Regulatory T-cell inhibition versus depletion: the right choice in cancer immunotherapy

Abstract

Tumour-induced expansion of regulatory T (T_Reg) cells is an obstacle to successful cancer immunotherapy. The potential benefit of T_Reg-cell depletion through the interleukin-2 receptor is lost by the concurrent elimination of activated effector lymphocytes and possibly by the de novo induction of T_Reg-cell replenishment. In theory, the functional inactivation of T_Reg cells will maintain them at high numbers in tumours and avoid their replenishment from the peripheral lymphocyte pool, which has the capacity to further suppress the effector lymphocyte anti-tumour response.

Main

Since their discovery in the 1960s as suppressive T cells¹, regulatory T (T_Reg) cells have been extensively studied in a wide range of both physiological and pathological conditions in mouse and man. It is now well established that T_Reg cells are a distinct lymphocyte lineage endowed with regulatory properties that affect a variety of immune cells in the innate and adaptive, and priming and effector phases of the immune response^2,3 (Fig. 1). It is now also widely acknowledged that inhibiting T_Reg-cell function in patients with cancer is an essential step if new therapies, especially immunotherapies, are to be broadly successful in the clinic. But how should the function of T_Reg cells be inhibited? Initial investigations have indicated that depleting T_Reg cells from cancer patients might be a valid approach, but more recent, preliminary data has raised the hypothesis that functionally inactivating T_Reg cells might be better. This Opinion explores this hypothesis.

Figure 1: Regulatory T cells in the network of immunosuppression.

Tumour cells produce suppressive factors that directly act on regulatory T (T_Reg) cells or license myeloid-derived suppressor cells, tumour-associated macrophages and dendritic cells to expand T_Reg cells. Such expansion occurs through the proliferation of pre-existing T_Reg cells and the conversion of naive precursors into de novo induced T_Reg cells. Tumour-associated T_Reg cells suppress both the innate and the adaptive immune response, including the priming and effector phase of both CD4 (green) and CD8 (brown) T cells. IDO, indoleamine 2,3-dioxygenase; TGFβ, transforming growth factor β.

Extinguishing the T _Reg cell 'friendly fire'

T_Reg cells are currently classified into two main subsets, according to their origin and suppressive activity⁴. Natural T_Reg cells originate in the thymus by high-affinity interaction of the T-cell receptor (TCR) with antigens expressed in the thymic stroma⁵. They have been classically described to suppress the proliferation of responder T cells in vitro in a contact-dependent, cytokine-independent manner. Natural T_Reg cells constitutively express CD25, cytotoxic T-lymphocyte antigen 4 (CTLA4), glucocorticoid-induced tumour necrosis factor (TNF) receptor (GITR) and OX40, molecules that characterize activated lymphocytes^6,7,8,9,10. However, the high-level expression of the transcription factor FOXP3 (forkhead box p3) is the most distinctive marker for the regulatory lineage, at least in the murine system². Although incompletely characterized to date, the network of natural T_Reg-cell suppression mechanisms includes surface molecules such as CTLA4, membrane-bound transforming growth factor β (TGFβ) and the pericellular generation of adenosine^3,11,12. Adaptive T_Reg cells are antigen-specific suppressive cells that are generated in the presence of interleukin 10 (IL10) or TGFβ in case of T regulatory 1 or T helper 3 subsets, respectively^13,14. In contrast to natural T_Reg cells, adaptive T_Reg cells exert suppression mostly through soluble factors (IL10 and/or TGFβ) and their suppressive function is not strictly associated to high-level FOXP3 expression.

The incompletely understood panoply of T_Reg-cell-derived immunosuppressive factors regulates the response of nearby cells in a dominant fashion. T_Reg cells require TCR triggering to become functional, but once activated they suppress T cells blindly, with no regard to the antigen specificity of targeted cells. This implies that the number and function of T_Reg cells should be tightly controlled to maximize immunity against foreign pathogens and preserve self tolerance. When the number of T_Reg cells increases, the immune response is unbalanced and favours unresponsiveness. This condition is evident in many murine and human cancers up to the critical point at which tumours are no longer attacked by the immune cells. The available data indicate that T_Reg-cell expansion can occur through the proliferation of pre-existing T_Reg cells^15,16, and from the conversion of peripheral naive precursors into newly derived T_Reg cells^17,18 (Table 1). Thus, through mechanisms that still need to be examined in detail, tumours turn the host T_Reg cells from helpful guardians against exacerbated immunity into detrimental suppressors of the anti-tumour immune response.

