Purification and characterization of an interleukin-1beta-converting enzyme family protease that activates cysteine protease P32 (CPP32).

CPP32, a member of the interleukin-1β-converting enzyme (ICE) family of cysteine proteases, cleaves poly(ADP-ribose) polymerase and sterol regulatory element binding proteins during apoptosis. CPP32 normally exists in the cytosol as a 32-kDa inactive precursor and only becomes activated when cells are undergoing apoptosis. The activation is a proteolytic event that generates a p20/p11 heterodimer. We report here the identification, purification, and characterization of a hamster CPP32-activating protease (CAP) that cleaves and activates CPP32. The biochemical properties of CAP suggest that it is another member of the ICE family of proteases. Purified CAP consists of two prominent polypeptides of 19 and 13 kDa. Protein sequencing revealed that CAP is derived from the hamster homolog of Mch2α, a member of the ICE family recently identified based on the sequence conservation among the ICE family members. CAP activity is inhibited by CrmA, a cowpox virus protein that prevents host cell apoptosis. CAP itself is also activated through proteolytic cleavage. These data are consistent with the idea that the activation of the ICE family of proteases during apoptosis proceeds through a cascade of proteolytic events.

CPP32, 1 an interleukin-1␤-converting enzyme (ICE)-like cysteine protease, has been implicated in the pathway of apoptosis in mammalian cells based on several observations. First, CPP32 is closely related to an apoptosis promoting gene (ced-3 of Caenorhabditis elegans) in terms of both sequence similarity and substrate specificity (1,2). Second, CPP32 activity is markedly elevated in cells undergoing apoptosis induced by a variety of reagents (3)(4)(5). Third, a cowpox virus protein CrmA and a baculovirus protein p35, both of which prevent cells from undergoing apoptosis, inhibit the activity of CPP32 (2,4,6). Finally, a tetrapeptide aldehyde inhibitor that specifically inhibits CPP32 activity also blocks the ability of cytosol from apoptotic cells to induce apoptosis-like changes in normal nuclei in vitro. (3).
When activated, CPP32 specifically cleaves poly(ADP-ribose) polymerase (PARP) (3,4) and sterol regulatory element binding proteins (SREBPs) (5,7). PARP is an enzyme implicated in DNA repair and genome surveillance and integrity. The proteolytic cleavage of PARP during the onset of apoptosis by CPP32 results in the separation of its DNA binding and catalytic domains. This cleavage prevents the catalytic domain of PARP from being recruited to sites of DNA damage and presumably disables the ability of PARP to coordinate subsequent repair and genome maintenance events (3). Furthermore, the Ca 2ϩ /Mg 2ϩ -dependent endonuclease implicated in the internucleosomal DNA cleavage, a hallmark of apoptosis, is negatively regulated by polyADP-ribosylation, and this inhibition may be retrieved when PARP is cleaved (8). SREBPs are a family of transcription factors that stimulate transcription of genes involved in cholesterol and fatty acid metabolism, including the low density lipoprotein receptor, 3-hydroxy-3-methylglutaryl-CoA synthase, and fatty acid synthase genes (9 -12). The amino-terminal halves of SREBPs are bona fide basic helix-loop-helix zipper transcription factors. Unlike any other transcription factors, they are linked to extended carboxylterminal halves by two trans-membrane domains that anchor the proteins to the membranes of the endoplasmic reticulum and nuclear envelope. In cells starved for cholesterol or undergoing apoptosis, a proteolytic cleavage event between the leucine zipper and the membrane attachment region frees the amino-terminal fragment which enters the nucleus and activates its target genes (5,13). The CPP32-mediated cleavage of SREBPs during apoptosis is not regulated by cellular cholesterol content and occurs at a different site compared with that of sterol-regulated proteolysis (5). The physiological function of activated SREBPs in apoptotic cells is still obscure. Nevertheless, since the activated SREBPs should have a profound impact on cellular lipid metabolism, it has been speculated that cleavage of SREBPs by CPP32 during apoptosis is involved in preserving the cytoplasmic membrane integrity of apoptotic cells, and/or preparing the membrane for phagocytosis (5).
