Targeting Regulatory T Cells in Cancer

Optimism has been renewed that many cancers can be cured or forestalled by immune-based therapies, used either alone or as part of multimodal programs. This optimism originates from an improved understanding of tumor immune interactions and the availability of gene, cell, and ligand-based technologies that promote effector antitumor responses. Most tumors develop in the face of normal immune function, and antitumor responses of varying strength result. A strong immune response against the primary tumor is associated with clearance and induced dormancy of metastatic cancer cells, with a resulting enhanced prognosis. Conversely, global immune deficiencies secondary to disease or therapy are associated with an increased frequency, earlier recurrence, more rapid progression of tumors, and poorer prognosis. Responses to chemotherapy and oncolytic virotherapy may, in part, be immune determined, and evidence is persuasive that an intact immune system, specifically determined by CD4⁺ T cells, is required for sustained tumor regression following oncogene inactivation therapies (1).

Adaptive antitumor immune responses are durable, tumor antigen specific, and acquired through the integrated intercellular responses of the innate and adaptive immune systems (Fig. 1, center; ref. 2). Tumor-infiltrating T cells, especially CD8⁺ CTLs and IFN-γ–secreting CD4⁺ (Th1) cells, are central to effective immune containment. Adaptive immune responses are initiated when cells of the innate immune system [natural killer T (NKT), γδ T, NK, and macrophages] are recruited to the tumor microenvironment; the continued process of tumor remodeling results in the shedding of cancer cells and debris with a consequent induction of inflammatory signals. The production of IFN-γ (initially from NK and NKT cells) seems critical because it creates a positive feedback loop by inducing some tumor cell death, the further activation of NK cells and macrophages, and the production of chemokines and cytokines, which are also tumoricidal and antiangiogenic. Immature dendritic cells (DC) are activated following uptake of tumor debris and/or antigens and migrate to the regional lymph nodes where they present the tumor antigens to naive T cells, which can differentiate into Th1, Th2, Th17, or regulatory T cells (T_regs), depending on the cytokine environment. Th1 cells can license DCs to induce tumor-specific CTLs by cross-presentation of antigen on MHC class I. Antigen-specific CD8⁺ T cells traffic to the tumor, where cell-mediated killing of tumor cells is augmented by Th1- and Th17-derived cytokines. However, these effector responses can be inhibited by T_regs, induced by or recruited to the growing tumor (3).

Immune evasion is a hallmark of cancer, resulting from both passive and active tolerizing conditions that subvert antitumor immune responses (4). Passive tolerization may result from downregulation of MHC class I expression on the tumor cells and/or low antigenicity secondary to immune editing and selective cell growth. Other tolerizing mechanisms involve inhibition of immune cells in the tumor domain by depletion of tryptophan by the enzyme indoleamine 2,3-dioxygenase (IDO). Active tolerization involves suppression of antitumor cell-mediated responses by tumor-infiltrating T_regs and myeloid-derived suppressor cells (5).

Overall, T_regs are considered the most powerful inhibitors of antitumor immunity and the greatest barrier to successful immunotherapy (6). In the early stages of cancer, T_regs are concentrated in the tumor mass, resulting in concomitant immunity, whereby the primary tumor can progress because of local inhibition of effector immune responses, but metastatic cells are eliminated by uninhibited systemic antitumor immune responses. In advanced-stage disease or for poorly immunogenic cancers, the number of T_regs is increased systemically, and there is an absence of concomitant antitumor immunity (7). Although a correlation between increased T_reg number and survival, either negative or positive, remains equivocal, the ratio of T_reg to effector T cells (T_effector) in the tumor mass seems to have greater prognostic significance (8).

T_regs can be classified into a number of subtypes (8), including natural CD4⁺ T_regs (nT_reg), which originate in the thymus, express CD25, FOXP3, CTLA-4, LAG3, and GITR, and suppress innate and adaptive immune cells. Induced CD4⁺ T_regs (iT_reg) control immune responses to tissue antigens, including tumor antigens, and include CD4⁺ nT_reg–like, Tr1, and Th3 cells that suppress through production of interleukin (IL)–10 and TGF-β. The iT_regs develop in the periphery following engagement of the T-cell receptor (TCR) of naïve T cells and under the influence of innate IL-10 and TGF-β. Their cell-surface markers are often indistinguishable from those of nT_regs, and they differ principally in their mechanism of suppression. Although less well characterized, there are also populations of natural and induced CD8⁺ T_regs.

