Tregs serve as a central regulator of immune homeostasis, with their complex heterogeneity stemming from diverse developmental origins, phenotypic characteristics, and functional mechanisms. Characterized by high immunosuppressive capacity, Tregs constitute approximately 10% of CD4⁺ T cells in healthy individuals. They are defined by the expression of transcription factor FoxP3 (8), and are critical for controlling immune responses and maintaining peripheral tolerance (12–15). In humans, Tregs are often defined by the combined expression of CD3⁺CD4⁺CD25^hiFOXP3⁺CD127^low/− (16). Based on the expression patterns of FoxP3, CD25, and CD45RA, they can be further divided into three distinct populations. Fraction I (Fr. I) comprises CD45RA⁺FoxP3^loCD25^lo naive Tregs, which are considered bona fide Tregs characterized by a demethylated Treg-specific demethylated region (TSDR) and full suppressive capacity. Upon activation, nTregs differentiate into Fraction II (Fr. II), the effector Tregs, which exhibit a CD45RA^-FOXP3^hi CD25^hi phenotype and possess potent suppressive activity. In contrast, Fraction III (Fr. III; CD45RA^-FOXP3^loCD25^lo) represents a highly heterogeneous population composed of a mixture of genuine Treg and non-Treg cells (17). Recent studies have successfully refined this subset using additional surface markers such as CD127, CCR4, and CD49d. For instance, the CD127⁺CD25^lo subpopulation exhibits a fully methylated TSDR and high inflammatory cytokine production, which categorized them as non-Tregs. Furthermore, the CD49d⁺CCR4⁻ phenotype can distinguish an inflammatory subpopulation capable of secreting IFN-γ and IL-17. Importantly, circulating T follicular regulatory (cTfr) cells are also found within Fr. III. These cells show stable FoxP3 expression, a demethylated TSDR, and suppressive function, confirming their identity as authentic Tregs and accounting for about 30% of Fr. III cells in healthy individuals (18). Therefore, Fr. III should no longer be considered as a uniform entity but must be precisely delineated using these newly defined markers (19). However, human and mouse Tregs are heterogeneous in terms of developmental origin, functional activity, and activation status. Based on cellular origin, Tregs are classified into two main types: thymus-derived Tregs (tTregs), which develop in the thymus and comprise the majority of the Treg pool in secondary lymphoid organs. By contrast, peripherally induced Tregs (pTregs) originate from naive FoxP3⁻CD4⁺ T cells at inflammatory sites in the periphery where they acquire FoxP3 expression (20). Conventional T cells (Tconv) can also differentiate in vitro to form induced Tregs (iTregs), though these cells typically lack the complete epigenetic programming for stable Treg gene expression and exhibit functional instability (20, 21). Therefore, the pTreg category encompasses both in vivo-derived cells and in vitro-generated iTregs (20).

Tregs can be further categorized by activation state into natural/thymic Tregs (nTregs), activated Tregs (aTregs), and effector Tregs (eTregs). CD62L⁺CD44⁻TCF1⁺ nTregs stimulated by TCR signaling in the presence of IL-2 transit into CD62L⁻CD44^(mid/hi)TCF1⁺ aTregs, which subsequently differentiate into CD62L⁻CD44^hiTCF1⁻ eTregs (20). Although the newly produced nTregs possess suboptimal immunosuppressive capacity, upon TCR engagement in lymph nodes, they proliferate and differentiate into highly suppressive eTregs. These eTregs suppress the antigen-specific maturation of APCs such as dendritic cells (DCs) (22). Conversely, eTregs deplete IL-2 through increased sensitivity to the IL-2 receptor and secrete inhibitory cytokines including IL-10, TGF-β, and IL-35, thereby suppressing immune function in an antigen-nonspecific manner (23). A specialized Treg subset, type 1 regulatory T (Tr1) cells (FoxP3⁻IL-10⁺ CD4⁺ T cells), do not express FoxP3 but secrete high levels of IL-10, playing a vital role in promoting peripheral tolerance (24). Notably, elevated levels of Tr1 cells are associated with better management of blood glucose levels in new-onset T1D patients (25), and preclinical studies in autoimmune mice confirm their protection can be harnessed through antigen-specific induction via peptide-MHC-coated nanoparticles (26). These findings suggest that a strategy of fostering antigen-specific Treg populations directly within the patient could be a viable way to exploit Tr1 cells for therapy (27).

