Starfish Apaf-1 activates effector caspase-3/9 upon apoptosis of aged eggs

Caspase-3-related DEVDase activity is initiated upon apoptosis in unfertilized starfish eggs. In this study, we cloned a starfish procaspase-3 corresponding to mammalian effector caspase containing a CARD that is similar to the amino terminal CARD of mammalian capsase-9, and we named it procaspase-3/9. Recombinant procaspase-3/9 expressed at 15 °C was cleaved to form active caspase-3/9 which has DEVDase activity. Microinjection of the active caspase-3/9 into starfish oocytes/eggs induced apoptosis. An antibody against the recombinant protein recognized endogenous procaspase-3/9 in starfish oocytes, which was cleaved upon apoptosis in aged unfertilized eggs. These results indicate that caspase-3/9 is an effector caspase in starfish. To verify the mechanism of caspase-3/9 activation, we cloned starfish Apaf-1 containing a CARD, a NOD, and 11 WD40 repeat regions, and we named it sfApaf-1. Recombinant sfApaf-1 CARD interacts with recombinant caspase-3/9 CARD and with endogenous procaspase-3/9 in cell-free preparations made from starfish oocytes, causing the formation of active caspase-3/9. When the cell-free preparation without mitochondria was incubated with inactive recombinant procaspase-3/9 expressed at 37 °C, DEVDase activity increased and apoptosome-like complexes were formed in the high molecular weight fractions containing both sfApaf-1 and cleaved caspase-3/9. These results suggest that sfApaf-1 activation is not dependent on cytochrome c.

Meiosis reinitiation of oocytes in starfish (Asterina pectinifera) is stimulated by the hormone 1-methyladenine (1-MA), which is a prerequisite for fertilization. Without insemination or fertilization, endogenous caspase-3-like activity increases in aged eggs ∼10 h after 1-MA stimulation, followed by blebbing and apoptotic body formation [35][36][37] . Starfish eggs develop the competence to die when high extracellular signal-regulated kinase (ERK) activity is maintained for several hours 36,38 . After this ERK-dependent period, ERK is spontaneously inactivated, and apoptosis follows 36 . If starfish oocytes are not treated with 1-MA, they are alive over several days in seawater. Thus, hormonal stimulation leads to apoptosis, whereas fertilization blocks the apoptotic program.
As the starfish is one of the species located close to the evolutionary branching point between vertebrates and nematodes, it should help to provide clues to elucidate the relationship of apoptosis between vertebrates and nematodes. We therefore report here on the molecular mechanisms of starfish apoptosis, including the identification of caspase and its activation.

Results
Cloning and activity of starfish caspase-3/9 gene. In our previous studies, we detected caspase-3 (DEVDase) activity in unfertilized starfish eggs ∼30 min before blebbing 35,36,39 . To identify the responsible enzyme, we first cloned a caspase cDNA from a cDNA library of starfish ovaries using degenerate primers against caspase-3. We obtained a complete cDNA encoding a protein of 452 amino acids with a predicted molecular weight of 50.8 kDa. The deduced amino acid sequence contained the catalytic cysteine site (C305) in the pentameric QAC(R/G)G motif, which are perfectly conserved across species (Fig. 1a) 40 . Two putative cleavage sites are located at Asp318 and Asp356 (Fig. 1a). Comparing the predicted protein with those from other species using a BLAST search tool, we found that it exhibits a high level of sequence identity to effector caspases such as caspase-3 and -7, although it has a caspase recruitment domain (CARD) that is similar to those of initiator caspases such as caspase-2 and -9 (Fig. 1b). Starfish caspase-3/9 has caspase-9-like sequence in the N-terminal side, and caspase-3-like sequence in the C-terminal side ( Supplementary Fig. S1 a and b). We hypothesized that the starfish caspase has properties of both initiator and effector caspase. To examine this possibility, we prepared recombinant starfish procaspase-His 6 . Full-length starfish procaspase (about 60.8 kDa in size) was observed at 37 °C (Fig. 2a, lane 2), whereas two cleaved fragments (about 40 kDa and 48 kDa) were detected at 15 °C (Fig. 2a,  lane 3), suggesting that active recombinant caspase was formed at this temperature. The molecular weights of full-length caspase were slightly larger than the predicted procaspase (50.8 kDa), probably due to the presence of acidic amino acids 41 . When we mixed the cleaved caspase expressed at 15 °C with various substrates, it hydrolized only DEVD sequence that is recognized by an active mammalian caspase-3 (Fig. 2b), indicating that the activity of starfish caspase is very specific. Because starfish caspase has a caspase-9-like CARD and caspase-3-like DEVDase activity, we named it "caspase-3/9". Recombinant procaspase-3/9-His 6 expressed at 37 °C was rather inactive ( Fig. 2c), suggesting that much proportion of the proteins has undergone unfolding at this temperature, as normal environmental temperatures for starfish is around 20 °C in seawater.

