Harnessing the biology of regulatory T cells to treat disease

Abstract

Regulatory T (T_reg) cells are a suppressive subset of CD4⁺ T cells that maintain immune homeostasis and restrain inflammation. Three decades after their discovery, the promise of strategies to harness T_reg cells for therapy has never been stronger. Multiple clinical trials seeking to enhance endogenous T_reg cells or deliver them as a cell-based therapy have been performed and hint at signs of success, as well as to important limitations and unanswered questions. Strategies to deplete T_reg cells in cancer are also in active clinical testing. Furthermore, multi-dimensional methods to interrogate the biology of T_reg cells are leading to a refined understanding of T_reg cell biology and new approaches to harness tissue-specific functions for therapy. A new generation of T_reg cell clinical trials is now being fuelled by advances in nanomedicine and synthetic biology, seeking more precise ways to tailor T_reg cell function. This Review will discuss recent advances in our understanding of human T_reg cell biology, with a focus on mechanisms of action and strategies to assess outcomes of T_reg cell-targeted therapies. It highlights results from recent clinical trials aiming to enhance or inhibit T_reg cell activity in a variety of diseases, including allergy, transplantation, autoimmunity and cancer, and discusses ongoing strategies to refine these approaches.

Introduction

Regulatory T (T_reg) cells constitute ~2–5% of the adult peripheral blood CD4⁺ T cell repertoire¹. Whereas the majority of immune system cells function to promote inflammation to fight pathogens and cancers, T_reg cells keep immunity in check to maintain homeostasis and prevent pathology. Dysregulation of T_reg cell function is linked with many diseases and the failure of these cells to control antigen-specific effector T cells can lead to autoimmune diseases, allergies and various other hyperinflammatory diseases. When a patient receives a transplant, T_reg cells prevent alloreactive effector T cells from attacking the foreign tissue. In contrast, tumours enhance the T_reg cell response so that anticancer immunity is suppressed. Therapies focused on T_reg cells thus aim to either enhance their function (in the case of autoimmunity, allergy, transplantation and other hyperinflammatory diseases) or hinder it (in patients with cancer). Establishing an appropriate balance between T_reg cells and effector T cells is thus critical to human health and underlies the broad clinical potential for T_reg cell-based therapies.

T_reg cells are characterized by constitutively high expression of FOXP3 and CD25 (the IL-2 receptor α-chain; IL-2Rα). As their lineage-defining transcription factor, FOXP3 is essential for T_reg cell development. The absence of functional FOXP3 from birth results in deficient T_reg cell function and a disease characterized by widespread autoimmunity and allergy, known as immune dysregulation, polyendocrinopathy, enteropathy and X-linked (IPEX) syndrome^2,3. FOXP3 has both transcriptional activation and repression functions; for example, it can simultaneously promote CD25 and suppress IL2 transcription, resulting in the expression of CD25 (ref. ⁴) and reliance on environmental IL-2 (refs. ^5,6,7,8), which are both characteristic of T_reg cells.

Like all other T cells, T_reg cells express a T cell receptor (TCR), which allows them to detect antigens presented by major histocompatibility complex II (MHC II) on antigen-presenting cells (APCs; commonly dendritic cells or macrophages). Upon recognition of their cognate antigen–MHC on an APC, T_reg cell immune modulation can be broadly categorized into three main phases (Fig. 1). First, interactions with APCs reduce their capacity to activate effector T cells. Second, activation-stimulated release of cytokines and metabolites by T_reg cells diminishes the pro-inflammatory activity of surrounding immune cells. Third, this re-shaped environment favours the expansion and de novo development of new T_reg cell populations, thereby promoting tolerance. T_reg cell-focused therapies attempt to target one or more of these modes of action to either enhance cell function (in the case of autoimmunity, allergy or transplantation) or to suppress it (in the case of cancer). An additional therapeutic goal is to change the abundance of T_reg cells to re-establish their healthy balance with effector T cells.

Fig. 1: Three phases of T_reg cell-mediated immune suppression.

a, In phase 1, regulatory T (T_reg) cells suppress the ability of antigen-presenting cells (APCs) to present antigen to and co-stimulate effector T (T_eff) cells. The T cell receptor (TCR) of a T_reg cell forms a tight immune synapse with the antigen–major histocompatibility complex (MHC) complex displayed on the APC, physically blocking effector T cell access to the same antigen. Additionally, T_reg cell CTLA4 preferentially binds to CD80 and CD86 on the APC, blocking effector T cell access to co-stimulatory signalling through the lower affinity receptor, CD28. Upon retreat from the APC, the T_reg cell removes co-stimulatory proteins and antigen–MHC complexes, thereby further abrogating effector T cell activation. These T_reg cell–APC interactions are thought to occur predominantly in secondary lymphoid organs (such as lymph nodes) but might also occur in tissues. b, In phase 2, T_reg cells release a variety of regulatory cytokines into their microenvironment to suppress inflammation. Most notable are IL-10 and TGFβ, which suppress APCs and promote FOXP3 expression, respectively. TGFβ is cleaved from its latent form into its active form by various mechanisms, notably integrin αvβ8 on T_reg cells. T_reg cells express high levels of CD25, which has a higher affinity for IL-2 than the dimeric IL-2 receptor expressed by conventional T cells. Therefore, T_reg cells behave as an ‘IL-2 sink’, restricting the amount available for conventional T cells. T_reg cells also express the ectonucleotidases CD39 and CD73, which work together to convert ATP to adenosine, which has anti-inflammatory properties. c, In phase 3, T_reg cells mediate ‘infectious tolerance’ by inducing the expansion of existing T_reg cells and/or by converting effector T cells into T_reg cells, establishing a long-lasting tolerogenic balance.

In this Review, we first provide a brief overview of the development and function of T_reg cells, then outline the challenges associated with their use as a biomarker of immune function and discuss therapeutic approaches to boost endogenous T_reg cells. We next discuss T_reg cellular therapy as an exciting prospect for the treatment of autoimmunity and to induce transplant tolerance, and describe how the detrimental effects of T_reg cells can be targeted to fight cancer. We highlight new, more precise T_reg cell-targeting approaches that are beginning clinical testing. Although we focus on FOXP3⁺ T_reg cells, we acknowledge that other types of suppressive T cells, such as IL-10-producing type 1 T_reg cells⁹ and CD8⁺ T_reg cells¹⁰, also have important roles in immune homeostasis. We review knowledge derived from the study of human T_reg cells but refer to selected studies in model systems using mice or non-human primates.

The biology of human T_reg cells

T_reg cell development in the thymus and periphery

T_reg cells arise via two major developmental pathways: selection in the thymus to generate thymic T_reg (tT_reg) cells (sometimes also referred to as natural T_reg cells) or differentiation in the periphery from conventional T cells to generate peripheral T_reg (pT_reg) cells. In the thymus, T_reg cell differentiation is driven by the pattern and strength of TCR and cytokine signalling^11,12, with a general consensus that T_reg cells arise from cells bearing ‘intermediate’ affinity TCRs with hydrophobic antigen–MHC-binding regions that bind self-antigens¹³. These cells with intermediate affinity TCRs bifurcate into two populations: cells receiving continuous antigen signalling develop into autoreactive, IL-2-producing effector T cells, whereas less continuous TCR signalling, for example, interrupted by TGFβ-mediated inhibition, results in cells expressing FOXP3 (ref. ¹⁴). The development of these FOXP3⁺ T_reg cells is then locked in by IL-2 produced by their surrounding autoreactive counterparts^14,15. Studies in mice suggest that most T_reg cells bear TCRs that recognize common tissue antigens, rather than highly restricted tissue-specific antigens, resulting in tissue-agnostic migration patterns of the cells¹⁶. Although FOXP3 is necessary for tT_reg development, ongoing expression in fully committed T_reg cells might not be essential because knockout of FOXP3 does not significantly disrupt their transcriptome or methylome¹⁷. Accordingly, transcription factors other than FOXP3 (for example, TCF1 and SATB1) also have major roles in tT_reg cell development, and collectively multiple proteins contribute to lineage-specific epigenetic organization^18,19,20.

The development of pT_reg cells has been best characterized in intestinal tissues, where cells differentiate to induce tolerance to dietary antigens and commensal microbiota. Recently, three studies reported the role of a new type of intestinal APC^21,22,23, termed RORγt⁺ APCs, which share features of both dendritic and epithelial cells. These APCs express integrins αvβ8 and αvβ3, which convert latent TGFβ to its active form. TGFβ then converts conventional T cells into FOXP3⁺ T_reg cells, which proceed to restrain gut-resident T cells. In humans, RORγt⁺ APCs are enriched in mesenteric lymph nodes²³ and might be dysfunctional in inflammatory bowel disease²¹. Interestingly, although FOXP3 expression in pT_reg cells is essential for suppression of a transcriptional programme associated with T_H17 cells, commitment to the pT_reg lineage is reported to be FOXP3 independent²⁴. Nevertheless, consistent evidence that people with genetic mutations in FOXP3 suffer from colitis²⁵ indicates that this transcription factor has a non-redundant role in maintaining intestinal tolerance.