Table 1 Evidence for proliferation and conversion of T_Reg cells in tumours

Several strategies have been proposed to combine T_Reg-cell neutralization with conventional anti-tumour therapies (Table 2). This might be achieved by targeting T_Reg-cell development, frequency, migration or activity. The choice of the most suitable approach will depend on the type and stage of tumour, because the mechanisms of immune suppression and the susceptibility to different treatments can vary.

Table 2 Seminal works of T_Reg-cell targeted therapies

Physiological T _Reg -cell homeostasis

T_Reg cells account for 5–10% of the circulating CD4⁺ T-cell population. In naive mice, the vast majority of peripheral T_Reg cells are of thymic origin, as suggested by the observation that the TCR repertoire of peripheral and thymic T_Reg cells displays a 90% overlap¹⁹. However, which antigens are recognized by the TCR repertoire of T_Reg cells is still poorly understood. Several reports support the idea that natural T_Reg cells have a diverse TCR repertoire that significantly differs from conventional naive T cells and preferentially recognizes self-antigens^19,20,21, in keeping with their exposure to antigens expressed in the thymic stroma⁵. Self-reactive regulatory thymic emigrants are maintained in the periphery through continuous stimulation by self-peptides expressed at a low level by antigen-presenting cells²². Co-stimulatory molecules, such as B7.1 (also known as CD80) and CD40, are required for peripheral T_Reg-cell maintenance^23,24, and interleukin 2 (IL2) is crucial for T_Reg-cell homeostasis; IL2-null or IL2 receptor-null mice develop multiorgan fatal autoimmunity that can be reversed by reconstitution with functional T_Reg cells or by exogenous IL2 (Ref. 25). IL2 also restores T_Reg-cell number in CD40-deficient mice²³. Throughout life, the major role of naturally arising T_Reg cells is to avoid autoimmunity that is generated from fortuitous activation of autoreactive T-cell clones that have escaped negative selection (deletion of high affinity self-reactive immature T cells) in the thymus. Indeed, both neonatal and adult mice in which T_Reg cells are acutely ablated succumb to generalized autoimmunity²⁶.

During life, exposure to neo-antigens might enrich the TCR repertoire of T_Reg cells by conversion, a process in which mature, conventional, non-regulatory CD4⁺ T cells acquire the regulatory phenotype and gain suppressive functions following peripheral stimulation under certain conditions²⁷. Adoptive transfer of congenic, TCR transgenic or fluorochrome-labelled non-regulatory CD4⁺ T cells enables their phenotypic changes to be tracked in recipient mice. Such changes include expression of FOXP3 and suppressive functions. These experiments also enable the characterization of the conditions that drive conversion, such as suboptimal antigen stimulation in combination with TGFβ²⁸. TGFβ and retinoic acid-mediated conversion has an important role in establishing intestinal tolerance to exogenous antigens^29,30,31. Moreover, recent findings support the notion that peripheral conversion is involved in establishing local immune privilege in transplanted organs³². Therefore, whereas natural T_Reg cells are centrally induced against the predetermined self-antigen repertoire, T_Reg cells that developed in the periphery derive from the encounter with any available cognate antigen that is presented in a sub-immunogenic context, regardless of its origin, thus allowing flexible adaptation of host tolerance to external challenges.

Thymus involution occurs late in the mouse lifetime but early, at puberty, in man. This implies that, contrary to mice, the human peripheral pool of T_Reg cells should decrease with age, unless continuously replenished by peripheral conversion. This has been suggested to occur in elderly people whose effector memory cells turn into memory T_Reg cells³³. This finding predicts that, during the lifespan, the T_Reg-cell repertoire shifts from the centrally determined self recognition to the peripherally acquired memory of foreign antigens. Whether this hypothesis has evolutionary significance is still unclear.

