Relationship between CD4 Regulatory T Cells and Anergy In Vivo

The self-tolerance mechanisms of T cells can be broadly characterized as central or peripheral (Fig. 1). Central tolerance mechanisms destroy high-affinity self-reactive T cells during thymic development, or else induce their differentiation into a regulatory T (T_reg) cell lineage (1). Nonetheless, central tolerance appears to be insufficient to clear all self-reactive T cells, and other peripheral tolerance mechanisms are necessary (2, 3). Peripheral tolerance may rely on: 1) ignorance, wherein autoreactive T cells never encounter their cognate Ag; 2) deletion, whereby self-specific peripheral T cells are destroyed after TCR engagement; 3) anergy, which is a state of functional unresponsiveness induced upon self-Ag recognition; and/or 4) Foxp3⁺ T_reg cell–mediated suppression of dangerous T cell responses against self-Ag. Each of these potential tolerance mechanisms has been clearly defined in numerous in vivo experimental systems using TCR-transgenic responder T cells, typically at abnormally high cell frequencies. However, much less is known about self-tolerance in the natural polyclonal CD4⁺ T cell repertoire.

Two recent studies took advantage of experimental Ags (peptides derived from either the P1 bacteriophage Cre recombinase [Cre] or the Aequorea victoria enhanced GFP [eGFP]) whose expression in mice was directed by transgenic tissue-specific promoters, and additionally made use of peptide MHC class II (pMHCII) tetramers to detect polyclonal CD4⁺ T cells that respond to these model self-Ags (4, 5). Both experiments demonstrated that ubiquitous (both thymic and peripheral) expression of self-Ag leads to deletion of 60–95% of the highest affinity self-specific CD4⁺ T cells, with the remaining low-affinity cells left functionally unresponsive to Ag (Fig. 1). In contrast, restriction of self-Ag expression exclusively to nonthymic tissues leads to a tolerance that appears to rely solely on ignorance by naive autoreactive T cells. The strongest evidence for an active peripheral self-tolerance mechanism in these studies came in the form of self-Ags that were only weakly expressed in the thymus and, in particular, were otherwise restricted to mucosal tissues (e.g., Cre protein expression in the intestinal brush border driven by a Vil1-Cre transgene, or expression in airway Clara cells driven by a Scgb1a1-Cre transgene). These Ags induce little central deletion, and instead self-tolerance relies on the generation of self pMHCII-binding Foxp3⁺ T_reg cells (5). Importantly, these T_reg cells appear to suppress the conventional CD4⁺ T cell responses to self-Ag, because systemic ablation of Foxp3⁺ T_reg cells at the time of a self-Ag immunization led to a restoration of the conventional T cell clonal expansion and cytokine production response to near-control levels (5).

In addition to T_reg cells, several other suppressor cells have been described, such as the Tr1 cells, myeloid-derived suppressor cells, as well as the regulatory B cells. However, it is well established that Foxp3⁺ T_reg cells maintain immune homeostasis and prevent adverse immune responses throughout the life span of an individual. Foxp3 is an essential lineage-defining transcription factor of T_reg cells (6, 7), because mutations or deletion of the Foxp3 gene lead to impaired generation of T_reg cells and cause severe autoimmunity in humans and mice (6, 8). Furthermore, instability of Foxp3 expression allows for the transdifferentiation of T_reg cells to T effector cell lineages capable of causing autoimmunity (9, 10). Although Foxp3 is considered to be the master regulator of T_reg suppressive function, the expression of Foxp3 is not sufficient to maintain a stable T_reg cell lineage (11, 12). Numerous T_reg cell–specific genes are in fact expressed independently of Foxp3 protein, including Il2ra (the gene for CD25), Ctla4, Ikzf4 (Eos), and neuropilin 1 (Nrp1) (11, 13, 14). These data suggest that additional lineage-defining factors act with Foxp3 to ensure the generation of a stable, functional T_reg cell compartment.

