The Ced-3/interleukin 1beta converting enzyme-like homolog Mch6 and the lamin-cleaving enzyme Mch2alpha are substrates for the apoptotic mediator CPP32.

Recent evidence suggests that CPP32 is an essential component of an aspartate-specific cysteine protease (ASCP) cascade responsible for apoptosis execution in mammalian cells. Activation of CPP32 could lead to activation of other downstream ASCPs, resulting in late morphological changes such as lamin cleavage and DNA fragmentation, observed in cells undergoing apoptosis. Here we describe the identification and cloning of a novel human ASCP named Mch6 from Jurkat T lymphocytes. We demonstrate that the pro-enzymes of Mch6 and the lamin-cleaving enzyme Mch2alpha are substrates for mature CPP32. Site-directed mutagenesis revealed that CPP32 processes pro-Mch6 preferentially at Asp330 to generate two subunits of molecular masses 37 kDa (p37) and 10 kDa (p10). However, CPP32 processes pro-Mch2alpha at three aspartate processing sites (Asp23, Asp179, and Asp193) to produce the large (p18) and small (p11) subunits of the mature Mch2alpha enzyme. The CPP32-processed Mch2alpha is capable of cleaving the VEIDN lamin cleavage site, indicating that CPP32 can, in fact, activate pro-Mch2alpha. Granzyme B at a concentration that allows processing and activation of CPP32 failed to process pro-Mch2alpha. However, incubation of pro-Mch2alpha with granzyme B in the presence of a cellular extract containing pro-CPP32 resulted in activation of pro-CPP32 and subsequent processing of pro-Mch2alpha. Interestingly, granzyme B can also process pro-Mch6 but at a site N-terminal to that cleaved by CPP32. These data suggest that Mch2alpha and Mch6 are downstream proteases activated in CPP32- and granzyme B-mediated apoptosis. This is the first demonstration of a protease cascade involving granzyme B, CPP32, Mch2alpha, and Mch6 and evidence that the lamin-cleaving enzyme Mch2 is a target of mature CPP32.

of the new family of aspartate-specific cysteine proteases (ASCPs) 1 (1)(2)(3)(4). The prototype of this family is the interleukin 1␤ converting enzyme (ICE), which exists as a 45-kDa proenzyme and is cleaved to form two subunits (p20 and p10), which associate as a tetramer (p20) 2 /(p10) 2 to form the active enzyme (5,6). ICE bears no resemblance structurally to other known cysteine proteases and is unique in its substrate specificity in that it cleaves after certain aspartates. In light of the homology to Ced-3, an expanded role for ASCPs as positive regulators of apoptosis was hypothesized and subsequently supported when it was shown that overexpression of ICE in fibroblasts resulted in their death by apoptosis (7). This process could be blocked by crmA (a specific viral inhibitor of ICE) or the Bcl-2 proto-oncogene product. Several new members of this family have been described recently, including CPP32 (8), Nedd2/Ich-1 (9,10), Mch2␣ and Mch2␤ (11), Mch3␣ and Mch3␤ (12), Mch4 (13), Mch5 (13), TX (ICH-2, ICErel-II) (14 -16), ICErel-III (16), and four alternatively spliced isoforms of ICE (17). Like ICE, all of these enzymes require processing to large and small subunits for activity, cleave after specific aspartates, and possess the unique and highly conserved pentapeptide active site containing an essential cysteine. Like Ced-3, overexpression of these ICE homologs is capable of inducing apoptosis in various in vivo models.
Granzyme B is a cytoplasmic granule-associated serine protease expressed by cytotoxic T and natural killer cells and is the only other enzyme with similar substrate specificity to ASCPs (18). Release of granzyme B by these cells has been shown to be essential for rapid degradation of DNA and induction of apoptosis in susceptible target cells (19). It now appears that this activity may be mediated, at least in part, by multiple ASCPs that are activated by granzyme B (13).