The tumour perturbs T _Reg -cell homeostasis

One of the distinctive features of all living organisms is the ability to react to external stimuli while preserving the stability of the internal environment. This homeostatic property results from continuous adjustments to the dynamic equilibrium, within tolerable limits, between host and foreign challenges. On this basis, the increased level of T_Reg cells that occurs in the presence of tumours, could be interpreted as the highest level compatible with prolonged tumour–host interaction, allowing the tumour to escape the immune attack but preserving the host from exaggerated and indiscriminate immune suppression. Unfortunately, the benefit of such an equilibrium seems temporary for the host, who eventually succumbs to the tumour burden³⁴.

The mechanisms driving T_Reg-cell expansion in patients with cancer are not fully understood³⁵. Early studies by Robert North demonstrated that, during tumour progression, the initial development of concomitant immunity is followed, and finally subverted, by the establishment of CD4⁺ T-cell-mediated immune suppression³⁶. This state of hyporesponsiveness is maintained for up to 28 days after the surgical removal of the primary tumour, as is the persistence of the T-cell population that is responsible for the suppression³⁷. This dual-phase model suggested that effector and suppressor subpopulations might be functionally predominant at different stages of tumour progression. A more recent study in mice with and without tumours has correlated immune status with predominant lymphocyte subsets in lymph nodes. The initial increase of activated effector T cells is followed by a rapid decline that coincides with the increase in the number of T_Reg cells³⁸. In this model, the subversion of immunity into suppression occurs at the time when T_Reg cells outnumber effector T cells.

The increase in T_Reg cells is not unlimited and never exceeds 50% of the CD4⁺ population^{17,39,40,41,42,43,44,45}. Once established, the upper limit of tumour-induced T_Reg cells remains constant during the late phase of tumour progression. This has been observed in gastric, colorectal and ovarian carcinomas in which the most prominent increase in T_Reg cells occurs during the early stages of tumour progression and remains almost unchanged at later stages^42,44. Furthermore, T_Reg-cell alterations in peripheral blood of head and neck cancer patients do not normalize in the absence of evident disease after therapy⁴⁵, suggesting that post-operative systemic immune suppression might contribute to disease relapse and metastasis.

In addition to analyses that examined overall numbers of T_Reg cells, several studies have shown that T_Reg-cell accumulation and compartmentalization can vary among different types of human cancer. In patients with ovarian cancer, T_Reg cells progressively accumulate within the tumour mass and in the ascitic fluid, and decrease in the draining lymph nodes. T_Reg cells are recruited to the tumour by CCL22, which is produced by tumour cells and tumour-associated macrophages⁴². Accumulation of T_Reg cells at the tumour site might increase suppression locally when T_Reg-cell expansion has reached the highest possible level. In breast, pancreatic and colorectal tumours, regional lymph nodes remain heavily infiltrated by T_Reg cells^41,43. However, despite variability among tumour types, some common mechanisms involved in T_Reg-cell amplification can be identified, as described in the following sections.

Proliferation of pre-existing T_Reg cells. The vast majority of tumour antigens come from self-molecules and so it is conceivable that such an antigenic assortment will preferentially activate natural T_Reg cells rather than effector T cells. On activation, T_Reg cells can proliferate while gaining suppressive activity. Tumour-induced T_Reg-cell proliferation has been reported in several experimental settings^15,16,18,46 (Table 1). In the case of influenza haemagglutinin (HA) used as surrogate tumour antigen, T_Reg cells are the main proliferating cell fraction in the total CD4⁺ T-cell population expressing HA-specific TCR when transferred into mice with A20-HA expressing mouse lymphoma¹⁶. In addition, when mixed with CD4⁺CD25⁻ (the naive CD4⁺, non-regulatory T cell, population) from the same TCR-transgenic mice, T_Reg cells become the prevalent population on transfer into tumour-bearing mice¹⁸. These experiments clearly show that the transferred T_Reg cells proliferate in preference to naive cells in tumour bearing hosts. Which signals drive T_Reg-cell proliferation in the tumour setting? IL2 is unlikely because few activated effectors or mature dendritic cells can provide it in the context of general immune suppression. Rather, TGFβ is the cytokine that is thought to foster T_Reg-cell amplification. Both tumour cells directly⁴⁷ or 'tumour-educated' immune cells can locally produce large amounts of TGFβ (Fig. 1). Some mouse and rat tumours actively induce myeloid immature dendritic cells to secrete TGFβ and this promotes T_Reg-cellproliferation¹⁵. There is also substantial evidence that indicates the involvement of TGFβ in T_Reg-cell conversion.

