Anti-tumor immunity and autoimmunity: a balancing act of regulatory T cells

Abstract

Regulatory T (Treg) cell activity has been observed in anti-tumor and autoimmunity since the 1970s. Functional and molecular characterization of Treg cells has been made possible by the recent association of cell markers, such as CD25, CTLA-4, GITR, and Foxp3 gene product, with immunoregulatory activity. Here the influence of Treg cells in both anti-tumor immunity and autoimmunity was measured in BALB/c mice. Depletion of CD4⁺CD25⁺ Treg cells with CD25 mAb resulted in mammary tumor regression and increased susceptibility to thyroiditis. This in vivo priming to both tumor-associated antigens and self-thyroglobulin attests to the presence of otherwise undetectable immune effectors which are under negative regulation. Modulation of Treg cells is a powerful strategy in cancer therapy, but may potentiate autoimmune complications. Murine models exhibiting breakable tolerance to tumor-associated antigens, such as ErbB-2 (HER-2/neu), and increased susceptibility to autoimmunity following Treg-cell depletion are being established to test new vaccination or therapeutic strategies involving Treg-cell modulation.

Introduction

The quest to control cancer by immunological means has met with only occasional success. There has been much speculation on this inadequacy, and many studies have been designed to delineate this difficulty [5, 15, 44]. For example, tumor cells may lose major histocompatibility antigen class I (MHC I), disabling their recognition by cytotoxic T cells [6, 44]. They may produce immunosuppressive molecules like TGF-β [9, 32] to dampen immune stimulation. Solid tumors may be encapsulated, posing a physical barrier to immune cells [49]. To overcome these obstacles and elicit a functional anti-tumor immunity, a variety of strategies have been proposed. One of the most promising involves the modulation of regulatory T (Treg) cells which may inhibit the induction of anti-tumor immunity.

Regulatory T (Treg) cells in anti-tumor immunity

Immunoregulatory or suppressor activity was observed in many laboratories in the 1970s [1, 3, 10]. Elaborate suppressor pathways were proposed to illustrate how suppressor cells and factors may inhibit anti-tumor immunity [40, 41]. It was suggested that tumor antigens preferentially activate the suppressor pathway [41], and that cyclophosphamide abrogates some, but not all, suppressive mechanisms [40]. At the time, the suppressor cells and factors were characterized mainly by their functions, since they were difficult to isolate and molecular markers were unavailable. Further studies regarding the nature of suppressor cells merely raised questions of their very existence, and the pursuit of suppressor cells was abandoned in most labs. With new molecular markers, suppressor-like or regulatory cells have been thrust into new light. Both myelocytic [27, 59] and lymphocytic regulatory cells have been described. The discussion here will be focused on Treg cells.

Regulatory T (Treg) cells in autoimmunity

Similar to suppressor cells in anti-tumor immunity, suppressor cells in autoimmune diseases were described in the mid-1970s. In both the mouse [16] and rat [28], a low incidence of autoimmune thyroiditis arose spontaneously after postnatal thymectomy and irradiation to deplete putative suppressor T cells. A decade later, cotransfer studies in nude mice showed a T-cell subset capable of suppressing the emergence of autoimmune diseases, such as gastritis, oophoritis, orchitis, as well as thyroiditis [37]. We demonstrated in a murine model of experimental autoimmune thyroiditis (EAT) that regulatory cells were induced in 2–3 days after tolerance induction with soluble mouse thyroglobulin (mTg) and that subsequent induction of EAT was prevented [18]. These regulatory cells were CD4⁺ [19]. The rapid induction of regulatory activity suggested the amplification of an existing Treg population mediating natural tolerance. The recent identification of the CD25 (IL-2 receptor α) marker on cells with regulatory activity has permitted further characterization of Treg cells mediating natural tolerance. Whereas the transfer of CD25^- cells into nude mice resulted in autoimmune diseases, including thyroiditis and gastritis, CD25⁺ cells blocked such development [38]. Similarly, absence of CD4⁺CD25⁺ Treg-cell activity led to autoimmune gastritis [42] and transfer of Treg cells delayed diabetes in CD28-deficient NOD mice [39]. The CD4⁺CD25⁺ T cell was thus recognized as a candidate Treg cell in autoimmunity.