Whether or not tT_reg and pT_reg cells are distinguishable in humans has been debated for many years. In mice, surface expression of neuropilin 1 (NRP1) is often used to define tT_reg cells as it is not expressed by pT_reg cells²⁶ but, in humans, this receptor does not effectively identify T_reg cellular origin²⁷. Alternatively, expression of the transcription factor HELIOS (encoded by the IKZF2 gene) has been proposed as a characteristic of tT_reg cells and, in humans, blood-derived T cells co-expressing high levels of FOXP3 and HELIOS are enriched with stable, highly suppressive T_reg cells²⁸.

T_reg lineage stability

The stability of the T_reg cell lineage and how the biology of thymic versus peripherally derived cells differs are key questions in this field. Understanding T_reg cell stability is fundamental for addressing the risk of their conversion into pathogenic effector T cells as well as the longevity of therapeutic effects. The best marker of T_reg cell stability is the level of CpG DNA methylation at the T_reg cell-specific demethylated region (TSDR), a non-coding DNA sequence element located in the first intron of FOXP3 (ref. ²⁹). This locus is stably demethylated in T_reg cells but not in activated conventional T cells²⁹. Of note, since FOXP3 is on the X chromosome, in females one allele is methylated due to X-inactivation, making it important to account for sex during the analysis. In the context of inflammation, some studies in mice show stable FOXP3 expression and lineage stability³⁰, whereas others report loss of FOXP3 and gain of an effector T cell phenotype^31,32. These contrasting results likely arise from differences in the lineage-tracing systems used and because FOXP3 is transiently expressed in activated effector T cells that could be present in putative T_reg cell populations³³.

Similar concerns have also been raised about the stability of human T_reg cells, particularly in the context of T_reg cell therapy products³⁴ (discussed below). Patients with IPEX have autoreactive, TSDR-demethylated effector T cells that resemble destabilized T_reg cells³⁵. These destabilized T_reg cells do not emerge in the presence of wild-type FOXP3⁺ T_reg cells, showing that FOXP3 is important for T_reg cell stability and that functional T_reg cells dominantly assert tolerance in a heterogeneous population of unstable progenitors³⁵. A consideration is that it is difficult to attribute changes in phenotype or function to true lineage instability versus effects driven by heterogeneous cell populations that include conventional T cells. Nevertheless, numerous studies in humans found that T_reg cells resist lineage instability and maintain suppressive function in the presence of inflammatory cytokines, including IL-6, TNF or IL-12 (refs. ^36,37). In addition, multiple preclinical models of adoptive T_reg cell therapy have not revealed significant loss of tolerogenic properties^38,39,40.

HELIOS expression is also associated with lineage stability, potentially via its ability to suppress IL-2, IFNγ and TNF production, as demonstrated in patients with biallelic or dominant negative IKZF2 mutations^41,42. Indeed, ex vivo culture of human FOXP3⁺ HELIOS⁺ T_reg cells in pro-inflammatory cytokines does not destabilize their phenotype and rather enhances their proliferation^36,37. However, knockout of IKZF2 in human T_reg cells does not change cell phenotype or function, suggesting that HELIOS might be a T_reg stability marker but its ongoing expression is not required for fully functional cells²⁸.

Overall, the conflicting reports regarding T_reg cell stability are likely attributed to different T_reg cell enrichment and tracking protocols used between studies as well as to differences between mice and humans. To date, there are no reports of a T_reg cell product losing its suppressive capacity in patients, suggesting that current clinical T_reg cell isolation protocols derive functionally stable products.

Mechanisms of action

Multiple mechanisms are associated with the suppression and/or promotion of tissue repair by T_reg cells (Box 1). Gene knockout studies in mice support mechanistic roles for IL-2 consumption⁴³ and the expression of CTLA4 (ref. ⁴⁴), IL-10 (ref. ⁴⁵) and/or TGFβ⁴⁶. Below, we summarize some of the best-characterized mechanisms of action grouped into three phases (Fig. 1). More comprehensive summaries of additional mechanisms are found in recent reviews^47,48. Precisely which mechanisms control immune homeostasis in different tissues and disease contexts, especially in humans, remains undefined.

Box 1 T_reg adoptive cell therapy in non-traditional diseases

Regulatory T (T_reg) cells control immune tolerance but also generally suppress inflammation and contribute to tissue regeneration through cytokines such as amphiregulin, a low-affinity epidermal growth factor receptor ligand^105,262. Leveraging the ability of T_reg cells to reduce inflammation while promoting tissue repair offers a new approach to managing a range of inflammatory diseases.

Human studies

• During the COVID-19 pandemic, polyclonal allogeneic T_reg cell therapy was shown to be safe and potentially effective at reducing mortality in patients with acute respiratory distress syndrome¹⁴³.

• In a mouse model of Alzheimer disease, β-amyloid-specific T_reg cells accumulated in the brain, reducing β-amyloid depositions and microglial-driven inflammation, while enhancing cognitive function²⁶³. This T_reg cell therapy is now in a phase I clinical trial (NCT05016427).

• T_reg cell frequency in the blood inversely correlated with amyotrophic lateral sclerosis (ALS) progression in humans²⁶⁴ and now multiple clinical trials are under way for the treatment of ALS, including the REGALS trial (NCT05695521; Cellenkos) using cryopreserved, allogenic cord blood-derived T_reg cells with neurotropic homing markers.

Mouse studies

• After myocardial infarction in mice, T_reg cells reduce inflammation and release factors that promote cardiomyocyte proliferation, thereby reducing pathological burden and promoting repair²⁶⁵. Similarly, T_reg cells promote neovascularization following ischaemic injury in models of obesity and type 2 diabetes mellitus²⁶⁶.

• T_reg cells reside in the mouse brain, where they reduce neuroinflammation and promote tissue repair in models of ischaemic stroke, traumatic brain injury and multiple sclerosis, reducing disease severity and accelerating healing^267,268. Similarly, co-transplanting T_reg cells with midbrain dopamine neurons promotes neuronal survival and function in a preclinical model of Parkinson disease²⁶⁹.

• Attempts to expand endogenous T_reg cells in mice have been able to delay the onset of hindlimb paralysis in a preclinical model of ALS²⁶⁴.

Phase 1: APC modulation

T_reg cells inhibit activation of effector T cells by physically blocking their access to cognate antigens and depleting co-stimulatory proteins from APCs, which primarily reside in secondary lymphoid organs, including lymph nodes (Fig. 1a). Live microscopy reveals that T_reg cells form tight bonds with APCs that are presenting their cognate antigen, thereby blocking effector T cells with the same antigen specificity from accessing the APC and preventing T cell activation⁴⁹. Furthermore, when T_reg cells pull away from APCs they remove the antigen–MHC complex from the APC surface, further preventing effector T cell activation⁴⁹. Beyond these effects on antigen presentation, T_reg cells limit the ability of APCs to provide co-stimulatory signals (via CD80 and CD86) to effector T cells by expressing high levels of the coinhibitory receptor CTLA4, which outcompetes the effector T cell co-stimulatory receptor CD28 for binding to CD80 and CD86 and depletes them from the APC surface^50,51. As a result, T_reg cells from people with genetic mutations in CTLA4 have decreased suppressive capacity, contributing to their autoimmune phenotypes⁵². Moreover, combined blockade of CD28 and CTLA4 with the CTLA4–Ig fusion protein inhibits the therapeutic effect of T_reg cells in a humanized mouse model of skin transplantation⁵³.

Phase 2: microenvironment regulation

T_reg cells produce multiple anti-inflammatory cytokines and metabolites, most notably TGFβ and IL-10 (Fig. 1b). TGFβ is produced by many T cells but often remains in its latent form, tethered to the cell membrane. However, T_reg cells express αvβ8 integrin, which can release TGFβ from its inhibitory complex into its active form⁵⁴. Activated autocrine TGFβ then maintains and promotes FOXP3 expression^46,55. T_reg cells also produce IL-10, which suppresses production of innate, pro-inflammatory cytokines (such as TNF, IL-6 and IL-1β) and APC function^45,56. Furthermore, T_reg cells produce IL-35 (Ebi3–p35 heterodimer), which diminishes inflammation and promotes T_reg cell function⁵⁷. An additional anti-inflammatory mechanism involves expression of the ectonucleotidases CD39 and CD73, which work together to convert extracellular ATP into immunosuppressive adenosine^58,59, contributing to T_reg cell suppression.

Not only do T_reg cells produce anti-inflammatory mediators, they also behave as an ‘IL-2 sink’. In T_reg cells, the IL-2Rα subunit (CD25) complexed with IL-2Rβ and IL-2Rγ subunits form a trimeric receptor with 100-fold higher affinity for IL-2 than the dimeric IL-2Rβ–IL-2Rγ form that is typically expressed by effector T cells⁶⁰. Therefore, T_reg cells deplete IL-2 from their environment and deprive conventional T cells of this essential cytokine. Experiments in mice with a T_reg cell-specific deletion of CD25 revealed that T_reg cell-mediated IL-2 deprivation was primarily required for suppression of CD8⁺ T cells⁴³, whereas the effect on CD4⁺ T cells was nuanced. Specifically, only CD4⁺ T cells with weak TCR signalling and low IL-2 production are suppressed via this mechanism⁶¹. A therapeutic consideration is that combining T_reg cell adoptive transfer with low-dose IL-2 therapy might be deleterious⁶² because provision of the exogenous cytokine could override the IL-2 sink effect.