Although ASCPs appear to play fundamental roles in inflammation and apoptosis, their individual contribution to the apoptosis mechanism is not yet clear. Additionally, their mechanism of activation, enzymatic specificity, and physiologically relevant apoptotic substrates remain to be established. Recently, we provided evidence for the involvement of multiple ASCPs in apoptosis (13,20). We also demonstrated that lamin A is cleaved by Mch2␣, but not CPP32, at the same site that is cleaved in apoptotic S/M extracts (20). In one study, we demonstrated that the newly identified FADD-like ASCP (Mch4) can process and activate CPP32 and Mch3 (13). pro-Mch4 and pro-Mch5 are the only known ASCPs that contain FADD-like death effector domains in their long N-terminal pro-domain, suggesting possible involvement in Fas and other apoptotic pathways (13). It appears that proteases, such as Mch4 and Mch5 with long pro-domains, might be upstream and could interact with the proximal signal transduction machinery of apoptosis. Others with short pro-domains, such as CPP32, Mch3, and Mch2␣, are downstream or distal, operating at or near the cell death effector level.
CPP32 appears to be an important central intermediary in the cell death pathway. This is evident from its ability to be activated by granzyme B and by the upstream FADD-like proteases Mch4 and Mch5 (13) to cleave the well characterized cell death substrate poly(ADP-ribose) polymerase (PARP) (11,12) and induce the morphological changes of apoptosis in isolated nuclei (21). Identification of ASCPs that are targets for activation by mature CPP32 should help us to understand the biochemical events that lead to the apoptotic morphology. The most likely targets for CPP32 are those ASCPs that contain the processing site DXXD between their large and small subunits. These include Nedd2/Ich-1, Mch2␣, and Mch6 (this study). Recently, we demonstrated that Nedd2 is a substrate for mature CPP32 and granzyme B (22). In this study, we report the identification and cloning of a new member of the family of mammalian ASCPs that we have designated Mch6 (mammalian ced-3 homolog 6). We provide evidence that pro-Mch6 is a natural substrate for granzyme B and mature CPP32. Furthermore, we show that mature CPP32 can directly process and activate pro-Mch2␣ by cleaving at three Asp cleavage sites. These data suggest that activation of CPP32 in apoptosis could lead to activation of pro-Mch2␣ and pro-Mch6.

MATERIALS AND METHODS
Cloning of Human Mch6 -To clone Mch6, a 10-l aliquot of human Jurkat Uni-ZAP XR cDNA library (8) containing ϳ10 8 plaque-forming units was denatured at 99°C for 5 min and used as a template for PCR amplification with PCR primer Mch6-pr1 (CTCAACGTACCAG-GAGCC) derived from the GenBank-Established Sequence Tag sequence T97582 and the T3 vector-specific primer (Stratagene). A 10-l aliquot of the primary amplification product was then used as a template for a secondary PCR amplification with primer Mch6-pr2 (CCT-GGGAAAGTAGAGTAGG) derived from the same Established Sequence Tag sequence and the SK-Zap vector-specific primer (11) located downstream of the T3 primer. The secondary amplification products were cloned into a SmaI cut pBluescript II KS ϩ vector. The partial cDNA was then excised from the vector, radiolabeled, and used to screen the original Jurkat cDNA library. Positive clones were purified, rescued into the pBluescript II SK Ϫ plasmid vector, and sequenced.
Northern Blot Analysis-Tissue distribution analysis of Mch6 mRNA was performed on Northern blots (Clontech) using Mch6 cDNA as a probe, as described previously (12,13).
Expression of CPP32 and Mch2␣ in Bacteria and Assay of Enzyme Activity-Recombinant human CPP32 and Mch2␣ with a C-terminal His 6 tag were expressed in Escherichia coli and assayed as described recently (11)(12)(13). The recombinant proteases were purified on a Ni 2ϩaffinity resin.
Purification of Granzyme B-Granzyme B was purified by immunopurification from human natural killer cell lysates using granzyme B-specific monoclonal antibody and assayed as described previously (23,24).
