TRAIL Signaling and Synergy Mechanisms Used in TRAIL-Based Combination Therapies

Since the discovery of members of the TNF family and their so-called death receptors, the possibility of a new avenue for apoptosis-inducing cancer therapies has emerged. The death ligand–dependent extrinsic apoptotic pathway can induce apoptosis independent of the transcription factor p53 and, therefore, can circumvent the apoptosis resistance acquired by many tumors through loss-of-function mutations in this tumor suppressor. However, TNFα and agonistic antibodies activating the Fas receptor were found to be highly cytotoxic toward primary hepatocytes and other nontransformed cells when applied at clinically relevant concentrations (1, 2). In contrast, soluble TRAIL as well as leucine zipper–trimerized TRAIL potently induce apoptosis in a range of cancer cell lines, can inhibit tumor xenografts in mice, and were found to be of very low systemic toxicity in mice and nonhuman primates (3, 4). These findings sparked tremendous interest in exploring the potential of TRAIL as an anticancer therapeutic for a broad range of human malignant neoplasms. In this review, we briefly describe the key signaling events during TRAIL-induced apoptosis initiation, give a short overview on the status of TRAIL-based clinical research and trials, and describe the mechanisms of signaling processes that evoke synergistic apoptosis responses in TRAIL-based combination treatments.

Endogenous TRAIL is expressed as a 281–amino acid type II trans-membrane protein, which is anchored to the plasma membrane and presented on the cell surface. TRAIL was independently identified by Wiley and colleagues and Pitti and colleagues in 1995 and 1996, respectively, and sequence alignments indicated its close relation to other death ligands, with highest sequence similarities reported for Fas ligand (FasL; refs. 5, 6). TRAIL is expressed by natural killer cells, which, following the establishment of cell–cell contacts, can induce TRAIL-dependent apoptosis in target cells (7). Physiologically, the TRAIL-signaling system was shown to be essential for immune surveillance, for shaping the immune system through regulating T-helper cell 1 versus T-helper cell 2 as well as “helpless” CD8⁺ T-cell numbers, and for the suppression of spontaneous tumor formation (8–10).

TRAIL induces apoptosis through ligation with its cognate death receptors TRAIL-R1 (also known as DR4) and TRAIL-R2 (DR5; Fig. 1; refs. 11–13), and the trimerization of TRAIL around a central zinc atom via cysteine²³⁰ is essential for its apoptotic potential (14, 15). TRAIL binding is followed by receptor trimerization, and groups of trimerized death receptors can further cluster together to form bigger aggregates. These aggregates predominantly accumulate in lipid raft microdomains in the plasma membrane (16). The molecular composition of the receptor trimers has been controversially discussed, and it has not been fully resolved whether, at physiologic conditions, exclusively homotypic interactions or also heterocomplexes can be observed (17–20). Following trimerization of TRAIL-R1 or TRAIL-R2, the adaptor protein Fas-associated death domain containing protein (FADD) can bind to the intracellular death domains of the receptors and promote the recruitment and activation of initiator caspase-8 within the death-inducing signaling complex (DISC; Fig. 1D). In most cells, caspase-8 then initiates apoptosis through Bid cleavage and the mitochondrial apoptosis pathway (Fig. 1E). Both the formation of the caspase-activating DISC, as well as the subsequent signaling network leading to apoptosis execution, can be modulated through a multitude of complex regulatory processes (21, 22).

