Mitochondrial pathway of apoptosis is ancestral in metazoans

Abstract

The mitochondrial pathway of apoptosis is the major mechanism of physiological cell death in vertebrates. In this pathway, proapoptotic members of the Bcl-2 family cause mitochondrial outer membrane permeabilization (MOMP), allowing the release of cytochrome c, which interacts with Apaf-1 to trigger caspase activation and apoptosis. Despite conservation of Bcl-2, Apaf-1, and caspases in invertebrate phyla, the existence of the mitochondrial pathway in any invertebrate is, at best, controversial. Here we show that apoptosis in a lophotrochozoan, planaria (phylum Platyhelminthes), is associated with MOMP and that cytochrome c triggers caspase activation in cytosolic extracts from these animals. Further, planarian Bcl-2 family proteins can induce and/or regulate cell death in yeast and can replace Bcl-2 proteins in mammalian cells to regulate MOMP. These results suggest that the mitochondrial pathway of apoptosis in animals predates the emergence of the vertebrates but was lost in some lineages (e.g., nematodes). In further support of this hypothesis, we surveyed the ability of cytochrome c to trigger caspase activation in cytosolic extracts from a variety of organisms and found this effect in cytosolic extracts from invertebrate deuterostomes (phylum Echinodermata).

The mitochondrial pathway of apoptosis, characterized by mitochondrial outer membrane permeabilization (MOMP) and the activation of caspases by cytochrome c, has been extensively described in vertebrates, including mammals (1, 2) and amphibians (3, 4). However, a role for MOMP and cytochrome c in invertebrate apoptosis remains controversial. Although both apoptosis and the proteins of the mitochondrial pathway of apoptosis are extensively described among invertebrate phyla (5), the available evidence suggests that the mitochondrial pathway arose with the vertebrates.

Current understanding of invertebrate apoptosis is based on extensive studies of two ecdysozoans, Caenorhabditis elegans and Drosophila. In the nematode C. elegans, apoptosis is induced when the APAF-1 homolog, CED-4, is released from the Bcl-2 protein, CED-9, by the BH3-only protein, EGL-1, allowing CED-4 to activate the caspase, CED-3 (6). No role for cytochrome c or MOMP has been found, and, indeed, CED-4 lacks the WD region that is predicted to be the cytochrome c-interacting region of the APAF-1 molecule (7). Antiapoptotic Bcl-2 proteins from humans do not show an ability to interact with APAF-1 (8) and instead inhibit apoptosis by blocking MOMP (3, 9), a function not observed for Ced-9 (10). However, a role for mitochondria in C. elegans apoptosis was suggested by an observation that mitochondria undergo fission upon EGL-1 expression (11), although no MOMP has been observed (10), and inhibition of the fission machinery affected programmed cell death (11). Thus, although elements of the mitochondrial pathway of apoptosis are present in C. elegans, the pathway as defined by MOMP and the role of cytochrome c does not appear to be conserved.

In Drosophila, the role of MOMP and cytochrome c in apoptosis is more controversial. It has been observed that cells undergoing apoptosis in Drosophila do not undergo MOMP as detected by release of cytochrome c (12), although this finding has been challenged (13). However, no release of cytochrome c was observed in Drosophila S2 cells during stress-induced apoptosis (14). ARK, the Drosophila APAF-1 homolog, is required for the activation of DRONC (15), the presumptive initiator caspase and caspase 9 ortholog, which, in turn, activates the effector caspase DRICE. The mechanism of ARK activation is unknown. ARK possesses a WD region (16); however, attempts to show cytochrome c-mediated activation of caspases in cytosolic extract from Drosophila cells (14, 17) have consistently failed. Indeed, knockdown of cytochrome c in Drosophila S2 cells had no effect on ARK-dependent caspase activation or apoptosis (14, 17, 18). In contrast, genetic studies have suggested that caspase activation involved in Drosophila sperm differentiation (19) and in apoptosis in pupal eye development (20) may depend on cytochrome c. However, biochemical evidence that cytochrome c activates ARK in any setting is lacking. Although it has been suggested that ARK forms an oligomeric apoptosome in the absence of cytochrome c (21), there is no evidence for caspase activation by this structure. Reaper and Hid induced changes in mitochondrial morphology (13, 22) and cytochrome c distribution (13); however, cytochrome c redistribution was shown to be caspase-dependent and is thus unlikely to be involved in the initiation of the apoptotic program. Nevertheless, a role for the mitochondrial fission machinery in apoptosis was identified, suggesting that at least some aspects of the mitochondrial pathway of apoptosis exist in insects. Therefore, a role for cytochrome c in activating the APAF1 homolog in insect apoptosis remains, at best, controversial.

