Caspase-3 Is the Primary Activator of Apoptotic DNA Fragmentation via DNA Fragmentation Factor-45/Inhibitor of Caspase-activated DNase Inactivation*

Caspase-3 initiates apoptotic DNA fragmentation by proteolytically inactivating DFF45 (DNA fragmentation factor-45)/ICAD (inhibitor of caspase-activated DNase), which releases active DFF40/CAD (caspase-activated DNase), the inhibitor's associated endonuclease. Here, we examined whether other apoptotic proteinases initiated DNA fragmentation via DFF45/ICAD inactivation. In a cell-free assay, caspases-3, -6, -7, -8, and granzyme B initiated benzoyloxycarbonyl-Asp-Glu-Val-Asp (DEVD) cleaving caspase activity, DFF45/ICAD inactivation, and DNA fragmentation, but calpain and cathepsin D failed to initiate these events. Strikingly, only the DEVD cleaving caspases, caspase-3 and caspase-7, inactivated DFF45/ICAD and promoted DNA fragmentation in an in vitro DFF40/CAD assay, suggesting that granzyme B, caspase-6, and caspase-8 promote DFF45/ICAD inactivation and DNA fragmentation indirectly by activating caspase-3 and/or caspase-7. In vitro, however, caspase-3 inactivated DFF45/ICAD and promoted DNA fragmentation more effectively than caspase-7 and endogenous levels of caspase-7 failed to inactivate DFF45/ICAD in caspase-3 null MCF7 cells and extracts. Together, these data suggest that caspase-3 is the primary inactivator of DFF45/ICAD and therefore the primary activator of apoptotic DNA fragmentation.

Caspase proteinases drive apoptotic signaling and execution by cleaving critical cellular proteins solely after aspartate residues (reviewed in Ref. 1). Caspases exist as latent zymogens, but apoptotic death stimuli activate the initiator caspases, caspase-8 and caspase-9. Death receptors such as Fas induce caspase-8 activation via the adapter molecule Fas-associated death domain protein. Chemotherapeutic agents and UV irradiation cause release of mitochondrial cytochrome c, which then binds the adapter molecule apoptotic proteinase activating factor-1 (APAF-1), 1 and this complex along with adenine nucleotides promotes caspase-9 autoactivation. Once activated, initiator caspases in turn activate the executioner caspases, caspases-3, -6, and -7. The active executioners promote apoptosis by cleaving cellular substrates that induce the morphological and biochemical features of apoptosis (1).
Besides caspase-3, caspases-6, -7, -8, -9, the cytotoxic T cell proteinase granzyme B, the calcium-dependent proteinase calpain, and the lysosomal proteinase cathepsin D may function during apoptosis (9 -12). These proteinases have all been suggested to participate in apoptotic DNA fragmentation; however, the mechanism(s) by which they promote DNA fragmentation remain unclear. Here, we examined whether these proteinases induced DNA fragmentation by inactivating DFF45/ICAD. We find that caspase-3 and caspase-7 are the only direct inactivators of DFF45/ICAD. However, endogenous levels of caspase-7 failed to inactivate the inhibitor or promote DNA fragmentation in vitro or in intact cells, suggesting that caspase-3 is the primary regulator of apoptotic DNA fragmentation via proteolysis of DFF45/ICAD.
Drs. Guy Salvesen (The Burnham Institute) and Vishva Dixit (Genentech) provided plasmids encoding histidine-tagged caspases-3, -6, -7, and -8 (9,14). A construct encoding 6ϫ-histidine tagged caspase-9 was produced using the polymerase chain reaction and a topoisomerase-based cloning system (Invitrogen). Briefly, the coding region of caspase-9 was amplified from the pRSV-LacZ-caspase-9 plasmid (15) (Dr. Emad Alnemri, Thomas Jefferson University). The polymerase chain reaction product was subjected to agarose gel electrophoresis and purified by gel extraction (Qiagen). The gel-purified product was then cloned into the pTrcHis2TOPO vector according to the manufacturer's instructions. The fidelity of the construct sequence was confirmed by nucleotide sequencing.
