Modulation of costimulation to enhance tumor immunity

Introduction

It is widely accepted that two signals are required for efficient T-cell activation. T cells receive their first, antigen-specific signal following ligation of TCR by MHC-peptide complexes. The second signal, which is antigen-independent, is delivered by ligation of CD28 by CD80 or CD86 (B7-1 or B7-2). The importance of this second, costimulatory signal has been clearly demonstrated. T cells receiving an antigen-specific signal without costimulation may become unresponsive. In addition to CD28, many other important costimulatory pathways have been described.

Understanding T-cell activation via TCR ligation and costimulation is crucial to the development of novel strategies to enhance antitumor immune responses. The complex organization of the immune system allows effective elimination of a nearly infinite repertoire of foreign invaders while self/non-self discrimination prevents damage of self. Unfortunately, the immune response to tumors is often ineffective. A major mechanism of immune evasion is the lack of costimulatory ligand expression, which renders tumor-antigen-specific T cells impotent [21]. To counteract this, many studies have exploited the various costimulatory pathways to elicit a more potent antitumor immune response.

Costimulatory molecules critical for initiating T-cell activation

B7-1 and B7-2

B7-1 and B7-2, expressed predominately on professional antigen-presenting cells (APCs), are the ligands for both CD28 and CTLA-4 [5, 25, 54]. CD28 is constitutively expressed on T cells, whereas CTLA-4 is up-regulated following T-cell activation. The ligation of CD28 by B7-1/B7-2 delivers a well-characterized, and perhaps the most potent, antigen-independent costimulatory signal. The effects of CD28 ligation include an increase in the production of IL-2 and other cytokines, expression of CD25 (the IL-2 receptor), and prevention of activation-induced cell death (AICD) [8, 44, 60]. Mice lacking CD28 have extremely impaired T-cell responses, underscoring the importance of this pathway [76].

CTLA-4 was identified in the late 1980s [12] and defined as another receptor for B7-1 and B7-2 in the early 1990s [55]. Interestingly, it was discovered that although CTLA-4 and CD28 share common ligands, CTLA-4 has a much higher affinity for B7 [56]. In contrast to CD28, CTLA-4 ligation has an inhibitory effect on T-cell activation [45, 46, 91]. Owing to the role of CTLA-4 in inhibiting T-cell responses, CTLA-4–deficient mice suffer from a lymphoproliferative disorder that results in death within 1 month of birth [80, 93].

Given that B7/CD28 interactions enhance and B7/CTLA-4 interactions inhibit T-cell activation, these pathways have been exploited to enhance the immune response to tumors. Many tumor cells lack B7 expression, which means that they cannot effectively prime T cells. In fact, if T cells receive chronic TCR stimulation without costimulation, they may become anergic [66, 75]. While many investigators demonstrated that conferring B7 expression on tumor cells augmented the antitumor response to tumors of varied origins [6, 16, 51, 81], it was apparent that provision of B7 expression was not sufficient to induce immunity against nonimmunogenic tumors [17]. In addition, therapy with single-chain antibody molecules designed to deliver CD28 costimulation effectively controlled the growth of colon carcinoma cells [34]. Combining B7 expression with cytokine expression (IFN-γ, IL-12, GM-CSF) also led to enhanced tumor immunity in some models [19, 39, 102]. However, therapeutic efficacy was often restricted.

Given its role in attenuating T-cell responses, blockade of CTLA-4 was tested as a way to enhance immunity to recently established tumors. [24, 47, 48]. Several models showed a synergy between CTLA-4 blockade and GM-CSF [40, 41, 82]. This suggested that enhancement of APC activation by GM-CSF could synergize with enhancement of antitumor CTLs following CTLA-4 blockade. Coexpression of both B7-1 and IFN-γ significantly enhanced antitumor immunity to a murine lymphoma following administration of anti-CTLA-4 [83]. These murine studies have led to the initiation of human clinical trials testing the efficacy of CTLA-4 blockade against melanoma and prostate cancer. A recent report demonstrated that treatment with a human anti-CTLA-4 MAb (MDX-CTLA-4) is sufficient to enhance antitumor immune responses in melanoma patients previously vaccinated with GVAX, a GM-CSF–expressing cell-based vaccine [36]. Ongoing studies with MDX-CTLA-4 in other therapeutic settings will help to confirm the feasibility of this promising approach.