Key molecular markers of Tregs not only establish their cellular identity but are also critical determinants of their functional stability and suppressive capacity. The most canonical identity markers include FoxP3, the critical transcription factor, and surface markers such as CD25 and CTLA-4 (8). Among which, FoxP3 is indispensable for Treg function (12, 28, 29). Loss-of-function mutations in FoxP3 cause fatal autoimmune pathology. In scurfy mice, such mutations lead to severe systemic autoimmunity, while in humans, FoxP3 mutations result in functional Treg deficiency and cause Immune dysregulation, Polyendocrinopathy, Enteropathy, and X-linked (IPEX) syndrome. IPEX manifests as multi-organ autoimmunity, including diabetes, thyroiditis, and allergies (e.g., eczema), and is typically fatal within the first year of life without bone marrow transplantation (30–32). Prior to FoxP3 expression in early tTreg precursors, the genomic organizer SATB1 (Special AT-rich sequence-binding protein 1) cooperates with unknown pioneer factors to bind super-enhancers governing the FoxP3 locus and other Treg signature genes. This binding establishes a permissive epigenomic landscape essential for the induction of key Treg genes (14).The DNA demethylation occurring during this reprogramming is indispensable for the long-term stability of the Treg phenotype and its suppressive function (33, 34). However, FoxP3 function and the Treg-specific epigenome operate through distinct yet complementary mechanisms (Table 1). Of note, the preservation of the TSDR within the Foxp3 CNS2 enhancer is essential for maintaining the stable lineage identity and suppressive function of Tregs. Deletion of CNS2 triggers progressive erosion of FoxP3 expression stability during Treg proliferation, ultimately resulting in cell extinction (35–38). Treg-specific epigenetic alterations not only facilitate the lineage establishment and functional persistence of Tregs, but also serve as a genomic signature to distinguish them from Tconv cells (34, 39). Although both Tregs and activated Tconv cells express CD25 and CTLA-4, but Tregs constitutively express these molecules at high levels, while they only express in Tconv cells upon TCR stimulation. This disparity is mechanistically linked to unique CpG methylation profiles in the TSDR of a panel of genes, including FoxP3, CD25, CTLA-4, and Helios (34). Hence, a hypomethylated epigenome is a vital criterion for delineating functionally competent Tregs, safeguarding their identity amid fluctuations in FoxP3 expression.

Other than their lineage-defining transcription factor FoxP3, Treg cells could be further defined by the sustained highly expressed molecules essential for their function, in which, CD25 (IL-2Rα) and CTLA-4 are particularly critical (40–42). CD25 combines with the β and γ chains to form a high-affinity interleukin-2 receptor complex, enabling Treg cells to efficiently sense and consume IL-2 (43, 44), while CTLA-4 serves as a crucial co-inhibitory receptor that empowers them to directly suppress immune activation (45–47). Germline deletion or Treg-specific loss of CTLA-4 leads to fatal autoimmunity resembling that in scurfy mice (47). The core inhibitory mechanisms involve dynamic interactions for signal perception and regulation. A central pathway entails a negative feedback loop related to IL-2 sensing: through the consumption of IL-2 via CD25, Treg cells can rapidly locate and suppress self-reactive T cells early in an immune response, and utilize this cytokine to expand their own population and enhance their suppressive activity, thereby self-reinforcing their regulatory function (48, 49). Strikingly, Treg cells are programmed for antigen-specific suppression during their thymic development. They are not only positively selected by self-antigens but also instructed to mature into functional suppressive cells within the thymus. This pre-established immunological specificity allows them to curb Tconv activation at the onset of an immune response, thereby averting autoimmune pathology (43, 44). Similarly, CTLA-4 mediates “disarming” of APCs, whereby Treg cells physically remove the co-stimulatory molecules CD80/CD86 from the APC surface via trans-endocytosis. This deprivation of co-stimulatory signals not only inhibits immune responses, but also actively dictates the fate of responding T cells. Depending on TCR affinity, high-affinity T cells undergo apoptosis, medium-affinity T cells are induced into an anergic state, and low-affinity T cells remain quiescent. Therefore, Treg cells can orchestrate the survival and outcome of self-reactive T cells, which establishes durable immunosuppressive tolerance. As a result, identifying the mechanisms underlying the reprogramming of Tconv cells into Treg-like suppressors has become a key research objective (50). Even without FoxP3, experimentally eliminating IL-2 production, enforcing constitutive high CTLA-4 expression, and providing TCR stimulation can endow Tconv cells with regulatory functions, indicating that these molecular events constitute pivotal mechanisms of Treg-mediated suppression, which is essential for the prevention of autoimmunity (51).

Tregs regulate immune responses through multiple cell-contact-dependent and -independent mechanisms. In general, Treg cells broadly utilize mechanisms like IL-2 consumption and expression of CTLA4, CD73, or CD39. The induction of alternative pathways including the production of IL-10, TGF-β, IL-35, or indoleamine 2,3-dioxygenase activity, however, is often restricted to particular microenvironments or triggered by specific extracellular cues (47–49). Importantly, many inhibitory mechanisms rely on antigen-specific interactions between Tregs and APCs. High-intensity sustained contact between Tregs and APCs can physically sequester pMHC and CD80/86 co-stimulatory molecules on APC surfaces, thereby disrupting the ability of DCs to interact with and activate cognate effector T cells (45, 46). Additionally, Tregs can secrete apoptotic mediators such as granzymes and perforin to eliminate APCs, conventional T cells, and natural killer (NK) cells (50, 52, 53). Notably, Integral to most Treg functions (54), TCR activation thereby affirms that antigen specificity is indispensable for optimal suppressive efficacy. Studies also indicate that Treg internalization of pMHC requires antigen specificity sharing with the suppressed effector T cells (46). This is particularly critical in T1D, where effector T cells targeting neoepitopes often lack corresponding antigen-specific Tregs for their control (27).

