Inducing death in tumor cells: roles of the inhibitor of apoptosis proteins

The heterogeneous group of diseases collectively termed cancer results not just from aberrant cellular proliferation but also from a lack of accompanying homeostatic cell death. Indeed, cancer cells regularly acquire resistance to programmed cell death, or apoptosis, which not only supports cancer progression but also leads to resistance to therapeutic agents. Thus, various approaches have been undertaken in order to induce apoptosis in tumor cells for therapeutic purposes. Here, we will focus our discussion on agents that directly affect the apoptotic machinery itself rather than on drugs that induce apoptosis in tumor cells indirectly, such as by DNA damage or kinase dependency inhibition. As the roles of the Bcl-2 family have been extensively studied and reviewed recently, we will focus in this review specifically on the inhibitor of apoptosis protein (IAP) family. IAPs are a disparate group of proteins that all contain a baculovirus IAP repeat domain, which is important for the inhibition of apoptosis in some, but not all, family members. We describe each of the family members with respect to their structural and functional similarities and differences and their respective roles in cancer. Finally, we also review the current state of IAPs as targets for anti-cancer therapeutics and discuss the current clinical state of IAP antagonists.


Introduction to inhibitor of apoptosis family of proteins
The inhibitor of apoptosis protein (IAP) family is a functionally and structurally related group of proteins that serve as endogenous inhibitors of programmed cell death, or apoptosis. In addition, some family members are regulators of another form of programmed cell death termed necroptosis 1,2 . Whilst some of the IAPs have also been shown to be involved in immune regulation 3,4 , chromosome segregation, and cytokinesis 5-7 , this review will focus on their roles in explicitly regulating apoptosis. Although the various IAPs have somewhat differing functions, they are linked by one unique domain: membership of the IAP family is ascribed if the gene/protein in question possesses a baculovirus IAP repeat (BIR) domain. Indeed, as the name suggests, BIR domains were first identified in a baculoviral protein capable of inhibiting cell death in virally infected cells [8][9][10] . BIR domains are zinc finger domains and invariantly contain three cysteines and one histidine, which co-ordinate the zinc ion 10 , and these domains are involved in various protein-protein interactions (PPIs). IAPs were subsequently identified and characterized by various techniques in yeast, worms, insects, and mammalian cells 5,11-15 . The first human IAP revealed was neuronal apoptosis inhibitory protein (NAIP or BIRC1), which was serendipitously discovered in a search for genes involved in the autosomal recessive condition spinal muscular atrophy (SMA) 16 . The next human IAPs to be characterized were the cellular IAPs 1 and 2: cIAP1 (or BIRC2) and cIAP2 (or BIRC3). These proteins were discovered to have a role in tumor necrosis factor receptor (TNFR) signaling through association with the adaptor proteins TRAF1 and TRAF2 17-20 . The fact that several proteins shared the common BIR domain led to the identification of more family members via traditional homologymatching database searches (reviewed in 21). Notably, many of these proteins were further shown to be involved in the regulation of apoptosis 15,[22][23][24][25][26][27] . Rounding out the group of eight human BIRcontaining proteins are XIAP (BIRC4), Survivin (BIRC5), Apollon (BIRC6), Melanoma IAP (ML-IAP or BIRC7), and IAP-like protein 2 (ILP-2 or BIRC8) 15,22,25,[28][29][30][31][32][33][34][35][36][37] . A schematic of the general IAP family structure is shown in Figure 1. Figure 2 shows the intracellular signaling interplay of IAPs with respect to cell survival and apoptosis.
