Two Arabidopsis Metacaspases AtMCP1b and AtMCP2b Are Arginine/Lysine-specific Cysteine Proteases and Activate Apoptosis-like Cell Death in Yeast*

Metacaspases in plants, fungi, and protozoa constitute new members of a conserved superfamily of caspase-related proteases. A yeast caspase-1 protein (Yca1p), which is the single metacaspase in Saccharomyces cerevisiae, was shown to mediate apoptosis triggered by oxidative stress or aging in yeast. To examine whether plant metacaspase genes are functionally related to YCA1, we carried out analyses of AtMCP1b and AtMCP2b, representing the two subtypes of the Arabidopsis metacaspase family, utilizing yeast strains with wild-type and the disrupted YCA1 gene (yca1Δ). Inducible expression of AtMCP1b and AtMCP2b significantly promoted yeast apoptosis-like cell death of both the wild-type and yca1Δ strains, relative to the vector controls, during oxidative stress and early aging process. Mutational analysis of the two AtMCPs revealed that their cell-death-inducing activities depend on their catalytic center cysteine residues as well as caspase-like processing. In addition, the phenotype induced by the expression of two AtMCPs was effectively prevented when the cells were pretreated with a broad-spectrum caspase inhibitor N-benzyloxycarbonyl-Val-Ala-Asp-fluoromethyl-ketone. These results suggest that the two subtypes of Arabidopsis metacaspases are functionally related to Yca1p with caspase-like characteristics. However, we found that bacterial and yeast extracts containing AtMCP1b, AtMCP2b, or Yca1p exhibit arginine/lysine-specific endopeptidase activities but cannot cleave caspase-specific substrates. Together, the results strongly implicate that expression of metacaspases could result in the activation of downstream protease(s) with caspase-like activities that are required to mediate cell death activation via oxidative stress in yeast. Metacaspases from higher plants may serve similar functions.

initial execution phase for this developmental PCD (13). However, direct biochemical evidence for the activity of the mcII-Pa encoded protein is lacking.
The budding yeast Saccharomyces cerevisiae has emerged as a useful system for PCD or apoptosis studies, because it is readily amenable to genetic and molecular analysis (14,15). Indeed, apoptosis-like cell death of yeast occurs in certain cdc48 mutant (16) or upon treatment with low doses of hydrogen peroxide (17). In addition, ectopic overexpression of proapoptotic factors such as Bax induces yeast apoptosis; the effect can be counteracted by the simultaneous overexpression of anti-apoptotic proteins such as Bcl-X L (18), suggesting a striking parallelism with apoptosis in metazoans. The single metacaspase gene in S. cerevisiae, termed yeast caspase-1 (YCA1), was shown to mediate oxidative stress-induced and age-related apoptosis in yeast (19). Overexpression of YCA1 stimulates apoptotic cell death, whereas disruption of this gene prevents it, indicating that this metacaspase is required for apoptosislike cell death in yeast. Interestingly, the wild-type Yca1p but not one with a point mutation at the catalytic center Cys 297 residue was shown to be proteolytically activated upon oxidative stress, which correlated with the appearance of new caspase-like activities that have similar substrate specificity with initiator caspases (caspase-6 and caspase-8). However, conclusive evidence demonstrating that Yca1p indeed exhibits caspase-like activity using purified or recombinant enzymes is still lacking. On the other hand, in Trypanosoma brucei metacaspase gene family is consisted of five different genes (TbMCA1-5) (20). Of these, heterologous expression of Tb-MCA4 in the budding yeast led to growth inhibition, mitochondrial dysfunction, and cell death, suggesting that this metacaspase may have a role in controlling cellular proliferation coupled to mitochondrial function. Further, mutation studies of TbMCA4 suggested that this metacaspase could function as a cysteine protease, whereas enzymatic properties of TbMCA4, such as its substrate specificity, remain unknown (20).
In this work, we addressed the question of whether members of the two subtypes of Arabidopsis metacaspase gene family are functionally related to YCA1, which is a well characterized metacaspase gene (19,21). Using the wild-type and yca1⌬ yeast strains, we performed a detailed analysis of the activities of AtMCP1b, AtMCP2b, and Yca1p during oxidative stress-stimulated and age-related apoptosis in yeast. Furthermore, we performed an initial biochemical characterization of these metacaspases produced in yeast and bacteria, to compare their enzymatic properties, including substrate preference.
The pYES2.1/V5-His metacaspase vectors and the corresponding vector control without any foreign gene were introduced into both the wild-type (KFY715) and yca1⌬ (KFY729) strains with an S.C. Easy-Comp transformation kit (Invitrogen) and selected on agar plates containing synthetic complete medium lacking uracil, consisting of 0.67% yeast nitrogen base with ammonium sulfate (Invitrogen) and 30 mg/ liter of all amino acid (except 80 mg/liter histidine and 400 mg/liter leucine), 30 mg/liter adenine with 2% glucose as carbon source (SC glucose media).