Mutagenesis, in Vitro Transcription/Translation, and Cleavage Assays-Potential aspartate processing sites in pro-CPP32, pro-Mch2␣, or pro-Mch6 were mutated by site-directed mutagenesis using overlapping PCR mutagenic oligonucleotides. The resulting PCR products were subcloned in pBluescript II KS ϩ vector (pro-Mch2␣) or pET-21b vector Western Blot Analysis of CPP32-Processing of CPP32 in 697 cellular extracts treated with granzyme B was determined by Western blot analysis, using a rabbit polyclonal anti-human CPP32 antibody. This antibody was raised against a recombinant p20 subunit (amino acids 1-175) of human CPP32.

RESULTS AND DISCUSSION
Cloning of Human Mch6 -A search of the GenBank Established Sequence Tags for sequences related to CPP32 and Mch2␣ identified a short established sequence tag with accession number T97582. Based on this sequence information, an ϳ2-kilobase cDNA was cloned from a human Jurkat T-lymphocyte cDNA library using a similar methodology as described recently (11,12). This cDNA contains an open reading frame of 1248 base pairs that encodes a 416-amino acid protein, named Mch6, with a predicted molecular mass of ϳ46.2 kDa (Fig. 1A).
Mch6 Belongs to the Aspartate-specific Cysteine Protease Ced-3 Subfamily-The Ced-3-like ASCP subfamily includes Ced-3, CPP32, Mch2␣, Mch3, Mch4, and Mch5 (13). Sequence alignment of all known ASCPs revealed that Mch6 belongs to this subfamily (Fig. 1B). Within the subfamily, Mch6 shows highest homology to CPP32 (ϳ37% identity and 57% similarity). Mch6 is structurally similar to other ASCPs. A mature enzyme could be derived from the precursor pro-enzyme by cleavage at highly conserved Asp residues (Asp 315 and/or Asp 330 ) located between the two subunits ( Fig. 1B). Other potential aspartate cleavage sites are present in the pro-domain region of pro-Mch6 that could also be cleaved to remove the pro-domain (Fig. 1A). The major difference between this enzyme and other family members, however, is that the fourth residue in its active site pentapeptide (QACGG) is a Gly instead of Arg or Gln (Fig. 1B).
The pro-domain of pro-Mch6 has high homology to the prodomains of CED-3 and Nedd2/Ich-1. There is evidence suggesting that the pro-domain of some ASCPs could play a role in the activation of the pro-enzyme (25). Interestingly, we discovered recently that the pro-domains of pro-Mch4 and pro-Mch5 contain FADD-like death effector domains (13). These domains could interact with FADD or other cellular FADD-like proteins, thus connecting the proximal apoptotic signal transduction pathways (i.e., Fas pathway) to the downstream ASCP activation cascade.
The crystal structure of ICE revealed that His 237 , Gly 238 , and Cys 285 are involved in catalysis, whereas Arg 179 , Gln 283 , Arg 341 , and Ser 347 are involved in binding the carboxylate side chain of the substrate P1 aspartate (5,6). These residues are identical in all family members including Mch6, except in Mch5, where there is a Ser to Thr conservative substitution for the residue corresponding to Ser 347 of ICE (Fig. 1B). Another Ser to Thr conservative substitution can also be seen in Mch4 in the region corresponding to Ser 236 of ICE, which is one of the residues that participates in binding the substrate P2-P4 residues. Other residues that might participate in binding the substrate P2-P4 residues are not widely conserved, suggesting that they may determine substrate specificity.
Tissue Distribution of Mch6 -The pro-Mch6 riboprobe detected three major mRNA species (ϳ1.0, ϳ2.4, and ϳ4.4 kilobases) in most human tissues (Fig. 2). The highest expression is seen in the heart, with moderate expression in liver, skeletal muscle, and pancreas. Lowest expression was observed in the other tissues. The presence of multiple mRNA species was observed previously with ICE mRNA (26) and is suggestive of alternative splicing or polyadenylation.
