Caspases are powerful destructive agents which, once activated, lead to certain death. Or do they? According to a report in Nature by Emad Alnemri and colleagues, a cell can be snatched from the jaws of death through the recruitment, by activated caspase-9, of its own inhibitor.

Death is, as Tennyson famously said, the end of life. It's no surprise, then, that caspase pathways must be tightly regulated. The so-called intrinsic death pathway, which responds to most death stimuli, is switched on through recruitment of the weakly active procaspase-9 by an adaptor molecule called Apaf-1. Procaspase-9 is then processed by autoproteolytic cleavage to yield active caspase-9.

Alnemri and colleagues knew that caspase-9 can be inhibited by the X-linked inhibitor of apoptosis protein ( XIAP). To find out how this inhibition works, they reconstituted, in vitro, caspase-9–Apaf-1 complexes containing either fully processed caspase-9 or the procaspase form, and then studied their interactions with XIAP. Although both complexes were catalytically active, only the fully processed caspase-9 could be inhibited by XIAP. This, it turns out, is because XIAP cannot associate with the unprocessed procaspase-9.

Does this mean that processing is required for inhibition by XIAP? To test this idea, the authors identified a conserved motif of four amino acids, which becomes exposed at the amino terminus of the caspase-9 small subunit after proteolytic processing. This motif also has considerable homology to a sequence that binds IAPs in the Drosophila death proteins Grim, Reaper and Hid. Systematic mutation of the residues in this motif confirmed that they are needed for binding of XIAP to caspase-9.

Inhibition of caspase-9 can be overcome by a protein called Smac/DIABLO, which also interacts with XIAP and also contains the conserved four residues at its amino terminus. Alnemri and colleagues wondered whether the binding of Smac/DIABLO and caspase-9 to XIAP is mutually exclusive, and found that, indeed, the caspase-9 IAP-binding motif abolishes the binding of Smac/DIABLO to XIAP (and vice versa).

The inference, say the authors, is that “Smac competes with caspase-9 for binding the same pocket on the surface of XIAP, which could explain the ability of Smac to promote the catalytic activity of caspase-9 in the presence of XIAP”. They also point out that, unlike for other caspases, proteolytic processing of caspase-9 is required for its inhibition rather than its activation. And, as Donald Nicholson discusses in the accompanying News and Views article, these results could have exciting implications for cancer therapy, as synthetic mimics of XIAP-binding peptides might sensitize cancer cells to apoptotic stimuli.