Disarming suppressor cells to improve immunotherapy

Abstract

Human tumors can use many different mechanisms to induce dysfunction in the host immune system. Accumulations of inducible regulatory T cells (iTreg, Tr1) are commonly seen in the tumor microenvironment. These Treg express CD39 and up-regulate CD73 ectonucleotidases, hydrolyze exogenous adenosine triphosphate (ATP) to AMP and adenosine and produce prostaglandin E₂ (PGE₂). Most tumors also express CD39/CD73 and COX-2 and thus contribute to immune suppression. Pharmacologic inhibitors can be used to eliminate adenosine/PGE₂ production by Tr1 as well as the tumor or to block binding of these factors to their receptors on Teff or to selectively block cAMP synthesis in Teff. These pharmacologic blocking strategies used alone or in combination with conventional treatments or immunotherapies could disarm Tr1, at the same time restoring antitumor functions of Teff.

Introduction

Regulatory T cells (Treg) represent a small subset (~5%) of human CD4⁺ T cells. Their major characteristic is the capability to regulate functions of conventional T lymphocytes (Tconv) in order to prevent excessive immune activation. More than 35 years ago, Gershon coined the term “suppressor cells” to describe CD8⁺ T lymphocytes capable of blocking functions of other T cells [1]. Today, it is abundantly clear that another subset of T cells, CD4⁺ Treg, assumes this responsibility when called to exercise immune control. Treg comprise several subsets of phenotypically similar cells that mediate suppression via distinct and often unexpected mechanisms [2–5].

Treg in humans

At least two Treg subsets have been recognized in humans: (a) natural Treg (nTreg), which originate in the thymus, mediate suppression by cell contact-dependent mechanisms involving the granzyme B/perforin or Fas/FasL pathways and constitute the major regulatory T-cell subset responsible for maintaining peripheral tolerance to self [6]; (b) inducible Treg (iTreg) also referred to as type 1 regulatory T cells (Tr1), which are induced in the periphery following chronic antigenic stimulation in the presence of IL-10 derived from tolerogenic antigen-presenting cells [7]. They differentiate into active suppressor cells that mediate suppression via contact-independent mechanisms largely involving TGF-β1, IL-10 or other soluble immunoinhibitory factors such as adenosine (ADO) or prostaglandin E₂ (PGE₂) [8, 9]. The phenotype of human Treg is not firmly defined. Most nTreg are CD3⁺CD4⁺CD25^highFOXP3⁺, while Tr1 cells have a somewhat different phenotype characterized by high expression levels of inhibitory cytokines, TGF-β1 and IL-10, and the absence of IL-4 [8]. Treg accumulate at inflammatory tissue sites and in human tumors [10]. Treg recruited to tumors appear to be resistant to death and mediate higher levels of suppression than Treg in the peripheral circulation [11]. Considerable controversy exists as to the phenotype and the role of Treg in tumor progression. It is important to remember that Treg-mediated regulation of activated immune cells represents a physiologically normal response designed to maintain a homeostatic balance and prevent undesirable immune activation. Treg found in pathologic conditions, such as cancer or chronic infections, may be recruited to affected tissues and conditioned to suppress excessive activation of immune cells, including antitumor Teff, thus favoring tumor growth. Unexpectedly, Treg were found to be significantly increased in the frequency and activity after various immune therapies, including anti-cancer vaccines [12, 13]. This observation reported by several different investigators suggests that Treg might interfere with antitumor immune responses generated as a result of therapy and thus promote tumor progression. If Treg play a role in tumor progression, then their silencing could eliminate or block tumor escape. However, attempts to remove Treg using various depleting agents have only partially and transiently reduced their frequency [13]. In contrast, by directly inhibiting the production of immunoinhibitory factors, e.g., ADO or PGE₂, the two of many such factors produced by iTreg and contributing to tumor escape [14], it might be possible to avoid or prevent suppression or to protect T effector cells (Teff) from inhibitory effects of these factors. Such strategies might be able to restore effective antitumor immunity in patients with malignancies.