Cell-free Reactions-Cytosolic extracts from Jurkat cells and MCF7 cells were prepared as described (13,17). Cell-free reactions were conducted in the following buffers (as indicated): CAD buffer (10 mM HEPES-KOH, pH 7.2, 50 mM NaCl, 20% v/v glycerol, 2 mM MgCl 2 , 5 mM EGTA, 5 mM dithiothreitol, 1 mg/ml bovine serum albumin), calpain buffer (CAD buffer lacking EGTA and containing 2.5 mM calcium chloride and 0.1 mg/ml bovine serum albumin), and cathepsin D buffer (10 mM PIPES-KOH, pH 6.7, 50 mM NaCl, 20% (v/v) glycerol, 2 mM MgCl 2 , 5 mM dithiothreitol, 1 mg/ml bovine serum albumin). Since preliminary experiments demonstrated that calcium present in calpain buffer promoted DNA fragmentation even without calpain addition, we performed cell-free reactions in two steps. First, extracts were treated with proteinases or cytochrome c plus dATP as indicated in the figure legends. Aliquots of each reaction mixture were then withdrawn for caspase assay and Western blot analysis (13). Second, 10 6 rat liver nuclei and 5 mM EGTA were added to the extracts and the reactions then continued for the indicated times. DNA fragmentation was subsequently visualized via agarose gel electrophoresis and ethidium bromide staining (17). For the kinetic analysis described in the legend to Fig. 4, cell-free reactions were performed by mixing cytosolic extract, nuclei, and cytochrome c plus dATP in a single step as described in the figure legend.
In Vitro Transcription and Translation-35 S-DFF45/ICAD was prepared from pBS-ICAD L using [ 35 S]methionine (Amersham Pharmacia Biotech) and a reticulocyte lysate transcription and translation system (Promega). Since functional DFF40/CAD can only be produced in the presence of DFF45/ICAD (3, 7), we included 200 ng of GST-DFF45/ ICAD in the transcription/translation reaction to prepare 35 S-pro-DFF40/CAD. Aliquots (2 l) of 35 S-DFF45/ICAD and 35 S-pro-DFF40/ CAD were incubated with proteinases in caspase buffer (20 mM PIPES, 100 mM NaCl, 10 mM dithiothreitol, 1 mM EDTA, 0.1% CHAPS, 10% sucrose, pH 7.2) as indicated in the figure legends and the products subjected to SDS-PAGE and autoradiography.
In Vitro DFF40/CAD Assay-In vitro DFF40/CAD assays were performed as described (3). Briefly, pro-DFF40/CAD was prepared by in vitro transcription and translation in the presence of 200 ng of GST-DFF45/ICAD. Aliquots (2 l) of pro-DFF40/CAD were incubated with proteinases for 30 min at 30°C. 10 6 nuclei and EGTA (final concentration 5 mM) were added to the reaction mixture and after 2 h DNA fragmentation was analyzed via agarose gel electrophoresis and ethidium bromide staining (17). In control experiments, we determined that the reticulocyte lysate used for transcription and translation did not have DNase activity either in the absence or presence of proteinases or cytochrome c plus dATP.

Caspases, Granzyme B, and Cytochrome C plus dATP Promote DFF45/ICAD Cleavage and DNA Fragmentation in a
Cell-free Assay-We first assessed whether various apoptotic proteinases could promote DNA fragmentation in a cell-free assay. Proteinases were incubated with cytosolic extracts from Jurkat T cells and subsequently examined for DNase activity using rat liver nuclei as a substrate. We also examined whether the proteinase-treated extracts cleaved DFF45/ICAD and the fluorogenic caspase substrate DEVD-AFC.