De novo T_Reg-cell induction. The pool of T_Reg cells in tumour bearers includes newly generated T_Reg cells in addition to proliferating natural T_Reg cells (Table 1). Conversion, described above in physiological contexts, is enforced in the presence of tumours such that the size of the pool of T_Reg cells is increased at the expense of potential effector T lymphocytes. Evidence for tumour driven conversion comes from many mouse and human studies. For example, when HA-specific CD4⁺CD25⁻ T cells are transferred into tumour-bearing mice, new T_Reg cells arise with fully suppressive capability^16,18. In addition, polyclonal CD4⁺CD25⁻ T cells that are purified from non TCR-transgenic naive mice convert into T_Reg cells in the lymphoid organs of tumour-bearing mice¹⁷. In this setting, if antigen encounter is required for conversion, the naive T cells should recognize the few tumour-specific non-self molecules or the more abundant tumour-associated self-antigens. The latter possibility suggests that autoreactive clones, that have escaped thymic deletion, are the most likely T_Reg-cell precursors in tumour-induced peripheral conversion. As in physiological conditions, conversion does not seem to require proliferation, as lymphocytes from mice treated with doses of vinblastine that induce cytostasis convert into T_Reg cells if transferred into tumour-bearing mice¹⁷. This observation implies that targeting T_Reg-cell proliferation by cyclophosphamide (see below) might be ineffective because conversion would become the dominant mechanism of T_Reg-cell expansion.

In addition to TGFβ, a suboptimal level of antigen stimulation favours peripheral conversion²⁸. These two conditions are concomitant in the tumour: antigen presentation is impaired, tumour antigens are poorly immunogenic and TGFβ is constantly available. Another molecule that induces T_Reg-cell conversion is indoleamine 2,3-dioxygenase (IDO), an enzyme that degrades tryptophan leading to tryptophan starvation and the accumulation of kynurenines, which are toxic catabolites of tryptophan (Fig. 1). Tolerogenic dendritic cells, such as the plasmacytoid subset, can show IDO enzymatic activity, thus promoting T_Reg-cell conversion⁴⁸. Tumour cells can also directly degrade tryptophan by IDO expression. For instance, the T_Reg-cell conversion that is induced in vitro and in vivo by A20 lymphoma cells is abrogated in the presence of an IDO inhibitor⁴9. Moreover, IDO expression by human leukaemic cells induces T_Reg-cell conversion in vitro and correlates with the increase in the number of T_Reg cells in the peripheral blood of patients with acute myeloid leukaemia⁴⁹. Therefore, different molecular pathways can contribute to the generation of a suppressive network in which T cells preferentially convert into T_Reg cells rather than effector T cells.

Less is known about the mechanisms of T_Reg-cell expansion in cancer patients. Indirect evidence supports the hypothesis of peripheral conversion for T_Reg-cell expansion. The most expanded T_Reg-cell subset in patients with multiple myeloma has the naive rather than the memory phenotype that characterizes pre-existing memory T_Reg cells or effector/memory T cells⁵⁰. Furthermore, there is evidence that naive T_Reg cells that have expanded in cancer patients are not recent thymic emigrants, but cells that have already proliferated and colonized the periphery before conversion⁵⁰.