Most Foxp3⁺ T_reg cells undergo their terminal differentiation in the thymus and are referred to as thymic T_reg (tT_reg) cells, whereas others originate in the periphery (particularly at mucosal barrier surfaces exposed to food Ags and commensal organisms) from conventional Foxp3⁻ CD4⁺ T cells and are consequently called peripheral T_reg (pT_reg) cells (15, 16). It is believed that the self pMHCII specificity and suppressive functions of these two T_reg subsets complement one another in preventing immunopathology (17). tT_reg cells are primarily responsible for maintaining general T cell immune homeostasis, whereas pT_reg cells control immunopathology that is directed against tissue-restricted Ags in mucosal tissues such as the lung and gut (17).

Natural T_reg (nT_reg) cells are defined as all T_reg cells generated in vivo, and thus include tT_reg cells and some pT_reg cells (16). For the case of most natural T_reg cells, epigenetic changes associated with the development of a unique T_reg cell methylome (T_reg-me) are thought to be necessary for stable Foxp3 expression. The T_reg-me consists of demethylated DNA CpG motifs at four T_reg cell–related gene loci, including Tnfrsf18 (the gene for GITR), Ctla4, Ikzf4, and the Foxp3 conserved noncoding DNA sequence (CNS) 2 (18). Therefore, the expression of Foxp3 and the development of the T_reg-me are independent and complementary events. TCR signaling is required for T_reg-me induction, which in turn helps maintain the function and stability of T_reg cells (12, 18). It has been proposed that within the thymus, the strength of TCR signaling controls Foxp3 expression, whereas the duration of signaling establishes the demethylations that are associated with the T_reg-me (12, 18). In contrast, quantitative demethylation of the T_reg-me loci (including Foxp3 CNS2) appears unnecessary for some pT_reg cell differentiation. Rather, Foxp3 transcription in mucosal tissue T_reg cells can be stabilized by Tgfb1- and Smad3-dependent activation of the CNS1 enhancer element (19, 20).

Even though pT_reg cells have been found to play an essential role in the suppression of tissue-specific immunopathology in adoptive transfer experiments (17), the experimental data of Malhotra et al. (4) and Legoux et al. (5) have seemed to point toward a thymic origin for the increased self-Ag–specific T_reg cells observed when tissue-restricted Ag expression in the thymus is only modest. Vil1 promoter–driven Cre-specific T_reg cells express high levels of Helios and Nrp1, both markers compatible with a tT_reg cell origin (5). In addition, genetic disruption of the CNS1 sequence from the Foxp3 locus fails to inhibit the generation of Cre-specific Foxp3⁺ T_reg cells or restore the functional responsiveness of the conventional Cre-reactive CD4⁺ T cell population in Vil1-Cre transgenic mice (5). Interestingly, Foxp3⁺ T_reg cells specific for eGFP are more frequently detected in the thymus of Ins2 promoter–driven eGFP transgenic mice (4), whereas the number of Cre-specific T_reg cells in Vil1-Cre and Scgb1a1-Cre transgenic mice is similar to the wild type (5). Taken together, the experiments suggest that weak expression of self-Ag in the thymus (presumably as a consequence of Aire-mediated transcription in medullary thymic epithelial cells) is insufficient to elicit central deletion of all high-affinity autoreactive CD4⁺ T cells, yet facilitates the maintenance of self-tolerance by inducing the differentiation of highly functional self pMHCII-specific T_reg cells that suppress peripheral responses. In contrast, CNS1-dependent pT_reg cells may be more important to suppress mucosal T cell reactivity to food and commensal Ags introduced only after thymic development has completed.

Notably, neither of these elegant studies suggested a role for anergy in the development of tolerance to peripherally expressed self-Ags. Should one conclude, therefore, that the induction and maintenance of anergy are unimportant to natural peripheral self-tolerance? Or is it possible that anergy plays some role in the establishment of self-tolerance in concert with Ag-specific Foxp3⁺ T_reg cells?

