Barrier sites are constantly exposed to external antigens and face the challenge of ensuring a potent immune response against pathogens while maintaining tolerance to commensals and harmless antigens. This challenge also extends to Tregs, which at these sites are largely specific for microbial antigens and to a large degree differentiate into Tregs at the barrier site itself (23, 64, 65).

In the gut, the microbiota and the host tissue are separated by a physical barrier of intestinal epithelial cells (IECs). An extensive crosstalk exists between immune cells in the host tissue and IECs that conveys information from the gut lumen to the immune compartment and from immune cells to the IECs and the microbiota in the gut lumen ( Figure 3 ). Foxp3⁺ Tregs that reside in the intestinal tissue are in their great majority induced from circulating conventional T cells, but there is also a fraction that originates in the thymus and homes to the intestine (66). In the lamina propria of the small intestine, a Treg niche has been characterized and some studies have addressed the possibility that these cells migrate to and are maintained in this organ independently from other immune system cues such as antigen encounter (67). In vivo studies with mice lacking peripheral MHC II expression have shown that, despite a reduction in the total CD4⁺ T cell count, adult MHC II KO mice have an increased frequency of Foxp3⁺ Tregs in the lamina propria (67). Despite the absence of MHC II, these Tregs show high CD44 expression, indicative of an activated phenotype. Furthermore, intestinal Tregs have a reduced expression of the IL-2 receptor subunit CD25 and, unlike other Tregs, their proliferation and maintenance is independent of IL-2 (15, 67). Instead, continued ICOS signaling is important for their maintenance (15). In addition, the alarmin IL-33, which is constitutively expressed by epithelial cells at barrier sites (68), acts as an intestinal survival factor for Tregs. The IL-33 receptor ST2 is enriched on colonic Tregs and IL-33 promotes expansion of these ST2⁺ Tregs (24, 69). Furthermore, IL-33 signaling also enhances TGF-β-mediated differentiation of Tregs and thus promotes Treg accumulation and maintenance in the inflamed gut tissue (69).

Interestingly, IL-33 not only plays an important role as a local survival factor for Tregs, but can also modulates their function by inducing Areg production (12). Areg promotes tissue regeneration by signaling through the epidermal growth factor receptor and multiple studies have demonstrated the ability of Tregs to produce Areg and promote tissue regeneration (11, 12, 70, 71). However, whether ST2⁺ Tregs in the gut also produce Areg and exert similar tissue protective and regenerative functions still remains to be determined. Besides Areg, which would directly act on barrier tissues, Tregs also produce immunomodulatory cytokines to promote homeostasis. One such anti-inflammatory cytokine is IL-10, which plays a crucial role in maintaining tolerance to commensal microbiota in the gut (5). IL-10 promotes tissue integrity by dampening effector immune responses against bacteria, viruses and parasites and hence dampening immuno-pathology. At the same time, it can also enable pathogen persistence, which may result in pathogen-induced pathology (72). Selective deficiency of IL-10 in Foxp3⁺ Tregs results in spontaneous colitis (73), suggesting that IL-10 is a key effector molecule for Treg-mediated tissue regeneration and protection. This likely not only holds true for the regulation of immune responses against microbiota, but also for those elicited upon infection ( Figure 3 ). Overall, intestinal Tregs represent a highly specialized Treg subset that is essential for limiting inflammation through its secretion of anti-inflammatory cytokines. Furthermore, additional mediators of tissue protection, such as Areg, may enable them to actively promote tissue regeneration and thereby maintain tissue integrity.