Microinjection of recombinant caspase-3/9 into immature oocytes induces blebbing.
To determine whether caspase-3/9 is sufficient to trigger apoptosis in starfish oocytes, we microinjected purified active caspase-3/9-His 6 (1.1 or 0.56 µg/mL final concentration) expressed at 15 °C or control buffer into the cytoplasm of immature oocytes. Blebbing was initiated within 1-2 h after microinjection of purified active caspase-3/9-His 6 ( Fig. 2d upper panels and 2e), whereas all oocytes injected with control buffer were alive (Fig. 2d lower panels and 2e). These results suggest that caspase-3/9 is an executor of apoptosis.
In our previous studies, we reported that starfish apoptosis is induced by spontaneous inactivation of extracellular signal-regulated kinase (ERK) followed by activation of p38 MAPK 36 . Because artificial inactivation of ERK accelerated the timing of apoptosis 36 , we treated pre-apoptotic eggs with the MEK inhibitor U0126. As expected, apoptosis induction and procaspase-3/9 cleavage were observed earlier in the U0126-treated eggs than in the untreated eggs (Supplementary Fig. S2a and b). When we checked the timing of caspase-3/9 cleavage as well as inactivation/activation of ERK and p38 MAPK, we found that cleaved caspase-3/9 appeared after ERK  inactivation, prior to p38 MAPK activation (Fig. 3c). Thus, it is likely that ERK inactivation induces the activation of both caspase-3/9 and p38 MAPK.
Cloning of starfish Apaf-1. In mammalian apoptosis, the CARD of procaspase-9 interacts with the CARD of Apaf-1, which is followed by procaspase-9 cleavage and activation 13,14 . This caspase activation mechanism, including the formation of caspase multimers with Apaf-1/CED-4/Dark, is conserved from nematodes to mammals 19 . As starfish caspase-3/9 has CARD, starfish eggs may express starfish Apaf-1, which would interact with caspase-3/9 CARD upon apoptosis.
To generate starfish Apaf-1 cDNA, we used RT-PCR. The resulting complete cDNA encoded a protein of 1,238 amino acids with a predicted molecular weight of 138.5 kDa (Fig. 4a). Comparing the cDNA with other species using a BLAST search tool, it showed 36% identity with human Apaf-1, 23% identity with D. melanogaster dark, and 22% identity with C. elegans ced-4. These results strongly support the idea that the cDNA we generated encodes starfish Apaf-1, which is evolutionarily conserved ( Supplementary Figs S3 and S4). Starfish Apaf-1 has one putative nucleotide-binding site (GXXGXGK) and several related motifs, CARD, a nucleotide-binding oligomerization domain (NOD), and 11 WD40 repeat regions (Fig. 4b). We predicted that starfish Apaf-1 interacts with caspase-3/9, causing activation of caspase-3/9 in a way similar to mammalian caspase-9.

Starfish Apaf-1 binds to caspase-3/9 via a CARD-CARD interaction.
