Molecular Cloning and Characterization of Two Novel Pro-apoptotic Isoforms of Caspase-10*

Caspase-10/a (Mch4) and caspase-10/b (FLICE2) are related death effector domain-containing cysteine aspartases presumed to be at or near the apex of apoptotic signaling pathways. We report the cloning and characterization of two novel proteins that are splice isoforms of the caspase-10 family. Caspase-10/c is a truncated protein that is essentially a prodomain-only form of the caspase that lacks proteolytic activity in vitro but efficiently induces the formation of perinuclear filamentous structures and cell death in vivo. Caspase-10/c mRNA is specifically up-regulated upon TNF stimulation, suggesting a potential role of this isoform in amplifying the apoptotic response to extracellular stimuli such as cytokines. Caspase-10/d is a hybrid of the known caspases Mch4 and FLICE2, as it is identical to FLICE2 except for the small (p12) catalytic subunit, which is identical to Mch4. Caspase-10/d is proteolytically activein vitro and also induces cell death in vivo, although it is less active than Mch4. The mRNAs for all known isoforms of caspase-10 are abundantly expressed in fetal lung, kidney, and skeletal muscle but are very poorly expressed or absent in these tissues in the adult, implying a possible role for the caspase-10 family in fetal development.

Caspase-10/a (Mch4) and caspase-10/b (FLICE2) are related death effector domain-containing cysteine aspartases presumed to be at or near the apex of apoptotic signaling pathways. We report the cloning and characterization of two novel proteins that are splice isoforms of the caspase-10 family. Caspase-10/c is a truncated protein that is essentially a prodomain-only form of the caspase that lacks proteolytic activity in vitro but efficiently induces the formation of perinuclear filamentous structures and cell death in vivo. Caspase-10/c mRNA is specifically up-regulated upon TNF stimulation, suggesting a potential role of this isoform in amplifying the apoptotic response to extracellular stimuli such as cytokines. Caspase-10/d is a hybrid of the known caspases Mch4 and FLICE2, as it is identical to FLICE2 except for the small (p12) catalytic subunit, which is identical to Mch4. Caspase-10/d is proteolytically active in vitro and also induces cell death in vivo, although it is less active than Mch4. The mRNAs for all known isoforms of caspase-10 are abundantly expressed in fetal lung, kidney, and skeletal muscle but are very poorly expressed or absent in these tissues in the adult, implying a possible role for the caspase-10 family in fetal development.
Apoptosis is a process of regulated cell suicide crucial for the development and homeostasis of multicellular organisms that is characterized by chromatin condensation, DNA fragmentation, cell shrinkage, and plasma membrane blebbing (1). The gene products of ced-3 and ced-4 regulate the apoptotic death of cells during normal development of the nematode Caenorhabditis elegans (2). Subsequently, CED-3 was shown to have a mammalian homologue named interleukin 1␤-converting enzyme (ICE 1 ; caspase-1) (3), which was the first of a series of proteases to be identified with an active-site cysteine and a unique aspartate-Xaa substrate specificity. In recognition of these properties, this family of apoptotic proteases was renamed the caspases, for cysteine-aspartic acid proteases (4). With the recent discovery of MICE (5), the number of caspases now stands at 14, including murine proteases.
Caspases are present as inactive proenzymes comprising a prodomain and a catalytic protease domain that can be further processed to give a large and a small subunit. Crystallographic data (6,7) has revealed (for caspases 1 and 3 and by extrapolation to the other caspase family members) that the activated caspase exists as a heterodimeric tetramer of two large and two small subunits, with each substrate binding pocket spanning both subunits. Evidence for the sequential activation of caspases has led to the concept of a caspase cascade (8 -11), with initiator caspases at the apex, transmitting signals to executor caspases that cleave a wide variety of substrates. In this scheme, caspases 2, 8, and 10 (12-16) have been demonstrated to associate with the death receptors Fas and TNF 1 receptor 1 and are presumably initiator caspases.