Survivin and ML-IAP are small proteins with only one BIR domain, yet their functions are enigmatic, having been ascribed to various processes, including apoptosis inhibition. As such, their roles, especially with respect to cancer, will be discussed in greater detail later in this review. ML-IAP additionally possesses an E3 ubiquitin ligase domain named RING (really interesting new gene), a domain also present in other IAPs and believed to be important in many signaling events [38][39][40][41] . ILP-2, likewise with only one BIR domain, also contains a RING domain and a ubiquitinassociated (UBA) domain. Whilst ILP-2 shows high homology to XIAP, it is a product of a separate gene and its expression in healthy tissues appears to be restricted to the testes 36 . Overexpression studies have shown that ILP-2 has no effect on extrinsic death receptor-induced apoptosis but that it can inhibit intrinsic (also known as mitochondrial) apoptosis through a potential interaction with caspase-9, an apical protease involved in mitochondrial apoptosis 36 . However, others have shown that its BIR domain is unstable and, as such, it is only a weak binder of caspase-9, at least in the absence of other cellular factors 42 . The representation of the homology between the different BIR domains of the IAP family reflects the accepted designation of BIR1, BIR2, and BIR3. The BIR domains of Survivin (BIRC5) and Apollon (BIRC6) can be aligned with either BIR1 or BIR2, depending on the specific alignment criteria, but owing to their uniqueness they are colored and labeled accordingly. CARD, caspase recruitment domain; cIAP1, cellular inhibitor of apoptosis protein 1; cIAP2, cellular inhibitor of apoptosis protein 2; ILP-2, inhibitor of apoptosis protein-like protein 2; LRR, leucine-rich repeat; ML-IAP, melanoma inhibitor of apoptosis protein; NAIP, neuronal apoptosis inhibitory protein; RING, really interesting new gene; UBA, ubiquitin-associated; UBC, ubiquitin-conjugating. Apollon is an extremely large protein (approximately 528 kDa) containing only one BIR domain that is thought to be membraneassociated as well as a C-terminal ubiquitin-conjugating (UBC) domain 43 . It has been shown to attenuate apoptosis 44-46 and to directly engage and interfere with both the second mitochondriaderived activator of caspases (Smac, discussed in greater detail below) and caspase-9 43,47 . Others have confirmed that Apollon is involved in caspase-9-mediated apoptosis but that it can also regulate p53 and is essential in murine embryo development 48 .
The remaining four IAPs each possess three BIR domains in tandem and are the most studied members of the IAP family. Whilst XIAP and cIAP1 and 2 each contain a UBA and a RING domain, NAIP differs in that it has neither of these features but instead contains a "NACHT" domain and a C-terminal leucinerich repeat (LRR). The NACHT domain is so named because of its presence in NAIP, C2TA, HET-E, and TEP1, and it is predicted to be a nucleoside-triphosphatase (NTPase) domain 49 . Whilst its original discovery as the causative gene in SMA proved erroneous, NAIP has been shown to attenuate apoptosis in multiple models 15,50,51 . The main role of NAIP, however, appears to be in the regulation of innate immunity. Thus, NAIP, which is also part of the NOD-like receptor (NLR) family, is important for NLRC4 inflammasome activation in response to certain bacterial ligands 52-54 . cIAP1 and 2, whilst showing similar architecture to XIAP, also possess a caspase recruitment domain (CARD). Somewhat confusingly, however, the CARD of the cIAPs does not bind to caspases, but it appears to function in an auto-inhibitory manner to block the cIAP RING domain's E3 ubiquitin ligase activity 55 . The cIAPs are structurally very similar to each other with only a short linker sequence difference, are functionally redundant 56,57 , and are believed to have resulted from a recent evolutionary gene duplication. As with XIAP, they contain three BIR domains, and BIR1 is essential for binding to the TNFR adapter protein TRAF2 58,59 . The third BIR domain (BIR3) in these three proteins, as well as the homologous BIR domain in ML-IAP, all potently bind to Smac, a negative regulator that will be discussed in much greater detail in subsequent sections of this review.
