Research Article

The PIDDosome, a Protein Complex Implicated in Activation of Caspase-2 in Response to Genotoxic Stress

See allHide authors and affiliations

Science  07 May 2004:
Vol. 304, Issue 5672, pp. 843-846
DOI: 10.1126/science.1095432


Apoptosis is triggered by activation of initiator caspases upon complex-mediated clustering of the inactive zymogen, as occurs in the caspase-9–activating apoptosome complex. Likewise, caspase-2, which is involved in stress-induced apoptosis, is recruited into a large protein complex, the molecular composition of which remains elusive. We show that activation of caspase-2 occurs in a complex that contains the death domain–containing protein PIDD, whose expression is induced by p53, and the adaptor protein RAIDD. Increased PIDD expression resulted in spontaneous activation of caspase-2 and sensitization to apoptosis by genotoxic stimuli. Because PIDD functions in p53-mediated apoptosis, the complex assembled by PIDD and caspase-2 is likely to regulate apoptosis induced by genotoxins.

Caspase-2 is the most evolutionarily conserved caspase (1), but its role in apoptosis remains enigmatic and controversial, despite the fact that it was the second mammalian caspase to be identified (2, 3). Although a role for caspase-2 in apoptosis initiated by β-amyloid cytotoxicity (4), trophic factor deprivation (5), and granzyme B (6) has been proposed, the interest in this caspase has been dampened due to the lack of an overt phenotype of caspase-2–deficient mice and limited knowledge of its mode of activation and downstream targets.

Nevertheless, caspase-2 is required in stress-induced apoptosis (7), and caspase-2–deficient germ cells and oocytes are resistant to cell death after treatment with chemotherapeutic drugs (6). It acts upstream of mitochondria, by inducing Bid cleavage, Bax translocation to mitochondria, and subsequent cytochrome c release (79). Caspase-2–induced apoptosis requires caspase-9 (9), and when added to purified mitochondria, only caspase-2 can directly cause cytochrome c release (8, 9).

Caspase-2 has a caspase-recruitment domain (CARD) pro-domain which, when overexpressed, associates with RAIDD, an adaptor protein containing a CARD and a death domain (DD) (10). RAIDD was proposed to interact with RIP1 and thereby to link tumor necrosis factor R1 (TNF-R1) activation with caspase-2–mediated apoptosis (10, 11). This notion could not be confirmed, however (6, 12).

We therefore sought to identify the molecular complex responsible for caspase-2 activation. Because caspases with CARD pro-domains usually are activated by CARD-containing proteins, we expressed CARD-containing proteins and tested their possible interaction with the CARD of caspase-2. The only interaction partner identified was RAIDD (13), in agreement with previous results (10, 11), which interacted specifically with caspase-2 and not with any of the other CARD-containing initiator caspases (Fig. 1A). Using a similar approach, we identified an interaction partner that bound the DD of RAIDD. Of the more than 20 different proteins analyzed, PIDD (p53-induced protein with a DD) was the only protein that was found to interact with RAIDD (Fig. 1B). Through overexpression, a PIDD–RAIDD–caspase-2 trimolecular complex could be reconstituted (Fig. 1C). A cooperativity between all three components to generate a stable complex was observed.

Fig. 1.

Formation of a complex containing caspase-2, RAIDD, and PIDD. (A) Specific interaction of RAIDD with caspase-2. CARD-containing DYDDDK epitope (FLAG)–tagged caspases were overexpressed in 293T cells along with VSV (vesicular stomatitis virus epitope)-tagged RAIDD. Immunoprecipitates from an antibody against FLAG (αFLAG or anti-FLAG) were analyzed for the presence of RAIDD by Western blot analysis. ASC represents control of a CARD-containing, noncaspase protein. (B) Interaction of the RAIDD-DD with the DD of PIDD. FLAG-tagged death domains (DD) containing proteins were overexpressed in 293T cells along with the VSV-tagged DD of RAIDD, and immunoprecipitates from an antibody against FLAG were analyzed for the presence of RAIDD. (C) Formation of a stable complex RAIDD–caspase-2 in the presence of PIDD. The indicated proteins were overexpressed in 293T cells, and proteins interacting with the CARD of caspase-2 were analyzed. (D) Domain structure of PIDD showing LRRs, the ZU5 domain, and DD. (E) Within cells, PIDD is processed between the two ZU5 domains [cleavage site (D)]. Left, Western blot of HeLa cells stably expressing tagged PIDD (N terminus, VSV; C terminus, FLAG). Right, Analysis of the VSV–PIDD–FLAG construct expressing Jurkat T cells. PIDD-C was detected with an antibody directed against the DD of PIDD. This antibody detects two additional bands (asterisk) that are absent in nontransfected cells.