In terms of their substrate specificities, the caspases can be classified into three groups (17,18), all of which have an absolute requirement for an aspartate residue in the P1 position but otherwise show varying degrees of flexibility. Group I caspases (caspases 1, 4, and 5) prefer the tetrapeptide sequence WEHD; group II caspases (caspases 2, 3, and 7) have a DEXD specificity, whereas group III caspases (caspases 6, 8, 9, and 10) cleave after (I/L/V)EXD. It should be noted, however, that these are only the preferred substrate motifs and are not absolute, being dependent on factors such as concentration, reaction time, and accessibility. Multiple redundancies are common among the caspases, with many caspases having the ability to cleave the same substrates, at least in vitro (17,18). There is an additional level of complexity with the discovery that multiple isoforms are present for many caspases, including caspases 1, 2, 6, and 8 (19). These isoforms are most likely splice variants (20) or variants derived from post-translational modifications, but the biological significance of these variants is mostly unclear.
In this paper we describe the identification and characterization of two novel isoforms of caspase-10 (caspase-10/c and caspase-10/d) that incorporate features of both Mch4 (caspase-10/a) (14) and FLICE2 (caspase-10/b) (15). We describe their tissue expression and activity both in vitro and in vivo and examine the effect of exogenous stimulation by TNF on their mRNA expression.

EXPERIMENTAL PROCEDURES
Cell Lines and Reagents-Expression vectors coding for PARP or caspase-3/Yama and the 293T cell line were generous gifts from Dr. V. Dixit. Antibodies against the large and small catalytic subunits of caspase-10 were obtained from Research Diagnostics, Inc. The polyclonal serum directed against a caspase-10/c-specific peptide sequence (EGSCVQDESEPQRPL) was produced by Genosys, Inc. Chemiluminescent Western blots were developed using the Immun-Star kit from * This work was supported by the Institute of Molecular and Cell Biology, Singapore. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AF111344 (caspase-10/c) and AF111345 (caspase-10/d).
Cloning and Identification of Caspase-10 Variants-PCR was performed on human spleen, thymus, and peripheral blood leukocytes cDNA libraries (CLONTECH) using primers 5MCH4 (AAA GGA TCC GCT AGC ATG AAA TCT CAA GGT CAA CAT TGG TAT TCC) and 3MCH4 (AAA GAA TTC CTA TAT TGA AAG TGC ATC CAG GGG CAC AGG) to obtain the coding regions of Mch4, caspase-10/c, and caspase-10/d. Advantage-HF polymerase (CLONTECH) was used with cycling conditions recommended by the manufacturer. The final clones were from spleen (caspase-10/c) and thymus (caspase-10/d). The PCR products were inserted into pCITE-4a(ϩ) (Novagen) for in vitro expression as S-Tag fusion proteins or pCI-neo (Promega) for in vivo eukaryotic expression.
mRNA Expression Analysis-The primers C614 (CCG AGT CGT ATC AAG GAG AGG AAG AAC) and 3BRIDGE (TAT ATG CAC TGT GAA CCC AAG CCA) were used in PCR and RT-PCR experiments to obtain short DNA fragments corresponding to Mch4, caspase-10/c, and caspase-10/d. RNA from various cell lines was purified using the Qiagen RNAeasy kit, and reverse-transcription was performed using Super-Script II enzyme with oligo(dT) primers from Life Technologies, Inc. PCR was then performed for 30 cycles on the resulting cDNA at an annealing/extension temperature of 68°C using Advantage-HF polymerase or on cDNA from adult and fetal multiple tissue cDNA panels (CLONTECH), and the products were analyzed in ethidium bromidestained 2.5% agarose gels.