The cIAPs have been highly characterized in signaling events associated with a subset of TNFR superfamily members, called the death receptors (DRs), and it appears that their E3 ubiquitin ligase activity is especially pertinent in this regulation. DRs are categorized on the basis of the presence of a so-called death domain (DD). DDs are approximately 80-amino-acid alpha-helical structures that recruit adapter proteins capable of binding multiple other proteins in supramolecular complexes that regulate distinct signaling pathways based on composition (reviewed in 60-63). cIAPs recruited to these complexes can be involved in both degradative K48-and non-degradative K63-branched ubiquitination. Indeed, cIAP1 has been shown to control the levels of cIAP2 via degradative signaling, as depletion of cIAP1 results in a "rebound" of cIAP2 levels 64 . Similarly, levels of nuclear factorkappa B (NF-κB)-inducing kinase (NIK or MAP3K14) are tightly controlled by cIAPs, and these protein levels are almost undetectable when the E3 ubiquitin ligase activity of the proteins is available 64-66 . Much more significant, however, are the cIAPmediated non-degradative K63-branched ubiquitination and ensuing signaling. This ubiquitination of receptor-interacting protein 1 (RIP1) results in the formation of a signaling complex that can recruit further ubiquitin ligases and kinases that ultimately result in classical NF-κB activation 67,68 (and reviewed in 69). Indeed, recruitment of RIP1 to these complexes has led to the coining of the term "RIPoptosome" to describe them 70-72 . When cIAPs are absent-owing to genotoxic stress or chemical depletion with Smac mimetics (see below), for example-and the relevant receptor agonist is engaged, RIP1 is not degraded but forms a death signaling RIPoptosome 72 with apoptosis effected via the apical caspases-8 or -10 or both. Furthermore, in the absence of these caspases (or upon their inhibition), necroptosis can occur 1,2 . Necroptosis has been demonstrated to be dependent on RIP1 and specifically on its kinase activity (reviewed in 73). RIP1 phosphorylation of RIP3 results in the activation of mixed lineage kinase domain-like protein (MLKL) 74-76 , which induces necroptotic death by rupturing of the plasma membrane 77-79 .
In sum, the cIAPs are integral components of multiple signaling complexes emanating from TNFR superfamily members and, as a consequence, can regulate diverse cellular responses such as cell survival, apoptosis, and necroptosis via the RIPoptosome 80,81 .
XIAP is by far the most studied and highly characterized member of the IAP family. It is a potent inhibitor of apoptosis as judged by multiple model systems and techniques and has been clearly demonstrated to effect such inhibition due to direct binding of caspases 24 . BIR2 and a short linker section between BIR1 and BIR2 are essential for binding and sequestration of the effector caspases -3 and -7 24,82-84 , whilst BIR3 is crucial for binding to the apical caspase-9 85-87 . As with the cIAPs, the BIR3 of XIAP also binds Smac, and this interaction results in caspase de-repression 85,88-90 . Thus, XIAP BIR3 binding of Smac has been shown to result in the release of active caspases from the XIAP protein complex and thus BIR3-Smac interaction is permissive for apoptosis induction 88,91 . As such, Smac is not a direct activator of caspases, despite its name, but rather an "inhibitor of the inhibitor". Smac effects this displacement of factors from the BIR domains because of a four-amino-acid sequence of Ala-Val-Pro-Ile (AVPI) in Smac. Exposure of cells to this peptide motif can therefore sensitize cells to apoptotic stimuli or, in the case of cIAPs, result in their auto-degradation and subsequent switch from inhibitory to pro-apoptotic events from TNFRs. Owing to these effects, the AVPI tetrapeptide sequence has drawn much attention as a potential anticancer agent, and multiple Smac mimetics have been developed with a view to promoting apoptosis in tumor cells, where normal apoptotic signaling is perturbed. The current clinical progress of these agents is described in detail later.
In summary, the IAP family, whilst small in number, contains a series of diverse members with differing but somewhat overlapping biological roles. The most relevant of these roles in tumors is apoptosis inhibition, and the mechanisms governing how each member is involved are somewhat unique. The next section will discuss the roles of these proteins in cancer, and finally we will discuss the application of IAP inhibitors (Smac mimetics) as potential anti-cancer agents.

