Viewpoint: Therapeutic Implications of CTLA-4 Compartmentalization


  • Miren L. Baroja,

    1. The Biotherapeutics and Transplantation and Immunobiology Groups, The John P. Robarts Research Institute, and the Departments of Microbiology and Immunology, and Medicine, The University of Western Ontario, London, Ontario, Canada N6A 5K8
    Search for more papers by this author
  • Joaquín Madrenas

    Corresponding author
    1. The Biotherapeutics and Transplantation and Immunobiology Groups, The John P. Robarts Research Institute, and the Departments of Microbiology and Immunology, and Medicine, The University of Western Ontario, London, Ontario, Canada N6A 5K8
    Search for more papers by this author

* Corresponding author: J. Madrenas, madrenas


Understanding the regulatory events involved in the activation and inactivation of T cells is crucial to develop therapeutic approaches for autoimmune diseases and for organ transplantation. Co-stimulatory signals delivered through the CD28 receptor and inhibitory signals through CTLA-4 are required for the proper modulation of T cell responses and the induction and maintenance of peripheral tolerance. Manipulation of these signals is emerging as a potential strategy to prevent allograft rejection in different animal models. Recent data on the compartmentalization and the structural features of CTLA-4 within T cells provides critical information not only on the molecular basis of T cell inactivation by CTLA-4, but also on the key requirements for the successful development of therapeutic strategies targeting this molecule.


Activation of T cells requires not only signals from the T cell antigen receptor (TCR) complex but also signals provided by co-stimulatory molecules, the best characterized being CD28. As activation proceeds into proliferation and differentiation of antigen-specific T cells, regulatory mechanisms gain pre-eminence. One of these regulatory mechanisms is the induction of expression of the cytotoxic T lymphocyte-associated antigen 4 (CTLA-4), a membrane glycoprotein that inhibits T-cell responses when co-ligated with the TCR. A remarkable feature of this arrangement is that the opposite effects of co-engagement of the TCR with CD28 or with CTLA-4 (T-cell activation vs. inactivation) result from binding of these two molecules to the same ligand on the surface of antigen-presenting cells (APC) – the B7 family of molecules CD80 (B7-1) and CD86 (B7-2) – through a relatively conserved molecular basis (1).

Sharing of ligands between CD28 and CTLA-4 poses the challenge of explaining how the engagement of the TCR with CD28 or with CTLA-4 is regulated to ensure the appropriate biological response. Two factors are clearly implicated. One is the temporal pattern of surface expression for these molecules: CD28 is expressed on resting T cells whereas CTLA-4 is expressed after T-cell activation (2). Consistent with this profile of expression, T-cell activation is the primary event that leads to the induction of CTLA-4 expression and subsequent T-cell inactivation. A second factor that regulates the balance between CD28 vs. CTLA-4 engagement is the kinetic features of the binding of B7 molecules with CD28 and CTLA-4 (3). CTLA-4 binds B7 with higher affinity and avidity than does CD28. Also, recent data have elegantly demonstrated that, at least in solution, there is a bias for B7.2 – the B7 form expressed early in immune responses – to bind CD28 and for B7.1 – the B7 form expressed at a later phase – to interact with CTLA-4 (4). In this review, we propose that in addition to the temporal and kinetic constraints, there is an additional factor that controls TCR-CD28 vs. TCR-CTLA-4 co-engagement and thus regulate the outcome of T-cell stimulation. This third factor is the spatial arrangement and compartmentalization of CD28 and CTLA-4 on the surface of a T cell, particularly in relation to the formation of an immun- ological synapse (IS) with the APC (5), and in the context of signaling-associated, lipid raft microdomains (6). Here, we will discuss recent data about the compart mentalization of CTLA-4, and link it to the mechanistic evidence of negative signaling by this molecule. Using this, we will then outline some aspects that should be fulfilled to generate feasible therapeutic strategies for the modulation of T-cell activation through CTLA-4. Such strategies are of great value in Transplantation Medicine given the central role that T cells play in allo-responses and the ability of CTLA-4 to down-regulate them. We refer the reader to recent elegant reviews that specifically deal with the structure and function of B7 family members, CD28 and CTLA-4, the CTLA-4/B7 interactions, and the targeting CTLA-4 in tumor immunotherapy (1–3, 7–9).

