Brain regulatory T cells

Abstract

The brain, long thought to be isolated from the peripheral immune system, is increasingly recognized to be integrated into a systemic immunological network. These conduits of immune–brain interaction and immunosurveillance processes necessitate the presence of complementary immunoregulatory mechanisms, of which brain regulatory T cells (T_reg cells) are likely a key facet. T_reg cells represent a dynamic population in the brain, with continual influx, specialization to a brain-residency phenotype and relatively rapid displacement by newly incoming cells. In addition to their functions in suppressing adaptive immunity, an emerging view is that T_reg cells in the brain dampen down glial reactivity in response to a range of neurological insults, and directly assist in multiple regenerative and reparative processes during tissue pathology. The utility and malleability of the brain T_reg cell population make it an attractive therapeutic target across the full spectrum of neurological conditions, ranging from neuroinflammatory to neurodegenerative and even psychiatric diseases. Therapeutic modalities currently under intense development include T_reg cell therapy, IL-2 therapy to boost T_reg cell numbers and multiple innovative approaches to couple these therapeutics to brain delivery mechanisms for enhanced potency. Here we review the state of the art of brain T_reg cell knowledge together with the potential avenues for future integration into medical practice.

Introduction

The brain is increasingly recognized to have connections to the immune system. The immunological compartment of the brain includes both tissue-resident immune cells and leukocytes recruited from the blood, even though these immune populations are less prevalent in the brain than in most other organs. The recruited leukocytes include regulatory T (T_reg) cells, which are a specialized subset of T cells that have a critical role in maintaining immune tolerance and promoting an anti-inflammatory environment. Defects in the number of T_reg cells, or in their suppressive functions, have been implicated in the pathology of many common immunological diseases¹, including neuroinflammatory and neurodegenerative conditions².

T_reg cells in the brain need to be considered in the broader context of tissue-resident T_reg cells found across other non-lymphoid tissues³. Although primarily found in the blood and lymphoid tissues, T_reg cells migrate into non-lymphoid tissues in the steady state, and accumulate under inflammatory conditions. These tissue-resident T_reg cells are thought to recognize self-antigens presented in the tissues, resulting in a limited T cell receptor (TCR) repertoire³. Numerous studies have shown that T_reg cells play diverse roles and exhibit distinct phenotypes in non-lymphoid tissues, including visceral adipose tissue⁴, muscle⁵, the lungs⁶, the gut⁷, the skin⁸ and the liver⁹. A common phenotype emerges across the tissue T_reg cell populations, distinct from that of lymphoid tissue T_reg cells, characterized by elevated expression of molecules such as CD69, IL-10, ST2, AREG, KLRG1 and CTLA4 (refs. ^3,10). An analogous population of tissue T_reg cells is found in the brains of mice^11,12 and humans¹¹, with a cellular phenotype distinct from that of circulating T_reg cells (Fig. 1). Nonetheless, the brain T_reg cell population is not homogeneous. Although the resting naive-like population of T_reg cells that is common in the circulation is near-absent in the brain, a diversity of phenotypes are observed within the activated T_reg cell population present in the brain. These phenotypes range from T_reg cells that show a generic activated phenotype — which can be seen in T_reg cells in lymphoid tissues and is characterized by elevated expression levels of canonical T_reg cell markers such as CD25, FOXP3 and CTLA4 — to T_reg cells that show a tissue-residency phenotype, which is characterized by the expression of the signature ‘residential’ markers observed across multiple tissues — CD69, KLRG1 and ST2 (ref. ¹⁰). It has further been proposed that additional, potentially brain-specific, gene expression is found within the brain-resident T_reg cell population¹¹. Such genes include those encoding various neuropeptides (for example, neuropeptide Y and preproenkephalin) and neuronal receptors (for example, 5-hydroxytryptamine receptor 7 (5-HT₇ receptor) and arginine vasopressin receptor), suggesting potential interactions between T_reg cells and the nervous system¹². The expression of the transcription factor BLIMP1 also enables T_reg cells in the brain to retain their regulatory capacity during inflammatory insult¹³.

Fig. 1: Key features of brain regulatory T cells.

The main identified functional molecules and associated markers for brain regulatory T cells (T_reg cells). These molecules have been classified according to whether they are putative effector molecules, molecules involved in cell migration, residency markers, mediators of homeostasis or neuroimmune interface proteins. AREG, amphiregulin; BDNF, brain-derived neurotrophic factor; CCR2, CC-chemokine receptor 2; TCR, T cell receptor.

Interest in these brain T_reg cells is driven, in part, by surprising findings in other tissues, where the tissue-resident T_reg cells not only control local inflammation but also have roles in tissue homeostasis and repair³. Growing evidence supports the idea that brain T_reg cells likewise have a critical role in neuroprotection and neurological repair, and that these activities are relevant to a range of neuroinflammatory, neurodegenerative and even psychiatric conditions. In this Review, we discuss our current understanding of the T_reg cell populations found in the brain, focusing on the cellular kinetics of T_reg cells entering and residing in the brain, the anti-inflammatory and reparative functions these cells can have, and the likely pathway towards using these properties of T_reg cells in the clinic.