The ability of activated CPP32 to trigger apoptosis implies that cells must have a tight mechanism to control this activation to prevent unwanted cell death. CPP32 is activated by multiple proteolytic cleavages at aspartic acid residues. The eventual result is cleavage of the 32-kDa precursor into the 20/11-kDa active form (3,5). The mechanism that triggers CPP32 activation is not known. Partially purified active CPP32 from HeLa cell extracts was able to cleave the CPP32 precursor in vitro (5). This reaction was partially, but not completely, inhibited by a CPP32-specific tetrapeptide inhibitor, suggesting autocatalytic activation as well as the existence of another activating enzyme (5). ICE has also been shown to be able to cleave and activate CPP32 in vitro (4). However, since ICE knockout mice have no general defects in apoptosis (14,15), ICE does not appear to be a general mediator of apoptosis.
Accordingly, since CPP32 is activated by cleavage at aspartic acid residues, a hallmark of ICE-like proteases (16), a cascade of ICE-like proteolytic cleavages leading to apoptosis has been proposed (4,5). Such a protease cascade would provide both the regulation and signal amplification necessary for a highly controlled yet rapid and irreversible process of apoptosis.
In this paper, we report the identification and purification of a CPP32-activating protease (CAP) from hamster liver extract. This enzyme specifically cleaves and activates CPP32. The biochemical properties of this protease suggest that it is an ICE-like cysteine protease distinct from ICE and CPP32, the two enzymes that have previously been implicated in the activation of CPP32 (4,5). The protein sequence of purified CAP revealed that CAP is derived from the hamster homolog of Mch2␣, a member of the ICE family recently cloned by PCR based on sequence conservation of the ICE family (17). Mch2␣ may represent the upstream protease acting on CPP32 and may initiate the ICE-like protease cascade leading to apoptosis. We also find that CAP activity is more sensitive to inhibition by CrmA than is CPP32, defining a new and more efficient target for CrmA blockage of the onset of apoptosis.

General Methods and Materials
We obtained male Golden Syrian hamsters (ϳ150 g) from Sasco (Omaha, NE); [ 35 S]methionine was from Amersham Corp.; N-ethylmaleimide (NEM), iodoacetamide (IAA), phenylmethylsulfonyl fluoride (PMSF), imidazole, and aprotinin were from Sigma; Ac-Tyr-Val-Ala-Asp-aldehyde (Ac-YVAD-CHO) were from Bachem Bioscience, Inc.; Ac-Asp-Glu-Ala-Asp-aldehyde (Ac-DEAD-CHO) was from Julio C. Medina of Tularik, Inc. (7); molecular weight standards for SDS-PAGE and gel-filtration chromatography were from Bio-Rad. cDNA clones of human SREBP-2 and hamster CPP32 were described in the indicated references. HeLa cell cytosol was prepared as described (13). Protein concentration was determined by the Bradford method. Silver staining was carried out using a Silver Stain Plus kit from Bio-Rad. Plasmids were purified using a Megaprep kit from Qiagen.
In Vitro Translation of SREBP-2, CPP32, and CrmA SREBP-2 was translated in a TNT SP6 transcription/translation kit from Promega as described (7). A PCR fragment encoding amino acids 29 -277 of hamster CPP32 (5) was cloned into NdeI and BamHI sites of pET 15b vector (Novagen). The resulting fusion protein of six histidines with hamster CPP32 (amino acids 29 -277) was translated in a TNT T7 transcription/translation kit in the presence of [ 35 S]methionine according to the manufacturer's instruction. The translated protein was passed through a 1-ml nickel affinity column (Qiagen) equilibrated with buffer A (20 mM Hepes-KOH, pH 6.8, 10 mM KCl, 1.5 mM MgCl 2 , 1 mM Na-EDTA, 1 mM Na-EGTA, 1 mM dithiothreitol (DTT), and 0.1 mM PMSF). After washing the column with 10 ml of buffer A, the translated CPP32 was eluted with buffer A containing 250 mM imidazole. CrmA cDNA was cloned into the EcoRI site of pBK-CMV vector (Stratagene) and translated in a TNT T3 transcription/translation kit in the presence of [ 35 S]methionine. The translated CrmA (200 l) was purified by passing the translation mixture through a 10-ml Sephadex G-25 gelfiltration column equilibrated with buffer A. The translated proteins contained within the exclusion volume of the column were collected.
