Following T cell activation through the B7:CD28 signaling axis, CTLA-4 receptors are upregulated and bind to B7 molecules with higher affinity relative to CD28. By outcompeting CD28 for B7 ligands, CTLA-4 attenuates the T cell response, primarily through the inhibition of IL-2 and by blocking cell cycle progression (2). Similarly, constitutive CTLA-4 expression on T regulatory cells (Tregs) also reduces the level of B7 ligands on APCs, further inhibiting T cell immunity (3). Although the appearance of CTLA-4 on T cells during an acute antigen exposure is transient, chronic antigen exposure, as in the case of cancer, leads to a sustained expression of CTLA-4 (4). The inhibitive properties of CTLA-4 on T cell activation and its function in Treg suppression, then, are believed to contribute to the immunosuppressive phenotype observed in tumor-bearing hosts, and to cancer immune evasion.

Given the role of CTLA-4, substantial effort has been made to investigate CTLA-4 blockade in the hopes of rescuing T cell responses during cancer growth. Treatment with anti-CTLA-4 antibodies has shown to be effective in ameliorating disease in multiple murine tumor models (5,6). Collectively, these studies demonstrated improved tumor regression and survival, with concomitant increases in T cell activity. Notably, anti-CTLA-4 antibody therapy alone has been proven to be most effective with immunogenic tumor models, since low or non-immunogenic tumor models show no improvement upon CTLA-4 inhibition (7). Combinatorial therapy has been successful at circumventing this obstacle, as administration of anti-CTLA-4 antibodies alongside cancer vaccines, chemotherapeutic agents, radiation, and other therapies was shown to create a synergistic effect, allowing weakly immunogenic tumors to be targeted by activated T cells (8,9,10,11). In particular, the synergistic effects of CTLA-4 blockade and the engagement of another CD28 family T cell co-stimulatory molecule, ICOS, have shown promising results (see below).

Two human anti-CTLA-4 antibodies have been in clinical investigations for the past decade and have shown success as viable cancer immunotherapeutics particularly in melanoma. While both antibodies can neutralize CTLA-4 and enhance T cell responses, the efficacy appears to differ. Ipilimumab (IgG1 isotype) was demonstrated to enhance the expression of activation markers on circulating lymphocytes (12) and augment antigen-specific immune responses in patients with melanoma who were co-administered peptide vaccines (13). In 2010, a seminal phase III clinical trial of ipilimumab monotherapy provided encouraging results, as previously-treated metastatic melanoma patients saw increased overall survival (14). Another phase III study in 2011 observed similar results in untreated metastatic melanoma patients administered ipilimumab with dacarbazine (15). Based on these results, the U.S. FDA and European Medicines Agency have approved ipilimumab for metastatic melanoma therapy in 2011 and currently, studies examining ipilimumab as a monotherapy and in conjunction with other therapies are underway.

In contrast, studies involving tremelimumab (IgG2 isotype) have resulted in moderate success: in a phase I study, a cohort of melanoma patients saw long-term benefits upon tremelimumab treatment (16). These results were contrasted by a 2013 study, in which tremelimumab-treated melanoma patients had similar objective response rates to control patients (17). Possible explanations for the differences between ipilimumab and tremelimumab efficacy may stem from the properties of the immunoglobulin isotype, as murine studies have demonstrated that anti-CTLA-4 antibodies expressing the IgG2a isotype (equivalent to ipilimumab isotype) have enhanced antitumor activity relative to treatment with anti-CTLA-4 antibodies expressing other isotypes (18).

In light of studies in mouse models demonstrating the importance of tumor immunogenicity on the impact of CTLA-4 blockade, emerging clinical evidence also suggests that patients with ongoing immune responses prior to and during immunotherapy respond better upon CTLA-4 treatment. Patients who exhibited NY-ESO-1-specific T cell activity were better responders to ipilimumab therapy compared to patients that were sero-negative (19). Further, in bladder cancer patients, the accumulation of tumor-specific, IFN-γ-producing CD4⁺ICOS⁺ T cells following treatment was associated with enhanced effector to Treg cell ratio, making this subset an attractive biomarker for predicting patient outcome (20).