Two Mechanisms of Action of CTLA-4 with Two Different Structural Requirements

The inhibition of T-cell responses by CTLA-4 can be mediated by, at least, two mechanisms (10). One mechanism is the delivery of a negative signal that directly or indirectly inhibits early signaling from the TCR, resulting in the down-regulation of expression of those genes normally associated with CD28 co-stimulation (11). The precise molecular basis of this negative signal remains elusive. However, the search for a signaling pathway emanating from CTLA-4 has been revitalized by recent discoveries on the structural arrangement of CTLA-4 upon engagement with its B7 ligand. Analysis of the crystal structures of CTLA-4 engaging B7-1 and B7-2 has shown that each dimer of CTLA-4 can engage two B7 molecules each on independent dimers (12–14). This may conduce to the formation of a lattice structure at the T cell–APC interface with an intermembrane spacing of 100–140 Å. This distance is of particular significance as it would allow engaged CTLA-4 to fit within the core of the IS. The arrangement of engaged CTLA-4/B7 into oligomers within the IS could be relevant to explain how signaling through CTLA-4 can be initiated as this molecule is expressed on the cell surface as homodimer and because there is no evidence that signaling through this molecule could result from a conformational change after its binding to B7.

From a molecular point of view, negative signaling through CTLA-4 requires the cytoplasmic tail of CTLA-4, but it is not contingent on the presence of the potential motifs described in this tail, nor on phosphorylation of the tyrosine residues in this portion of the molecule (10,15–18). The molecular partners that mediate negative signaling by CTLA-4 remain unclear. CTLA-4-mediated negative signaling can occur at low levels of surface expression of this molecule. This would be expected given that only a minority of the CTLA-4 molecules expressed by an activated T cell is on the cell surface (less than 5%) due to efficient internalization of nonphosphorylated CTLA-4 by a clathrin/AP2-dependent mechanism (19–21).

Of interest, CTLA-4-mediated signaling can induce a state of unresponsiveness in primary CD4+ T cells (22–24). This state resembles T cell anergy. However, contrary to conventional anergy, this unresponsiveness state resulting from cell cycle arrest is refractory to exogenous IL-2, and can still be induced in the presence of CD28 co-stimulation. A common feature of the anergy induced by cell cycle arrest and the in vivo tolerance induced by CTLA-4 is the elevated levels of the cell cycle inhibitor p27kip1 (23,24).

The other mechanism of T-cell inactivation by CTLA-4 is antagonism of B7-dependent, CD28 co-stimulation. This mechanism results from active sequestration of B7 molecules by CTLA-4 due to the higher affinity and avidity of CTLA-4 for the B7 ligands (4,25). As expected, B7 sequestration by CTLA-4 translates into antagonism of CD28 co-stimulation, and is apparent under conditions in which B7-delivered co-stimulation is required for activation of T cells (26,27). The ability of CTLA-4 to sequester B7 is directly proportional to the levels of surface expression of CTLA-4, i.e. the higher the surface expression of CTLA-4, the more potent is the CTLA-4-mediated inhibition of T-cell activation. B7 sequestration by CTLA-4 does not require the cytoplasmic tail of CTLA-4 as B7 sequestration is strictly dependent on B7 binding (10). The contribution of this mechanism in an in vivo context is unknown, but it is clearly important in in vitro models, especially when using mutant CTLA-4 molecules (e.g. tailless forms) expressed on the cell surface at very high levels due to the lack of AP-2-mediated internalization (10). As expected under conditions of limited CD28-mediated co-stimulation, sequestration of B7 molecules by CTLA-4 may lead to the induction of conventional (i.e. IL-2 recoverable) T cell anergy (23).

CTLA-4 Compartmentalization and Signaling in T cells

A strategy to characterize the signaling pathway triggered by CTLA-4 and narrow the search for its components has been to examine the subcellular compartments where CTLA-4 is located, in particular during its co-ligation with the TCR. Work over the past few years has demonstrated the formation of a highly ordered interface – the IS – between the T lymphocyte and the APC during antigen-driven stimulation leading to T-cell activation. The IS involves not only the TCR/CD3 complex but also co-stimulatory molecules, adhesion molecules, cytoskeleton-associated molecules, and signaling molecules. In its mature stage, the IS has a central core (known as central supramolecular activation complex or c-SMAC) containing TCR, CD28, the tyrosine kinase lck and protein kinase C (PKC)-θ, and a peripheral ring (known as p-SMAC) that is enriched for adhesion molecules such as LFA-1, and cytoskeletal proteins such as talin (28–30).