Origin, entry and residency of T_reg cells in the brain

Origin

Most brain-resident T_reg cells are of thymic origin, based on their expression of helios and neuropilin 1 (ref. ¹¹). TCR sequencing of T_reg cells isolated from the brains of mice that have experienced ischaemic stroke reveals a limited diversity of TCR clones, including overlapping TCR sequences between different mice¹², suggesting thymic selection of T_reg cells against a restricted pool of antigens present in brain. In agreement with this, in the experimental autoimmune encephalomyelitis (EAE) mouse model of multiple sclerosis, T_reg cells in the central nervous system (CNS) showed over-representation of specific TCR sequences, suggesting oligoclonal expansion of T_reg cells against self-antigens present in the CNS¹⁴. Independent data suggest that this expansion is based on the incoming T_reg cell pool, rather than being induced de novo from conventional autoreactive T cells¹⁵. Despite this, the self-antigens recognized by brain T_reg cells remain unknown. It is also possible that the TCR repertoire of brain T_reg cells changes over time, with, for example, the proposed early neonatal output consisting of distinct specificities¹⁶, although the high turnover would dilute any effect of such early events by adulthood (see ‘Residency’ below).

As cells of thymic origin, brain T_reg cells need to traffic to their destination organ via the blood. A prerequisite of this brain entry appears to be a prior antigen-stimulation event, likely in lymphoid organs, to impart the trafficking activation status, with TCR-restricted naive-like T_reg cells less able to enter the brain, in both physiological¹¹ and pathological conditions¹². The brain T_reg cell population is dynamic over the life course of mice, with the first detection of these cells occurring around the time of birth and an increase in the population observed with both age and environmental enrichment¹¹. This suggests that the brain-resident T_reg cell population may be influenced by various environmental factors and age-related changes, potentially contributing to its functional diversity in different contexts. To date, relatively few studies have compared T_reg cells within the brain barrier compartments with those within the brain parenchyma, but a phenotypic analysis of meningeal and parenchymal T_reg cells, in mouse and human tissues, found similar residential marker expression¹¹. Although inconclusive, these data are consistent with the barrier compartment and parenchymal cells sharing an origin, being linked through direct migration flows or simply responding to similar microenvironmental signals.

Barrier compartments and T_reg cell entry

Compared with tissue T_reg cells resident in other organs, brain-destined T_reg cells face a daunting task to transition from the bloodstream into tissue residency. Indeed, until the first half of the twentieth century, the brain was considered by many scientists to be immune privileged and lymphocytes were assumed to be excluded from the healthy brain, with the intact blood–brain barrier preventing entry. The exception to this was in the setting of neuroinflammation, such as during multiple sclerosis, where autoreactive T cells entering barrier compartments were licensed for parenchymal entry by local antigen recognition¹⁷. This licensing model prompted a reinterpretation of classical experiments showing long-term survival of foreign tissues grafted into the brain parenchyma. Even these classical immunoprivilege experiments demonstrated that prior exposure to the graft tissue in the periphery allowed for rapid rejection of the graft in the brain¹⁸, demonstrating the ability of antigen-experienced lymphocytes to enter and function in the brain. The immune privilege of the brain is thus relative and not absolute. The brain entry of T_reg cells, as self-reactive T cells, activated in lymphoid organs and capable of recognizing self-antigen in the barrier compartments, is therefore entirely compatible with the observed gated brain immunoprivilege.

Although not impenetrable, the barriers separating the brain from peripheral organs are highly sophisticated and multi-faceted¹⁹. Despite this, there are multiple potential migration routes allowing blood-borne cells to enter brain tissue (Fig. 2a). First, at the blood–brain barrier, the parenchyma of the brain is walled off from the perivascular space and brain interfaces by the glia limitans, an additional layer of basement membrane that is supported by astrocytic end-feet. At the brain surface barrier, this glia limitans borders the meningeal layers (dura mater, arachnoid mater and pia mater), which separate the brain from the skull. Analogous structures are present at the other key interface sites, the choroid plexus and cribriform plate. At the blood–cerebrospinal fluid (CSF) barrier, ependymal cells separate the brain parenchyma and the CSF, whereas epithelial cells form a barrier between the CSF and the choroid plexus vasculature. Finally, at the cribriform plate a thinned border structure lines the olfactory neuron bundles that project through the nasal lymphatics. Each of these barrier surfaces can be used for entry by lymphocytes under some circumstances; however, as different subsets of T cells have differential capacity to cross brain barrier surfaces in vitro²⁰, the dominant route used by T_reg cells in vivo remains unknown. Furthermore, pathological contexts will change the expression of molecular determinants of adhesion and extravasation, likely changing the relative importance of each route. Regardless of the route, a small number of T_reg cells cross over these barriers and are found within the perivascular spaces and barrier compartments (the meninges, choroid plexus/CSF and cribriform plate), and, to a lesser extent, the brain parenchyma¹¹.

Fig. 2: Potential migration routes for brain regulatory T cells.

a, Four barrier surfaces act as the interface between the brain and the peripheral immune system. These barrier surfaces act as immunological niches, able to host immune cells, and as potential routes for migration further into the brain. b, The blood–brain barrier provides an interface between the brain and circulation. It is formed by tight junctions between endothelial cells lining cerebral capillaries, pericytes and astrocytic end-feet that form the glia limitans. To enter via this route, regulatory T cells (T_reg cells) would need to extravasate and reach the perivascular space (1) and then access the brain parenchyma by crossing the glia limitans (2). T_reg cells can activate mechanisms of tissue retention and be recruited to inflamed or tumour sites (3). c, The choroid plexus represents the blood–cerebrospinal fluid (CSF) barrier. The choroid plexus consists of an epithelial layer equipped with tight junctions, surrounding a stromal interface that contains fenestrated blood vessels. The ependymal cells are at the interface between the brain parenchyma and the ventricles that are filled with CSF. To enter via this route, T_reg cells would need to translocate the choroid plexus epithelium or circumvent via the basal fold (4) and complete their multistep transmigration from the CSF into the brain parenchyma (5). d, At the surface of the brain, the meninges surround the brain and the spinal cord. The meninges are represented by three layers, the dura mater, the arachnoid mater and the pia mater. The dura mater is characterized by the presence of lymphatic vessels and sinuses. T_reg cells in the dural lymphatics are drained into the deep cervical lymph nodes (6). The dural sinuses collect blood from cerebral veins. It has been proposed that sinuses drain CSF from the arachnoid villi originating from the subarachnoid space, where most of the meningeal T_reg cells reside, although this role has recently been questioned. Direct channels link the bone marrow of the skull to the surface of the arachnoid barrier, providing a potential, but unproven, route for T_reg cell entry (7). e, The cribriform plate barrier provides a direct portal from the nasal cavity to the olfactory bulb within the brain for T_reg cells to potentially use. Lymphatic vessels run through the cribriform plate along olfactory cranial nerves penetrating the cribriform plate. During inflammation, T cells originating from the mouse nasal-associated lymphoid tissue migrate along sensory axons towards the olfactory bulb (8). Lymphatic vessels near the cribriform plate can drain T_reg cells from the CSF (9). f, Candidate molecular mediators of brain T_reg migration across the different barrier compartments, with the numbers (1–9) corresponding to the labels in panels b–e. T_reg cells are likely to engage different receptor–ligand pairs depending on the route of entry or exit, and these are likely to change between homeostasis and pathological conditions. CCR2, CC-chemokine receptor 2; CXCR4, CXC-chemokine receptor 4; GR, glucocorticoid receptor; TCR, T cell receptor.