Growth Conditions and Induction of Expression-Yeast cells were grown at 28°C and liquid cultures were agitated at 260 rpm. For each experiment, strains were streaked out freshly on an SC glucose plate from a stock stored at Ϫ80°C and incubated for 2 days at 28°C. Yeast culture harboring expression vectors were grown overnight in 3 ml of noninducing selective media (SC glucose), diluted in 10 ml of fresh SC glucose media (adjusted an OD of 0.05), and then cultured until reaching an OD of 0.4 -0.6. The cells collected by brief centrifugation were resuspended in inducing selective medium (SC containing the same contents but 2% galactose instead of glucose). For stimulation of yeast apoptosis, H 2 O 2 was added to a final concentration of 0.4 mM (for all strains in wild-type background) or 1.2 mM (for all strains in yca1⌬ background). For inhibition of caspase-like activities in yeast cells, zVAD(OMe)-fmk (Calbiochem) from 20 mM stock in 1:1 Me 2 SO/ethanol was added to a final concentration of 20 M. All strains were cultured for up to 96 h after the galactose shift and harvested at specified intervals.
Cell Viability Assays-After culturing in SC galactose medium for various times (0 -96 h), an aliquot of cells was removed for trypan blue dye exclusion assay, counting ϳ700 total (live and dead) cells as described previously (22). Dead cells have lost plasma membrane integrity and as a result cannot exclude dyes and thus are stained. The ability of cells to exclude trypan blue indicates that they are still alive (22). Alternatively, survival of the cell was determined by counting unstained cells, and 10 3 cells were spread on YEPD plates as described (19). All experiments were repeated at least in triplicate.
Test for Apoptotic Marker and Microscopy-TdT-mediated dUTP nick end labeling (TUNEL) test were performed as described previously (17,21). To determine the frequency of morphological phenotypes (TUNEL), at least 300 cells of three independent experiments were evaluated. For DAPI (4Ј,6-diaminido-2-phenylidole) staining, stationary phase cells were harvested, resuspended in phosphate saline buffer (20 mM sodium phosphate, 140 mM NaCl, pH 7.4), stained with DAPI at the concentration of 1 g/ml for 5 min at room temperature, and rinsed 5 times with distilled water. A DeltaVision restoration microscope system (Applied Precision) equipped with a TE200 microscope (Nikon) was used to observe DAPI-stained nuclei in stationary phase cells. The images of nuclei were analyzed by softWoRx software (Applied Precision) on an Octane Work station (Silicon Graphics). For determination of reactive oxygen species (ROS) accumulation in the cell, an aliquot of cells were stained with dihydrorhodamine (DHR) (Sigma-Aldrich) at the concentration of 5 g/ml for 2 h at room temperature or dihydroethidium (DHE) at the concentration 5 g/ml for 10 min (17,21). DHR and DHE fluorescence were observed under an inverted fluorescent microscope (model TE300, Nikon Inc.), equipped with an EYFP filter set (excitation 510/20 nm, emission 560/40 nm, beamsplitter 530 nm; Chroma Technology Corp.) and a DsRed filter set (excitation 546/11 nm, emission 605/75 nm, beamsplitter 560 nm; Chroma Technology Corp.), respectively, and the images were taken with a charge-coupled device camera (Optronics Inc.). All images were finally processed by Adobe Photoshop 5.5 (Adobe Systems) on a PowerMac G4 computer (Apple Inc.) for the final display.
Enzyme Assay-To assay for metacaspase activity in crude cell extracts of galactose-induced yeast cultures (15-50 ml), the cells were suspended in lysis buffer (50 mM potassium phosphate, pH 7.5, 500 mM NaCl, 1 mM EDTA, 10 mM dithiothreitol, 1% (w/v) CHAPS, and 5 g/ml aprotinin). Following cell lysis by vortexing with glass beads, the cell lysates were centrifuged (20,000 ϫ g, 10 min, 4°C) and the resulting supernatant was subjected to measurement of in vitro metacaspase activity (see below).
To assay enzyme activity of recombinant AtMCP1b, AtMCP2b, and Yca1p produced in Escherichia coli BL21(DE3) strain (Invitrogen), the coding region of each gene, amplified by PCR, was cloned in-frame in the bacterial expression vector pET23a (Novagen), resulting in double fusion proteins (N-terminal fusion with a T7 tag and C-terminal fusion with His 6 ). The recombinant metacaspases were induced by adding 0.4 mM isopropyl-1-thio-␤-D-galactopyranoside to cell culture with an A 600 of 0.4 -0.6. Then, the cell culture was shaken (260 rpm) at 28°C for 1 h (for AtMCP2b) or 3 h (for AtMCP1b and Yca1p). Cells were harvested, washed with distilled water, and resuspended in extraction buffer (50 mM potassium phosphate, pH 7.4, 500 mM NaCl, 0.1 mM dithiothreitol, 1% (w/v) CHAPS, and 10% (w/v) glycerol, 5 g/ml aprotinin). Cells were broken by brief sonication, and insoluble materials were removed by centrifugation (20,000 ϫ g, 20 min, 4°C). The resulting supernatants were used as active lysates for direct measurement of metacaspase activity in vitro. Alternatively, recombinant AtMCP2b proteins were purified. The soluble extracts from the recombinant AtMCP2b-expressing strain, supplemented with 20 mM imidazole, was mixed with Ni 2ϩ chelating Sepharose FF (Amersham Biosciences) equilibrated with 20 mM imidazole in extraction buffer and incubated for 3 h at 4°C. After washing with extraction buffer supplemented with 50 mM imidazole, bound metacaspases were eluted with 250 mM imidazole in extraction buffer. The purified samples were further passed through an NAP-10 column (Amersham Biosciences) equilibrated with sample buffer (50 mM HEPES-KOH, pH 7.5, 150 mM NaCl, 10 mM dithiothreitol, 1 mM EDTA, 0.1% (w/v) CHAPS, and 10% (w/v) glycerol).