To determine whether a CARD-CARD interaction occurs between starfish Apaf-1 (sfApaf-1) and caspase-3/9, we first expressed GST-sfApaf-1-CARD 1-134aa (GST-A-CARD) and His 6 -caspase-3/9-CARD 1-130aa (His-C-CARD). Either purified GST-A-CARD or control GST was bound to glutathione Sepharose 4B beads, followed by incubation with purified His-C-CARD. When the glutathione Sepharose 4B beads were subsequently treated with glutathione elution buffer, GST-A-CARD and His-C-CARD were co-eluted (Fig. 5a, right panel), whereas GST alone without His-C-CARD was eluted in the control experiment (Fig. 5a, left panel). These results indicate that A-CARD interacts with C-CARD. Next, we investigated whether recombinant A-CARD interacts with endogenous caspase-3/9 by pull-down assays using cell-free preparations made from starfish oocytes. GST-A-CARD and control GST beads were treated with cell-free preparation, and precipitated proteins with beads were analyzed by SDS-PAGE and western blotting with anti-caspase-3/9 antibody. We found that endogenous procaspase-3/9 in cell-free preparations was efficiently precipitated by purified GST-A-CARD, but not by GST alone (Fig. 5b). These results indicate that caspase-3/9 can interact directly with sfApaf-1 through their CARDs.
Starfish Apaf-1 CARD associates with endogenous caspase-3/9. Mammalian Apaf-1 associates with the apoptosome-activating caspase-9, and more importantly, the sole Apaf-1 CARD can increase caspase-9 activity by forming a large hetero-oligomer of Apaf-1-CARD/caspase-9 complex 42 . To examine whether GST-A-CARD can activate endogenous procaspase-3/9, we incubated purified GST-A-CARD in cell-free preparations made from starfish oocytes. We found that the DEVDase activity increased (Fig. 6a cell-free + GST-A-CARD) and endogenous procaspase-3/9 was cleaved in the presence of GST-A-CARD (Fig. 6b, bottom panel), but we detected no activity increase and no cleavage of procaspase-3/9 in the presence of the GST control ( Fig. 6a cell-free + GST; 6b, upper panel). These results suggest that starfish Apaf-1 CARD activates caspase-3/9 in a manner similar to human Apaf-1 CARD. To examine whether GST-A-CARD in cell-free preparations made from starfish oocytes forms a large hetero-oligomer containing caspase-3/9, we performed the gel filtration analysis of GST-A-CARD incubated with or without the cell-free preparation. The GST-A-CARD (39.7 kDa) was eluted in a peak centered around the low molecular weight fraction 32 in the absence of the cell-free preparation (Fig. 6c, top), whereas it assembled into a large complex of roughly 0.7-1.4 MDa containing cleaved caspase-3/9 with high DEVDase activity ( Fig. 6c: fractions 15-22 and 6d, yellow column). Instead, in the control gel filtration of cell-free preparation in the absence of A-CARD, the basal DEVDase activity of endogenous caspase-3/9 was low (Fig. 6d, blue column) without cleaved caspase-3/9 (Fig. 6c, bottom). Starfish Apaf-1-CARD therefore induces the formation of a large complex involving caspase-3/9, and the catalytic activity of caspase-3/9 is enhanced in this complex. In addition, these results suggest that endogenous procaspase-3/9 CARD is exposed in order to interact with exogenous sfApaf-1 CARD.