Inhibitor of apoptosis and cancer
The evasion of apoptosis is one of the hallmarks of cancer 92-95 , and, as noted above, the IAP family of proteins plays an important role in attenuating programmed cell death pathways, predominantly through modulation of the caspase cascade (extensively reviewed in 27,96-103). Furthermore, IAPs are often upregulated in cancers 104 and are believed to underlie the resistance of many tumors to chemotherapeutics 105,106 . Ablation or antagonism of IAPs is therefore an attractive strategy to sensitize or re-sensitize tumor cells to apoptosis induced by other agents. The roles that the eight IAPs found in humans play in cancer are discussed below.
NAIP NAIP (BIRC1) was first identified and named in 1995 by Roy et al. 16 as a potential modulator of the neuronal apoptotic pathway. As noted earlier, the main biological role for NAIP appears to be the regulation of innate immunity. Nevertheless, NAIP has been weakly linked to unfavorable prognosis in esophageal cancer 107 , breast cancer 108 , prostate cancer 109 , and neuroblastoma 110 . The precise role of NAIP in the dysregulation of apoptosis in cancer and its value as a potential therapeutic target need further study.
cIAP1, cIAP2, and XIAP (BIRC2, BIRC3, and BIRC4) As noted above, XIAP is a very potent binder and inhibitor of caspase-3. Accordingly, research by pharmaceutical companies has primarily focused on antagonizing this protein for oncology applications (reviewed in 111,112). cIAP1 and cIAP2 have also been implicated in cancer, and their role in the modulation of the NF-κB signaling pathway has been investigated in detail 113 . It was also found that cIAP1 can protect cancer cells from the lethal effect of TNF through synergy with the MYC oncogene, thus driving tumorigenesis 114-116 . As cIAPs suppress TNF-induced cell death, it is likely that increased levels of cIAPs support tumor cell survival by modulating cellular responses to TNF. cIAPs and XIAP are additionally thought to contribute to cancer cell invasion and metastasis through their ability to drive NF-κBmediated expression of genes involved in cell motility, migration, and invasion 117,118 . Similarly, in lymphomas, cIAP2 is often found as a fusion protein with mucosa-associated lymphoid tissue 1 (MALT1), resulting in the activation of NF-κB signaling (reviewed in 119). IAPs have been shown to be overexpressed in many cell lines from the NCI60 panel as compared with the corresponding normal tissue 120 . XIAP overexpression in turn has been reported in childhood acute myeloid leukemia (AML) 121 , renal carcinoma 122,123 , multiple myeloma (MM) 124 , and bladder cancers 125 , and AML patients with low levels of expression of XIAP were shown to have a statistically significant survival advantage compared with those patients with higher levels 120,126 . In summary, the role and importance of these members of the IAP family of proteins in cancer have been extensively investigated and reviewed.
Survivin Survivin (BIRC5) was first identified by Altieri et al. as an anti-apoptosis gene expressed in various cancer cells 25,127 . Survivin is an example of one of the earliest IAP proteins strongly implicated in oncogenesis 128 and has been well established as a prognostic marker with a negative correlation on outcome in many cancers 129-133 (reviewed in 134-138). Consistent with this, excessive levels of Survivin inhibit both intrinsic and extrinsic pathways of apoptosis 25,139-141 . Of note, however, Survivin is only a weak apoptosis inhibitor at physiological concentrations and may in fact exert anti-apoptotic activity through stabilization of XIAP 142 . Recent studies have shown Survivin to be an important regulator of cell division, and this appears to be its main biological function 5-7,143 . The role of Survivin in cancer has recently been reviewed extensively 144-147 . Therapeutic targeting of Survivin has been mostly confined to non-small-molecule strategies 148,149 and repression of protein translation 150-152 (reviewed in 153), and only recently have small-molecule inhibitors been reported 154,155 .