Metacaspase activity was measured by fluorometric detection of hydrolysis of peptidyl substrates (Z-Arg-2NA, Z-Arg-Arg-2NA, Z-Ala-Arg-Arg-2NA, Z-Gly-Gly-Arg-2NA, Z-Val-Lys-Lys-Arg-2NA, and Boc-Val-Leu-Lys-MCA, Sigma-Aldrich, or Boc-Gly-Arg-Arg-MCA, Bachem Bioscience Inc.) in crude cell extracts or purified recombinant proteins. The enzyme reaction was performed in a reaction buffer (100 l) containing 25 mM HEPES-KOH (pH 7.5), 150 mM NaCl, 10 mM CaCl 2 , 10 mM dithiothreitol, 10% (w/v) glycerol, 0.1% (w/v) CHAPS, 5-10 g of total proteins (crude cell lysate) or 0.1 g of purified recombinant protein, and 100 M peptidyl substrate. After incubation of 30 -60 min at 30°C, the reaction was stopped by the addition of 100 l of stop solution (10% acetic acid in 150 mM sodium acetate), and an aliquot of sample (50 l) was diluted in 2 ml of distilled water. Then, the amount of AMC or 2NA liberated during reaction was measured fluorometrically (excitation of 380 nm and emission of 440 nm for AMC, and excitation of 335 nm and emission of 410 nm for 2NA) with a fluorescence spectrophotometer (F-4500, Hitachi). Inhibitory effects of protease inhibitors were assayed under the same reaction conditions, using concentrations as indicated.

AtMCP1b and AtMCP2b Can Be Efficiently Expressed in
Yeast-To facilitate a comparative analysis of two subtypes of Arabidopsis metacaspases (AtMCP1b and AtMCP2b) and Yca1p, a plasmid encoding each gene C-terminally tagged with a V5 epitope and hexahistidine sequence under the GAL1 promoter control was created. The inducible expression of each metacaspases was verified by immunoblotting using anti-V5 epitope antibody. When all metacaspases were individually expressed in the yca1⌬ mutant, a dominant band of about 47 kDa (AtMCP1b), 53 kDa (AtMCP2b), and 54 kDa (Yca1p), which likely correspond to the predicted molecular mass of each metacaspase tagged with a V5 epitope and hexahistidine (44.9, 46.7, and 52.4 kDa, respectively) were detected at 6 h after the galactose shift (Fig. 1A). In contrast, no band was detected in the control strain. In the strains expressing AtMCP2b or Yca1p, additional protein bands were also detected at this time point, suggesting the possibility of their partial proteolysis by endogenous yeast protease activities or by itself (autocatalytic processing). The levels of AtMCP1b and AtMCP2b were comparable to the level of Yca1p when expressed in yca1⌬ mutant. Similar results were obtained with the wild-type yeast strain (data not shown). From these, we concluded that AtMCP1b and AtMCP2b can be efficiently expressed in yeast.