Assay for CPP32-activating Protease 10 l of purified, 35 S-labeled, in vitro translated hamster CPP32 was incubated at 37°C for various times with the indicated enzyme fractions in a final volume of 25 l of buffer A. At the end of the incubation, SDS sample buffer was added to each tube. After boiling for 3 min, the samples were subjected to 15% SDS-PAGE. The gel was subsequently transferred to a nitrocellulose filter and exposed to a Kodak X-Omat AR x-ray film.

Purification of CAP from Hamster Liver
All purification steps were carried out at 4°C. All the chromatography steps except the SP-Sepharose column were carried out using an automatic fast protein liquid chromatography station (Pharmacia Biotech Inc.).
Step 1: Preparation of S-100 Fraction-Livers from 25 hamsters were rinsed twice with cold buffer A and homogenized for 45 s in the same buffer (0.5 g/ml) in a Waring blender followed by three strokes of a motor-driven homogenizer. The homogenates were centrifuged at 10 5 ϫ g for 1 h in a SW 28 rotor (Beckman). The resulting supernatant (S-100) fraction was dialyzed overnight against three changes of 4 liters of buffer A.
Step 2: SP-Sepharose Chromatography-The S-100 fraction from step 1 was applied to a SP-Sepharose column (200 ml) equilibrated with buffer A. The flow-through fraction was supplemented with DTT to a final concentration of 10 mM and incubated at 30°C for 2 h before being loaded onto a fresh SP-Sepharose column (100 ml) equilibrated with buffer A. After washing with 4-column volumes of buffer A, the column was eluted with buffer A containing 400 mM NaCl.
Step 3: Ammonium Sulfate Precipitation-Solid ammonium sulfate was added to the SP-Sepharose column eluate (150 ml) to 40% saturation. After stirring for 1 h, the precipitate was collected by centrifigation at 15,000 rpm in a JA 20 rotor (Beckman) for 20 min. The pellet was resuspended in 10 ml of buffer A.
Step 4: Superdex-200 Gel-filtration Chromatography-The resuspended ammonium sulfate pellet was loaded onto a Superdex-200 gel filtration column (Phamacia, 300 ml) equilibrated with buffer A. The column was eluted with the same buffer, and fractions of 10 ml were collected and assayed for CAP activity.
Step 5: Mono Q Chromatography-The active fractions from Superdex 200 column were pooled and loaded onto a Mono Q 5/5 column after adjusting the pH to 7.6 with 1 M KOH. The column was equilibrated with buffer B (buffer A adjusted to pH 7.6) and eluted with buffer B containing 400 mM NaCl. The flow-through and the bound materials were assayed for CAP activity.
Step 6: Heat Treatment-The flow-through fraction from the Mono Q column was adjusted to pH 6.8 with 1 N HCl followed by incubation at 55°C for 15 min. The denatured proteins were pelleted by centrifigation at 15,000 rpm in a JA 20 rotor for 20 min.
Step 7: Mono S Chromatography-The supernatant after heat treatment was loaded onto a Mono S 5/5 column equilibrated with buffer A. The column was eluted with a 20-ml linear salt gradient of buffer A to buffer A containing 440 mM NaCl. Fractions of 1 ml were collected and assayed for CAP activity.

NH 2 -terminal Sequencing Analysis of CAP
The CAP peak fractions from Mono S column (step 7) were subjected to electrophoresis in a 15% SDS gel and then transferred onto a piece of poly(vinylidene difluoride) membrane (Millipore). The 19-and 13-kDa subunits were visualized by Coomassie Blue staining and excised for direct NH 2 -terminal sequencing on a sequencer (Biosystem).