Although the field is still in its infancy, evidence is emerging that inhibition of T_regs may help in tumor containment, especially when combined with appropriate immunotherapies that activate effector T cells. Systemic T_reg depletion in patients induced regression of melanoma metastases (9), and in mice, when combined with immunogene stimulation of intratumoral immune effector cells, resulted in cure of 90% of animals who had large and weakly immunogenic sarcomas (10). The clinical objective will be to provide sustained reduction of T_reg function, particularly in the tumor environment, allowing enhancement of antitumor effector functions and with minimal risk of developing systemic autoimmune diseases.

Depletion strategies are not specific to the T-cell subset, but they have a selective advantage when the T_reg accumulation provides functional dominance in the tumor environment (Fig. 1A). T_reg-depletion strategies have focused on monoclonal antibodies or ligand-directed toxins targeted to cell-surface receptors, such as CD25. Daclizumab and basiliximab are anti-CD25 antibodies that invoke cell death by cytokine deprivation (IL-2) and by triggering antibody-dependent cell-mediated cytotoxicity or complement-dependent cytotoxicity. Results from an ongoing clinical trial have shown that daclizumab reduces T_regs and, thereby, enhances CTL responses to tumor antigen induced by vaccination (11).

Denileukin diftitox is a fusion protein of human IL-2 and the enzymatically active and membrane-translocating domains of diphtheria toxin. After binding to CD25 and internalization, release of the toxin is cytocidal. Clinical data on the use of denileukin diftitox for alternative indications have led to its application for CD25 targeting of T_regs and the emergence of similar CD25-targeted immunotoxins LMB-2 and RFT5-SMPT-dgA. With one exception, denileukin diftitox depleted T_reg numbers, albeit transiently, with T_reg nadirs persisting for less than 3 weeks (11). The T_reg elimination was mirrored by a concomitant increase in the prevalence of IFN-γ⁺CD3⁺ T cells in the blood and de novo appearance of melanoma antigen-specific CD8⁺ T cells (9). However, the clinical benefits were modest. Regression of melanoma metastases in 5 out of 16 patients represents the most promising outcome (9). Consistency of response is an issue, as 2 patients who developed antigen-specific T cells failed to show any tumor regression, and another study in patients with melanoma failed to yield a single objective clinical response (11).

Denileukin diftitox is the subject of numerous clinical trials, but to date, it fails to realize its clinical promise. Because CD25 is expressed on activated T_effector cells, denileukin diftitox may also restrain protective antitumor immune responses. Denileukin diftitox transiently depleted various T subsets, including tumor antigen-specific CD8⁺ T cells (9). Its indiscriminate effects on CD4⁺CD25⁻ cells are difficult to rationalize.

Low-dose oral metronomic cyclophosphamide induced a profound, selective reduction in T_regs and restored T- and NK-cell function in patients with advanced cancer (12). This restoration of T- and NK-cell function invoked temporary disease stabilization in a number of patients without clinical improvement. The mechanism underpinning its selective toxicity toward T_regs is unexplained. Metronomic cyclophosphamide also has antiangiogenic and direct cytotoxic effects that contribute to tumor stabilization or shrinkage.

Depleting T_regs may have further consequences aside from an unintended treatment-mediated elimination of activated T_effector cells (13). Their depletion leads to an increase in tumor-mediated T_effector to T_reg conversion, with a diminution in antitumor immune responses, which does not seem to occur with the other T_reg modulation approaches.