Caspase-like Processing of Two AtMCPs-Most animal caspases are initially produced in cells as catalytically inactive zymogens and must undergo proteolytic activation during apoptosis. The inactive zymogens consist of an N-terminal prodomain, a large catalytic subunit of ϳ20 kDa, and a small Cterminal catalytic subunit (ϳ10 kDa) that become activated by scaffold-mediate transactivation or by cleavage via upstream proteases in an intracellular cascade (3,5,6). Similarly, it was shown that such a caspase-like processing would be essential for activation of Yca1p that is initiated by the proteolytic removal of the small subunit, which probably corresponds to the band of ϳ12-kDa and that depends on its catalytic center Cys 297 (19). We therefore wanted to address the question of whether the caspase-like processing of AtMCP1b and AtMCP2b can be observed in a similar manner to that of Yca1p. To this end, we first monitored by immunoblotting with anti-V5 epitope antibody the expression patterns of AtMCP1b and AtMCP2b as well as Yca1p in yca1⌬ mutant during prolonged FIG. 1. Immunoblot analysis of AtMCP1b, AtMCP2b and Yca1p tagged with V5-His 6 at their C terminus. A, extracts from yeasts strains, with plasmid encoding each metacaspase as indicated and were grown in SD-Ura media containing 2% galactose for 0 and 6 h, were immunoblotted and probed with anti-V5 antibody. B, caspase-like processing of AtMCP1b, AtMCP2b, and Yca1p during prolonged culture in yeast. Metacaspase expression in exponentially grown yca1⌬ cells, transformed with the indicated constructs, was induced in SC-Ura media containing 2% galactose (t ϭ 0). At the time indicated, aliquots were subjected to immunoblot analysis using anti-V5 antibody. culture until 72 h after the galactose shift. Interestingly, a strong band of C-terminal fragment of AtMCP1b and AtMCP2b (ϳ12 kDa) could also be detected over time of expression, although the band patterns were largely different in each strain (Fig. 1B). To demonstrate that their potential catalytic residues are essential for the formation of the C-terminal fragment, the conserved Cys 220 of AtMCP1b and Cys 139 of AtMCP2b were changed to alanine. In fact, mutations in these residues completely blocked the formation of their C-terminal fragments (see Fig. 3B). A caspase-like processing of AtMCP1b expressed in yca1⌬ mutant with additional stimulation by H 2 O 2 clearly depended on its catalytic center Cys 220 (Figs. 1 and 3B). In contrast, it seems that caspase-like processing of AtMCP2b depended on its catalytic center Cys 139 but did not require an additional apoptotic stimulus, thus implicating an autocatalytic mechanism (Fig. 3B). Interestingly, formation of the ϳ25-kDa C-terminal fragment of AtMCP2b was not detectable after mutation of Cys 139 , suggesting the possibility that a larger C-terminal fragment (ϳ25 kDa) might be an intermediate form to generate a smaller C-terminal fragment (ϳ12 kDa) that corresponds to p10 subunit of mammalian caspases and Yca1p.
Expression of AtMCP1b and AtMCP2b in Synergy with Prolonged Culture or Oxidative Stress Leads to Cell Death in Yeast-To assess the effect of AtMCP1b and AtMCP2b on the viability of wild-type yeast cells, cells growing in liquid culture were stained with trypan blue, and cell death was quantified by microscopy. Similarly, effect of YCA1 overexpression on cell viability was also examined in parallel. In the wild-type background, no significant loss of cell viability was observed in all strains tested at 24 h after the galactose shift ( Fig. 2A). However, after 48 h the viability of AtMCP1b or AtMCP2b-expressing cells gradually started decreasing and finally reached to Ͻ20% at 72 h in a manner resembling the effect of overexpressing YCA1 ( Fig. 2A). In contrast, the viability of control cells decreased more slowly (61% at 72 h). Similar results were also obtained with yca1⌬ cells (Fig. 2B). In yca1⌬ background, expression of AtMCP1b or AtMCP2b could accelerate cell death progression in a similar manner to that of YCA1. However, it should be noted that the timing of cell death activation and progression of cell death were somewhat delayed compared with the wild-type background, probably due to the absence of endogenous Yca1p as discussed previously (19). The phenotype induced by AtMCP1b, AtMCP2b, and YCA1 were partially prevented by pan-caspase specific inhibitor zVAD-fmk (Fig. 2C) or by site-specific mutations with the AtMCP1b C220A , AtMCP2b C139A , or yca1 C297A variants ( Fig. 2A, and additional data not shown), suggesting that cysteine-dependent protease activity of AtMCP1b and AtMCP2b could contribute to activation of cell death as in the case of Yca1p. Therefore, it is possible that inducible expression of AtMCP1b or AtMCP2b can partially restore a yca1⌬ mutation by functionally phenocopying YCA1.
We next addressed the question of whether AtMCP1b or AtMCP2b might mimic the function of YCA1 in H 2 O 2 -mediated apoptosis. As previously reported, low external doses of reactive oxygen species stimulate induction of yeast apoptosis (17). However, yca1⌬ mutant shows better survival after treatment with 1.2 mM H 2 O 2 in comparison to the wild-type, indicating that disruption of YCA1 protected the cells from H 2 O 2 -activated apoptosis (19) (Fig. 3A). Conversely, sensitivity of yca1⌬ mutant to H 2 O 2 can be restored by overexpression of YCA1 in the mutant (19) (Fig. 3A). Treatment with 1.2 mM H 2 O 2 for 26 h significantly affected the viability of yca1⌬ cells expressing AtMCP1b (58%) or AtMCP2b (75%), but their degree of cell viability reduction was apparently lower than those of YCA1overexpressing strain (25%). In contrast, treatment with 0.4 mM H 2 O 2 did not cause any activation of cell death in yca1⌬ mutant harboring AtMCP1b, AtMCP2b, YCA1 and the corresponding vector control after 26 h (data not shown). Furthermore, mutational analysis and inhibitor study using zVAD-fmk suggested that the phenotypes induced by AtMCP1b, AtMCP2b, and YCA1 depend on protease activities (Fig. 3). Similar experiments were also run with the wild-type strain. In the wild-type background, overexpression of YCA1 in synergy with treatment with mild oxygen stress (0.4 mM H 2 O 2 ) can effectively induce cell death, whereas the vector control shows perfect viability against this treatment (ϳ90%) (Fig. 3C). Interestingly, the degree of cell death in wild-type background induced by expression of AtMCP1b or AtMCP2b in synergy with H 2 O 2 was very similar to the levels observed in the yca1⌬ mutant background (data not shown). Therefore, these results strongly suggest that AtMCP1b or AtMCP2b can partially restore the sensitivity of yca1⌬ mutant to H 2 O 2 by functionally replacing YCA1 rather than by providing an additional stimulus from a cryptic pathway.