PIDD was initially identified as a prime p53 target gene in an erythroleukemia cell line that undergoes G1 phase arrest and subsequent apoptosis after p53 expression (14). Independently, PIDD was also described as a DD-containing protein with unknown function (15). The N-terminal region of PIDD contains seven leucine-rich repeats (LRRs), a protein interaction motif found in various proteins with diverse functions (Fig. 1D), followed by two ZU-5 domains and a C-terminal DD. Prototypical ZU-5 interaction domains are present in the tight junction protein ZO-1, ankyrin, and Unc5-like netrin receptors. Expression of PIDD is widespread, but it is highest in the spleen (14). Expression of PIDD is elevated in cells transiently expressing p53 (14). The full-length protein contains 910 amino acids, migrates at ∼90 kD, and is frequently cleaved into two fragments of 48 kD (N-terminal region) and 51 kD (C-terminal region) [Fig. 1E and (15)]. The size of the fragment indicates that an unknown protease may cleave between the two ZU-5 domains [cleavage not inhibitable by z-Val-Ala-Asp-fmk (zVAD-fmk) (13)]. Processing does not result in the dissociation of the two fragments (13).

We next investigated whether the PIDD–RAIDD–caspase-2 complex (the PIDDosome) also forms under more physiological conditions. The characterization of a caspase-2–containing complex that is distinct from the Apaf-1–caspase-9 apoptosome was described (16). Formation of the caspase-2–containing complex occurred spontaneously in cell lysates after incubation at 37°C for 1 hour (16), which is in marked contrast to apoptosome formation that requires the addition of 2′deoxyadenosine 5′-triphosphate (dATP) and cytochrome c. In agreement with the published data, spontaneous caspase-2 complex formation was detectable in cell extracts from HeLa cells (human cervical carcinoma), Jurkat T cells (human leukemia T cell), 293T cells (human embryonic kidney cells), or Ramos B lymphocytes (human Burkitt's lymphoma) [Fig. 2, A and B, and (13)]. When cell extracts incubated at 4°C were analyzed by gel-filtration chromatography, caspase-2 eluted mainly in fractions corresponding to the monomeric protease (50 kD), whereas incubation of extracts at 37°C caused more caspase-2 to elute in fractions corresponding to a molecular mass of >670 kD. The same shift was observed for RAIDD. Complex formation was also observed when only the PIDD-interacting DD of RAIDD, but not the CARD of RAIDD, was expressed, which suggests that assembly of RAIDD into a complex is PIDD-dependent (fig. S1). Unlike caspase-2 and RAIDD, PIDD was already present in a complex of ∼550 kD prior to incubation at 37°C (Fig. 2B). The size of the PIDD complex further increased in response to a temperature elevation. Taken together, these data suggested that PIDD, RAIDD, and caspase-2 spontaneously assemble into a high-molecular-weight complex, when all three components are present in the same cell extract.

Fig. 2.

Spontaneous complex formation of PIDD, RAIDD, and caspase-2. (A) 293T cell lysates were subjected to size exclusion chromatography after incubation for 1 hour at 4°C or 37°C, respectively. Fractions were analyzed for the presence of caspase-2, RAIDD, and FADD (control). The elution positions of the molecular size markers are indicated. (B) As (A), but cell extracts of Jurkat T cells were analyzed for the presence of caspase-2, RAIDD, and PIDD.

Caspase-2 is an initiator caspase for which proteolytic processing is not absolutely required for activation (17, 18). To detect unequivocally active caspase-2, we incubated cell extracts of 293T cells overexpressing various PIDDosome components with biotinylated VAD-fmk, an inhibitor that irreversibly binds to the exposed active site of caspases (Fig. 3A). Subsequent streptavidin precipitates revealed that overexpression of either RAIDD or PIDD resulted in the generation of enzymatically active caspase-2. Activation by PIDD was dependent on the presence of endogenous RAIDD, because ablation of RAIDD expression by small, interfering RNA (siRNA) blocked caspase-2 processing (Fig. 3A). Overexpression of other apoptosis inducers, such as RIP1 or caspase-9, resulted in strong activation of caspase-3 but not of caspase-2 (Fig. 3A). Thus, PIDD-induced caspase-2 activation appears not to rely on caspase-3, a conclusion supported by the observation that PIDD-mediated caspase-2 processing was not impaired in caspase-3–deficient MCF7 cells (fig. S2).