Costimulatory molecules involved in sustaining and enhancing T-cell responses

CD40

CD40 is a member of the TNFR superfamily and is expressed on B cells, macrophages, and dendritic cells. The ligand for CD40, CD40L (CD154), is expressed on activated T cells. Ligation of CD40 enhances the expression of costimulatory molecules on dendritic cells and stimulates the production of IL-12 [14].

CD40/CD40L costimulation is important for effective antitumor responses. Blocking CD40/CD40L interactions prevented the generation of a protective antitumor response following vaccination with GM-CSF–expressing B16 melanoma cells [59]. Several methods for incorporating CD40L into tumor cell–based vaccines have significantly enhanced responses to poorly immunogenic tumors [29, 33, 58, 70, 78, 97]. In addition, Diehl et al. demonstrated that ligation of CD40 was sufficient to overcome tolerance to tumor antigen and to enhance protective antitumor responses [22].

Combining CD40/CD40L costimulation with additional immunomodulators has been somewhat successful. C-26 colon carcinoma cells transduced by GM-CSF and CD40L enhanced the survival of BALB/c mice as compared with C-26 cells transduced with either individually [18]. Combining CD40L and IFN-γ was advantageous in a lung cancer model [70], and adenovirus expressing CD40L and IL-2 was more protective than CD40L alone in a murine melanoma model [73]. A synergistic response was also reported between anti-CD40 and CTLA-4 blockade in a murine leukemia model. The administration of both anti-CD40 and anti-CTLA-4 MAbs prolonged postchallenge survival more than either MAb alone [43].

4-1BB

Another member of the TNFR family, 4-1BB (CD137), is rapidly expressed by activated CD4⁺ and CD8⁺ T cells [85]. The ligand for 4-1BB, 4-1BBL, is expressed on mature dendritic cells, activated B cells, and macrophages [20, 85]. When coupled with a strong signal through the TCR, 4-1BB ligation, like CD28 ligation, induces potent IL-2 production by naïve T cells [74]. Ligation by 4-1BB via anti-4-1BB MAb also prevents AICD of previously activated T cells [38]. Although both CD4⁺ and CD8⁺ T cells express 4-1BB, ligation of 4-1BB may preferentially enhance proliferation and production of IFN-γ in CD8⁺ T cells [77].

The potential for 4-1BB ligation to enhance antitumor responses was evaluated in several murine tumor models. Treatment of mice with anti-4-1BB antibodies 3 and 6 days after tumor challenge resulted in elimination of P815 mastocytoma and Ag104A sarcoma tumors [63]. In addition, vaccination with Ag104A sarcoma tumors co-expressing 4-1BBL and B7-1 enhanced subsequent rejection of Ag104A cells, which was in contrast to the ineffectiveness of transfection with either 4-1BBL or B7-1 alone [64].

OX40

OX40 is a member of the TNFR family that is expressed on activated T cells, with preferential expression on CD4⁺ T cells. Induction of OX-40 expression is dependent on CD28 ligation [49, 89]. Peak OX40 expression is seen 24–48 h following activation. The ligation of OX40 enhances the survival of CD4 cells during the initial response and increases the number of memory cells [28] by preventing AICD [61]. In addition, OX40 ligation increases the production of cytokines [27, 49, 89]. The ligand for OX40, OX40L, is induced on B cells and dendritic cells following CD40 ligation [71, 94].