As shown in Fig. 1, all proteinases examined promoted DFF45/ICAD cleavage as demonstrated by Western blotting. However, not all cleavage events inactivated the inhibitor. Caspases-3, -6, and -7 produced DFF45/ICAD cleavage products of ϳ23 and ϳ12 kDa. Note that the ϳ23-kDa band likely represents intermediate cleavage products resulting from cleavage at one of the two caspase cleavage sites; whereas, the broad band at ϳ12 kDa probably represents a composite of the three DFF45/ICAD fragments observed when the inhibitor is cleaved at both caspase sites (2).
The control lane shows a small amount of the intermediate cleavage products, likely due to background caspase activity. With granzyme B and cytochrome c plus dATP, only the ϳ12-kDa fragments were observed, suggesting cleavage had occurred at both caspase cleavage sites. By contrast, calpain prompted loss of DFF45/ICAD immunoreactivity and cathepsin D produced a ϳ19-kDa DFF45/ICAD fragment. Strikingly, only the caspases, granzyme B, and cytochrome c plus dATP elicited DEVD cleaving caspase activity (Fig. 1A), DFF45/ICAD inactivation, and DNA fragmentation (Fig. 1C). As in apoptotic cells (2), DNA fragmentation correlated with the production of ϳ12-kDa DFF45/ICAD fragments. Although calpain and cathepsin D induced DFF45/ICAD cleavage, they failed to inactivate the inhibitor, induce DNA fragmentation, or activate DEVD cleaving caspases. This suggests that DEVD cleaving caspase activity is not necessary for DFF45/ICAD cleavage, but is required for inactivation of the inhibitor and DNA fragmentation.
We also examined whether the initiator caspases functioned in the cell-free assay. Caspase-8 (50 nM) induced DEVDase activity, DFF45/ICAD cleavage, and DNA fragmentation (not shown). Caspase-9 failed to promote these events; however, our caspase-9 preparation showed little activity against the fluorogenic caspase-9 substrate LEHD-AFC. This is probably due to lack of cofactors since caspase-9 demonstrates little activity in the absence of cytochrome c, dATP, and cyotsolic factors (APAF-1) (18).
Caspase-3 and Caspase-7, but Not Other Apoptotic Proteinases, Promote DFF45/ICAD Inactivation and DNA Fragmentation in Vitro-We next asked whether the preceding apoptotic proteinases directly cleaved DFF45/ICAD or DFF40/CAD that was bound to the inhibitor. To accomplish this, we treated 35 S-DFF45/ICAD and 35 S-pro-DFF40/CAD with each proteinase and examined the products by SDS-PAGE and autoradiography. As shown in Fig. 2A, DFF45/ICAD was susceptible to each proteinase; however, the extent and pattern of DFF45/ ICAD cleavage varied with each proteinase.
To determine whether DFF45/ICAD cleavage inactivated the inhibitor and released active DFF40/CAD, we treated pro-DFF40/CAD (prepared by in vitro transcription and translation in the presence of GST-DFF45/ICAD) with each proteinase and then examined DNase activity against rat liver nuclei. Strikingly, although each proteinase cleaved DFF45/ICAD, only caspases-3, and -7 inactivated the inhibitor, prompting release of active DFF40/CAD and DNA fragmentation (Fig. 2C). The initiator caspase, caspases-8, also failed to inactivate DFF45/ ICAD, even at concentrations up to 200 nM (not shown). DFF45/ ICAD inactivation and DNA fragmentation correlated with production of 24-, 22-, and 12-kDa DFF45/ICAD fragments that were only observed with caspase-3 and caspase-7 or when 35 S-DFF45/ICAD was incubated with cytochrome c plus dATPtreated cytosol (Fig. 2A). Although the other proteinases cleaved DFF45/ICAD, they failed to inactivate the inhibitor. Since granzyme B, caspase-6, and caspase-8 promoted DFF45/ ICAD inactivation and DNA fragmentation in the cell-free assay (Fig. 1), they probably function in this system via activation of caspases-3 and/or -7.