Similar to CD25, CTLA-4 is not exclusively expressed on T_regs; it is also found on activated CD4⁺ and CD8⁺ T cells (Fig. 1B; ref. 8). CTLA-4 inhibits antigen priming of T_effectors by competing with CD28 for the costimulation of CD80/CD86 on antigen-presenting cells (APC). Furthermore, it induces IDO in DCs (14). The consequent depletion of tryptophan and production of tryptophan metabolites, such as kynurenines and picolinic acid, inhibit T_effector proliferation and function. The anti–CTLA-4 antibodies ipilimumab (MDX-010) and tremelimumab (CP-675206) are currently undergoing clinical evaluation. Ipilimumab, as monotherapy or in combination with peptide vaccination, improved survival in patients with previously treated metastatic melanoma (15). Tremelimumab promotes antitumor responses, but recently these have been shown to result from T_effector activation rather than T_reg modulation (16). This finding may also be true of ipilimumab, because its clinical mode of action has yet to be fully defined and could be ascribed to direct effects on T_regs, or T_effectors, or on both.

The glucocorticoid-induced TNF receptor (GITR) is constitutively expressed on T_regs and at lower levels on activated T_effectors. Intratumoral injection of an agonistic antibody to GITR (DTA-1) invoked potent antitumor immunity and eradicated established tumors in mice (17). The exact mechanism by which this approach achieves its effects is controversial. One study showed that the benefit of DTA-1 was T_reg mediated, facilitated by their selective modulation (18). However, a more recent study suggested T_effector costimulation as the predominant outcome (19). Regardless of mechanism of action, GITR approaches have yet to recapitulate these promising findings in humans. Receptor activator of NF-κB (RANK) ligand (RANKL) expression on T_regs engages the RANK receptor on cancer cells and promotes metastases (20). Inhibitors of RANK signaling, such as the anti-RANKL antibody denosumab, already used against osteoclastic-mediated bone resorption, may block direct T_reg-induced metastases of certain cancers.

Targeting FOXP3, the essential transcription factor of T_regs, by RNA interference (RNAi) could also modulate their function. Lentiviral-mediated delivery of miR-31 (a negative regulator of FOXP3) to T_regs abolished their suppressor capability (21). Translation to clinical application is challenging, as miR-31 would need to be delivered specifically to T_regs, because FOXP3 is also transiently expressed on activated human T_effectors. FOXP3 is also expressed (both mRNA and protein) in numerous cancer cell lines (22), but the effects of its downregulation are unknown and could even be counterproductive.

Further options for disrupting T_reg function include Toll-like receptor (TLR) modulation, OX40 stimulation, or interference with the adenosinergic pathway. Exposure of T_regs to the TLR8 ligand poli-G10 abolished their suppressive influence on CD8⁺ T cells, leading to improved antitumor immunity (23). More recently, a synthetic TLR1/TLR2 agonist, an analogue of bacterial lipoprotein, mediated a dose-dependent tumor regression and a long-lasting protective response against tumor rechallenge through a reciprocal downregulation of T_regs and upregulation of CTL function (24). These findings suggest that TLR signaling is a worthwhile pursuit, but caution is advised because TLR agonists can promote regulatory as well as effector responses (25). Stimulation of OX40 (a costimulatory member of the TNF receptor family) inhibits the suppressive function of T_regsin vitro (by downregulation of FOXP3) and abolishes protection against graft-versus-host disease in mice (13). The paradoxical stimulatory effects on T_effectors make it an enticing target for cancer immunotherapy. Another potential target on T_regs is ectonucleotidase activity, which facilitates local generation of adenosine, which has immunosuppressive capability. Ectoenzyme inhibitors, such as ARL67156, and other modulators of the adenosinergic pathway, such as inhibitors of the A2A adenosine receptor, have been shown to block T_reg-induced immunosuppression (26).

Chemokine–chemokine receptor and integrin–integrin ligand interactions attract T_regs to the tumor, a phenomenon first observed for the CCL22–CCR4 interaction in ovarian cancer (Fig. 1C; ref. 27). Importantly, CCL22 expression was not confined to tumor cells but also included bystander cells, such as tumor-associated macrophages. Further chemokines and/or integrins have been implicated in the selective recruitment and retention of T_regs at tumor sites, including CXCR4, CD103, and CCR2 (8). Because chemoattraction is ubiquitous in the immune system, efforts to block T_reg recruitment to the tumor mass may be limited by the concurrent effects on T_effectors. Nevertheless, disruption of CCR5/CCL5 signaling blocks T_reg migration to tumors and inhibits pancreatic tumor growth in mice (28). Methyl gallate has also recently been shown to inhibit infiltration of T_regs into tumors, resulting in reduced tumor growth and prolonged survival rates (29).

