J. Exp. Med.
180
:
631
-640.
22
Lenschow, D. J., T. L. Walunas, J. A. Bluestone.
1996
. CD28/B7 system of T cell costimulation.
Annu. Rev. Immunol.
14
:
233
-258.
23
Stamper, C. C., Y. Zhang, J. F. Tobin, D. V. Erbe, S. Ikemizu, S. J. Davis, M. L. Stahl, J. Seehra, W. S. Somers, L. Mosyak.
2001
. Crystal structure of the B7-1/CTLA-4 complex that inhibits human immune responses.
Nature
410
:
608
-611.
24
Schwartz, J. C., X. Zhang, A. A. Fedorov, S. G. Nathenson, S. C. Almo.
2001
. Structural basis for co-stimulation by the human CTLA-4/B7-2 complex.
Nature
410
:
604
-608.
25
Riley, J. L., M. Mao, S. Kobayashi, M. Biery, J. Burchard, G. Cavet, B. P. Gregson, C. H. June, P. S. Linsley.
2002
. Modulation of TCR-induced transcriptional profiles by ligation of CD28, ICOS, and CTLA-4 receptors.
Proc. Natl. Acad. Sci. USA
99
:
11790
-11795.
26
McHugh, R. S., W. D. Ratnoff, R. Gilmartin, K. W. Sell, P. Selvaraj.
1998
. Detection of a soluble form of B7-1 (CD80) in synovial fluid from patients with arthritis using monoclonal antibodies against distinct epitopes of human B7-1.
Clin. Immunol. Immunopathol.
87
:
50
-59.
27
Jeannin, P., G. Magistrelli, J. P. Aubry, G. Caron, J. F. Gauchat, T. Renno, N. Herbault, L. Goetsch, A. Blaecke, P. Y. Dietrich, et al
2000
. Soluble CD86 is a costimulatory molecule for human T lymphocytes.
Immunity
13
:
303
-312.
28
Ip, W. K., C. K. Wong, T. F. Leung, C. W. Lam.
2005
. Elevation of plasma soluble T cell costimulatory molecules CTLA-4, CD28 and CD80 in children with allergic asthma.
Int. Arch. Allergy Immunol.
137
:
45
-52.
29
Wong, C. K., L. C. Lit, L. S. Tam, E. K. Li, C. W. Lam.
2005
. Aberrant production of soluble costimulatory molecules CTLA-4, CD28, CD80 and CD86 in patients with systemic lupus erythematosus.
Rheumatology
44
:
989
-994.
30
Hock, B. D., J. L. O’Donnell, K. Taylor, A. Steinkasserer, J. L. McKenzie, A. G. Rothwell, K. L. Summers.
2006
. Levels of the soluble forms of CD80, CD86, and CD83 are elevated in the synovial fluid of rheumatoid arthritis patients.
Tissue Antigens
67
:
57
-60.
31
Dennehy, K. M., F. Elias, G. Zeder-Lutz, X. Ding, D. Altschuh, F. Luhder, T. Hunig.
2006
. Cutting edge: monovalency of CD28 maintains the antigen dependence of T cell costimulatory responses.
J. Immunol.
176
:
5725
-5729.
32
Evans, E. J., R. M. Esnouf, R. Manso-Sancho, R. J. Gilbert, J. R. James, C. Yu, J. A. Fennelly, C. Vowles, T. Hanke, B. Walse, et al
2005
. Crystal structure of a soluble CD28-Fab complex.
Nat. Immunol.
6
:
271
-279.
33
Hunig, T., K. Dennehy.
2005
. CD28 superagonists: mode of action and therapeutic potential.
Immunol. Lett.
100
:
21
-28.
34
Harper, K., C. Balzano, E. Rouvier, M. G. Mattei, M. F. Luciani, P. Golstein.
1991
. CTLA-4 and CD28 activated lymphocyte molecules are closely related in both mouse and human as to sequence, message expression, gene structure, and chromosomal location.
J. Immunol.
147
:
1037
-1044.
35
Lafage-Pochitaloff, M., R. Costello, D. Couez, J. Simonetti, P. Mannoni, C. Mawas, D. Olive.
1990
. Human CD28 and CTLA-4 Ig superfamily genes are located on chromosome 2 at bands q33–q34.
Immunogenetics
31
:
198
-201.
36
Howard, T. A., J. M. Rochelle, M. F. Seldin.
1991
. Cd28 and Ctla-4, two related members of the Ig supergene family, are tightly linked on proximal mouse chromosome 1.
Immunogenetics
33
:
74
-76.
37
Fallarino, F., P. E. Fields, T. F. Gajewski.
1998
. B7-1 engagement of cytotoxic T lymphocyte antigen 4 inhibits T cell activation in the absence of CD28.
J. Exp. Med.
188
:
205
-210.
38
Lin, H., J. C. Rathmell, G. S. Gray, C. B. Thompson, J. M. Leiden, M. L. Alegre.
1998
. Cytotoxic T lymphocyte antigen 4 (CTLA4) blockade accelerates the acute rejection of cardiac allografts in CD28-deficient mice: CTLA4 can function independently of CD28.
J. Exp. Med.
188
:
199
-204.
39
Engelhardt, J. J., T. J. Sullivan, J. P. Allison.
2006
. CTLA-4 overexpression inhibits T cell responses through a CD28-B7-dependent mechanism.
J. Immunol.
177
:
1052
-1061.