Further Analysis of the Reaction of Caspase-3 and Caspase-7 with DFF45/ICAD and Pro-DFF40/CAD-We next examined the reaction of caspase-3 and caspase-7 with DFF45/ICAD in detail since these proteinases were the only direct inactivators of the inhibitor. As shown in Fig. 3A, when equal amounts (40 nM) of caspase-3 or caspase-7 were reacted with 35 S-DFF45/ICAD, caspase-3 cleaved the inhibitor more rapidly than caspase-7. After 30 min at 37°C, aliquots (40 g of protein) of each reaction mixture were analyzed for DEVD cleaving caspase activity (A) and subjected to Western blotting (12% gels) with anti-DFF45/ICAD antibodies (B). 10 6 rat liver nuclei and EGTA (final concentration 5 mM) were added to each reaction and the samples then incubated for an additional 2.5 h at 37°C. DNA fragmentation was then assessed by agarose gel electrophoresis and ethidium bromide staining (C). Unless indicated otherwise, reactions were conducted in CAD buffer. Note that DFF45/ICAD has a predicted molecular mass of 36.5 kDa based on amino acid sequence (2); however, the protein and its alternatively spliced short isoform run as bands of ϳ45 and ϳ38 kDa, respectively (2). Cleavage at the N-terminal caspase site gives predicted products of ϳ23.7 and 12.8 kDa; whereas, cleavage at the C-terminal caspase cleavage site yields products of ϳ24.8 and 11.7 kDa. Hydrolysis at both caspase sites therefore yields fragments of ϳ12.8, 12.0, and 11.7 kDa. In B, L and S indicate the long and short DFF45 isoforms, I represents the intermediate caspase cleavage products (ϳ23.7, 24.8 kDa), and the asterisk (*) denotes the ϳ12-kDa DFF45 fragments that correspond with DNA fragmentation. In A, error bars represent the S.E., n ϭ 3. In B and C, the data are representative of three independent experiments. 35  DFF45/ICAD cleavage was evident within 10 min following caspase-3 treatment and complete by 2 h. By contrast, caspase-7 did not produce detectable DFF45/ICAD cleavage until 2 h and only ϳ50% of the DFF45/ICAD was cleaved at 4 h. In concentration dependence experiments (Fig. 3B), caspase-3 also cleaved DFF45/ICAD more efficiently than caspase-7. Following a 30min incubation, DFF45/ICAD cleavage was detectable with as little as 8.8 nM caspase-3; however, DFF45/ICAD cleavage by caspase-7 was not detectable until the caspase concentration was 88 nM.

FIG. 2. Limited proteolysis of DFF45/ICAD by caspase-3 and caspase-7, but not other apoptotic proteinases, releases active DFF40/CAD in vitro.
To determine if DFF45/ICAD cleavage corresponded with release of active DFF40/CAD, we incubated various concentrations of caspase-3 or caspase-7 with pro-DFF40/CAD and rat liver nuclei and then monitored DNA fragmentation. As shown in Fig. 3C, caspase-3 induced detectable DNA fragmentation at 50 nM, whereas with caspase-7, DNA fragmentation required 150 nM caspase-7. Together, the data indicate that caspase-3 inactivates DFF45/ICAD and induces DNA fragmentation more efficiently than caspase-7.
Caspase-7 Activation Does Not Induce DNA Fragmentation in Caspase-3 Null MCF7 Cells or Cytosolic Extracts-To determine whether endogenous levels of caspase-7 could cleave DFF45/ICAD and promote DNA fragmentation in the absence of caspase-3, we used caspase-3 null MCF7 cell extracts in a cell-free assay. The MCF7 cells lacked caspase-3 as demonstrated by Western blotting, although they contained caspases-6, -7, -8, -9, and APAF-1 at comparable levels to Jurkat cells (not shown). We treated the extracts with cytochrome c plus dATP and then examined DEVDase activity, procaspase-7 processing, DFF45/ICAD cleavage, and DNA fragmentation as a function of time. For comparison, we also analyzed these events in Jurkat cell extracts, which contain caspase-3.