Phase 3: infectious tolerance

‘Infectious tolerance’ refers to the establishment of independent, long-lasting tolerance in response to a treatment⁶³ (Fig. 1c). The concept is that enhanced T_reg cell function leads to a permanent change in the balance of T_reg versus effector T cells, favouring tolerance induction. An additional angle is that enhancing T_reg cells with one antigen specificity can induce infectious tolerance to a distinct antigen via a process known as ‘linked suppression’, which is mediated by local APC modulation. The notion of infectious tolerance is crucial for T_reg cell therapy because it raises the possibility that a transient treatment could lead to long-lasting tolerogenic effects. Although the mechanisms by which tT_reg and/or pT_reg cells mediate infectious tolerance remain to be determined, the process is likely TGFβ dependent^64,65 and enhanced in the absence of CD28 signalling⁶⁶. In humans, it is difficult to study infectious tolerance in vivo, and therefore evidence is from studies of autoimmunity and transplantation in mice^67,68,69. With more sophisticated cell tracing and gene-editing tools emerging, more research to fully understand this process is warranted.

Assessing human T_reg cells in clinical studies

Identifying T_reg cells

Perhaps one of the biggest challenges to the study of human T_reg cells is the difficulty in tracking their numbers and function in health versus disease, or in response to therapy, due to the lack of a single definitive phenotypic marker (Fig. 2). Human T_reg cells are commonly identified as CD25^high, CD127^– and FOXP3⁺ (ref. ⁷⁰); however, these proteins are technically difficult to measure via flow cytometry as they have gradients of expression rather than clear positive and negative populations (Fig. 2a). Moreover, this pattern of expression is also characteristic of activated effector T cells. Thus, multiple phenotypic markers need to be assessed to increase confidence in accurate T_reg cell identification. Simple approaches include co-staining for HELIOS and/or CD45RA⁷⁰. CD45RA is particularly useful in separating resting (CD45RA⁺FOXP3^low) or activated (CD45RA^–FOXP3^high) T_reg cells from conventional T cells expressing low levels of FOXP3 (CD45RA^–FOXP3^low)⁷¹. More complex approaches, such as ascertaining the absence of IL-2 production³³ and/or using gene signatures, can further help identify T_reg cells. For example, nanoString or RNA sequencing has been used to identify and track changes in the transcriptomic signature of T_reg cells in people with type 1 diabetes mellitus (T1DM)⁷² or heart transplant recipients⁷³ (Fig. 2b).

Fig. 2: Tracking T_reg cells in clinical studies.

a, Human regulatory T (T_reg) cells can be identified based on their surface expression of CD4⁺CD25^highCD127^low, intracellular expression of FOXP3 and HELIOS, and lack of IL-2 production (inhibited by FOXP3 and HELIOS). b, Stable T_reg cells have demethylated DNA at the CNS2 region of FOXP3 (the T_reg cell-specific demethylation region (TSDR)) and a characteristic gene signature comprised of a combination of high and low gene expression compared to effector T (T_eff) cells. c, T_reg cell function can be evaluated ex vivo by measuring reduced expression of co-stimulatory molecules on co-cultured antigen-presenting cells (APCs) and/or the ability to reduce T_eff cell proliferation and IL-2 production. d, Antigen-specific T_reg cells can be tracked with fluorescently labelled multimers or antigen-induced marker assays to quantify antigen-stimulated expression of activation proteins, including CD134 and CD137.

Measuring T_reg cell antigen specificity

T_reg cell therapy is thought to be most effective if it enriches the function of disease-relevant, antigen-specific cells. While TCR sequencing provides information on clonality changes, the use of multimers and/or activation-induced marker (AIM) assays can quantify antigen-specific T_reg cells (Fig. 2d). Multimers are complexes of antigen–MHC conjugates that bind to antigen-specific T_reg cells so they can be detected by flow cytometry. Multimers have been used to track T_reg cells in T1DM, revealing that people with protective MHC haplotypes have an increased frequency of islet-specific T_reg cells⁷⁴. However, multimer-based methods are limited by manufacturing difficulties as well as limited knowledge of relevant antigens. Moreover, this technology is highly influenced by cell freezing⁷⁵, making its use in human studies challenging.

As an alternative, many groups are turning towards AIM assays in which whole proteins or mixtures of peptides are added to blood or peripheral blood mononuclear cells, resulting in rapid upregulation of antigen-stimulated surface proteins. Antigen-specific T_reg cells can be identified using a variety of markers, including CD25, CD134 (OX40) and CD137 (4-1BB), and are distinguished from conventional T cells based on the absence of CD154 (CD40L) expression^76,77,78,79. With AIM assays growing in popularity⁷⁶, we predict that they will increasingly be incorporated into immune monitoring for antigen-specific T_reg cells.

Functional assays

T_reg cells are functionally defined by their suppressive capability; their study was revolutionized by Thornton and Shevach who developed an in vitro suppression assay, which, in its original form, tested for the ability of T_reg cells to prevent proliferation and IL-2 production by stimulated effector CD4⁺ T cells⁸⁰ (Fig. 2c). Although many variations on this assay have since been developed and the assay is considered the ‘gold standard’ to assess T_reg cell function, it is limited by two major caveats. First, the mechanisms of suppression measured in this assay and the assay’s physiological relevance are unclear. Experiments testing roles for TGFβ, IL-10, CTLA4 and CD39–CD73 in suppression generally show no, or only partial, effects. Second, activated effector T cells can also exhibit suppressor-like activity via unknown mechanisms⁸¹. An alternate functional assay that is gaining popularity is to co-culture T_reg cells with APCs and assess the subsequent downregulation of co-stimulatory markers on the latter⁸² (Fig. 2c). The development of improved in vitro assays where T_reg cell mechanisms of action are defined and reflect relevant in vivo effects is urgently needed. Systems such as organoid-based co-cultures⁸³, live tissue slices⁸⁴ and organs on chips⁸³ could offer new approaches to these old issues.

Boosting endogenous T_reg cells

Given that T_reg cells are essential for maintaining immune homeostasis and that most hyperinflammatory diseases can, in part, be ascribed to insufficient T_reg cell efficacy, there is significant interest in devising ways to increase T_reg cell activity. In this section, we focus on strategies to enhance T_reg cells in vivo either by harnessing natural mechanisms or by pharmacological approaches.

Harnessing natural mechanisms

Allergen immunotherapy

Allergen immunotherapy (AIT) for foreign antigens, including food and aero-antigens^85,86, seeks to reverse or prevent allergic symptoms mediated by antigen-specific T helper 2 (T_H2) cell immunity and enhance T_reg cells by repeated, low-dose allergen exposure to restore tolerance (Table 1). AIT is particularly effective for peanut allergy, as shown in the large PALISADE trial⁸⁷, where 67.2% of treated patients, compared to only 4% of placebo-treated patients, could eat peanut protein without symptoms at the end of the trial. Although the effects of AIT are presumed to be at least partly due to the induction of T_reg cells, consistent evidence supporting this mechanism in humans is lacking, possibly because few studies tracked changes in antigen-specific T_reg cells or considered changes in ratios of T_reg cells to allergen-specific TH2 cells (Table 1). For example, the PALISADE trial did not find increased peanut-specific circulating T_reg cells, although there was a decrease in peanut-specific T_H2 cells⁸⁷, potentially resulting in a beneficial balance between these cells. Indeed, a study of T_reg cells in birch allergy found that ratios of antigen-specific T_reg to T_H2 cells best correlated with clinical phenotypes⁷⁷. On the other hand, another study of peanut AIT did find an increase in circulating antigen-specific T_reg cells using assays that showed decreased TSDR methylation and enhanced in vitro suppression⁸⁸. Administration of peanut AIT in combination with omalizumab, an anti-IgE antibody that reduces allergen-triggered inflammation, also led to enhanced antigen-specific in vitro suppression by T_reg cells⁸⁹. Of note, continued efficacy of AIT seems to be reliant on consistent antigen exposure, suggesting that, if T_reg cell-mediated tolerance is enhanced, it might not be permanently re-set for life, contrary to the theory of infectious tolerance.

Table 1 Selected clinical studies reporting effects of oral allergen-specific immunotherapy on antigen-specific T_reg cells

Limitations of AIT include significant risks, side effects and patient compliance; therefore, various modifications of AIT are being explored. For example, a short regimen of allergen peptides modified to be less inflammatory increases the proportion of activated T_reg cells and decreases grass allergy symptoms⁹⁰. Similarly, less immunogenic mannan-coupled ‘allergoids’ that are taken up by dendritic cells and induce T_reg cells⁹¹ are effective in treating patients with dust mite allergy⁹².

Microbiome and environment

T_reg cells are strongly influenced by the microbiome and also control immune homeostasis to commensal bacteria. Therefore, there is significant interest in developing microbiota or other environmental interventions to enhance T_reg cell function. As an example, allergic infants lacking a series of Clostridiales species had fewer T_reg cells in their blood than healthy controls⁹³. When these bacteria species were introduced to allergy-prone mice, there was an increase in protective T_reg cells. The reverse approach is also possible, with T_reg cell-promoting therapies facilitating beneficial microbiome changes. In mice, faecal microbiota transplant from animals previously treated with low-dose IL-2 to expand T_reg cells offered increased protection from colitis and T1DM compared to controls⁹⁴. Similarly, in a small number of patients with autoimmune disease, treatment with low-dose IL-2 promoted a T_reg cell-permissive intestinal microbiome⁹⁴.