It is likely that entry of T_reg cells into the brain is a multistep process. The relative paucity of resting naive-like T_reg cells, the behaviour of TCR transgenic T_reg cells¹¹ and entry rate modelling of parabiosis data¹⁰ suggest that an activation stage is required prior to brain entry, presumably within secondary lymphoid organs. Likewise, the activation of the Nur77 reporter for TCR signal engagement within the CD69⁺ phenotypically tissue-resident fraction of brain T_reg cells suggests that secondary in situ antigen recognition occurs and may drive phenotypic differentiation and prolong the dwell time of T_reg cells in the brain. The exact site of this secondary engagement with antigen is unknown; however, it is reasonable to consider that the engagement would need to occur anatomically close to the entry site given the highly limited dwell time of the CD69⁻ fraction of T_reg cells¹⁰. The barrier compartments are thus critical for brain T_reg cell entry, being both the site of initial translocation and the probable site for the secondary stimulation likely to be required for extended residency and further migration into the parenchyma. The entry process is likely to differ depending on the entry route, with highly distinct physical and molecular structures present, but insufficient evidence is currently available.

Molecular candidates for T_reg cell entry

The molecular mediators of brain T_reg cell migration across the different barrier compartments remain largely unknown in the homeostatic state. In the absence of direct experimental evidence for the routes and mechanisms of T_reg cell entry, many of the candidate genes for entry via the potential barrier compartments need to be derived from expression patterns of key molecules, be extrapolated from the migration of other leukocyte subsets or be based on migration under different pathological contexts.

Despite being the primary interface of the brain, remarkably little is known about the extent and mediators of lymphocytes crossing from the vasculature directly across the blood–brain barrier under homeostasis (Fig. 2b). Under pathological conditions such as stroke or injury, the damaged vasculature and interface is a major route for lymphocyte entry. CC-chemokine receptor 2 (CCR2), CCR5, CCR7, CXC-chemokine receptor 4 (CXCR4) and sphingosine-1-phosphate receptor are chemotactic receptors that are likely involved at early stages of this migration^{21,22,23,24,25}, with CCR6 and CCR8 gaining importance at later stages following the upregulation of their chemokine ligands by activated glia¹².

The blood–CSF barrier (Fig. 2c) has been proposed as a plausible entry point for T cells into the brain, supported by the presence of T cells in the CSF even in healthy individuals^26,27. Across the blood–CSF barrier, T_reg cells entering from the blood need to cross the choroid plexus epithelial barrier. It has been proposed that immune cells can exit the stroma at the base of the choroid plexus where it folds out from the ventricular wall, without passing any endothelial or epithelial outer brain barrier^28,29. During inflammation, this transmigration may be enhanced by ICAM1-mediated and VCAM1-mediated processes. In rodents, these adhesion molecules are expressed on the apical side of choroid plexus epithelial cells³⁰ and upregulated during EAE³¹. Notably, the number of leukocytes in the CSF in mice correlates with the levels of leukocyte-trafficking molecules expressed (specifically, the levels of ICAM1, CXCL10 and CCL2) in the choroid plexus epithelium, suggesting that the adhesion of inflammatory T cells to the apical side of the epithelial surface increases the leakiness of the tight junctions, enhancing the flow of leukocytes from the choroid plexus stroma to the CSF. Also involved as a negative regulator of migration, although via an unknown mechanism, is glucocorticoid receptor (GR), with knockdown of this receptor on the choroid plexus enhancing the influx of T_reg cells^32,33. In EAE, infiltrating pro-inflammatory T helper 17 (T_H17) cells use CCR6 to follow a gradient of the chemokine CCL20 produced by the choroid plexus³³. This allows the inflammatory cells to rapidly distribute into the CSF. As T_reg cells also express CCR6, conservation of this migratory pathway makes it a candidate for T_reg cell entry during inflammation. Migration into the brain parenchyma would then require a subsequent translocation, from the CSF across the ependymal layer at the ventricular wall, or the glia limitans at the subarachnoid space, similar to what occurs at the brain surface (discussed below).

