Regulatory CD4 T (T reg) cells control immune responses to self-antigens and pathogens. However, where T reg cells act to curtail effector T cells in vivo and what stage of effector T cell activation or differentiation is targeted by T reg cells remain unknown. Furthermore, a requirement for direct effector T cell–T reg cell contact in vivo has not been ascertained. Varying answers to these important questions have been provided by several new studies.
During the decade since their discovery, T reg cells have become major culprits for immunosuppression associated with chronic infection and cancer. At the same time, T reg cell deficiency has been invoked as a cause of pathogenic immune hyperreactivity and inflammation associated with various forms of autoimmunity and allergies. Numerous studies suggested that, in addition to responses to autoantigens, commensal flora, and environmental and tumor antigens, exuberant adaptive immunity to viruses, bacteria, parasites, and at least some aspects of innate immunity are also kept in check by CD4+ T reg cells, which can be identified by expression of the forkhead transcription factor Foxp3 (1). These results suggest that many different cell types might be targeted by T reg cells. Indeed, in vitro and in vivo studies have provided evidence that T reg cells can suppress activation, proliferation, differentiation, and even effector function of multiple immune cell types including CD4+ and CD8+ T cells, B cells, natural killer cells, and dendritic cells (DCs) (1–3). It is unclear, however, whether the same or different effector mechanisms are used by T reg cells to control such a variety of cell types. Also unknown is whether the suppressive action of T reg cells is executed in secondary lymphoid organs or at the site of inflammation or both. The latter question is of immediate relevance to another question—the stage of activation and differentiation of the cells that are subjected to T reg cell–mediated control.
Several potential mechanisms of T reg–mediated suppression have been suggested. The results of in vivo analysis favor the production of immunosuppressive cytokines such as TGF-β and interleukin (IL)-10 or cell–cell contact–dependent suppression mediated by the inhibitory receptor CTLA-4. In contrast, according to the majority of studies (with some notable exceptions), in vitro suppression was shown to be independent of IL-10, TGF-β, or CTLA-4, but required cell–cell contact (2, 4–8). An additional suppression mechanism entertained by several investigators is granzyme or perforin-dependent killing of “suppressed” cells by T reg cells (9, 10). A limitation of the available in vivo studies is that they are subject to alternative interpretations, whereas the in vivo relevance of results obtained in in vitro systems is left open to question. Several new in vivo studies have provided novel important information pertinent to the mechanisms of T reg cell–mediated suppression operating in vivo.
Modification of migratory behavior of antigen-specific T cells in the lymph nodes
To address some of the aforementioned outstanding questions, a study by Tadokoro et al. published in this issue (p. 505) (11) and another recent joint study by Tang et al. published in Nature Immunology (12) used the increasingly popular technique of two-photon laser scanning microscopy to examine potential interactions between T cell receptor (TCR) transgenic regulatory and nonregulatory T cells and antigen-bearing DCs in lymph nodes.
In the Tadokoro et al. study, fluorescently labeled naive T cells specific for myelin basic protein (MBP) peptide Ac1-11 were introduced into TCR transgenic recipient mice containing or lacking T reg cells of the same specificity. The rate of migration of naive T cells was not affected by T reg cells in the absence of the antigenic peptide. However, upon provision of MBP Ac1-11 peptide, a notable arrest of activated MBP-specific T cells interacting with antigen-loaded DCs was observed in the absence of antigen-specific T reg cells, whereas in the presence of T reg cells arrest of T cells was significantly diminished. Polyclonal CD4+CD25+ T reg cells were also capable of diminishing antigen-mediated arrest of activated T cells, albeit to a smaller degree as compared with antigen-specific TCR transgenic T reg cells (11).
A similar suppressive effect of T reg cells on the formation of stable clusters of diabetogenic T cells with DCs in isolated pancreatic lymph nodes was reported by Tang et al. (12). These investigators transferred diabetogenic TCR transgenic BDC2.5 T cells into nonobese diabetic (NOD) mice containing endogenously generated T reg cells or into CD28−/− NOD mice, which were previously shown to have drastically diminished numbers of T reg cells (13). As in the Tadokoro study, cotransfers of naive BDC2.5 T cells with in vitro–expanded T reg cells expressing the BDC2.5 TCR or polyclonal CD25+CD4+ T cells into CD28−/− NOD mice also relieved T cell activation-associated arrest (12).
Both MBP- and pancreatic antigen-specific T cells showed diminished cytokine production in the presence of T reg cells in the lymph nodes. Notably, both studies failed to reveal stable T reg–T effector cell couples or triplets formed by T reg cells, activated T effector cells, and antigen-presenting DCs; only T reg–DC and T effector cell–DC interactions were registered.
One potential difficulty in interpreting the set of experiments by Tadokoro et al. is that essentially all T cells in the recipient mice were specific for a MBP peptide, which may lead to systemic production of cytokines and chemokines in response to MBP. And interpretation of the Tang et al. study is complicated by the large number of in vitro–expanded antigen-specific T reg cells that were provided (11). Nevertheless, the very similar observations reported in these two studies support the view that T reg cells exert their suppressive action on DCs in secondary lymphoid organs, specifically in draining lymph nodes where, via a yet unidentified mechanism, they diminish the ability of DCs to form stable contacts with autoreactive T cells and induce their activation.