Vitamin D, obtained from the diet or generated in the epidermis upon ultraviolet B light exposure, also influences T_reg cells. Daylight influences the number of circulating T_reg cells in healthy people, and a study conducted in the northern hemisphere showed that peak T_reg cells in July and August correlated with the highest blood vitamin D levels⁹⁵. A systematic review of the impact of oral vitamin D supplementation concluded that it increases the number of circulating T_reg cells relative to placebo and enhances their suppressive function⁹⁶. A study of more than 25,000 healthy people showed supplementation with 2,000 IU of vitamin D per day resulted in a 22% drop in the incidence of autoimmune disease compared to placebo after 5 years⁹⁷. Although these studies remain correlative, they suggest that manipulation of micronutrients could be a promising approach to enhance T_reg cell function.

Diet and exercise

A so-called Western lifestyle consisting of high-fat, salt and caloric intake might be a causative factor driving rising rates of autoimmunity. High levels of salt inhibit T_reg cells, which might be due to effects on mitochondrial respiration⁹⁸. In terms of obesity, adipose tissue in overweight mice and humans is characterized by low-grade inflammation and decreased T_reg cell frequency and function relative to lean individuals^99,100,101. Lean individuals fed a high-fat diet for 2 weeks had decreased T_reg cell frequencies and elevated markers of inflammation in their subcutaneous fat, relative to baseline¹⁰⁰. In a pilot trial that studied the effects of diet and the microbiome in people with multiple sclerosis, 2 weeks of intermittent fasting led to microbiome changes that were similar to those seen in mice following the same diet and to a small but significant increase in proportions of circulating T_reg cells¹⁰².

In mice, exercise promotes T_reg cell function by inducing expansion of the muscle T_reg cell compartment, thereby limiting the production of IFNγ, an inflammatory cytokine that impairs exercise-induced performance enhancement¹⁰³. Therapeutic boosting of T_reg cells in mice improves insulin sensitivity in obesity⁹⁹ and muscle regenerative capacity^104,105. At least some of the beneficial effects of exercise on T_reg cells are mediated by IL-6 released from muscle during exercise¹⁰⁴, challenging the concept that IL-6 might destabilize T_reg cells and aligning with human T_reg cell-based studies showing beneficial effects of inflammatory cytokines³⁶. Further investigation into how T_reg cells promote muscle recovery might reveal relevant mechanisms that could be harnessed to treat diseases such as sarcopenia, sterile muscle injury and tocilizumab-associated muscle weakness.

Pharmacological approaches

Drawbacks to the non-pharmacological interventions discussed above include poor compliance and highly heterogeneous, non-T_reg cell-specific effects. Thus, a significant area of growth is in developing traditional drug-based approaches to boost T_reg cells in vivo.

Low-dose IL-2

IL-2 is a critical cytokine that is indispensable for T_reg cell function and survival. Although T_reg cells are unable to produce their own IL-2, they express the high-affinity trimeric IL-2 receptor (including CD25)⁴; therefore, multiple clinical trials have tested the effect of administering low doses of IL-2 to promote T_reg cell expansion¹⁰⁶. T_reg cells with their high-affinity IL-2 receptor should outcompete CD25^– effector T cells for IL-2 administered at low doses. Although low-dose IL-2 therapy is well tolerated, evidence for its efficacy is mixed. In trials for chronic graft-versus-host disease (GVHD)¹⁰⁷ and hepatitis C virus-induced vasculitis¹⁰⁸, more than half of patients showed improved symptoms after IL-2 therapy. Similarly, the TRANSREG study, which enroled patients with 11 different autoimmune diseases, reported decreased disease activity¹⁰⁹. Other studies treating rheumatoid arthritis¹¹⁰, systemic lupus erythematosus (SLE)^111,112 and T1DM¹¹³ also reported beneficial effects. In transplantation, low-dose IL-2 administered to two face transplant recipients resulted in enhanced T_reg cell frequency in the skin graft and higher suppressive capacity¹¹⁴.

On the other hand, several studies found no benefit or detrimental effects of low-dose IL-2. The LITE trial, which aimed to decrease immunosuppression in liver allograft recipients, was halted due to a potentially increased risk of rejection with no evidence for alloreactive T_reg expansion¹¹⁵. In the TILT study, autologous polyclonal T_reg cells plus low-dose IL-2 were combined to treat new-onset T1DM but showed no metabolic benefit and increased circulation of pro-inflammatory cells⁶².

These heterogeneous results might be partly explained by the widely varying dosing regimens between studies, both in terms of timing and amount of cytokine. Given the short half-life and potential negative feedback pathways of the cytokine, small differences in these parameters could have major effects on outcomes¹¹⁶. Another consideration is that low-dose IL-2 could also expand other, non-suppressive immune cells leading to deleterious impacts. It consistently causes an increase in circulating CD56^bright natural killer cells and eosinophils, although these appear transient and not harmful¹¹⁷. However, a study found that low-dose IL-2 increases inflammatory granzyme B⁺ lymphocytes and clonally expanded CD8⁺ T cells⁶², potentially fuelling hyperinflammation. Overall, the lack of a true T_reg-specific effect and undesirable pharmacodynamics makes low-dose IL-2 a challenging strategy to boost endogenous T_reg cells.

Modified IL-2 for T_reg selective effects

Given the challenges with natural IL-2, strategies to modify IL-2 to increase its specificity for T_reg cells and extend its half-life are being pursued with three general approaches. The first is to mutate the IL-2 protein so it is more selective for CD25, resulting in so-called ‘muteins’. Mutant forms of IL-2 are often combined with a second approach of adding a half-life-extending moiety. Efavaleukin alfa (previously AMG 592) is a mutated form of IL-2 with decreased binding to the IL-2Rβ subunit and increased reliance on the IL-2Rα subunit (CD25) of the IL-2R; it is also fused to an immunoglobulin Fc domain to extend its half-life. This drug was in clinical trials for ulcerative colitis, but was terminated due to meeting a prespecified futility rule (NCT04987307). An earlier trial of efavaleukin alfa in SLE (NCT03451422) showed T_reg cell expansion with minimal off-target effects on conventional T cells and natural killer cells¹¹⁸, but clinical development was also halted due to a likelihood of it being ineffective. Similarly, development of other Fc-fusion IL-2 muteins was stopped by Roche and Bristol Myers Squibb due to a lack of efficacy in ulcerative colitis and psoriasis, respectively. Merck, Cugene and Xencor also performed trials with Fc-fusion IL-2 muteins in ulcerative colitis (NCT04924114) or healthy volunteers (NCT05328557, NCT04857866), with results pending.

Exemplifying a second approach to modify IL-2, rezpegaldesleukin (NKTR-358) consists of wild-type IL-2 conjugated to polyethylene glycol to extend its half-life. Compared to unmodified IL-2, it has improved T_reg cell selectivity, a longer half-life and consistently expands T_reg cells, with only a few people exhibiting CD56^bright natural killer cell expansion¹¹⁹. A phase II study to treat SLE (NCT04433585) showed some symptom improvement¹²⁰ and there is reported success in treating patients with eczema (NCT04081350) or psoriasis (NCT04119557)¹²⁰.

A third approach, which is still in preclinical development, is to couple IL-2 with another moiety for increased functionality. For example, an anti-human IL-2 antibody (F5111) was converted into a single-chain antibody and complexed with human IL-2, creating a so-called ‘immunocytokine’ (licensed by Cartesian Therapeutics). This immunocytokine preferentially activates and expands T_reg cells, showing efficacy in mouse models of colitis and immune-checkpoint inhibitor-induced diabetes mellitus¹²¹. In another approach, IL-2 has been fused to other beneficial cytokines. For example, IL233 is a hybrid cytokine linking IL-2 and IL-33 (licensed by Slate Bio) so that IL-2 can induce IL-33R expression on T_reg cells, making them IL-33 responsive. The combined effects of IL-2 and IL-33 expand T_reg cells in vivo, protect against autoimmunity and reduce inflammation^122,123.

Overall, boosting T_reg cells with IL-2, using either the native cytokine or various modified forms, has not resulted in the major success originally hoped for based on data from preclinical models. Apart from the continuing challenge that T_reg cells are not the only cells expressing CD25, a fundamental question with this approach is whether transiently increasing T_reg cells in an antigen non-specific way is sufficient to induce tolerance. The answer to this question likely depends on the disease context and patient-specific factors. More work is needed to understand how IL-2 therapy might enhance disease-relevant T_reg cells and how it could be combined with other approved therapies for synergistic effects.

Nanomedicine-based therapy to boost T_reg cells

Thanks to the success of the COVID-19 mRNA vaccines, there has been an explosion of interest in using biodegradable nanomedicines to target T_reg cells, with most approaches seeking antigen-specific T_reg cell induction or expansion¹²⁴ (Table 2). In mice, nanoparticles loaded with autoantigen-encoding mRNA¹²⁵ or complexed with autoimmune peptides¹²⁶ prevent progression of experimental autoimmune encephalomyelitis and T1DM, respectively, leading to T_reg cell expansion. Nanoparticles can also be conjugated with antibodies to target them to certain locations. For example, antigen-complexed nanoparticles targeting scavenger and mannose receptors on liver sinusoidal endothelial cells promote T_reg cell expansion and subsequent suppression of airway inflammation in mice¹²⁷. Moreover, nanoparticles can be loaded with T_reg-promoting substances such as rapamycin¹²⁸, TGFβ¹²⁹ or F5111 immunocytokine¹³⁰.