Western Blot Analysis
A polyclonal antibody against hamster CPP32 was produced as described (5). A monoclonal antibody against human CPP32 was purchased from Transduction Laboratories. Immunoblot analysis of CPP32 was performed with horseradish peroxidase-conjugated anti-rabbit immunoglobulin G using Enhanced Chemiluminescence Western blotting detection reagents (Amersham Corp.) as described previously (13).

Expression and Purification of Recombinant His 6 -tagged CPP32 and CrmA
The plasmid containing fusion protein of six histidine and hamster CPP32 (amino acid 29 -277) was the same as used for the in vitro translation described above. The plasmid was transformed into DE3 competent cells (Novagen). The entire coding sequence of CrmA was PCR-amplified from the plasmid p996 containing cowpox virus crmA gene and cloned into the SalI and HindIII sites of pQE 30 vector (Qiagen). The plasmid was transformed into the M15 competent cells (Qiagen). The bacteria cultures (1 liter for each plasmid) were grown at 37°C until the density reach A 600 reading of 0.6. Isopropyl-1-thio-␤-Dgalactopyranoside was then added to the final concentration of 2 mM. After 1-h induction, the bacteria were pelleted and lysed in buffer A through sonication. After centrifigation, the supernatants were loaded onto two 3-ml nickel-Sepharose (Qiagen) columns equilibrated with buffer A. The columns were washed with 10 ml of buffer A, followed by 10 ml of buffer A containing 500 mM NaCl, and again with 10 ml of buffer A. The fusion proteins were eluted with buffer A containing 250 mM imidazole. The peak protein fractions of the nickel column eluates were further purified through a FPLC Superdex-200 16/30 column equilibrated with buffer A, and 1-ml fractions were collected. The column fractions of the recombinant CPP32 were assayed for its enzymatic activity by incubating with 35 S-labeled SREBP-2 as described previ-ously (7). The purity of recombinant CPP32 enzyme and CrmA were analyzed by SDS-PAGE followed by silver staining.

RESULTS
Identification of CAP Activity-CPP32 in hamster liver extracts can be specifically activated in vitro by incubating at 30°C (7). The activation is due to the cleavage of its 32-kDa precusor into the p20/p11 active form (5). To search for the protease(s) that catalyze this specific cleavage, we translated hamster CPP32 mRNA with a six histidine tag in the presence of [ 35 S]methionine. The translated CPP32 was purified on a nickel affinity column and incubated with the hamster liver S-100 extract to identify enzyme activity that cleaves the in vitro translated, 35 S-labeled CPP32 precursor into the p20/p11 active form.
Incubation of 35 S-labeled CPP32 with hamster liver S-100 did not result in any significant cleavage (Fig. 1, lane 1). Little activity was detected in the flow-through fraction, and the high salt eluate after the S-100 fraction was passed through an SP-Sepharose column (lanes 2 and 3). However, after preincubating the S-100 fraction or the flow-through fraction from the SP-Sepharose column at 30°C for 2 h, significant cleavage of the CPP32 precursor into p20 and p11 subunits was observed (lanes 4 and 5). The material that bound to the SP column showed little activity even after preactivation (lane 6). After preincubation, all of the activated CAP bound to the column (lane 8) and recovered in the high salt eluate (lane 9). The above experiment demonstrates that CAP, like CPP32, is present in the liver extract in an inactive form that can be activated by preincubation. CAP precursor and the biochemical components required for its activation are not retained on the SP-Sepharose column while the active CAP is. By applying the starting material to the SP-Sepharose column before activation and repeating the column step after activation, we achieved a significant purification of CAP (Table I).
CAP Is a New ICE-like Protease That Cleaves CPP32 Precursor-The activated CAP eluted from the SP-Sepharose column was precipitated by 40% ammonium sulfate (Fig. 2, panel A, S and P). The resuspended ammonium sulfate pellet was loaded onto a Superdex 200 gel-filtration column, and the eluted fractions were assayed for CAP activity (fraction [6][7][8][9][10][11][12][13][14][15][16][17][18][19][20]. The CAP activity eluted from the gel-filtration column at fraction 16. In contrast, activated CPP32 eluted at fraction 10 on the same column detected by Western blot analysis (panel B) and the cleavage of 35 S-labeled SREBP (data not shown). The CAP therefore appeared to run smaller than CPP32 on the gelfiltration column and behaved anomalously small compared with the molecular weight standard.