Transfections, Killing Assays, and Solubility Studies-For the killing assays, MCF7 cells were seeded at a density of 1 ϫ 10 5 cells/60-mm dish in 5 ml of RPMI 1640 supplemented with 10% fetal bovine serum and 1% penicillin/streptomycin. At ϳ60 -80% confluence, each dish of cells was transfected with 4 g of pCI-neo expression vector with or without the CASP insert and 1 g of pCMV-␤gal reporter plasmid (CLON-TECH) using Tfx-20 reagent (Promega) as directed by the manufacturer, except that the cells were washed once with PBS before incubation with the DNA-transfection reagent complex, and incubation was for 45 min at 37°C. Where required, tetrapeptide inhibitors Ac-YVAD-CHO or Ac-DEVD-CHO (Biomol) were added after this incubation. Approximately 16 -24 h post-transfection, the cells were fixed in 0.25% glutaraldehyde and stained with 0.2% 5-bromo-4-chloro-3-indolyl-␤-Dgalactopyranoside (X-gal), 2 mM MgCl 2 , 5 mM K 4 Fe(CN) 6 ⅐3H 2 O, 5 mM K 3 Fe(CN) 6 for 2 to 10 h at 37°C. Blue-stained cells were scored as apoptotic on the basis of morphology. Four random microscopic fields per dish were counted with a minimum of 400 blue cells per data set. For the concurrent determination of PARP cleavage, transfected cells were boiled in 500 l of Laemmli sample buffer (62.5 mM Tris-HCl, pH 6.8, 2% SDS, 7.5% Ficoll Type 400, 5% ␤-mercaptoethanol, 0.001% bromphenol blue), and the extracts were immunoblotted using polyclonal anti-PARP antiserum (Upstate Biotechnology, Inc.).
For the solubility studies, 293T or MCF7 cells were seeded at a density of 5 ϫ 10 6 cells/100-mm dish in 10 ml of growth medium. At approximately 80% confluence, the cells were transfected with 12 g of plasmid DNA, with the addition of Ac-DEVD-CHO inhibitor at a final concentration of 20 M to delay cell death. At 24 h post-transfection, the cells were scraped and pelleted by centrifugation. Total lysates were obtained by boiling the pellets in 1 ml of Laemmli sample buffer. In experiments separating soluble from insoluble fractions, pellets were first incubated in lysis buffer (0.1% Nonidet P-40, 250 mM NaCl, 50 mM Hepes pH 7.4, 5 mM EDTA, Complete Protease Inhibitor mix (Boehringer Mannheim)) for 30 min at 4°C, then centrifuged for 20 min at 10,000 ϫ g. The supernatant was harvested as the soluble fraction, whereas the pellet was washed three times with lysis buffer, then boiled in 300 l of Laemmli sample buffer and used as the insoluble fraction. Immunoblotting was carried out using either commercial goat anti-Mch4 p12 subunit antibody at 1:500 dilution or custom anti-caspase-10/c polyclonal antiserum at 1:1000 dilution in blocking buffer (5% milk/PBS, 0.05% Tween 20).
Confocal Microscopic Visualization-Constructs of Mch4, caspase-10/c, and caspase-10/d fused to the N terminus of EGFP were made by cloning the respective cDNAs into pEGFP-N3 (CLONTECH). MCF7 cells were transfected in 60-mm dishes using 2 g of pEGFP-N3-derived plasmids in the presence of 50 g of Ac-DEVD-CHO. Approximately 16 h post-transfection, the cells were fixed in 4% paraformaldehyde, PBS for 15 min at 25°C, then equilibrated in 2ϫ SSC buffer (15 mM sodium citrate, 150 mM NaCl) and incubated for 30 min at 37°C with 100 g/ml DNase-free RNase in 2ϫ SSC. After rinsing with 2ϫ SSC, the cells were stained for 2 min in propidium iodide for nuclear visualization, mounted in Vectashield (Vector Laboratories), and visualized under confocal laser scanning microscopy (Bio-Rad MRC 1024).
TNF Stimulation and mRNA Expression-MCF7 cells were seeded at 5 ϫ 10 6 cells/100-mm dish in 10 ml of growth medium, then grown to high density, usually 48 h post-seeding. Recombinant human TNF␣ (BIOSOURCE) was added directly to the conditioned medium at a final concentration of 50 ng/ml where required. Treated cells were harvested, and RT-PCR was performed essentially as described above.
Cytotoxicity Assay-MCF7 cells were seeded at a density of 3 ϫ 10 4 cells/well in 96-well plates. Approximately 48 h post-seeding, recombinant human TNF␣ was added at a final concentration of 50 ng/ml. At various time points, the medium was aspirated, and the cells were washed with 100 l of PBS, fixed in 5% formaldehyde, PBS for 5 min, then stained for 20 min at room temperature in a solution of 0.5% crystal violet, 20% methanol. After thorough washing to remove unincorporated dye, the plates were dried completely at room temperature. The stained cells were then lysed using 50 l of 33% acetic acid per well, and the absorbance at 550 nm was measured in a microplate reader.
