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

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^8–10. BIR domains are zinc finger domains and invariantly contain three cysteines and one histidine, which co-ordinate the zinc ion¹⁰, 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 homology-matching database searches (reviewed in 21). Notably, many of these proteins were further shown to be involved in the regulation of apoptosis^15,22–27. Rounding out the group of eight human BIR-containing proteins are XIAP (BIRC4), Survivin (BIRC5), Apollon (BIRC6), Melanoma IAP (ML-IAP or BIRC7), and IAP-like protein 2 (ILP-2 or BIRC8)^{15,22,25,28–37}. 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.

Figure 1. Domain structures of all known members of the human inhibitor of apoptosis protein (IAP) family, with a focus on the different baculovirus IAP repeat (BIR) domains.

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.

Figure 2. Schematic of pertinent inhibitor of apoptosis signaling pathways relevant to tumor cell survival and apoptosis.

Dashed lines indicate potential degradative events (blue = ubiquitin-mediated, black = caspase-mediated). cIAP1, cellular inhibitor of apoptosis protein 1; cIAP2, cellular inhibitor of apoptosis protein 2; ML-IAP, melanoma inhibitor of apoptosis protein; NAIP, neuronal apoptosis inhibitory protein; TNF, tumor necrosis factor.

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^38–41. ILP-2, likewise with only one BIR domain, also contains a RING domain and a ubiquitin-associated (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 testes³⁶. 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 apoptosis³⁶. 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 factors⁴².

Apollon is an extremely large protein (approximately 528 kDa) containing only one BIR domain that is thought to be membrane-associated as well as a C-terminal ubiquitin-conjugating (UBC) domain⁴³. It has been shown to attenuate apoptosis^44–46 and to directly engage and interfere with both the second mitochondria-derived 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 development⁴⁸.

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 leucine-rich 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) domain⁴⁹. 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 activity⁵⁵. 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 levels⁶⁴. Similarly, levels of nuclear factor-kappa 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 cIAP-mediated 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 RIPoptosome⁷² 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 caspases²⁴. 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 anti-cancer 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 cancers¹⁰⁴ 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.

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 caspase-suppressing 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 small-molecule inhibitors of such PPIs has been quite difficult. Most of these interactions are devoid of the classic druggable binding pockets (about 300–500 Ų) with which most drug discovery scientists are familiar²⁰². 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 Ų) 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 mimetics²⁰³. Mutation studies further confirmed the importance of the post-translationally 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 BIR3⁸⁹. 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^204–206. 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 2004²⁰⁷ 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 apoptosis²⁰⁸. A summary of the collective structure-activity-relationship (SAR) conclusions from Smac mimetic medicinal chemistry work is shown in Figure 4.

Figure 3. Crystal structure of Ala-Val-Pro-Ile (AVPI), a Smac core motif, bound to the BIR2 domain of XIAP (Protein Data Bank code = 4J46).

Binding is strongly driven by hydrogen-bond formation (dashed cyan lines) and non-polar interactions. Hydrophobic surface properties of the BIR2 domain are shown in yellow. Note that the color scheme of the tetrapeptide sequence is maintained for the subsequent figure.

Figure 4. Structure-activity relationship of Smac mimetics is largely based on the original amino acid positions from the Ala-Val-Pro-Ile (AVPI) peptide.

A number of research groups from both academia and industry have initiated programs in the space since these early discoveries, focusing on Smac mimetics^{66,199,209–256} (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 pre-clinical data in several cancer cell lines is shown in Figure 5¹⁹⁹. 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.

Figure 5. Structures of the Smac mimetic SBP-0636457 being developed by Sanford Burnham Prebys Medical Discovery Institute and the bivalent agent JP1201 from Joyant Pharmaceuticals.

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 construct) were observed for the first reported bivalent IAP inhibitor (JP1201) from the Wang and Harran labs (Figure 5)²⁰⁸.

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 events¹¹². 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, carboplatin²⁶² and paclitaxel²⁶³), radiation therapy²⁶⁴, 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).

Figure 6. Chemical structures of inhibitor of apoptosis inhibitor compounds in active clinical trials.

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 pre-clinically 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^{64,199,208,279–283}. 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 models²⁸⁹. Indeed, the use of Smac mimetics and attenuated oncolytic viruses as an anti-cancer strategy has shown promising results in some models²⁹⁰. 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.