When the cell-free preparation without stimulation was gel filtered, followed by western blotting using the antibody against sfApaf-1 CARD, distribution of sfApaf-1 (Fig. 7b: fractions 31 and 32) was not always the same as that of procaspase-3/9 (Fig. 7b: fractions 32-36), suggesting that sfApaf-1 did not interact with procaspase-3/9 in the cell-free preparation before stimulation of apoptosis. Thus, it is likely that endogenous sfApaf-1 CARD was not exposed to interact with procaspase-3/9 CARD. Instead, endogenous procaspase-3/9 CARD should be exposed because GST-sfApaf-1-CARD could pull down procaspase-3/9 (Fig. 5b). In mammalian cells, Survivin-HBXIP (hepatitis B X-interacting protein) complexes or TUCAN bind procaspase-9, preventing procaspase-9 activation 44,45 . If procaspase-3/9 in starfish oocytes was blocked by such endogenous inhibitor proteins, recombinant procaspase-3/9-His 6 may absorb the possible inhibitors suppressing activation of endogenous procaspase-3/9. As expected, DEVDase activity in the cell-free preparation increased after the addition of procaspase-3/9-His 6 expressed at 37 °C (Fig. 7c), which was rather inactive initially (Fig. 2c). To determine whether dimerization of procaspase-3/9 46 occurred in the cell-free preparation which had been incubated with inactive recombinant procaspase-3/9-His 6 , we performed gel filtration analysis. To our surprise, procaspase-3/9,  (2) with IPTG induction at 37 °C. (b) Ultracentrifuged cell-free preparations were fractionated by gel filtration chromatography. Endogenous procaspase-3/9 and sfApaf-1 were detected by western blotting with anti-caspase-3/9 and anti-sfApaf-1 antibodies. (c) Activation of endogenous caspase-3/9 in cell-free preparations by treatment with procaspase-3/9-His 6 . A time course of DEVDase activity was measured at the indicated times after adding either procaspase-3/9-His 6 or buffer (control). (d) Ultracentrifuged cell-free preparations were incubated with recombinant procaspase-3/9-His 6 , and fractionated by gel filtration chromatography. Fractions were analyzed by western blotting with the anticaspase-3/9 antibody and anti-sfApaf-1 antibody. (e) DEVDase activity in fractions was measured by the cleavage of Ac-DEVD-MCA. The red column is from gel filtered cell-free preparations with procaspase-3/9-His 6 , and the blue column is from gel filtered cell-free preparations without recombinant protein. cleaved caspase-3/9, and sfApaf-1 were eluted in the high molecular weight fractions corresponding to an apparent molecular weight of 0.7-1.4 MDa (Fig. 7d). Those fractions had DEVDase activity (Fig. 7e, red column), suggesting that sfApaf-1 formed the apoptosome. Because the apoptosome-like complex was formed in cell-free preparations, which had been ultracentrifuged to remove mitochondria, cytochrome c may not be required for sfApaf-1 activation. Correspondingly, endogenous procaspase-3/9 was not activated by cytochrome c/dATP addition to cell-free preparations ( Supplementary Fig. S6), suggesting that starfish apoptosis is triggered by mechanisms other than cytochrome c release. Zhou et al. (2015) determined the three-dimensional structure of human Apaf-1 in complex with horse cytochrome c 47 . The interactions were established between the WD40 repeat region of Apaf-1 and cytochrome c as a whole, and the specific amino acid residues involved in the interaction in the WD40 repeat region can be identified (Fig. 8a,b). The interface between the WD40 repeat region and cytochrome c mainly consists of hydrogen bonds and van der Waal's contacts. The amino acid residues in the WD40 repeat region involved in the interface show high conservation in other vertebrates, but not in starfish (Fig. 8a). Thus, this low conservation of amino acid residues in the interface in the WD40 repeat region precludes a possibility of similar interactions, if any, between starfish Apaf-1 and cytochrome c. In addition, amino acid identities between the human and rat Apaf-1 interface interacting with cytochrome c were higher than that between the human and rat WD40 repeat The three-dimensional structure of human Apaf-1 (WD40 repeat region) in complex with horse cytochrome c (PDB ID: 3jbt chains A and B). WD40 repeat region is colored from green to red and cytochrome c in black. Amino acid residues in WD40 repeat that interact with cytochrome c are depicted in stick model. The interaction Structure of human Apaf-1 WD40 repeat in complex with horse cytochrome c and the characteristics of the interface. The interaction was calculated based on ∆accessiblity and chose the residues that have difference in solvent accessible area, when the protein interacts with the partner or not. Two loops in gray protruding from WD40 repeat to cytochrome c are deleted in sfApaf-1. (c) Percentage identity in WD40 repeats (blue) and that in cytochrome c binding residues (orange) between Apaf-1 of human and other animals. Note that the values of percentage identity reverse in starfish Apaf-1. (d) The amino acid sequence alignment of cytochrome c from animals. The sequences were obtained from UniPort and the ID is shown at the end of each sequence. The sequence identities are between 73 (starfish and human) and 100 (rat and mouse) %, which are much higher than those in WD40 repeat of Apaf-1. region of Apaf-1 (Fig. 8c, left two columns). Similarly, in mice, frogs, and zebrafish, amino acid identities of the Apaf-1 interface interacting with cytochrome c were higher than those of the WD40 repeat region. However, in starfish, amino acid identities of the residues of sfApaf-1 corresponding to those in the human Apaf-1 interface were lower than that of the WD40 repeat region (Fig. 8c). These results indicate that conservation of the interface residues is high in vertebrates, whereas conservation of the surface residues of sfApaf-1 relating to the interface of Apaf-1 is low in starfish. This supports the hypothesis that sfApaf-1 does not interact with cytochrome c. The sequence identity of cytochrome c among different species including starfish is very high (Fig. 8d), which precludes the possibility of covariation between Apaf-1 and cytochrome c in starfish that could have evolved unique interactions between these proteins in starfish. Thus, the structural bioinformatics analysis has reached a conclusion that is consistent with the experimentally suggested scenario.