Expression of AtMCP1b and AtMCP2b Induce an Apoptosislike Cell Death in Yeast-Because overexpression of YCA1 resulted in the activation of apoptosis in yeast (19), we next wanted to investigate whether the loss of viability observed in stationary yeast cells expressing AtMCP1b or AtMCP2b was also related to apoptotic events. However, to address the question of whether AtMCP1b and AtMCP2b can induce apoptosis by itself, we expressed the two AtMCPs and YCA1 in yca1⌬ mutant.
One of the key events in triggering yeast apoptosis is the accumulation of ROS (17,(23)(24)(25). To detect ROS production, we treated stationary yca1⌬ cells expressing AtMCP1b, AtMCP2b, YCA1, or the vector control with DHR. In the presence of ROS, DHR is oxidized to fluorescent chromophore rhodamine 123, which depends on the redox status of mitochondrial membrane potential and can then be visualized under a fluorescence microscope. We monitored the formation of ROS in a time course during the early aging process until 96 h. Exponentially growing yca1⌬ cells expressing AtMCP1b, AtMCP2b, YCA1, and the corresponding vector control at 24 h showed no DHR-positive fluorescence upon staining (data not shown). At 48 h, DHRpositive cells but with low fluorescence became visible in AtMCP1b-, AtMCP2b-, and YCA1-expressing cells, and strong intracellular DHR fluorescence was observed by 72 and 96 h (Fig. 4A). In contrast, control yca1⌬ cells showed weak fluorescence signal from 72 to 96 h, but the level of ROS accumulation was remarkably lower than that in AtMCP1b-, AtMCP2b-, or YCA1-expressing cells (Fig. 4A). In addition, expression of sitespecific mutated variants (AtMCP1b C220A , AtMCP2b C139A , or yca1 C297A ) did not result in significantly increased ROS level (not shown). As shown in Fig. 4B, similar results were also obtained with another ROS indicator, DHE, which is oxidized specifically by superoxide ion (O 2 . ), and the oxidized form (ethidium) shows red fluorescence in cell cytoplasm (19). From 48 h the ratio of ROS-accumulating cells increased in AtMCP1b-, AtMCP2b-, and YCA1-expressing strains; after 72 h Ͼ50% cells showed strong high ROS level while the control strains did not. It appears that the ratio of DHE-positive cells correlated with increased cell death in all strains (Fig. 2B). In addition, we found that the production of ROS induced by AtMCP1b, AtMCP2b, and YCA1 was partially prevented by pretreatment with zVAD-fmk (Fig. 4B). These results thus suggest that ectopic expression of each AtMCP and overexpression of YCA1 resulted in increased ROS accumulation in yca1⌬ cells, which is dependent on downstream caspase-like activities, and promotes yeast apoptosis-like cell death. A positive feed-back loop thus may exist whereby activation of meta-caspases by ROS could in turn trigger additional ROS production via zVAD-sensitive proteases.
Other hallmarks of apoptosis in yeast are cleavage of DNA in the nucleus and nuclear fragmentation (16,(23)(24)(25). DNA nicking can be visualized by TUNEL staining, which reveals free 3Ј-OH ends originated by DNA strand breaks, whereas nuclear fragmentation can be detected by DAPI staining (see "Experimental Procedures"). Fig. 4C shows the percentages of TUNEL staining and trypan-blue staining cells over 96 h in strains expressing each metacaspase or vector control after galactoseshift without H 2 O 2 treatment. Initially, ϳ10% of the cells expressing AtMCP1b, AtMCP2b, YCA1, or the vector control stained positive for TUNEL at 24 h. However, after 48 h populations of TUNEL-positive cells were remarkably increased and in good agreement with the populations of dead cells in each metacaspase-expressing strains but not in the vector control strain (Fig. 4C, upper panel). DAPI staining revealed that the majority of cells expressing the vector control showed a normal and single round-shaped nucleus, whereas an abundance of abnormally shaped and fragmented nuclei (50ϳ70%) was observed in the cells expressing AtMCP1b, AtMCP2b, or YCA1 at 96 h (Fig. 4C, lower panel). In addition, expression of site-specific mutated variants (AtMCP1b C220A , AtMCP2b C139A , or yca1 C297A ) did not result in significantly increased TUNELpositive cells or abnormally shaped and fragmented nuclei in yca1⌬ cells (not shown). Taken together, the results strongly suggest that ectopic expression of AtMCP1b or AtMCP2b in yeast can induce an apoptotic-like cell death correlated with an increased generation of ROS in a similar manner to that of YCA1.
