A mitosis-sensing caspase activation platform? New insights into the PIDDosome

ABSTRACT

In contrast to the apoptosome and death-inducing signaling complex, the PIDDosome remains an orphan caspase activation platform unassigned to a specific apoptotic pathway. We found that DNA damage-induced PIDDosome formation is blocked by the mitotic checkpoint factor BUBR1 (budding uninhibited by benzimidazole-related 1), via a direct interaction that disrupts the PIDDosome core scaffold. This inhibition occurs at the kinetochore, thus physically connecting the mitotic and apoptotic machineries.

Abbreviations

APC/C

anaphase promoting complex/cyclosome

ATM

ataxia telangiectasia mutated

BIM

bcl-2 interacting mediator of cell death

BUB

budding uninhibited by benzimidazole

BUBR1

budding uninhibited by benzimidazole-related 1

CDC20

cell division cycle 20

CDK

cyclin-dependent kinase

DD

death domain

IF

immunofluorescence

IR

ionizing radiation

KT

kinetochore

MAD

mitotic arrest deficient

MCC

mitotic checkpoint complex

MCL-1

myeloid cell leukemia sequence

PIDD

p53-inducible protein with DD

RAIDD

RIP-associated Ich-1/CED homologous protein with DD

RNAi

RNA interference

SAC

spindle assembly checkpoint.

The PIDDosome¹ is composed of the death domain (DD) proteins p53-inducible protein with a DD (PIDD, a scaffold protein) and RIP-associated Ich-1/CED homologous protein with DD (RAIDD, a caspase adaptor molecule), and their client caspase, caspase-2. Similar to classical caspase-activation platforms, assembly of the PIDD–RAIDD–caspase-2 complex promotes a local increase in the concentration of the inactive caspase precursor, enabling its activation via homodimerization and autocatalytic cleavage.¹

To identify negative regulators of PIDDosome formation, we initiated screens for genes whose RNA interference (RNAi): (1) enhances caspase-2 cleavage in response to ionizing radiation (IR), as assessed in an immunoblot-based primary screen; (2) does so in a PIDD- and RAIDD-dependent manner, as verified in PIDD- and RAIDD-deficient cells; and (3) effectively triggers formation of the complex, as assessed by the inclusion of PIDDosome components into high molecular weight fractions of approximately 660 kDa. These efforts² identified Budding Uninhibited by Benzimidazole 1B (BUB1B), which encodes the mitotic checkpoint factor BUBR1 (mitotic arrest deficient 3 [Mad3] in yeast, C. elegans, and Arabidopsis). BUBR1 is a core component of the mitotic checkpoint complex (MCC), a tetramer of BUBR1–budding uninhibited by benzimidazole 3 (BUB3)–mitotic arrest deficient 2 (MAD2)–cell division cycle 20 (CDC20) that prevents anaphase-promoting complex/cyclosome (APC/C) activation until the spindle microtubules are properly attached to kinetochores (KTs)—a process termed mitotic checkpoint or spindle assembly checkpoint (SAC).^3,4

Suppression of PIDDosome activity by BUBR1 is biologically significant: knockdown or mutational removal of the gene is sufficient to trigger PIDDosome-mediated apoptosis in irradiated human, mouse, and zebrafish cells.² In further support of a genuine role for BUBR1, a mitotic protein, in PIDDosome control, we found that (IR)-induced PIDDosome assembly is tightly linked to mitotic progression. In these latter experiments using synchronized HeLa cells, we asked whether inhibitors of mitotic entry or exit (the CDK inhibitor RO-3306 and microtubule poison nocodazole, respectively) would affect RAIDD recruitment to PIDD as assessed by co-immunoprecipitation. We envisioned 3 scenarios: (1) RAIDD recruitment would be unaffected by either drug, indicating that the complex forms before mitosis, during the interphase; (2) RAIDD recruitment would be affected by RO-3306 but not nocodazole, indicating that that PIDDosome formation occurs during mitosis; and (3) RAIDD recruitment would be affected by both RO-3306 and nocodazole, indicating that the complex assembles after mitosis. RAIDD recruitment to PIDD clearly matched scenario 3—i.e., occurs upon mitotic exit—an observation consistent with the timing of APC/C-mediated physiologic degradation of BUBR1.^2-4

Although the data supported a direct role for BUBR1 in PIDDosome control, it remained possible that the observed effects were an indirect consequence of BUBR1′s canonical role in MCC-mediated APC/C inhibition. Indeed, Kornbluth and colleagues had shown that CDK1–CYCLIN B1, a key target for destruction by APC/C, phosphorylates caspase-2 on S340 during mitosis to prevent its activation until satisfaction of the SAC (caspases 9 and 8 are also CDK1 targets, see Fig. 1).⁵ One would therefore predict that loss of BUBR1, which leads to deregulated APC/C activity and thus excess destruction of CDK1-CYCLIN B1,^3,4 would act to disable mitotic suppression of caspase-2. This would in turn account for the increase in caspase-2 cleavage observed in BUBR1-ablated cells. However, 3 observations ruled out such an indirect effect. First, this model predicts that depletion of the other essential MCC component, MAD2, would also lead to enhanced caspase-2 activation, which was not the case.² Second, while the BUBR1–APC/C–CDK1–caspase-2 connection might, in principle, affect PIDDosome formation at the level of the RAIDD-caspase-2 interaction, it is unlikely to explain the loss of the PIDD–RAIDD interaction observed in response to nocodazole² (see above). Third, and most importantly, we found that BUBR1 in fact acts upstream of RAIDD–caspase-2, via a direct interaction with the PIDD DD that interferes with RAIDD recruitment to the scaffold.² The data therefore identified BUBR1 as a direct, competitive inhibitor of PIDDosome formation, acting in a non-canonical role independent of the MCC (Fig. 1A).