As shown in Fig. 4, cytochrome c plus dATP initiated procaspase-7 processing and the onset of DEVD cleaving caspase activity in Jurkat and MCF7 extracts. However, procaspase-7 processing occurred earlier and was more extensive in the Jurkat extracts. Procaspase-3 processing occurred with identical kinetics to procaspase-7 processing in the Jurkat extracts (not shown). The Jurkat extracts demonstrated ϳ10-fold greater DEVDase activity than the MCF7 cells at 2 h, likely due to the combined activity of caspase-3 and caspase-7. Al-though DEVDase activity declined with time in the Jurkat extracts, the Jurkat DEVDase activity remained substantially greater than the MCF7 DEVDase activity at all time points. Caspase-7 processing and DEVDase activity correlated with the extent of DFF45/ICAD cleavage and DNA fragmentation in both extracts (Fig. 4, B and C). In the Jurkat extracts, DFF45/ ICAD cleavage was complete by 2 h; whereas, even after 6 h, only ϳ50% of DFF45/ICAD had been cleaved in the MCF7 extracts. Strikingly, while DNA fragmentation had occurred by 2 h in the Jurkat extracts, little DNA fragmentation had occurred in the MCF7 extracts even after 6 h of incubation, despite caspase-7 activation and partial DFF45/ICAD cleavage. Thus, cytochrome c plus dATP initiate caspase-7 activation in the MCF7 extracts, but this does not inactivate DFF45/ICAD or promote significant DNA fragmentation. Similarly, we detected caspase-7-like DEVDase activity in extracts prepared from apoptotic MCF7 cells, but we detected no DNA fragmentation by agarose gel analysis or propidium iodide staining and fluorescence-activated cell sorter analysis (not shown).
We next incubated caspase-3 with MCF7 extracts and nuclei to determine whether this proteinase could initiate DNA fragmentation. Fig. 4D shows that 125 nM caspase-3 induced DNA fragmentation in the extracts. Similarly, addition of exogenous caspase-7 to the extracts initiated DNA fragmentation, although this required a higher concentration than caspase-3 (not shown). Together, these data suggest that in the absence of caspase-3, endogenous levels of caspase-7 do not inactivate DFF45/ICAD.

DISCUSSION
In this paper, we examined whether various apoptotic proteinases could promote internucleosomal DNA fragmentation by inactivating DFF45/ICAD, thereby releasing active DFF40/ CAD. Of the eight proteinases examined, we find that only caspase-3 and caspase-7 are direct inactivators of DFF45/ ICAD. However, caspase-3 was the more efficient inactivator and endogenous levels of caspase-7 failed to release active DFF40/CAD, suggesting that caspase-3 is the primary activator of apoptotic DNA fragmentation. These findings confirm and extend those of Liu et al. (19) who recently demonstrated that caspases-3 and -7, but not caspases-6 and -8, inactivate recombinant DFF45, releasing active DFF40 (19). Additionally,

FIG. 3.
Caspase-3 inactivates DFF45/ICAD and activates DNA fragmentation more efficiently than caspase-7. In A, 35 S-DFF45/ICAD (2 l) was incubated with caspase-3 (40 nM) or caspase-7 (40 nM) for the indicated times at 37°C. Reactions were terminated by denaturing samples in Laemmli sample buffer and the products subjected to SDS-PAGE (12% slabs) and autoradiography. In B, 35 S-DFF45/ICAD (2 l) was incubated with the indicated concentrations of caspase-3 and caspase-7 for 30 min at 37°C. Samples were then denatured and analyzed by SDS-PAGE (12% slabs) and autoradiography. In C, pro-DFF40/CAD was incubated with 10 6 rat liver nuclei for 2.5 h at 30°C and the indicated concentrations of caspase-3 and caspase-7. DNA fragmentation was then assessed by agarose gel electrophoresis and ethidium bromide staining. The data are representative of three independent experiments. our work emphasizes the central importance of caspase proteinases, particularly caspase-3, as the principal mediators of apoptosis.
