The gamma subunit of the rod photoreceptor cGMP phosphodiesterase can modulate the proteolysis of two cGMP binding cGMP-specific phosphodiesterases (PDE6 and PDE5) by caspase-3

We have investigated whether the proteolysis of members of the cGMP binding phosphodiesterases (PDE6, PDE5A1, and PDE10A2) by caspase-3 is modulated by the gamma inhibitor subunit of PDE6. We show here that purified caspase-3 proteolyses PDE6, an enzyme composed of two nonidentical catalytic subunits (termed alpha and beta) with molecular mass of 88 and 84 kDa. The proteolysis of PDE6 produced a single fragment with a molecular mass of 78 kDa. This corresponds to the possible cleavage of the caspase-3 consensus DFVD site (amino acids: 164-168) in the alpha subunit and leads to a 50% decrease in the cGMP hydrolysing activity of the enzyme. The addition of rod PDEgamma to the incubation completely blocked the cleavage of PDE6 by caspase-3. In contrast, rod PDEgamma converted PDE5A1 (molecular mass of 98 kDa) to a better substrate for caspase-3. This resulted in the formation of four major fragments with molecular mass of 82-83, 67, 43, and 34 kDa. In addition, caspase-3 induced an approximately 80% reduction in the activity of a partially purified preparation of PDE5A1 in the presence of rod PDEgamma. Caspase-3 also cleaved PDE10A2 (molecular mass of 95 kDa) to a single 48-kDa fragment. This was consistent with cleavage of the DLFD site (amino acids: 312-315) in PDE10A2. In contrast with both PDE6 and PDE5A1, rod PDEgamma was without effect on this enzyme. These data show that rod PDEgamma interacts with at least two members of the cGMP binding PDE family (PDE5A1 and PDE6) and can exert differential effects on the cleavage of these enzymes by caspase-3.     

Ceramide and reactive oxygen species generated by H2O2 induce caspase-3-independent degradation of Akt/protein kinase B

This study was designed to elucidate the mechanisms leading to down-regulation of the Akt/protein kinase B (PKB) survival pathway during H2O2-induced cell death. H2O2 produced early activation of Akt/PKB and also DNA damage that was followed by stabilization of p53 levels, formation of reactive oxygen species (ROS), and generation of ceramide through activation of a glutathione-sensitive neutral sphingomyelinase. These events correlated with long term dephosphorylation and subsequent degradation of Akt. A membrane-targeted active Akt version attenuated apoptosis but not necrosis induced by H2O2 and was more resistant to dephosphorylation and proteolysis induced by apoptotic concentrations of H2O2. Proteolysis of Akt was prevented by exogenous addition of glutathione, indicating a role of ROS and ceramide in Akt degradation. However, Akt was degraded similarly in cells transfected with wild type and dominant negative p53 mutant, indicating that degradation of Akt under oxidative injury may be p53-independent. Specific inhibitors of caspase groups I and III prevented proteolysis of Akt/PKB and poly(ADP-ribose) polymerase in cells submitted to apoptotic but not necrotic H2O2 concentrations. Surprisingly, in caspase-3-deficient MCF-7 cells Akt was more sensitive to H2O2-induced degradation than the caspase-3 substrate poly(ADP-ribose) polymerase. Moreover, the Akt/PKB double mutant Akt(D108A,D119A), which is not cleaved by caspase-3, and a triple mutant (D453A,D455A,D456A), which lacks the consensus sequence for caspase-3 cleavage, were also degraded in H2O2-treated cells. Our results suggest that strong oxidants generate intracellular ROS and ceramide which in term lead to down-regulation of Akt by dephosphorylation and caspase-3-independent proteolysis.     