In this study, we examined the mitochondrial pathway of apoptosis in a lophotrochozoan, the freshwater planarian (phylum Platyhelminthes), and in an invertebrate deuterostome, the purple sea urchin, Strongylocentrotus purpuratus (phylum Echinodermata). Our results support a revised phylogeny of the mitochondrial pathway of apoptosis, in which this pathway is ancestral in the animals and may have been lost in some phyla, including those of the Ecdysozoa.

Results

Planarian cells die in response to several agents, including γ-radiation, and this death proceeds via apoptosis, as determined by ultrastructural changes and DNA fragmentation (by terminal deoxynucleotidyl transferase dUTP nick end labeling) (23). To determine whether planarian cells, including the cycling planarian stem cells known as neoblasts (24, 25), undergo death via the mitochondrial pathway of apoptosis, a single cell suspension enriched in planarian (Schmidtea mediterranea) neoblasts was prepared and cultured with or without the broad-spectrum caspase inhibitor qVD-OPh, followed by γ-radiation. Apoptosis was assessed 24 h later by the binding of Annexin V to externalized phosphatidylserine before loss of plasma membrane integrity [assessed by uptake of propidium iodide (PI)] and the loss of mitochondrial membrane potential, ΔΨm. We observed an increase in Annexin V⁺, PI⁻ cells following irradiation (Fig. 1A), which was effectively blocked by preincubation with qVD-OPh (Fig. 1B). We also observed a reduction of ΔΨm in irradiated cells (Fig. S1).

Fig. 1.

γ-radiation induces phosphatidylserine externalization, caspase activation, and cytochrome c release in planaria. (A) A cell fraction enriched in planaria neoblasts (S. mediterranea) was subjected to 50 Gy of γ-radiation (γ-rad). Twenty-four hours later, cells were stained with Annexin V-FITC and PI and analyzed by flow cytometry. Representative dot plots are shown. (B) As in A, except that cells were incubated in the presence or absence of 10 μM qVD-OPh 1 h before γ-radiation. Each data point represents two or three independent experiments. Error bars represent SD. (C–E) Planaria (D. dorotocephala) were subjected, or not, to 100 Gy of γ-radiation, and cytosolic extract was prepared 24 h later. (C) Rate of cleavage of Ac-DEVD-afc by cytosolic extract from untreated or irradiated planaria, preincubated in the presence or absence of a broad protease inhibitor mixture (Materials and Methods) and incubated with or without 10 μM qVD-OPh. RFU, relative fluorescence units. Error bars represent SD. (D) ³⁵S-labeled wild-type iCAD or mutant iCAD (D117E/D224E) (m-iCAD) was incubated with extract from untreated or irradiated planaria preincubated with a protease inhibitor mixture, in the presence or absence of 10 μM qVD-OPh (Left) or with activated human recombinant caspase 3 (rCasp3) (Right). Samples were resolved by SDS/PAGE and analyzed by autoradiography, and percent cleavage was determined by densitometry analysis. Arrowhead indicates cleaved product. (E) Cytosolic extracts from untreated and irradiated planaria were examined for cytochrome c (cyt c) and actin content by Western immunoblot.

Cytosolic extracts prepared from another planarian (Dugesia dorotocephala), 24 h after irradiation of the whole organism, cleaved the synthetic caspase substrate Ac-DEVD-afc (Fig. 1C). This cleavage was blocked by qVD-OPh, but not by a mixture of inhibitors that block members of other protease families (Fig. 1C), suggesting that irradiation had induced a caspase responsible for orchestrating apoptotic death. To test this idea, cytosolic extracts from untreated or irradiated planaria were incubated with mammalian iCAD, a known caspase substrate, or iCAD mutated at its caspase cleavage sites (26). The cytosolic extracts from irradiated animals cleaved wild-type iCAD to the same fragments observed with recombinant human caspase 3. Although some substrate cleavage was observed in the extract from untreated animals, cleavage was increased in the extract from irradiated animals (Fig. 1D). The cleavage was blocked by qVD-OPh and was not observed with the mutant substrate (Fig. 1D). To assess release of cytochrome c, as an indication of MOMP during apoptosis, we examined the cytosolic extracts by immunoblot and found that irradiation effectively induced release of cytochrome c into the cytosolic fraction (Fig. 1E).