At the brain surface barrier (Fig. 2d), where the meninges surround the brain, lymphocytes are mostly found in the dura mater layer, just below the bone of the skull. This dura mater layer is, however, further separated from the brain by the arachnoid barrier, which may serve as a substantive barrier for lymphocyte migration into the deeper brain structures. Indeed, functional impairment of the dural lymphatic vessels has surprisingly little effect on neuroinflammation outcomes, and the dura is less responsive to neuroinflammation than proximal brain surfaces³⁴. The leptomeninges/subarachnoid space, by contrast, is more closely aligned with neuroinflammatory events, and inflammation of these regions has a prominent role in neuropathology, for example in the development of cerebral cortical grey matter pathology during progressive multiple sclerosis³⁵. In terms of mechanistic molecules, meningeal T cells, and in particular meningeal T_reg cells, use CCR7 to egress from the tissues²²; however, the molecules that mediate T_reg cell trafficking from the meninges to the brain parenchyma remain unknown³⁴.

Lymphocyte migration across the cribriform plate (Fig. 2e) is poorly understood. Because of discontinuity of the arachnoid membrane near the cribriform plate³⁶, this represents a potentially unrestricted access route to the CSF. Although largely speculative, axonal breaches may provide a potential route for T_reg cell migration through the cribriform plate. Entry of T_reg cells from the olfactory mucosa to the olfactory bulbs is therefore, potentially, a migration route into other brain regions. Such a route is likely to be of increased importance during inflammation, when inflammatory mediators such as cytokines can directly and indirectly increase blood–brain barrier permeability³⁷. In addition, during inflammation, meningeal lymphatics near the cribriform plate acquire a phenotype that is permissive for crosstalk with plasmacytoid dendritic cells and T cells. This phenotype includes the upregulation of antigen presentation molecules, expression of PDL1 and expression of podoplanin on the lymphatics³⁶, although whether this pathway is shared between conventional T cells and T_reg cells remains to be determined.

Finally, it should be noted that the development of a brain tumour alters the entry dynamics of brain T_reg cells, driving large accumulations of this otherwise rare cell type. CCL2, CCL22 and CCL28 are upregulated in the tumour environment, with selective involvement of CCR4, CCR8 and CCR10 chemokine receptors for T_reg cell recruitment^38,39,40,41. The anatomical route for the initial brain entry in these cases is largely unknown.

These different barrier compartments provide a diverse set of potential routes for T_reg cells into the brain, although not all have yet been directly demonstrated and additional routes may yet remain to be discovered. Multiple candidates for molecular mediators of T_reg cell passage along these routes exist (Fig. 2f), although it should be emphasized that knowledge here is particularly incomplete, and reliant on mechanisms observed in related cell types. Systematic evaluation of T_reg cell migration into these compartments, and the influence of pathology and T_reg cell activation status on these events, remains a key gap in knowledge for exploration and potential exploitation.

Residency

Although referred to as tissue-resident, the limited data available for brain T_reg cells suggest that their residency is highly transitory in nature. Whereas myeloid cells can remain resident in the mouse brain for the life span of the cell, with certain myeloid lineages remaining resident for the entire life span of the organism, T_reg cells are only retained in the mouse brain for a period of weeks. Parabiosis studies in mice show rapid host replacement of brain T_reg cells during steady-state conditions, with even the phenotypically resident T_reg cell population (that is, those cells expressing CD69) having a dwell time of only ~3–4 weeks on average^10,11.

The elevated rate of steady-state replacement is consistent with either tissue egress — such as migration of T_reg cells through the draining lymphatics of the brain⁴² — or with high apoptosis rates, consistent with very low local concentrations of IL-2 in the brain, a rate-limiting survival factor for brain T_reg cells⁴³. Even when brain IL-2 levels were synthetically supplemented to match those found in other tissues, parabiosis experiments demonstrated unchanged replacement turnover rates of T_reg cells from the brain⁴³, suggesting that migration is independent of the number of T_reg cells that could be supported by the relevant survival factors present in the brain. Currently, no parabiotic experiments have been performed dissecting all of the barrier compartments, and thus diversity in residency times is possible across these regions. Whereas retention rates for T_reg cells in the brain may change in the settings of neuroinflammation and neuropathology, the data currently available are most consistent with a transient residency of T_reg cells in the brain that is driven by continual turnover. This dynamic process may play a role in shaping the functional diversity of the brain-resident T_reg cell population.

Brain T_reg cell functions

Immunological functions of brain T_reg cells

The tightly regulated presence, and highly modified phenotype, of brain T_reg cells suggests that these cells have biological functions within brain tissue. Of the known functions of T_reg cells outside the brain, the prototypical role is that of suppression of adaptive immune responses. Brain T_reg cells express all of the identified T_reg cell-associated immunosuppressive molecules; indeed, many of these molecules are expressed at higher levels by brain T_reg cells than by circulatory T_reg cells. It is therefore axiomatic that brain T_reg cells should be capable of controlling the same type of adaptive immune responses controlled in the periphery. These classical anti-inflammatory and immunoregulatory roles of brain T_reg cells are now well established in a range of inflammatory settings, in particular in models of EAE, and have been reviewed in depth elsewhere^44,45. In the homeostatic state it remains an open question as to whether brain T_reg cells control local adaptive immune responses. Although consistent with the known functions, the homeostatic numbers of brain-resident T_reg cells and brain-resident conventional T cells are both so low that it is plausible that meaningful interactions are rare enough to render such effects negligible. Analysis of the role of brain-resident T_reg cells in controlling local adaptive immunity is made more difficult due to the potent role of peripheral T_reg cells in controlling peripheral priming of adaptive immune cells that later migrate to the brain.