T_Reg-cell conversion versus proliferation. In tumour bearers, conversion and proliferation are mutually independent processes. Indeed, T cells do not require proliferation to convert¹⁷ and proliferation of natural T_Reg cells does not limit or increase the rate of conversion¹⁸. The relative contribution of conversion to reach and maintain the upper homeostatic level of T_Reg cells in tumour bearers becomes evident when T_Reg cells are depleted. In this case converted T_Reg cells replace most of the pre-existing natural T_Reg cells, a process with no obvious benefit for the host. In terms of the TCR repertoire, it is conceivable that, beside tumour antigens, any other available antigen, from normal tissues or pathogens, could guide the cognate lymphocytes through conversion into T_Reg cells rather than into effector T cells. This allows the peripheral generation of T_Reg cells with a wide TCR repertoire that increases the likelihood of encountering the antigen activating their suppressive activity. Also, conversion redirects T-cell precursors recognizing tumour antigens into T_Reg cells rather than into T effectors (Fig. 2). In addition, T_Reg cells that are derived from conversion contain a higher fraction of cells with CD62L^lowCD103⁺ effector or memory phenotype than pre-existing T_Reg cells¹⁷. This subset has recently been shown to rapidly migrate to inflammatory sites and to suppress in vitro proliferation without further TCR triggering, thus potentially representing a ready-to-suppress population⁵¹. Because of the above reasons, depletion of tumour-expanded T_Reg cells could reduce the host immune response further (Fig. 2). If this is shown to be the case in at least some patients with tumours, depletion-based therapies might then need to be replaced by alternative strategies aimed at the functional inactivation of T_Reg cells.

Figure 2: Repertoire of regulatory T cells before and after CD25⁺ cell depletion.

In naive conditions, natural regulatory T (T_Reg) cells (round) display an anti-self T-cell receptor (TCR) repertoire (orange). The tumour microenvironment induces the expansion of pre-existing T_Reg cells and the conversion of naive T cells (oval) into T_Reg cells, which might have a non-self (green) or self-reactive (pink) TCR. When CD25⁺ T cells are depleted, conversion accounts for most of the regenerated T_Reg cells, which now show a wide TCR repertoire.

Depleting T _Reg cells: fewer pros than cons

The alpha-subunit of the IL2 receptor CD25, the first surface marker identified on T_Reg cells, was soon considered a promising target for T_Reg-cell neutralization in cancer immunotherapy. The availability of the anti-CD25 monoclonal antibody, PC61, enabled the effects of T_Reg-cell depletion to be tested in several murine cancer models^{39,52,53,54,55,56} (Table 2). Despite some efficacy^{39,52,53,54,55,56}, intrinsic limitations apply when PC61 is used to treat established tumours. Time-course experiments in two models showed that the efficacy of PC61 is lost as tumours progress^40,54. These observations suggest that, in the naive context before tumour challenge, the only CD25⁺ tumour-specific cells that can be ablated are natural T_Reg cells. Under these conditions, immunogenic tumours are rejected when transplanted into mice. However, in mice with established tumours, T_Reg cells and activated effector T cells that express CD25 are depleted by PC61. Therefore, targeting CD25⁺ cells in situations where the tumour is well established might be ineffective if both regulatory and effector lymphocytes are eliminated^40,54. Interestingly, following the injection of anti-CD25 antibody into tumour-free mice, natural FOXP3⁺CD4⁺ T cells do not seem to be completely depleted, but instead are maintained at low levels for up to 6 weeks⁴⁰. In the presence of a growing tumour, however, depletion seems relatively short with T_Reg cells being replaced in less than 2 weeks¹⁷. Depleting T_Reg cells also raises the possibility of autoimmunity. However, different doses, schedules and experimental settings of PC61 administration indicate that the generation of autoimmunity is variable^40,53.