PDL-1 and PDL-2 are additional members of the B7 family, and represent the two known ligands for the PD-1 receptor. PD-L1 is found on hematopoietic cells and parenchymal cells, and PD-L2 is restricted to macrophages and dendritic cells (DCs) (21). The PD-1 receptor is induced on activated T cells, but can also be found on B cells and NK cells (22). While CTLA-4 limits the amplitude of early T cell responses, PD-1 suppresses T cell function in peripheral tissues through the activation of phosphatases that inhibit kinases in the upstream of TCR signaling cascade. Additionally, PD-1 is crucial for the development and function of Tregs, representing an indirect mechanism leading to dampened T cell responses (23). Since PD-1 is selectively upregulated in conditions of persistent antigen exposure and chronic inflammation and has been shown to drive cellular exhaustion, the expression of PD-1 on T cells has become one of the hallmarks of exhausted T cells. Indeed, in multiple human tumors, a significant proportion of TILs have been shown to express PD-1, and often, this expression has been associated with impaired CD8+ T cell function (24,25). Rescue of T cell activity by PD-1/PD-L1 blockade has been demonstrated in models of chronic viral infections (26), providing evidence for the use of PD-1 blockade to restore T cell function amongst exhausted PD-1+ TILs.

While PD-1 is often associated with TILs in numerous cancers, PD-1 ligands are frequently observed on the surface of multiple human tumors and murine cancer cell lines, with PD-L1 as the ligand that is most commonly expressed (27,28,29). This is not surprising, as PD-L1 can be upregulated by inflammatory cytokines and via oncogenic signalling (23). PD-L2 expression has mainly been documented in B cell lymphomas, such as primary mediastinal B-cell lymphoma and Hodgkin's lymphoma (30,31). In hepatocellular carcinoma, melanoma and breast cancer, PD-L1 positivity was correlated with worse prognosis (32,33,34), consistent with its physiological role in negatively regulating T cell responses. Other studies have found that PD-L1 status on tumors has either a positive or no correlation with patient prognosis (23,35,36). These observations may be attributed to differences in cancer type, stage of cancer, and/or patient treatment history, indicating the requirement for further studies examining PD-1 ligand positivity and patient outcome.

In numerous murine models of cancer, inhibition or absence of PD-1 has been demonstrated to rescue anti-tumor T cell responses, resulting in diminished tumor burden (37,38). Likewise, blockade of PD-L1 could also rescue T cell function and enhance tumor regression in a variety of murine cancer models (27,39). Studies on PD-L2 blockade, however, have provided contradicting results regarding the inhibitory or stimulatory role of PD-L2 (39,40), and may indicate the involvement of an undiscovered co-stimulatory T cell receptor specific for PD-L2.

While the blockade of CTLA-4 has been characterised in human patients, clinical studies examining PD-1 blockade have only been initiated recently, and thus limited data is available. Nonetheless, initial clinical trials showed improved disease progression upon administration of anti-PD-1 antibodies. In the first clinical study involving a fully human IgG4 anti-PD-1 antibody (MDX-1106), PD-1 blockade elicited partial responses in patients with melanoma and renal cell carcinoma (41). An additional study with another anti-PD-1 antibody (BMS-936558) also demonstrated objective responses in patients with non-small cell lung cancer, melanoma and renal cell cancer, which did not occur in patients whose tumors were negative for PD-L1 (42). Treatment with lambrolizumab, another blocking PD-1 antibody, has yielded sustained tumor regression in patients with advanced melanoma (43).

Studies examining the effects of blocking PD-1 ligands in humans are scarce, although preliminary studies have been initiated and are presently ongoing. The first clinical trial with anti-PD-L1 antibodies (BMS-936559) demonstrated tumor regression and disease stabilization in non-small cell lung cancer, melanoma and renal cell cancer at 24-weeks (44). An ongoing clinical trial documenting the efficacy of another PD-L1 antibody (MPDL3280A) has also yielded objective responses in non-small cell lung cancer patients. This is particularly notable, as these patients had tumors that were difficult to treat, as evidenced by their treatment history (investigator update http://www.roche.com/investors/ir_update/inv-update-2013-09-29.htm). To date, only one reagent targeting PD-L2 has been tested in humans. A recombinant PD-L2-Fc fusion protein (AMP-224) has been developed, and in individuals with partial or mixed responses, AMP-224 treatment reduced PD-1^hi exhausted cells and enhanced functional T cells in patients with advanced solid tumors (2013 ASCO Annual Meeting poster abstract, http://meetinglibrary.asco.org/content/117257-132).