In Vitro Cleavage Analysis-The coding regions of Mch4, caspase-10/c, and caspase-10/d starting from Ile (194) (14) were each inserted into the pGEX-2TK bacterial expression plasmid (Amersham Pharmacia Biotech), enabling their expression as GST fusion proteins in Escherichia coli JM109. The pellets from 200 ml of 1 mM isopropyl-␤-Dthiogalactopyranoside-induced culture were resuspended in 20 ml of ICE cleavage buffer (25 mM Hepes, pH 7.4, 5 mM EDTA, 2 mM dithiothreitol, 0.1% CHAPS) and sonicated for 3 ϫ 20 s. After centrifugation for 30 min at 12,000 ϫ g, the supernatants were collected and used as the source of active caspase. Between 2 and 5 l of induced sonicate (depending on the amount of processed p12 small subunit, as described under "Results") was added to 3 l of [ 35 S]methionine-labeled substrates expressed in vitro using the T7 Quick TnT system (Promega). ICE cleavage buffer was then added to a final volume of 30 l. Where used, the tetrapeptide inhibitors Ac-YVAD-CHO and Ac-DEVD-CHO were added to a final concentration of 1 M. In some experiments, the radiolabeled in vitro translated proteins were first affinity-purified by mixing the entire translation product with 200 l of S-protein agarose slurry (Novagen) and using 10 l of the washed, protein-bound matrix as substrates.

RESULTS
Cloning and Identification of Caspase-10 Variants-Previous reports had indicated the possible existence of multiple isoforms of caspase-10 (15,21). Our initial cloning by PCR of what appeared to be a single band of approximately 1500 bp emerged on sequencing to consist of three individual DNA fragments, wild type Mch4 (1440 bp), caspase-10/c (1477 bp), and caspase-10/d (1569 bp). It was possible to distinguish electrophoretically between these three DNAs by designing PCR primers (5MCH4/3BRIDGE) spanning a mid-insert region (comprising exons 6 and 7) common to caspase-10/d and FLICE2 (Fig. 1). Bands of 230, 260, and 350 bp in size corresponded to Mch4, caspase-10/c, and caspase-10/d, respectively (see below).
In the recent paper on the characterization of Usurpin (22), the authors described the sequencing of the death effector domain (DED)-caspase gene cluster on human chromosome 2 band q33-34, which contains the genes for Usurpin, caspase-10/a/Mch4, and caspase-8/a/FLICE. Based on the genomic map, the mid-insert region mentioned above consists of exons 6 and 7 (Fig. 1). Hence Mch4 lacks exons 6, 7, and 11. In contrast, FLICE2 contains exons 6 and 7 but skips exon 10 and contains part of exon 11. Caspase-10/c contains exon 6 but not 7, whereas caspase-10/d contains both exons 6 and 7 (summarized in Fig. 1).
The predicted amino acid sequences show that caspase-10/d is a hybrid of FLICE2 and Mch4 (Fig. 2). Interestingly, the insertion of exon 6 in caspase-10/c leads to a frameshift at the protein level, resulting in a protein that is truncated shortly after the DED-containing prodomain. Thus, caspase-10/c is essentially a prodomain-only protein (Fig. 2). We confirmed the authenticity of the predicted protein sequences by translation in vitro and by performing immunoblotting on the products. As predicted, in vitro translated caspase-10/d was slightly larger in size than Mch4 and was recognized by antibodies against both the large and small protease subunits of Mch4 (Fig. 3). In vitro translated caspase-10/c was a much smaller protein of approximately 40 kDa on SDS-polyacrylamide gel electrophoresis (Fig. 3, upper panel) that was not recognized by either of the anti-protease subunit antibodies (Fig. 3, lower panel).