Apollon
Apollon (BIRC6, the human homolog of murine BRUCE) was first identified in 1999 by Chen et al. 32 as a marker in brain and ovarian cancer cell lines that is linked to resistance to various anticancer drugs. A number of subsequent studies have concluded that elevated levels of Apollon are linked to poor prognosis in a range of cancers, such as leukemia 156,157 , breast 158 , neuroblastoma 159 , prostate 160-162 , lung 163 , ovarian 164 , colorectal 165 , hepatocellular 166 , and head and neck 167 cancers. It is largely understood that the role of Apollon as an oncogene is centered on its role in modulating Smac and caspase-9 levels, where overexpression of Apollon leads to increased silencing of apoptosis through Smac degradation as well as to attenuation of the caspase cascade by targeting caspase-9 for ubiquitination and subsequent degradation 43,45,47 . Based on these predictive findings, the role of Apollon as a therapeutic target has been evaluated in a number of studies using functional genomic approaches, since no appropriate smallmolecule tool has yet been developed 158,159,166,168 . Undoubtedly, the development of potent and selective small-molecule antagonists to Apollon will allow detailed elucidation of its potential as a therapeutic target in oncology.

ML-IAP
ML-IAP (BIRC7, also known as Livin or KIAP) was first identified as a member of the IAP family because of its single BIR domain 33,35 . The ML-IAP BIR domain is also responsible for apoptosis inhibition, and small molecules that target this region could potentially re-sensitize cancer cells to chemotherapeutics.
In particular, the RING domain of ML-IAP has been shown to function as an E3 ubiquitin ligase facilitating the ubiquitination and subsequent degradation of itself 169,170 and, more importantly, of Smac 170 , the natural caspase antagonist that modulates apoptotic signaling. Thus, inhibition of ML-IAP leads to a direct increase of Smac and a re-sensitization of cells to apoptotic stimuli. Both protein and mRNA levels of ML-IAP are low to undetectable in most adult tissues 171 but are highly expressed in several cancers 33,171-179 , including various lung cancers, melanoma, liver cancer, glioblastoma, and oral squamous cell carcinoma. This protein is also highly expressed in renal cell carcinoma 180,181 , and this is why the original name of kidney IAP (KIAP) was coined. ML-IAP maps to chromosome 20q13, a region frequently implicated in the mutagenic etiology of lung cancers 33 . ML-IAP levels have been shown to be highly relevant as a prognostic biomarker in lung 172,173,182,183 and other 174,175,177,179,180,184-187 cancers. These studies have consistently reported that high ML-IAP expression correlates with a poor outcome but that lower levels predict a more favorable prognosis. A number of recent studies have clearly shown the considerable therapeutic potential of ML-IAP inhibition to treat cancer. A wealth of data has been presented in cellular contexts [188][189][190][191][192][193][194][195] as well as in xenograft studies 196,197 . In particular, the mouse xenograft studies by Chen et al. 196 and the cell-line-based work by Zhuang et al. 198 showed a substantial benefit gained from BIRC7 gene ablation in models of lung cancer. However, all of these studies inhibited ML-IAP through RNA knockdown approaches because of the unavailability of a selective and potent smallmolecule antagonist. Recently, however, potent and uniquely selective ML-IAP inhibitors have been reported, which will help more comprehensive elucidation of the role of ML-IAP in cancers 199 .
ILP-2 ILP-2 (IAP-like protein-2 or BIRC8) was originally detected only in the testis and lymphoblastoid cells 36 . However, some recent work has established a tenuous link to breast cancer 200 , and it will be of interest to see whether this link gains further support to establish ILP-2 as a novel biomarker in human malignancies as well as a potential target for therapy.