Similar results were obtained by using another stressor, heat, to which these organisms are likely exposed under natural conditions. As with γ-radiation, mild heat stress induced caspase activation and cytochrome c release (Fig. S2).

Because cytochrome c mediates activation of caspases in the mitochondrial pathway of apoptosis in vertebrates, we hypothesized that it might also induce caspase activation in planaria. Cytosolic extracts were prepared from untreated D. dorotocephala, and caspase activity was measured. Addition of mammalian cytochrome c rapidly triggered DEVDase activity, which was inhibited by broad-spectrum caspase inhibitors zVAD-fmk and qVD-OPh (Fig. 2 A and B). Cytochrome c-activated extracts cleaved wild-type iCAD, but not the caspase-uncleavable mutant, in the same manner as recombinant human caspase 3 (Fig. 2C). Untreated extract induced a variable background of substrate cleavage, perhaps due to contamination with low levels of endogenous cytochrome c; however, addition of cytochrome c induced markedly more cleavage, and this cleavage was blocked by qVD-OPh. Similar results were observed when cytochrome c-activated extracts were incubated with wild-type human PARP or its caspase-uncleavable mutant (27) (Fig. 2D). Although general degradation of the substrate was observed, only cytochrome c-activated extract induced cleavage of the substrate to an 85-kDa band, dependent on the caspase cleavage site. In vertebrate extracts, cytochrome c from mammals and insects triggers caspase activation, whereas that from yeast does not (28), and this pattern was also observed in planaria extracts (Fig. 2E). Cytochrome c did not induce other protease activities, as assessed by cleavage of other synthetic caspase, calpain/proteasome, or cathepsin substrates (Fig. S3 A and B).

Fig. 2.

Cytochrome c activates caspases in planaria cytosolic extract (D. dorotocephala). (A) Planaria cytosolic extract was preincubated in the presence or absence of 10 μM mammalian (horse heart) cytochrome c. Ac-DEVD-afc was added, and cleavage was measured. (B) Rate of cleavage of Ac-DEVD-afc by planaria cytosolic extract preincubated in the presence or absence of 10 μM cytochrome c and either DMSO (vehicle), zVAD-fmk (1 nM to 100 μM), or qVD-OPh (1 nM to 100 μM). (C) ³⁵S-labeled wild-type iCAD or mutant iCAD (D117E/D224E) was incubated with planaria cytosolic extract preincubated with a protease inhibitor mixture in the presence or absence of 10 μM cytochrome c and 10 μM qVD-OPh (Left) or with activated human recombinant caspase 3 (Right). Samples were resolved by SDS/PAGE and analyzed by autoradiography, and percent cleavage was determined by densitometry analysis. Arrowhead indicates cleaved product. (D) Wild-type or mutant PARP (D214A) was incubated with planaria cytosolic extract in the presence or absence of 10 μM cytochrome c and 10 μM qVD-OPh or with activated human recombinant caspase 3, resolved by SDS/PAGE, and examined by immunoblot. Arrowhead indicates cleaved product. (E) Rate of cleavage of Ac-DEVD-afc by planaria extract in the absence or presence of 1, 10, or 100 μM mammalian (horse heart), insect (Manduca sexta), or yeast (S. cerevisiae) cytochrome c. (F) Rate of cleavage of Ac-DEVD-afc by planaria cytosolic extract preincubated or not with supernatant from enriched planaria (D. dorotocephala) mitochondria disrupted with water. Data are representative of at least three independent experiments. Error bars represent SD. (G) Cytochrome c-induced DEVDase activation in mammalian cytolic extract is inhibited by the CARD domains of human APAF1 or caspase-9, but not by the CARD domain of planaria APAF1. Cytosolic extract from 293 was incubated with 10 μM cytochrome c plus doubling dilutions of the indicated recombinant proteins (starting concentrations were equivalent before addition, as indicated by immunoblot; Fig. 2H, Right). (H) Cytochrome c-induced DEVDase activation in planaria cytolic extract is inhibited by the CARD domain of planaria APAF1, but not by the CARD domains of human APAF1 or caspase-9. Cytosolic extract from planaria was incubated with 10 μM cytochrome c plus doubling dilutions of the indicated recombinant proteins (starting concentrations were equivalent before addition, as indicated by immunoblot; Right).