Additional TRAIL receptors, which are incapable of transducing the death signal because they lack a functional intracellular death domain, have been identified. These receptors are TRAIL-R3, TRAIL-R4, and osteoprotegerin (22). TRAIL-R3 and TRAIL-R4 are alternatively also referred to as decoy receptors 1 and 2 (DcR1, DcR2). TRAIL-R4 (DcR2) bears huge similarities to TRAIL-R1 and TRAIL-R2; however, it contains only a truncated cytosolic death domain (Fig. 1A; ref. 23). TRAIL-R3 (DcR1) consists of an extracellular cysteine-rich structure that resembles the proapoptotic TRAIL receptors and is associated with the plasma membrane through a COOH-terminal glycosyl-phosphatidylinositol anchor (Fig. 1A; refs. 11, 24). It is still highly debated whether overexpression of TRAIL-R3 or TRAIL-R4 correlates with cellular apoptosis resistance upon TRAIL exposure, and high expression levels are rarely found naturally in isolated cancer cell lines. The exact molecular details of how decoy receptors inhibit TRAIL-induced apoptosis in addition to their obvious role as TRAIL scavengers are not fully resolved and may differ between TRAIL-R3 and TRAIL-R4. For example, it was reported that TRAIL-R3 exclusively forms homotrimers upon TRAIL binding, whereas TRAIL-R4 may also aggregate into heterotrimers with activated TRAIL-R1 and TRAIL-R2 (Fig. 1B; ref. 25). Such heterotrimers are incapable of forming a functional DISC required for apoptosis initiation. Relatively little is known about osteoprotegerin, the fifth receptor capable of TRAIL binding. Osteoprotegerin negatively regulates osteoclastogenesis and is largely secreted as a soluble protein that might act as a scavenger for soluble TRAIL (26). Uncertainties remain about the molecular mechanisms of death and decoy receptor interaction, and it remains to be conclusively shown whether the relative abundance of death versus decoy receptors could serve to indicate cellular susceptibility to TRAIL-induced apoptosis.

TRAIL treatment strategies largely build on soluble variants of recombinant human TRAIL (rhTRAIL, a 60-kDa homotrimer TRAIL protein without its membrane anchor), as well as on monoclonal agonistic antibodies, which specifically target death receptor TRAIL-R1 or TRAIL-R2, thereby potentially circumventing decoy receptor–mediated resistance, and which in addition have the benefit of a significantly longer plasma half-life (27, 28). Additionally, to increase the local concentration and cancer cell–directed cytotoxicity of TRAIL, fusion proteins have been designed to translocate TRAIL to specific tissues or cells, for example, the epidermal growth factor receptor (EGFR)–selective delivery of soluble TRAIL, combined with an EGFR-targeting antibody fragment (29). Soluble TRAIL fused to a melanoma-associated chondroitin sulfate proteoglycan (MCSP)–specific antibody fragment combines inhibition of MCSP tumorigenic signaling with TRAIL-mediated apoptosis induction in melanoma cells (30). A multitude of studies using models of human tumor xenografts in mice showed a high anticancer activity of TRAIL or receptor-specific ligands in vivo, and TRAIL efficacy can be potentiated in combination treatments with kinase and proteasome inhibitors, genotoxic drugs, histone deacetylase inhibitors, and others drugs (20, 31).

It has frequently been debated whether TRAIL can be cytotoxic to untransformed cells in vivo. In contrast to the high tolerability of TRAIL found in the seminal studies of Ashkenazi and colleagues and Walczak and colleagues (3, 4), other studies reported toxicities of TRAIL in isolated human hepatocytes (32, 33) and human brain tissue (34). However, significant TRAIL hepatotoxicities seem to be restricted to ex vivo cultures of primary liver cells or to highly aggregated TRAIL variants carrying flag or polyhistidine tags (35, 36). In contrast, rhTRAIL, as well as leucine or isoleucine zipper variants of TRAIL, was shown to be well tolerated in animal models. This finding is in stark contrast to the generally high hepatotoxicities that were reported for death ligands FasL and TNFα (3, 4). As a note of caution, however, it was shown that liver steatosis or hepatitis C infection can increase human hepatocyte sensitivity toward TRAIL (37), suggesting that patients from these risk groups should be excluded from TRAIL-based therapies.

Soluble rhTRAIL (also named dulanermin), the TRAIL-R1–targeting monoclonal agonistic antibody mapatumumab, and the TRAIL-R2–targeting monoclonal agonistic antibodies lexatumumab, conatumumab, tigatuzumab, and DAB4 (also named PRO95780) have entered clinical studies. A large number of phase I and phase II clinical trials have been undertaken with these agents by now or are still ongoing, either as monotherapy or in combination with other chemotherapeutic drugs. These trials target a wide range of both solid and nonsolid malignant neoplasms.