Notably, combinatorial therapy involving PD-1 and CTLA-4 blockade has shown exceptionally promising results, as a phase I trial showed 53% of advanced melanoma patients had an objective response with tumor reduction of at least 80% (45). These observations have led to the initiation of phase III clinical studies, which is expected to yield encouraging data. These findings support the targeting of T cell co-inhibitory molecules involved in multiple phases of the T cell response, including those responsible for early T cell priming and those mediating late-stage effector functions.

In the majority of patients responding to PD-1 or PD-L1 blockade, the expression of PD-L1 in the tumor was correlated with treatment responsiveness. While it may appear counterintuitive that a T cell inhibitor may predict better prognosis, the upregulation of PD-L1 is often induced by cytokines produced by effector immune cells. Evidence for this in the context of tumor growth was provided in one study that demonstrated the localization of IFN-γ to the same area where PD-L1⁺ tumors associated with TILs (35), and studies in mice have shown that the induction of PD-L1 in the tumor relied upon IFN-γ-producing tumor-associated T cells (46). This suggests a negative feedback loop, as blockade of PD-1 signaling would enhance the production of inflammatory cytokines by immune cells, which would then promote the upregulation of PD-L1 in the tumor. Expression of PD-L1 would presumably suppress PD-1⁺ TILs, and this mechanism of tumor-mediated immunosuppression following T cell attack has been termed "adaptive immune resistance". This is in contrast to the constitutive expression of PD-L1 on tumor cells which is driven by oncogenic signalling, and suggests that the blockade of PD-1 and PD-L1, along with the careful timing of treatments, would produce the most effective anti-tumor response. These observations also imply that the expression of PD-L1 in the tumor milieu is indicative of pre-existing immune responses, and may be useful in predicting the responsiveness of immunotherapies that rely on the presence of immunosurveillance in the growing tumor.

The inducible co-stimulatory receptor (ICOS) shares much homology with CD28, yet key differences in signaling mechanisms and unique expression patterns of ICOS ligand suggest non-redundant functions. Similar to CTLA-4, ICOS is induced following T cell activation (2). During the initial priming of naïve T cells, the contribution of ICOS appear minimal relative to that of CD28; T cells lacking ICOS show minor differences in proliferation (1,2,47,48). However, while both CD28 and ICOS can bind phosphoinositide 3-kinase (PI3K), engagement of ICOS has been shown to induce greater PI3K signaling than CD28 co-stimulation (49). The significance of this finding was demonstrated by our group and others in being crucial for the production IL-21 and IL-4, key cytokines involved in the differentiation and function of T follicular helper (Tfh) cells (50,51,52). ICOS has also been detected on Tregs, where its expression was shown to influence the homeostasis and function of this subset (1).

The ICOS receptor is engaged by ICOSL, another member of the B7 family. ICOSL is expressed in APCs (B cells, macrophages, dendritic cells) and can be induced by inflammatory cytokines in non-hematopoietic cells including endothelial cells and epithelial cells (53).

ICOS and ICOSL have been observed in several human cancers, yet how this signaling pathway contributes to the anti-tumor response remains unclear. In the peripheral blood of colon cancer patients, co-stimulatory genes such as ICOS were significantly diminished and this reduction was linked to lymph node metastasis and aggressive tumor invasion (54). Moreover, high ICOS expression on TILs in metastatic melanoma lesions was associated with post-recurrence survival (55). While these findings support the co-stimulatory role of ICOS:ICOSL in facilitating the anti-tumor T cell response, others have revealed an inhibitory, pro-tumor role for ICOS signaling related to its function in Treg homeostasis. In one study, acute myeloid leukemia patients exhibiting ICOSL positivity had significantly decreased survival (56). In freshly isolated human melanomas, 25% of samples expressed ICOSL and 50% of metastatic samples demonstrated high ICOSL expression (57). The ICOSL present in these melanoma samples were able to engage ICOS on activated Tregs and stimulate suppressive functions, thus providing a means of tumor immune evasion. Notably, in melanoma patients, the specific expansion of ICOS⁺ Tregs following the first cycle of high-dose IL-2 therapy is correlated with worse clinical outcome (58).