Collectively, Treg heterogeneity arises from diverse developmental origins and functional states, underpinned by distinct molecular identity markers. Defining these identity signatures is critical, not only to resolve functionally discrete Treg subpopulations but also to inform the rational design of antigen-specific Treg therapies. Core molecules like FoxP3 (stabilized by TSDR demethylation), CD25, and CTLA-4 establish Treg identity while enabling potent immunosuppression. Critically, TCR activation is indispensable for Treg-mediated suppression. However, in T1D settings, frequent deficits in cognate antigen recognition specifically impair this process, underscoring the therapeutic imperative for antigen-specific interventions.

As mentioned earlier, Tregs can be classified into three main subsets: tTregs, pTregs, and Tr1. This is evidence that all three populations are able to suppress islet autoimmunity (58). However, the absence of definitive markers to discriminate tTregs from pTregs hampers the assessment of their individual contributions to T1D pathogenesis. Therefore, delineating the identity markers of Tregs is crucial for determining the origin(s) of reactive Treg populations, which warrants the elucidation of the allelic susceptibility to Treg dysfunction and the upstream immune defects in T1D setting (27).

The progression and remission of T1D involve specific Treg subsets and their distinct transcriptomic profiles. Benoist and colleagues conducted studies to compare the transcriptomes between pancreatic and splenic Tregs in NOD mice, and identified that the pancreatic Tregs are featured by the upregulation of suppressive molecules (such as IL10 and LAG3), specific chemokine receptors (CXCR3 and CCR5), and activation-associated transcription factors (Nr4a2, Fos, and Jun) (112). Beyond this shared effector profile, a signature more specific to pancreatic Tregs is enriched in genes governing cell growth and proliferation, suggesting a higher local turnover rate (69, 113). Additionally, this unique identity is further underscored by a lack of typical tissue-resident transcription factors (112). Subsequent analysis revealed that genes associated with Treg activation signatures are significantly upregulated in the pancreatic Treg transcriptome. Furthermore, pancreatic Tregs also displayed a bias toward the BLIMP-1 and IRF4-dependent Treg transcriptomic signatures; these transcription factors are critical for Treg effector functions and their maintenance within tissues (35, 107). However, the most pronounced differences were observed in the expression profile of CXCR3⁺ Tregs. The induction of CXCR3 depends on the expression of the transcription factor T-bet in Tregs and other leukocyte subsets (9, 24, 37). CXCR3 and T-bet profoundly influence the dynamic migration of Tregs (to be discussed in detail later). Collectively, these results indicate that Treg cells located at sites of autoimmune responses possess a specific identity characterized by enhanced effector function, accelerated cell turnover, and a prominent CXCR3⁺ Treg signature program, which may bolster their capacity to control type 1 inflammatory responses within the islets.

Emerging technologies have substantially advanced the classification of Treg subsets. In humans, Mass cytometry using 26 well-established Treg-associated markers has resolved 22 distinct subsets (114), while transcriptomic analyses have corroborated their differentiation into Th-like lineages and defined subset-specific molecular signatures for both Th cells and Tregs (115). However, none of these classification schemes rely on unsupervised global gene expression profiling. Recent advances in single-cell RNA sequencing (scRNA-seq) have provided novel insights into T cell and Treg heterogeneity at single-cell resolution (116–120). Using scRNA-seq data, Treg cells were subdivided into 6 clusters in healthy human peripheral blood and 5 clusters within the human breast cancer microenvironment (116). Despite these findings, in-depth single-cell studies of human Treg cells under steady-state or disease conditions remain limited. The identity, functional/steady-state characteristics, differentiation pathways, and interrelationships of distinct Treg subsets are still incompletely defined. scRNA-seq has revealed significant dynamic shifts in immune cells at different stages of T1D, particularly during partial remission. The study identified marked fluctuations in the proportions of TIGIT⁺CCR7⁻ Tregs and CD226⁺CCR7⁻CD8⁺ T cells and validated, using machine learning based algorithms, that these two subsets serve as biomarkers for declining β-cell function (121). Treg heterogeneity in humans has been further resolved by single-cell transcriptomic analysis. In particular, eTregs can be subdivided into a FOXP3^hi subset with potent suppressive function and a highly proliferative MKI67^hi subset, developing along two distinct differentiation trajectories (Path I/II) (21).Although these findings are primarily based on peripheral blood data, they still provide crucial insights for understanding the features of pancreas trTregs.

