Type 1 diabetes (T1D) is an autoimmune disease, during which immune homeostasis is destroyed and immune cells selectively attack pancreatic β cells. The development of T1D involves a complex crosstalk among immune cells of both the innate and adaptive immune systems (1). Of note, the dual roles of immune cells are often observed in β cell destruction depending on cell subsets, activation pathways, and immune microenvironment (2, 3). Thus, the specialized suppressive subsets that antagonize over-exuberant immune responses are important in inhibiting β cell destruction.

Regulatory T cells (Tregs) are critical regulators by performing suppressive functions through several well-established mechanisms (4). Studies in animal models, particularly in non-obese diabetic (NOD) mice, have demonstrated a strong association of Tregs and T1D. These cells deficiency in NOD mice accelerated disease progression (5). Therapies targeted to Tregs have also been reported in early phase clinical trials (6–8). However, different studies in T1D patients have reported conflicting results, with respect to the frequency or absolute numbers, as well as functional defect of Tregs from peripheral blood (9–13). As such, considering heterogeneity and diversity of circulating Tregs, do distinct subpopulations of Tregs engage as an important player and regulate the pathophysiology in the home tissue of T1D?

Actually, distinguishable Treg subsets that reside in tissues have attracted interest. Those tissue-resident Tregs exhibit specific phenotype and function in response to local cues, thereby promoting tissue homeostasis (14, 15). The specialized distribution of Tregs has provoked the assessment of how target tissue-resident Tregs control the development of diabetes. Now, it is evident that pancreatic-resident Treg subsets distinct from the Treg population that exist in peripheral blood and spleen (9, 16, 17). In this review, we focused on the unique features of Tregs residing in the target tissues, namely pancreatic tissues and pancreatic lymph nodes (PLNs), and how these special Tregs were maintained and functioned throughout diabetes progression.

Regulatory T cells constitute 10–15% of the total CD4⁺ T cell population in whole body but constitute a much higher proportion in local tissues such as visceral adipose tissue and intestinal tissue (18, 19). The question of whether proportion of pancreatic-resident Tregs has a similar tendency has been examined with NOD mice. In NOD mice, a small population of forkhead box P3 (Foxp3)⁺ Tregs (8–20% of CD4⁺ T cells) resided in pancreatic tissues, in comparison with 10–15% Tregs within CD4⁺ T cell compartment in lymphoid organs such as PLNs and spleen (20, 21). It gradually fluctuated with age and disease progression both in PLNs and pancreatic tissues (20–22). Controversial results were obtained when detecting Foxp3 transcripts (16), whose expression was important for the establishment and maintenance of Treg features (4). The study demonstrated that the levels of Foxp3 failed to be detected in the PLNs of prediabetic and diabetic mice (16). Divergent results are likely to reflect the mice with different gene manipulation (NOD mice versus Foxp3-GFP reporter mice), disease stages, timing and technology of Treg purification, and Treg subpopulation involved. It will be important to obtain a more global view of Treg proportion with age and disease development in T1D.

Human Treg in local pancreas has been studied only to a limited extent, but available evidence indicates that amount of Foxp3 expressed in CD4⁺ T cells and in CD25^bright T cells was similar within the PLNs of diabetic and non-diabetic donors (9). With difficult access to human tissues, these findings accomplished by experimental model are difficult to validate in humans. Therefore, the precise changes of pancreatic-resident Treg subsets need to be elucidated in T1D.

Comparative transcriptome analysis revealed that pancreatic-Tregs have quite different profile from their counterparts in spleen, including differential expression of suppressive mediators (Il10, Fgl2, and Lag3), chemokine receptors (Cxcr3 and Ccr5), and transcription factors that typically denote cell activation (Nr4a2, Fos, and Jun) (21). Although some of their major suppressive mediators such as Il10 and Lag3 also engaged in other tissue-resident Tregs, principal components analysis revealed that differential expression genes in pancreatic-Tregs involved in cell growth and proliferation (21). Transcriptional profiling studies which compared pancreatic-Tregs and effector T cells (Teffs) helped to further dissect differential roles of pancreatic-Tregs. Again, genes encoding functional molecules such as glucocorticoid-induced TNF receptor (GITR), CD103, Nrp-1, IL-10, and CTLA-4 were overexpressed, genes encoding transcription or signaling factors such as OBF-1, Tcf7, Eomesodermin, and Smad1 were underrepresented (23).