Characterization of Treg cells

Thymus-derived CD25⁺ T cells represent ~10% of peripheral CD4⁺ T cells and encompass a major Treg-cell activity. However, CD25 is expressed on both Treg and effector T cells. In vivo treatment of mice with a CD25 mAb, PC61, depletes Treg-cell activity in naïve mice, but depletes both Treg and effector T cells in immunized mice. Other markers associated with CD4⁺CD25⁺ Treg cells include CTLA-4 [34, 50], CD45RB^low [12, 34], and toll-like receptors 4, 5, 7, and 8 [4]. Recently, two additional markers have shown high correlation with Treg activity. One is the glucocorticoid-induced tumor necrosis factor receptor family–related gene (GITR) [26]. Similar to CD25, elevated GITR is expressed on both Treg and activated T cells [48]. GITR is a 228 a.a., cysteine rich, type 1 transmembrane protein of the TNF family [26] and forms homodimers on the cell surface. A nondepleting GITR mAb, DTA-1, triggers T cell signaling and abolishes suppressive activity without depleting cells [48]. Systemic treatment of BALB/c mice with GITR mAb resulted in spontaneous gastritis [48], indicating the induction of autoimmunity by functionally disabling Treg cells.

The most specific marker of Treg cells to date is scurfin, a transcription factor encoded by forkhead box P3 gene Foxp3 [7, 11, 14, 43]. Scurfin binds to the promoter region of cytokine genes and attenuates the production of activation-induced cytokine, such as IL-2 [43], keeping Treg cells from proliferating when stimulated. When transduced with Foxp3 gene or isolated from Foxp3 transgenic mice, even CD4⁺CD25⁻ T cells exerted suppressive activity [11, 14]. Conversely, Foxp3 knockout mice developed lymphoproliferative disease, similar to Foxp3-mutant Scurfy mice or CTLA-4 [52] deficient mice, further supporting the association of Foxp3 with negative regulation [7]. Overexpression of Foxp3 in CTLA-4 deficient mice delayed lymphoproliferative disease. These findings support CD4⁺CD25⁺Foxp3⁺ cells as a distinct lineage of Treg cells.

Functionally, Treg cells exert suppressive activity in a nonspecific fashion when their TCRs are triggered [51]. Cell-cell contact appears necessary [51], and membrane-bound TGF-β may contribute to the activity [34]. This function may suppress self-reactive T lymphocytes that escape thymic clonal deletion [45, 46] or prevent over-reaction to pathogens [36]. Other T cells may exist and regulate immune function through IL-10, but they may represent cell populations distinct from Foxp3⁺ Treg cells [20].

Treg cell depletion on anti-tumor immunity and the induction of autoimmunity

Treg-like CD4⁺CD25⁺ T cells with suppressive activity have been described in lung [58], pancreas, and breast cancer patients [22], although it is unclear if these cells are of Foxp3 lineage. In mice, liberation from negative regulation by CD25 mAb treatment led to regression of leukemia and fibrosarcoma [47]. In another study, depletion of CD25⁺ Treg cells resulted in lower incidence or slower growth rate of B16F10 tumor [13]. Both CD4⁺ and CD8⁺ T cells contributed to this anti-tumor activity. Priming to B16F10-associated antigens was evidenced by B16F10 tumor rejection in naïve mice which received adoptively transferred immune CD4⁺ T cells. Immune priming was further demonstrated by specific recognition of a self-antigen, tyrosinase, in mice which rejected B16F10 tumor with CD25 mAb treatment.

The amplification of anti-tumor immunity through Treg-cell modulation may be complicated by immune reactivity to non-tumor-associated, self-antigens. In 14 patients with metastatic melanoma, there were two complete and one partial responders following treatment with human mAbMDX-010 to block CTLA-4 engagement to B7, and s.c. vaccination with two modified HLA-A*0201-restricted peptides [29]. In six patients (43%), grade III/IV autoimmune manifestations were observed, including dermatitis, enterocolitis, hepatitis, and hypophysitis. The three patients with objective cancer regression all developed severe autoimmune symptoms requiring treatment. The trial with a two-stage design was intended to accrue 21 patients in the first stage. The accrual was ceased after 14 patients because of these autoimmune complications.

By modulating Treg cells, cancer immunotherapy is encountering an unprecedented opportunity, yet one with a clear and present danger. Although autoimmunity is being handled in patients as it arises, there is an urgent need to analyze in a comprehensive manner the two sides of Treg-cell modulation. New and innovative strategies must be developed to implement cancer control without falling prey to autoimmunity. Toward this end, animal model systems to assess simultaneously anti-tumor and autoimmunity following Treg-cell modulation have to be established.

Results and discussion

Treg-cell depletion on tumor growth in BALB/c mice

Many tumor models have been developed in BALB/c mice. Thus, the effect of Treg-cell depletion on tumor growth was tested in these mice using a D2F2/E2 tumor cell line. D2F2 is a mouse mammary tumor line derived from a spontaneous mammary tumor that arose in a BALB/c hyperplastic alveolar nodule (HAN) line D2, originally induced by prolactin stimulation [23, 57]. To introduce a defined tumor-associated antigen into the cell line, D2F2 cells were transfected with pCMV/E2 which encodes wild-type human ErbB-2 (HER-2) and a stable D2F2/E2 line was generated [57]. Human ErbB-2 (HER-2) is overexpressed in breast, ovarian, and several other cancer types. In stage IV breast cancer patients, therapeutic efficacy of HER-2 mAb, Herceptin, has been demonstrated. HER-2 based vaccines, including the HER-2 DNA vaccine generated in our lab, are being tested in clinical trials. In human ErbB-2 DNA immunized mice, rejection of D2F2/E2 tumor was mediated primarily by T cells, although significant humoral response was induced [31, 33].