Inhibitor of apoptosis inhibitor development for cancer therapeutics
In the mid-1990s, it was shown that the BIR domains were necessary and responsible for the anti-apoptotic and caspasesuppressing activity of the IAP proteins 10,14,84 . With the subsequent discovery of the endogenous IAP ligand Smac in 2000 88,201 , the path toward the development of small-molecule inhibitors of the IAPs unfolded. Historically, however, the development of smallmolecule inhibitors of such PPIs has been quite difficult. Most of these interactions are devoid of the classic druggable binding pockets (about 300-500 Å 2 ) with which most drug discovery scientists are familiar 202 . Rather, these PPIs typically derive their binding energy from a large number of intermolecular interactions along a relatively flat and large (about 1,000-2,000 Å 2 ) surface.
It was a critical observation made by Xiadong Wang et al. regarding the loss of Smac activity upon the addition of a glutathione s-transferase (GST) fusion to its N-terminus that paved the way for the current crop of Smac mimetics 203 . Mutation studies further confirmed the importance of the posttranslationally processed and flexible N-terminus of mature Smac. Perhaps equally important was the contribution from Fesik et al. that year, generating the first nuclear magnetic resonance structure of truncated Smac bound to one of the IAPs, XIAP BIR3 89 . Specifically, four residues, AVPI, that bind to a surface groove on the IAP BIR domains proved indispensable for activity. As shown in Figure 3, there exists in the IAP BIR domains a negatively charged cleft of perfect size to accept the alanine. Furthermore, the proline of Smac allows for a crucial reverse turn feature to maintain close contacts with the binding site. These are two key elements represented in nearly all of the reported IAP inhibitors. Early on, several groups showed that synthetic oligopeptides (4-9-mers) exhibit better binding affinity than native Smac for XIAP BIR3 and are notable for their apoptosis-inducing ability [204][205][206] . These oligopeptides served an important role as a drug discovery proof-of-concept: that mimicking a small portion of Smac is a viable strategy to target the IAPs. Subsequent reports took this concept a step further and focused on developing more drug-like peptidomimetics of the N-terminal AVPI tetrapeptide binding motif to disrupt the IAP-caspase PPI, and thus far this has proven to be the most popular and successful tactic. The first true medicinal chemistry work reported by Fesik et al. in 2004 207 laid the groundwork for the advances that would follow in subsequent years, and, also in 2004, seminal work from Wang and Harran showed that a small-molecule Smac mimetic could potentiate TNF-induced and TNF-related apoptosis-inducing ligand (TRAIL)-induced apoptosis 208 . A summary of the collective structure-activity-relationship (SAR) conclusions from Smac mimetic medicinal chemistry work is shown in Figure 4.
A number of research groups from both academia and industry have initiated programs in the space since these early discoveries, focusing on Smac mimetics 66,199, (also reviewed in 257,258). Some of these compounds remain in pre-clinical testing, whereas others have entered but are no longer active in clinical trials. Our laboratories are currently testing a series of Smac mimetics developed by us at Sanford Burnham Prebys Medical Discovery Institute. A representative compound with encouraging preclinical data in several cancer cell lines is shown in Figure 5 199 . SBI-0636457 has demonstrated potent cell-killing effects in several subtypes of breast, ovarian, and prostate cancer cell lines but only when the DR ligand TRAIL or another such apoptosis inducer is co-administered. Furthermore, SBI-0636457 administered as a single agent exhibited no toxicity to normal human fibroblasts.