Since CPP32 is activated by cleavage after aspartic acid residues, CAP might be either an ICE-like protease (16) or a serine protease related to granzyme (18). However, it is un-likely that CAP is a serine protease since its activity was insensitive to serine protease inhibitors such as PMSF, leupeptin, and aprotinin (data not shown). If CAP is another member of the ICE family, it should be sensitive to cysteine-alkylating reagents such as NEM or IAA (3,7). As shown in Fig. 3, CAP activity was completely abolished by 1 mM NEM (lane 11) or 100 M IAA (lane 5), and the inhibiting activity was prevented by an excess of DTT (lanes 7 and 12). ICE and CPP32 can be inhibited specifically by nM concentration of tetrapeptide aldehydes corresponding to the amino acid sequences preceding their cleavage sites (3,7,19). The CAP activity was resistant to both ICE and CPP32 inhibitors up to 1 M (lanes 13 and 14), indicating that CAP is a new ICE-like protease distinct from CPP32 and ICE. Equal amounts of the CPP32-specific inhibitor completely blocked the cleavage of SREBP by CPP32 (Fig. 5,  lane 8).
Purification of CAP from Hamster Liver Extracts-To complete the purification of CAP, the gel-filtration column fractions containing CAP activity (fractions 16 -17) were pooled and loaded onto a Mono Q column. CAP activity was found in the flow-through while the CPP32 activity was bound (data not shown). The Mono Q column flow-through fraction was then heated at 55°C for 15 min. CAP activity was resistant to this heat treatment (Fig. 4, lanes 1 and 2). After pelleting the heat-denatured proteins, the supernatant was loaded onto a Mono S column and eluted with a 40 -440 mM linear NaCl FIG. 1. In vitro activation of CAP. Hamster liver extracts (S-100) were prepared and passed through a SP-Sepharose column as described under "Experimental Procedures." 2 l of the S-100 and the SP column flow-through and bound fractions were assayed before (lanes 1-3) and after (lanes 4 -6) preincubation at 30°C for 2 h. The activated SP-Sepharose flow-through fraction was passed through the SP-Sepharose column again, and the flow-through and bound material were collected and assayed as before (lanes 7-9). The assays were conducted by incubating indicated fractions with aliquots (10 l) of in vitro translated, 35 S-labeled CPP32 in 25 l of buffer A. After incubation at 37°C for 15 min, the samples were subjected to 15% SDS-PAGE. The gel was transferred to a nitrocellulose filter and exposed to film for 12 h at Ϫ80°C.

TABLE I Purification of CAP from hamster liver cytosol
Hamster liver cytosol was prepared from 25 hamster livers. CAP activity of fractions was assayed by cleavage of 35 S-labeled CPP32 at four concentrations of protein, and the cleaved products were quantified by a Fuji 1000 phosphorimager machine.
Step gradient. As shown in Fig. 4 (panel A), CAP activity was eluted from the Mono S column around 185 mM NaCl. The same fractions were also subjected to SDS-PAGE followed by silver staining (panel B). Two predominant polypeptides with apparent molecular masses of 19 and 13 kDa eluted precisely with the CAP activity. Table I shows quantitative estimates of a typical purification procedure, starting with liver cytosol from 25 hamsters (5.1 g of protein). As shown in Fig. 1, the S-100 fraction, the SP-Sepharose flow-through, and bound material did not have a significant amount of CAP activity before preincubation at 30°C for 2 h. The SP-Sepharose column flow-through fraction was subsequently activated in vitro and passed through the SP-Sepharose column again. About 129 mg of protein was bound and eluted from the SP-Sepharose column with a 49% recovery of CAP activity. The CAP activity from the SP-Sepharose column was precipitated by the addition of 40% ammonium sulfate. After gel filtration, Mono Q chromatography, heat treatment, and finally Mono S chromatography, a 12,000-fold purification was achieved with 6% recovery of the starting activity.