Purification and partial characterization of membrane-associated type II (cGMP-activatable) cyclic nucleotide phosphodiesterase from rabbit brain

Membrane-associated, Type II (cGMP-activatable) cyclic nucleotide phosphodiesterase (PDE) from rabbit brain, representing 75% of the total homogenate Type II PDE activity, was purified to apparent homogeneity. The enzyme was released from 13,000 x g particulate fractions by limited proteolysis with trypsin and fractionated using DE-52 anion-exchange, cGMP-Sepharose affinity and hydroxylapatite chromatographies. The enzyme showed 105 kDa subunits by SDS-PAGE and had a Stokes radius of 62.70 A as determined by gel filtration chromatography. Hydrolysis of cAMP or cGMP showed positive cooperativity, with cAMP kinetic behavior linearized in the presence of 2 microM cGMP. Substrate concentrations required for half maximum velocity were 28 microM for cAMP and 16 microM for cGMP. Maximum velocities were approx. 160 mumol/min per mg for both nucleotides. The apparent Kact for cGMP stimulation of cAMP hydrolysis at 5 microM substrate was 0.35 microM and maximal stimulation (3-5-fold) was achieved with 2 microM cGMP. Cyclic nucleotide hydrolysis was not enhanced by calcium/calmodulin. The purified enzyme can be labeled by cAMP-dependent protein kinase as demonstrated by the incorporation of 32P from [gamma-32P]ATP into the 105 kDa enzyme subunit. Initial experiments showed that phosphorylation of the enzyme did not significantly alter enzyme activity measured at 5 microM [3H]cAMP in the absence or presence of 2 microM cGMP or at 40 microM [3H]cGMP. Monoclonal antibodies produced against Type II PDE immunoprecipitate enzyme activity, 105 kDa protein and 32P-labeled enzyme. The 105 kDa protein was also photoaffinity labeled with [32P]cGMP. The purified Type II PDE described here is physicochemically very similar to the isozyme purified from the cytosolic fraction of several bovine tissues with the exception that it is predominantly a particulate enzyme. This difference may reflect an important regulatory mechanism governing the metabolism of cyclic nucleotides in the central nervous system.     

Direct photolabeling of the cGMP-stimulated cyclic nucleotide phosphodiesterase

cGMP-stimulated phosphodiesterase (PDE) has been directly photolabeled with [32P]cGMP using UV light. Sequence analysis of peptide fragments obtained from partial proteolysis or cyanogen bromide cleavage indicate that two different domains are labeled. One site, on a Mr = 36,000 chymotryptic fragment located near the COOH terminus, has characteristics consistent with it being close to or part of the catalytic site of the enzyme. This peptide contains a region of sequence that is highly conserved in all mammalian cyclic nucleotide PDEs and has been postulated to contain the catalytic domain of the enzyme. The other site, on a Mr = 28,000 cyanogen bromide cleavage fragment located near the middle of the molecule, probably makes up part of the allosteric site of the molecule. Labeling of the enzyme is concentration dependent and Scatchard analysis of labeling yields a biphasic plot with apparent half labeling concentrations of about 1 and 30 microM consistent with two types of sites being labeled. Limited proteolysis of the PDE by chymotrypsin yields five prominent fragments that separate by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) at Mr = 60,000, 57,000, 36,000, 21,000, and 17,000. Both the Mr = 60,000 and 57,000 apparently have blocked NH2 termini suggesting that the Mr = 57,000 fragment is a subfragment of the Mr = 60,000 fragment. Primary sequence analysis indicates that both the Mr = 21,000 and 17,000 fragments are subfragments of the Mr = 36,000 fragment. Autoradiographs of photolabeled then partially proteolyzed enzyme show labeled bands at Mr = 60,000, 57,000, and 36,000. Addition of 5 microM cAMP prior to photolabeling eliminates photolabeling of the Mr = 36,000 fragment but not the Mr = 60,000 or 57,000 fragments. The labeled site not blocked by cAMP is also contained in a Mr = 28,000 cyanogen bromide fragment of the enzyme that does not overlap with the Mr = 36,000 proteolytic fragment. Limited chymotryptic proteolysis also increases basal activity and eliminates cGMP stimulation of cAMP hydrolysis. The chymotryptic fragments can be separated by either ion exchange high performance liquid chromatography (HPLC) or solid-phase monoclonal antibody treatment. A solid-phase monoclonal antibody against the cGMP-stimulated PDE removes the Mr = 60,000 and 57,000 labeled fragments and any intact, unproteolyzed protein but does not remove the Mr = 36,000 fragment or the majority of activity. Ion exchange HPLC separates the fragments into three peaks (I, II, and III). Peaks I and II contain activity of approximately 40 and 100 units/mg, respectively. Peak II is the undigested or slightly nicked native enzyme.(ABSTRACT TRUNCATED AT 400 WORDS)     

In vivo differential prenylation of retinal cyclic GMP phosphodiesterase catalytic subunits.