Recent studies that integrate single-cell transcriptomic data from humans and mice have systematically elucidated the conserved features and species-specific adaptations of Treg cells throughout evolution. In terms of conservation, Treg cells in both species express core transcriptional markers such as FOXP3 and IKZF2, which allow for a clear distinction from Tconv cells. Additionally, both species possess a “furtive” Treg subset characterized by low suppressive activity (119). Furthermore, Treg cells exhibit transcriptional dynamics along a tissue-adaptation continuum and display conserved expression programs between homeostasis and disease, as well as between mice and humans (117). However, notable interspecies molecular differences exist within this conserved regulatory framework. Certain signaling molecules demonstrate paralogous gene substitution; for example, Pim1 in mice is analogous to Pim2 in humans. Moreover, the expression profiles of some tissue-homing receptors and effector molecules reveal species-specific preferences (117). These findings indicate that although Treg adaptation is conserved among species, its molecular mechanisms exhibit evolutionary plasticity. This has significant implications: the conserved features support the use of mouse models to investigate fundamental aspects of human Treg biology, whereas species-specific differences require the validation of human therapeutic targets. Future integration of cross-species single-cell data will facilitate the identification of conserved pathways and distinct human characteristics, thereby advancing precision immunotherapy.

The homing and localization of pancreatic Tregs depend on a complex chemokine-receptor network, and their dynamic migration patterns are critical for local immune suppression. Notably, pancreas trTregs often exhibit islet antigen specificity, which is closely linked to their dynamic migratory behavior. The accumulation pattern of Tregs within the pancreas, particularly in the islets, is finely regulated by adhesion molecules, chemokines, and their receptors, whose expression is influenced by both Treg-intrinsic factors and the pancreatic microenvironment. Importantly, emerging evidence highlights that endothelial cells (ECs), comprising both blood (BECs) and lymphatic (LECs) endothelial cells, are integral and active components of the tissue microenvironment, serving as essential gatekeepers of Treg trafficking and localization (122). Through bidirectional crosstalk involving molecules such as LTα1β2–LTβR, PD1–PDL1, and S1P–S1PRs, endothelial cells not only facilitate Treg transmigration but also modulate their stability, differentiation, and immunosuppressive function (122). In the context of T1D, the accumulation of CXCR3⁺ Tregs within the islets is impaired. This phenomenon has significant functional consequences, as evidenced by that CXCR3-deficient mice fail to effectively suppress Th1 activity and develop diabetes earlier (123). The deficit caused by impaired CXCR3 signaling is partly counteracted by the activity of additional chemokine receptors, notably CCR2, CCR8, and CXCR6, which are also present on pancreatic Tregs and support their accumulation within islets. Pioneering work on Treg trafficking in transplanted islets demonstrated that inflamed islets provide an essential “educational” environment for Tregs prior to their migration to the draining lymph nodes (124). Differential usage of homing mechanisms guides Tregs to distinct sites: migration to the inflamed islets implicates CCR2, CCR4, CCR5, and P-/E-selectin ligands, whereas recruitment to the draining lymph nodes employs CCR2, CCR5, and CCR7 (124). Furthermore, Treg positioning is dependent on islet-derived factors (e.g., CCL12, MAdCAM-1) and Treg-expressed integrins (α4 and β7) (125–128). Pancreas-resident APCs, particularly F4/80⁺ macrophages and CD11c⁺ dendritic cells, are a primary source of CXCL9, CXCL10, and CXCL11. These ligands mediate the specific recruitment of CXCR3-expressing Tregs into the pancreatic islets (129). This empirical evidence underscores that the functional efficacy of tissue Tregs is inextricably linked to their migratory capacity, which is enabled by a series of factors operating with substantial redundancy. This functional compensation, however, raises a central unanswered question: what is the hierarchical contribution of these various molecules to Treg migration specifically between the pancreatic parenchyma and the pancreatic lymph nodes (PLN). Similarly, it remains unclear whether these migration patterns executed by pancreatic Tregs are primarily associated with T1D or represent a general phenomenon occurring in other inflammatory responses. Further investigation into the endothelial-Treg crosstalk within the pancreas may unveil novel targets for restoring immune balance in T1D and beyond.