Monoclonal antibodies to human CD25 that are available for clinical use, such as daclizumab, block IL2 and receptor interaction, thus they are used to treat haematological malignancies⁵⁷. However, none of these antibodies has been tested for their efficacy in the depletion of T_Reg cells, and daclizumab has been reported to maintain T_Reg-cell function in recipients of heart transplants⁵⁸. So far, most studies in humans have used the immunotoxin denileukin diftitox (Ontak), a fusion protein between the IL2 and the diphtheria toxin, to selectively kill lymphocytes expressing the IL2 receptor. The in vivo anti-tumour efficacy is still under preclinical and clinical investigation, and discrepant results have been reported so far. The open issues are the extent to which denileukin diftitox depletes T_Reg cells and its therapeutic efficacy in treating different types of tumour based on the administration schedule^59,60,61,62 (Table 2). A recent observation of autoimmune retinopathy indicates potential localized adverse effects⁶³. Currently, we can speculate that, similar to PC61, denileukin diftitox might eliminate both effector T cells and T_Reg cells, and this might be followed by peripheral conversion. Moreover, it is possible that triggering of the IL2 receptor might occur through the IL2 portion of the fusion protein⁶⁴ thus activating rather than depleting T_Reg cells. Cyclophosphamide is an alkylating agent that profoundly impairs lymphocyte proliferation. Early studies have shown that suppressor lymphocytes from tumour-bearing hosts are susceptible to this drug⁶⁵. However, single high-dose administration might result in only a transient T_Reg-cell depletion. This might be corrected by low-dose metronomic administration, which constantly targets T_Reg cells as they are induced to proliferate^66,67,68. In conclusion, the peripheral generation of T_Reg cells has profound implications for choosing the right therapeutic option.

Alternative ways to target T _Reg cells

Inhibition of tumour-specific T_Reg-cell expansion. The above discussion suggests that T_Reg-cell elimination might, in some circumstances, be a relatively blind approach that is unable to distinguish between normal and pathological mechanisms of T_Reg-cell development. Such approaches could potentially leave uncontrolled autoreactive effectors that would damage normal tissues and tumour tissues alike. In theory, the selective targeting of tumour-associated T_Reg-cell expansion would preserve the physiological T_Reg-cell context.

This goal could be achieved by inhibiting, for instance, the IDO pathway (Fig. 3). Several preclinical data confirm that the administration of an IDO inhibitor significantly decreases the rate of peripheral conversion and dramatically impairs tumour growth^49,69,70.

Figure 3: Different strategies to neutralize regulatory T cells and their possible outcomes.

Depletion of CD25⁺ cells results in the elimination of both regulatory T (T_Reg) cells (orange) and activated effector T cells (CD4 green, CD8 brown). Following T_Reg-cell depletion, conversion rescues a state of immune tolerance. This could be avoided by targeting mechanisms that are involved in conversion, such as indoleamine 2,3-dioxygenase (IDO) and transforming growth factor β (TGFβ). TGFβ blockade could also inhibit T_Reg-cell proliferation, without affecting pre-existing T_Reg-cells. Alternatively, T_Reg-cell functional inhibition inactivates their suppression (white) without inducing a new wave of conversion, thus leading to effective tumour rejection. TLR8, Toll-like receptor 8.

Another possible target is TGFβ, involved in both proliferation and conversion of T_Reg cells in tumour bearers (Fig. 3). Mice that are genetically engineered to express a dominant negative form of the TGFβ receptor II on lymphocytes show reduced, if not absent, growth of several transplanted tumours^15,71, owing to insensitivity of effector T cells to TGFβ suppression and impairment of peripheral T_Reg-cell generation. A similar effect is seen with TGFβ silencing in tumour cells, which also enables unimpaired triggering of the immune response and impairs tumour engraftment⁷².

A certain degree of specificity in targeting tumour-associated T_Reg cells can be obtained by confining the treatment to the tumour. Intratumour injection of a monoclonal antibody that depletes CD4⁺ lymphocytes induces rejection of late-stage tumours⁷³. Also, the intratumour administration of an antibody triggering GITR eliminates incipient tumours, in the absence of autoimmunity, but this effect seems to be the result of stimulating effector T cells rather than inhibiting T_Reg-cell function^74,75.

An additional possibility is the inhibition of T_Reg-cell recruitment to the tumour, an approach that has been proposed but not translated into practice yet⁴².