These data indicate that pancreatic-Treg populations have a tissue-specialized transcriptome, which may reflect their unique features in proliferation, retention, and function within pancreas in the context of T1D. Of note, there exists no information on the phenotypic particularities of Tregs residing in human pancreas, additional analysis of T1D patients are needed. In addition, skewing TCR repertoire from PLN of T1D patients has been reported (24), but no data demonstrated whether pancreatic Treg TCRs were involved.

Regulatory T cells can generate in the thymus (termed thymic Treg, tTreg) and induce in the periphery by conversion of CD4⁺ Foxp3⁻ T cells into CD4⁺Foxp3⁺ Tregs (termed peripheral Treg, pTreg). Where and how T cells are developed into pancreatic-resident Tregs is currently unknown, although recent studies indicate some specific factors exist for their maintenance in pancreas. For example, CXCR3⁺ Tregs in pancreas are crucially dependent on transcription factor T-bet (21), possible positive feedback loop of IL-2–ICOS is important for pancreatic ICOS⁺ Treg homeostasis (36). Furthermore, the possibility remains that tTregs constitute a substantial proportion of pancreatic-resident Tregs. This hypothesis is supported by the fact that pancreatic CXCR3⁺ Tregs were regulated by T-bet (21), but this regulation was only observed in pancreatic natural Tregs (64). However, further investigation is required to comprehensively define the origin of pancreatic-resident Tregs that coordinate protective effects in T1D.

The pattern of Treg accumulation in pancreas is governed by adhesion molecules, chemotactic molecules and chemoattractant receptors that expressed or produced by Tregs and pancreatic environmental cues (Figure 1). In the context of T1D, CXCR3⁻ Tregs show impaired accumulation in pancreatic islets. This effect has functional consequences, as mice defective in CXCR3 could not effectively suppress Th1 activity and developed earlier onset of diabetes (76). Besides CXCR3, pancreatic Tregs express several other chemokine receptors such as CCR2, CCR8, and CXCR6, which were thought to compensate for the absence of CXCR3 in promoting Tregs localization to the islets. For example, early studies described the migration pattern of Tregs in transplanted pancreatic islets and showed that Tregs needed to be educated first in the inflamed islets before entering the draining lymph node (77). Tregs migrated from blood to the inflamed islets depending on CCR2, CCR4, CCR5, and P- and E-selectin ligands, whereas Tregs migrated to draining lymph nodes in a CCR2, CCR5, and CCR7 fashion (77). Other molecules such as CCL12 and MAdCAM-1 expressed by islet, SDF and Integrin α4β7 expressed by Tregs were required for Treg localization in the pancreas (16, 78–80). Furthermore, pancreatic-resident antigen-presenting cells, specifically F4/80⁺ macrophages and CD11c⁺ DCs, were commonly expressed CXCL9, CXCL10, and CXCL11, which specifically induced CXCR3⁺ Tregs migration in islets (35). These data suggest that substantial migration is required for Treg suppressive function in local setting, this process involves in an array of molecules that act in a redundant fashion. No detailed insights exist as to which of these molecules are more important for Treg migration between pancreas and PLNs. It is also not clear whether these migration patterns performed by pancreatic Tregs are primarily in the context of T1D or any other inflammatory responses.