Characterization of CAP-To obtain the amino acid sequence of CAP, the two polypeptides (p19/p13) in the Mono S column peak fractions were isolated from the SDS-PAGE, and the NH 2 -terminal sequences of both proteins were obtained by Edman degradation. As shown in Table II, the NH 2 -terminal sequences from both p19 and p13 share a high degree of identity with the recently reported sequence of human Mch2␣, a member of the ICE family whose cDNA was isolated by a polymerase chain reaction strategy based on the sequence conservation among ICE family members (17). Four out of 25 amino acids of the NH 2 -terminal of p19 polypeptide are different from the human Mch2␣, with the serine at the third posi- FIG. 2. Purification of CAP by ammonium sulfate precipitation followed by gel-filtration chromatography. CAP activity eluted from the SP-Sepharose column was precipitated with 40% ammonium sulfate and dialyzed against buffer A. Panel A, an aliquot (2 l) of supernatant (S) and resuspended pellet (P) were assayed for CAP activity. The rest of the resuspended pellet was loaded onto a gelfiltration column, and 2-l aliquots of the fractions eluted from the column were assayed for CAP activity (fractions 6 -20). Panel B, the CPP32 in the same column fractions was detected by Western blotting using a polyclonal antibody against hamster CPP32.

FIG. 3. Inhibition of CAP by N-ethylmaleimide and iodoacetamide.
Aliquots (5 l) of partially purified hamster liver CAP (through gel-filtration step) were incubated with NEM (lanes 3-7) or IAA (lanes 8-12) at the indicated concentration in a 20-l reaction. In (DTT) lanes 7 and 12 a final concentration of 10 mM was added to the reaction before the addition of NEM or IAA. In lanes 13 and 14, the tetrapeptide aldehydes Ac-DEAD-CHO or Ac-YVAD-CHO were dissolved in buffer A containing 3% (v/v) dimethyl sulfoxide and added to the CAP fraction in a final concentration of 1 M. After incubation at 30°C for 15 min, aliquots (10 l) of in vitro synthesized, 35 S-labeled CPP32 in buffer A were added to each reaction. After an additional 15-min incubation at 37°C, the samples were subjected to SDS-PAGE, and the gel was transferred to a nitrocellulose filter and exposed to film for 12 h at Ϫ80°C.

FIG. 4. Heat treatment and Mono S column separation of CAP.
The flow-through fraction from the Mono Q column was heated at 55°C for 15 min and loaded onto a Mono S column and eluted as described under "Experimental Procedures." Panel A, aliquots (1 l) of the Mono S fractions were incubated with 10 l of in vitro synthesized, 35 S-labeled CPP32 at 37°C for 15 min in a 25-l reaction volume. The samples were then subjected to SDS-PAGE and transferred to a nitrocellulose filter and exposed to film for 12 h at Ϫ80°C. Panel B, 30-l aliquots of the same fractions were subjected to 15% SDS-PAGE followed by silver staining. subunits, comparison with human Mch2␣ Sequences were obtained from the NH 2 -terminal of purified p19 and p13 polypeptides by Edman degradation. The proteins were isolated on 15% SDS-PAGE followed by electrotransfer onto a polyvinylidene difluoride membrane. The sequence of human Mch2␣ was reported by Fernandes-Alnemri et al. (17). Numbers in parentheses denote the amino acid position in the cDNA sequence of human Mch2␣. The (Ϫ) denotes an amino acid residue gap in human Mch2␣. Asterisks (*) denote residues that differed in two sequences. Purified CAP Cleaves and Activates CPP32- Fig. 5 demonstrates that purified CAP cleaves and activates CPP32 that are present in crude HeLa cytosol. The cleavage of CPP32 was monitored by Western blot analysis (lanes 1-3), and the activity of CPP32 was assayed by the cleavage of 35 S-labeled SREBP-2 (lanes 4-9). HeLa cytosol alone incubated at 37°C for 30 min was not able to activate its CPP32 (lanes 2 and 6). CAP itself had very little SREBP cleavage activity (lane 5). However, incubation of HeLa cytosol with purified CAP resulted in the cleavage of CPP32 (lane 3) and activation of CPP32 (lane 7). The activated CPP32 was sensitive to its specific tetrapeptide aldehyde inhibitor Ac-DEAD-CHO (7) (lane 8) but not to ICE inhibitor (lane 9).