In addition to the suppression of adaptive immunity, brain-resident T_reg cells may mediate neuroprotection by suppressing inflammatory gliosis. Both microgliosis and astrogliosis have been implicated in neuroinflammatory damage. Although microglia activation can be reparative, providing for the clearance of apoptotic bodies and debris, excessive activation can result in the production of harmful mediators⁴⁶. Likewise, the initial response of astrocytes to inflammation has been observed to have beneficial effects, whereas prolonged astrogliosis drives damaging inflammation⁴⁷ and leaves behind a glial scar that physically interferes with axonal growth⁴⁸. The ability of brain T_reg cells to influence the balance between harmful and reparative glial phenotypes is likely to be a key aspect of neuroprotection⁴⁹. In a mouse model of Parkinson disease, adoptive transfer of polyclonally activated T_reg cells reduced dopaminergic neuronal loss in vivo⁵⁰. T_reg cell-mediated neuroprotection was associated with reduced microglial neurotoxicity and increased levels of brain-derived neurotrophic factor (BDNF) and glial cell line-derived neurotrophic factor (GDNF), which are likely produced by astrocytes, demonstrating that T_reg cell administration resulted in phenotypic shifts in both glial populations⁵⁰. T_reg cells have also been shown to inhibit neurotoxic, and promote neuroprotective, microglial phenotypes in an in vivo mouse model of amyotrophic lateral sclerosis⁵¹. In a similar manner, T_reg cells are protective in mouse models of stroke, with IL-10 production implicated⁵², AREG shown to be a potential modulator of astrogliosis¹² and osteopontin involved in reparative microglial changes⁵³. In a model of traumatic brain injury, osteopontin was also implicated, although in that case the elevated production of osteopontin was indirect, being produced by microglia in T_reg cell-rich environments⁴³. Together these data suggest that brain T_reg cells have a rich network of anti-inflammatory effects, including direct effects on astrocytes and microglia, that polarize these cells away from destructive inflammatory states, and indirect amplification of these effects through glial-produced effector molecules.

Reparative functions of brain regulatory T cells

Beyond the immunological functions of brain-resident T_reg cells, there are emerging data suggesting that these cells have a reparative role. Although difficult to untangle from the anti-inflammatory properties described above, there is growing evidence that brain T_reg cells can enhance the survival of neurons following toxic exposures, can drive the regenerative process of remyelination following demyelination and can activate reparative angiogenesis and neurogenesis.

Direct neuroprotective effects of T_reg cells have been demonstrated in an in vitro model of neurotoxicity of ventral mesencephalic neurons. Co-culture of primary neurons with activated T_reg cells resulted in reduced neuronal cell death following treatment with the neurotoxin 1-methyl-4-phenylpyridinium, an agent used to model the neurotoxicity that occurs in Parkinson disease⁵⁴. Neuroprotection in this model was contact-mediated and involved interaction between CD47 on T_reg cells and signal regulatory protein-α (SIRPα) expressed on ventral mesencephalic neurons, which led to activation of the RAC1–AKT signalling pathway in the neurons⁵⁴. These results suggest that the neuroprotection observed following T_reg cell transfer in a Parkinson disease model may not be purely due to anti-inflammatory properties, but may also be due to T_reg cells directly mitigating neurotoxic effects⁵⁵.

One of the most prominent regenerative roles of T_reg cells in the CNS is their support of myelin regeneration. This may be, in part, mediated by their roles in resolving pro-inflammatory, tissue-damaging immune responses and enabling the phagocytic process that is a prerequisite for myelin regeneration⁵⁶. In addition, however, brain T_reg cells directly signal to oligodendrocyte progenitor cells to differentiate into myelin-producing oligodendrocytes, a crucial step in the process of remyelination⁵⁷. Once differentiated, mature oligodendrocytes will engage with demyelinated axons and re-ensheath these axons with myelin to restore neurological function and neuroprotection⁵⁸. This capacity of T_reg cells to drive oligodendrocyte differentiation and remyelination has been demonstrated directly in animal models of toxin-induced⁵⁷, autoimmune⁵⁹ and viral-induced demyelination⁶⁰, as well as in stroke⁵³. CCN3 production is important⁵⁷ but not essential⁶¹ for this process, with additional molecular mediators potentially including ITGA2 and MCAM influencing oligodendrocyte precursors⁶², and osteopontin working via microglial reprogramming⁵³. As IGF1 and oncostatin M are also expressed by brain T_reg cells⁵³ and aid oligodendrocyte precursor differentiation^63,64, they are potential candidates in the process too. This resulting re-ensheathment of neuronal axons may, in turn, aid protection from inflammation, as remyelinated axons are less susceptible to neurodegeneration than demyelinated axons^65,66.

Finally, a role has been reported for brain T_reg cells in reparative neurogenesis and angiogenesis. In a mouse model of stroke, T_reg cells promote neuronal stem cell proliferation, aiding neurogenesis⁶⁷. This process is reported to be mediated by IL-10 (ref. ⁶⁷) and galectin 1 (ref. ⁶⁸), with BDNF another key candidate⁶⁹. Although evidence at this stage is scarce, T_reg cells are associated with preservation of the blood–brain barrier and repair of damaged vasculature. This may be through inhibition of neutrophil⁷⁰ and microglial cell⁷¹ activation, yet effector molecules secreted by brain T_reg cells, including IL-10, AREG and galectin 1, are also known to support vascular remodelling^72,73, suggesting that a more direct effect may be occurring. Although requiring further exploration, both enhanced neurogenesis and vascular repair are likely non-immunological benefits provided by brain T_reg cells.

These results demonstrate the neuroprotective capacity of brain T_reg cells through multiple modalities: preventing adaptive immune activation, dampening innate inflammatory responses among resident glia, direct neuroprotection and enhancement of regenerative processes. In many cases, however, the relative contribution of these neuroprotective pathways remains completely unknown. For example, the protective capacities of brain T_reg cells in ameliorating cognitive decline in ageing mice⁷⁴ could be driven through their blockade of harmful CD8⁺ T cell responses, reprogramming of activated microglia and astrocytes, direct protection of neurons from harmful inflammatory mediators or an improved regenerative pathway⁷⁵. As primary correction of one aspect is likely to also drive secondary improvements in other read-outs, untangling the causative mode of neuroprotection for brain T_reg cells will remain problematic.