The expression of transcription factors also influences the migration of Treg cells in T1D, with T-bet exhibiting the most pronounced effect. However, its function is context-dependent and cannot be simply categorized as either detrimental or beneficial to Treg suppressive capacity. On the one hand, T-bet plays a guiding role in Tregs by promoting their migration to inflammatory sites (130, 131). For instance, it upregulates CXCR3 expression, facilitating Treg recruitment to Th1-inflammatory regions such as pancreatic islet, where CXCL9 and CXCL10 are enriched (131, 132). Moreover, T-bet-specific knockout in Tregs inhibits IFN-γ production and impairs their ability to suppress Th1 effector cells (102, 130). Some studies even suggest that T-bet⁺ Tregs may specialize in suppressing effector T cells that also express T-bet (102). On the other hand, multiple studies indicate that T-bet deficiency does not impair, and may even enhance, the overall suppressive function of Tregs (131). T-bet-deficient Tregs have been shown to suppress effector T cells as effectively as or more effectively than wild-type Tregs, both in vitro and in vivo (133–135). This apparent paradox, where functional maintenance or enhancement coexists with altered migratory behavior, can be explained by multiple migration and retention mechanisms regulated by T-bet. Specifically, although loss of T-bet impairs CXCR3-dependent migration to the pancreatic islets, T-bet⁻/⁻ Tregs upregulate CCR4, enabling compensatory migration and accumulation in the pancreas via CCL22-mediated signaling (124, 136). Concurrently, T-bet deficiency upregulates CD103 (αE integrin), which binds to E-cadherin and promotes anchorage of Tregs within epithelial tissues, thereby enhancing retention of T-bet⁻/⁻ Tregs in the islets (132, 137–139). Moreover, T-bet regulates S1PR1 expression, and its deficiency leads to reduced S1PR1 levels, which impairs Treg egress from pancreatic islets back into the bloodstream, thereby further promoting Treg local retention (140, 141). Collectively, in the context of T1D, T-bet does not act as a functional “on/off switch”, rather as a “navigation commander” for Tregs. Its absence does not fundamentally disrupt Treg suppressive capacity but instead reprograms their migratory behavior and tissue-localization properties. T-bet⁻/⁻ Tregs tend to be recruited via alternative pathways and retained longer within target tissues. In fact, this is consistent with the concept that T-bet⁺ Tregs differentiate from T-bet⁻ precursors (102). In conclusion, the impact of T-bet deficiency in Tregs should not be simply interpreted in binary terms of benefit or harm, rather its essential role lies in regulating the spatiotemporal distribution and migratory dynamics of Tregs. These effects may yield diverse consequences in the complex immune landscape of T1D, and understanding how T-bet fine-tunes the tissue-specific localization of Tregs is critical for evaluating its potential as a therapeutic target.

Importantly, the fundamental mechanisms governing Treg migration are not unique but are common to pathogenic effector cells. As illustrated in NOD mice, high α4β7 expression on islet-infiltrating lymphocytes indicates that MAdCAM-1 within the inflamed islet microenvironment also mediates the recruitment of pathogenic Teffs, implying a common homing pathway (128, 142). During T1D progression, a mixed population of immune cells including lymphocytes with both naïve and memory phenotypes, dynamically and continuously infiltrates into the pancreatic islets to mediate β cell destruction. It is worth noting that the active migration of Tregs into these insulitic lesions is lower compared to their conventional CD4⁺ counterparts (143). Therefore, it would be interesting to dissect when and how preferential recruitment of Tregs into the pancreatic islet occurs. A more comprehensive understanding of the detailed migration mechanisms required for Treg accumulation within the pancreas remains to be elucidated. This knowledge is of great importance for the expansion and development of therapeutic Tregs capable of responding to tissue-specific environmental cues.

Insulin is the most abundant protein in pancreatic β cells. Insulin autoantibodies (IAAs) are typically the first to appear during the pathogenesis of T1D, highlighting the pivotal role of insulin in initiating T1D autoimmunity (144). Therefore, inducing insulin-specific Tregs holds promise for effectively controlling autoimmunity. T1D vaccine therapy, also known as autoantigen-specific immunotherapy, aims to restore immune tolerance and induce antigen-specific Tregs by targeted delivery of autoantigens (e.g., insulin, GAD65). Its core mechanism involves the processing of subimmunogenic doses of antigens (e.g., insulin mimotope peptides) by APCs, which subsequently activate islet antigen-specific CD4⁺ T cells to differentiate into FoxP3⁺ Tregs, rather than Teffs. These Tregs exhibit high expression of CTLA-4, IL-2Rα, and TIGIT, and they precisely regulate autoimmune responses by suppressing Teff activity and promoting the expression of immune tolerance-associated genes (e.g., FoxP3, IL-10) (145). For instance, in an HLA-DQ8 transgenic humanized mouse model, an insulin vaccine successfully induced stable and functional Treg clones that effectively suppressed β-cell destruction (145). The key advantages of this approach lie in its antigen specificity, safety profile, and preventive potential. Vaccines can precisely target pathogenic autoimmune pathways, circumventing broad immunosuppression, and thereby minimizing disruption to normal immune function (145). Furthermore, the use of subimmunogenic vaccine doses avoids triggering Teffs activation, significantly reducing the risk of adverse effects (145). Vaccine therapy is also applicable to high-risk individuals, such as autoantibody-positive relatives, and has the potential to delay or even prevent the onset of certain diseases like T1D (146). However, vaccine therapy may manifest several limitations. First, its efficacy has been modest. For example, oral insulin failed to effectively preserve residual β-cell function in patients with new-onset T1D and did not markedly delay the progression of disease in at-risk populations with pre-diabetes (146, 147). Second, vaccines often suffer from low bioavailability, and oral administration is particularly susceptible to gastrointestinal degradation, leading to insufficient antigen presentation efficiency (147). Additionally, patient heterogeneity poses a significant challenge, as vaccine efficacy may be restricted to specific subgroups (e.g., those with high IAA levels), necessitating precise stratification for optimal therapeutic outcomes (146). Therefore, the clinical application of antigen-specific Treg therapy continues to face obstacles in the context of autoimmune diseases. Moreover, the ability to directly identify and characterize human insulin-specific FoxP3⁺ Tregs in vitro is crucial for assessing responses to insulin-specific vaccination.