To determine whether planarian mitochondrial proteins such as cytochrome c can similarly trigger caspase activation, planaria (D. dorotocephala) mitochondria were enriched and then disrupted with water. The supernatants induced caspase activation in both planaria cytosolic extract (Fig. 2F) and in Xenopus egg extract (Fig. S3C). This result is consistent with an ability of planaria cytochrome c to induce DEVDase activity. We suggest that the release of cytochrome c, upon induction of apoptosis in planaria, can induce caspase activity, thus implicating the mitochondrial pathway in cell death induced by γ-radiation and heat shock in planaria.

In vertebrates, cytochrome c triggers caspase activation via oligomerization of the adapter protein APAF1 (29). A planarian protein with homology to APAF1 was identified in the S. mediterranea genome (Fig. S4), and the region containing the caspase-recruitment domain (CARD) was cloned and expressed in Escherichia coli as a GST-fusion protein. The corresponding human APAF1 region was similarly expressed. In mammalian cytosolic extracts, the human, but not the planarian, APAF1 CARD fragment inhibited caspase activation in response to mammalian cytochrome c (Fig. 2G and Fig. S5). Therefore, it is likely that the human APAF1 CARD-containing fragment acts as a competitive inhibitor of the human APAF1 apoptosome, presumably by competing for the caspase-9 CARD domain. This hypothesis is further supported by the ability of recombinant human caspase-9 CARD region to inhibit in this assay (Fig. 2G and Fig. S5). In contrast, the planarian, but not the human, APAF1 CARD fragment inhibited the ability of mammalian cytochrome c to trigger caspase activation in planarian extracts (Fig. 2H and Fig. S5). Thus, it is likely that cytochrome c triggers caspase activation in planaria extracts via the planarian APAF1 homolog.

Recently, antiapoptotic and proapoptotic Bcl-2 family proteins were identified in another lophotrochazoan, Schistosoma mansonii (30). We interrogated the genome of S. mediterranea and identified several candidates for proapoptotic Bcl-2 effector proteins (Table S1). The mammalian Bcl-2 effector proteins, Bax and Bak (31), trigger cell death in the yeast Saccharomyces cerevisiae, and we therefore tested the planarian candidates for such killing. Of these, one candidate (Smed-Bak-2) effectively triggered cell death (Fig. S6A), which we designate herein as “Smed Bak” based on its homology to human Bak. A mutation in Smed Bak that corresponds to an inactivating mutation in human Bak was introduced (32), and we found that this mutation inactivated its ability to kill yeast (Fig. 3A). We then tested whether this killing was antagonized by either human BCL-xL or a described S. mediterranea antiapoptotic Bcl-2 protein (23). Both human and planarian antiapoptotic Bcl-2 proteins blocked killing by Smed Bak and mouse Bax (Fig. 3B and Fig. S6B).

Fig. 3.

Planaria Bak localizes to mitochondria and induces MOMP and cell death. (A and B) S. cerevisiae were transformed with the indicated constructs, plated in serial dilution, and induced to express the transformed gene(s). (C and D) Bax/Bak-deficient MEFs stably expressing Omi-mCherry were transfected with the indicated constructs in the presence of qVD-OPh (32 μM). At 24 h, cells were analyzed by confocal microscopy (C) or flow cytometry (D) for MOMP. Displayed is the percentage of cells that have undergone MOMP relative to the transfected population, averaged over three independent experiments ± SD. (E) HeLa cells were transfected with Venus-Smed Bak and mitochondrial targeted Cerulean (fused to the C-terminal 20 amino acids of human Bcl-xL). Cells were permeabilized with digitonin and imaged by confocal microscopy. In the merged image, the cerulean is false colored red.