Discussion
Stimulation of starfish oocytes by the hormone 1-MA is a prerequisite for fertilization and development. If mature eggs remain unfertilized, the 1-MA-mediated signaling pathway eventually triggers death via the activation of caspase-3-like DEVDase 36,39 . Although the 1-MA receptor has not been identified, it is likely to be a seven-transmembrane domain receptor without the mammalian Fas-like death domain, because 1-MA stimulation activates a heterotrimeric GTP-binding protein, which is sensitive to the pertussis toxin 48,49 . Starfish Gα i subsequently dissociates from Gβγ, which activates PI3-kinase 50,51 , and is followed by cdk1 and ERK activation 52 . Several hours after 1-MA stimulation, apoptosis is induced by spontaneous ERK inactivation 36,38 .
Mammalian Apaf-1, Apaf-1 homologs of C. elegans CED-4, and D. melanogaster Dark interact with their procaspases through CARDs, and activate caspases 12,14,28,31 . The regulation of Apaf-1, however, differs significantly from the regulation of Apaf-1 homologs. In mammals, cytochrome c, which is released from mitochondria, binds to the WD40 repeat region at the carboxyl terminus of Apaf-1 7-10 . This binding is a trigger for forming the apoptosome and activating procaspase-9 13,14 . On the other hand, no cytochrome c is required for the regulation of either CED-4 or Dark 20,21,27,28 . Because DEVDase activity was not increased by the addition of cytochrome c/ dATP to cell-free preparations made from starfish oocytes (Supplementary Fig. S6) and procaspase-3/9 was activated in ultracentrifuged cell-free preparations lacking mitochondria (Figs 6 and 7), sfApaf-1 apparently can be activated without cytochrome c during starfish egg apoptosis. The structural and bioinformatics analyses support this hypothesis (Fig. 8).
Before the CARD-CARD interaction, mammalian procaspase-9 and its D. melanogaster procaspase-9 homolog, pro-Dronc, are inactivated by the binding of inhibitors such as Survivin-HBXIP complexes or TUCAN, and DIAP1 44,45,56 . The release of such inhibitors is required for activating procaspases. When inactively expressed procaspase-3/9-His 6 at 37 °C was incubated in cell-free preparations, DEVDase activity increased and cleavage of procaspase-3/9 occurred (Fig. 7). These results suggest that an inhibitor for procaspase-3/9 may be absorbed by the recombinant protein, causing the dimerization and activation of endogenous procaspase-3/9 as demonstrated for mammalian procaspase 46 .
Caspase proteolytic activity assay. Caspase activity was determined by the cleavage of the peptide substrates Ac-DEVD-MCA, Ac-IETD-MCA, and Ac-LEHD-MCA (Peptide Institute, Inc., Osaka, Japan). The substrates dissolved in dimethyl sulfoxide at 10 mM were added to the samples to a final concentration of 0.1 mM. Fluorescence intensity was measured at 380 nm for excitation and at 460 nm for emission using FluoroMax-4 (HORIBA, Ltd., Kyoto, Japan).

SDS-PAGE and western blotting.