Expression of ectonucleotidases by human T cells

It has been known for a long time that immune cells participate in exogenous adenosine triphosphate (ATP) hydrolysis to immunosuppressive ADO, because they express surface ectonucleotidases, CD39 (an ectonucleotidase triphosphate diphosphorylase) and CD73 (an ecto-5′-nucleotidase) [15]. Immune cells also respond to ADO, because they bear receptors for ADO, which are mainly of the A2a/A2b type [16]. More recently, it has been reported that Treg in the mouse are enriched in enzymatically active CD39 and CD73 relative to Teff and that the latter shows a higher expression of ADO receptors than Treg [17–19]. This introduced a possibility that differences might exist in the way Treg and Teff to handle ADO and respond to its signaling. In fact, we observed that CD39 was expressed on the surface of the great majority (>90%) of human CD4⁺CD25^hiFOXP3⁺ Treg and only minor subsets of CD4⁺ and CD8⁺ T cells (Fig. 1). We also found that while Teff expressed an abundance of CD26 (dipeptidyl peptidase IV) and adenosine deaminase (ADA), highly purified Treg did not [20]. Further, it appeared that CD73 was not readily detectable by flow cytometry on the surface of Treg, although it was present on the surface of small subsets of other immune cells (Fig. 1) and on tumor cells we have tested. These data have led us to propose that (a) CD39 could be a useful cell marker for the detection and isolation of human Treg; (b) the presence of CD39 and the absence of the CD26/ADA complex in Treg suggest these cells might effectively use adenosine monophosphate (AMP) and ADO for suppression, while being relatively resistant to inhibition mediated by these products; (c) Teff that lack CD39 are unable to hydrolyze exogenous ATP and might be susceptible to ATP-mediated toxicity; and (d) Teff express A2a/A2b receptors for ADO, are highly sensitive to ADO-mediated suppression and use the CD26/ADA complex for protection by deaminating ADO to inosine. The emerging hypothesis is that Treg and Teff differ in the way they utilize the adenosinergic pathway with biologically significant consequences.

Fig. 1.

Differential expression of surface CD39, CD73 and CD26 on lymphocyte subsets in the peripheral blood mononuclear cells obtained from normal donors (n = 15). Data are mean values ± SD. In human lymphocytes, CD73 is expressed on a small proportion of CD4⁺CD25^high Treg. The P values are for differences between marker expression on Treg and conventional CD4⁺ T cells. Data reproduced from ref. [20]

Suppression mediated by CD39⁺ Treg

To begin investigating the possibility that the observed differences between Treg and Teff phenotype underline their distinct functional profiles, we have been working with highly purified subsets of human CD4⁺CD25^hiFOXP3⁺ Treg and CD4⁺CD39⁻CD25⁻ Teff isolated from the peripheral blood of normal donors and patients with cancer [21]. We first showed that CD39 expressed on these Treg is enzymatically active (Fig. 2a). The CD39⁺ Treg suppressed proliferation of autologous responder T cells in CSFE-based assays (Fig. 2b). Both ATP hydrolysis and suppressor activity of CD4⁺CD39⁺ Treg were blocked by ARL67156, a pan inhibitor of ecto-ATPases [20]. To show that CD39⁺ Treg participate in ADO production, we co-incubated these cells with CD73⁺CD4⁺ T cells in the presence of 20 μM ATP for 45 min and measured ADO in the supernatants by mass spectrometry. We observed that CD39⁺ Treg supplied with CD73 were excellent ADO producers (data not shown). In aggregate, these results showed that human CD4⁺CD39⁺ Treg can hydrolyze exogenous ATP to AMP and ADO and utilize these factors to suppress proliferation of autologous Teff responding to stimulation with anti-CD3/anti-CD28 Abs.

Fig. 2.

Hydrolysis of exogenous ATP by human CD4⁺CD39⁺ Treg in (a) and suppression of proliferation of autologous responder T cells stimulated with anti-CD3/anti-CD28 Abs by CD4⁺CD39⁺ Treg in (b). In a, CD4⁺CD25^hiCD39⁺ Treg were incubated with exogenous ATP for 30 min. Unhydrolyzed ATP was measured. Asterisks indicate significantly lower levels of unhydrolyzed exogenous ATP in the absence of the inhibitor, ARL67156. Representative data from one of 3 experiments performed with Treg isolated from the peripheral blood of different normal donors. In b, responder T cells (R) were labeled with CFSE and co-incubated with autologous Treg (S) in the presence or absence of ARL67156. Suppression mediated by CD4⁺CD39⁺ Treg was significantly (P < 0.05) inhibited in the presence of ARl67154. The data are mean values ± SC obtained in 3 independent experiments. Data reproduced from ref. [20]

Adenosine and prostaglandin E2 (PGE2) production by iTreg (Tr1)

To generate human iTreg (Tr1), an in vitro assay system (IVA) was developed [22], in which a starting population of CD4⁺ CD25- T cells cultured for 9 days in the presence of autologous dendritic cells, irradiated tumor cells, and low concentrations of IL-2, IL-10 and IL-15 differentiates into Tr1-expressing CD39, CD73, CTLA-4, CD132, IL-10, TGF-β and COX-2 [23]. These Tr1 were variably enriched in CD4⁺FOXP3⁺ cells [8, 23]. They suppressed TcR-driven proliferation of autologous CD4⁺ responder cells and ARL67156 as well as an antagonist of A2a receptors, ZM241385, significantly diminished this suppression (Fig. 3) [23]. These observations were consistent with the conclusion that ADO was responsible for suppression mediated by Tr1. Indeed, CD39⁺COX2⁺ Tr1 were able to hydrolyze exogenous ATP to ADO and to produce PGE₂ after stimulation of the cells with anti-CD3/anti-CD28 Abs (Fig. 4).

Fig. 3.

Suppression of autologous effector T-cell proliferation in response to TCR-mediated signals mediated by inducible Treg (Tr1) ± ARL67154, a selective inhibitor of ATP hydrolysis or ZM 241385, an antagonist of A2a receptor. Data reproduced from ref. [23]

Fig. 4.