Whether the structural organization within an IS has intrinsic functional implications is still debatable. Based on the kinetics of IS formation, it is apparent that initial signaling from the TCR can occur without formation of a mature IS (31). In fact, this early signaling from the TCR, including the activation of the ZAP-70 kinase, is critical for subsequent T-cell polarization and the recruitment of additional signaling molecules (32). It has been proposed that the IS may play a role in the maintenance and/or integration of TCR-mediated signaling by facilitating the concentration of relevant molecules for TCR-mediated signaling (33). The formation of a mature IS may also contribute to the overall polarization of the effector T cell function including polarized cytokine secretion or polarized release of the contents of cytotoxic granules (34–36). Finally, the IS may regulate TCR-mediated signaling by directing TCR internalization through lipid rafts.

The formation of a mature IS between the APC and the responding T cell also correlates with compartmentalization of TCR-dependent signaling molecules into discrete cell membrane microdomains known as lipid rafts, and subsequent aggregation of these microdomains. Lipid rafts are characterized by enrichment in glycosphingolipids and by detergent insolubility [reviewed in (37)]. It is important to note that lipid rafts are enriched with the src kinases lck and fyn, the linker for activation of T cells (LAT), PKC-θ and cytoskeletal regulatory proteins (38). Such arrangement is consistent with raft clustering at the core of the IS. The biological contribution of lipid rafts to signal transduction is purely based on correlative data indicating that raft integrity is required for efficient T-cell activation (39), and one should keep in mind the possibility that lipid rafts may play a role in down-regulation of T-cell signaling. This may take the form of providing sites for storage and/or internalization of inactive signaling molecules and receptors (40,41).

Determining the spatial and temporal compartmentalization of CTLA-4 in the context of these two emerging paradigms of cell activation, i.e. formation of an IS and subcellular compartmentalization of some molecules within lipid rafts, may help us to understand how and when this receptor down-regulates T-cell responses. As indicated above, most of the CTLA-4 molecules produced upon T-cell activation remain inside the cell in a yet-to-be defined compartment/s; only a small fraction of CTLA-4 is exported to the T cell surface. CTLA-4 exported to the cell surface is polarized to the site of TCR engagement as initially shown by Linsley et al. using antibody-mediated ligation (42) and confirmed by Egen and Allison using an antigen-dependent system (43). Surface CTLA-4 co-localizes with CD3 and PKC-θ at the IS under conditions in which it inhibits T-cell activation after stimulation with antibodies (42), nominal antigen (43), and superantigen (44). The translocation of CTLA-4 to the IS is regulated by the strength of TCR signaling and is more efficient under conditions of T-cell stimulation with antigenic peptides with strong agonist properties than under conditions of stimulation with weak agonist peptides (43). It is unclear what determines the correlation between the strength of TCR signaling and CTLA-4 polarization to the IS, but it may be due to CTLA-4 retention on the cell membrane close to engaged TCR resulting from phosphorylation of the tyrosine residues in the cytoplasmic region of CTLA-4 induced by TCR signals (16). This modification retains CTLA-4 on the cell surface by preventing its AP-2 depending internalization.

Most CTLA-4 molecules in T lymphocytes are detected within the detergent soluble fraction of the cell lysate (44). However, a significant amount of the CTLA-4 on the cell surface partitions within lipid rafts (44,45). Interestingly, the amount of CTLA-4 within lipid rafts and on the cell surface increases under conditions of T-cell stimulation, concomitantly with migration of CTLA-4 to the IS and with clustering of lipid rafts at the core of the synapse (44,45). This observation is consistent with the proposal that TCR signaling induces retention of CTLA-4 on the membrane at the core of the IS.