BALB/c mice were injected i.p. with CD25 mAb on 2 consecutive days. Depletion of CD4⁺CD25⁺ cells was verified by flow cytometry. On day 5 after the second mAb injection, when CD4⁺CD25⁺GITR⁺ T cells were at a minimum, mice were injected s.c. with 2×10⁵ D2F2/E2 cells. In untreated mice, D2F2/E2 tumors were palpable in 1 to 3 weeks after inoculation, and they grew to ~500 mm³ in 6 weeks, when the mice were sacrificed (Fig. 1). Following Treg-cell depletion, six of eight mice developed palpable tumors in less than 2 weeks, but the tumors started to regress before they were 50 mm³ in volume. By week 4, all tumors regressed completely. The mice were monitored for another 10 weeks without any sign of tumor recurrence. The same tumor regression was observed when mice received unmodified D2F2 tumor cells following CD25 mAb treatment (not shown).

Fig. 1.

Tumor regression following CD4⁺CD25⁺ T-cell deletion. BALB/c mice were injected i.p. with CD25 mAb PC61 (solid triangle) 5 and 6 days before they received s.c. 2×10⁵ D2F2/E2 cells. Control mice were treated with normal mouse sera or PBS (open triangle). There were eight mice each in PC61 treated and control groups

The dramatic increase in anti-tumor immunity was induced by simply depleting Treg cells. This induction of functional immunity to a growing tumor (i.e., in situ priming) challenges the long-standing paradigm that solid tumors are poor at priming the immune system and cannot be rejected effectively by immune cells. Rather, their immunogenicity is thwarted by the regulatory pathway.

There is a finite window in which CD25 mAb can be used to selectively deplete Treg cells because activated T cells also express CD25 and are sensitive to this antibody. It would be detrimental if CD25 mAb is administered after effector T-cell activation. It may be possible to further amplify the striking anti-tumor immunity with reagents like GITR antibody which removes Treg-cell activity without depleting effector cells. These are powerful tools and there is significant concern that such treatment can open the gate to autoimmunity. To assess the extent of this possibility, we have chosen autoimmune thyroiditis as a prototype autoimmune disease to gauge the potential complications.

Thyroiditis and Treg cells

Autoimmune thyroiditis was among the first autoimmune complications recognized in the mid-1970s, following the removal of putative suppressor T cells in rodents [16, 28]. Because experimental autoimmune thyroiditis (EAT) is inducible with a self-antigen, mouse thyroglobulin (mTg), a homologue of human thyroglobulin, and shares well-delineated features with human Hashimoto’s thyroiditis (HT), it has served as a prototype autoimmune disease model [17, 55]. HT is a well-characterized, organ-specific disease with known thyroid antigens. Similar to EAT with susceptibility linked to the mouse MHC class II genes, HT susceptibility is strongly influenced by the human MHC class II genes, such as HLA-DRB1*0301, supplemented by CTLA-4 genes in disease development [53, 54]. Clinically, HT is an easily diagnosed hypothyroid syndrome, characterized by elevated thyroid-stimulating hormone (TSH) and decreased thyroid hormone levels. This syndrome results from destruction of the thyroid gland by mononuclear cell infiltration [56]. Early autoimmune responses can be monitored by autoantibody production and T-cell proliferative response in vitro to thyroid antigens. Although HT progresses chronically, it can be managed by thyroid hormone replacement therapy.

The role of Treg cells in EAT tolerance has long been recognized. Treg activity in EAT can be specifically enhanced by elevating the circulatory mTg level for 2–3 days, either by pretreating susceptible mice with mTg [18], as mentioned above, or by TSH infusion to release endogenous mTg [21]. Subsequent induction of EAT with mTg and adjuvant such as complete Freund’s adjuvant or lipopolysaccharide is markedly suppressed in such tolerized mice. Tolerance is mediated by CD4⁺ Treg cells [19] which exert their suppressive action for >10 weeks [8]. The interval of 2–3 days for TCR engagement with elevated mTg antigen is sufficient for the activation and expansion of Treg cells, such that the animals can withstand EAT induction with mTg plus an adjuvant. It is important to note that immunization with mTg plus proinflammatory cytokines, such as IL-1 [25] or IL-12 [60], was not sufficient to overcome this tolerant state. We have ruled out T_h2-related influence on Treg activity, since IL-4 and IL-10, separately or together, did not participate in mTg-induced tolerance [61].