Fig. 3.

Activation of caspase-2 during PIDDosome assembly. (A) Cell extracts of 293T cells transiently expressing the indicated activators of caspases were incubated with Biotin-VAD and streptavidin immunoprecipitates were analyzed for the presence of caspase-2 and cleaved caspase-3 (left four panels). The right panel shows the analysis of cell extracts of cells that were transfected with two RAIDD siRNA constructs [RAIDD.1 nonactive control, RAIDD.2 active, see (D)]. (B) Complex formation between RAIDD, PIDD, and active caspase-2. PIDDosome assembly was spontaneously induced in cell extracts from Jurkat T cells and active caspase-2 was precipitated with biotin-VAD. The presence of RAIDD, PIDD, and caspase-3 was determined in immunoblots (left). The anti-PIDD antibody used cross-reacts with two proteins (asterisk) that migrate close to processed PIDD (51-kD fragment). Bottom gel, PIDD incorporation into the assembling caspase-2 complex in Ramos B cells. Right, analysis of cell extracts. (C) Recruitment of caspase-2 and RAIDD into the assembling PIDD complex. PIDDosome assembly was spontaneously induced in cell extracts from 293T cells stably expressing FLAG-tagged PIDD, and PIDD was immunoprecipitated with FLAG-specific Agarose beads. Association with caspase-2, RAIDD, FADD, and cytochrome c was analyzed in immunoblots. (D) Requirement of PIDD and RAIDD for the activation of caspase-2. 293T cells were transfected with siRNA constructs for RAIDD or PIDD oligonucleotides, and active caspase-2 precipitated with biotin-–AD-fmk after 48 hours. siRNA-mediated reduction of RAIDD and PIDD was assayed in 293T cells after transfection with VSV–PIDD-N and VSV-RAIDD constructs (bottom).

We next examined whether caspase-2 activation occurs not only when the concentration of PIDD is artificially increased, but also in the spontaneously assembling PIDDosome in cell extracts of Jurkat T cells. Even though VAD-fmk is not a specific inhibitor of caspase-2, the remaining initiator caspases are not spontaneously activated in Jurkat T cells under the conditions of cell rupture that we used [(16) and see below]. Thus it was feasible to use biotin–VAD-fmk to immunoprecipitate only the active form of caspase-2. The amount of caspase-2 bound to biotin-VAD increased over time and reached a plateau after 60 min (Fig. 3B). In agreement with the fact that cleavage of initiator caspases is not required for activity, the majority of caspase-2 was found in its precursor form. Along with caspase-2, RAIDD and PIDD were coprecipitated, which indicated that caspase-2 is activated during complex formation. Recruitment of RAIDD was not detectable in murine embryonic fibroblasts deficient in caspase-2 (fig. S3), which confirmed that caspase-2 is the only caspase that spontaneously interacts with RAIDD when cells rupture. Spontaneous assembly of PIDD, RAIDD, and caspase-2 was also observed when PIDD, instead of caspase-2, was immunoprecipitated (Fig. 3C). Recruitment of RAIDD and caspase-2 was observed in a time-dependent manner. Unlike recruitment of caspase-2 with biotin-–VAD-fmk, which partially blocks processing of caspase-2 in cell extracts, the processed form of caspase-2 was detected in the PIDD complex. Caspase-2 activation was dependent on PIDD and RAIDD. Decreasing RAIDD or PIDD expression with siRNA abrogated the time-dependent caspase-2 activation, which indicated that the PIDDosome plays an important role in caspase-2 activation (Fig. 3D).