Table 2 Clinical-stage nanomedicines to boost T_reg cells

This preclinical success has led to multiple companies entering clinical-stage testing of such nanomedicines to treat allergy and autoimmunity (Table 2). Each approach uses nanoparticles with immunomodulatory properties expected to increase T_reg cell activity. For example, both Moderna and Cartesian Therapeutics have developed nanoparticles that deliver IL-2 muteins. Furthermore, rapamycin nanoparticles (ImmTOR) that reduced the development of anti-drug antibodies specific for uricase (used to treat gout)¹³¹ were shown to induce significant T_reg cell expansion when combined with an IL-2 mutein¹³². TOPAS Therapeutics and COUR (in partnership with Takeda) are developing nanoparticles loaded with tolerogenic autoantigens that induce T_reg cells to treat a variety of autoimmune disorders and allergies. Given the broad interest in nanoparticle-based therapies as well as in modulating T_reg cell activity, there are likely to be increased numbers of clinical trials in this area in the future.

T_reg cellular therapy

The strategies discussed above to expand T_reg cells in vivo are relatively low cost and feasible but the effects could vary significantly depending on the immunological history of a patient and the environmental context. An alternate approach to bolster T_reg cells is to administer them as a cell-based therapy (Fig. 3). Numerous approaches are being developed for obtaining, expanding and engineering T_reg cells for optimal therapeutic application.

Fig. 3: T_reg cell therapy steps and considerations.

a, Regulatory T (T_reg) cells are commonly sorted from blood, umbilical cord blood or thymus based on their expression of CD4⁺CD25^highCD127⁻ and, commonly, CD45RA⁺. Alternatively, conventional CD4⁺ T cells can be genetically engineered to express FOXP3. b, T_reg cells can be expanded ex vivo either using a polyclonal general T cell stimulus or by antigen-presenting cells (APCs) presenting disease-relevant antigens. The latter approach also enriches T_reg cells for antigen specificity. c, T_reg cell antigen specificity can also be modified by genetically engineering cells to express a T cell receptor (TCR) or chimaeric antigen receptor (CAR). d, T_reg cell therapies are often administered to patients in combination with existing immunosuppressive regimens. In the future, gene editing could be used to incorporate orthogonal receptors and/or remove deleterious proteins. e, Bespoke T_reg cell products with optimal function, stability and persistence hold promise to treat many diseases.

Sources of natural T_reg cells for therapy

Early clinical trials investigating the safety and feasibility of T_reg cell therapy were focused on autologous T_reg cells isolated from peripheral blood. T_reg cells were enriched from peripheral blood mononuclear cells using magnetic beads to deplete CD8⁺ T cells and enrich CD25⁺ cells. However, because of the relatively low purity of T_reg cells achieved with this approach, these cells must be expanded in the presence of rapamycin to limit growth of contaminating conventional T cells and yield a potent, immunosuppressive T_reg cell product¹³³. An alternate approach is to use flow cytometry-based cell sorting to obtain a purer starting population of T_reg cells, with sorting typically based on a combination of CD4, CD25 and CD127. Sorting strategies can also further select naive T_reg cells based on CD45RA^134,135, CD226 (ref. ¹³⁶) or GPA33 (ref. ¹³⁷) to enrich for cells co-expressing FOXP3 and HELIOS.

In an effort to decrease the cost and complexity of T_reg cell therapy, several groups are exploring the possibility of manufacturing T_reg cells from allogeneic donors, with a focus on two tissues that contain a large proportion of naive T_reg cells: umbilical cord blood (UCB) and thymus. Advantages of UCB T_reg cells include an increased TCR repertoire diversity and lineage stability relative to blood-derived cells^138,139. UCB T_reg cell therapy also decreases the risk of acute GVHD^140,141. Furthermore, fucosylation (the addition of a Siayl-Lewis X moiety onto P-selectin) has been used to alter the homing potential of UCB-derived T_reg cells, with promising preliminary data and an ongoing phase II clinical study¹⁴². A key question is the requirement for HLA matching with these allogeneic cells. In the studies by Brunstein et al.^140,141, the administered T_reg cells were partially HLA matched (4–6 of 6 HLA antigens), whereas a recent study of UCB T_reg cells for COVID-19-associated acute respiratory distress syndrome did not perform intentional HLA matching and did not observe an increase in anti-HLA antibodies compared to the placebo group¹⁴³. UCB-derived polyclonal T_reg cells developed by Cellenkos are being used to treat refractory bone marrow failure syndrome (NCT03773393). Preclinical studies are also testing UCB T_reg cells in a variety of non-traditional contexts such as traumatic brain injury¹⁴⁴ and lung inflammation¹⁴⁵.

Thymuses are routinely removed from infants during cardiac surgery procedures and contain large numbers (0.3–3 billion) of T_reg cells. Thymus-derived T_reg cells are more suppressive than adult blood-derived cells¹⁴⁶, leading to the development of protocols for their isolation and expansion that are compatible with good manufacturing practices^147,148,149. Recently, the first testing of thymus-derived T_reg cells was reported in a single patient, with the delivery of 20 × 10⁶ cells/kg autologous cells 9 days after a heart transplantation. Notably, despite ongoing treatment with calcineurin inhibitors as immunosuppressants, the patient maintained T_reg cells at higher levels than before the transplantation throughout the 2-year follow-up period¹⁵⁰, suggesting this approach stably increased the T_reg cell pool.

Generating therapeutic T_reg cells from conventional T cells

Given the difficulties in isolating pure populations of T_reg cells, another approach is to re-programme the more numerous conventional T cells into T_reg cells. To increase FOXP3 expression, conventional CD4⁺ T cells can be cultured in the presence of rapamycin, retinoic acid and/or TGFβ to generate induced T_reg (iT_reg) cells, which express FOXP3 for a period of time. Therapeutic administration of iT_reg cells (CD4⁺CD25^– T cells cultured in rapamycin and TGFβ1) was shown to be safe in patients at risk of GVHD¹⁵¹. However, there is some doubt about the stability of iT_reg cells because their levels of TSDR demethylation never approach those of ex vivo T_reg cells and they do not express HELIOS^152,153.

Recently, two groups generated iT_reg cells via differentiation of induced pluripotent stem (iPS) cells. CD4⁺ T cells were first derived from iPS cells, and then FOXP3⁺, TSDR-demethylated T_reg cells were generated by expanding the CD4⁺ T cells in a cocktail of CDK8 and CDK19 inhibitor, rapamycin, TGFβ, and an agonistic TNFR2 antibody¹⁵⁴ or, as reported in a preprint, by expanding them in TGFβ plus all-trans retinoic acid¹⁵⁵. Given that iPS cells can theoretically give rise to unlimited numbers of well-characterized cells, these methods will undoubtedly be of significant interest.

T_reg cells can also be obtained by overexpressing FOXP3 in conventional T cells. To achieve the desired effect, FOXP3 must be controlled by a strong, constitutive promoter so that expression does not significantly diminish as cells enter a resting state^7,156. CRISPR-based methods are also being attempted, whereby a strong promoter is inserted immediately upstream of the endogenous FOXP3 locus to drive constitutive expression¹⁵⁷. FOXP3 manipulation can also promote T_reg cell differentiation from haematopoietic stem and progenitor cells (HSPCs). Interestingly, constitutive overexpression of FOXP3 in UCB-derived CD34⁺ HSPCs promotes stem cell quiescence and impairs T cell differentiation¹⁵⁸, but this can be overcome by delivering FOXP3 with its regulatory elements to recapitulate physiological, non-constitutive expression¹⁵⁹. Further, genome engineering to co-express FOXP3 and HELIOS can generate more functional T_reg cells than engineering to express FOXP3 alone¹⁶⁰.

An outstanding question is whether overexpression of FOXP3 will fully recapitulate all the functions of natural T_reg cells. So far, the functional properties of human, polyclonal conventional T cells overexpressing FOXP3 have only been tested using in vitro suppression assays or in vivo xenogeneic models of GVHD and IPEX¹⁶¹. T cells that constitutively express FOXP3 following lentivirus transduction are being investigated to treat IPEX (NCT05241444), and data from this study will be key for understanding the therapeutic potential of FOXP3-expressing conventional T cells as a replacement for T_reg cells.

Expanding polyclonal or antigen-specific T_reg cells

To acquire a clinically relevant number of cells, autologous T_reg cells are usually stimulated and expanded ex vivo. Most clinical trials have used polyclonally expanded cells (Fig. 4), which have been tested in many different diseases (see Bluestone et al. for a recent comprehensive review¹⁶²). Of particular interest is the ONE study, an international, multi-centre study that tested several cell therapies in living donor kidney transplantation with the goal of minimizing the requirement for immunosuppressive drugs. Patients who received T_reg cell therapy did not undergo increased levels of rejection despite not being given the complete standard-of-care immunosuppressive therapy and, moreover, they had lower rates of viral infections¹⁶³. As part of the ONE study, Roemhild et al. showed that most (8 out of 11) T_reg-treated patients achieved stable immunosuppression with a single drug, whereas a reference group remained on standard dual or triple immunosuppressive drugs¹⁶⁴. Interestingly, T_reg cells in the blood shifted to a less diverse TCR repertoire, suggesting there could be an alloantigen-driven selection process for these cells. The ongoing TWO study (ISRCTN: 11038572) aims to confirm the benefit of polyclonal T_reg cell therapy in transplantation by testing whether immunosuppressive drug treatment can be tapered to a low-dose of tacrolimus alone following polyclonal T_reg cell infusion in kidney transplant recipients¹⁶⁵. A subset of patients from this trial showed successful transition to tacrolimus monotherapy without transplant rejection¹⁶⁶.