Treg-cell depletion on thyroiditis

The potential of using EAT as a prototypical autoimmune disease to assess the effect of Treg-cell manipulation was tested by depleting Treg cells in tolerized mice prior to EAT induction. EAT-susceptible CBA mice were used initially. Mice received the tolerogenic regimen of two doses of deaggregated mTg (dmTg). Treg cells were then depleted with two doses of CD25 mAb, 4 days apart. The mice were subsequently challenged with the EAT-inducing regimen of mTg and LPS. Depletion of CD4⁺CD25⁺ T cells abrogated protection against EAT induction; 100% (6/6) of tolerized mice became susceptible to EAT induction with thyroid damage (thyroid infiltration of 10% or more), compared with 17% (1/6) mice in the tolerized, IgG control group (Fig. 2).

Fig. 2.

In vivo depletion of CD4⁺CD25⁺ T cells in mice with established tolerance abrogates protection against EAT. Mice were given 100-μg deaggregated mTg or saline, followed by CD25 or IgG control as shown. EAT was induced with 40-μg mTg + 20-μg LPS, and thyroid pathology was examined on day 28

With the establishment that Treg cells are involved in the immune regulation of EAT-susceptible CBA mice [24], the role of Treg cells was tested in a relatively EAT-resistant BALB/c strain. In a pilot experiment, anti-CD25 treatment prior to EAT induction led to marked thyroid damage in 100% of the mice, compared with mild inflammation in 43% of control mice (data not shown). Therefore, Treg cells are critical in maintaining EAT resistance in BALB/c mice and the effect of Treg-cell modulation on autoimmune thyroiditis can be assessed in these mice.

Tolerance to tumor-associated antigens

The pressing need to modulate Treg cells stems from the observed tolerance to tumor-associated antigens. Although profound anti-tumor immunity to tumors like D2F2/E2, which contains a foreign human ErbB-2 antigen, can be readily induced in BALB/c mice by one-time, transient depletion of Treg cells (Fig. 1), the same may not be true for an overexpressed self-antigen in human cancer. More strenuous Treg-cell modulation may be required to achieve therapeutic effect, creating an even greater risk of autoimmunity [29]. A test system with tolerance to tumor-associated antigen, e.g., ErbB-2, is needed to assess the effect of Treg-cell modulation. BALB NeuT (NeuT) mice are suited for this purpose. NeuT mice express a transforming rat Neu under MMTV promoter and have been backcrossed to BALB/c mice for >12 generations [2]. NeuT females develop atypical hyperplasia, carcinoma in situ, and palpable mammary tumors at ~5, 10, and 17 weeks of age, respectively. Males develop salivary tumors at ~7 months of age. Spontaneous tumors express abundant transgenic Neu. These NeuT mice are tolerant to Neu and respond weakly to Neu vaccination [35]. Blockade of spontaneous tumorigenesis in NeuT mice is a significant challenge, particularly after carcinoma has been well established in situ.

In our lab, transgenic mice expressing wild-type human ErbB-2 (HER-2 Tg) have been established [30] and have been backcrossed with C57BL/6 mice for >12 generations. These mice do not develop spontaneous tumor, but are tolerant to human ErbB-2. It may be fruitful to test the effect of Treg-cell modulation on ErbB-2-positive tumor growth and to monitor autoimmune complications in BALB NeuT or C57BL/6 HER-2 Tg mice.

Balancing anti-tumor and autoimmunity when manipulating Treg cells

New strategies to induce strong anti-tumor immunity without significant autoimmunity will be the next milestone in tumor immunotherapy. To achieve anti-tumor immunity, it will be important to eliminate Treg cells to tumor-associated antigen. To avoid autoimmunity, it is advantageous to amplify Treg cells to self-antigens. The available tools for manipulating Treg cells, such as mAbs to CD25, GITR, and CTLA-4, do not distinguish one Treg cell from another. Will these tools be useful in cancer therapy? Perhaps the first order of business is to determine whether all Treg cells in a given individual are created equal and are therefore equally sensitive to these modulating agents. Using thyroiditis as a model system with known MHC class II–based susceptibility, one can test if Treg cells generated during thyroglobulin-induced tolerance are qualitatively different from resident Treg cells in naïve mice. Alternatively, it may be advantageous to inactivate tumor-specific Treg cells locally at the vaccination site to amplify priming to tumor-associated antigens without systemic down-modulation of Treg cells, thus minimizing the priming to self-antigens. The insight on Treg cell biology, the available molecular tools, and the awareness of autoimmune complications will be critical in cancer immunotherapy.