Bax and Bak are essential for MOMP in mammalian cells (33). We therefore tested whether Smed Bak could induce MOMP in Bax, Bak double-deficient mouse embryo fibroblasts (MEFs), which do not undergo MOMP in response to apoptotic signals (33). We used cells expressing the mitochondrial intermembrane-space marker Omi-mCherry, which upon MOMP is released to the cytosol and degraded by the proteasome (34). We transiently expressed fluorescent Venus-tagged Smed Bak and observed the induction of MOMP in the cells expressing the green Venus (Fig. S7). Quantitative confocal analysis revealed that Smed Bak was at least as effective as human Bax in inducing MOMP (Fig. 3C). Importantly, both mutant Smed Bak and Smed Bak coexpressed with human Bcl-xL failed to induce MOMP. We obtained similar results using flow cytometry to detect cells that had undergone MOMP [detected by lower Omi-mCherry expression (34)] (Fig. 3D). To determine whether Smed Bak localized to mitochondria, Venus Smed Bak was coexpressed with mitochondrial-targeted Cerulean in HeLa cells (35).

Before imaging, cells were treated with digitonin to selectively permeabilize the plasma membrane and thereby release soluble cytosolic proteins. By using this approach, Smed Bak was found to colocalize with mitochondria (Fig. 3E). These results demonstrate that Smed Bak can localize to the mitochondria and induce MOMP in a Bcl-xL–regulated manner.

Our findings suggest that the mitochondrial pathway of apoptosis, as defined by cytochrome c-induced caspase activation, exists not only in vertebrates but also in lophotrochozoans. Therefore, other invertebrate phyla may display this pathway. In keeping with this idea, we found that egg cytosolic extracts from two echinoderms, purple sea urchin (S. purpuratus) and sand dollar (Dendraster excentricus), both showed robust cytochrome c-induced DEVDase activity, which was blocked by caspase inhibitors (Fig. 4 A and B and Fig. S8), but not by a mixture of other protease inhibitors (Fig. S8A). Cytochrome c-activated sea urchin egg extracts cleaved iCAD, but not its caspase-uncleavable mutant (Fig. 4C). As with vertebrate (28) and planaria cytosolic extracts, mammalian and insect cytochrome c triggered DEVDase activity in sea urchin egg extracts, whereas yeast cytochrome c did not (Fig. 4D). Cytochrome c-induced caspase activity was enhanced by the addition of exogenous dATP (Fig. S8 B and C). Cytochrome c did not induce cleavage of other synthetic caspase, calpain/proteasome, or cathepsin substrates (Fig. S8D).

Fig. 4.

Cytochrome c activates caspases in sea urchin egg cytosolic extract (S. purpuratus). (A) Cleavage of Ac-DEVD-afc by sea urchin egg cytosolic extract preincubated in the presence or absence of 10 μM mammalian (horse heart) cytochrome c and 1 mM dATP. (B) Rate of cleavage of Ac-DEVD-afc by sea urchin egg cytosolic extract preincubated in the presence or absence of 10 μM cytochrome c/1 mM dATP and either DMSO (vehicle), zVAD-fmk (1 nM to 100 μM), or qVD-OPh (1 nM to 100 μM). (C) ³⁵S-labeled wild-type or mutant iCAD (D117E/D224E) (m-iCAD) was incubated with sea urchin egg cytosolic extract preincubated with a protease inhibitor mixture in the presence or absence of 10 μM cytochrome c/1 mM dATP and 10 μM qVD-OPh or with activated human recombinant caspase 3. Samples were resolved by SDS/PAGE and analyzed by autoradiography, and percent cleavage was determined by densitometry analysis. Arrowhead indicates cleaved product. Note that the rightmost four lanes are also used as controls in Fig. 2C, Right, as they derive from the same experiment. (D) Rate of cleavage of Ac-DEVD-afc by sea urchin egg extract in the absence or presence of 1, 10, or 100 μM mammalian (horse heart), insect (Manduca sexta), or yeast (S. cerevisiae) cytochrome c and 1 mM dATP. Data are representative of at least three independent experiments. Error bars represent SD.