pro-Mch6 Is a Substrate of Mature CPP32-pro-Mch6 contains two potential processing sequences between its large and small subunits (Fig. 1). The 312 PEPDA 316 site is a potential granzyme B cleavage site because it contains an acidic residue at the P3 position. The 327 DQLDA 331 site is a potential CPP32 cleavage site, because it is similar to the DEVDG site in PARP and DNA-PKcs (27) and it contains an Asp residue at the P4 position. To test whether pro-Mch6 is a substrate for CPP32, pro-Mch6 was translated in vitro in the presence of [ 35 S]methionine and then incubated with purified recombinant human CPP32. Time course analysis (Fig. 3A) revealed that CPP32 cleaves pro-Mch6 at one site to generate two cleavage products of molecular masses ϳ37 kDa (p37) and ϳ10 kDa (p10). These products can be detected within the first 15 min (lanes 2 and 3).
Longer incubation results in a decrease in the intensity of the full-length 46-kDa band and an increase in the intensity of the p37 and p10 products (lanes 5 and 6). No significant processing of the p37 product was observed upon prolonged incubation, suggesting that CPP32 cleaves preferentially at one site within the pro-Mch6 polypeptide. The sizes of the cleavage products are consistent with cleavage at the DQLDA site, which contains Asp 330 . This was confirmed by site-directed mutagenesis of Asp 315 and Asp 330 . As expected, CPP32 was unable to process the Asp 330 mutant to the p37 and p10 products (Fig. 3C). On the other hand, it was able to process the Asp 315 mutant to generate the p37 and p10 products (Fig. 3E). These products are indistinguishable from those obtained with the wild-type pro-Mch6 (Fig. 3A). This establishes that Asp 330 is the CPP32 processing site.
Granzyme B Cleaves pro-Mch6 Preferentially at the PEPDA Site-One way by which granzyme B can induce apoptosis in target cells is by activation of ASCPs. Recently, we demonstrated that granzyme B is able to process multiple members of the ASCP family (13). To test whether granzyme B can process pro-Mch6, 35 S-labeled pro-Mch6 was incubated with granzyme B and then analyzed by SDS-PAGE and autoradiography. Fig.  3B shows that granzyme B cleaves pro-Mch6 preferentially at one site to generate two cleavage products: a large ϳ35-kDa (p35) product and a small ϳ12-kDa (p12) product. The large  granzyme B-cleaved product migrates faster than the large CPP32-cleaved product, suggesting that the two enzymes cleave at two different sites. The granzyme B cleavage site is N-terminal to the CPP32 cleavage site, and it is most likely to be Asp 315 within the PEPDA site. This was confirmed using the Asp 315 and Asp 330 mutant pro-Mch6 (Fig. 3, D and F). Unlike CPP32, granzyme B was able to cleave the Asp 330 mutant to generate the p35 and p12 products (Fig. 3D). Mutation of Asp 315 prevented granzyme B from cleaving pro-Mch6 at the PEPDA site, as evidenced from the absence of the p35 and p12 products (Fig. 3F ). Interestingly, granzyme B could still cleave this mutant at the DQLDA site, but less efficiently than CPP32, to generate faint p37 and p10 products. This establishes that granzyme B processes pro-Mch6 at Asp 315 and Asp 330 with preference for Asp 315 over Asp 330 . A double mutation of Asp 315 and Asp 330 completely blocked granzyme B and CPP32 processing of pro-Mch6 (data not shown).