Bivalent Smac mimetics take advantage of the homodimeric nature of native Smac and are able to bind both the BIR2 and the BIR3 domains. The consequence of this improved binding mode is poorer drug-like properties, as the Smac mimetics must adopt a larger molecular size in order to access both binding sites. Impressive binding data (K d = 300 pM for the BIR2-BIR3

Current clinical status of inhibitor of apoptosis inhibitors in oncology
In the US, several monovalent Smac mimetic compounds and one bivalent compound have entered the clinic and are still active in clinical trials (Figure 6). All of the compounds for which clinical data have been reported so far demonstrated generally favorable safety profiles in phase I, and amylase/lipase elevation, alanine and aspartate transaminase (ALT and AST) elevation, cytokine release syndrome (CRS), and Bell's palsy were the dose-limiting serious adverse events 112 . Of note, however, the Bell's palsy toxicity has been observed only with bivalent and not with monovalent Smac mimetics. It has been suggested that CRS may result from the Smac mimetic-induced degradation of cIAP1 and the consequent activation of the NF-κB pathway and an autocrine/paracrine TNF signaling loop. Other possibilities exist, however, as work from Silke and Vaux suggests that triple knockdown of cIAP1, cIAP2, and XIAP results in a hyperactive inflammatory state through still-undefined mechanisms (reviewed in 259). While TNF release potentially enables the efficacy of Smac mimetics as single agents in cancer therapy, the possibility of inducing a "cytokine storm" may render this approach less desirable compared with a combination approach (TNFR agonists + Smac mimetics), especially for indications outside of cancer 260,261 . Indeed, Smac mimetics have demonstrated synergy with other modes of treatment, including cytotoxic agents (that is, carboplatin 262 and paclitaxel 263 ), radiation therapy 264 , and cell DR ligands (TRAIL analogues) 265 . These synergies are well defined in pre-clinical models, but, so far, they have been less successful in clinical settings (see below). In general, any treatment that stresses the cells, such as standard chemotherapy or radiation therapy, and induces either intrinsic or extrinsic apoptosis via upstream activation could be combined with the caspase-liberating effect of IAP inhibitors to kill cancer cells. Although a number of Smac mimetics have already entered clinical trials, we shall focus our discussion here on those for which trials are currently active (Table 1).

Birinapant
Birinapant is a bivalent Smac mimetic developed by Tetralogic Pharmaceuticals and currently owned by Medivir. Owing to the size of this molecule, only administration by intravenous line has been reported for birinapant, both as a single agent and in combination with several chemotherapeutics (azacitidine, gemcitabine, irinotecan, and conatumumab) 266 . The data released so far have been lackluster, and poor efficacy has been demonstrated in the completed studies. In NCT01681368, no complete nor partial response was observed in solid tumors of 11 patients, and accrual was terminated for lack of detected clinical benefit 267 . Good, though muted, news came in the phase I/II trial NCT01188499. In patients with metastatic colorectal cancer who previously failed irinotecan treatment, the combination of irinotecan with birinapant resulted in disease stabilization in 62% of patients, higher than the 41% rate shown by the recently approved kinase inhibitor regorafenib 268 . The other silver lining for these results is that the response rate as measured by tumor regression was higher, albeit small, at 8% compared with regorafenib at 1% 269 and that the enrolled patients had previously failed all available treatments. When comparing these two sets of trial data, one must bear in mind that the regorafenib data come from an earlier and much larger phase III trial. The failure of birinapant versus placebo in a study (NCT02147873) investigating its capacity to treat myelodysplastic syndrome resulted in the folding of Tetralogic and transfer of assets to Medivir, where trials are ongoing 270 . Given the previous failures, it will be critical to see favorable clinical outcomes for the conatumumab (DR5 agonist) combination therapy trial (NCT01940172) as well as the trial with pembrolizumab (PD-1 inhibitor) as co-treatment (NCT02587962). CRS has so far proven to be a relatively minor adverse event in patients who received birinapant, and only 10% of patients reported low-grade symptoms 266 .

LCL161
LCL161 is a monovalent IAP inhibitor developed by Novartis that is currently in active clinical trials and has also shown generally good safety up to a 1,800 mg dose; CRS was the major  Expanding on these findings is recent evidence that immune checkpoint blockade combined with Smac mimetics is efficacious in pre-clinical models of glioblastoma 276 . As such, the immune regulatory roles of IAPs may also be of much therapeutic relevance.