Analysis of the structural requirements for CTLA-4 compartmentalization into lipid rafts and relocation to the IS indicates that both processes are dependent on the cytoplasmic domain of CTLA-4 (44). Recombinant tailless CTLA-4 molecules, which are expressed on the T cell surface at very high levels, fail to partition within lipid rafts and do not aggregate at the IS. This is remarkable as recombinant GPI-anchored CTLA-4 molecules, which lack a cytoplasmic tail, are enriched in lipid rafts; one could have argued that the transmembrane region plays a role in determining lipid raft relocation. However, the presence of a transmembrane region without a cytoplasmic tail is not sufficient for relocation of CTLA-4 to lipid rafts. Therefore, the cytoplasmic tail of CTLA-4 is required not only for trafficking to and from the cell membrane or for negative signaling, but also for compartmentalization into lipid rafts within the T cell surface.

The two mechanisms of action of CTLA-4 – negative signaling and B7 sequestration – fit very well the data about its compartmentalization. As presented, negative signaling through CTLA-4 requires its cytoplasmic tail, which is also essential for partitioning within lipid rafts and at the IS. On the other hand, B7 sequestration does not require the cytoplasmic tail of CTLA-4 and, based on the effects observed with tailless CTLA-4 molecules, does not require CTLA-4 partitioning within lipid rafts or its polarization to the IS. Thus, the formation of CTLA-4-based lattices may have different mechanistic implications inside or outside the IS. Within the IS, it may provide the appropriate architecture for effective CTLA-4 signaling and down-regulation of early TCR-mediated signaling. On the other hand, the CTLA-4-based lattice outside lipid rafts may actively regulate the access of B7 to the IS where it could potentially interact with CD28. It will be of interest to determine if any differences in the B7-1/B7-2-binding properties to CTLA-4 can also affect relocation of CTLA-4 into the IS. This may prove very valuable in assessing temporal changes in the availability of CTLA-4 or its ligand at the IS. As Schwartz et al. suggest, the balance between the availability of B7 or of CTLA-4 at a given time, at the IS or its vicinity, may have important mechanistic implications (1). For example, an excess of CTLA-4 (or a limitation in B7) may favor appropriate ligation of separate CTLA-4 dimers and facilitate negative signaling. Alternatively, an excess of ligand may favor the formation of lattices and primarily work by preventing relocation of surface receptors to the IS, by analogy with what is seen with galectin-1 (46), and antagonize B7-dependent co-stimulation.

Functional Implications of CTLA-4 Compartmentalization

The presence of surface CTLA-4 within lipid rafts and its migration to the IS has at least two implications. One is that compartmentalization of CTLA-4 into lipid rafts may bring the signaling molecules required for the down-regulation of early TCR signaling, either on TCR-ζ or CD3 chains or on downstream steps (e.g. ERK activation). Lee et al. have shown that the cytoplasmic tail of CTLA-4 can associate with the TCR-ζ and the SH2-domain-containing tyrosine phosphatase (SHP-2), and this interaction promotes dephosphorylation of the TCR-ζ chain (47). Indeed, recent data from Chikuma et al. have reported that CTLA-4 interacts with phosphorylated TCR-ζ within lipid rafts, and that the amount of this phosphoTCR-ζ accumulated in the rafts decreases after TCR/CTLA-4 co-ligation. These results imply that CTLA-4-mediated inhibition of TCR signaling can occur at the IS where CTLA-4 can associate with TCR-ζ within the lipid raft fraction. In conditions of TCR/CTLA-4 co-engagement, CTLA-4 recruitment to the raft and the IS may attenuate TCR signaling by driving TCR out of the raft fractions (45). Another TCR downstream intracellular pathway regulated negatively by CTLA-4 is the ras/mitogen-activated protein kinases (MAPK) pathway. Several reports have shown that TCR/CTLA-4 co-engagement inhibits the activation of extracellular signal-regulated kinase (ERK) and c-Jun N-terminal kinase (JNK) (16,48). LAT and ras are involved in the formation of multimolecular complexes that activates the MAPK pathway in response to agonist ligands of the TCR. Thus, inhibition of this pathway by CTLA-4 is consistent with the compartmentalization of both these molecules and CTLA-4 in lipid rafts (49).