To overcome vaccine limitations, IL-2 therapy exploits Treg high-affinity IL-2R dependency to expand and enhance their function. Tregs exhibit heightened sensitivity to fluctuations in IL-2 concentration due to their expression of the high-affinity IL-2R (44, 148). Administration of IL-2 has been demonstrated as an effective strategy to induce in vivo Treg expansion and prevent autoimmunity in numerous mouse models, such as in NOD mice (149). However, IL-2 exhibits dual anti-inflammatory and pro-inflammatory effects, as it can also non-specifically expand pro-inflammatory immune subsets. This differential cellular response depends on the hierarchical expression levels of IL-2R across distinct lymphocyte subsets (150). Under normal physiological conditions, IL-2 produced by activated Teffs expands and sustains the Treg population, while Tregs reciprocally maintain immune homeostasis by feedback-inhibiting Teff responses. Disruption of this reciprocal interaction leads to dysregulation of the Treg/Teff ratio and promotes autoimmune responses in NOD mice. Notably, within the context of T1D, Tregs residing in the pancreas exhibit greater sensitivity to IL-2 compared to Tregs in the PLNs and other sites (69). IL-2 treatment increases the proportion of Tregs in the pancreas of prediabetic mice; however, in mice with new-onset diabetes, where pancreatic Treg proportions are already markedly elevated, IL-2 therapy fails to further augment their numbers (69). Studies have excluded proliferation as the primary mechanism underlying IL-2-mediated increases in Treg numbers. Furthermore, IL-2 signaling promotes Treg survival (69), recruitment, and synergizes with TGF-β to induce the conversion of CD4⁺ T cells into Tregs (151). Beyond increasing pancreatic Treg numbers, IL-2 also directly enhances the activity of pancreatic Tregs by upregulating the expression of molecules critical for Treg function including CD25, FoxP3, CTLA-4, ICOS, and GITR (152), coupled with immunosuppression within the islets during T1D progression. It is therefore critical to elucidate the mechanisms through which IL-2 can selectively engage Tregs, particularly those localized in the pancreatic islets. To this end, Mark and colleagues engineered a β-cell-targeted IL-2 delivery system using an adeno-associated viral (AAV) vector to direct localize the expression of IL-2 within the islets of NOD mice (153). Consistent with systemic IL-2 therapy, β-cell-specific IL-2 delivery persistently suppressed β-cell autoimmunity and provided long-term protection against diabetes onset in both preclinical and advanced disease stages. However, it showed no therapeutic efficacy in NOD mice with new-onset diabetes (153). Furthermore, this approach failed to induce sustained proliferation of islet-resident Tregs (153). Functionally, pancreatic Tregs converted to a CD62L^highFoxp3⁺ Treg phenotype, exhibiting enhanced fitness and suppressive capacity. Notably, this effect was not associated with increased levels of IL-10 or TGF-β (153). Although β-cell-specific IL-2 demonstrated efficacy comparable to systemic IL-2 in preventing T1D, these data suggest that a β-cell-targeted strategy may exhibit advantage for effectively sustaining the functional capacity of intra-islet Tregs within the context of T1D.

Of note, several alternative strategies have been developed to achieve more specific targeting of Tregs. For example, the use of specific anti-IL-2 monoclonal antibodies (mAbs) to block the IL-2 binding site on the β-chain (CD122) enables the formation of IL-2/antibody complexes that preferentially target Tregs. This selectivity stems from the fact that Tregs, unlike most other immune cells, primarily rely on CD25 for IL-2 binding. These engineered IL-2 complexes exhibit the ability to preferentially expand both murine and human Tregs in vitro, representing a promising approach for future targeted immunomodulatory therapies (154). Another strategy involves engineered IL-2 muteins, which are designed with reduced affinity for CD122 to increase CD25 dependency (150). However, even with enhanced selectivity, CD25 can be expressed on a range of immune cells beyond Tregs, potentially leading to off-target effects and associated adverse events (155).