CAP Is Inhibitable by CrmA-Cowpox virus protein CrmA is a potent and specific inhibitor of ICE-like proteases (20) and can protect cells against apoptosis in a variety of systems, including growth factor withdrawal (21), detachment of mammary epithelial cells from the underlying extracellular matrix (22), treatment of cytotoxic T-lymphocytes, and activation of either Fas or the tumor necrosis factor receptor (23,24). CrmA inhibits ICE-like proteases by serving as a substrate for these enzymes whose cleaved products remain bound to the enzyme after cleavage reaction (2, 6). Fig. 6 shows that the purified CAP was capable of cleaving in vitro translated, 35 S-labeled CrmA into one major band of 25 kDa and a minor band of 35 kDa (lanes 2-4). Under the same condition, the same amount of purified recombinant CPP32 had little CrmA cleavage activity, even though it was able to cleave SREBP (see Fig. 7). This result is consistent with the observation that CrmA is a poor inhibitor of CPP32 (3). Fig. 7 shows an experiment in which we tested directly for inhibition of CAP activity by CrmA. Purified CrmA was incubated with purified CAP or CPP32 followed by the addition of 35 S-labeled CPP32 or SREBP-2, respectively. About 50% of CAP was inhibited by a 2-fold molar excess of CrmA, while it took five times more CrmA to reach the same level of inhibition of CPP32.

DISCUSSION
Activation of CPP32 during Apoptosis-Activation of the ICE-like protease CPP32 is associated with several forms of programmed cell death (3)(4)(5). The activation is due to proteolysis that cleaves the CPP32 precursor after aspartic acid residues to generate the p20/p11 active form (3,5). To search for the protease that carries out this cleavage, we identified and purified the CPP32-activating protease from hamster liver extracts. The purified CAP activity contained two prominent protein bands that migrated at 19 and 13 kDa on SDS-PAGE. This is a typical p20/p10 pattern that is shared by other ICElike proteases (3,7,19). Protein sequencing data revealed that the two polypeptides are originated from the hamster homolog of human Mch2␣, an enzyme whose cDNA was cloned based on the sequence conservation among ICE family of proteases (17). The NH 2 -terminal sequences of CAP p19 and p13 show no significant homology with any other known ICE-family of proteases including Mch3/CMH-1/ICE-LAP 3 (25)(26)(27).
ICE exists in the cytosol of monocytes as an inactive 45-kDa precursor that is cleaved at several aspartic acid positions to generate an NH 2 -terminal 20-kDa fragment and a COOH-  3 and 7-9) in 20 l of buffer A at 30°C for 30 min. In panel A, the indicated samples were directly subjected to 15% SDS-PAGE followed by Western blot analysis using a monoclonal antibody against human CPP32. In panel B, the resulting samples were incubated with aliquots (5 l) of in vitro synthesized, 35 S-labeled SREBP-2 for an additional 15 min at 30°C. In lanes 8 and 9, 1 M final concentration of the indicated tetrapeptide aldehydes dissolved in buffer A containing 3% (v/v) dimethyl sulfoxide was added to the reaction mixture before the addition of 35 S-labeled SREBP-2. The samples were subsequently subjected to 8% SDS-PAGE, and the gel was dried and exposed to a film for 12 h at Ϫ80°C. After an additional 15-min incubation at 37°C, the samples were subjected to 15% (panel A) or 8% (panel B) SDS-PAGE. The 15% gel was then transferred to a nitrocellulose filter and the 8% gel was dried. The filter and gel were exposed to film for 12 h at room temperature. The radioactivity in the filter and dried gel was also quantified in a Fuji 1000 phosphorimager, and the results were plotted in panel C.
terminal 10-kDa fragment (19). The p20 and p10 fragments remain associated as a heterodimer, and two heterodimers associate to form the tetrameric active enzyme (28,29). CAP resembles ICE in several ways: (i) it is sensitive to inhibition by cysteine alkylating reagents; (ii) it cleaves substrate after aspartic acid residues (3); and (iii) it is synthesized as a precursor that becomes activated through cleavage after aspartic acid residues. One interesting note is that CAP behaved as a much smaller protein on the gel-filtration column, even though the size of the two subunits are similar to that of ICE and CPP32. The anomalous behavior of CAP on the gel-filtration column may have been due to an interaction between CAP and the column resin. However, the molecular basis for this behavior remains to be determined.