Three lines of evidence support our finding that caspase-3 is the primary inactivator of DFF45/ICAD and thus the primary activator of apoptotic DNA fragmentation. First, only proteinases that initiated DEVDase activity in Jurkat extracts, which is due primarily to caspase-3 inactivated DFF45/ICAD and initiated DNA fragmentation (Fig. 1). Second, caspase-3 cleaved DFF45/ICAD and promoted DNA fragmentation more effectively than caspase-7, the only other direct DFF45/ICAD inactivator (Figs. 2 and 3). Third, caspase-3 null MCF7 cells failed to fragment DNA during apoptosis and cytochrome c plus dATP-activated MCF7 extracts did not inactivate DFF45/ICAD or promote DNA fragmentation, despite caspase-7 activation (Fig. 4). However, addition of caspase-3 to these extracts restored their caspase-dependent DNase activity. These findings are consistent with the delayed or absent apoptotic DNA fragmentation observed in caspase-3 null cells and mice (20 -23). Caspase-3 activation is therefore fundamentally important for DFF45/ICAD inactivation.
Although caspase-7 initiated DNA fragmentation in a cellfree assay (Fig. 1) and inactivated DFF45/ICAD in an in vitro nuclease assay (Fig. 2), our data suggest that caspase-7 plays a secondary role in inactivating the inhibitor. Like caspase-3, caspase-7 prefers DEVD-based peptide substrates (1), however, caspase-3 cleaved DFF45/ICAD more effectively than caspase-7 (Fig. 3) in vitro and the low level of caspase-7 DEV-Dase activity generated in MCF7 cells and extracts (Fig. 4) did not inactivate DFF45/ICAD. This suggests that the two caspases cleave macromolecular substrates with different effi- FIG. 5. A model of apoptotic DNA fragmentation. The model emphasizes release of active DFF40/CAD via caspase-3 dependent inactivation of DFF45/ICAD and highlights how other apoptotic proteinases may intersect this pathway. Note that the initiator caspases and granzyme B function primarily via activation of executioner caspases. Active caspase-3 is the primary inactivator of DFF45/ICAD, but caspase-7, which may be localized to mitochondria, plays a secondary role in DFF45/ICAD inactivation. Caspase-3 in turn activates caspase-6, which aids in nuclear apoptosis via cleavage of nuclear lamins. Caspase-6 can also amplify caspase-3 activity via activation of procaspase-3. Cathepsin D and calpain apparently function independently of caspases and DFF40/CAD, although calpain could potentially initiate high molecular weight DNA fragmentation via release of mitochondrial apoptosis inducing factor.
FIG. 4. Endogenous levels of caspase-7 cleave DFF45/ICAD inefficiently and fail to promote DNA fragmentation. Cytosolic extracts were prepared from Jurkat cells and caspase-3 null MCF7 cells as described in the text. Extracts (final concentration 12 mg of protein/ ml) were incubated at 37°C with buffer or cytochrome c (10 M) plus dATP (1 mM) and 10 6 rat liver nuclei. At the indicated times, the nuclei were pelleted by centrifugation and analyzed for DNA fragmentation (C). Aliquots of the supernatant (40 g protein) were analyzed for DEVD cleaving caspase activity (A) and subjected to Western blotting with antibodies against caspase-7, DFF45/ICAD, and actin (B). In D, MCF7 extracts were incubated with 10 6 rat liver nuclei and the indicated concentrations of caspase-3. After 2.5 h at 37°C, DNA fragmentation was assessed by agarose gel electrophoresis and ethidium bromide staining. In A, error bars represent the S.E., n ϭ 3. In B, C, and D, the data are representative of three independent experiments. ciencies and that endogenous levels of caspase-7 do not initiate DNA fragmentation. Furthermore, since active caspase-7 localizes to mitochondrial and microsomal membranes (24), the caspase may be physically sequestered from DFF45/ICAD and therefore unable to react with the inhibitor. Overall, the data suggest that caspase-7 is not a physiologic inactivator of DFF45/ICAD and that critical caspase-7 substrates remain unidentified.