Fig. 4: Clinical trials of T_reg cell therapy.

Time course of the number of clinical trials investigating regulatory T (T_reg) cell therapy, based on data available from ClinicalTrials.gov and EudraCT. Ongoing (striped bars) refers to the number of clinical trials that were registered as recruiting in each given year. Complete (solid bars) refers to the number of clinical trials that were recorded as complete or terminated in each given year. Trials are separated into disease areas. GVHD, graft-versus-host disease.

Clinical trials exploring the use of polyclonal T_reg cells to inhibit GVHD following HSPC transplantation have also yielded promising results, with most treated patients showing clinical improvement and decreased GHVD activity^{167,168,169,170,171}. An interesting new approach is the manipulation of the graft itself by infusing a polyclonal T_reg cell-enriched product, a strategy being investigated by Orca Bio. In several studies, patients received a CD34⁺-selected HSPC graft supplemented with T_reg cells (sorted from the graft), followed by infusion of defined numbers of donor-derived conventional T cells 2 days later^167,170. Patients treated with this product had early myeloid engraftment, and low rates of GVHD and relapse compared to standard of care^167,170,171. A phase III study known as Precision-T (NCT05316701), has completed enrolment with results expected in 2025. These intriguing results suggest that more research is needed to understand how T_reg cells can benefit HSPC engraftment. Indeed, the bone marrow is known to be a rich source of T_reg cells, yet little is known about their biology and function in this tissue.

In T1DM, there have been mixed reports about the clinical efficacy of autologous polyclonal T_reg cells to preserve β-cell function. Some studies showed preservation of insulin production and reduced dependence on exogenous insulin^172,173, particularly when combined with a B cell-depleting antibody (rituximab)¹⁷⁴, whereas others failed to find clinical impacts^62,175,176. These outcome variations could be due to differences in T_reg doses, number of infusions or the timing of treatment with respect to disease onset.

Polyclonal T_reg cells have also been sporadically tested in single individuals for the treatment of SLE¹⁷⁷ and ulcerative colitis¹⁷⁸, further confirming the wide applicability and excellent safety profile of this strategy. These studies noted an enrichment of T_reg cells in the skin and intestine, respectively, but the low number of patients treated limits the ability to draw meaningful conclusions about therapeutic efficacy. An ongoing phase II clinical trial in amyotrophic lateral sclerosis (NCT05695521) will reveal potential utility beyond traditional immune-mediated diseases (Box 1).

There have been attempts to expand antigen-specific T_reg cells using APCs ex vivo. For the majority of autoimmune conditions and inflammatory disorders, the feasibility of this expansion approach is poor as only a small number of disease-relevant T_reg cells naturally reside in the periphery. However, in a transplant context, approximately 1–10% of peripheral T_reg cells can be activated by donor antigens¹⁷⁹. Consequently, ex vivo stimulation with graft donor B cells expands autologous T_reg cells that predominantly recognize donor-derived antigen–MHC complexes, termed donor-alloantigen reactive T_reg (darT_reg) cells¹⁸⁰. This approach has been successfully tested in three kidney transplant recipients¹⁸¹. However, a separate study in liver transplantation emphasized the difficulties of manufacturing clinically relevant doses using this strategy, with 44% and 33% of recruited participants yielding insufficient or only partial doses of T_reg cells, respectively¹⁸². Furthermore, despite a promising safety profile from these clinical studies, data from a recent non-human primate heart transplantation model suggested that darT_reg cells might not be stable as they lose signature T_reg cell markers following infusion¹⁸³.

Genetic engineering of T_reg cells

The expansion phase of the T_reg cell manufacturing protocol provides an opportunity for genetic manipulation and enhancement. Engineering can improve T_reg cell function, stability, trafficking and/or persistence in vivo following adoptive transfer. Most research has focused on improving T_reg cell potency by conferring antigen specificity with TCRs or chimaeric antigen receptors (CARs). Key considerations for this approach are whether overexpression of a TCR or CAR is more suitable (Table 3) and selection of the target antigen (Box 2).

Table 3 Comparison of key features of engineered TCRs versus CARs

Box 2 Selecting a CAR T_reg cell target antigen

Chimaeric antigen receptor (CAR) regulatory T (T_reg) cells incorporate an antibody-derived antigen-targeting moiety, enabling them to engage a range of possible target antigens in a major histocompatibility complex (MHC)-independent manner. Below are some factors that should be considered when selecting a target antigen for CAR T_reg cells.

Disease relevance

In oncology, CARs redirect T cells towards antigens that might not be strictly disease-specific, leading to the potential for so-called ‘collateral damage’. For T_reg cells, whether CAR target antigens need to be disease relevant is unclear. To date, most CAR T_reg cell studies selected a disease-relevant antigen, for example, in transplantation targeting allogeneic MHC molecules known to be targets of rejection or in autoimmunity selecting known autoantigens. Because T_reg cells are not expected to mediate collateral damage, it might be feasible to redirect specificity in an organ-specific or tissue-specific way but not necessarily in a disease-specific way.

In vivo location of suppression

Another consideration for CAR target selection is knowing where in vivo suppression should occur. In organ transplantation, evidence suggests suppression is needed both in the allograft and in associated lymph nodes. Redirecting CAR T_reg cells to allogeneic MHC molecules would be expected to induce suppression in both these locations due to exosome-mediated cross-decoration of donor MHC on host antigen-presenting cells. For autoimmunity, it remains to be defined if CAR T_reg cells targeting tissue-specific antigens will also need to have activity in lymph nodes and, if so, how the relevant antigens will need to be displayed.

TCR engineering

Antigen engagement and subsequent TCR signalling by T_reg cells drive retention of the cells in the relevant tissue and enhanced suppression; thus, a logical approach to redirect T_reg cell specificity is to introduce new TCR α-chain and β-chain genes. This idea has been explored in a variety of contexts, most notably in mouse models of T1DM¹⁸⁴. Initial studies in human cells focused on modifying blood-derived T_reg cells¹⁸⁵, and TCR-engineered blood-derived T_reg cells are now being pursued for the treatment of multiple sclerosis by Abata Therapeutics (ABA-101). More recently, TCR delivery has been combined with FOXP3 editing¹⁵⁷ to generate islet-specific T_reg cells from conventional T cells¹⁸⁶. Mouse TCR-transgenic Foxp3-edited T_reg cells were shown to home to the pancreas, stably persist, and prevent disease development in an adoptive transfer model of T1DM. This multi-editing approach is now being pursued commercially for the treatment of T1DM by GentiBio (GNTI-122), using conventional T cells engineered to express FOXP3, an islet-specific TCR and a rapamycin-activated, chemically induced IL-2 signalling complex (allowing rapamycin to drive IL-2 signalling in these cells)¹⁸⁷.

CAR engineering

CARs are synthetic fusion proteins that typically bypass the requirement for antigen–MHC interactions through the use of an antibody-derived antigen-targeting moiety attached to TCR signalling domains¹⁸⁸. Early studies investigating the potential of CARs to confer specificity to human T_reg cells were performed in the context of transplantation, where HLA-A2 was targeted as a clinically relevant human MHC molecule that is commonly mismatched. In vivo studies of mouse or human HLA-A2-specific CAR (A2-CAR) T_reg cells showed that CAR expression controlled T_reg cell homing to transplanted HLA-A2⁺ skin or islet grafts, in turn delaying transplant rejection^189,190,191.

Multiple groups also refined CAR design for optimal T_reg cell function by testing various intracellular signalling domain configurations. These studies uniformly showed that CARs encoding CD28 co-stimulatory signalling domains but not the CD137 co-stimulatory domain are optimal for T_reg cells^82,192,193. Interestingly, in a skin transplant model, a first-generation CAR with no co-stimulation domain was also effective, revealing that CAR T_reg cells respond to endogenous co-stimulation¹⁹³. Overall, the potent effects of A2-CAR T_reg cells led to the rapid development of good manufacturing practice-compatible protocols for CAR T_reg generation³⁹, and two clinical trials are testing the safety and efficacy of A2-CAR T_reg cells in kidney (NCT04817774) and liver (NCT05234190) transplant recipients.

In the context of GVHD, the B cell antigen CD19 has been pursued as a CAR T_reg target^194,195 with the rationale that systemic CD19⁺ B cells will enable widespread CAR T_reg cell stimulation. Remarkably, CD19-specific CAR T_reg cells not only reduced GVHD severity but also controlled the growth of CD19⁺ tumour cells. Importantly, this was achieved in the absence of cytokine release syndrome, a common adverse event associated with CAR T cell therapy but not with CAR T_reg cell therapy, given the latter’s relative lack of inflammatory cytokine production¹⁹⁶. CD19-CAR T_reg cells might also be suitable for the treatment of autoimmune diseases in which B cells and/or antibodies play pathological roles. Recently, FOXP3-overexpressing, CD19-CAR T_reg cells reduced autoantibody generation and delayed lymphopenia in a humanized mouse model of SLE¹⁹⁷.

