In humans, two apical caspases are implicated in triggering the extrinsic death pathway: Caspase-8 and caspase-10 (27
). Despite reports that caspase-10 is recruited to the DISC and able to initiate apoptosis (30
), most previous research on the role of caspases in the extrinsic pathway has focused on caspase-8 – indeed there are over 20 times more citations in PubMed on caspase-8 than on caspase-10. One reason for this could be that caspase-10 is not expressed in the rodent lineage and therefore less amenable to genetic studies. However, protein sequence alignment of the two caspases shows high similarity, and even though there is not yet a crystal structure of caspase-10, one would expect a similar folding and function of the protein.
Surprisingly, the recombinant cleavage site mutant was not processed although there is a second conserved cleavage site in the linker region. This site is apparently not processed by autoproteolysis, which distinguishes caspase-10 from human caspase-8, where two sites undergo autoproteolysis (17
) (). Indeed, even after full activation of the mutant in optimal kosmotrope we failed to see any cleavage, and material remained as a single chain. It is possible that the second potential cleavage site in caspase-10 is a remnant of the twin caspase-8 cleavage sites, but since the role of the twin cleavage sites in human caspase-8 is unknown, further speculation is fruitless at this point. Enzymatic activity of caspase-10 was greatly enhanced by kosmotrope-induced dimerization, although the overall activity of the mutant was significantly lower as compared to the wildtype. These results agree with the hypothesis that dimerization is the activating event, whereas cleavage of the linker stabilizes the dimer - shifting the equilibrium towards a fully active enzyme - as demonstrated for caspase-8 (16
To address the important caveat of using kosmotropic salts, which do not represent physiological conditions, we employed an inducible dimerization system. We show that the wildtype caspase-10 is fully activated with stoichiometric amounts of the dimerizer AP20187, whereas the cleavage site mutant had no measurable activity with AP20187 alone, even though size-exclusion chromatography confirmed that the mutant becomes dimeric in the presence of the compound. Interestingly, the positive control in sodium citrate displayed activity of the fusion proteins of both wildtype and cleavage site mutant. Because simple dimerization of the cleavage site mutant failed to activate caspase-10 we considered the possibility that citrate activates the zymogen by somehow generating an active site in the monomer. However, this seems unlikely given that Fv-mediated dimerization substantially lowers the concentration of citrate required to activate caspase-10. This implies that dimerization of the Fv-domain is necessary but not sufficient to fully activate the cleavage site mutant, which serves as a model of the initial form of the caspase-10 zymogen recruited to the DISC.
The simplest explanation is that the dimerized Fv-domain is not in an optimal conformation to allow the same dimer conformation of the catalytic domain that would be driven by the native DEDs in the DISC, which the Fv-domain simulates. The Fv-mediated dimer is simply not organized precisely enough, and the Fv-mediated wildtype overcomes this deficit because of additional interactions between the cleaved portions of the structure (). Yet, as outlined above, the consequence of this line of reasoning is that kosmotropes such as citrate promote dimerization of the monomer zymogen of caspase-10 with an additional role in stabilizing active site loops. Taken together, our data strongly suggest that caspase-10 follows the proximity-induced dimerization model for apical caspases.
It is possible that the decrease in substrate hydrolysis of the non-cleavable protease is utilized in vivo
to convey substrate preferences. To investigate this possibility we tested the cleaved and non-cleavable caspase-10 on a P1 Asp-fixed positional scanning substrate library in the presence of 1.0 M sodium citrate to provide robust activity of both proteins. The library revealed very similar subsite preferences (P4-P3-P2) for both the wildtype and mutant caspase-10. However, when we compared the catalytic parameters of the cleaved versus the noncleavable enzyme on small peptide substrates and protein substrates there seemed to be a big difference in the efficiency of protein substrate processing. While the efficiency reflected by kcat
on three different peptide substrates, Ac-LEHD-AFC, Ac-IETD-AFC and Ac-DEVD-AFC, was quite comparable, the difference on protein substrates was striking. These findings are summarized in . The wildtype caspase-10 was able to process each protein with relatively good kcat
in a range from 1.3 × 103
to 2.5 × 104
, with Bid and RIPK1 as the best substrates. Cleavage of the four proteins tested supports the pro-apoptotic function of caspase-10. Intriguingly, although the cleavage site mutant processed pro-caspase-3, pro-caspase-7, and RIPK1 very poorly, in comparison, Bid appeared to be a bona fide substrate with a kcat
of 4.6 × 103
. The substantial difference in the activity of the cleavage site mutant on Bid vs. RIPK1 despite their very similar cleavage site motifs (LQTD/G vs. LQLD/G) is plausible through a different mode of presentation of these sequences. Previous work (62
) has shown that linkers or loops and their lengths significantly influence the presentation of the cleavage motifs and thus play an important role for the efficiency by which protease substrates get processed. In Bid we find an extended long protruding loop containing the cleavage site LQTD/G, presented by a Bcl-2 fold scaffold which can serve as an appropriate basis for the near optimal behavior as a caspase substrate. If we consider that cleavage of Bid has been reported to be an exclusively pro-death signal (53
), we come to the hypothesis that the uncleaved form of caspase-10 has a pro-death role, which is in contrast to data on the role of cleavage in caspase-8.
Non-cleavable caspase-8, expressed in mouse cells, was not able to trigger apoptosis, but could still perform non-apoptotic functions, such as proliferation of T-lymphocytes and B-lymphocytes or macrophage differentiation (16
). Another approach to study the role of cleavage in caspase-8 was taken by reconstituting the DISC in vitro
and this has lead to the conclusion that caspase-8 cleavage is necessary to induce DISC-mediated apoptosis (18
) and that uncleaved caspase-8 has a restricted substrate repertoire that supports cell survival. Our data argue that caspase-10 does not demonstrate a specific alteration in specificity upon cleavage that would adjust its specificity towards a non-death role, distinguishing it from caspase-8.
In conclusion we established the activation mechanism and primary specificity of caspase-10. Our results reveal a similar activation mechanism and intrinsic substrate preference to caspase-8, though with restricted tolerance in the P4 pocket. Importantly, we observe that cleavage of the caspase allows for substantially accelerated cleavage of protein substrates, with the exception of Bid, for which caspase cleavage was not required, and which is an excellent substrate for both cleaved and uncleaved caspase-10. Though it is yet too early to support in detail, this finding tends to disagree with a potential non-death role for uncleaved caspase-10.