Debio 1143
Debio 1143 is another monovalent Smac mimetic in ongoing clinical trials for a number of different malignancies. It was developed in its early stages by the Wang group at the University of Michigan and later at Ascenta Therapeutics, ultimately being licensed to Debiopharm. Phase I safety studies were in line with the previously reported Smac mimetics reported above, when tested as a monotherapy: generally mild adverse events with a highest tested dose of 900 mg 277 . On-target pharmacodynamic modulation was achieved at doses above 80 mg, as measured by cIAP1 degradation. Preliminary efficacy data from the trial indicated that 20% of patients exhibited stable disease as the best response. With the encouraging safety data, a phase I/II trial (NCT02022098) with cisplatin and radiotherapy as co-treatment was undertaken for squamous cell carcinoma of the head and neck, and the expected completion date is 2019.

ASTX660
UK-based Astex Pharmaceuticals recently initiated their own phase I/II trial (NCT02503423) for the small-molecule ASTX660 for solid tumors and lymphomas. Envisioning ASTX660 as part of a two-pronged cell death approach, Astex screened a number of breast, colorectal, ovarian, leukemia, and melanoma cell lines for their response to monotherapy versus co-treatment with TNF-alpha 278 . It will be interesting to see how well the pre-clinical data correlate with the clinical data expected in 2018.
While the ability of the reported Smac mimetics to induce cancer cell death in pre-clinical models was exciting and held much promise, so far the first-in-human studies have presented lackluster results. Several compounds that were able to induce cancer cell death and thus partial or complete remission in tissue culture and animal studies have not had similar success in trials as a monotherapy. However, given the promising clinical pharmacodynamics and safety data, further research and development efforts are certainly warranted.

Conclusions and future work
As detailed above, the IAPs are at the nexus of cancer cell survival and, conversely, apoptosis. As such, the inhibition of pertinent family members would be expected to afford a valuable therapeutic intervention strategy for cancers, as these diseases are largely conditions of increased proliferation and impaired apoptosis. As often occurs, however, the reality has proven vastly more complicated than first envisioned. As detailed above, although Smac mimetics are safe and well tolerated, they have shown little single-agent activity in clinical trials. Intuitive, yet not extensively preclinically verified, combinations of IAP antagonists such as Smac mimetics with standard-of-care chemotherapeutics have likewise proven unfruitful to most degrees, although there have been some responses, as described above. Perhaps most encouraging have been pre-clinical studies showing that IAP antagonists are potent sensitizers to certain TNFR family agonists 64,199,208,[279][280][281][282][283] . Additionally, it has been shown that this can be effected not only by the natural ligands themselves but also by agonistic antibodies to TRAIL receptors developed by several pharmaceutical companies 284-288 . Targeting TRAIL receptors with simultaneous IAP inhibition not only is toxic to cancer cells but also leaves non-transformed cells untouched, a "holy grail" of anti-cancer therapy. Expanding on these observations are studies by Beug et al., who show that concomitant induction of an immune response when IAPs are inhibited can produce a profound tumor regression in animal models 289 . Indeed, the use of Smac mimetics and attenuated oncolytic viruses as an anti-cancer strategy has shown promising results in some models 290 . As such, the notion of targeted activation of certain TNFRs in combination with IAP inhibition is a potential potent intervention point in many cancers. Already, clinical trials of just such a combination are underway (discussed above), and the results should further assist in our progress toward more targeted therapies using these phenomena.
In sum, whilst the clinical application of IAP antagonists has to date not produced the panacea desired, the ongoing development of next-generation agents and pertinent combinations bodes well for the future. "Inhibiting the inhibitors" 291 may soon be a viable anti-cancer strategy.

Competing interests
The authors declare that they have no competing interests.

Grant information
This work was supported by National Institutes of Health grant R01CA195227 to DF and NDPC.
The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. PubMed Abstract | Publisher Full Text | F1000 Recommendation