Transcriptional factors were also observed to affect Treg migration in T1D. T-bet^−/− natural Treg expressed more CCR4 and migrated better to CCL22, which contributed to T-bet^−/− Treg in the islet (64, 77). Also, T-bet affected expression of sphingosine-1-phosphate (S1P) receptor 1 (81), which bonded to S1P and promoted migration of Tregs from tissues into the blood (82). Thereby, T-bet deficiency resulted in a defective migration of Tregs across lymphatic endothelial cells and was accompanied by retaining T-bet^−/− Treg in pancreatic tissues (81). Similarly, CD103 (also known as αE integrin) expression can be influenced by T-bet (36, 64). CD103 formed a heterodimer with β7 integrin and contributed to specific retention of lymphocytes in tissues by binding to E-cadherin on epithelial cells (83, 84). CD103 has been used to identify tissue-resident lymphocytes (85); hence, CD103 upregulated by T-bet deficiency was functionally involved in retaining T-bet^−/− Tregs in islets (64). In addition, as described above, T-bet induced CXCR3 expression on Tregs, which was prone to residing in pancreas (76). Altogether, chemokine and chemokine receptors are regulated by T-bet in Tregs, and their function in terms of ligand interactions indicate that T-bet is involved in the retention of Tregs in pancreatic tissues.

It should be noted that mechanisms underlying Treg migration are shared with those of pathogenic effector cells. For example, α4β7 was highly expressed on islet-infiltrating lymphocytes in NOD mice, suggesting that the MAdCAM-1 in inflamed islets also drove the migration of pathogenic Teffs into islets (78, 86). It has been described that immunocytes moved into the pancreas in a dynamic manner during disease progression. Insulitic lesion was continuously infiltrated by a mixed influx of immunocytes: both naïve- and memory-phenotype lymphocytes could traffic to the insulitis, but Tregs had less active migration than their conventional CD4⁺ counterparts (87). Thus, there is a lot of interest to know when and how preferential recruitment of Tregs was happened in pancreas in the context of T1D. A more complete understanding of the detailed migration mechanism that is necessary for Treg accumulation in pancreas remains to be elucidated and will be important for expansion and development of therapeutic Tregs in response to the tissue-specific environment.

Regulatory T cells potently suppress the activation and function of other immune cells through directly cell-to-cell contact or indirectly secreted factors. In addition to regulate immunological process, growing evidence showed that Tregs, particularly tissue-resident Tregs, also regulated non-immunological functions, which has been well studied in muscle and visceral adipose tissue. Thus, in the context of T1D, non-traditional roles should be considered and evaluated as an important facet of pancreatic-resident Tregs.

Although there are no data so far to directly show non-immunological role of pancreatic-resident Tregs, there is evidence suggesting a role for Treg-derived mediators in regulating β cells. First, ICOS⁺ Tregs expressed IFN-γR at higher levels and prevented IFN-γ production. Insulin⁺ β cells expressing IFN-γR could respond to IFN-γ, thereby, promoting apoptosis of islet cells. Thus, Tregs may have an indirect role to prevent the destruction of the islets (35, 88). Second, in PLNs CXCR4/SDF-1 axis was involved in retaining Tregs, these molecules also contributed to the regeneration of autologous β cells in the antea-diabetic model (16). Third, TGF-β has been viewed as a protective cytokine in β cell mass, and consequently, TGF-β signaling has an important role in pancreatic development and islet homeostasis (89–91). However, a recent study described that inhibition of TGF-β pathway promoted β cell replication, which suggests that Tregs-secreted TGF-β might have a detriment effect on maintenance and expansion of β cell mass (91). This study found that TGF-β signaling induced age-related accumulation of p16^INK4a that led to pancreatic β-cell proliferation decline; by contrast, treatment with TGF-β inhibitor exhibited a significant replication of β cells in both mouse and human (91). Although less well characterized, based on our current understanding, it is plausible that Tregs that are enriched in the pancreas may contribute to protecting β cells at least in part by non-immune-mediated ways.

However, these studies raise the question whether pancreatic-resident Tregs could secrete some factors that specifically target to β cell and promote their repair and regeneration. These tissue-specific factors will allow Tregs to reside in pancreatic tissues and act appropriate effector function in response to β cell destruction. However, given the distinct pancreatic-resident Treg subsets already found in T1D, it seems that their primary role is to prevent β cell destruction by locally restraining inflammatory responses of both innate and adaptive immune cells (49, 92–94).