Despite contradictory observations regarding the role of ICOS signaling in cancer progression, recent findings support the role of ICOS as a biomarker for host responsiveness to CTLA-4 blockade. Ipilimumab treatment in patients with urothelial carcinoma of the bladder led to an increased frequency of CD4⁺ICOS^hi T cells (59), and administration of both tremelimumab and exemestane in hormone-responsive advanced breast cancer patients resulted in enhanced circulating ICOS⁺ T cells (60). CD4⁺ICOS^hi T cells obtained from bladder cancer patients treated with anti-CTLA-4 antibodies also showed IFN-γ production, the ability to recognize the NY-ESO-1 tumor antigen, and an increase in the ratio of T effector cells to Tregs (20). Correlations between ICOS positivity following CTLA-4 blockade and clinical benefit has also been examined, albeit to a lesser extent. In a small cohort of metastatic melanoma patients treated with ipilimumab, retrospective analysis showed that a sustained increase in CD4⁺ ICOS^hi T cells was associated with increased likelihood of clinical benefit (59). Findings from murine studies further proposes that ICOS engagement may be required for the optimal effects of CTLA-4 inhibition, as ICOS-deficient mice bearing B16 melanomas exhibited drastically diminished anti-tumor T cell responses upon CTLA-4 blockade (61). Recently, modified ICOS-based chimeric antigen receptors generated tumor-specific IL-17⁺ T cells with greater persistence (62). Collectively, these findings support the requirement of ICOS for maximal benefits of CTLA-4 inhibition, and prompt further studies examining the synergy between CTLA-4 blockade and ICOS stimulation in generating optimal anti-tumor T cell immunity.

B7-H3 is a B7 family whose receptor remains unidentified. Despite this, one candidate, the triggering receptor expressed on myeloid cells (TREM)-like transcript 2 (TLT-2), has been proposed. While the engagement of B7-H3 with TLT-2 resulted in enhanced proliferation and IFN-γ production (63), a subsequent report disputed these findings in mouse and human systems (64). Consequently, contradictory findings regarding the co-stimulatory or co-inhibitory function of B7-H3 persists, as evidence demonstrating both co-stimulatory and co-inhibitory functions have been reported by numerous groups (65,66,67,68). In terms of expression, B7-H3 is constitutively found on murine APCs, but must be induced on human immune cells (69). B7-H3 is not restricted to immune cells and is found on osteoblasts, fibroblasts, epithelial cells and other cells of non-lymphoid lineage (70).

While in vitro studies and murine disease models have provided support for both the co-stimulatory and co-inhibitory capacities of B7-H3, experiments with tumor models appear to largely support a co-stimulatory role for B7-H3 in the regulation of anti-tumor immunity. For instance, P815 tumors transfected with B7-H3 showed enhanced immunogenicity and tumor regression (71), and intratumoral administration of a B7-H3 expression plasmid led to the significant reduction of EL-4 tumors (72).

Despite the evidence from murine studies implicating B7-H3 as a positive regulator of anti-tumor immunity, expression of B7-H3 in human cancers tends to favor poor prognosis. To date, only a handful of clinical studies have reported a co-stimulatory function and good prognosis associated with B7-H3 (73,74). The majority of reports demonstrate an inverse correlation between B7-H3 staining and patient outcome (75,76). In one instance, strong B7-H3 positivity was observed in colorectal carcinomas and correlated with tumor grade and decreased TILs; notably, TNF-α, an inflammatory protumorigenic molecule, could induce the shedding of soluble B7-H3 by colon cancer cell lines, suggesting another method of cancer immune evasion via B7-H3 (77).

To date, a monoclonal antibody against a human isoform of B7-H3 (8H9) has been tested in early-phase clinical trials. High-risk patients with solid tumors and central nervous system metastasis exhibited prolonged survival upon 8H9 administration (78), no doubt prompting further inquiries into the efficacy of targeting B7-H3 in other cancer types.