The other functional implication of CTLA-4 compartmentalization is that lipid raft relocation of CTLA-4 and co-ligation with the TCR at the immunological synapse may release CTLA-4 from its own negative regulation. Recent evidence has shown that CTLA-4 interacts with the serine/threonine phosphatase PP2A (50,51). Two roles have been attributed to this interaction. One is that PP2A on CTLA-4 may restrain the activation of kinases induced by TCR and CD28 ligation. The other is that PP2A is a negative regulator of CTLA-4 function, as mutant CTLA-4 molecules with a defect to bind PP2A are more potent at inhibiting T-cell activation. The data supporting this second model include the observation that although PP2A is associated with CTLA-4 in resting T cells, co-ligation of TCR with CTLA-4 induces tyrosine phosphorylation of PP2A and subsequent dissociation. The release of PP2A would permit CTLA-4 to exert its inhibitory function on early TCR-mediated signaling (51).

There may be additional complexities in the trafficking of CTLA-4 from intracellular compartments to the cell surface and into lipid rafts as suggested by a report claiming that TCR/CD28 and CTLA-4 co-engagement inhibits the formation of lipid raft microdomains (52). This finding correlated with a reduction in the levels of raft-associated LAT. Several explanations could be considered for this observation. The simplest is that CTLA-4 has a direct role in regulating the generation of signal permissive lipid rafts. Alternatively, CTLA-4 signaling may also have an effect on the function of the cytoskeleton and regulate the dynamics of lipid raft movement. Either one may occur as a direct effect of CTLA-4 on early TCR-mediated, ZAP-70-dependent signaling (32,53), or as an indirect effect on CD28 co-stimulation (54) and its involvement in the regulation of cytoskeletal function (55–57).

A Model to Link CTLA-4 Compartmentalization with CTLA-4-mediated Signaling

From the previous evidence, one can propose a model that puts the mechanism of action of CTLA-4 in the context of trafficking and compartmentalization of this molecule during T-cell activation (Figure 1). Three fundamental aspects will be emphasized in this model: the need for co-ligation of CTLA-4 with the TCR for T-cell inactivation, the proper CTLA-4 compartmentalization, and the regulation of CTLA-4 function by PP2A. In resting T cells, expression of CTLA-4 is only found in the memory pool. In these cells, CTLA-4 is expressed at significant levels although the majority of molecules remain in intracellular vesicles, and only a minority of these molecules is exported to the cell surface likely as dimers (P. Darlington, M. Kirchhof and J. Madrenas, unpublished observations, London, Ontario, Canada, January, 2003). The exported CTLA-4 molecules in these cells are associated with PP2A. This phosphatase regulates the function of CTLA-4 and makes CTLA-4 inactive. On the cell surface, CTLA-4 compartmentalizes within lipid rafts, and in the absence of phosphorylation of their cytoplasmic tyrosine residues, it is rapidly internalized by an AP-2-dependent mechanism in clathrin-coated pits. Internalization of surface CTLA-4 may be facilitated by localization within lipid rafts as these microdomains have been linked to receptor internalization.

Figure 1.

A model to link CTLA-4 compartmentalization with CTLA-4-mediated signaling. The diagram represents the events at the T cell surface at three different time points upon antigen stimulation. t1 represents the activation of a naïve T cell that leads to up-regulation of CTLA-4 expression. Only a minority of CTLA-4 is exported to the cell surface, where it is associated with PP2A and partitions within lipid rafts (shown as the dotted area). Most of the surface CTLA-4 is rapidly internalized into a compartment still uncharacterized. However, in the presence of concomitant TCR signaling (as in t2), CTLA-4 is phosphorylated and retained on the cell surface. Such signaling also causes phosphorylation of PP2A, which translates into its dissociation from CTLA-4. Under these conditions (t3), CTLA-4 is active. The fate of the released PP2A is unknown. In this model, we suggest that it may be transferred to CD28, inhibiting the co-stimulatory function of this receptor and thus down-regulating T-cell responses. Implicit in this model are the three main aspects emphasized in this article: need for TCR-CTLA-4 co-ligation, compartmentalization of CTLA-4, and regulation of CTLA-4 function by PP2A.