Historically, anergy has been defined as a state of functional inactivation wherein CD4⁺ T cells lose the capacity to produce growth factors and proliferate in response to pMHCII recognition (21–23). But it has been unclear how long this state lasts and/or whether it is an intermediate state in vivo. Anergy can be induced in CD4⁺ T cells when TCR signaling is unaccompanied by strong CD28 costimulatory receptor ligation (24–26). The binding of coinhibitory receptors such as CTLA4 and PD-1 reinforces the induction and/or maintenance of anergy (27, 28). In vivo, anergy is observed in CD4⁺ T cells after systemic, repeated exposure to a soluble Ag or superantigen in the absence of infection or adjuvant (29). Likewise, the adoptive transfer of naive self-Ag–specific CD4⁺ T cells to normal mice bearing relevant self pMHCII complexes leads to the development of functional unresponsiveness (30, 31). This is in contrast with the adoptive transfer of self-reactive CD4⁺ T cells into lymphopenic hosts lacking a population of Foxp3⁺ T_reg cells, whereby a failure of anergy induction typically leads to dangerous T effector cell clonal expansion and differentiation, with consequent severe immunopathology (31, 32). It is important to note that there are many independent paths to anergy, which result in different types of functional unresponsiveness. The different ways that assays are performed lead to varied readouts because the measurement of anergy has not been standardized. We define anergy in this review as defective proliferation in response to pMHCII recognition by CD4 T cells.

Multiple biochemical signaling defects have been ascribed to the CD4⁺ T cell anergic state, including blocked signaling to Ras and MAPKs 1 and 8, and impaired expression of functional AP-1 transcription factor complexes (21, 33, 34). Upregulation of counterregulatory gene products such as Cblb, Dgka, Rap1, Rnf128, Itch, Dtx1, and Ndrg1 (which in most cases lie downstream of NFAT- and Egr2/Egr3-dependent transactivation) are understood to underlie this development of functional unresponsiveness (35–39). Despite the presence of multiple counterregulatory molecules and signal transduction defects, anergic CD4⁺ T cells retain their capacity to recognize pMHCII and undergo anergy reversal in its absence (29).

Anergy is also reversed in the setting of T cell lymphopenia, regardless of the continued presence of Ag (40). This may relate to T_reg cell deficiency, as well as to increased availability of homeostatic cytokines in lymphopenic hosts. Consistent with this, anergy induction and maintenance are strongly antagonized by IL-2R signaling that leads to downstream PI3K, Akt, and mTORC1 activation, and that promotes a shift in metabolism away from oxidative phosphorylation and toward aerobic glycolysis (41, 42). The activation, clonal expansion, and differentiation of aggressive CD4⁺ T effector cells is associated with unique metabolic demands that require mTORC1 activity to integrate environmental, nutrient, and intracellular signaling cues (43). If activated CD4⁺ T cells fail to increase their glycolytic metabolism because of inefficient mTORC1 activation, they can become anergic (44).

It is worth noting that inhibition of mTORC1 signaling not only induces anergy, but also promotes the differentiation of Foxp3⁺ T_reg cells from conventional precursors and stabilizes the expression of Foxp3 (45–47). Not surprisingly, T_reg cells themselves appear to be anergic, because they do not produce IL-2 or proliferate when stimulated unless exogenesis IL-2 is provided. Nonetheless, mTORC1 activity is also necessary for proliferation by T_reg cells. In fact, it has been suggested that alternate activation and subsequent inhibition of mTORC1 facilitate optimal T_reg cell function and lineage stabilization (48).

Our own investigations have led to the discovery of two additional anergy factors: ecto 5′-nucleotidase (Nt5e; hereafter referred to as CD73) and folate receptor 4 (FR4; Izumo1r). Both molecules are found expressed at moderate levels on Foxp3⁺ T_reg cells (as well as at lower levels on conventional CD4⁺ naive T cells and T follicular helper cells), but CD73 and FR4 are most highly expressed on conventional Ag-experienced CD44^hi CD4⁺ T cells after the induction of anergy (31, 49, 50). Little is understood about the function of FR4 on T cells; however, the gene encoding FR4 (Izumo1r) was recently shown to encode Juno, the receptor for Izumo1 (51). In reproductive biology, Izumo1 expression on sperm and Juno on eggs guide normal sperm–egg fusion during fertilization (51). It remains unclear how the expression of FR4 relates to anergy.