Non-canonical Treg functions are strictly associated with their localization and tissue adaptation as well as a fine-tuned crosstalk with the surrounding tissue. A good example of this is the gut, where continuous exposure to microbe- and food-derived antigens shapes the Treg niche to undergo a specialization driven by the tissue and food antigens present at this site. As such, intestinal Tregs can lose expression of Foxp3 and convert into non-inflammatory CD4⁺ intestinal epithelium lymphocytes (IELs) once they have migrated to the damaged epithelium ( Figure 2 ). In this environment, they can co-operate with the remaining Foxp3⁺ Tregs, performing complementary roles in balancing intestinal inflammation and promoting an anti-inflammatory state (74). Notably, these local Tregs strictly depend on the intestinal microenvironment, i.e. microorganism-derived signals and defined components of the microbiota, for proper function and development (75). In vitro studies using a reporter cell line for TCR engagement showed that TCR activation in Tregs did not occur in the absence of microorganism-derived antigens (64). Nevertheless, germ-free mice have similar Treg frequencies and numbers as specific pathogen-free mice (76). In contrast, antigen-free mice harbor a dramatically reduced Treg population. Despite this dramatic reduction, antigen-free mice still maintained a small intestinal Treg population suggesting that not the entire Treg population in the gut is induced by local antigen (76). Besides providing antigen, commensal microorganisms in the gut also produce metabolites, such as the short-chain fatty acid (SCFA) butyrate, that promote Treg differentiation in vivo (77). A similar function has been assigned to food-derived fatty acids (78, 79). Dietary-derived SCFAs expand gut Tregs by suppressing the JNK and p38 signaling pathways (80). Conversely, dietary long-chain fatty acids (LCFAs) can induce Th1/Th17 differentiation, while decreasing SCFAs in the gut (80). LCFA treatment resulted in exacerbated autoimmune disease in the EAE model, which could be rescued by treatment with SCFAs via differentiation of lamina propria-derived Tregs (80). In addition to directly enhancing Treg differentiation, SCFAs produced by the microbiota enhance TGF-β1 production by IECs (81) ( Figure 3 ). TGF-β1 then directly drives the conversion of conventional T cells into Foxp3⁺ Tregs in peripheral organs (82). Thus, SCFA-induced TGF-β1 production generates an environment that also indirectly promotes Treg induction in vivo. Specific symbionts in the mouse gut microbiome have been found to induce a subset of Tregs that co-expresses the transcription factor Rorγ together with Foxp3 (65, 83). This Treg subset plays an important role in controlling colonic inflammation. Intestinal Rorγ⁺ Tregs are mostly Helios^- and express high levels of IL-10, ICOS and CTLA-4 (65, 83), which mediate the strong suppressive capacity of these specialized Tregs. Nevertheless, whether these microbiota-induced Tregs only contribute to tissue homeostasis by limiting effector T cell activation or whether they also directly promote tissue protection and regeneration through release of tissue regenerative factors such as Areg still needs to be investigated. However, the studies we summarized here highlight the heterogeneity and plasticity of T cells and Tregs in the intestine as well as the fact that these Tregs are shaped by microbiota- and food-derived antigens and specialize to limit inflammation of the intestinal tissue and potentially participate in its renewal when damaged.

The lung represents another non-lymphoid organ where Tregs have shown a strong involvement in tissue maintenance beyond the canonical immune homeostasis ( Figure 2 ). Like the gut, the lung is highly exposed to external pathogens and microorganisms, which can potentially cause harm. Adaptive immune cells including Tregs colonize the respiratory tract and like in the gut, the Tregs have to maintain the delicate balance between allowing efficient immunity against pathogenic threats while maintaining tissue homeostasis and integrity.

Non-classical roles of Tregs in the lung tissue are still under investigation, but several independent studies have demonstrated their active involvement in lung epithelium regeneration (84, 85). Indeed, multiple studies have shown the ability of lung Tregs to produce Areg in response to IL-33 (11, 12, 70, 71). During lung inflammation, ST2⁺ Tregs exert tissue protective and regenerative functions through Areg production. Furthermore, Foxp3⁺ Tregs in the lung produce keratinocyte growth factor upon LPS stimulation. This factor, together with Areg, induces epithelial cell proliferation and subsequent tissue regeneration upon injury (85) ( Figure 3 ).

Tregs have also been studied in acute lung injury (ALI), which leads to a progressive influx of CD3⁺ T cells including Tregs into the respiratory tract. Importantly, Treg influx into the alveolar compartment correlates with the transition from injury to resolution and Treg infusion promotes resolution of LPS-induced lung injury in mice, resulting in increased survival and reduced BAL protein and cell count (84). Interestingly, Tregs directly act on alveolar epithelial cells during the recovery phase of ALI by enhancing their proliferation and thereby accelerating resolution and lung tissue regeneration (86) ( Figure 3 ). Furthermore, IL-10-producing Tregs play an important role in limiting type 2 inflammation in the lung. In an OVA-induced lung inflammation model, lack of Treg-derived IL-10 results in exacerbated immune pathology characterized by increased mucus production, Goblet cell expansion, edema, as well as lymphocyte and eosinophil infiltration (73). Nevertheless, Tregs can also have a negative impact on the lung tissue by promoting fibrosis through secretion of IL-10 and TGF-β (87). This once again highlights the need for a fine-tuned balance to allow prevention of immune-pathology while promoting tissue regeneration.