Detection of Arginine/Lysine-specific Endopeptidase Activities in Extracts from AtMCP1b-, AtMCP2b-, or Yca1p-expressing Bacteria or Yeast Cells-Recently, it was reported that two Arabidopsis type II metacaspases, Atmc4 (AtMCP2d) and Atmc9 (AtMCP2f), exhibit arginine/lysine-specific endopeptidase activities that do not cleave caspase-specific peptide sub- strates (Ac-YVAD-MCA, Ac-DEVD-MCA, Ac-VEID-MCA, and Ac-IETD-MCA). Also, these metacaspases do not show any sensitivity to caspase-specific inhibitors, including zVAD-fmk (26). Independently of this report, we also found that purified recombinant type II metacaspases (AtMCP2b and AtMCP2d) can cleave arginine/lysine-specific substrates such as Boc-GRR-MCA in a Ca 2ϩ -dependent manner (at sub-millimolar range), whereas they were unable to cleave caspase-specific synthetic substrates. Also, we found that the endopeptidase activity of purified recombinant AtMCP2b is strongly inhibited by two arginal protease inhibitors (leupeptin and antipain), oligopeptide inhibitor (benzyloxycarbonyl-Phe-Lys-trimethylbenzoyloxymethylketone), and tosyl-lysyl-chloromethylketone but not aprotinin and phenylmethylsulfonyl fluoride (serine protease inhibitors), pepstatin (aspartic protease inhibitor), oligopeptide inhibitor (Z-FA-fmk), and zVAD-fmk. 2 To examine whether AtMCP1b and Yca1p also have arginine/lysine-specific endopeptidase activities, we first attempted to produce active recombinant proteins as double fusions with T7 epitope-(at the N terminus) and His 6 -tagged (at the C terminus) in E. coli. Immunoblot analysis using anti-T7 epitope antibodies revealed that molecular masses of the recombinant AtMCP1b, AtMCP2b, and Yca1p separated by SDS-PAGE are in good agreement with the results obtained with expression of these metacaspases in yeast (Fig. 5A). Bacterial extracts containing recombinant AtMCP1b or Yca1p exhibited significantly increased endopeptidase activities hydrolyzing arginine-and/or lysine-containing peptidyl substrates, which are largely different in degree (1.5-to 4-fold), as compared with control extracts (Fig. 5B). However, their activities were apparently much lower than AtMCP2b (Fig. 5C). Interestingly, strict Ca 2ϩ dependence of AtMCP1b and Yca1p in Boc-GRR- MCA-hydrolyzing activity was not observed, but these activities were slightly stimulated (2-to 3-fold) under the presence of 10 -100 mM Ca 2ϩ (not shown). In addition, bacterial extracts containing recombinant AtMCP1b and Yca1p did not display any increased caspase-like activity against some of caspasespecific substrates tested, including Ac-YVAD-MCA, Ac-DEVD-MCA, Ac-VEID-MCA, and Ac-IETD-MCA (not shown).
Our results using the bacterial expression system suggested the possibility that enzymatic activities of AtMCP1b and Yca1p are potentially lower than that of type II metacaspases, and/or recombinant AtMCP1b and Yca1p were largely inactive. Therefore, we next assessed the question of whether arginine/lysinespecific endopeptidase activities of all metacaspases examined in this study might be increased in yca1⌬ cells with or without low doses of H 2 O 2 . We found that AtMCP1b or YCA1 overexpression for 24 h without additional apoptotic stimuli did not result in significant increase of arginine/lysine-specific endopeptidase activities in yeast extracts (data not shown). However, significantly increased endopeptidase activities (3-to 7-fold) toward paired basic amino acids-containing substrates (RR, ARR, GRR, and VKKR) occurred after incubation with low doses of H 2 O 2 (Fig. 6A). In contrast, AtMCP2b expression for 24 h without additional apoptotic stimuli resulted in dramatically increased endopeptidase activities toward all substrates examined in this assay (9-to 280-fold) (Fig. 6B). In addition, we found that expression of site-specific mutated variants (AtMCP1b C220A , AtMCP2b C139A , or yca1 C297A ) did not result in any increased arginine/lysine-specific endopeptidase activities (not shown). Taken together, the results shown by using bacterial and yeast expression systems strongly suggest that AtMCP1b, AtMCP2b, and Yca1p have intrinsic arginine/lysinespecific endopeptidase activities.