Given the limitations of endogenous immunomodulatory therapies, such as IL-2 therapy, the adoptive transfer of exogenous Tregs offers a promising new approach for treating T1D. This innovative immunomodulatory strategy involves isolating immunosuppressive CD4⁺CD127^lowCD25⁺ Tregs from the patient’s own peripheral blood, followed by large-scale ex vivo expansion and subsequent reinfusion into the patient (156). Its primary objectives are to restore immune tolerance and suppress the autoimmune attack by Teffs on pancreatic β cells, thereby delaying disease progression. Significant progress has been made in this therapy. Technically, a GMP-compliant expansion protocol utilizing artificial APCs, such as the KT64/86 system, has been successfully established. This protocol achieves over 3000-fold expansion of Tregs while maintaining high FoxP3⁺ expression and robust suppressive function (157, 158). Clinically, despite concerns regarding Treg lineage instability raised previously, Phase I trials confirmed that a majority of infused CD25⁺CD127^low Tregs expanded ex vivo from autologous polyclonal sources, maintain phenotypic fidelity in T1D patients. These expanded Tregs demonstrate favorable safety (no serious adverse events) and durable persistence, with ~25% of peak cell levels detectable at 1-year post-infusion. Critically, sustained C-peptide preservation beyond two years in 38% of recipients provides compelling rationale for advancing to Phase II efficacy trials (156).

For Treg cell isolation, cells enriched by fluorescence-activated cell sorting (FACS) in Fr. I represent a more suitable starting population for expansion. This preference stems from the ability to avoid potential contamination by non-Treg cells present in Fr. III (159), although the Fr. I population itself is not entirely homogeneous. Current mainstream methods commonly use extensive polyclonal or antigen-driven expansion to propagate Tregs in an undifferentiated state. A potential alternative strategy involves purifying specific functional Treg subsets, such as follicular regulatory T cells or CXCR3⁺ Th1-like Treg cells, to expand them ex vivo. This approach aims to more precisely modulate the autoimmune responses mediated by Tfh or Th1 cells, respectively. However, the efficacy of this targeted strategy remains to be fully validated. Furthermore, it is still unclear whether its benefits outweigh those of maintaining Tregs in an undifferentiated state, an approach that preserves their functional plasticity and enables them to respond to a diverse array of signals following adoptive transfer in vivo.

Although the safety of autologous Treg adoptive transfer has been established, its therapeutic efficacy remains constrained by factors such as antigen non-specificity, low in vivo persistence, and insufficient cell numbers. While the overall efficacy of polyclonal Treg therapy has been modest, the observation of long-lived Tregs underscores the need to develop novel strategies to enhance Treg functionality, survival, and proliferative capacity within the host (156, 160). In recent years, genetically engineered Tregs (EngTregs) has emerged as an innovative solution for precision therapy in T1D. Compared to biologics and small-molecule drugs, living cells, particularly immune cells, can sense combinatorial environmental signals and orchestrate sophisticated, controllable therapeutic responses accordingly. In other words, molecular drugs function akin to single-task tools, whereas cells resemble programmable devices capable of deploying context-appropriate combinations of tools based on the specific situation. Consequently, engineering immune cells, such as EngTregs, holds promise for achieving smarter and more dynamic immunomodulation.

Prospective clinical studies serve as the core driving force behind the clinical translation of Treg therapy. As of July 2019, 51 clinical trials involving Treg cells were registered on ClinicalTrials.gov. These trials were summarized by Ferreira et al (216). With continued advancements in the field, the number of Treg cell clinical trials focused specifically on T1D had reached 14 by November 2025. Among these, six studies have been completed (NCT01210664, NCT02772679, NCT01827735, NCT03444064, NCT02691247, NCT02265809), four are actively recruiting participants (NCT06688331, NCT02932826, NCT06708780, NCT06427421), one has not yet started recruitment (NCT05973734), one has been withdrawn (NCT03236558), and two have an unknown status (NCT00173641, NCT03011021).