In addition to limiting lung pathology in settings of sterile inflammation, Tregs also play an important role in mitigating lung damage upon infection. Like upon injury, lung Tregs release Areg in response to Influenza Virus infection, thereby counteracting pathogen-induced as well as immune-mediated pathology (12). Importantly, this tissue protective effect is independent of the immune suppressive function of Tregs and their classical regulatory role (12). In addition, Tregs have been addressed in the context of SARS-CoV-2 infection and several studies revealed that in patients with severe COVID-19, marked by potent cytokine storm, Tregs might be beneficial for the lung environment, dampening the pro-inflammatory cues and restoring tissue integrity (88) ( Figure 3 ). On the other hand, excessive immune regulation by Tregs may hamper an effective immune response against SARS-CoV-2 and thereby exacerbate disease severity (89). Of note, Tregs in COVID patients are phenotypically distinct from Tregs observed upon Influenza Virus and Respiratory Syncytial Virus infection (89, 90). As mentioned earlier, Tregs up-regulate CXCR3 in response to the infection, which directs them to the lung as the site of an ongoing immune response (44). While the recruitment of T-bet-expressing CXCR3⁺ Tregs to the site of type 1 immune responses is essential for the efficient control of these responses (44, 45), it is still unknown whether the same Treg subset also exerts the tissue protective functions observed upon infection (12). Understanding the exact functions of each Treg population will be particularly important in infectious settings in which immune-mediated damage correlates with disease severity, such as Influenza or COVID-19 (91, 92). In conclusion, Treg function may contribute to pathology if they excessively limit immunity to infection. Nevertheless, Tregs also play an important role in limiting sterile, immune-mediated, or pathogen-induced tissue damage by shaping the immune environment but also by acting directly on epithelial cells.

As discussed, the main function of Tregs at barrier sites is to maintain tolerance where continuous environmental cues are faced. In addition, accumulating evidence for a memory phenotype of Tregs at these sites has emerged. A clear phenotypical characterization of a Treg memory subset is still lacking but, in the past years, evidence of the persistence of memory-like Tregs in peripheral lymphoid and non-lymphoid tissues has arisen, especially at barrier sites (93). Functional Treg memory was assessed in the lung, where adoptive transfer of memory but not naïve Tregs from SLOs attenuated lung pathology following Influenza infection (94). Similarly, memory Tregs capable of suppressing a secondary inflammatory response upon antigen re-encounter have been observed in pregnancy. In this context, fetal antigen-specific Tregs persist in SLOs even beyond the pregnancy. These Tregs re-expand following re-encounter with the same antigen in a subsequent pregnancy and maintain fetal tolerance (95). However, whether fetal-antigen-specific Tregs are also maintained as a tissue-resident memory population or are only present in SLOs still needs to be determined. Guo and colleagues defined a subset of CD4⁺CD25⁺Foxp3⁺ T cells in the lamina propria which has effector/memory-like properties and is characterized by low expression of CD62L and CD45RB while expressing CD44 and the chemokine receptors CCR4 and CCR9, which are not exclusively expressed on memory cells but mark circulating and tissue-infiltrating lymphocytes (96) ( Figure 1 ). However, whether these memory Tregs are maintained in the tissue and represent a truly tissue-resident population is unclear. Moreover, a mouse model allowing for the inducible expression of a self-antigen in the skin clearly showed that a memory Treg population is maintained following antigen clearance and contraction of the effector cytokine response (97). A follow-up study then also identified the corresponding memory Treg subset in humans (98). Moreover, IL-2 has been shown not to be required for the generation of memory Tregs in the skin, while IL-7 is essential for maintaining them at steady state in this organ (99). While these studies showed that memory Tregs persist in the skin, it is still unclear whether they are replenished from SLOs or whether they represent a truly tissue-resident memory population that is maintained locally.

The existence of memory Tregs has been explored for decades now, but a distinct characterization of their phenotype, maintenance, and mechanism of action in the peripheral tissue is still missing and will require further investigation. Nevertheless, as highlighted above, accumulating data suggest that Treg memory may play a key role in tissue protection in peripheral organs. However, as memory Tregs share phenotypical characteristics and functional features with effector Tregs, further studies are needed to clearly define their markers to allow for a clear identification of these cells. Furthermore, it will be important to determine whether memory Tregs residing in peripheral tissues can recirculate or whether they are tightly restricted to the tissues and represent a tissue-resident population.

In contrast to barrier tissues, the central nervous system (CNS) is shielded from external antigens and holds immune privilege, conferring protection from potential damage of the peripheral immune system to this site. As such, Treg numbers in brain tissue at steady state are very low, under 100 cells in most of the cases. Despite this low count, Tregs are present and show a similar phenotype as those observed at barrier sites in association with tissue repair ( Figure 2 ). Like barrier Tregs, CNS Tregs express the IL-33 receptor ST2. In addition, they also express the serotonin receptor as a tissue specific receptor and can survive and proliferate in absence of IL-2 upon IL-33 and serotonin sensing (40).