A fundamental advantage of using CARs to redirect T cell specificity is the ability to engage target antigens in an MHC-independent manner (Table 3). However, in some cases, ideal antigens might be intracellular or predominantly soluble, making them difficult CAR targets. This limitation was recently overcome by generating so-called ‘TCR-like’ CARs specific for an insulin peptide presented by MHC class II. In two parallel studies, T_reg cells expressing insulin-MHC-specific CARs suppressed pathogenic T cell proliferation in vitro and significantly delayed or prevented T1DM in vivo^198,199.

Depending on the structure and environmental context of an antigen, it is possible to use CARs to redirect T_reg cell specificity to non-membrane-bound proteins. For example, CAR T_reg cells targeting the blood coagulation factor VIII are efficacious in mouse and humanized mouse models of haemophilia A^200,201. More recently, CAR T_reg cells targeting flagellin, the protein component of bacterial flagella, were explored as a treatment option for inflammatory bowel disease²⁰². Flagellin was selected as it is naturally oligomeric and primarily accessible to immune cells during periods of inflammation and gastrointestinal damage. Flagellin-specific CAR expression promoted intestinal trafficking, and CAR T_reg cells were significantly more suppressive and could promote intestinal epithelial cell integrity in the presence of their target antigen.

Ensuring stability of antigen receptor-engineered T_reg cells

A general concern with T_reg cells, particularly those modified to express antigen-specific receptors, is lineage stability. Injection of human CAR T_reg cells into mice systemically expressing HLA-A2 did not reveal significant loss of FOXP3 (refs. ^38,39,190). Similarly, in a non-human primate model of islet transplantation, infusion of autologous Bw6-specific CAR T_reg cells into Bw6⁺ animals did not result in any overt toxicity despite ubiquitous target antigen expression in vivo⁴⁰. Nevertheless, CAR-induced lineage changes are possible as evidenced by the loss of HELIOS (but not FOXP3) expression in mouse and human T_reg cells expressing CARs encoding TNFR family co-receptors^82,193.

Strategies to safeguard against instability and promote stable FOXP3 expression include expanding cells in the presence of FOXP3-promoting molecules (such as rapamycin or TGFβ) or genetic manipulation to overexpress FOXP3. For example, McGovern et al. developed a method to transduce an enriched population of T_reg cells with a vector encoding FOXP3 and a myelin basic protein-specific TCR under the same promoter, restricting transgenic TCR expression to FOXP3⁺ cells²⁰³.

Another factor that could influence T_reg cell stability is dysfunction induced by overstimulation. In studies of tonic-signalling CARs in T_reg cells, chronic stimulation caused an exhausted phenotype and loss of function²⁰⁴ and, in another study, was associated with loss of FOXP3 and HELIOS expression (although the cells did not acquire an effector phenotype)²⁰⁵. This tonic signalling can be mitigated by careful selection of CAR co-stimulatory domains and, possibly, by also using CRISPR technology to insert CARs into the TCRα constant (TRAC) locus, resulting in TCR-like transcriptional control and a more physiological level of CAR expression¹⁹⁰.

Combining engineered T_reg cells with immunosuppression

A consideration yet to be explored is how to condition patients so they are ideally suited to receive engineered T_reg cells. Ideally, T_reg cell therapy should be integrated with existing regimens and opportunities identified to combine it with new types of less toxic immunosuppression. As examples, CAR T_reg cells can work in synergy with rapamycin¹⁹¹, and TCR-transgenic T_reg cells can be combined with low-dose IL-2 (ref. ²⁰⁶) or anti-CD3 (ref. ²⁰⁷). Ongoing work is also defining the optimal strategy to combine T_reg cell therapy with anti-thymocyte globulin²⁰⁸ or CD28 blockade⁵³. Development of combination therapies is a particularly important consideration in transplantation, where patients receive drug-based immunosuppression that could reduce the function of T_reg cell therapy²⁰⁹. For example, gene editing to knock out FK506-binding protein 12 or CD52 could result in T_reg cells that are resistant to tacrolimus²¹⁰ or alemtuzumab (an anti-CD52 antibody)²¹¹, respectively. One can also envision bespoke CAR engineering combined with specific types of immunosuppression, for example, to deliver signals blocked by antibodies or recombinant proteins.

Other gene-editing strategies

Building on the success of TCR and CAR engineering, additional gene-editing strategies are being explored to further enhance the efficacy, survival and/or stability of adoptively transferred T_reg cells. As previously discussed, IL-2 therapy is an effective strategy to bolster T_reg cells. Limitations related to lack of T_reg specificity can be overcome by using engineered, so-called ‘orthogonal’, systems in which synthetic versions of IL-2 and the IL-2 receptor interact with each other but not the endogenous versions^212,213. Engineered expression of an orthogonal IL-2 receptor in adoptively transferred T_reg cells allows selective stimulation in vivo by administration of orthogonal IL-2. The benefits of this approach have been demonstrated in mouse models of GVHD²¹⁴ and heart transplantation²¹⁵.

A similar approach is the use of chimaeric cytokine receptors, or switch receptors, which detect one extracellular cytokine and deliver the stimulatory signal of another. Although this approach has not yet been reported in T_reg cells, a similar concept was developed whereby extracellular detection of pro-inflammatory cytokines triggered CD3ζ and CD28 signalling in T_reg cells²¹⁶. T_reg cells expressing these so-called artificial immune receptors were significantly more effective at alleviating GVHD in a mouse model.

CAR T_reg cell therapy can delay but not prevent tissue rejection^191,217, suggesting the need to further enhance T_reg cell-suppressive mechanisms. In an early test of this concept, A2-CAR T_reg cells were genetically modified to constitutively express IL-10 (ref. ²¹⁸), which is normally expressed at relatively low levels in T_reg cells compared to IL-10-producing type 1 T_reg cells²¹⁹. A similar strategy could be used to introduce other beneficial molecules such as TGFβ¹⁸⁸.

Removal of deleterious genes from therapeutic T_reg cell products can also be achieved with gene editing. In oncology, PD1 ablation improves the efficacy of CAR T cells^220,221 and, given the negative impact of PD1 signalling on T_reg cells⁸², adapting this approach could be beneficial. This strategy could also be used to enhance the function of allogeneic T_reg cells to minimize their immunogenicity. For example, in work reported in a preprint, McCallion et al.²²² used a humanized mouse model of skin transplantation to show that CD8⁺ T cell-mediated killing of allogeneic T_reg cells could be overcome by CRISPR–Cas9-mediated silencing of MHC. Moreover, expression of a non-polymorphic HLA-E–β2-microglobulin fusion protein further protected T_reg cells lacking MHC expression from natural killer cell-mediated killing. Other strategies to evade natural killer cells, such as expression of HLA-E^223,224,225, Siglec ligands²²⁶ or CD47 (ref. ²²⁷), are also likely to have applications in T_reg cells.

Overall, it is clear that, to fully maximize the potential of T_reg cell therapies, gene editing should be employed. Fortunately, the extensive work on gene editing of conventional T cells can be leveraged for T_reg cells, albeit for a different purpose.

Depleting T_reg cells to treat cancer

Cancer evades immunity by creating an immunosuppressive environment that promotes T_reg cells. The prognostic value of T_reg cell abundance in tumours varies depending on the type of cancer but, generally, a higher frequency of intratumoural T_reg cells is associated with decreased overall survival (reviewed in ref. ²²⁸). Therefore, many cancer therapies aim to deplete T_reg cells in order to invigorate the anticancer functions of other immune cells.

Targeted T_reg cell depletion

As discussed earlier, there have been many attempts to selectively expand T_reg cells based on their high expression of CD25. Targeting CD25 has also been used to deplete T_reg cells in cancer. Denileukin diftitox (ONTAK) is a human IL-2 protein fused to diphtheria toxin that was tested for many years; however, T_reg cell depletion was transient and, in some cases, expansion of antigen-specific effector T cells was halted²²⁹. E7777 is a new, purer IL-2–diptheria toxin fusion protein under investigation. A recent clinical trial treating patients with relapsed or refractory lymphoma showed an objective response in one-third of participants; however, half of patients experienced a serious adverse event²³⁰. A different approach tested RG6292, a non-blocking, CD25-depleting antibody that binds CD25 to trigger antibody-dependent cellular cytotoxicity of T_reg cells but leaves the substrate-binding domain of CD25 open to receive IL-2 signalling on remaining effector T cells²³¹. It exhibited preclinical efficacy without immune-related toxicities and a monotherapy clinical trial in patients with advanced solid tumours was recently concluded (NCT04158583).

Because all therapies targeting CD25 have a risk of off-target effects, several new approaches have been developed to more selectively target T_reg cells. For example, AstraZeneca developed an antisense oligonucleotide targeting FOXP3 (AZD8701) that is in clinical testing (NCT04504669). Preclinical modelling showed effective knockdown of FOXP3 throughout the body, including the tumour, as well as reduced T_reg cell-suppressive capacity and a parallel boost in CD8⁺ T cell antitumour activity²³². Preclinical models have also demonstrated the efficacy of reducing intratumoural T_reg cells by targeting CCR8 (refs. ^233,234), a strategy now in clinical testing by Shionogi as a combination therapy with the PD1 inhibitor pembrolizumab (NCT05101070). Interestingly, the controversial relevance of NRP1 in human T_reg biology has recently been re-examined, with Chuckran et al.²³⁵ finding its expression is prevalent in tumour-resident T_reg cells, supporting the rationale for anti-NRP1 therapy (NCT03565445).