Tissue-and Cell Line-specific mRNA Expression of Caspase-10 Isoforms-To determine the constitutive mRNA expression profiles of the caspase-10 variants, PCR and RT-PCR was performed on RNA extracted from various cell lines and on commercially available cDNA panels. Because the CLONTECH multiple tissue cDNA panels are prenormalized against several housekeeping genes, we were able to directly compare the levels of mRNA expression between samples. The bands representing Mch4, caspase-10/c and caspase-10/d were 230, 260, and 350 bp in size, respectively (Fig. 4). Their identity was confirmed by subcloning and sequencing two independent clones of each DNA fragment from spleen, thymus, and peri- pheral blood leukocytes cDNA libraries. Although this method did not distinguish between FLICE2 and caspase-10/d expression (because they both contain the same exon 6 and 7 insert), it is clear that expression levels of caspase-10-related mRNAs are generally higher in fetal than in adult tissues (Fig. 4A). In particular, Mch4, caspase-10/c, and FLICE2/caspase-10/d are highly expressed in fetal skeletal muscle, lung, and kidney, whereas their expression in these tissues is virtually undetectable in the adult.
Expression of the caspase-10 isoforms was highly variable among various cell lines (Fig. 4B). There was no obvious pattern in their distribution. The 350-bp FLICE2/caspase-10/d band was consistently highly expressed in all cell lines tested, whereas caspase-10/c and Mch4 were present at very low levels in the MCF7 breast carcinoma and SW480T adenocarcinoma cell lines (Fig. 4B). Caspase-10/c was not absent, as increasing the number of PCR cycles resulted in easier visualization (not shown). Caspase-10/d but Not Caspase-10/c Is Proteolytically Active in Vitro-Crude sonicated lysates of bacterial cultures expressing recombinant Mch4 and caspase-10/d were proteolytically active. Immunoblotting of the bacterial sonicates using antibodies against the Mch4 p12 small subunit revealed that autoprocessing had occurred in the case of Mch4 and caspase-10/d with the release of the p12 subunit but not with caspase-10/c (Fig. 5, panel A).
The bacterial sonicates containing Mch4 and caspase-10/d cleaved in vitro translated full-length Mch4 and caspase-10/d proteins, demonstrating the possibility of intermolecular processing of caspase-10 isoforms (Fig. 5, panel B) and also cleaved in vitro translated PARP and caspase-3 (Fig. 5, panels C and  D). Again, caspase-10/c was not proteolytically active in any of these experiments. The activity of Mch4 or caspase-10/d was completely abrogated by the addition of 1 M specific caspase inhibitor Ac-DEVD-CHO but not Ac-YVAD-CHO (Fig. 5, panels  C and D). The cleavage activity (and the inhibition thereof) is because of the expressed recombinant caspase and not to the activation of any other endogenous proteases because (i) bacteria do not have endogenous caspase activity (Ref. 23 and vector controls in Fig. 5, panels B-D), (ii) affinity-purified recombinant caspases are proteolytically active (14,23), 2 and (iii) the in vitro translation products used as substrates in some of our experiments were first purified from any contaminating reticulocyte components.
In addition, we consistently observed that caspase-10/d was less active than Mch4 after normalization of the starting quantities of bacterial sonicates used based on the amount of processed p12 subunit. We quantitated this activity using the caspase substrate DEVD-pNA and confirmed that Mch4 was approximately twice as active as caspase-10/d (not shown).
Caspases 10/c and 10/d Are Highly Toxic to MCF7 Cells-Despite the lack of a protease domain, caspase-10/c was highly efficient in causing apoptosis when transiently overexpressed in the MCF7 cell line. Caspase-10/d was also pro-apoptotic, as was Mch4 used as a control (Fig. 6). This killing by all three isoforms was caspase-dependent, as shown first by the dosedependent inhibitory effect of the tetrapeptide inhibitor Ac-DEVD-CHO (Fig. 6, panel A) and second by the observation that PARP was cleaved in cells transfected with Mch4, caspase-10/c, or caspase-10/d (Fig. 6, panel B, lanes 3-5). Furthermore, the fact that Ac-DEVD-CHO but not Ac-YVAD-CHO inhibited killing indicated that apoptosis was transduced by downstream caspase-3-like enzymes, and caspase-1-like proteases were probably not involved (17,18).