As a more recent addition to the B7 family, B7-H4 represents a T cell co-inhibitory molecule whose expression pattern in the tumor microenvironment has garnered significant attention. B7-H4 transcripts are ubiquitous in healthy individuals, and B7-H4 mRNA has been found in both lymphoid and non-lymphoid organs such as the lung, liver, spleen, thymus, kidney, pancreas, and other tissues. B7-H4 protein, however, remains highly restricted in both mice and humans, indicating a tightly regulated translational mechanism (79,80). On hematopoietic cells, B7-H4 protein can be induced following in vitro stimulation of human T cells, B cells, monocytes, and DCs (79). Similarly, tumor-associated macrophages were observed to express B7-H4, and this upregulation was attributed to the presence of IL-6 and IL-10 in the tumor microenvironment (81).

Functionally, B7-H4 has been well-documented to inhibit T cell responses. In vitro studies on murine and human T cells revealed an inhibitory role for B7-H4 in regulating cell cycle progression, proliferation and cytokine secretion (82,83). Accordingly, mice deficient in B7-H4 displayed upregulated Th1 response upon Leishmania major infection, but did not show enhanced hypersensitive inflammatory responses or increased CTL activity during viral infections, suggesting that the role of B7-H4 may be one of a fine tuner (84). B7-H4 has also been found to regulate the activity of myeloid cells, as B7-H4 knockout mice showed enhanced CD11b⁺Gr-1⁺ neutrophils and lower Listeria monocytogenes burden, concordant with the observations that in vitro administration of B7-H4 fusion protein diminishes the expansion of neutrophil progenitors (85). We have also observed enhanced suppressive capacities of CD11b⁺Gr-1⁺ myeloid-derived suppressor cells (MDSCs) in the absence of B7-H4 (86). Notably, despite the restricted profile of B7-H4 in healthy individuals, an array of human cancers overexpress B7-H4 molecules. These findings have prompted the study of B7-H4 on cell growth outside of their capacity to modulate immunity. For instance, overexpression of B7-H4 on cancer cells appear to enhance proliferation, adhesion, migration and protect from apoptosis in vitro, and accordingly, tumor cells overexpressing B7-H4 showed enhanced growth in SCID mice (87,88).

To date, the receptor for B7-H4 has not been identified, and although initial studies pointed to BTLA as the B7-H4 binding partner, a subsequent study showed a lack of direct binding between the two (89). Recent studies attributing a function for B7-H4 in influencing myeloid cell activity further suggest the presence of a B7-H4 receptor on this subset. Indeed, neutrophils and tumor-associated neutrophils have been shown to bind B7-H4 protein (85,90).

While B7-H4 protein has limited expression in healthy individuals, numerous human cancers exhibit aberrant B7-H4 expression (91). In the majority of instances, B7-H4 positivity is linked with poor prognosis and is thus believed to have potential as a biomarker (88,92,93). These findings provide a basis for targeting B7-H4 in cancer, and several studies performed in mice have further validated this notion.

In an experimental model of lung metastasis, host B7-H4 was demonstrated to reduce the level of tumor-specific T cell responses and augment the infiltration of immunosuppressive subsets into the lung. As a result, B7-H4 WT mice exhibited more lung metastases relative to B7-H4 knockout mice (90). We have also shown that in the 4T1 tumor transplantation model, B7-H4-deficiency leads to enhanced IFN-γ in the tumor microenvironment, but no difference in tumor burden was observed between WT and B7-H4 knockout mice (86). This may be explained by our finding that in the absence of B7-H4, MDSCs showed enhanced immunosuppressive activity, in line with previous data suggesting a negative role for B7-H4 in the regulation of neutrophils which share similar markers. Notably, B7-H4 deficiency and treatment with gemcitabine, a current chemotherapeutic agent which was shown to decrease the level of MDSCs without affecting other immune subsets (94), was seen to drastically reduce 4T1 tumor burden relative to gemcitabine treatment alone, and thus provides support for combinatorial therapy. In a humanized murine model of ovarian cancer, administration of anti-B7-H4 single-chain fragments variable (scFv) was able to delay tumor growth; the same reagent was also able to rescue tumor-specific T cell activity from inhibition mediated by B7-H4⁺ APCs in vitro (95). To date, no anti-human B7-H4 blocking antibody has been tested in the clinic, yet based on the findings in murine tumor models and the high degree of B7-H4 staining in cancers relative to non-cancerous tissue, it is of great interest to see if blockade of B7-H4 can be used in combination with other immunotherapeutics or other chemotherapies.