DISCUSSION
Molecular phylogenic analysis suggests that metacaspases in plants, fungi, and protozoa may be at the evolutionary root of caspases (10). Accumulating evidence suggest that a yeast metacaspase Yca1p serves to regulate yeast apoptosis triggered by some apoptotic stimuli, including oxidative stress (19), low pH (19), and high salinity stress (27). It has also become increasingly clear that there is a connection between YCA1 and age-related apoptosis in yeast cells (19,21). Although all members of the Arabidopsis metacaspase family contain a caspaselike domain, it remains to be determined whether they are functionally related to YCA1 as a caspase-like enzyme, or have other biological functions. Here we provided evidence that both AtMCP1b and AtMCP2b, representing the two subtypes of Arabidopsis metacaspases, can partially substitute for Yca1p function in mediating oxidative stress-induced and age-related apoptosis in yeast. In yeast apoptosis, cell-death-inducing activity of AtMCP1b and AtMCP2b depends on both cysteine protease activity and caspase-like processing in a similar manner to Yca1p. To our knowledge, this is the first study experimentally implicating that the two subtypes of AtMCP and YCA1 belong to a conserved protease family that has a similar molecular function between plants and yeast. This suggests an evolutionarily-conserved relationship between PCD and the metacaspase family. However, using recombinant enzymes produced in bacteria and yeast, we also found that the two AtMCPs and Yca1p have arginine/lysine-specific endopeptidase activities but failed to cleave caspase-specific synthetic substrates under our in vitro assay conditions. Indeed, various synthetic substrates containing an arginine/lysine residue at the P1 position could be cleaved by all metacaspases examined in this study. These data, in turn, suggest that AtMCPs and Yca1p are not bona fide homologues of animal caspases.
AtMCP1b and YCA1 belong to the same type-I subtype of metacaspase gene family, whereas AtMCP2b is a type II subtype that is not found in yeast. Nevertheless, we found that AtMCP2b can also partially replace the function of YCA1 in mediating oxidative stress-induced and age-related apoptosis in yeast. However, the potency of cell-death-inducing activity of AtMCP2b seems to be lower than that of AtMCP1b and Yca1p (Fig. 3A). This difference between type I and type II metacaspases could be explained by two possibilities: one is that the prodomain of AtMCP1b and Yca1p might contain an important regulatory function to initiate apoptosis effectively in yeast, and the other is that AtMCP2b may have lower activity or different substrate specificity relative to Yca1p and AtMCP1b. Our initial characterization of enzymatic activity of the three metacaspases suggest that AtMCP1b and Yca1p have a similar preference for paired basic amino acid residues at P1 and P2 positions in RR, ARR, GRR, and VKKR substrates and apparently display similar levels of endopeptidase activity (Figs. 5 and 6). In contrast, AtMCP2b has a strong preference for a GRR substrate relative to the others, but apparent endopeptidase activity of this metacaspase against various synthetic substrates was much stronger than that of both AtMCP1b and Yca1p under our enzyme assay condition. Therefore, it seems likely that the strength of endopep- tidase activities and substrate preferences are not the critical determinants for their biochemical functions required for the activation of yeast apoptosis.
Proteases such as caspases are synthesized as zymogens that await activation at a suitable time to initiate a cascade of subsequent proteolytic events that act to orchestrate a critical cellular event like PCD (3,6,7,9,28). The data presented in this study indicate that activation of AtMCP1b and AtMCP2b are initiated by the proteolytic removal of the small C-terminal fragment, which probably corresponds to the p10-like subunit of Yca1p (Figs. 1B and 3B). In the case of AtMCP2b, this processing did not require an additional apoptotic stimuli in yeast and bacteria, thus an autocatalytic mechanism exists (Figs. 1B and 3B). It seems likely that autocatalytic maturation of all type II Arabidopsis metacaspases (Atmc4 -9) could occur in bacterial cells (26). Among these metacaspases, it was shown that site-specific processing at Arg-183 of Atmc9 is essential for the conversion of the proenzyme into mature form, resulting in generation of the p10-like subunits of Atmc9 consisting of amino acids 184 -325 (ϳ15.4 kDa). It is noteworthy that the deduced amino acid sequence of the p10-like region shows significant similarity to other type II AtMCPs that possess a conserved putative processing site Arg or Lys. On the other hand, caspase-like processing of AtMCP1b for 24 h requires apoptotic stimuli induced by exposure to sub-lethal dose of H 2 O 2 , which is very similar with the case of Yca1p (Fig. 3B). However, this processing could occur spontaneously during prolonged culture (Fig. 1B), suggesting the existence of a certain late-limiting step for their activation in yeast. AtMCP1b contains one zinc-finger domain and one proline-rich domain at its N terminus, whereas Yca1p contains only one proline-rich domain. So far, zinc-finger proteins and proline-rich proteins have been implicated in cytoplasmic protein-protein interaction and/or binding to nuclear DNA (29 -31). For example, the human zinc-finger protein (ZNF198) contains a proline-rich region that constitutes a self-association domain and confers the oligomerization of ZNF198 with fibroblast growth factor receptor 1, thereby leading to tyrosine kinase activation of this receptor (31). We speculate that the prodomain of AtMCP1b and Yca1p might mediate homomeric interactions between AtMCP1b or Yca1p, leading to their activations via local increases in their concentrations as a result of oligomerization or aggregation (19). This hypothesis is consistent with the fact that bacterial overexpression could result in AtMCP1b and Yca1p autocatalytic processing with concomitant activation (Fig. 5A). However, the process of how these type-I metacaspases are converted into the mature forms remains to be elucidated.