Discussion

Apoptosis and the molecules of the mitochondrial pathway for this form of cell death are conserved among the animals, but evidence has not supported the existence of cytochrome c-induced caspase activation beyond the vertebrates (5). Our studies herein suggest that if apoptosis via APAF-1–like molecules proceeds independently of MOMP in some organisms, this mechanism is likely to have derived from an ancestral pathway in which MOMP-mediated release of cytochrome c activates caspases—that is, the mitochondrial pathway as seen in vertebrates. Indeed, the most parsimonious scenario is that the mitochondrial pathway of apoptosis arose once, before the emergence of the deuterostomes, and that all or portions of the pathway have been lost in some lineages. This hypothesis is supported by the observation that proapoptotic and antiapoptotic Bcl-2 proteins in the lophotrochazoan, S. mansonii, promote or inhibit apoptosis in mammalian cells (30), whereas the nematode Bcl-2 protein, CED-9, does not (30).

Apoptosis, by morphological criteria, has unambiguously been described throughout the animals, including Sponges, Cnidaria, Platyhelminthes, Mollusks, Nematodes, Arthropods, Echinoderms, and Chordates, both invertebrate (Ciona) and vertebrate (5). In addition to the C. elegans and Drosophila homologs, homologs of caspases are described in many of these other phyla, as are Bcl-2 homologs (Table S1). It is clear, however, that sequences of genes and proteins relating to the mitochondrial pathway of apoptosis do not predict the function of this pathway in organisms of a given phylum. Our results suggest that, in addition to sequence-based phyletic analysis, an understanding of the underlying mechanism of cellular processes requires cell and biochemical approaches to determine whether complex processes such as the mitochondrial pathway are, indeed, conserved.

Materials and Methods

Animals.

S. mediterranea were maintained in Montjuic salts (36) (pMedia). D. dorotocephala were purchased from Carolina Biological Supply and maintained in pMedia. S. purpuratus and D. excentricus were collected locally in San Diego.

DEVDase Assay.

For each data point, 100 μg of cytosol extract was used. Samples were activated by 1 μL of 100 μM horse heart cytochrome c (Sigma) for 30 min at room temp [or insect (M. sexta), yeast (S. cerevisiae), or mitochondria lysate as indicated]. The human cytosol included 1 μL of 10 mM dATP (Invitrogen). DEVDase activity was determined by adding 100 μL of Caspase Buffer [20 mM Pipes at pH 7.4, 100 mM NaCl, 1 mM EDTA, 0.1% CHAPS, 10% (wt/vol) sucrose, and 10 mM DTT] with 0.1 mM Ac-DEVD-afc (SM Biochemicals). Ac-DEVD-afc cleavage was measured in a SpectraMax Gemini XPS plate reader (Molecular Devices) with excitation at 400 nm and emission at 505 nm. Readings were measured each minute for 30 min. Results are expressed as RFU per minute (37).

Caspase Activation Assay.

To analyze cleavage of protein substrates, cytosolic extract was added to in vitro transcribed/translated wild-type or mutant (D117E, D224E) iCAD or wild-type or mutant (D214A) PARP. Samples were then subjected to SDS/PAGE and analyzed by autoradiography (iCAD) or immunoblot (PARP).

Yeast Death Assays.

S. cerevisiae were transformed with pYES2.1 with or without pADH plasmids with the indicated inserts and selected on the appropriate dropout noninducing medium. pYES2.1 has an inducible promoter (GAL1), and pADH has a constitutive promoter (alcohol dehydrogenase). Colonies were picked, suspended in water, serially diluted 10-fold, and plated on both inducing and noninducing medium. Plates were photographed at the indicated times.

MOMP Assay.

Bax/Bak double-knockout MEFs stably expressing Omi-mCherry were transfected with the indicated plasmids by using Lipofectamine 2000 (Invitrogen). Transfected cells were cultured with qVD-OPh at a concentration of 32 μM for 20–23 h. MOMP was evaluated by flow cytometry (LSR II; BD Biosciences) and analyzed by using FloJo software or a spinning disk confocal microscope (Zeiss) and Slidebook software (Intelligent Imaging Innovations).

Supporting Information.

Detailed methods and methods not described here can be found in SI Materials and Methods.