Toxicity studies as part of an rhTRAIL phase I trial provided encouraging results. In a dose-escalating monotherapy in patients with advanced cancers, rhTRAIL was well tolerated at serum concentrations that were shown to be effective against cancer cells in preclinical models (38). Similarly, TRAIL-R1–specific mapatumumab seems to be well tolerated, and in a phase II study in patients with relapsed or refractory non-Hodgkin lymphoma, approximately one third of all patients responded to mapatumumab (28, 39, 40). In a phase II clinical trial, mapatumumab was well tolerated in patients with colorectal cancer (41). Lexatumumab, conatumumab, tigatuzumab, PRO95780, and other TRAIL-R2–targeting antibodies, likewise, were investigated in phase I and II clinical trials (42–44). Like mapatumumab, these antibodies only caused low-to-moderate side effects in a small number of patients. Several patients showed partial responses or attenuated disease progression when treated with TRAIL-R2–targeting antibodies (45, 46). In general, TRAIL and TRAIL receptor–binding antibodies, therefore, seem to be well tolerated, and only a few cases of adverse responses have been documented. Furthermore, promising results in patients with advanced or refractory cancers underline the therapeutic potential of TRAIL receptor–targeting molecules. However, as a caveat, signaling through TRAIL death receptors may also induce antiapoptotic responses and proliferation signaling. This alternative response was shown to require the activation of kinases, such as RIP1, IKK, c-jun-NH₂-kinase, or p38 in signaling complexes secondary to the DISC (47) and to induce transcription-dependent prosurvival responses that attenuate apoptosis signaling. Indeed, subpopulations of primary cells isolated from children with acute leukemia showed increased proliferation rates following TRAIL treatment (48). These prosurvival responses may, therefore, limit the usability of TRAIL in monotherapies.

Cancers like melanoma, leukemia, multiple myeloma, breast, bladder, prostate, renal, colon, and others would be expected to be largely TRAIL sensitive judging on the basis of preclinical drug response data. However, cases of TRAIL resistance are frequently being observed and often could be associated with the overexpression of antiapoptotic proteins or the low expression of TRAIL receptors. An attractive and preclinically successful strategy, therefore, is to identify combination treatments that sensitize otherwise resistant cancers to TRAIL. Although this strategy holds risks because of the potential sensitization of nontransformed cells to the treatment, initial results are very promising. In a phase Ib clinical study to determine safety, pharmacokinetics, and maximum dose tolerance levels, patients with non–small cell lung cancer were treated with rhTRAIL (dulanermin), combined with paclitaxel, carboplatin, and bevacizumab. No dose-limiting toxicity was observed, and the treatment was well tolerated, with 1 complete response and a high percentage of partial responses observed (49). Other early phase Ib trials assessing combination treatments with mapatumumab and paclitaxel or carboplatin, as well as mapatumumab together with gemcitabine or cisplatin, in patients with advanced solid tumors reported high tolerability of mapatumumab (50, 51). By now, many more additional phase I and II clinical trials have been conducted or are still ongoing in other cancer types and using additional U.S. Food and Drug Administration–approved drugs, such as kinase inhibitors, proteasome inhibitors, histone deacetylase inhibitors, or genotoxic drugs, in combination with TRAIL (Table 1; ref. 45). Even though cases of adverse effects and toxicities were reported, overall, TRAIL combination therapies seem to be remarkably well tolerated. Still, outstanding results from ongoing phase II clinical trials will soon allow assessment of whether patients may benefit from such alternative treatment regimes and whether patient responses are indicative of multidrug synergies that were reported in vitro. The mechanisms bringing about synergistic multidrug responses are very complex. However, significant progress in understanding the underlying signaling cross-talk has been made in recent years, as outlined in the following discussion.