OX40 expression was reported on tumor-infiltrating lymphocytes (TILs) of patients with melanoma and head and neck carcinomas, which suggests that the OX40/OX40L pathway could be a therapeutic target [84]. OX40 ligation significantly increased survival following a lethal challenge with MCA 303 by expanding tumor-specific CD4⁺ T cells [95] and enhanced immunity to a poorly immunogenic murine mammary carcinoma [65]. Additional studies suggest that antitumor immune responses can also be enhanced by combining OX40 ligation with other immunostimulatory approaches such as GM-CSF, 4-1BB, and IL-12 [31, 72].

ICOS

The inducible costimulator (ICOS) is also expressed on recently activated T cells [42]. The ICOS gene is clustered with CD28 and CTLA-4 on mouse chromosome 1 but does not share their ligands. The ICOS ligand, B7h, is expressed on B cells, macrophages, and dendritic cells [53, 99, 100] as well as in nonlymphoid tissues [79]. B7h expression is induced by IFN-γ and TNF-α [3, 10, 79]. ICOS ligation increases proliferation and cytokine production, particularly of IL-10 [42, 62]. Like OX40, ICOS expression is dependent on CD28 ligation [7, 62], which suggests that the ICOS/B7h interaction is involved in potentiating effector functions while CD28 ligation is essential for the activation of naïve T cells [52].

As with B7-1 and B7-2, B7h expression on tumor cells was demonstrated to enhance recognition of tumor cells by T cells. B7h expression on Sa1/N fibrosarcoma and J558 plasmocytoma cells enhanced tumor rejection in a CD8-dependent mechanism [57, 90]. Similar to enhancement of immunity by B7, engagement of ICOS effectively promoted regression of immunogenic tumors but not weakly immunogenic tumors [4].

PD-1

Programmed death 1 (PD-1) is expressed by activated T and B cells [2], as well as by activated macrophages [68]. Like CTLA-4, the function of PD-1 appears to be inhibitory, as suggested by the autoimmune-related diseases reported in PD-1–deficient mice [67, 69]. Furthermore, ligation of PD-1 inhibits proliferation and cytokine production [26, 50, 68]. PD-L1 (B7-H1) and PD-L2 (B7-DC) are both ligands for PD-1. Although a wide variety of tissues express mRNA for one or both of the PD-1 ligands [13], the protein expression is more restricted. The PD-L1 protein is constitutively expressed on lymphocytes and professional antigen-presenting cells and can also be increased on T cells (anti-CD3), macrophages (LPS, IFN-γ, GM-CSF, and IL-4), and dendritic cells (IFN-γ, GM-CSF, and IL-4). The PD-L2 protein is not constitutively expressed but can be induced on macrophages and dendritic cells following stimulation with IFN-γ, GM-CSF, and IL-4 [98]. Interestingly, there is recent evidence that suggests that PD-L1 and PD-L2 may bind to another receptor (not PD-1) and deliver a costimulatory, rather than an inhibitory, signal [23, 92].

Given that PD-1/PD-L interactions have an inhibitory effect on T cells, the finding that a variety of tumors express PD-L1 suggests another mechanism by which tumor cells could evade antitumor T-cell responses [11, 23]. Thus, blocking PD-1 ligation may be a useful strategy to enhance antitumor T-cell responses. Antibodies that block PD-1/PD-L interactions do in fact enhance T-cell responses [11]. Anti-PD-L1 MAb resulted in a transient regression of J558L myeloma cells (which naturally express PD-L1). Finally, B7-1–expressing P815 tumors completely regress but the addition of PD-L1 restores the growth of this tumor in syngeneic mice [23].

B7-H3

B7-H3 is the most recently described member of the B7 family and is expressed in a number of tissues and tumor cell lines [15]. Protein expression can be induced on T cells, macrophages, and dendritic cells but not B cells. Although the receptor for B7-H3 presently remains undefined, it is not CD28, CTLA-4, ICOS, or PD-1 [15]. A B7-H3-Ig chimeric fusion protein was used to demonstrate that the B7-H3 receptor is present on activated CD4⁺ and CD8⁺ T cells and when engaged, it enhances T-cell proliferation and IFN-γ production [15]. In addition, B7-H3–expressing 624mel melanoma cells significantly increased the effector function of 624mel-specific CTLs [15]. To date, there have been no in vivo reports of B7-H3–mediated enhancement of tumor immunity, either alone or in combination with other immunomodulatory approaches.