Potential limitations to the reparative functions of brain regulatory T cells

Of note, the majority of studies of T_reg cells in vascular and neural regeneration focus on T_reg cell function in acute phases of the response and at relatively short timeframes after insult. These phases generally feature increased blood–brain barrier permeability and, often, glial scars have not yet formed, facilitating recruitment of T_reg cells to damaged tissue. Less attention has been paid to later stages of disease pathogenesis when blood–brain barrier function is restored and glial scars may have developed, potentially impeding the recruitment, access and bioactivity of T_reg cells into damaged tissue. Whether T_reg cells can still drive tissue regeneration in these settings is not well understood.

Harnessing brain regulatory T cells for advanced therapeutics

Amplification of brain regulatory T cells to reverse neuroinflammation

Exploiting the anti-inflammatory, pro-repair modalities of brain T_reg cells has been proposed in a plethora of conditions — including injuries (traumatic brain injury), vascular conditions (stroke), autoimmune diseases (multiple sclerosis), neurodegenerative conditions (Alzheimer disease, Parkinson disease and amyotrophic lateral sclerosis), psychiatric disorders (depression) and the cognitive decline observed in ageing (Fig. 3a). There are two basic modalities of T_reg cell therapy that are currently the focus of most attention — cell therapy (Fig. 3b) and IL-2 therapy (Fig. 3c).

Fig. 3: Harnessing brain regulatory T cells to treat neuroinflammation.

Potential beneficial effects of regulatory T (T_reg) cell-centric therapies could span neuroinflammatory injuries such as traumatic brain injury and stroke, autoimmune diseases such as multiple sclerosis, neurodegenerative conditions ranging from Alzheimer disease to Parkinson disease and amyotrophic lateral sclerosis, psychiatric disorders such as depression and the cognitive decline experienced in normal ageing. a, There are currently three main T_reg cell cellular products developed for cell therapy: polyclonal T_reg cells isolated from blood and expanded in vitro using anti-CD3/CD28 antibodies and IL-2 (left); astrocytes used to stimulate T_reg cells from the recipient in vitro in the presence of IL-33 and serotonin to confer part of the properties of brain-derived T_reg cells (expression of ST2 and 5-HT₇) (middle); and expansion of polyclonal engineered T_reg cells expressing chimeric antigen receptor (CAR) or artificial T cell receptor (TCR) that recognizes a target antigen (right). b, Mechanism of T_reg cell therapy. T_reg cells infused via the blood enter the brain. The cell therapy is proposed to reduce pathology by exerting immunosuppressive effects on effector T cells, repolarizing microglia and astrocytes, and enhanced repair processes. Key mediators are likely to vary based on pathology, but may include AREG, IL-10, osteopontin and TGFβ. c, Main modalities of IL-2 therapy. IL-2 as a biologic can be delivered systemically in low doses or can be intrathecally injected in a slow release carrier. T_reg cells are highly sensitive to IL-2, accumulating in number (top). Gene delivery systems, such as adeno-associated viral vector (AAV), are capable of crossing an intact blood–brain barrier and delivering the DNA needed for local production in IL-2. Accumulated brain T_reg cells rewire the brain into an anti-inflammatory environment (bottom).

Cell therapy for neuroinflammation

Cellular therapies with T_reg cells have been carried out in both animal models and clinical trials across peripheral organ transplantation and autoimmunity. The basic premise of T_reg cell therapy is that the use of autologous purification, in vitro expansion and, finally, infusion back into the circulation can increase the total number of T_reg cells present in the system, reducing inflammation⁷⁶. Although such approaches can temporarily boost T_reg cell numbers, the tight homeostasis control over the total number of T_reg cells in the system⁷⁷ suggests that a polyclonal T_reg cell therapy will be best suited for neuroinflammatory conditions where a short-term immunological suppression can lead to long-term clinical benefit. Potential routes to circumvent this limitation include the use of antigen-specific T_reg cell expansion, where polyclonal displacement will enrich the repertoire for desired T_reg cell TCR specificities, and genomic engineering of T_reg cells, imparting biological properties for improved retention, expansion or function (Fig. 3b). In each case, the desired outcome will be an enrichment of brain T_reg cells, driving pro-repair and anti-inflammatory processes (Fig. 3c). The latter two approaches have the theoretical benefit of fewer off-target systemic effects, as the approaches are intended to impart brain-specific homing or accumulation.

Two examples serve to demonstrate the potential approaches of advanced T_reg cell therapies. In Parkinson disease, preclinical testing has built upon standard T_reg cell transfer⁵⁰ to create an ex vivo T_reg cell induction approach that enriches the repertoire for antigen-specific brain T_reg cells⁷⁸. By stimulating naive T_reg cells from the spleen in the presence of primary astrocytes, IL-33 and serotonin, the resulting expanded population had transcriptional profiles closer to those observed in bona fide brain T_reg cells. This enriched population demonstrated preferential brain entry and more effective control of neurodegeneration in models of Parkinson disease⁷⁸. In principle, a single enrichment protocol such as this could have broad utility across any neuroinflammatory condition. Whether this protocol can be replicated through co-culture with other brain cells, and the degree to which it mimics the physiological process of in situ differentiation, remains unknown. In multiple sclerosis, one small clinical trial was conducted to examine the potential of T_reg cell-based therapy. This phase Ib/IIa clinical trial involved patients with relapsing–remitting multiple sclerosis. Among these participants, 11 individuals received T_reg cells that were expanded ex vivo and administered intravenously, whereas 3 individuals received freshly isolated T_reg cells delivered intrathecally. None of the participants experienced adverse effects following the treatment. Intriguingly, whereas participants who received T_reg cells intravenously experienced disease relapses and increased disability, this was not observed in the three participants who received T_reg cells intrathecally⁷⁹. Although the trial had a limited number of participants, these findings suggest that in individuals with multiple sclerosis it may be the local effects of brain-resident T_reg cells on neuroinflammation that are most potent, and therapeutic approaches may need to focus on boosting this tissue-resident population rather than the systemic circulating population.