While the CNS is well guarded from immune-mediated insults, some pathogens may exploit the immune privilege and target the CNS to expand. When that happens, Tregs (and other T cells) can infiltrate the brain upon inflammation, following the CCL1 and CCL20 gradient produced by astrocytes (40). Effector Th1 cells and memory cells may enter the CNS through the blood-brain barrier, while Tregs and Th17 can reach the CNS through the choroid plexus (100), emphasizing a difference between these subsets of CD4⁺ T cells. This variation in the entry site could influence their final location, and may therefore influence their main function and purpose as well (101, 102). After infiltration, the main role of Tregs is to suppress ongoing immune responses and dampen inflammation to protect neural integrity. Nevertheless, other functions have been attributed to them, making Tregs a key protective cell population of the nervous system ( Figure 2 ). To start with, Tregs can control overreactive microglia, either by reducing their activation and cytokine secretion, by decreasing their production of neurotoxic factors (e.g. nitric oxide, NO, and inducible nitric oxide synthase, iNOS) through driving a shift to the anti-inflammatory M2 phenotype or by inducing their apoptosis (103–105). As NO is one of the major causes of neuronal apoptosis, this role of Tregs is crucial. On the other hand, astrogliosis (or abnormal astrocyte growth) can also be dampened by Tregs through the release of Areg (40) ( Figure 3 ). Moreover, Areg can potentiate neural stem cell proliferation (106). Again, the fact that Areg also acts in immune privileged tissues highlights the importance of Treg-derived Areg in tissue protection. During sterile CNS damage as it may occur during a stroke, Tregs can contribute to the resolution of the acute phase either directly or by promoting macrophages to adopt a M2 phenotype (107, 108). Moreover, Treg depletion during EAE impairs remyelination, a property that can be regained upon Treg transfer (109), and explains why a diet restricted in SCFAs reduces axonal damage through Treg induction, as discussed in detail in an earlier section (80). In addition, the fact that Treg numbers do not decrease during the remission phase, as it occurs for CD4⁺ effector T cells, highlights that the role of Tregs in tissue protection continues beyond the inflammatory phase (110).

Despite their important role in maintaining CNS homeostasis, Tregs are a double-edged sword when it comes to CNS infection: their proliferation and activation can dampen effector immune responses, allowing e.g. viruses to spread in the CNS (111). On the other hand, Treg depletion can allow excessive infiltration and activation of effector T cells in the brain, causing strong immune pathology and damage (112). Furthermore, Tregs have been observed to directly act on infected innate immune cells and drive their apoptosis via the caspase-3 and the perforin/granzyme B pathway (61). Hence, Tregs can be involved in viral clearance from the brain, as observed in Human Immunodeficiency Virus (HIV) infection (113). Moreover, during HIV infection, brain-infiltrating Tregs can potentially promote the conversion of M1 to M2 macrophages, which adopt an anti-inflammatory phenotype, and thereby protect the morphology of the dendrites (114). IL-10 secretion by Tregs has also been associated with neuroprotection upon viral infection, independent from immune suppression (115). In the case of Herpesvirus, Tregs from acute murine cytomegalovirus (MCMV)-infected animals prevent neural damage by restraining microgliosis and astrogliosis and can avoid hippocampal injury and cognitive impairment (116). Interestingly, Tregs also promote memory T cell formation in the CNS after MCMV-infection (117). Therefore, besides acting on effector T cells to suppress collateral T cell damage, Tregs can actively support formation of immunological memory. Treg depletion prior infection with the model coronavirus MHV (mouse hepatitis virus) did not interfere with T cell infiltration but led to increased apoptosis of neurons and myelin loss (118), suggesting that rather than being mainly involved in T cell suppression, CNS Tregs have a central role in tissue protection after infection. In line with this, it is tempting to hypothesize that brain Tregs could ameliorate the neurological symptoms derived from COVID-19 infection (119).

All in all, the potential beneficial roles of Tregs in CNS will depend on the circumstances. Most studies focus on the role of Tregs in limiting effector T cell activation and immunopathology, while little is known about their direct impact on neural cells. Regarding viral infections, we understand that Tregs hamper viral clearance in the periphery and may thereby facilitate virus infiltration in the CNS. Nevertheless, depending on the type of virus, Tregs can also promote anti-viral effects and may play a favorable role, particularly during the late stage of encephalic disease, when their tissue protective functions take the lead.