Caspase-6 and granzyme B induced DFF45/ICAD inactivation and DNA fragmentation in Jurkat extracts (Fig. 1); however, although they cleaved DFF45/ICAD, they failed to inactivate the inhibitor in an in vitro nuclease assay (Fig. 2). Since these proteinases effectively activate caspase-3 (16,25), these results suggest that they inactivate DFF45/ICAD via caspase-3 activation. Non-activating DFF45/ICAD cleavage was also observed with calpain and cathepsin D (Figs. 1 and 2); however, since these proteinases do not activate caspases ( Fig. 1 and Ref. 13), they were unable to initiate DNA fragmentation in a cellfree assay (Fig. 1). Caspase-6, granzyme B, calpain, and cathepsin D probably cleave DFF45/ICAD at or near the C-terminal caspase cleavage site (Asp 224 ) since DFF45/ICAD mutants that can only be cleaved at this site retain DFF40/ CAD inhibitory activity (6). Non-activating cleavage of DFF45/ ICAD by these proteinases could be synergistic with activating cleavage events mediated by caspase-3 or caspase-7 if the partially cleaved inhibitor is more susceptible to cleavage by caspases-3 and -7. Alternatively, non-activating cleavage might alter interaction of DFF40/CAD with cofactors such as histone H1 (19) or target the endonuclease for destruction. These possibilities are currently under investigation.
In summary, our data indicate that caspase-3 is the primary inactivator of DFF45/ICAD and suggest that proteolytic pathways that induce apoptotic internucleosomal DNA fragmentation must involve this proteinase. Fig. 5 summarizes our findings and presents a model of apoptotic DNA fragmentation. In this model, the initiator caspases, caspases-8 and -9, promote apoptotic signaling and activate the executioner caspases, which in turn degrade apoptotic substrates. Similarly, granzyme B functions primarily to activate executioner caspases and not to cleave apoptotic substrates. Studies demonstrating that granzyme B activates caspase-3 more efficiently than caspase-8 (16) and the lack of DNA fragmentation observed in the absence of caspase activation during granzyme B-mediated apoptosis (9) support this hypothesis. Once activated, caspase-3 plays a central role in apoptotic DNA fragmentation by inactivating DFF45/ICAD, thereby releasing active DFF40/ CAD. Caspase-3 also activates caspase-6 (26), which in turn promotes nuclear apoptosis by degrading lamins (27). Caspase-6 can also activate procaspase-3 (25), providing the opportunity for amplification of caspase-3 activity. Caspase-7, which may be localized to mitochondria and microsomes plays a secondary role in DNA fragmentation as a back up DFF45/ ICAD inactivator. By contrast, calpain and cathepsin D do not activate caspases and function downstream or independently of caspase cascades. However, calpain, which localizes to mitochondria during apoptosis (28), could potentially induce high molecular weight DNA fragmentation via release of mitochondrial apoptosis inducing factor. Although lysosomes release cathepsin D during some forms of apoptosis (29), this proteinase does not inactivate DFF45/ICAD and its cellular substrates remain undefined. Thus, many proteinases may function in nuclear apoptosis, but caspase-3 is the key determinant of DFF45/ICAD inactivation and apoptotic internucleosomal DNA fragmentation.