Activation of CPP32 by Granzyme B in a Cell Lysate Leads to Processing of pro-Mch6 at Asp 315 and Asp 330 Simultaneously-Because CPP32 and granzyme B process pro-Mch6 preferentially at two different sites, it is possible that activation of pro-CPP32 by granzyme B in cell lysates could result in cleavage of pro-Mch6 at both sites simultaneously. To test this possibility, 35 S-labeled pro-Mch6 was mixed with cell extract from the 697 lymphocytic cell line and then incubated with granzyme B. The proteins were analyzed at different time points by Western blotting using the CPP32-p20 antibody to detect activation of endogenous pro-CPP32 (Fig. 4A, upper  panel) and by autoradiography to detect processing of pro-Mch6 (Fig. 4A, lower panel). The p20/p19 doublet of mature CPP32 was detected in less than 15 min simultaneously with the disappearance of the 32-kDa pro-CPP32 (Fig. 4A, upper panel,  lane 2). Subsequently, the p20 product was autocatalytically processed to the p19 (lanes 3-6). In the presence of the DEVD-CHO inhibitor, only the p20 can be detected (lane 7) because CPP32 was inhibited under these conditions. Autoradiographic analysis of the same samples revealed a time-dependent processing of pro-Mch6 to the p35 (granzyme B product) and p37 (CPP32 product) bands (Fig. 4A, lower panel). Also detectable was the p12/p10 band. Interestingly, when CPP32 was inhibited by DEVD-CHO, only the granzyme B-generated p35 and p12 bands were detected (lane 7). Because cell lysates contain all of the cytoplasmic components necessary for apoptosis (21, 28 -30), the observed products (p37, p35, p12, and p10) are most likely produced in cells undergoing granzyme B-mediated apoptosis. Therefore, the mature Mch6 enzyme is derived from the pro-enzyme by cleavage at Asp 315 and Asp 330 to generate the large (p35) and small (p10) subunits (Fig. 4B). Because there was no evidence of efficient processing in the pro-domain in cell lysate (Fig. 4A), our data suggest that p35 and p10 are the large and small subunits of mature Mch6, respectively. However, we cannot exclude the possibility that under different conditions, the p35 product may be further processed to remove the N-terminal pro-domain.
pro-Mch2␣ Is Also a Substrate for Mature CPP32-In cells undergoing apoptosis or in nuclei incubated with apoptotic extracts, lamin cleavage occurs subsequent to PARP cleavage and is much slower than PARP cleavage (31,32). CPP32 and Mch3 are the most efficient PARP-cleaving enzymes (12,33), and they are probably responsible for PARP cleavage in cells undergoing apoptosis. However, the two enzymes are not able to cleave lamin (20). This evidence suggests that the lamincleaving enzyme is downstream of CPP32 and Mch3, and it is activated by one or both enzymes. Thus far, the only known ASCP that can cleave lamin is Mch2␣ (20). To test whether pro-Mch2␣ is a substrate for CPP32, pro- sis (Fig. 5A) revealed that CPP32 cleaves pro-Mch2␣ at multiple sites to generate four cleavage products of molecular masses ϳ21, ϳ18, ϳ13, and ϳ11 kDa. These products can be detected within the first 15 min (Fig. 5, lanes 2 and 3). Longer incubation results in a decrease in the intensity of the 21-and 13-kDa products and an increase in the intensity of the 18-and 11-kDa products (lanes 5 and 6). This suggests that the 18-and 11-kDa products are derived from the 21-and 13-kDa products, respectively, by further proteolytic processing. No further processing was observed upon prolonged incubation, suggesting that the 18-and 11-kDa products are most likely to be the large and small subunits, respectively, of the mature Mch2␣ enzyme. Unlike CPP32, an equal amount of recombinant bacterially expressed Mch3 was unable to process pro-Mch2␣ (data not shown).
Primary Structure of Mch2␣-The exact aspartate processing sites in the pro-Mch2␣ polypeptide chain have not yet been determined. Two potential processing sites between the two subunits of Mch2␣ ( 176 DVVDN 180 and 190 TEVDA 194 ) and one in the pro-domain ( 20 TETDA 24 ) are very similar to the CPP32 tetrapeptide substrate DEVD. To determine whether CPP32 can cleave pro-Mch2␣ at the proposed processing sites, mutant pro-Mch2␣ variants with a P1 D to A or R substitution in these sites were generated. The wild-type and mutant pro-Mch2␣ variants were in vitro translated in the presence of [ 35 S]methionine, incubated with purified CPP32, and then analyzed by SDS-PAGE and autoradiography. As expected, CPP32 was able to cleave the Asp 23 -mutated pro-Mch2␣ between the two subunits but not within the pro-domain (Fig. 5B). This generated the 21-, 13-, and 11-kDa products but not the 18-kDa product. The 13-kDa product was further processed to the 11-kDa product, indicating that there is an additional processing site between the two subunits of Mch2␣. However, the N-terminal 21-kDa product was not further processed to the final 18-kDa product. This establishes that Asp 23 is the processing site within the pro-domain and that the 18-kDa product is generated from the 21-kDa product.