Upon TCR ligation, CTLA-4 is tyrosine phosphorylated in an lck-/ZAP-70-dependent manner. This phosphorylation plays two roles. One is to prevent the targeting of intracellular CTLA-4 to lysosomal degradation, by a mechanism that involves the AP-1 adapter/targeting protein complex. The other is to prevent AP2-mediated internalization of surface CTLA-4. These two effects combined lead to increased levels of CTLA-4 retained on the cell surface. In addition, both the intracellular pool of CTLA-4 and the surface pool of CTLA-4 polarize towards the IS. In particular, the raft-associated surface CTLA-4 dimers migrate towards the IS where they interact with B7 homodimers and form extended lattices of engaged CTLA-4/B7 oligomers. Under these conditions of proximity, i.e. of co-ligation of TCR with CTLA-4, TCR-mediated signaling induces tyrosine phosphorylation of the serine/threonine phosphatase PP2A and subsequent dissociation from CTLA-4. The release of PP2A from CTLA-4 permits CTLA-4 to become functionally active and inhibit T-cell activation by acting either on early steps of TCR-mediated signaling such as on activation of ERK or more proximal events of TCR-dependent signaling such as phosphorylation of the TCR-ζ chain.

The molecular consequences of the dissociation of PP2A with CTLA-4 are unknown. PP2A binds also CD28. Based on in vitro data, there may be subtle differences in the cellular effects of such binding compared with that with CTLA-4 because the interaction with the former involves the catalytic subunit of this phosphatase, while the interaction with CTLA-4 involves primarily the regulatory subunit of PP2A. It is tempting to speculate that PP2A may be involved in a regulatory interplay of signals emanating from CD28 and from CTLA-4. This interplay would then determine the balance between CD28-co- stimulatory signals and CTLA-4-inhibitory signals during T-cell activation. This could explain why under some conditions CTLA-4 targets CD28 effects (11) while under different conditions it can act independently of CD28 (58).

What are the Therapeutic Implications of CTLA-4 Compartmentalization?

The vast amount of data supporting the critical role of co-stimulatory signals for full T-cell activation and the ability to induce T cell unresponsiveness by blocking these signals has prompted many investigators to consider the blockade of co-stimulatory molecules as a therapeutic strategy to prevent graft rejection and induce transplantation tolerance. Within the array of co-stimulatory molecules on T cells, two have taken primary stages: CD28 and CD40 ligand (CD154). Blockade of these receptors either using monoclonal antibodies against them or against their ligands and/or using soluble recombinant ligands for these molecules (e.g. CTLA-4-Ig) has been relatively successful in preventing rejection of allografts in small animal models [reviewed in (59)]. However, the efficacy of these protocols has not been as impressive in nonhuman primate studies. The reasons for such disparate results are unclear. It may include the presence of alternative co-stimulatory mechanisms that are active in vivo as suggested by the prolonged graft survival upon blockade of B7 in CD28 knockout mice and in CD28/CTLA-4 double knockout mice (60,61). Another explanation for the relatively low success of these protocols is the presence of large pools of memory T cells with relatively low dependency on co-stimulation for activation in large animals and humans.

An alternative to passive blockade of co-stimulatory signals is to enhance the function of negative regulatory molecules that directly inhibit T-cell activation. And it is within this framework that CTLA-4 becomes a prime therapeutic target given the multilevel inhibitory function of this molecule in vitro and in vivo in the context of transplantation. Net CTLA-4 function can be increased in two ways. One is to increase CTLA-4 expression. Ideally in transplantation, this should be achieved without concomitant T-cell activation. Such an option is not far-fetched given the recent report that monoclonal antibodies against CD45RB that can induce transplantation tolerance also up-regulate CTLA-4 expression (62). Increasing CTLA-4 expression may have beneficial immunomodulatory effects on graft survival not only by limiting B7-dependent co-stimulation causing conventional T cell anergy but also by acting on the APC through ligation of B7, for example by inducing tryptophan catabolism on dendritic cells and decreasing immune responsiveness (63).

Another way to increase CTLA-4 function is through agonist ligands of this receptor. By turning on CTLA-4 function in T cells, these compounds can directly down-regulate activated T cells and memory T cells (64), regulate the immune response towards a Th1 or Th2 microenvironment (18,65–67), and/or can induce of T cell unresponsiveness by conventional and nonconventional anergy or enhanced generation of CD4+/CD25+ regulatory T cells (23,68–70). Despite the appeal, development of CTLA-4 agonists to inhibit immune responses has been unsuccessful, something remarkable considering the success of blockade of surface CTLA-4 function to boost immunity (2).