TRICOM

Given the unique function of many of the costimulatory molecules, optimal priming and expansion of tumor-reactive T cells is likely to be achieved by simultaneously exploiting several mechanisms. TRICOM is a fowlpox-based viral vector that encodes a triad of stimulatory ligands (B7-1, ICAM-1, and LFA-1). The use of TRICOM in combination with CEA (a tumor-associated antigen expressed in many carcinomas) expression significantly enhanced the activation of murine T cells compared with treatment with ligand alone [35, 101]. The CEA/TRICOM vaccine, when given along with rGM-CSF and IL-2, significantly enhanced survival following tumor challenge [1, 32]. Similar strategies have been evaluated in the MIN mouse model of intestinal tumors [30] and B cell lymphoma [9], which demonstrates the therapeutic potential of combinatorial strategies.

The efficacy of the TRICOM combinatorial strategy is currently being investigated in several clinical trials. These trials are aimed at patients with a variety of primary and metastatic cancers. Several of the trials are administering a TRICOM vaccine in combination with other treatments such as GM-CSF, surgery, radiation, and chemotherapy.

Human clinical trials

In addition to the TRICOM clinical trials, there are numerous on-going trials testing the efficacy of modulating various costimulatory axes. Anti-CTLA-4, which blocks the inhibitory signals delivered to T cells, is currently being used to enhance antitumor T-cell responses in phase I and phase II clinical trials. A recent report demonstrated that in metastatic melanoma and ovarian cancer patients previously vaccinated with autologous GM-CSF–expressing tumor vaccines, administration of anti-CTLA-4 (MDX-CTLA-4) induced significant antitumor responses [36]. Enhancing antitumor responses via CD40L-expressing tumor cell vaccines is also being evaluated in clinical trials. In a phase I trial for treating chronic lymphocytic leukemia, patients receiving an autologous tumor cell vaccine expressing CD40L had a significant reduction in lymph node size and the number of circulating leukemic cells [96]. Encouraging results have also been reported in cancer patients treated with recombinant human CD40L [88].

B7 is another costimulatory molecule that is currently being tested in clinical trials. B7 expression by tumor cell vaccines is the primary mechanism of CD28 engagement. Initial results suggest that vaccination with a canarypox virus expressing B7-1 and CEA, effectively induces CEA-specific immune responses [37, 86, 87]. Expansion of these trials to include the TRICOM vaccine may prove even more successful.

Conclusions

The American Cancer Society has predicted that there will be 1,334,100 new cancer cases in 2003 with more than 1,500 cancer-related deaths each day. These statistics obviously warrant a significant research commitment to the development of novel cancer therapies. One promising category of therapeutic strategies is immune modulation.

This review provides a list of many different costimulatory receptors that are candidates for eliciting more potent antitumor immune responses. It is clear that these receptors can provide a variety of therapeutic avenues. Our understanding of how they contribute to the complex array of signals that integrate to result in a robust T-cell response will clarify how they may be combined to enhance immunity to tumors.

The initiation and potentiation of T-cell activation is a complex process. For an antitumor immune response to be efficacious, tumor antigen–specific T cells require costimulation to initiate and sustain effector functions. This can be accomplished by therapies that target costimulatory pathways in both of these stages of T-cell activation. Some of the costimulatory functions do not appear to be redundant. Therefore, enhancing multiple costimulatory pathways while blocking inhibitory signals could generate a more potent antitumor response that can be used for cancer therapies.

Modulation of costimulatory pathways alone may not be directed to the only targets of combinatorial strategies. The most efficacious cancer therapy may be one that combines classic treatments, such as chemotherapy and radiation, with immunomodulatory cytokines, such as IL-2 and GM-CSF, and costimulatory molecules. As we advance our appreciation of the complex array of costimulatory interactions involved in T-cell priming, their application to cancer will be translated into more effective therapies.