Mch4 and Caspase-10/d Are Soluble Proteins, Whereas Caspase-10/c Is Insoluble and Localizes to Filamentous Perinuclear Structures-Lysates of mammalian cells transiently overexpressing Mch4
, caspase-10/c, or caspase-10/d were prepared as detergent-soluble and -insoluble fractions, and immunoblotting was performed using either anti-Mch4 p12 antibodies to detect Mch4 and caspase-10/d or custom anti-caspase-10/c polyclonal antiserum. In both MCF7 and 293T cell lines, Mch4 and caspase-10/d were completely soluble, whereas caspase-10/c was predominantly insoluble (Fig. 7). These results are consistent with recent evidence that DED-only proteins might form insoluble filamentous structures (death effector filaments, DEF) either on their own or in association with as yet unidentified cytoskeletal elements (24). To determine whether DEFs were formed by caspase-10 isoforms, Mch4, caspase-10/c and caspase-10/d were ectopically expressed as EGFP fusion proteins in MCF7 and 293T cells and visualized using confocal laser scanning microscopy. Distinct perinuclear filamentous structures were seen in all cells expressing caspase-10/c-EGFP (Fig. 8, panel A). In most cases these filaments were closely bundled together, but more haphazard arrangements were sometimes also observed. EGFP alone was present in both the nucleus and diffused in the cytoplasm (Fig.  8, panels N1 and N2). Mch4-EGFP and caspase-10/d-EGFP were mostly diffused throughout the cytoplasm (Fig. 8, panels WT1 and B), but filaments could be seen in some cells (Fig. 8, panels WT2 and B), although these were never as distinct as those seen in caspase-10/c-transfected cells. Killing assays performed on MCF7 cells transiently overexpressing EGFPcaspase-10 fusion proteins confirmed that these EGFP fusion proteins were functionally active and induced apoptosis (data not shown).
Caspase-10/c mRNA Is Up-regulated by TNF-It has been shown that Mch4 and FLICE2 can associate with TNF receptor 1 via homotypic DED interactions with Fas-associated death domain protein (FADD) (15). It was of interest to know whether the expression of the mRNAs of any of the caspase-10 isoforms was altered by TNF stimulation. The addition of TNF resulted in a specific up-regulation of caspase-10/c mRNA. This upregulation was transient and typically observed at 18 h after TNF stimulation (Fig. 9, panel A), although it was sometimes observed at other times between 12 and 24 h post-stimulation (data not shown). The peak of mRNA induction occurred before or during the majority of apoptotic cell death (Fig. 9, panel B) and was not observed in parallel control experiments in which TNF was omitted (Fig. 9, panel A).  4 -7) and harvested at various times as indicated. RT-PCR was performed on RNA prepared from these cells as detailed in the text, and the products were electrophoresed in 2.5% agarose gels. Lanes 8 -11, parallel experiment without TNF. The figure shows one set of results that are representative of at least three similar experiments. WT, 10/c, 10/d, control PCR reactions using plasmids coding for Mch4, caspase-10/c, or caspase-10/d, respectively. GAPDH, control RT-PCR using primers specific for glyceraldehyde-3phosphate dehydrogenase. B, TNF was added to MCF7 cells grown in 96-well plates, and the extent of killing at various time points was measured as described under "Experimental Procedures." 31) has been controversial, because in some studies, MRIT had an inhibitory effect on apoptosis, whereas in others, MRIT induced cell death. Interestingly, this protein was reported to exist as two splice-isoforms, with the shorter variant essentially comprising just two DED domains, analogous in structure to the pro-apoptotic FLICE isoform MACH␤1 (12) and our caspase-10/c. The recent work on DEF (24) sheds some light on the reason for this controversy, suggesting that apoptotic activity is a concentration-dependent phenomenon. Expression of only the DED of FLICE (but not the full-length protein) above a certain concentration threshold resulted in the formation of an insoluble filamentous perinuclear structure that apparently acted as a scaffold to recruit and facilitate the autoactivation of cytoplasmic DED-containing pro-caspases (24). We also showed that pro-apoptotic caspase-10/c is present mainly in the insoluble fraction of cell lysates (in contrast to the predominantly soluble full-length Mch4 and caspase-10/d proteins) and that ectopically expressed caspase-10/c-EGFP fusion protein localizes intracellularly as distinct perinuclear structures identical to the DEFs described previously. These observations provide a plausible explanation for the mechanism of action of caspase variants containing only the DED and suggest that caspase-10/c and the ␤-isoforms of FLICE and MRIT may play a role in amplifying the apoptotic response through the formation of similar "activating scaffolds." How might caspase-10/c induce apoptosis after the possible formation of the DEF? We showed that cell death induced by caspase-10/c was inhibitable by Ac-DEVD-CHO, suggesting the requirement for caspase-3-like proteases. Yet, in MCF7 cells, caspase-3 is absent (32), and we were unable to detect specific caspase-10/c-induced activation of the two other known DEVDinhibitable caspases 2 and 7, 2 which is similar to the situation in TNF-induced apoptosis (33). Thus, the caspases that are activated by the overexpression of caspase-10/c in MCF7 cells remain to be identified.