Synergistic effects of chemotherapeutics in combination with TRAIL have been reported for various classes of drugs. In general, synergies arise from nonlinear cross-potentiation of drug effects. In the framework of apoptosis signaling, such potentiation may arise from the (often p53-dependent) transcriptional upregulation of proapoptotic proteins, from their accumulation due to impaired protein degradation, from alterations in regulatory posttranslational protein modifications, or from combinations of these processes. In the following paragraphs, we discuss the synergy mechanisms exerted by different drug classes. Related to this discussion, an overview of clinical trials aiming to use drug synergies in TRAIL-based combination treatments is provided in Table 1. Information on additional novel and promising cotreatment strategies that are still at the preclinical stage is provided in Supplementary Table S1.

Different genotoxic drugs have been suggested to enhance TRAIL signaling and its anticancer potential, with 5-fluorouracil and cisplatin having been studied most intensely. A typical stress response to genotoxic monotherapies is the transcriptional induction of BH3-only proteins, such as Puma, which is a potent inducer of apoptosis through the mitochondrial pathway (52). Such responses may synergistically complement the BH3-only protein Bid, which is cleaved and activated by caspase-8 in response to TRAIL. In this respect, similar synergy mechanisms would be relevant in cotreatment scenarios based on synthetic BH3-only protein mimetics, such as AT-101 (a gossypol derivate), ABT-263, ABT-737, and GX-15-070 (obatoclax; ref. 53). These novel drugs are already undergoing evaluation in monotherapy trials and may hold great potential for future combination treatments. However, synergies achieved with genotoxic drugs may also arise from upstream modulation of TRAIL signaling. For example, 5-fluorouracil can prime TRAIL-resistant hepatocellular carcinoma cells to apoptosis, at least partially, through the downregulation of the inactive caspase-8 homolog cFlip (Fig. 1C) and through the parallel upregulation of TRAIL-R2 at physiologically relevant doses (54). Importantly, cytotoxicity to normal cells seems to be limited to acceptable levels in this scenario (54). Similar scenarios of death receptor upregulation or cFlip downregulation were also reported in response to other genotoxic drugs, as well as to ionizing irradiation and epigenetic inhibitors, such as histone deacetylase inhibitors (55, 56). Also IFNs, which are frequently used in hematologic cancer therapy, were shown to potently promote TRAIL-induced apoptosis through evoking transcriptional responses. For example, IFN-α exposure can increase TRAIL and TRAIL-R2 expression in human hepatocellular carcinoma cells (57, 58), whereas IFN-γ can reduce cFlip protein levels (59).

Prolonged proteasome inhibition induces the expression and accumulation of BH3-only proteins, such as Bim, Bik, Puma, and Noxa (60), which leads to synergy mechanisms like those described above for genotoxic drugs. Furthermore, TRAIL-induced NF-κB activation and prosurvival signaling can be limited by proteasome inhibition because the degradation of NF-κB inhibitor IκBα is prevented (23, 61). Maybe more importantly, proteasome inhibition can lead to the direct accumulation of proteins that promote TRAIL-induced apoptosis. An accumulation of both TRAIL-R1 and TRAIL-R2 was described in response to proteasome inhibition in various cancers (62–64), potentially resulting in more efficient DISC formation and caspase activation upon TRAIL addition. The consequences arising from a general inhibition of protein degradation are naturally complex, due to the global disturbance of relative protein abundances in addition to active transcriptional stress responses. Attributing synergies in TRAIL and/or proteasome inhibitor cotreatments to the modulation of single proteins or processes, as proposed in several studies, therefore likely presents an oversimplification of the underlying molecular mechanisms that enhance TRAIL responsiveness. Nevertheless, the fact that proteasome inhibition can promote proapoptotic signaling, at least partially, independent of protein neosynthesis makes proteasome inhibitors particularly attractive for the treatment of cancers that present with deficiencies in transcriptional proapoptotic responses, as can arise from loss-of-function mutations in tumor suppressor p53.