The difficulties in generating CTLA-4 agonist ligands likely result from the complexity of CTLA-4-mediated inhibition, both in terms of spatio–temporal interactions with the TCR and in terms of mechanisms used by this molecule to inactivate T cells. The integration of information about the subcellular compartmentalization of CTLA-4, the binding requirements for CTLA-4, and the regulation of CTLA-4 function may provide us with an opportunity to identify the major attributes to design molecules with agonist properties on CTLA-4. Specifically, the three fundamental features emphasized above in the proposed model for CTLA-4 action are also critical to generate a CTLA-4 agonist. That is, a potential CTLA-4 agonist has to fulfill the need for co-ligation of CTLA-4 with the TCR in order to achieve negative signaling, the molecule has to work on the appropriate pool of CTLA-4, and the molecule has to overcome the mechanisms that regulate CTLA-4 function.

One of the essential features of T-cell inactivation by CTLA-4 is that it has to be co-ligated with the TCR in cis, i.e. the ligands need to be on the same surface. This is in sharp contrast with co-stimulation through CD28, which can work in trans, i.e. the ligation of the TCR and of the CD28 can result from ligands on different cell surfaces (71). Using chimeric ligands such as bivalent antibodies may fulfill the need for TCR-CTLA-4 co-ligation. However, the use of such compounds has to consider the timing of administration and the compartmentalization of CTLA-4 during that time window. From a biochemical point of view, the requirement for cis co-ligation suggests that CTLA-4 acts on early stages of TCR-dependent signaling, either directly on TCR subunits as previously suggested (47), or on the effects of CD28 on TCR as suggested by the results from DNA microarray analysis or from studies on the regulation of several metabolic effects of CD28 co-stimulation (11,72). From a cellular point of view, the need for co-ligation may reflect a requirement for appropriate compartmentalization of the signaling machinery coordinating TCR and CD28 signals. Both considerations certainly complicate the development of CTLA-4 agonists.

A second aspect to consider when designing CTLA-4 agonists is that they have to target the appropriate pool of CTLA-4. Current evidence points to the surface pool of CTLA-4 as the relevant pool. However, the therapeutic potential of targeting the intracellular pool of CTLA-4 should not be ignored given that this pool contains most of the CTLA-4 molecules and contributes to the regulation of CTLA-4 degradation and secretion (73–75). While the targeting of surface CTLA-4 may not be as challenging as the targeting of the intracellular pool, it may encounter the obstacle of fine-tuning the number of molecules being targeted to warrant net inhibition of T-cell activation. One could predict that low levels of CTLA-4 engagement under conditions of more abundant CD28-ligation by B7 may have a paradoxical enhancing effect on T-cell activation.

Finally, a CTLA-4 agonist has to release this receptor from its own negative regulators. This may be achieved by targeting the site of CTLA-4 where PP2A binds (51). Although this phosphatase is expressed as a trimeric molecule formed by a regulatory subunit (PP2AA), a structural subunit (PP2AB) and a catalytic subunit (PP2AC) (76), interference with the binding of the regulatory PP2AA subunit is sufficient to enhance CTLA-4 function (51). Such binding occurs at the lysine-rich region of the juxtamembrane portion of the cytoplasmic tail of CTLA-4 (lysine residues 152, 154, and 155 in human CTLA-4) (51). Based on this observation, one can predict that small molecules that bind the lysine-rich region of the tail of CTLA-4 will compete with PP2A binding to CTLA-4, and thus act as CTLA-4 agonists.

A concern of any approach targeting CTLA-4 to increase its function and thus induce T-cell unresponsiveness and/or immunodeviation is the lack of antigen specificity, particularly when targeting the binding of CTLA-4 to PP2A given the widespread distribution of this phosphatase (76). However, this concern may not be significant for several reasons. First, one should target the CTLA-4 site, not the PP2A site, so that potential drugs only interfere the PP2A function on CTLA-4 but not other PP2A functions. Second, one should keep in mind that CTLA-4 expression will mostly occur in activated T cells. Thus, during the early stages of an allo-response, the therapeutic effect will be primarily on alloreactive T cells and only under conditions of co-ligation of TCR with CTLA-4. In this way, concomitant broad immunosuppression is unlikely. In any case, testing the validity of therapeutic targeting of CTLA-4 will be a relatively small challenge compared with our current challenge of developing CTLA-4 agonists.