CD73 is an ectoenzyme that normally acts in tandem with CD39 to convert extracellular ATP to adenosine (52). Extracellular ATP is secreted by activated T cells and also accumulates at sites of tissue ischemia and necrosis (53–56). Although the exact mechanisms remain unclear, it is thought that the CD39/CD73-mediated depletion of extracellular ATP limits the triggering of purinogenic receptors such as P2X7. Activation of P2X7 in some systems stabilizes the expression of the glycolytic gene positive regulator Hif1α (57). CD73-dependent production of extracellular adenosine may also serve to resist glycolytic reprogramming through the suppressive effects of the adenosine A2a receptor and its intracellular second messenger cAMP on mTORC1 activity (54). Note that CD39 is not expressed on anergic conventional CD4⁺ T cells, and the role that CD73 plays in the induction or maintenance of anergy remains uncertain. Nevertheless, these findings may indicate one mechanism by which CD39⁺ T_reg cells facilitate the induction of anergy through their conversion of extracellular ATP to the more tolerogenic nucleotide adenosine (58, 59).

Tissue-restricted expression of Aire-regulated transgenic model Ags has clearly defined self pMHCII-specific Foxp3⁺ T_reg cells as an important barrier to peripheral self-reactivity by CD4⁺ T cells, whereas no evidence for peripheral anergy induction was obtained. Nonetheless, extrathymic Aire-regulated expression of self-Ags within secondary lymphoid organs has also been shown to functionally inactivate naive conventional autoreactive CD4⁺ TCR-transgenic T cells, as well as promote the accumulation of self-Ag–specific T_reg cells (60). We have questioned whether the detection of naturally occurring anergic CD4⁺ T cells has been hindered by the lack of sensitive and specific markers for detection of anergic T cells in the polyclonal repertoire.

Our previous discovery of CD73 and FR4 as reliable surface markers of anergic conventional T cells, together with the development of Ag tetramer technologies by our collaborators, offered us the opportunity to thoroughly investigate tolerant polyclonal CD4⁺ T cells (31, 49, 50, 61). In healthy B6.g7 mice, insulin B chain (InsB) pMHCII tetramer-binding conventional polyclonal CD4⁺ T cells were found to be naive in phenotype and generally ignorant of self-Ag (49, 50), consistent with the results of Legoux et al. (5) and Malhotra et al. (4). In contrast, conventional InsB/I-A^g7 tetramer-binding polyclonal T cells isolated from the pancreas-draining lymph node of nondiabetic NOD mice demonstrated: 1) evidence of previous Ag recognition (increased CD44 expression); 2) upregulation of FR4 and CD73; and 3) defective IFN-γ production, all consistent with the induction of anergy (49, 50). Notably, InsB/I-A^g7–specific Foxp3⁺ T_reg cells were also identified and found to be similar in number in both syngeneic strains. These data, therefore, generally supported the model that normal naive CD4⁺ T cells often ignore tissue-restricted self-Ags in the presence of a stable self-Ag–specific T_reg cell compartment. Nonetheless, NOD mice predisposed to autoimmune disease development apparently reveal InsB/I-A^g7 complexes to the naive peripheral T cell repertoire; then the anergy mechanism becomes available to maintain self-tolerance.