T_Reg-cell functional inactivation. Our discussion so far suggests that inhibiting the function of T_Reg cells so as to prevent T_Reg-cell conversion would be a viable approach (Table 2). The goal is to reach a situation in which inactive yet physically present T_Reg cells would still be 'sensed' and 'counted' by the homeostatic system, which although altered in the presence of a tumour still confines T_Reg-cell expansion to an upper limit (as described above). In theory, T_Reg-cell inhibition should avoid the new wave of peripherally converted T_Reg cells (Fig. 3).

T_Reg cells constitutively express CTLA4 and GITR. CTLA4 blockade or GITR triggering have been shown to reverse immune suppression as a result of T_Reg-cell function both in vitro and in vivo^7,8. However, recent data indicate that a reinterpretation of these initial findings might be required. Although, in experimental inflammatory bowel disease, CTLA4 blockade directly abolishes T_Reg-cell suppression⁷⁶, in tumours the same treatment is only partially dependent on T_Reg-cell inhibition, and is more active if coupled with T_Reg-cell depletion⁷⁷. CTLA4 blockade renders effector T cells resistant to T_Reg-cell suppression, preserves the T_Reg-cell intrinsic suppressive capacity and expands the pool of T_Reg cells⁷⁸. The mechanisms that mediate these effects are unknown. Similarly, the direct inhibition of T_Reg cells contributes only minimally to the anti-tumour efficacy of anti-GITR antibodies in vivo. Concomitant antibody-mediated GITR triggering has potent adjuvant effects during a tumour vaccination protocol, even if T_Reg cells have been previously depleted by PC61 (Ref. 75). Unexpectedly, GITR stimulation even expands the population of T_Reg cells within the tumour-draining lymph nodes⁷⁵. In summary, neither CTLA4 nor GITR seem the right target to strike at T_Reg-cell function in cancer immunotherapy.

Instead, the direct stimulation of Toll-like receptor 8 (TLR8) completely reverses T_Reg-cell-mediated suppression of human cells. Indeed, T_Reg cells that are pre-treated with the TLR8 ligand poli-G10 no longer inhibit the in vivo CD8⁺ response to a human tumour cell line that is grown in immunodeficient mice⁷⁹. This finding indicates that stimulators of the innate immune response, like TLR8 agonists, might concomitantly impair T_Reg-cell function.

Murine T_Reg cells constitutively express OX40, a co-stimulatory molecule belonging to the TNF receptor family. Triggering OX40 completely inhibits the suppressive function of T_Reg cells in vitro^9,10 and abolishes their ability to protect mice from experimental graft versus host disease in vivo⁹. This distinctive property of OX40, coupled with its well-known co-stimulatory function, fostering memory T-cell survival and recall response⁸⁰, supports the use of OX40-triggering molecules or antibodies in cancer immunotherapy. Accordingly, prevention and successful treatment of murine tumours by an agonistic antibody to OX40 seem to be mainly through the inhibition of T_Reg-cell-mediated suppression (S.P. and M.P.C., unpublished data).

Finally, the recent discovery that T_Reg cells possess ectonucleotidase activity that mediates immune suppression through the generation of pericellular immunosuppressive adenosine^11,12, indicates that ectoenzyme inhibitors or receptor antagonists, which are able to interrupt the adenosinergic axis, might be a promising therapeutic strategy for tumours. This follows from the observation that genetic or pharmacological inactivation of the A2A adenosine receptor rescues endogenous anti-tumour T-cell responses and induces successful tumour rejection⁸¹.

Conclusions and future perspectives

By inducing T_Reg-cell expansion, the tumour takes advantage of the inhibitory function that these cells exert on all the immune components. We suggest that avoiding the physical elimination of T_Reg cells would be potentially useful. This approach should prevent the induction of a new wave of peripherally converted T_Reg cells that are endowed with a wide TCR repertoire that will easily encounter a cognate antigen resulting in their activation. Conversion would also redirect potential effector T cells toward the T_Reg-cell phenotype. Alternatively, T_Reg-cell inactivation is a suitable strategy, which would functionally impair T_Reg-cell suppression without changing the TCR repertoire of the expanded T_Reg-cell population. Triggering of TLR8 or OX40, and potentially blocking adenosine, might improve the chances of neutralizing T_Reg-cell immunosuppression in cancer immunotherapy.