A cDNA for Mch2␣ was originally cloned from human Jurkat T lymphocytes using a degenerative PCR strategy based on the sequence conservation among known members of the ICE/ CED3 family of proteases (17). Mch2␣ cDNA encodes a 34-kDa protein that shares 38 and 35% identity with human CPP32 and C. elegans CED-3, respectively. An alternatively spliced version, Mch2␤, which is missing residues 15-103 of Mch2␣, was also cloned (17). The NH 2 -terminal sequence of the CAP p19 subunit corresponds to amino acids 26 -49 of human Mch2␣ and is therefore derived from the hamster homolog of Mch2␣. Interestingly, only Mch2␣ showed protease activity when expressed in bacteria. This protein was also able to induce apoptosis in sf9 cells when overexpressed using a baculovirus vector (17).
Autocatalytic Versus CAP-mediated Activation of CPP32-Partially purified CPP32 enzyme from HeLa cell extracts was able to cleave CPP32 precursor (5), indicating that CPP32 can be activated through autocatalysis. A similar mechanism is probably responsible for yielding active enzyme when CPP32 precursor is expressed in large quantity in bacteria (Refs. 2, 7, and "Experimental Procedures"). However, our experimental data suggest that Mch2␣ is the major protease in hamster liver extracts responsible for cleaving the CPP32 precursor into the p20/p11 active form. As shown in Fig. 2, even though the active CPP32 was present in column fraction 10, there was little CAP activity in this fraction compared with fraction 16, the main peak fraction for CAP. The reason for this observation, whether CPP32 is a preferred substrate for CAP or there is simply more CAP in the hamster liver extracts than CPP32, remains to be determined.
Inhibition of CAP by CrmA-CrmA, a poxvirus-encoded serpin, prevents host cells from undergoing apoptosis by a variety of stimuli that activate CPP32 (4,5). CrmA was able to directly inhibit CPP32 activity, a property that has been used to explain its prevention of apoptosis (4). However, on a quantitative basis, CrmA is a rather poor inhibitor for CPP32 (2,3). Our finding that CrmA is a better inhibitor for CAP may reconcile this controversy. By inhibiting CAP, CrmA could prevent the activation of CPP32 and therefore block the onset of apoptosis.
Activating the Activator-Our data also point out that Mch2␣ itself needs proteolytic processing to become active. The processing includes the removal of a short peptide from the NH 2terminal of the protein and cleavage at aspartic acid 193 (reference to human sequence). The NH 2 -terminal cleavage could be a result of cleavage after TETD followed by the sequential removal of two more amino acid residues (Table II). The NH 2terminal of p13 kDa subunit is preceded by TEVD. Both of these sequences resemble the recognition site IETD for Mch2␣ and DEVD for CPP32 (3). It is likely that the Mch2␣ precursor can be cleaved and activated by autocatalysis and/or by CPP32. In this scenario, the slight activation of Mch2␣ will trigger a cycle of amplification leading to the irreversible activation of CPP32, which in turn carries out the execution of apoptosis by cleaving death substrates PARP and SREBPs.
This hypothesis clearly needs to be confirmed in vivo. To do that, one may need to develop a CAP-specific inhibitor, like tetrapeptide aldehyde inhibitor Ac-YVAD-CHO for ICE (19). Alternatively, this issue could be studied in cells that have lost CAP by targeted deletion of mch2␣ gene. The prediction would be that cells lacking Mch2␣ should be impaired in forms of apoptosis that requires CPP32 activation.
The biochemical components that trigger the initial activation of Mch2␣ are still to be determined.