The apoptosome molecular timer synergises with XIAP to suppress apoptosis execution and contributes to prognosticating survival in colorectal cancer

The execution phase of apoptosis is a critical process in programmed cell death in response to a multitude of cellular stresses. A crucial component of this pathway is the apoptosome, a platform for the activation of pro-caspase 9 (PC9). Recent findings have shown that autocleavage of PC9 to Caspase 9 (C9) p35/p12 not only permits XIAP-mediated C9 inhibition but also temporally shuts down apoptosome activity, forming a molecular timer. In order to delineate the combined contributions of XIAP and the apoptosome molecular timer to apoptosis execution we utilised a systems modelling approach. We demonstrate that cooperative recruitment of PC9 to the apoptosome, based on existing PC9-apoptosome interaction data, is important for efficient formation of PC9 homodimers, autocatalytic cleavage and dual regulation by XIAP and the molecular timer across biologically relevant PC9 and APAF1 concentrations. Screening physiologically relevant concentration ranges of apoptotic proteins, we discovered that the molecular timer can prevent apoptosis execution in specific scenarios after complete or partial mitochondrial outer membrane permeabilisation (MOMP). Furthermore, its ability to prevent apoptosis is intricately tied to a synergistic combination with XIAP. Finally, we demonstrate that simulations of these processes are prognostic of survival in stage III colorectal cancer and that the molecular timer may promote apoptosis resistance in a subset of patients. Based on our findings, we postulate that the physiological function of the molecular timer is to aid XIAP in the shutdown of caspase-mediated apoptosis execution. This shutdown potentially facilitates switching to pro-inflammatory caspase-independent responses subsequent to Bax/Bak pore formation.


Supplementary Text 1: Parameterisation of models of the apoptosis execution phase
Kinetic values are based on the previously published model of allosteric and homodimerisation mediated cleavage at the apoptosome (ApoptoAll). All kinetic values in the ApoptoAll model are based on experimental data. Kinetic values for apoptosome formation were determined by a kinetic screening for KD, kon and koff values that allowed for 65% apoptosome formation within 5 min after CytC release in line with experimental data (1). Kinetic rates for the binding of PC9 and C9 to the apoptosome were determined by a kinetic screening of kon and koff value pairs to a experimentally determined KD value (2,3). All models require initial concentrations of PC3, PC9, Apaf-1, XIAP, mitochondrial Smac, mitochondrial cyt c and ATP (Table 1). Figure 1A, C & D and Figure 2D, E & H were based on HeLa cell quantification of these proteins, as reported previously (3,4). In parameter screens, cytochrome C (5) and ATP (6) levels were taken from HeLa cells as generally they are not thought to be rate limiting.

Table 1: List of starting concentrations in HeLa cells
Species that are not present at the start of the simulation, but are formed during apoptosis execution, are displayed in Table 2. The key reactions and kinetic parameters of the models are further summarised in Table 3 and Table 4. Most species are additionally modelled to undergo degradation with a specific degradation rate (Table 5). Additionally, Apaf-1, PC9, PC3 and XIAP are also formed de novo in the models. The formation rate is calculated by multiplying the kinetic rate of the degradation reaction of each protein with its initial concentration.  Fully assembled heptameric Apaf-1 complex with bound PC9 and/or C9. Both n and m can be any number between 0 and 7, however n + m ≤ 7.
[Apoptosome_P9_n_C9_m~XIAP] Apoptosome complex with bound XIAP. n can be any number between 0 and 7, m any number between 1 and 7, however n + m ≤ 7.
Formation of the apoptosome.
Binding of PC9 to the Apoptosome.
Binding of C9 to the Apoptosome.
Autocatalytic cleavage of PC9 monomers.   The models with only homodimerisation-mediated cleavage (ApoptoDimC) and cooperative recruitment (ApoptoCoop) models are based on the original ApoptoAll model and the differences between the models are highlighted in Table 6.