What is the physiological role of PIDDosome-induced caspase-2 activation? Transient overexpression of large amounts of PIDD resulted in slow cell death (14), associated with caspase-2 activation (Fig. 3A). Despite this, we were able to generate HeLa and B cell Ramos cells that stably expressed PIDD (Fig. 4A). In these PIDD-expressing cells, almost 50% of caspase-2 was processed, which indicated that an increase of PIDD concentration led to caspase activation (Fig. 4A). Spontaneous activation of caspase-2 in PIDD-expressing cells was confirmed when binding of biotin-VAD was analyzed. In contrast, cells expressing PIDD lacking the DD behaved like wild-type cells, and no caspase-2 activation was detectable, despite similar expression levels (13), which indicated that the RAIDD-interacting DD of PIDD was essential for caspase-2 activation.

Fig. 4.

Effect of increased PIDD expression on caspase-2 activation and sensitization of cells to genotoxin-induced cell death. (A) HeLa cells stably expressing equivalent levels of PIDD or a mutant PIDD lacking the DD (PIDDΔDD) were treated or not with the pan-caspase inhibitor zVAD (50 μM), and lysed. Cell extracts were incubated with biotin–VAD-fmk. Concentrations of zVAD-fmk that completely block the proteolytic activity of most caspases [see also (C)], do not block the PIDD-induced generation of the p19 fragment of caspase-2 when added to intact cells. Amounts of PIDD and PIDDΔDD expressed in cells were similar. Right, degree of caspase-2 processing in PIDD- and PIDDΔDD-expressing Ramos B cells. (B) Cytoplasmic extracts from wild-type (WT) HeLa cells or HeLa cells expressing PIDD were subjected to size-exclusion chromatography (Sephadex S-200, flow-rate 0.5 ml/min and 500-μl fractions) after incubation (1 hour) at 4°C or 37°C. Fractions were analyzed for the presence of caspase-2, RAIDD, and FADD (control). (C) HeLa cells subjected to doxorubicin (10 μg/ml) treatment were analyzed after different periods of time for cell death and caspase-3–dependent PARP cleavage (representative of three independent experiments).

Given that an increase of PIDD concentration triggers the activation of caspase-2 in living cells, we expect that PIDDosome complexes will be present in such cells expressing exogenous PIDD. Indeed, a fraction of caspase-2 and RAIDD were found in high-molecular-weight complexes without the need of prior incubation of cell extracts at 37°C (Fig. 4B).

A striking difference between PIDD-expressing cells and wild-type cells became apparent when cells were exposed to genotoxic stress. Within 5 hours after exposure to the topoisomerase inhibitors doxorubicin (Fig. 4C) or etoposide (13), 60% of the cells were dead, whereas in wild-type cells or cells expressing PIDDΔDD, most of the cells remained viable. This difference was also reflected at the level of caspase activation. In PIDD-expressing cells, the remaining, unprocessed caspase-2 was almost completely processed into the p19 fragment, and the activity of caspase-3 was highly increased as evidenced by the cleavage of poly(ADP-ribose) polymerase–1 (PARP1) (19).

Inhibition of PIDD expression attenuates p53-induced apoptosis, whereas overexpression of PIDD inhibits cell growth (14). Thus, PIDD appears to be a crucial target gene of a signaling pathway that is triggered when activated by p53 and ultimately leads to apoptosis. An increase of PIDD concentration promotes activation of caspase-2, as do stress conditions at low PIDD concentrations, as in our experiments on cell extracts. PIDD may tend to oligomerize, but the ultimate trigger leading to assembly seems to be a stress-related signal of unknown nature. Although the apoptosome-induced activation of caspase-9 usually leads to apoptosis, PIDDosome-based activation of caspase-2 is not toxic, and cells survive even if a notable fraction of the caspase-2 pool is activated. A second signal apparently is required for full commitment to caspase-2–mediated apoptosis. The caspase-2–activating PIDDosome complex, therefore, most resembles the inflammasome complex (20), which leads to caspase-1 activation without causing cell death. There is increasing evidence that even “apoptotic” caspases have pleiotropic effects. Caspase-8 activity is required for T cell proliferation (21), whereas caspase-3 regulates cell cycle progression in B cells (22). Similarly, active caspase-2 in cells expressing increased amounts of PIDD may have apoptosis-independent roles in processes such as DNA repair. Under genotoxic conditions, a fraction of nuclear caspase-2 might translocate to the cytoplasm, cleave Bid, and thereby accelerate apoptosis (8, 9). The identification of the PIDDosome as a platform for caspase-2 activation should facilitate the unraveling of these still poorly defined signaling pathways.

Supporting Online Material

Materials and Methods

Figs. S1 to S3

References and Notes

View Abstract

Navigate This Article