Although the data from the two imaging studies suggest an attractive model for T reg cell–mediated suppression, it seems likely that additional T reg cell–mediated suppressive mechanisms are operational in vivo, as suggested by a recent report from Chen et al. (14). These investigators transferred influenza virus hemagglutinin peptide (HA)-specific TCR transgenic T reg cells and naive HA-specific CD8 T cells into mice bearing HA-expressing tumors. Although HA-specific T reg cells in these experiments did not interfere with the expansion of HA-specific CD8 T cells and their differentiation into effector cells expressing high levels of the effector molecules FasL, IFN-γ, granzyme B, and perforin, killing of antigen-expressing target cells by what seemed to be competent effector CD8 T cells was blocked in the presence of T reg cells (14) (von Adrian and von Boehmer, personal communication). Notably, impaired TGF-β receptor signaling in HA-specific CD8 T cells conferred resistance to T reg–mediated suppression of cytolytic activity in vivo (14), in agreement with two recent reports (15, 16). These results emphasize an interplay between suppressive action of T reg cells and TGF-β1 on cells of the immune system, which remains to be elucidated in mechanistic terms.
T reg cell–mediated control of effector T cells in the tissues
In addition to their ability to block the initiation of immune responses in the secondary lymphoid tissues, several recently published papers indicate that Foxp3+ T reg cells also migrate to and function within nonlymphoid sites. There, they can effectively dampen immune responses directed toward both self- and foreign antigens. In contrast to the two-photon microscopy study reported by Tang et al. (12), using the same BDC2.5 TCR transgenic model of type 1 diabetes, Chen et al. recently showed that islet antigen-specific T reg cells had no effect on the priming of diabetogenic T cells in the pancreatic lymph node (17). Instead, BDC2.5 T reg cells prevented diabetes by inhibiting the function of effector T cells only in the pancreatic islets. Although it remains unclear as to why such differing results were obtained in two studies using the same TCR transgenic model of autoimmune diabetes, utilization of CD28−/− rather than Foxp3 mutant NOD mice as T reg cell–deprived recipients and differences in the numbers and activation status of the transferred T reg cells might account for the differing outcome of these studies.
Similarly to the study by Chen et al. (17), Belkaid et al. demonstrated several years ago that T reg cells in the skin dampen the immune response to the parasite Leishmania major during cutaneous infection (18). Indeed, T reg cell accumulation in the skin is impaired in mice deficient for the integrin αE (CD103), and this renders otherwise susceptible BALB/c mice resistant to parasite infection (19). Susceptibility is restored by addition of wild-type T reg cells, demonstrating that local paucity of T reg cells results in enhanced parasite clearance. In this issue (p. 777), this group takes this work further in a series of experiments indicating that during L. major infection T reg cells in the dermis and draining lymph node are not self-reactive, as might have been expected based on studies of T reg cell specificity, but instead are largely parasite specific (20).
Together, these results raise several intriguing questions about the specificity of T reg cells and the functional mechanisms used by these cells to limit autoimmune and inflammatory diseases. First, what are the functional mechanisms used by T reg cells in lymphoid versus nonlymphoid tissues? From the imaging data, it appears that T reg cells do not directly interact with their targets within the lymphoid tissues. Instead, by interacting with antigen-presenting cells (APCs) they appear to restrict the ability of APCs to form stable interactions with, and thus prime, naive T cells. However, the mechanisms used to dampen inflammatory responses mediated by effector T cells in nonlymphoid tissues are likely to be quite distinct. Here, there may be a role for antiinflammatory cytokines such as TGF-β and IL-10 in blocking the recruitment and activation of proinflammatory lymphoid and myeloid cells. Consistent with this idea, an increased level of mRNAs encoding these cytokines was found in pancreas-infiltrating T reg cells (17). Therefore, T reg cell localization in lymphoid versus nonlymphoid compartments may help explain the conflicting data regarding the effector mechanisms that T reg cells use in vivo.
In addition, the signals that direct T reg cell migration to nonlymphoid tissues remain to be identified. Effector T cells acquire the ability to migrate to nonlymphoid sites as they reprogram their homing receptor expression during their initial activation in the secondary lymphoid tissues. Importantly, this reprogramming requires not only antigen recognition, but also proinflammatory signals associated, for example, with microbial infection. T reg cells may undergo a similar phenotypic and functional differentiation. This leads to a model in which, under noninflammatory conditions, T reg cells in the secondary lymphoid organs can prevent initiation of T cell responses by keeping APCs “turned off,” thus inhibiting successful T cell priming (Fig. 1 A). Strong proinflammatory stimuli can overcome this regulation. Under these conditions, activated T reg cells migrate to nonlymphoid tissues, where they dampen effector responses, preventing collateral tissue damage and inflammation that can result in lasting tissue damage caused by excessive tissue remodeling or autoimmunity.
Finally, the question remains as to how T reg cells specific for foreign antigens develop. Several studies based on TCR transgenic mice that also express self-antigen led to the notion that Foxp3 expression is induced in a subset of self-reactive T cells that escapes negative selection during thymic development (21, 22). This is further supported by the analysis of the TCR repertoire of Foxp3+ T reg cells, which showed that these cells are inherently self-reactive (23, 24). One possibility is that the L. major–specific T reg cells in the study by Suffia et al. differentiated from antigen-specific nonregulatory T cells by acquiring Foxp3 expression upon encounter with the cognate antigen in the periphery under tolerogenic conditions. However, adoptive transfer experiments using congenically marked precursors of effector T cells and T reg cells demonstrated that the effector cells failed to appreciably up-regulate Foxp3 expression during L. major infection and that the parasite-specific T reg cells were derived from “naturally occurring” T reg cells found in the periphery of uninfected mice (20). This raises the possibility that thymic development and peripheral maintenance of T reg cells specific for foreign antigen such as L. major may be driven by recognition of cross-reactive self-antigens. Molecular identification of self- and pathogen-derived peptides recognized by these T reg cells, and a detailed analysis of the functional consequences of engagement of their TCRs by these ligands should shed light not only on development of these cells but also on the functional mechanisms they use to limit exuberant immune responses in the context of infection and autoimmunity.