Table 6: Differences in reactions and reaction rates between Apopto-All, -DimC and -Coop
In all the models used, PC9 and C9 were modelled to only bind to the fully assembled apoptosome, not to any precursor molecules. In the originally published ApoptoAll model (3) was implemented to cleave PC9 to C9. This reaction was turned off in all simulations, since C3 was shown to cleave both PC9 and C9, leading to the formation of new C9 species with different binding behaviour and activity that is not included in the models in this paper (7). Moreover, in the ApoptoAll model, C3-cleavage of PC9 to C9 had no notable effect due to the more rapid monomeric autocleavage of PC9 to C9 (data not shown). Additionally, the activity of PC9 was lowered in ApoptoCoop compared to the allosteric model. This was to reflect the fact that PC9 is not constitutively activated on the apoptosome but likely alternates between active heterodimer and inactive states. Although C9 also is activated as a heterodimer, C9's ability to dimerise is impeded after autocatalytic cleavage, moreover C9-35/12 has poor affinity for the apoptosome and is prone to falling off. Therefore, the majority of C9 activity likely comes from the original PC9 homodimer formed prior to cleavage (7,8). As a further validation of this effect, simulations matched experimental findings that C9-35/12 contributed approximately 85% of PC3 cleavage, as determined using the C9-35/12 specific inhibitor Bir3-RING derived from truncation of XIAP (data not shown) (8).
The model without the molecular timer (ApoptoCoop -Timer ) was implemented as ApoptoCoop with the following modification. The kon and koff of C9 binding to the apoptosome were set to the same as secondary binding of PC9, effectively preventing the molecular timer. This was validated by monitoring the IETDase activity with this alteration (ApoptoCoop -Timer ), where the activity is both increased and sustained compared to ApoptoCoop, demonstrating the lack of molecular timer activity.

Supplementary Text 2: Parameter Estimation
In order to implement cooperative recruitment, a parameter estimation was performed to determine kinetic values for primary binding of PC9, cooperative binding of PC9 and binding of C9 to the apoptosome. The upper and lower bounds used for the parameter estimation were chosen with regard to the SPR data (Fig 2) (Table ). Since the SPR data was based on binding to an individual CARD domain and the ApoptoAll model considers the heptameric apoptosome as a singular reactant, the SPR kinetic values were multiplied by seven to reflect the maximal availability of seven CARDs per apoptosome. In order to achieve a global parameter estimation approach, 100 different iterations of local parameter estimation were performed with local boundaries created using the lhsdesign function in Matlab for Latin Hypercube sampling. For the parameter estimation, simulated data was compared to two sets of experimental data with equal weighting using a least squares regression. Experimental data used was C3 substrate cleavage in HeLa cells, obtained previously from fitting a Boltzmann sigmoidal to FRET substrate data and the molecular timer data, extracted from Malladi and colleagues using ImageJ and normalised to the 5 minute time point (4,8). The cost function for each individual experiment was calculated via the least squares method with the following formula, where t is each time point with available experimental data: The cost functions for both fits were then combined to a total cost: The matlab function fmincon was used to minimise the cost function in each local estimation. The obtained optimal k-values were post-processed and the k-value combination chosen that fulfilled all of the following criteria: kon initial PC9 binding < kon cooperative PC9 binding koff monomer PC9 unbinding > koff dimer PC9 unbinding koff C9 unbinding > koff dimer PC9 unbinding The justification for this rule set is, in the case of cooperative recruitment, the kon value of the initial PC9 molecule binding to the apoptosome should be considerably lower than the kon value of the cooperative binding event. Furthermore, once PC9 is bound as a homodimer it is expected to be more stable on the apoptosome and therefore the koff value of cooperative binding should be lower than that of initial binding events. For the molecular timer to work it additionally is important that the koff value of C9 is higher than the koff value of PC9 and vice versa that the kon value of C9 is lower than the kon value of PC9. These assumptions are further supported by the SPR data extracted from Wu and colleagues, reported in Fig 2b. The resulting value sets were then chosen according to optimal cost function and implemented as ApoptoCoop.  Supplementary Figure 5. Non-cleavable PC9 sensitises cells to apoptosis more than the molecular timer alone. Survival curves with and without the molecular timer, without XIAP and with non-cleavable PC9 for complete MOMP (A) and minority MOMP (B). P-values from logrank test, n=1000 for all groups.