Immunostimulatory therapies may inadvertently promote tumoral homing of T_regs. Therapy with IL-2 can enhance CCR4 expression on T_regs, which stimulates their migration to the tumor mass and an upregulation of CXCR4, the receptor for CXCL12, a chemokine linked to development of organ-specific metastases (8). These findings endorse a more prudent use of IL-2 or perhaps its use in combination with agents such as AMD-3100, which antagonize the CXCR4–CXCL12 interaction.

The origins of iT_regs within the tumor microenvironment are diverse, because varying degrees of plasticity exist within the helper CD4⁺ T-cell population (T_Regs, Th1, Th2, Th17, Tfh; Fig. 1D; ref. 30). Predifferentiated T_regs may migrate under the influence of chemokines (27), and T_regs may arise from de novo generation via differentiation and expansion or may derive from conversion of CD4⁺CD25⁻ T cells. The plasticity inherent in each of these processes is a potentially exploitable therapeutic niche.

IL-6 is central to T-cell plasticity (30). It helps to convert FOXP3⁺ T_regs into IL-17–secreting T cells (Th17). It potently abolishes conversion of conventional T cells into iT_regs, and in its absence, no other cytokine can substitute for this inhibition. Thus, IL-6 merits further investigation as a therapeutic for cancer. TGF-β acts at the axis between T_reg and Th17 differentiation, enhancing the function of FOXP3 and inhibiting the function of RORγt, their essential transcription factors, respectively. TGF-β–induced FOXP3 expression is inhibited by proinflammatory cytokines (e.g., IL-6 and IL-21) in a STAT3-dependent manner. Thus, STAT3 may also represent a therapeutic option; indeed, forced expression of STAT3 augmented IL-17 production, most likely through increased RORγt expression (30). Redirecting differentiation toward a Th17 phenotype might also be achieved by direct introduction of RORγt, as this has been shown to induce IL-17 expression upon transduction of naive CD4⁺ T cells (30). Conversely, selective methylation at the FOXP3 locus would likely hinder differentiation along a suppressor pathway. Aside from the epigenetic level, targeting FOXP3 at the mRNA and protein levels would also be worthwhile. Other approaches include antagonists for retinoic acid receptors that facilitate differentiation into Th17 cells over T_regs (30). T_reg differentiation can be redirected toward lineages other than Th17. Specific inactivation of the transcription factor IFN regulatory factor 4 elevates Th2 cytokine production, whereas IL-4–driven growth factor independent 1 facilitates optimal Th2 differentiation (30).

Blocking T_reg proliferation is an obvious goal, which can be achieved by direct inhibition of TGF-β, inhibition of IDO directly with 1-methyl-d-tryptophan, or inhibition of IDO indirectly by CTLA-4 blockade. Aside from directly stimulating T_reg expansion, COX-2–derived PGE₂ facilitates tolerogenic APC-led T_reg recruitment and is itself a functional instrument of T_regs in certain tumors (8). Thus, use of COX-2 inhibitors like celecoxib may be justified. Alternatively, bevacizumab or blockade of PD1-L on T_regs with MDX-1106 (phase II) may halt T_reg proliferation.

Inhibiting the peripheral conversion of CD4⁺CD25⁻ T cells into CD4⁺CD25⁺ T_regs may be a useful therapeutic approach. The TGF-β–blocking antibody 1D11 abolished this conversion and reduced tumor burden in mice. Subsequently, other TGF-β modulators, including antibodies, soluble TGF-β receptors, and the antisense oligonucleotide AP-12009, have reached phase I and II clinical trials. However, systemic TGF-β blockade may carry the risk of developing autoimmune disorders. Furthermore, under subimmunogenic conditions, T-cell conversion can occur in the absence of TGF-β; IL-10 and IDO have also been shown to promote induction of T_regs (8).