Enzymatic assays of AtMCP1b, AtMCP2b, and Yca1p in this study have focused only on the repertoires of some synthetic substrates and assay conditions. We found that the amount of Ca 2ϩ required for the activation of AtMCP2b and stimulation of both AtMCP1b and Yca1p is remarkably higher than the estimated physiological concentration of Ca 2ϩ in the cytoplasm of plant cells (32). Similarly, all plant calcium-dependent proteases that have been characterized so far require sub-millimolar levels of Ca 2ϩ for optimal enzyme activation in vitro (33,34). However, all type II metacaspases may not require Ca 2ϩ as a cofactor or activator, because it appears that at least activation of Atmc9 (AtMCP2f) does not depend on the addition of Ca 2ϩ in the assay mixture (26). Interestingly, Atmc9 has different enzymatic characteristics, including substrate specificity, sensitivity to protease inhibitors, and pH dependence in comparison with Atmc4 (AtMCP2d). Among these differences, it is noteworthy that Atmc9 has an acidic pH optimum and is inactive at pH 7.0 -8.0, which is closer to the physiological cytoplasmic pH. Conversely, Atmc4, which is closely related to AtMCP2b in its deduced amino acid sequence, has a neutral pH optimum and is inactive under acidic pH (26). Therefore, metacaspases may have distinct enzymatic properties, including substrate specificities and requirement of specific reaction conditions. Further comparative studies will be required to scrutinize in more detail the enzyme properties of metacaspases including AtMCP1b and Yca1p by using purified enzymes.
In yeast apoptosis mediated by YCA1, activation of cell death was correlated with appearance of new caspase-like activities that have similar substrate specificity to initiator caspases (with VEID-and IETD-hydrolyzing activities) and could be abolished by a pan-caspase inhibitor zVAD-fmk (19). Several lines of evidence in the present study showed that AtMCP1b, AtMCP2b, and Yca1p do not exhibit caspase-like activity but arginine/lysine-specific endopeptidase activity in vitro (Figs. 5 and 6). Also, their activities could not be inhibited by zVADfmk in vitro (not shown). Nevertheless, activation of apoptosislike cell death in AtMCP1b-, AtMCP2b-, or YCA1-expressing yca1⌬ cells was effectively inhibited by pretreatment with this inhibitor (Figs. 2C and 3C), indicative of the contribution of caspase-like protease(s) in yeast apoptosis downstream of the metacaspases. In addition, it was recently reported that activation of caspase-8-like activity (VEIDase) correlated with massive cell death during embryonic pattern formation in plants. Silencing of a type II metacaspase, mcII-Pa, resulted in a significant reduction of both VEID-hydrolyzing activities and cell death (13). Together with our present findings, the above results suggest that these metacaspases may not be directly responsible for earlier reported caspase-like activities in plants and yeast. Namely, activation of unidentified caspase-like protease(s) triggered through either direct or indirect proteolysis by a metacaspase could mediate execution of yeast apoptosis or plant PCD.
Very little is known about the identities of caspase-like protease(s) in yeast and plants. Recently two investigations identified new types of subtilisin-like proteases (named saspase-A and -B) from oats (35) and a vacuolar processing enzyme from tobacco (36) that may play important roles as caspase-like proteases in the execution of PCD in plants. Interestingly, the two oat saspases possess strong initiator caspase-like activities (caspase-6, -8, and -9) with VKMD (Val-Lys-Met-Asp)-, IETD (Ile-Glu-Thr-Asp)-, LEHD (Leu-Glu-His-Asp)-hydrolyzing activity but not DEVD (Asp-Glu-Val-Asp)-and VEID-hydrolyzing activity, whereas these enzymes are very sensitive to numerous caspase-specific inhibitors (35). Because saspases do not appear to be cysteine proteases, they are not caspase-like proteases in a strict sense. In contrast, the cysteine protease vacuolar processing enzyme possesses unique characteristics with a caspase-1-like activity (YVAD (Tyr-Val-Ala-Asp)-hydrolyzing activity) and the ability to cleave the peptide bond at the carbonyl side of the asparagine residue responsible for maturation of various seed proteins in protein-storage vacuoles (36). In addition, these enzymes are sensitive to a pan-caspase inhibitor, zVAD-chloromethyl ketone or zVAD-fmk. Therefore, it would be interesting to see if yeast orthologues of plant saspases (the full-length sequence of which remains to be determined) or vacuolar processing enzyme are responsible for the caspase-like activities observed during yeast apoptosis.
In conclusion, our present work revealed that AtMCP1b and AtMCP2b representing two subtypes of Arabidopsis metacaspase genes have a similar molecular function to that of YCA1, thus implicating a conserved link between PCD and the metacaspase family. Furthermore, results from using bacteria and yeast expression systems in this work show that metacaspase representatives examined so far are arginine/lysinespecific cysteine proteases with no obvious caspase-like activity. Currently, the integrated function of plant metacaspases and the function of each individual member remain largely unknown. Also, there is no evidence of whether the two subfamilies of plant metacaspases act in series or in parallel pathways in plant cells. Using overexpression and/or silenced (knock-out) lines of various plant metacaspases should help in determining their possible roles in the regulation of PCD in plants.