A number of phototransducing proteins in vertebrate photoreceptors contain a carboxyl terminal -CXXX motif (where C = cysteine and X = any amino acid), known to be a signal sequence for their post-translational prenylation and carboxyl methylation. To study the roles of these modifications in the visual excitation process, we have utilized an intravitreal injection method to radiolabel the prenylated proteins of rat retinas in vivo. We showed that two of the major prenylated polypeptides in the rod outer segments are the PDE alpha and PDE beta subunits of cyclic GMP phosphodiesterase PDE alpha and PDE beta subunits of cyclic GMP phosphodiesterase (PDE). By chromatographic analyses of the amino acid constituents generated by exhaustive proteolysis of PDE alpha and PDE beta, we further demonstrated that they are differentially prenylated by farnesylation and geranylgeranylation, respectively. While a number of proteins ending with the -CXXX sequence have already been reported to possess either a farnesyl or a geranylgeranyl group, PDE is the first enzyme shown to be modified by both types of prenyl groups. The prenyl modification of PDE most likely plays a major role in membrane attachment and in correctly positioning the PDE molecule for phototransduction.     

Hydrolysis of cyclic GMP in rat peritoneal macrophages

Intact rat peritoneal macrophages (rPM) treated with 3-isobutyl-1-methylxanthine (IBMX), an inhibitor of phosphodiesterases (PDEs), accumulated more cGMP than untreated cells. A PDE activity toward [(3)H]cGMP was detected in the soluble and particulate fractions of rPM. The hydrolysis of cGMP was Ca(2+)/calmodulin-independent but increased in the presence of cGMP excess. Similar results were obtained when [(3)H]cAMP was used as a substrate. The hydrolytic activity towards both nucleotides was inhibited in the presence of IBMX. Therefore, the PDEs of families 2, 5, 10 and 11 are potential candidates for cGMP hydrolysis in the rPM. They may not only regulate the cGMP level in a feedback-controlled way but also link cGMP-dependent pathways with those regulated by cAMP.     

Nitric oxide-evoked transient kinetics of cyclic GMP in vascular smooth muscle cells

Cyclic-3',5'-guanosine monophosphate (cGMP) mediates the intracellular signaling cascade responsible for the nitric oxide (NO) initiated relaxation of vascular smooth muscle (VSM). However, the temporal dynamics, including the regulation of cGMP turnover, are largely unknown. Here we report new mechanistic insights into the kinetics of cGMP synthesis and hydrolysis in primary VSM cells by utilizing FRET-based cGMP-indicators [A. Honda, S.R. Adams, C.L. Sawyer, V. Lev-Ram, R.Y. Tsien, W.R. Dostmann, Proc. Natl. Acad. Sci. U S A 98 (5) (2001) 2437.]. First, 2-(N,N-Diethylamino)-diazenolate 2-oxide (DEA/NO) and 2,2'-(Hydroxynitrosohydrazono)-bis-ethanimine (DETA/NO) induced NO-concentration dependent, transient cGMP responses ("peaks") irrespective of their rates of NO release. The kinetic characteristics of these cGMP peaks were governed by the concerted action of the NO-sensitive guanylyl cyclase (GC) and phosphodiesterase type V (PDE5) as shown by their respective inhibition using 1H-[1,2,4]oxadiazolo[4,3-a]quinoxalin-1-one (ODQ) and Sildenafil. These responses occurred in the presence of moderately elevated cGMP (5-15% FRET ratio), and thus activated PKG and phosphorylated PDE5, suggesting a prominent role for GC in the maintenance and termination of cGMP peaks. Furthermore, cGMP transients could be elicited repeatedly without apparent desensitization of GC or by suppression of cGMP via long-term PDE5 activity. These results demonstrate a continuous sensitivity of the NO/cGMP signaling system, inherent to the phasic nature of smooth muscle physiology.     