The multitude of strategies discussed in this review deliver only marginal efficacy. Although some strategies have lacked potency, the majority flounder on specificity. This dearth of specificity is understandable, given the intersecting differentiation pathways shared by all cells of the T-cell lineage. Selective approaches to T_reg modulation are warranted. Simple depletion of T_regs may be naïve and the benefit short-lived, whereas inhibiting their migration to the tumor ignores the in situ generation of these cells. Thus, strategies focused on negating T_reg function or reprogramming their functional phenotype would seem more meritorious.

A unique cell-surface marker that facilitates selective targeting of T_regs has yet to be uncovered. Thus, targeting CD25 or CTLA-4 has been encumbered by a concomitant effect on T_effectors. Introducing a second layer of specificity, so-called dual specificity, to receptor targeting would likely be synergistic. This strategy is under investigation in our laboratory, whereby a relatively T_reg-specific gene therapy approach is coupled to ligand selectivity.

A global T_reg modulation is undesirable because it may increase susceptibility to autoimmunity. Tumor T_regs could be targeted via their antigen-specific TCRs; antigen-specific T_regs engaged melanoma-expressed LAGE1 and ARTC1 (31), and in patients with colorectal cancer, carcinoembryonic antigen, telomerase, HER2/neu, and MUC-1 reactive T_regs were detected in the peripheral blood (32). On a practical level, targeting tumor T_regs may be achieved by harnessing tetramer technology; saporin-coupled MHC class I tetramers specifically ablated islet-specific glucose-6-phosphatase catalytic subunit-related protein–autoreactive T cells and delayed diabetes in nonobese diabetic mice (33). Identification of CD4⁺ T_regs specific for a given tumor antigen would facilitate their targeting with MHC class II tetramers by similar means.

Although such agents would be specific for a given subset of T_regs, they would also target other CD4⁺ helper cells expressing the same antigen specificity; CD8⁺ cells would be unaffected. To circumvent this issue, the effector component attached to the tetramer could be modified to confer another level of specificity. It could be miR-31, as 100% of target cells internalize the tetrameric complexes (33). Although the consequence of FOXP3 knockdown in non-T_regs is unknown, TCR engagement in these cells may simply lead to activation, further augmenting the immune effector response.

Alternatively, one could target tumor T_regs indirectly by modulating DC activation. This indirect targeting could be achieved by blockade of DC p38 mitogen-activated protein kinase, COX-2, or phosphoinositide 3-kinase, which inhibit innate production of TGF-β and IL-10 and, thereby, suppress induction of T_regs. Such strategies enhance the efficacy of TLR agonists or HSPs as immunotherapeutics or adjuvants for DC vaccines and permit an unrestrained development of protective Th1 and Th17 cells (25).

T_reg inhibition in the cancer environment would permit an antitumor immune effector competency with containment or elimination of disease. Such responses would be tumor specific and durable and should be effective against systemic disease, particularly micrometastases. T_reg inhibitory strategies have clinical potential as part of multimodal programs or combined with targeted therapies or local immunogene stimulation of antitumor immune effector cells. The objective should be to selectively modulate T_regs within the tumor microenvironment, rather than their global depletion, to minimize the risk of autoimmune manifestations. Strategies targeting T_reg function or differentiation seem currently to be the best option because they are less susceptible to compensatory mechanisms. Emerging technologies, such as tetramer or RNAi approaches, should improve specificity and efficacy and, thus, favor the preferential inhibition of T_regs within the tumor environment.

No potential conflicts of interest were disclosed.

W.L. Byrne is funded by the Health Research Board in Ireland, PhD Scholars Program in Cancer Biology. K.H.G. Mills is funded by Science Foundation Ireland (PI award 06/IN.1/B87). J.A. Lederer is funded by an NIH grant (R01 GM-035633-23). G.C. O'Sullivan is funded by Science Foundation Ireland Research Frontiers Programme (08-RFP-NSC-1655).