Of course, these observations in disease-prone NOD mice begged the question whether anergy develops only when other immune tolerance mechanisms fail. This point was addressed in a series of experiments that examined fetal tolerance in healthy B6 pregnant mice, after mating to syngeneic B6 males made transgenic for a ubiquitously expressed 2W self-Ag (62, 63). Clonal expansion of 2W pMHCII-specific Foxp3⁺ T_reg cells is known to be necessary for fetal success in this system (64), with Foxp3 expression apparently stabilized by the CNS1 enhancer element (65). Given that the niche for T_reg cells having any particular Ag specificity appears limited (66, 67) and the observations of similar metabolic programming for both the anergic and T_reg fates (48), we hypothesized that during pregnancy both anergic T cells and Foxp3⁺ T_reg cells will be present among the 2W-specific T cells. 2W/I-A^b tetramer-binding CD4⁺ T cells were observed to undergo a 5-fold clonal expansion during pregnancy that resulted in approximately equal numbers of unresponsive CD44^hi FR4^hi CD73^hi anergic T cells and Foxp3⁺ T_reg cells (49). Interestingly, most of the anergic compartment disappeared during the postpartum period, perhaps reflecting a requirement for continuous TCR recognition of fetus-derived 2W/I-A^b complexes to maintain anergy or cell survival. Therefore, normal pregnancy is associated with CD4⁺ T cell anergy to fetal Ags.

The discovery of functionally unresponsive CD44^hi FR4^hi CD73^hi CD4⁺ T cells specific for self InsB/I-A^g7 or fetal 2W/I-A^b complexes lends support to the notion that peripheral self-tolerance can rely on CD4⁺ T cell anergy. To investigate the generality of these observations, we characterized the natural repertoire of polyclonal conventional CD4⁺ T cells that express these anergy markers in combination. A subpopulation of CD44^hi FR4^hi CD73^hi Foxp3⁻ cells was found to make up 2–5% of the polyclonal CD4⁺ T cell repertoire in the secondary lymphoid organs (but not thymus) of multiple normal mouse strains, and this subpopulation increased with age (49). Furthermore, expression of these markers was found to strongly correlate with proliferative arrest and defective cytokine production, the two hallmarks of anergy. Finally, loss of Aire-dependent gene expression and central deletion in the thymus of mutant Aire^−/− mice led to an increase in the proportion of peripheral CD4⁺ T cells that have this anergic phenotype (49).

Evidence of anergy in this polyclonal CD44^hi FR4^hi CD73^hi CD4⁺ T cell compartment cannot be taken as proof of self-Ag reactivity. However, loss of the anergic phenotype in fetal Ag-specific T cells postpartum after the expulsion of fetal tissues did suggest a requirement for continuous TCR engagement to maintain the anergic state (49). Consistent with this, steady-state polyclonal anergic T cells were shown to express high levels of PD-1, CTLA4, CD69, and Nrp1, all molecules whose expression can be induced and/or maintained by persistent TCR engagement. Uniformly increased levels of CD5 and a Nur77 reporter gene in anergic T cells also suggested that these cells have high-affinity TCRs specific for available self pMHCII complexes, similar to bona fide self-Ag–specific T_reg cells (68–70). In addition, the frequency and number of anergic cells was not different between germ-free mice and specific pathogen-free mice from our colony, suggesting that the functional inactivation observed here was not solely in response to commensal Ags (L.A. Kalekar, unpublished observations). Therefore, we now hypothesize that many or all anergic phenotype CD4⁺ T cells in secondary lymphoid organs have recently recognized self pMHCII.

Additional experiments were designed to formally test the self-reactivity of anergic phenotype CD4⁺ T cells by transferring them into autoimmune disease–prone Tcra^−/− mice and observing for the development of immunopathology. Initial experiments failed to demonstrate reliable autoimmune disease development, but instead led to the discovery that polyclonal anergic T cells can transdifferentiate into Foxp3⁺ T_reg cells (49). After the adoptive transfer of highly purified Foxp3⁻ CD44^hi FR4^hi CD73^hi CD4⁺ T cells into lymphopenic Tcra^−/− mice, as many as 25% of the resulting peripheral CD4⁺ T cells expressed Foxp3. Analogous to tT_reg cells that experience persistent high-affinity TCR engagements during their differentiation in the thymus, most of these anergy-derived T_reg cells also expressed Nrp1 and demonstrated a fully demethylated T_reg-me (including the Foxp3 CNS2).