Eggs were collected at various time points after 1-MA treatment (0-11 h). Eggs (n = 60) in 60 µL sample buffer 60 were boiled for 5 min at 95 °C. Samples containing 10 eggs were subjected to 12.5% SDS-PAGE. Proteins were blotted onto PVDF membranes (Immobilon-P, 0.45 µm, Millipore). Each membrane was blocked with PBS-T (0.05% Tween20-PBS) containing 1% BSA (Sigma-Aldrich), and was incubated with the anti-caspase-3/9 antibody, anti-ERK1/2 antibody (CST), and anti-active p38MAPK antibody (Promega) at a dilution of 1:2000 for 1 h at room temperature. After washing three times with PBS-T, membranes were incubated with the second anti-rabbit HRP antibody at a dilution of 1:2000 for 1 h at room temperature. After washing twice with PBS-T for 10 min and once with PBS for 10 min, proteins were detected using ECL Prime Western Blotting Detection System (GE Healthcare) and LAS-4000mini Luminescent image analyzer (Fuji Photo Film Co.). The results were analyzed by Image Gauge software (Fuji Photo FilmCo.).

Microinjection. Microinjections into oocytes and quantitation of injection volumes were performed
according to the methods of Hiramoto 61 . Oocytes were held between two coverslips separated by two pieces of double-sided tape during microinjection and observation 48 . Preparation of the oocyte homogenate and supernatant. Cell-free preparations were made as described previously 43 . De-jellied immature oocytes or mature eggs were washed twice in 10 volumes of ice-cold P11 buffer (150 mM Glycine, 100 mM EGTA, 200 mM HEPES-KOH, pH 7.0). After P11 buffer was removed, oocytes were homogenized by passing through a nylon mesh and centrifuged at 20000 g for 15 min at 4 °C. The supernatant was frozen with liquid nitrogen, and stored at −80 °C until use. For the pull-down assay and gel filtration analysis, cell-free preparation was 3-fold diluted with P11 buffer, and ultracentrifuged at 65000 g for 1 h. The supernatant was frozen with liquid nitrogen, and stored at −80 °C.
Assay for CARD-CARD interaction. Purified GST-A-CARD (2 nmol) and GST (2 nmol) were incubated with 100 µL washed Glutathione Magnetic Agarose Beads (Thermo Fisher) for 30 min at room temperature, and washed three times with 300 µL wash buffer according to the protocol. GST-A-CARD or GST beads were mixed with purified recombinant His-C-CARD (4 nmol) in PBS buffer. They were incubated for 30 min at room temperature, and magnetic beads were washed twice with 300 µL wash buffer to remove unbinding proteins. Bound proteins were eluted with 100 µL elution buffer, and CARD-CARD interactions were detected by SDS-PAGE with CBB staining.
Pull-down assay. Purified GST-A-CARD (2 nmol) and GST (2 nmol) were incubated with 100 µL washed Glutathione Magnetic Agarose Beads (Thermo Fisher) for 1 h at room temperature, and washed three times with 300 µL wash buffer. Each bead was mixed with ultracentrifuged cell-free preparations made from immature oocytes. They were incubated for 30 min at 4 °C, and precipitated magnetic beads were washed twice with 300 µL wash buffer. Sample buffer (100 µL) was added to the washed beads and boiled for 5 min at 95 °C, followed by SDS-PAGE and western blotting with the anti-caspase-3/9 antibody. Immunoprecipitation was performed as described 62 .

Gel filtration analysis.
Purified GST-A-CARD at a 2 mM final concentration was incubated with 800 µL ultracentrifuged cell-free preparation from immature oocytes for 30 min at room temperature. Gel filtration was performed by using a Superose TM 6 10/300 GL column (GE Healthcare) with the gel filtration buffer containing 10 mM HEPES (pH 7.5), 100 mM NaCl, and 2 mM dithiothreitol at 4 °C. The column was calibrated with molecular weight standards.
Purified procaspase-3/9-His 6 at the final concentration of 3 mM was incubated with 800 µL ultracentrifuged cell-free preparation from immature oocytes for 30 min at room temperature. Gel filtration steps are same as the above.
Accession codes. The data present in this work was deposited in NCBI's Gene Expression Omnibus (GEO) database under the accession number ACM46824 (caspase-3/9) and MF612046 (sfApaf-1).