Two steroidal saponins from Camassia cusickii induce L1210 cell death through the apoptotic mechanism.

Two steroidal saponins, tigogenin hexasaccharide-1 (TGHS-1, (25R)-5alpha-spirostan-3beta-yl 4-O-[2-0-[3-O-(alpha-L-rhamnopyranosyl)-beta-D-glucopyranosyl]-3-0-[4-0- (alpha-L-rhamnopyranosyl)-beta-D-glucopyranosyl]-beta-D-glucopyranosyl]-3-D- galactopyranoside) and tigogenin hexasaccharide-2 (TGHS-2, (25R)-5alpha-spirostan-3beta-yl 4-O-[2-0-[3-0-(beta-D-glucopyranosyl)-beta-D-glucopyranosyl]-3-0-[4-0- (alpha-L-rhamnopyranosyl)-beta-D-glucopyranosyl]-beta-D-glucopyranosyl]beta-D-galactopyranoside), were isolated from the fresh bulbs of Camassia cusickii. In murine leukemic L1210 cells, both compounds showed cytotoxicity with an EC50 value of 0.06 microM. The morphological observation revealed that TGHS-1 and TGHS-2 induced shrinkage in cell soma and chromatin condensation, suggesting apoptotic cell death. The cell death was confirmed to be apoptosis by Annexin V binding to phosphatidylserine in the cell membrane and excluding propidium iodide. A typical apoptotic DNA ladder and the cleavage of caspase-3 were observed after treatment with TGHS-1 and TGHS-2. In the presence of both the compounds, cells with sub-G1 DNA content were detected by flow cytometric analysis, indicating that TGHS-1 and TGHS-2 (each EC50 value of 0.1 microM) are the most powerful apoptotic saponins known. These results suggest that TGHS-1 and TGHS-2 induce apoptotic cell death through caspase-3 activation.     

Caspase-3 is required in the apoptotic disintegration of the nuclear matrix

Apoptotic breakdown of cellular structures is largely mediated by caspases. One target of degradation is a proteinaceous framework of the nucleus termed the nuclear matrix. We compared the apoptotic changes of the nuclear matrix in staurosporine-treated caspase-3-deficient MCF-7 cells transfected with intact CASP-3 gene (MCF-7c3) or an empty vector (MCF-7v) as a control. Nuclear Mitotic Apparatus protein (NuMA), lamin A/C and lamin B were used as markers for internal nuclear matrix and peripheral nuclear lamina, respectively. In both cell lines, staurosporine induced rapid cytoplasmic shrinkage and partial chromatin condensation. MCF-7c3 cells formed apoptotic bodies, whereas MCF-7v cells did not. NuMA and lamins were actively cleaved in MCF-7c3 cells following caspase-3 activation, but only minimal or no cleavage was detected in MCF-7v cells. Interestingly, lamin B but not lamin A/C was relocated into cytoplasmic granules in apoptotic MCF-7v cells. Pancaspase inhibitor, z-VAD-fmk, prevented the apoptotic changes, while caspase-3 inhibitor, z-DEVD-fmk, induced lamin B granules in both cell lines. These results show that caspase-3 is involved in the cleavage of NuMA and lamins either directly or by activating other proteases. This may be essential for disintegration of the nuclear structure during apoptosis.     

Caspase Cleavage of Keratin 18 and Reorganization of Intermediate Filaments during Epithelial Cell Apoptosis

Keratins 8 (K8) and 18 (K18) are major components of intermediate filaments (IFs) of simple epithelial cells and tumors derived from such cells. Structural cell changes during apoptosis are mediated by proteases of the caspase family. During apoptosis, K18 IFs reorganize into granular structures enriched for K18 phosphorylated on serine 53. K18, but not K8, generates a proteolytic fragment during drug- and UV light–induced apoptosis; this fragment comigrates with K18 cleaved in vitro by caspase-6, -3, and -7. K18 is cleaved by caspase-6 into NH₂-terminal, 26-kD and COOH-terminal, 22-kD fragments; caspase-3 and -7 additionally cleave the 22-kD fragment into a 19-kD fragment. The cleavage site common for the three caspases was the sequence VEVD/A, located in the conserved L1-2 linker region of K18. The additional site for caspases-3 and -7 that is not cleaved efficiently by caspase-6 is located in the COOH-terminal tail domain of K18. Expression of K18 with alanine instead of serine at position 53 demonstrated that cleavage during apoptosis does not require phosphorylation of serine 53. However, K18 with a glutamate instead of aspartate at position 238 was resistant to proteolysis during apoptosis. Furthermore, this cleavage site mutant appears to cause keratin filament reorganization in stably transfected clones. The identification of the L1-2 caspase cleavage site, and the conservation of the same or very similar sites in multiple other intermediate filament proteins, suggests that the processing of IFs during apoptosis may be initiated by a similar caspase cleavage.     