It is known that caspase transcripts are constitutively present in many cells, albeit at different levels of expression, and because apoptosis can occur in the presence of protein synthesis inhibitors such as cycloheximide (21), pro-caspases must already be present at sufficient levels to allow the apoptotic response to proceed. Nonetheless, it has been shown that caspase transcripts are up-and down-regulated in development. For example, caspase-2 (Ich-1/Nedd2) mRNA is more highly expressed in the mouse during embryonic development than in the adult, particularly in the brain, liver, lung and kidney (34). Our observations that the expression of Mch4 and caspases-10/c and -10/d/FLICE2 is considerably higher in fetal than in adult tissues, particularly in skeletal muscle, lung, and kidney, suggests a role for the isoforms of caspase-10 during development.
Caspase mRNAs may also be up-regulated by extracellular stimuli as well as in a genetic program during development. Caspase-1, -3, -4, -7, -8, and -10 transcripts are up-regulated within 16 h of interferon-␥ stimulation (35,36), and the CASP-2 and -3 genes are also up-regulated in response to etoposide treatment (37). We showed TNF induces the expression of caspase-10/c mRNA but not caspase-10/d or Mch4 transcripts, the first demonstration that the mRNA of a protein encoding only the DED is up-regulated by a specific stimulus. Overall, these studies suggest that cytokine-induced cytotoxicity could potentially involve the up-regulation of specific caspase isoforms.
Accordingly, we propose a hypothetical model (Fig. 10) wherein the binding of TNF to its receptors (most likely TNF receptor 1 (15)) results in the recruitment and activation of caspase-8 and/or -10 at the plasma membrane, and simulta-neously, an as yet undefined signaling pathway stimulates expression of caspase-10/c mRNA. A concomitant increase in caspase-10/c protein may be sufficient for DEF scaffold formation and subsequent intracellular caspase recruitment and activation, thereby amplifying the apoptotic response. However, the existence of such an amplification model in vivo remains to be proven, as we have been unable to detect any increase in caspase-10/c protein levels by Western blotting. First, the increase in caspase-10/c mRNA is transient, occurring at any time from 12 to 24 h after TNF treatment, and so the caspase-10/c protein synthesis may also be transient. Second, it is possible that any TNF-induced increase in caspase-10/c protein, although below the threshold of detection with our antibody, is nonetheless sufficient to induce cell death.
It appears that within each family of the DED-containing proteins (caspase-8, -10, and MRIT) there exists a cohort of isoforms (12,25) with variations in the composition of the protease domains depending on variations in splicing. These isoforms appear to be analogous in structure between the families, suggesting a common pattern of functionality. For instance, expression of DED-only isoforms is pro-apoptotic, whereas expression of longer isoforms (lacking catalytic residues) inhibits apoptosis (12). Together with the results presented here and elsewhere showing that caspases and caspase isoforms can be regulated spatially (depending on tissue type) and temporally (depending on stage of development), a picture emerges in which cells are allowed a high degree of control over the apoptotic process. Multiple caspases and caspase isoforms could potentially enable specialization in response to external stimuli in organ development and in cellular substrate cleavage, depending on local protein concentrations and subcellular localization. Multiple isoforms of the same caspase could also fine-tune caspase activity through hetero-oligomerization or differential expression, resulting in either an amplification or inhibition of the apoptotic response.
Acknowledgments-We thank Dr. V. Dixit for the 293T cell line and the plasmids encoding PARP and caspase-3/Yama. We are grateful to