Double mutation of Asp 179 and Asp 193 completely prevented processing between the two subunits of Mch2␣ (Fig. 5C) but did not affect processing within the pro-domain. CPP32 was able to cleave the 34-kDa double mutant pro-Mch2␣ at Asp 23 to generate a 31-kDa product (Fig. 5C). This establishes that Asp 179 and Asp 193 are the two processing sites between the two subunits of Mch2␣. This was further confirmed using single mutants of Asp 179 or Asp 193 pro-Mch2␣ (Fig. 5, D and E). Cleavage of the Asp 179 -mutated pro-Mch2␣ by CPP32 generated the 11-kDa product and a 20-kDa product but not the 13-or 18-kDa products (Fig. 5D). This suggests that the 11-kDa and the new 20-kDa products are generated by cleavage at Asp 23 and Asp 193 . Cleavage of the Asp 193 -mutated pro-Mch2␣ by CPP32 generated the 21-, 18-, and 13-kDa products but not the 11-kDa product (Fig. 5D). These combined data establish that the 11-kDa product is generated by cleavage at Asp 193 , and the 18-kDa product is generated by cleavage at Asp 23 and Asp 179 . Furthermore, cleavage at Asp 179 appears to be more efficient than cleavage at Asp 193 . This is because the 21-and 13-kDa products appear earlier than the 18-and 11-kDa products (Fig.  5A, lane 2). In addition, CPP32 requires less time to completely cleave wild-type pro-Mch2␣ (Fig. 5A) than the Asp 179 -mutated pro-Mch2␣ (Fig. 5C). According to these results, the mature Mch2␣ enzyme is derived from the pro-enzyme by cleavage at Asp 23 , Asp 179 , and Asp 193 to generate the large (p18) and small (p11) subunits (Fig. 5F).
Cleavage of pro-Mch2␣ by CPP32 Activates Mch2␣-A similar mutation analysis revealed that Mch2␣ is able to cleave its own precursor pro-Mch2␣ at Asp 23 and Asp 193 but not very well at Asp 179 (data not shown). Recently, we have shown that the granzyme B-activated CPP32 autocatalytically cleaves its own pro-domain once it is activated by granzyme B or Mch4 (13). Therefore, it is possible that once pro-Mch2␣ is cleaved by CPP32 at Asp 179 , it becomes active and could act synergistically with CPP32 to process itself at Asp 23 and Asp 193 . However, because both Mch2␣ and CPP32 can cleave pro-Mch2␣ at these sites, it is difficult to distinguish between the two activities. To overcome this problem, we introduced the lamin cleavage site (VEIDN) (20)  On the other hand, a very weak activity was observed with an equal amount of mature CPP32 (Ͻ5% cleavage) (lane 3). Consequently, when pro-Mch2␣ containing the lamin site was incubated with CPP32, pro-Mch2␣ was processed at Asp 179 and Asp 193 to generate the 21-, 13-, and 11-kDa products within the first 15 min (Fig. 6B, lane 2). Processing at Asp 23 followed more slowly, and after 2 h of incubation, about 65% of p21 was converted to p18 (lane 5). Since an equal amount of CPP32 generates less than 5% of cleavage at this site (Fig. 6A, lane 3), the remaining 60% of cleavage must be attributed to an autocatalytic activity of the activated Mch2␣. Also because the in vitro translated pro-Mch2␣ is present at a very low concentration in the reaction mixture, the observed autocatalytic activity could be interpreted as evidence for intramolecular processing. Therefore, these data clearly demonstrate that once pro-  2, upper panel). This is evident from the formation of the p20/p19 large subunit of CPP32. Autoradiographic analysis of the same samples revealed a time-dependent processing of pro-Mch2␣ (Fig. 7B,  lower panel). pro-Mch2␣ was processed to its individual subunits immediately after activation of the endogenous pro-CPP32 by granzyme B (lanes 3-5). Endogenous CPP32 was unable to process pro-Mch2␣ in the presence of 40 nM DEVD-CHO inhibitor (lane 6). Similarly, recombinant CPP32 was also unable to process pro-Mch2␣ in the presence of this inhibitor (Fig. 7A, lane 4). These data suggest that the enzyme responsible for processing of pro-Mch2␣ in cell extracts is most likely to be CPP32 itself. However, we cannot exclude the possible contribution of other granzyme B-activable, DEVD-CHO-inhibitable ASCPs present in the cell extract that can also process pro-  dant in most cell types (8), it is most likely to be the enzyme responsible for activation of Mch2␣ in most cases of apoptosis.