Nrp1 was originally thought to distinguish tT_reg cells from peripherally differentiated pT_reg cells (71, 72). However, this notion has since been challenged because activated T_reg cells (which include both tT_reg cells and pT_reg cells), as well as inducible T_reg cells generated in vitro from naive CD4⁺ T cells, can express Nrp1 (73, 74). Similar to natural polyclonal T_reg cells, anergy-derived polyclonal T_reg cells demonstrated an ability to protect lymphopenic Tcra^−/− mice from inflammatory bowel disease (49). Anergy-derived T_reg cells also suppressed the development of autoimmune arthritis and demonstrated a capacity to induce anergy in other self-Ag–specific CD4⁺ T cells (49).

Interestingly, the treatment of Tcra^−/− recipients of Foxp3^DTR anergic T cells with diphtheria toxin to destroy any developing Foxp3-expressing cells not only prevented the accumulation of anergy-derived T_reg cells, but also led to the development of severe wasting disease and the generation of tissue-specific autoantibodies, further demonstrating the self-reactivity of naturally anergic polyclonal CD4⁺ T cells (49). Taken together, the data suggest that anergic phenotype polyclonal CD4⁺ T cells have potentially dangerous TCRs that are specific for peripheral self pMHCII complexes, but these TCRs also make them ideal progenitor cells for the peripheral differentiation of Foxp3⁺ T_reg cells. Furthermore, these experiments indicate that anergy-derived T_reg cells cannot be readily distinguished from tT_reg cells based on phenotype or T_reg-me. Therefore, it is conceivable that some of the Nrp1⁺ Foxp3⁺ T_reg cells that preferentially expanded during the course of self-Ag immunization by Malhotra et al. (4) and Legoux et al. (5) were originally conventional CD4⁺ T cells that had undergone anergy induction after peripheral recognition of the same tissue-restricted self-Ag.

Nrp1 is a transmembrane glycoprotein on the surface of many cell types, including T_reg cells, dendritic cells (DCs), NKT cells, neurons, and endothelial cells (71, 72, 75, 76), and its function is important for axonal guidance in the developing nervous system (77). Nrp1 is a coreceptor for the soluble class 3 semaphorins (78), but can also promote angiogenesis by binding to vascular endothelial growth factor (76). More recently, Nrp1⁺ T_reg cells have been shown to bind to semaphorin-4a on plasmacytoid DCs, with subsequent recruitment of PTEN and inhibition of downstream Akt and mTORC1 signaling pathways (79, 80). The stability of the T_reg cell lineage is maintained by a Nrp1:semaphorin-4a axis, and Nrp1 can induce the expression of T_reg cell lineage-related genes independently of Foxp3 (79, 81).

As described earlier, a majority of polyclonal anergic CD4⁺ T cells express high levels of Nrp1 and demonstrate a unique pattern of partial T_reg-me DNA demethylation. Consistent with this coordinate expression of Nrp1 and demethylation of the T_reg-me, FR4⁺ CD73⁺ anergic CD4⁺ T cells sorted for high Nrp1 expression and transferred to lymphopenic Tcra^−/− hosts were found to be most efficient for the generation of Foxp3⁺ T_reg cells (49). Conversely, the Nrp1⁻ fraction of anergic T cells was found to be a poor source of T_reg cell progenitors, and preferentially differentiated toward a Th17 lineage that caused wasting disease after adoptive transfer to Tcra^−/− mice (49) (L.A. Kalekar, unpublished observations). A previous study similarly showed that autoreactive CD4⁺ T cells lacking Nrp1 would induce a more severe form of experimental autoimmune encephalitis, with their Nrp1⁻ CD4⁺ T cells also appearing biased toward the Th17 lineage (80). Loss of Nrp1 expression on T_reg cells limits the nuclear localization of Foxo1/3a, leads to a failure of Foxp3 expression, and allows for the upregulation of transcription factors such as RORγt that facilitate Th17 differentiation (79). Thus, the skewing of Nrp1⁻ polyclonal anergic CD4⁺ T cells toward a Th17 fate suggests that a similar Nrp1-directed genetic program may govern lineage selection in both anergic T cells and T_reg cells.