Combating the suppressive microenvironment

Immunosuppression is a hallmark of cancer; reducing immunosuppressive cells and/or signals can re-invigorate anticancer immunity. Cyclophosphamide, a cytotoxic alkylating agent used to treat malignancies, also selectively depletes T_reg cells²³⁶. Mechanistically, T_reg cells have less ATP than effector T cells and thus produce less glutathione, which detoxifies cyclophosphamide, making them more susceptible to this drug. Several clinical trials are testing CD39 and CD73 inhibitors in attempts to increase extracellular, pro-inflammatory ATP levels. One preclinical study showed that antibodies blocking CD39 and CD73 were effective in activating effector T cells and slowing tumour progression; however, there was no apparent impact on T_reg cells⁵⁸. Administration of oleclumab, an anti-CD73 antibody, together with an anti-PDL1 antibody (durvalumab) also increased overall response rate and progression-free survival relative to those treated with durvalumab alone²³⁷.

Tryptophan and kynurenine are metabolites that influence the balance of effector cells to T_reg cells. Their relative abundance is regulated by the intracellular enzyme indoleamine 2,3-dioxygenase (IDO1), which converts tryptophan into kynurenine. The latter metabolite decreases effector T cell proliferation and survival and promotes pT_reg cell differentiation. IDO1 is expressed in APCs and many tumour cells and has become an anticancer drug target. However, although there were high hopes for the IDO1 inhibitor epacadostat, it ultimately failed to have any effect in a phase III clinical trial²³⁸. Despite this disappointing result, which was possibly related to suboptimal trial design and lack of mechanistic understanding²³⁹, there are multiple new IDO1-targeting candidates in clinical trials (reviewed in ref. ²⁴⁰).

Galunisertib is a TGFβ type I receptor kinase small-molecule inhibitor that can reverse T_reg cell suppression in vitro²⁴¹. In preclinical models, it lowered T_reg cell abundance and increased lymphoma-related survival²⁴². In a phase II trial for patients with rectal cancer, galunisertib in combination with neoadjuvant chemoradiotherapy improved the complete response rate²⁴³. Interestingly, although the frequency of blood T_reg cells increased with treatment, intratumoural T_reg density and TGFβ signalling decreased in tumour biopsies over time²⁴³. Overall, many anticancer therapies are hypothesized to work, at least in part, via suppressing T_reg cell function. Continued investigation of the impact of these drugs on T_reg cells will provide mechanistic insights that could be leveraged for both pro-T_reg and anti-T_reg cell therapies.

ICIs and T_reg cells

Immune-checkpoint inhibitors (ICIs) are now a mainstay of cancer therapy, working to prevent effector T cell exhaustion and re-invigorate anticancer immunity. For example, inhibitors of the checkpoint proteins PD1 and CTLA4 block effector T cells from receiving inhibitory signals that would otherwise prevent them from killing cancer cells. The effects of these ICIs on T_reg cells are just starting to be elucidated. As with effector T cells, PD1 signalling inhibits T_reg cells, reducing their proliferation and suppressive effects^244,245,246. Thus, an unintended consequence of PD1 blockade can be enhanced T_reg cell function. A small fraction of patients treated with a PD1 ICI develop hyper-progressive disease associated with an increased frequency of proliferating T_reg cells in the blood²⁴⁵. Furthermore, patients who expressed PD1 on over half of their T_reg cells had poor survival following ICI treatment²⁴⁶. Of note, the ratio of tumour-resident PD1⁺ CD8⁺ T cells to PD1⁺ T_reg cells is predictive of clinical response to PD1 ICIs, with a higher ratio predicting favourable outcome²⁴⁶. Thus, in the future, identification of tumours with high PD1⁺ T_reg cells prior to treatment might enable personalized approaches to first deplete T_reg cells before administering PD1 ICIs.

CTLA4 inhibitors are another major class of ICI in clinical use. As discussed above, T_reg cells express constitutively high levels of CTLA4 so it is not surprising that anti-CTLA4 antibodies inhibit them^247,248. In mice, the depletion of intratumoural T_reg cells by CTLA4 ICIs required the presence of tumour macrophages expressing the IgG Fc receptor FcγRIV²⁴⁷. The human homologue of FcγRIV is FcγRIIIA and, in vitro, FcγRIIIA⁺ monocytes mediate the antibody-dependent cell cytotoxicity of CTLA4 ICI-treated T_reg cells²⁴⁸. Interestingly, the abundance of these monocytes is higher in tumours of CTLA4 ICI responders than in non-responders, suggesting that T_reg cell depletion contributes to clinical benefit. However, results are overall inconsistent, leading to the conclusion that the effects of CTLA4 ICIs on T_reg cells vary due to tumour type and time of sampling relative to ICI treatment^248,249,250.

Many patients treated with ICIs experience immune-related adverse events (IRAEs). These events vary in severity and are essentially due to localized and, in some cases, antigen-specific inflammation²⁵¹. Whether T_reg cells are implicated in IRAEs is unclear. In a study of ICI-induced colitis in patients with melanoma, single-cell sequencing analysis of the colon paradoxically found increased frequencies of T_reg cells in the patients with colitis compared with ICI-treated patients without colitis²⁵². However, these T_reg cells were skewed towards an inflammatory and potentially unstable phenotype, perhaps, as the authors suggest, in response to local inflammation²⁵². Interestingly, the risk of IRAEs is increased by a history of autoimmune disease or use of a CTLA4 ICI²⁵³, suggesting that T_reg cell depletion or destabilization (which would be stronger with a CTLA4 ICI) could tip the balance of tolerance in healthy tissues to become permissive of an inflammatory response. In accordance, IRAE incidence positively correlates with response rate and survival²⁵⁴, suggesting that a brief loss of tolerance might be required to allow anticancer immunity to function.

Conclusions and future directions

T_reg cells are undoubtedly a cornerstone of a healthy immune system; however, tracking their contribution to disease and harnessing their full therapeutic potential has remained somewhat elusive. Major limitations in tracking T_reg cells as biomarkers have included the lack of FOXP3 specificity and challenges in quantifying antigen-specific cells. Tools to more accurately identify T_reg cells, such as combining FOXP3 and HELIOS staining, measuring TSDR methylation, and gene signatures, can now be routinely applied to better discriminate between T_reg and conventional T cells. In terms of specificity, AIM assays and a growing repertoire of multimers provide new ways to follow antigen-specific cells (both in blood and other tissues) so that changes in disease-relevant cells can be more accurately quantified. Early use of such tools in allergy and autoimmunity clearly shows that changes in antigen-specific T_reg cells are relevant for human disease^74,77 and malleable in response to therapy (Table 1).

In terms of therapeutic targeting to increase T_reg cell activity, the past two decades have focused on in vivo modulation using relatively unspecific tools or non-antigen-specific adoptive cell therapy. These studies have hinted that it is possible to enhance tolerance and reduce reliance on non-specific immunosuppression. The field is now evolving to address challenges and to create more specific and effective approaches (Fig. 5). For adoptive T_reg cell therapy, although few phase II trials powered to measure efficacy have yet to be completed, it is a significant advance that polyclonal and darT_reg cells can be used to replace or reduce immunosuppressive drugs in the context of organ transplantation^163,181. Even if T_reg cell therapy does not induce long-term tolerance, the ability to reduce toxic immunosuppression would have a major impact on quality of life and morbidity. The potential to expand the application of polyclonal T_reg cell therapy to a variety of inflammatory diseases is also a very exciting new direction (Box 1).

Fig. 5: Evolution of T_reg therapies.

Over the last 10 years, regulatory T (T_reg) cell-directed therapies have primarily focused on in vivo (red) or polyclonal adoptive cell therapy-based ex vivo approaches (blue) to enhance disease-relevant function. These first-generation approaches are consistently safe but there are significant current challenges such as a lack of antigen specificity and poor control over the ultimate effects. The field is rapidly moving to next-generation approaches focused on more precise and specific T_reg cell targeting using protein engineering and nanomedicines for in vivo targeting and gene engineering for optimal adoptive cell therapy approaches. CAR, chimaeric antigen receptor; TCR, T cell receptor.

However, the most exciting potential lies in the evolution of methods to induce antigen-specific T_reg cells either directly in vivo or through adoptive cell transfer. Nanomedicine-based approaches represent a precision approach to tolerance induction; one could envision targeting nanomedicines specifically to T_reg cells, for example, through antibody-mediated targeting²⁵⁵ or mRNA-mediated or microRNA-mediated control of T_reg cell-specific expression²⁵⁶. These approaches have the potential to be significantly more potent than approaches such as AIT, yet still rely on the patient’s own immune system. Delivery of engineered T_reg cells by adoptive transfer offers an even more precise approach and we can foresee that, similarly to T cell therapy for cancer, cells that have been progressively more engineered will be tested over the next decade.

Tolerance is undoubtedly a balance and there is a significant opportunity to apply lessons learned from cancer to autoimmunity and transplantation and vice versa. Understanding the cellular mechanisms underlying the side effects of various ICIs has the potential to reveal significant insight into the biology of human T_reg cells and hence pathways that can be harnessed for tolerance.