Mch2␣ Can Activate CPP32-Interestingly, we observed that mature Mch2␣ can also process pro-CPP32 at the 172 IETDS 176 site between the two subunits of CPP32 (Fig. 8, lane 2). No cleavage was observed with the Asp 175 -mutated pro-CPP32 (lane 5), indicating that Mch2␣ processes CPP32 correctly. This processing was accompanied by activation of CPP32, as evident from some autocatalytic conversion of CPP32-p20 to p19 (lane 2). The autocatalytic processing occurs at Asp 9 , because it is inhibited by a D to A mutation at that site (lane 7). These results are similar to those obtained with Mch4 or granzyme B (13). Also, the autocatalytic processing of CPP32 at Asp 9 , but not the activity of Mch2␣, was inhibited by the DEVD-CHO inhibitor (40 nM) (lane 3), indicating that Mch2␣ is a non-DEVD-CHO-inhibitable protease. Higher concentrations of DEVD-CHO up to 1 M had no effect on Mch2␣ (data not shown).
Therefore, it is possible that once pro-CPP32 is activated in 1 apoptosis by upstream proteases such as Mch4 or Mch5 (13), it could activate pro-Mch2␣, which in turn may feed back on pro-CPP32, resulting in a protease amplification cycle (Fig. 9). Similarly, in a parallel apoptotic pathway, pro-Mch2␣ may be activated first by Mch4 and/or Mch5, 2 which then could lead to activation of the same protease amplification cycle. Activation of the upstream proteases Mch4 or Mch5 most likely depends on autocatalysis, which is the slowest step in the protease cascade. Autocatalysis is probably triggered by interaction with FADD or FADD-like death effector proteins that receive the apoptosis signal from the Fas receptor or other signal transduction pathways. We predict that because activation of all ASCPs require cleavage after Asp residues present between their large and small subunits, protease amplification cycles might be common among all members of this family. Once the slow autocatalysis step is completed, feedback from the downstream protease will initiate the protease amplification cycle to ensure efficient processing and activation of the remaining pro-enzymes involved in the cascade.
Conclusions-We have identified and cloned a novel ASCP that could play a role in the apoptotic protease cascade. We demonstrated that pro-Mch6 is a substrate for CPP32 and granzyme B. The mature Mch6 enzyme is composed of a large subunit (p35) and a small subunit (p10) that are derived from a single chain pro-enzyme by cleavage after two aspartate residues. We also demonstrated that the pro-enzyme of the lamin-cleaving enzyme Mch2␣ is a substrate for CPP32. The mature Mch2␣ enzyme is made of a large subunit (p18) and a small subunit (p11) that are derived from a single-chain proenzyme by cleavage after three aspartate residues. The ability of CPP32 to activate pro-Mch6 and pro-Mch2␣ suggests that these proteases are downstream proteases that might be responsible for the late morphological changes in apoptotic cells. One of these late morphological changes is lamin cleavage, resulting in collapse of the nucleus into apoptotic bodies. The ability of granzyme B to process/activate pro-Mch2␣ and pro-Mch6 through CPP32 is the first direct demonstration of an apoptotic protease cascade. These proteases might be active participants in granzyme B-mediated apoptosis.