Costimulatory signals controlling regulatory T cells

The concept of specialized suppressor T cells (now commonly called regulatory T cells or Tregs), which was abandoned by most immunologists in the early 1980s (1), was rescued by the description of a marker for their identification and isolation by Sakaguchi et al. (2), who, in 1995, described a CD4⁺CD25⁺ subpopulation of T cells that could prevent pathology in a mouse autoimmune disease model. However, the marker CD25, the α-chain of the high-affinity IL-2 receptor, is also induced on conventional T cells after activation. The dilemma arose as to how one could distinguish genuine Tregs from recently activated conventional T cells. Furthermore, the suppressor activity of CD4⁺CD25⁺ Tregs in vitro depended on their prior stimulation and could be overcome by strong costimulation (3, 4). What were the signals that controlled the activity of CD4⁺CD25⁺ Tregs in vivo that allowed them to inhibit certain immune responses but not others?

Glucocorticoid-Induced Tumor Necrosis Factor Receptor (GITR) and GITR Ligand (GITRL)

In a recent issue of PNAS, Tone et al. (5) reported the cloning of the murine ligand for GITR, a tumor necrosis factor receptor superfamily member that had previously been demonstrated to regulate CD4⁺CD25⁺ Treg function. Searching for novel Treg markers, two groups had independently found that freshly isolated CD4⁺CD25⁺ Treg cells but not conventional CD4⁺CD25^– T cells expressed uniformly high levels of GITR (6, 7). Activation of the CD4⁺CD25^– T cells in vitro rapidly increased GITR surface expression to levels comparable to that on CD4⁺CD25⁺ Tregs. GITR thus was no better than CD25 in distinguishing Tregs from activated conventional T cells. However, the addition of anti-GITR antibodies to the coculture of CD4⁺CD25^– and CD4⁺CD25⁺ T cells in vitro produced a dramatic functional effect, completely abrogating suppression (6, 7). This effect appeared to be caused by active signaling into the CD4⁺CD25⁺ Tregs rather than blockade of interaction with the unknown putative GITR ligand. In vivo, the administration of anti-GITR antibody to young mice caused autoimmune gastritis, a disease also seen with depletion of CD4⁺CD25⁺ Tregs (6). Although these studies established GITR as a receptor on the surface of CD4⁺CD25⁺ Tregs that could modulate Treg function, the physiological role of GITR signaling remained elusive. Tone et al. (5) identified the murine ligand for GITR (GITRL) and provided some initial insight into the regulation of its expression. They showed that GITRL was expressed by antigen-presenting cells (APCs), i.e., dendritic cells, macrophages, and B cells, but not by T cells. Stimulation of dendritic cells with lipopolysaccharide in vitro led to a transient up-regulation of GITRL expression. Furthermore, GITR ligation by its ligand not only affected Tregs, but GITR signaling additionally had a costimulatory effect on purified conventional CD4⁺CD25^– T cells.

The article by Tone et al. is an important step toward understanding the regulation of Treg function in vivo, but more work is clearly needed. It will be interesting to look in more detail at GITRL expression patterns on different types of APCs including subtypes of dendritic cells in vivo, the relationship between the expression of GITRL and other costimulatory ligands, and the kinetics of GITRL expression during immune responses in vivo. It is tempting to speculate on the phenotype of GITR or GITRL transgenic or knockout mice. Will overexpression of GITRL lead to a breakdown of peripheral tolerance by “switching off” CD4⁺CD25⁺ Treg cells? Could dysregulated expression of GITRL thus be a contributing factor in the development of autoimmune diseases? Another important aspect of the article by Tone et al. is the formal demonstration that GITR ligation has a costimulatory effect on conventional CD4⁺ T cells. This finding corrects the impression evoked by the two original papers describing GITR expression on CD4⁺CD25⁺ Tregs, that GITR signaling was unique to CD4⁺CD25⁺ Tregs. GITR ligation rather functions as a second signal for both CD4⁺CD25⁺ Tregs and conventional CD4⁺CD25^– T cells. Importantly, freshly isolated CD4⁺CD25^– T cells seem to be GITR^low rather than GITR^– much in contrast to CD4⁺CD8⁺ double positive thymocytes that appear to be truly GITR^– (6).