It remains unclear whether Nrp1 expression is essential for the induction and/or the maintenance of the anergic phenotype. Nrp1 was previously implicated in immunological synapse formation, and on T_reg cells Nrp1 enhances the duration of the T_reg interactions, with DCs resulting in higher sensitivity to small amounts of Ag (82). How the TCR repertoire and self pMHCII specificity of anergic T cells relates to this expression of Nrp1 remains uncertain. Nrp1 has also been shown to directly induce the expression of CD73 in T_reg cells (79). Nevertheless, our analysis of fetal Ag-specific anergic T cells suggested that Nrp1 is expressed only after the anergic phenotype (CD73^hi FR4^hi) is established. Only half of fetal Ag-specific FR4^hi CD73^hi CD4⁺ T cells expressed Nrp1 at day 10 of gestation, whereas Nrp1 expression on anergic phenotype T cells increased to 80–90% by day 18 (49) (L.A. Kalekar, unpublished observations). Thus, the level of Nrp1 on anergic T cells may simply indicate the degree of unresponsiveness. Alternatively, Nrp1 expression on anergic T cells may reinforce longer interactions with self pMHCII on DCs to promote quantitative demethylation of T_reg-me genes and induce the expression of Foxp3.

Despite recent advances in biological therapy, the treatment for autoimmune diseases remains problematic. For instance, the treatment of rheumatoid arthritis with drugs that suppress aberrant immune responses to self-Ags still poses risk for infection, because these drugs can have off-target effects leading to the suppression of T cells that recognize and destroy pathogens (2). A better approach for the treatment of autoimmune disorders would be to reinforce a stable T cell tolerance to self-Ags, while maintaining full responsiveness to non–self-Ags.

Anergy, a state of long-term functional unresponsiveness, is one such peripheral tolerance mechanism that has been studied extensively. Nevertheless, until recently it has never been shown that anergy can be induced in self Ag-reactive CD4⁺ T cells that escape negative selection in the thymus, mainly because of a lack of identifying markers specific for anergy development. It has also remained uncertain as to why the immune system would maintain viable anergic T cells long term. Our experiments have made use of a panel of predictive markers for anergy development in the natural polyclonal CD4⁺ T cell repertoire: CD44 expression to identify Ag-experienced T cells, the absence of Foxp3 expression to exclude T_reg cells, and a combination of elevated CD73 and FR4. The results now suggest that CD4⁺ T cells that have persistently recognized peripheral self pMHCII enter a CD44^hi CD73^hi FR4^hi unresponsive state. Moreover, anergy cannot be sustained in the absence of self-Ag recognition or in the setting of T_reg deficiency. Anergy reversal can lead to the differentiation of functional T_reg cells that suppress autoimmune disease or, alternatively, potentially pathogenic effector-memory T cells. Finally, the upregulation of Nrp1 expression on anergic T cells is predictive of a partially demethylated T_reg-me and serves as a marker for T_reg cell progenitors (Fig. 1).

Self-Ag–specific T_reg cell generation from anergic T cells is now reminiscent of the in vivo infectious tolerance model previously proposed by Kendal and Waldmann (83). Infectious tolerance is described as a process during which a tolerant state is passed on from one group of lymphocytes to another. Newly tolerant T cells would then reprogram, survey the immune system, and pass on their tolerant state to other T cell populations to continuously maintain self-tolerance. Because T_reg cells are important for inducing and/or maintaining anergy (31), and anergic T cells in turn can alter their epigenetic and transcriptional programs to become T_reg cells (49), anergic T cells may represent the intermediate reprogramming stage before themselves becoming surveying T_reg cells that maintain self-tolerance.

The authors have no financial conflicts of interest.