Human Tr1 cells generated in cocultures with COX-2⁺ or COX-2- tumor cells are CD4⁺ CD39⁺COX-2⁺ and have the capability to produce adenosine or PGE₂. Tr1 cells generated in the presence of COX-2⁺ tumor cells produced more adenosine or PGE₂ than Tr1 generated with COX-2- tumor cells. The reference cultures contain conventional CD4⁺ T cells stimulated with anti-CD3/anti-CD28 Abs and proliferating in the presence of IL-2. Data reproduced from ref. [23]

Expression and functions of CD39 and CD73 in the tumor microenvironment

Expression of surface ectonucleotidases in murine and human tumor cells has been previously reported [24]. Activated T cells, including iTreg, accumulate in tumors, and the presence of CD39⁺ iTreg in human tumors has been documented by us using confocal microscopy [23]. Therefore, it is reasonable to assume that tumor cells, which frequently are COX-2⁺, as well as activated immune cells contribute to ADO and PGE₂ production in the tumor microenvironment. These immunosuppressive factors signal via A2a/A2b receptors and EP receptors expressed by Teff, up-regulate cAMP levels in these cells and inhibit their functions (Fig. 5). Since a loss/inhibition of antitumor immune responses contributes to tumor escape, ADO and PGE₂ signaling promotes tumor progression. In addition, ADO and PGE₂ produced by Treg cooperate in inducing Teff dysfunction [25]. Silencing of ectonucleotidases and COX-2 in the tumor by small molecular inhibitors has been proposed as a strategy for controlling tumor growth [25]. The strategy has already been tried in experimental animals with some success [24]. Since iTreg are also a source of ADO and PGE₂, the elimination of Treg from the tumor microenvironment or blocking of their immunosuppressive effects mediated by these factors might be equally, if not more, important.

Fig. 5.

A schema of interactions between adenosine- and PGE2-producing Tr1 cells (CD4⁺CD39⁺CD73⁺) and effector T cells (Teff) expressing receptors for these factors. Teff that are positive for A2a receptors and EP2 receptors are highly sensitive to adenosine-mediated and PGE2-mediated suppression. Signaling via A2a and EP2 receptors converges on the intracytoplasmic adenylyl cyclase, leading to up-regulation of 3′5′-cAMP in Teff and inhibition of immune functions mediated by Teff. In a, blocking of ADO production from exogenous ATP by an ecto-ATPase, ARL 67156; in b, blocking of the binding of endogenous ADO produced by Tr1 to its receptors on Teff by, e.g, an antagonist ZM241385; or in c, blocking of PGE₂ production by Tr1 all are expected to rescue Teff from suppression. The red markers indicate blocking with pharmacologic inhibitors

The pharmacologic blockade to restore antitumor functions of effector T cells

Tumor cells and iTreg produce ADO and PGE₂ [23, 26] which bind, respectively, to A2a/A2b and EP receptors richly expressed on Teff. These signals converge on adenylyl cyclase (AC), stimulating its activity and up-regulating intracellular levels of 3′,5′-cAMP, a powerful inhibitor of Teff functions [27]. While AC-7, an AC isoform selectively expressed in hematopoietic cells, is a central regulator of cAMP levels, phosphodiesterase 4 (PDE4) expressed in Teff hydrolyzes cAMP to 5′-AMP. The ADA pathway operating downstream in Teff is responsible for the conversion of ADO to inosine. We speculate that disarming of the ADO/PGE₂ pathway by (a) blocking of ADO and EP receptors, (b) blocking of AC activity, (c) activating PDEs or (d) combining these strategies will enhance Teff functions while simultaneously decreasing/eliminating Treg activity. As indicated above, we have used various small molecular inhibitors to block the adenosine/PGE₂ pathway by, e.g., eliminating enzymatic activities of CD39 and CD73 ectonucleotidases or by blocking A_2A receptors expressed on Teff, thus reducing suppressor functions of Treg [20, 23]. However, it now appears that AC-7, the AC isoform selectively expressed in hematopoietic cells (our unpublished observation), may be an especially promising target for silencing the ADO/PGE₂ pathway. Using pharmacologic agents to block AC or stimulate PDE in purified, autologous Teff and Treg, we are examining the effects of these drugs on cAMP levels in these cells and on suppressor functions of Treg as well as on the ability of Teff to proliferate or produce cytokines in response to relevant tumor-associated antigens. We believe that the pharmacologic AC-7 blockade combined with PDE activation will protect Teff from Treg-mediated as well as tumor-mediated suppression and restores their antitumor activity. These coculture experiments are expected to provide a rationale for translation of this combinatorial drug immunotherapy first to in vivo models of tumor growth in mice and then to the clinic. By silencing Treg and concomitantly restoring or augmenting antitumor functions of Teff, this strategy could interfere with tumor progression and perhaps delay or eliminate tumor recurrence in cancer patients treated with conventional therapies.