Mutant forms of XIAP

To discriminate between the different functions of XIAP, point mutations were inserted into the XIAP sequence to abolish particular function. Since transfection of truncated forms like XIAP∆(BIR1/2) and XIAP∆(BIR3/RING) resulted in higher expression levels compared to XIAP, only one mutant was used lacking an entire domain. All mutations used in this study had been characterized ^155,156.

The employed mutants could be arranged according to their characteristics into three classes: mutants partly or entirely abolished in caspase inhibiting activity, a mutant

5 DISCUSSION

deficient in dispersed functions like E3 ubiquitin ligase, XIAP oligomerization ¹⁵⁶, and signalling properties and last a mutant deficient in E3 ubiquitin ligase activity.

To investigate the contribution of caspase inhibition to the protective function of XIAP in the HeLa cell model, mutants deficient in inhibition of caspase-3, / 7 or caspase-9 as well as the double deficient mutant were tested. Former unpublished experiments had revealed that XIAP∆(BIR1/2) and XIAP∆(BIR3/RING) fragments could only partially inhibit death receptor-mediated apoptosis. However, full-length mutants with the same caspase inhibiting abilities could inhibit cell death efficiently. These mutants showed only a weaker protection if low DNA amounts were transfected.

Consequently, the EC50 values were higher than that of the wild type.

In theory, after inhibition of caspase-9, caspase-3 should be still activated and vice versa (see Figure 1.1). Therefore, the protection should have been disturbed.

However, Silke et al. reported that protection conferred by these two mutants (deficient in caspase-3 ¹⁵⁵ and 9 ¹⁵⁶ inhibition) was comparable to that of the wild type.

In these studies, cell death has been induced by UV irradiation. Therefore, the explanation for the protection seemed to be quite plain since the stimulation was confined to the intrinsic pathway. However, in the results obtained in this study the extrinsic as well as the intrinsic pathway was activated. Therefore, an explanation is difficult. In literature, some hints pointed to possibly disturbed amplification loops ¹⁵⁶. For another explanation, the endogenous levels of XIAP had to be taken into consideration. HeLa cells have a high endogenous XIAP level ¹⁴⁶, which might be sufficient to inhibit caspases that were not inhibited by over-expressed mutants.

However, the reasons are far from clear indicating that the protection by XIAP is more complex.

In stark contrast, the mutant with abolished inhibition of all three caspases was no longer able to protect cells. These results clarify that the caspase binding sites of XIAP are essential for the protection conferred by XIAP. To exclude that the RING domain contributes to cell death ¹⁵⁷, a mutant unable to inhibit caspases and lacking the RING domain (XIAP DWE ∆RING) was tested, too. Cells transfected with this mutant were also not protected, indicating that only the mutated caspase binding sites contribute to the failure of protection.

These findings demonstrate that for protection by XIAP the caspase binding sites are essential. Since the mutations were localized in highly conserved regions and e.g.

5 DISCUSSION

both Smac/DIABLO and Omi/HtrA2 shared the same binding site ¹, it cannot be excluded that other proteins essential for protection were prevented from binding, too.

The contribution to protection by the RING domain was examined, too. Since the RING domain is involved in multiple functions (see 1.2.1.1.1), its deletion could alter the protection capability dependent on caspases (ubiquitination ^97,100) as well as independent of caspases (signal transduction ⁸³). To facilitate apoptosis induction by death receptors, HeLa cells were sensitized with cycloheximide. Since, cycloheximide is an inhibitor of translation, all protective mechanisms based on signal transduction followed by protein biosynthesis could be ruled out. For this reason, the very complex signal transduction properties of XIAP are excluded from the following considerations.

The experiments revealed a significant protection by the mutant lacking RING that was accompanied by a dramatic decrease in potency. In comparison to XIAP, the efficacy was fifty percent lower and the EC50 value was higher.

Creagh et al. published analyses employing the ∆RING mutant in a cell model of intrinsic apoptosis induction ⁹⁹. In this report, the same observations were made. Only the ineffectiveness appeared not to be as pronounced as in the presented study. This was the case although the auto-ubiquitination was markedly reduced. However, it cannot be excluded that differences in the experimental systems were responsible.

The results of the RING-deleted XIAP together with the results obtained with the caspase non-binding mutant, this ineffectiveness suggested an involvement of caspases. Therefore, experiments were focused on the E3 ubiquitin ligase.

For this purpose, the possibility had to be excluded that ubiquitin ligases other than that of over-expressed XIAP were capable to ubiquitinate proteins involved in cell death. To examine this, the auto-ubiquitination of XIAP was employed because wild type XIAP undergoes ubiquitination during apoptosis heavily. Therefore, auto-ubiquitination is a surrogate for auto-ubiquitination of target proteins. As illustrated by ubiquitination analysis, the absence of the RING domain prevented auto-ubiquitination of XIAP. It is important to emphasize that the lysine residues modified by ubiquitination are located within the BIR3 domain and therefore were still accessible. The caspase binding abilities were not affected by deletion of RING and consequently it was unlikely that other endogenous E3 ligases could bind and ubiquitinate caspases alternatively. Therefore, it is likely that the ineffectiveness of the mutant is due to its inability to ubiquitinate formed XIAP-caspase complexes. This

5 DISCUSSION

shifts the equilibrium towards free caspases. A mathematical bistability analysis of apoptosis supports this hypothesis ¹⁶⁹. Calculations of an apoptosis model showed that slower degradation of capase-3-IAP complexes would result in elevated levels of free active caspase-3. The observed effect was comparable to an increase of initial caspase-3 concentration and therefore enhanced the apoptosis rate.

To separate the ubiquitin ligase function from others of the RING domain, only one essential residue was exchanged in the active site of the E3 ubiquitin protein ligase.

Seeing that this mutant is protecting like the ∆RING mutant, the effects of the RING domain can be reduced to its ubiquitin ligase activity in this model.

Intriguingly, the mutant deficient in caspase inhibition was ubiquitinated to the same extent as the wild type. This finding indicates that the ability of XIAP for oligomerization was extant, which ought to be taken for granted for ubiquitin transfer.

For this caspase non-binding mutant, the fact that XIAP has to bind to caspases for ubiquitination leads to the assumption that caspases are not ubiquitinated by this mutant. Yet, the possibility cannot be excluded that others than caspase-dependent pathways are affected since e.g. NFκB activation is directly dependent on E3 ubiquitin ligase activity ⁸³. The exact mechanisms remain to be determined.

Considering the mentioned points, these experiments show the important contribution of the RING domain and especially the ubiquitination ability to the protecting capacity of XIAP after induction of apoptosis with TNFα. Moreover, a mechanistically proposal is made inasmuch as the degradation of the XIAP-caspase complexes are hold responsible for the effectiveness of protection.