Relevant to multiple sclerosis, additional studies in the preclinical EAE model have taken an alternative route of genome engineering to drive brain localization of T_reg cells. T_reg cell transfer experiments had long been known to suppress EAE severity⁸⁰, although determining whether this was due to brain residency impeding local inflammation or lymph node residency impairing peripheral priming of autoreactive cells has proven difficult. Antigen-specific T_reg cells, such as TCR transgenic T_reg cells, are more efficient suppressors of EAE. Whereas therapeutic use of TCR transgenesis is still in early clinical development, the use of chimeric antigen receptors (CARs) allows T cells to be imparted with a given antigen specificity, using an antibody fused to the TCR signalling apparatus⁸¹. CARs can therefore be used to impart monoclonal-like properties on a polyclonal repertoire, and can be used to create CAR T_reg cells either by starting with a polyclonal T_reg cell population or by coupling CAR expression with forced expression of FOXP3. Franssen et al. took the latter route, creating myelin oligodendrocyte glycoprotein (MOG)-specific CAR T_reg cells⁸². These cells were highly efficient at entering the brain after intranasal delivery, and impeded the neuroinflammatory EAE response⁸². A similar approach can transform T_reg cells isolated from patients with multiple sclerosis, imparting beneficial properties in EAE⁸³. Although CAR T_reg cells face substantial challenges in terms of scalability and cost, the genomic engineering step potentially enables the incorporation of additional design advantages. For example, CAR T_reg cells could be equipped with adhesion molecules to improve brain entry, supplemented via the IL-2 pathway to improve longevity, engineered to enhance their production of effector molecules or supplemented through the addition of biosensor pathways to drive activation at inflammatory sites⁸⁴. Due to these advantages, although the applications may be niche, CAR T_reg cells, or the related recombinant TCR T_reg cells, are likely to become part of the brain T_reg cell therapeutic toolbox.

IL-2 therapy for neuroinflammation

In the biologic space, the most developed therapies for enhancing T_reg cell-mediated effects are the IL-2 therapies. IL-2 is a key cytokine for T_reg cell homeostasis, setting the cellular niche size through control of proliferation and apoptosis⁸⁵. The provision of exogenous low-dose IL-2 is safe and well-characterized to drive an expansion of T_reg cell numbers, and there is emerging consensus on the utility of multiple low-dose IL-2 treatment for autoimmune diseases⁸⁶. A potential limitation of this approach is that the delivery of IL-2 systemically drives an expansion of T_reg cell populations in the peripheral blood and secondary lymphoid organs, which, although likely to be efficacious in the setting of systemic autoimmunity, may have limited benefits in neuroinflammation (in line with the data on intrathecal versus peripheral T_reg cell therapy trials). Nonetheless, a randomized clinical trial (NCT02424396) of low-dose IL-2 in multiple sclerosis has reported promising results, with peripheral T_reg cell activation and expansion⁸⁷. Despite the trial not being powered for efficacy, a trend towards improved clinical outcomes was also observed, where new active brain lesions showed a non-significant tendency to be reduced in patients treated with low-dose IL-2 (ref. ⁸⁷). These results have inspired a broad investigation of the effects of IL-2 in diseases with an immune component, even beyond the field of autoimmunity. In mouse models, low-dose IL-2 or IL-2–anti-IL-2 antibody complexes show promising results in Alzheimer disease⁸⁸, amyotrophic lateral sclerosis^88,89, migraine⁹⁰ and depression⁹¹, among other neurological conditions. This treatment has proceeded to clinical trials for Alzheimer disease (NCT05821153) and amyotrophic lateral sclerosis (NCT04055623, NCT02059759, NCT03039673, NCT03241784). The amyotrophic lateral sclerosis results have been promising to date. In a trial combining IL-2 with T_reg cell infusion, six out of eight patients showed slowed disease progression⁹². Another small study in patients with amyotrophic lateral sclerosis explored a combination of T_reg cells and IL-2. T_reg cells expanded ex vivo were administered intravenously concomitantly with low-dose IL-2 injection subcutaneously. The expanded autologous T_reg cell therapy suppressed oxidative stress responses and stabilized the clinical status⁹³. Finally, the MIROCALS trial, using low-dose IL-2 in a large cohort of patients with amyotrophic lateral sclerosis, demonstrated safety and a small decrease in death among the treated group with less severe diseases⁹⁴. The development of engineered forms of IL-2, with improved T_reg cell-promoting properties, may further enhance the prospects of direct IL-2 therapy in the neuro-space.