The completed trials primarily evaluated the adoptive transfer of polyclonal Tregs (e.g., NCT01210664, NCT03444064) and combination therapies with low-dose IL-2 (e.g., NCT02265809, NCT01827735). While the limitations of these approaches have been detailed previously, they collectively confirm the foundational safety of Treg therapy, thereby establishing a crucial groundwork for further investigation. Current research frontiers are focused on overcoming the key obstacles in translating Treg therapy from basic research to clinical application. First, patient heterogeneity and cell sourcing pose significant challenges to treatment accessibility. Patients with autoimmune diseases often receive long-term immunosuppressive treatments, which can compromise the number and function of their endogenous Treg cells. This makes it difficult to isolate and expand sufficient Tregs that meet therapeutic standards, representing a typical barrier in Treg transplantation trials (217). To address this, researchers are actively exploring alternative cell sources. For instance, umbilical cord blood (UCB)-derived Tregs have emerged as a promising option due to their more naive phenotype and greater proliferative potential (218). Notably, CD4 and CD25 markers are sufficient for their effective isolation from UCB (219). A clinical study (NCT02932826) evaluating the efficacy of ex vivo expanded UCB-Tregs for autoimmune diabetes is currently recruiting patients. Additionally, Tregs isolated from discarded thymic tissue obtained during pediatric cardiac surgery show considerable potential, with yields from a single donor comparable to those from adult peripheral blood (220). However, the clinical feasibility of this source requires further preclinical and clinical validation (220), and a study investigating thymus-derived Tregs in T1D has been withdrawn (NCT03236558), indicating remaining uncertainties for this pathway.

Second, refining therapy design toward greater precision is a key direction for improving efficacy. The field has placed strong emphasis on developing antigen-specific therapies. A clinical trial (NCT06708780) aiming to select antigen-specific Treg TCRs is currently underway. In the area of islet transplantation, another study (NCT06427421) is comparing the efficacy of recipient-derived Tregs versus donor bone marrow-derived immune cells in combination with islet transplantation to optimize immunomodulatory strategies. However, traditional pancreas or islet transplantation remains limited by donor scarcity and variable outcomes. In this context, new hope has emerged for numerous patients with long-standing, severe T1D in the form of Zimislecel (VX-880), Vertex Pharmaceuticals’ stem cell-derived, fully differentiated islet cell therapy, which offers the prospect of freedom from insulin dependence (221). This therapy also holds potential for future synergistic or complementary approaches with Treg therapy.

Of critical importance, the clinical translation of EngTreg therapies faces severe systemic challenges. Although engineered Treg therapies like ABA-201 are planned to initiate in 2025, as previously mentioned, no gene-engineered Treg therapy for T1D is currently registered on ClinicalTrials.gov, reflecting the significant gap between proof-of-concept and clinical implementation. In terms of safety, TCR-mediated off-target and cross-reactivity risks constitute a fundamental safety concern. Engineering TCRs is inherently complex, partly due to their native heterodimeric structure—introduced exogenous TCR subunits can mispair with endogenous TCRα or TCRβ chains, forming mixed TCRs with unknown specificity, thereby significantly increasing the risk of unpredictable off-target effects (216). Furthermore, even successfully designed high-affinity TCRs may cross-react with structurally similar self or foreign peptides, potentially leading to unintended autoimmune reactions or altered efficacy (222). While genome editing technologies offer potential solutions for precise TCR expression control, this path itself raises profound ethical considerations.

In terms of manufacturing scalability, the production process for EngTreg therapies is inherently complex, multi-step, lengthy, and costly. Achieving large-scale expansion compliant with GMP standards is a key constraint to broad accessibility. Notably, advanced manufacturing platforms with extensive experience in the CAR-T cell therapy field have widely adopted various systems to enhance process control and scale-up capability, ranging from semi-automated equipment (e.g., Miltenyi Biotec’s CliniMACS Prodigy, Cytiva’s Xuri system, ScaleReady’s G-REX) to fully automated integrated platforms (e.g., Lonza’s Cocoon Platform). These systems, utilizing closed operations and process standardization, help improve product consistency and reduce contamination risk, providing an important reference for future Treg scale-up production (223–225).

Regarding quality control and product release, although regulatory agencies have not stipulated a unified minimum cell viability threshold for cell therapy products, CAR-T cell clinical practice has established an important reference: clinical trials typically require a minimum cell viability of 70%, while commercial product release requires at least 80% viable cells (223). This standard provides a practical basis for Treg product quality specifications. Establishing comprehensive quality control and release standards throughout the entire process—from rigorous testing and qualification of starting materials (226) (e.g., patient cells) to precise control of critical process parameters during manufacturing—is fundamental to ensuring product batch-to-batch consistency, purity, and potency (223, 227). On the regulatory front, global regulatory agencies are actively working toward establishing harmonized frameworks for these advanced therapy products (227, 228). As living, genetically modified products, their long-term safety (e.g., tumorigenicity, insertional mutagenesis risk) and durability of efficacy are central to regulatory scrutiny. Some regulatory agencies (e.g., Japan’s PMDA) have introduced breakthrough review pathways like “Sakigake” to accelerate the approval of innovative therapies, but such pathways are also accompanied by stringent requirements for post-marketing studies to verify clinical benefits (229, 230).

In summary, although Treg therapy has achieved preliminary safety validation in the context of T1D, the comprehensive clinical translation of engineered Treg therapies still remains a considerable journey. Ensuring TCR specificity and safety, overcoming challenges in manufacturing, safety verification, and individualized application, and appropriately addressing associated ethical and societal issues are core problems that must be resolved to advance these next-generation treatments.