CASPASES AND IAPS

SUMMARY

In this issue of Molecular Cell, Ditzel et al show that DIAP1 polyubiquitinates DRONC, DCP-1 and drICE, leading to their non-degradative inactivation. Surprisingly, activation of DIAP1 requires caspase-mediated cleavage, revealing an elegant feedback mechanism by which caspases regulate their own fate.

Multicellular organisms maintain homeostasis by eliminating unnecessary cells using an apoptotic cell death program. A critical step in this program is the activation of caspases, which carry out apoptosis by cleaving specific cellular proteins, ultimately resulting in cell death. Caspases are synthesized as inactive zymogens and are activated by proteolytic removal of their small pro-domains (Riedl and Shi, 2004). The apoptotic program, and specifically the proteolytic cascade of caspase activation, is regulated at multiple levels by a variety of protein families, including the inhibitor of apoptosis (IAP) protein family.

IAPs can act both upstream and downstream of caspase activation to promote cell survival. Additionally, some studies indicate that IAPs may also act directly on caspases. The mammalian family member XIAP can directly inhibit caspases in vivo (Vaux and Silke, 2005). In this case, the physical interaction between caspases and XIAP seems to be sufficient to block apoptosis. In other cases, regulation of caspases by IAPs is more complex. For example, DIAP1, like many other IAPs, has a carboxy-terminal RING domain and DIAP1 requires the E3 activity of its RING domain to antagonize the initiator caspase DRONC and the effector caspase drICE (Zachariou et al., 2003). RING-mediated mono- and poly- ubiquitination of other caspases has also been described in Drosophila and mammals, although the consequences of caspase ubiquitination are unclear (Huang et al., 2000; Suzuki et al., 2001; Vaux and Silke, 2005). Indeed, although the RING domain of DIAP1 is absolutely required for its anti-apoptotic activity, the mechanistic details of how caspase ubiquitination blocks apoptosis have remained unclear.

In this issue of Molecular Cell, Ditzel et al show that DIAP1-mediated polyubiquitination of the caspases DRONC, drICE and DCP-1 is both necessary and sufficient for suppression of apoptosis. In vitro ubiquitination assays revealed that DIAP1 generates both Lys48- and Lys63-linked ubiquitin chains on drICE, but the Lys63-linked chains are significantly longer. As Lys48-linked chains are known to be degradation signals for the proteasome, DIAP1 activity might trigger degradation of drICE. However, in experiments with proteasome inhibitors, the authors show that polyubiquitination of drICE does not lead to proteasome-mediated degradation of the caspase as would be expected. Rather, the authors elegantly demonstrate that IAP-mediated ubiquitin conjugation alone is sufficient to suppress caspase activity. By combining an in vitro ubiquitination assay with a PARP cleavage reaction, they find that non-ubiquitinated drICE is much more efficient in cleaving its substrate PARP than drICE that has been ubiquitinated by DIAP1. Remarkably, DIAP1 seems to be promiscuous in its usage of lysine residues on drICE; substitution of up to eight of nine drICE lysines with arginine residues leaves DIAP1-mediated ubiquitination and suppression of drICE activity intact. Substitution of all nine lysine residues, however, results in a variant of drICE that, though it can still bind to DIAP1, cannot be ubiquitinated and is therefore resistant to DIAP1-mediated suppression.

Ditzel et al made another surprising observation: DIAP1 must be cleaved by caspases to generate fully competent DIAP1 (residues 21-438). It is this cleaved form of DIAP1 that can bind and ubiquitinate drICE. Caspase-mediated cleavage and N-end rule processing of DIAP1 have been described previously, but the functional significance of this processing is only now becoming clear. DIAP1 cleavage exposes a previously described binding site for UBR-domain-containing E3 ligases (Ditzel et al., 2003). The current study demonstrates that UBR binding is absolutely required for subsequent caspase inactivation, thereby providing a novel model of a negative feedback mechanism by which caspases regulate their own inhibition. Ditzel et al explored the importance of DIAP1 cleavage in vivo by generating a set of transgenic lines expressing variants of DIAP1 lacking a caspase cleavage site, UBR docking domain, or functional RING. Remarkably, only wild-type DIAP1 and N-terminally truncated DIAP1 (residues 21-438) were capable of suppressing apoptosis in the developing Drosophila retina. Therefore, both DIAP1’s intrinsic E3 ubiquitin ligase activity and its UBR docking domain are required for its anti-apoptotic function in vivo. Intriguingly, the authors identified five UBR-containing E3s that associate with DIAP1, namely UBR 1, 3 and 5-7. In addition to being required for its suppression of caspase activity in vivo, heterodimerization with other E3s could increase DIAP1’s potential pool of E3 ubiquitin ligase targets and therefore extend its function well beyond its known anti-apoptotic activity. The identification of ubiquitination targets of all the IAP E3 ligases is still ongoing, and there is much to learn about the diverse ubiquitin chain formations and their consequences.

Cleavage of XIAP and the cellular IAPs (c-IAP1, c-IAP2) by caspases after an apoptotic stimulus has been described (Clem et al., 2001; Deveraux et al., 1999). Based on the results of Ditzel et al, it is tempting to speculate that mammalian IAP function may be regulated by caspase-induced cleavage. This is an important question as IAPs have been the aim of intense investigation for the development of cancer therapeutics. Small molecule antagonists, mimicking the IAP:SMAC (an IAP antagonist) interaction, have been developed and been proven quite effective in inhibiting XIAP and the cellular IAPs, c-IAP1 and c-IAP2 (Vince et al., 2007; Varfolomeev et al., 2007). Given the results of Ditzel et al, it will be interesting to investigate if inhibition of caspase-mediated IAP cleavage affects mammalian IAP function in a similar way. Finally, although caspase ubiquitination by IAPs has been described previously, the finding that conjugation of polyubiquitin chains leads to the inhibition of substrate cleavage rather than proteasomal degradation of caspases is novel, and it will be interesting to explore polyubiquitination of caspases by IAPs in mammals.

Figure 1. The caspase-DIAP1 autoregulatory loop.

Basal proteolytic activity of drICE cleaves DIAP1 to generate processed, fully competent DIAP1, which is then capable of binding to and ubiquitinating both drICE and DRONC. Inactivation of drICE by DIAP1 requires the exposed UBR docking domain at the new amino-terminus but is independent of proteasomal degradation of drICE.