Transcription- or p53-independent synergisms of chemotherapeutics and TRAIL can also be achieved by targeting posttranslational protein modifications such as phosphorylation patterns. For example, inhibiting protein kinase CK2 impairs Bid phosphorylation and enhances Bid cleavage by caspase-8 (65, 66) and may be an attractive cotreatment strategy if outstanding safety tests yield satisfying results. Procaspase-8 itself can be phosphorylated by Src kinase, which impairs its binding to and activation at the DISC (67). In addition, formation of the DISC can be modulated upstream of caspase-8 recruitment through mitogen-activated protein kinases (MAPK; ref. 68). Furthermore, protein kinase C activity limits FADD recruitment into the DISC (69). Kinase inhibition can also directly or indirectly induce transcriptional responses, adding further complexity. For example, inhibiting phosphoinositide 3-kinase (PI3K)/AKT signaling by perifosine can restore TRAIL sensitivity in acute myelogenous leukemia through a p53-independent TRAIL-R2 upregulation and a concomitant cFlip and X-linked inhibitor-of-apoptosis protein (XIAP) downregulation (70). Multitarget kinase inhibitors, such as sorafenib, may therefore hold great potential in enhancing cancer cell responsiveness and are currently being investigated in combination treatments with TRAIL in phase I clinical trials (45). Furthermore, additional posttranslational modifications of the death receptors, such as palmitoylation, nitrosylation, and glycosylation, have been reported, and these modifications can influence cellular sensitivity toward TRAIL (22). Whether these modifications can be therapeutically targeted and used is currently unknown.

Antagonists of inhibitor-of-apoptosis proteins, such as synthetic Smac peptides and Smac mimetics, were shown to sensitize various cancer cell lines to TRAIL-induced apoptosis. In particular, antagonizing XIAP, the most potent inhibitor of executioner caspases, promotes the direct activation of caspase-3 by caspase-8 and results in efficient apoptosis execution (71, 72). Furthermore, Smac mimetics also bind to cellular IAP (cIAP)–1 and cIAP-2. As a consequence, cIAPs are rapidly degraded and cells may respond with secreting TNFα (73, 74). TNFα then may further enhance apoptosis by activating a parallel extrinsic apoptosis pathway that leads to caspase-8 activation. IAP inhibitors have been shown to synergistically kill pancreatic cancer cells in combination with the TRAIL-R1–targeting antibody mapatumumab, whereas the combined treatment with the TRAIL-R2–specific antibody lexatumumab was less efficient, indicating receptor-specific differences in synergy and susceptibility of certain cancers (75). Besides the requirement to better understand and use synergy mechanisms, means to predetermine differential sensitivity to either TRAIL-R1–or TRAIL-R2–transduced apoptosis will be paramount to case specifically identify the most effective treatment strategies in the future.

Experimental preclinical as well as clinical evidence shows that TRAIL has a high potential as a novel anticancer drug, both in monotherapies and in combination treatments (Table 1 and Supplementary Table S1). Given the comprehensive research activities currently focusing on identifying and deciphering the complex intracellular signaling cross-talk that evokes synergies in TRAIL combination treatments, we will soon better understand which conditions need to be fulfilled to more efficiently use the potential of different TRAIL-based treatment strategies. At this stage, TRAIL signaling is already one of the best-characterized apoptotic signaling pathways. In the coming years, this information may greatly assist in identifying potential biomarkers that will allow researchers to rationally predict tumor responsiveness to TRAIL-based therapies. This important, but still outstanding milestone, will need to be achieved before patient benefits can be maximized through appropriately tailored drug administration.

No potential conflicts of interest were disclosed.

This work was supported by grants from Science Foundation Ireland (05/RFP/BIM056), the Health Research Board Ireland (RP/2006/258; RP/2008/7), and the Irish National Biophotonics and Imaging Platform funded under the Irish Higher Education Authority Programme for Third Level Institutions (HEA PRTLI) Cycle 4, cofunded by the Irish Government and the European Union “Investing in your future.”