The identification of pancreatic-resident Tregs and recognition of their diverse phenotype and potent immunosuppressive activity have provided new insights into how Tregs function in tissue-specific environment and why the results of manipulating Tregs for the treatment of T1D have been disappointing in some cases. Although present in pancreas in relatively low numbers, the selectively higher proportion of some unique subsets in pancreatic infiltration and PLNs seems to confer a remarkable ability to regulate physiological process during T1D. Thus, the ability of pancreatic-resident Tregs to hamper β cell destruction clearly emphasizes its use as an immunotherapeutic strategy in T1D. Tailoring proliferation and apoptosis of these Tregs in response to pancreatic-specific inflammatory milieu would promote immune balance, allowing for preservation of remaining β cells in T1D. Indeed, islet-specific IL-2 has proved beneficial in the treatment of NOD mice, and increased pancreatic-resident Treg numbers (55). However, as the study of pancreatic-resident Tregs is a new field, many questions regarding their properties remain unaddressed.

Given the diversity of tissue-resident Tregs already found, transcriptional factors that are specific to corresponding tissues are contributed to their non-immunological activity, for example amphiregulin in the muscle, PPAR-γ in the VAT. The importance of this property failed to describe in the pancreatic-resident Tregs. In contrast to pancreatic tissues, tissues that already found specific-Tregs are under constant threat of invasion and prone to damage. It might be that islet surrounding microenvironment did not support the specific functions of Tregs, as individual tissues have unique challenges and immune response. It is also difficult to rule out the possibility that such novel Tregs residing in pancreas are unrecognized. It is reminiscent of the finding that BLIMP-1- and IRF4-dependent Treg signatures were biased in pancreatic Tregs compared with spleen Tregs, but their function has not been assessed (21). These two transcription factors BLIMP-1 and IRF4 have been reported for differentiation and function of effector Tregs in tissue homeostasis (95). Thus, future studies must identify the precise regulators of Tregs that mediates functional difference in the pancreatic microenvironment, knowing this will probably reflect pathophysiological complexity and therapeutic targets of T1D.

Despite evidence has revealed an important role of pancreatic-resident Tregs in T1D mice, very few current studies have performed detailed investigations of these cells in patients with T1D. Human CD4⁺ and CD8⁺ T cells in islets lack expression of CD25 in their healthy state (96), and only ~5% Tregs can be found in PLNs of T1D (24). In addition, limited knowledge is available with respect to immunosuppressive effector of pancreatic-Tregs, although functional defects in Tregs have been partly characterized by analysis of the Treg-specific demethylated region within the Foxp3 locus in PLN of T1D patients (9). These data raise the question whether Tregs recruit in pancreatic tissues only in physiological state in human. And the hypotheses that have aimed to ascribe human T1D pathophysiology to the defect of pancreatic-Tregs are confounded by our limited understanding of Treg population within the islet. Thus, a more complete understanding of features of human pancreatic-Tregs would be important for evaluating current efforts to target human Tregs with intra-islet in the context of T1D.

Finally, the question remains how important of pancreatic Tregs is for T1D development. The immune systems have evolved so that different cell subsets with specialized function collaborate to influence diabetes progression; therefore, it is important to consider dominant populations in this process. So far, the ability of pancreatic-Tregs to delay the onset of diabetes has been seen as a sign of superior function, but this phenotype could also represent systematic Tregs. In addition, it remains unclear how the pancreatic-Tregs modify immune responses by cooperating with immune cells of both the innate and adaptive immune systems. Increased in-depth understanding of these issues discussed above is need for the better development of therapeutic interventions in T1D.

JL, CZ, XZ, and LL participated in compilation and editing of the manuscript. JL, CZ, LL, XZ, WX, and CZ participated in literature and data collection. All authors read and approved the final version.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.