A further potential for improving IL-2 therapy in neurological applications comes from targeted delivery (Fig. 3c). The systemic administration route is well established for the expansion of circulating T_reg cells. This expansion likely feeds into the brain T_reg cell population via continual recruitment, but diminishing returns will be observed in the brain. Direct delivery of IL-2 into the brain, either through intrathecal or intracerebroventricular injection, allows for brain residency effects to be preferentially gained. Although the half-life of IL-2 is too short for such a delivery of the unmodified biologic to be viable, engineered forms of IL-2 with elevated longevity may be effective⁹⁵. Alternatively, numerous approaches for slow release or delayed production are being developed, including slow release microspheres⁹⁶ or occlusion bodies⁹⁷, lipid-encapsulated IL2 mRNA⁹⁸ and next-generation nanotechnology approaches⁹⁹. One promising form of IL-2 delivery is that of gene delivery. By using viral vectors, such as adeno-associated viral vectors (AAVs), the advantages of systemic delivery can be combined with brain-based production (Fig. 3c). The first such approach was an AAV-encoded IL2 vector, using a blood–brain barrier-crossing AAV capsid. Capable of driving both systemic and brain-based IL-2 production, this vector was beneficial in a mouse model of Alzheimer disease¹⁰⁰. A refinement of this approach encoded astrocyte-specific production, to eliminate off-target systemic effects, with the capacity to include a small molecule-inducible system⁴³. This approach was beneficial in mouse models of traumatic brain injury, stroke and EAE⁴³. Astrocyte-directed gene delivery was also beneficial in reducing the cognitive decline observed in ageing mice, although not in a mouse model of Alzheimer disease⁷⁴, suggesting divergent utility of global and brain-restricted IL2 gene delivery in neuroinflammation. Although the focus here has been on innovations in IL-2 therapy, other cytokines such as IL-33 can regulate the number of T_reg cells in the brain, albeit to a lesser extent, and may show similar utility¹⁰. Furthermore, as more research is performed on brain T_reg cells and their effector molecules are characterized (Fig. 1), many of these therapeutic modalities can be adapted to directly deliver the brain T_reg cell-associated effector molecules, bypassing the need for IL-2 and T_reg cells themselves. Such approaches have further advantages in that they may overcome potential mechanical barriers (such as glial scars) that could limit the entry and/or bioactivity of T_reg cells.

Depletion of brain regulatory T cells in cancer

Although the immunosuppressive role of brain-resident T_reg cells may provide a boon for the treatment of neuroinflammation, the reverse approach is likely to be needed in the cancer context. In the most common forms of primary cancers of the brain — glioblastomas — T_reg cells accumulate to much higher levels than observed in the healthy brain¹⁰¹, with the tumours producing soluble factors capable of enhancing proliferation and recruitment of T_reg cells¹⁰². The degree to which these resident T_reg cells interfere with the spontaneous immunological recognition and clearance of the tumours is debatable, with conflicting evidence on whether T_reg cell infiltration predicts survival outcomes^101,103. Brain T_reg cells, however, do appear to play a key role in the resistance of glioblastoma to immune checkpoint blockade treatment. In mouse models of glioblastoma, tumours are poorly responsive to immune checkpoint blockade involving anti-PD1 antibody treatment, with curtailed expansion of effector CD8⁺ T cells^104,105. Although previously explained by ‘immunological coldness’, these poor outcomes are also explained, in part, by the expansion of brain T_reg cells within the glioblastoma. Targeting of these T_reg cells through systemic treatment with anti-CD25 antibody to deplete circulating T_reg cells and brain-resident T_reg cells¹⁰⁴, or anti-GITR antibody treatment to convert T_reg cells into effector T_H1 cells¹⁰⁵, allows for improved cytotoxic responses and tumour eradication. In the patient context, it is likely that brain T_reg cell depletion will be required in order to improve the disappointing clinical results seen when treating glioblastoma with immune checkpoint blockade. Although numerous different modalities for T_reg cell depletion are being developed, one of the more promising avenues is the use of anti-CD25 antibody. An anti-CD25 antibody¹⁰⁶ has been trialled in patients with glioblastoma, with tolerable toxicity results and efficient T_reg cell depletion observed (NCT04158583). Follow-up clinical trials combining immune checkpoint blockade and T_reg cell depletion will determine whether the approach will be successful. This potential breakthrough in treating previously ‘cold’ brain tumours could be further developed through targeted approaches that deplete brain-resident T_reg cells while leaving the circulating population intact. Although several promising approaches are being developed to target biologics to a particular tissue, in this particular case the blood–brain barrier may work in favour of treatment: local infusion of T_reg cell-depleting agents may end up restricted to the brain, reducing off-target systemic effects. Considering the high mortality and poor therapeutic options for these patients, targeting T_reg cells in brain tumours may be a rapidly evolving field with high potential for therapeutic benefit.

Conclusions

Brain T_reg cells defy many of our preconceptions of brain cells — they are highly plastic, migratory and transient. The population of brain T_reg cells that is resident, temporarily, within the brain is continually replaced by incoming cells, only adopting their brain-based phenotype for a brief sojourn in the tissue. At the same time, the functions of brain T_reg cells defy the norms expected of immune cells. Although they possess the same capacity for classical immune suppression as peripheral T_reg cells, in the unique immunological environment of the brain the most important role of brain T_reg cells may be the unorthodox functions being unearthed: namely, calming down activated glial populations, enhancing the resiliency of neurons and enabling the remyelination function of mature oligodendrocytes. Yet it is these very properties of brain T_reg cells, defying the norms of both brain and peripheral immunity, that make brain T_reg cells enticing as a therapeutic target. As a transient and dynamic population, continually reseeding the brain from the periphery, brain T_reg cells are malleable targets, with the potential to intercede via the circulating population. Likewise, the multi-modality of brain T_reg cells, influencing immunity, gliosis, resilience and repair, provides far greater potential therapeutic gain than a single target treatment could achieve. With the potential to harness brain T_reg cells directly, via cell therapy, or indirectly, via IL-2 therapy, the active preclinical and clinical trial research in this space is likely to deliver novel therapies for a range of neurological pathologies. Finally, brain T_reg cells provide a learning template, as a black box waiting to be opened to identify the molecular mediators that could bypass the need for brain T_reg cells entirely and provide their therapeutic potential via as yet to be identified drugs.