The Two-Signal Model of T Cell Activation Revisited

A major paradigm of current immunology is that the initiation of an immune response requires an antigen-specific first signal and additional second signal(s) induced by “danger” or “foreignness.” A popular version of the two-signal model for T cell activation, described a few years ago, focused on B7-1 and B7-2 molecules expressed on APCs as the main providers of T cell costimulation (8). In this model, T cells are activated through the interaction of peptide/MHC complexes and B7-1/B7-2 molecules on the APC with, respectively, the T cell antigen receptor (first signal) and constitutively expressed CD28 (second signal). Full T cell activation induces expression of the negative regulatory molecule CTLA-4, an alternative ligand for B7-1 and B7-2, which then turns off the immune response. Over the past several years, it has become clear that the initiation of an immune response is a much more complex process. Additional B7 family members expressed on APCs and their respective ligands on T cells have been identified, including B7h/ICOS and the negative regulatory PDL1+2/PD-1 (9). Furthermore, other classes of receptors have been shown to be involved in the regulation of immune responses, including tumor necrosis factor receptor family members like OX40, 4-1BB, and their ligands (10). Some of these costimulatory ligand/receptor pairs are constitutively expressed, whereas others are induced upon activation. There appears to be a hierarchy of costimulatory ligand/receptor combinations in the control of immune responses, e.g., CTLA-4 knockout mice die within 3–4 weeks from a severe lymphoproliferative syndrome (11), whereas PD-1 knockout mice develop a far more subtle lupus-like disease later in life (12). In fact, costimulatory receptors appear to differentially affect certain aspects of the immune response such as proliferation, survival, and effector cell differentiation (10). Another layer of complexity is added to the two-signal model by the demonstration of the CD4⁺CD25⁺ Treg cells. The general implication is that the initiation of a T cell response involves the interaction of three different cell types (APCs, conventional T cells, and regulatory T cells, CD4⁺CD25⁺ Tregs being one example). Costimulatory signals provided by APCs influence the “behavior” of both the conventional T cells and the regulatory T cells. The interaction of the three (or more?) cell types will then “decide” about the initiation and phenotype of the response. An example of this complex interplay is the recent demonstration that conventional T cells need a certain cytokine milieu, with an absolute requirement for IL-6 produced by dendritic cells, to escape from the suppressive influence of the CD4⁺CD25⁺ Treg cells (13).

Costimulatory Molecules on CD4⁺CD25⁺ Treg Cells

Currently, no surface molecules are known that are truly specific for CD4⁺CD25⁺ Tregs, GITR being no exception. This means that Treg activity is regulated by the same ligands that regulate conventional T cells. What is known about costimulatory molecules and their functional effect on CD4⁺CD25⁺ Tregs? As discussed above, GITR ligation on CD4⁺CD25⁺ Tregs appears to abrogate their “suppressiveness.” How this works is not clear, in keeping with the fact that the suppressor mechanism of CD4⁺CD25⁺ T cells still awaits elucidation. The classical costimulatory molecule CD28 is expressed on the surface of CD4⁺CD25⁺ and CD4⁺CD25^– T cells at equivalent levels. As for anti-GITR, addition of anti-CD28 antibodies abrogates suppression in coculture systems in vitro (4); however, this appears to be caused by a costimulatory effect on the CD4⁺CD25^– T cells rather than inhibition of the Tregs (14). Interestingly, CD28 knockout mice and B7-1/B7-2 double knockout mice have a much reduced frequency of CD4⁺CD25⁺ Treg cells. On the autoimmune genetic background of the NOD mouse, this regulation correlates with earlier disease onset (15). Thus, an additional function of B7/CD28 interaction seems to be the regulation of the CD4⁺CD25⁺ Treg pool size (16). This finding is probably an indirect effect through the control of IL-2 production by conventional T cells, given that CD4⁺CD25⁺ T cells are incapable of producing IL-2 themselves but require IL-2 signals for their generation and/or homeostasis (17). CTLA-4, much like GITR, is highly expressed in freshly isolated CD4⁺CD25⁺ T cells although it predominantly resides intracellularly (18). It has been proposed that CTLA-4 signaling switches off CD4⁺CD25⁺ Tregs; however, this view is not universally accepted. CTLA-4 is up-regulated in CD4⁺CD25^– T cells upon activation. More recent concepts of CTLA-4 function suggest that it inhibits the expansion of high-affinity T cell clones within a polyclonal responder population (19). The role of CTLA-4 for CD4⁺CD25⁺ Treg cells could thus be to maintain a broad range of T cell antigen receptor specificities and, assuming a largely self-restricted repertoire, prevent the development of repertoire holes that could result in autoimmune responses. Finally, a number of other tumor necrosis factor receptor family members appear to be differentially expressed on CD4⁺CD25⁺ Tregs (20). However, no evidence has been reported yet that any of these can be used as additional Treg markers, or that they have functions similar to GITR. Signaling via GITR is unique in that it both turns off CD4⁺CD25⁺ Tregs and directly costimulates conventional T cells. The reported cloning of the murine ligand for GITR and initial characterization of GITRL expression is another step forward in understanding the regulation of Tregs and the complexities involved in the induction of immune responses.

GITR both turns off regulatory T cells and costimulates conventional T cells.