Figure 1.

Connecting the mitotic checkpoint machinery to caspase-activating platforms. (A) In the presence of intact levels of BUBR1, BUBR1 is mobilized to both the MCC (canonical function) and the PIDD scaffold (novel, non-canonical function independent of the MCC). These functions serve, in part, to generally prevent initiator–caspase activation via (i) CDK1-mediated inhibitory phosphorylation of caspase-2, −8, and −9, and (ii) blockade of RAIDD recruitment to PIDD, which further immunizes the cell against caspase-2. Note that the CDK1 route is constitutive whereas the PIDD pathway is only relevant in the context of DNA damage (symbolized by lightning rod), which is a prerequisite for PIDD^T788 localization at KTs and its ability to interact with either RAIDD or BUBR1 (ref. ²). (B) In the absence of BUBR1—through physiological degradation upon mitotic exit, genetic manipulation, or genetic alteration in cancer—both caspase-inhibitory axes are relieved. Caspase monomers are no longer phosphorylated, enabling their activation by cognate caspase activation platforms, as indicated. In addition, RAIDD is now free to associate with PIDD, triggering PIDDosome formation and further stimulating caspase-2–mediated apoptosis. It remains to be demonstrated whether full PIDDosome assembly occurs directly at the KT or in other cellular compartments (nucleoplasm, cytoplasm) during late mitosis or the next G1. BUBR1, budding uninhibited by benzimidazole-related 1; CDK, cyclin-dependent kinase; KT, kinetochore; MCC, mitotic checkpoint complex; PIDD, p53-inducible protein with a death domain; RAIDD, RIP-associated Ich-1/CED homologous protein with DD.

Perhaps the most striking observation from our study² was that PIDDosome inhibition by BUBR1 after IR occurs at the cellular site of mitotic checkpoint signaling, that is, the kinetochore (KT).^3,4 The spatial detection of endogenous PIDD, RAIDD or caspase-2 in cells was hampered by the lack of immunofluorescence (IF)-compatible antibodies. We therefore tested the antibody that we developed against PIDD phosphorylated on threonine 788, which detects the PIDD protein pool that has been primed for RAIDD recruitment via phosphorylation of the PIDD DD by ataxia telangiectasia mutated (ATM) in response to IR.⁶ PIDD^pT788 localized at double-strand DNA breaks during interphase, as expected for an ATM substrate. However, this was no longer the case in mitotic cells; instead, PIDD^pT788 was overwhelmingly detected at KTs where it colocalized with BUBR1 itself, consistent with their physical interaction. Notably, BUBR1 was required for the relocation of PIDD^pT788 from DNA breaks to KTs, and add-back experiments with WT versus KT-deficient BUBR1 constructs demonstrated that KT localization was essential for PIDDosome inhibition by BUBR1.²

Why would PIDD travel to KTs after having been primed for PIDDosome assembly by ATM at DNA breaks? In other words, what is the significance of PIDDosome control by BUBR1 after DNA damage? There is emerging evidence that the MCC can be co-opted by the DNA damage response as a backup to arrest compromised cells prior to anaphase onset, especially in situations of violated or weakened DNA damage checkpoints.⁷ Primed PIDD may thus relocate from DNA breaks to KTs in order to sense the functionality of the backup by surveying BUBR1 levels. At high levels of KT-recruited BUBR1, the backup is sensed as operational: BUBR1 both prevents APC/C-mediated anaphase entry and outcompetes RAIDD at the PIDD DD, thus arresting the cell and allowing it to survive (Fig. 1A). However, at low levels of BUBR1—as obtained physiologically, experimentally, or via mutation or downregulation as seen in some cancers³–the backup is assessed as compromised. Here, inhibitions of the APC/C and PIDDosome are simultaneously relieved and the cell evades arrest, but nonetheless the PIDDosome assembles and kills it, either shortly before or after it gives rise to daughter cells (Fig. 1B). Consistent with this view, caspase-2 has been implicated in mitotic catastrophe, a form of cell death that occurs during cell division or shortly thereafter.⁸

It has long been known that the SAC can trigger apoptosis as a consequence of prolonged activity or dysfunction; these connections are at the root of antimitotic cancer medicine and the development of mitotic kinase inhibitors, respectively.⁴ With the exception of BIM and MCL-1, which were recently identified as APC/C substrates,^9,10 the physical links between the mitotic checkpoint and apoptotic machineries had remained unknown.⁴ Our discovery of BUBR1 as a direct PIDDosome inhibitor operating at KTs suggests that the connection to apoptosis may be contained within the checkpoint apparatus itself, in the form of a caspase-activation platform (Fig. 1). The development of IF-compatible antibodies against RAIDD and caspase-2 should help test this hypothesis.

Disclosure of potential conflicts of interest

No potential conflicts of interest were disclosed.

Funding

Funding for this research was provided by The Searle Funds at the Chicago Community Trust (Grant ID: 11-SSP-196) and NIH/NCI (Grant ID: 1 RO1 CA178162-01).