The MAP kinase ERK2 inhibits the cyclic AMP-specific phosphodiesterase HSPDE4D3 by phosphorylating it at Ser579

The extracellular receptor stimulated kinase ERK2 (p42(MAPK))-phosphorylated human cAMP-specific phosphodiesterase PDE4D3 at Ser579 and profoundly reduced ( approximately 75%) its activity. These effects could be reversed by the action of protein phosphatase PP1. The inhibitory state of PDE4D3, engendered by ERK2 phosphorylation, was mimicked by the Ser579-->Asp mutant form of PDE4D3. In COS1 cells transfected to express PDE4D3, challenge with epidermal growth factor (EGF) caused the phosphorylation and inhibition of PDE4D3. This effect was blocked by the MEK inhibitor PD98059 and was not apparent using the Ser579-->Ala mutant form of PDE4D3. Challenge of HEK293 and F442A cells with EGF led to the PD98059-ablatable inhibition of endogenous PDE4D3 and PDE4D5 activities. EGF challenge of COS1 cells transfected to express PDE4D3 increased cAMP levels through a process ablated by PD98059. The activity of the Ser579-->Asp mutant form of PDE4D3 was increased by PKA phosphorylation. The transient form of the EGF-induced inhibition of PDE4D3 is thus suggested to be due to feedback regulation by PKA causing the ablation of the ERK2-induced inhibition of PDE4D3. We identify a novel means of cross-talk between the cAMP and ERK signalling pathways whereby cell stimuli that lead to ERK2 activation may modulate cAMP signalling.     

Control of the cyclic GMP phosphodiesterase of frog photoreceptor membranes

The light-activated cyclic GMP phosphodiesterase (PDE) of frog photoreceptor membranes has been assayed in isolated outer segments suspended in a low-calcium Ringer's solution. Activation occurs over a range of light intensity that also causes a decrease in the permeability, cyclic GMP levels, and GTP levels of isolated outer segments. At intermediate intensities, PDE activity assumes constant intermediate values determined by the rate of rhodopsin bleaching. Washing causes an increase in maximal enzyme activity. Increasing light intensity from darkness to a level bleaching 5 x 10(3) rhodopsin molecules per outer segment per second shifts the apparent Michaelis constant (Km) from 100 to 900 microM. Maximum enzyme velocity increases at least 10-fold. The component that normally regulates this light- induced increase in the Km of PDE is removed by the customary sucrose flotation procedures. The presence of 10(-3) M Ca++ increases the light sensitivity of PDE, and maximal activation is caused by illumination bleaching only 5 x 10(2) rhodopsin molecules per outer segment per second. Calcium acts by increasing enzyme velocity while having little influence on Km. The effect of calcium appears to require a labile component, sensitive to aging of the outer segment preparation. The decrease in the light sensitivity of PDE that can be observed upon lowering the calcium concentration may be related to the desensitization of the permeability change mechanism that occurs during light adaptation of rod photoreceptors.     

Peptide mapping of multiple forms of cyclic nucleotide phosphodiesterase

Purified multiple forms of 3':5'-cyclic-nucleotide phosphodiesterase (EC 3.1.4.17) were analyzed using two-dimensional tryptic pep]tide mapping of radioiodinated peptides. Comparisons of peptide maps of rat liver insulin-sensitive phosphodiesterase (PDE) with rat brain calmodulin-sensitive PDE suggest that some peptides co-migrate (31-43% co-migration). However, except for a single peptide, bovine retinal rod outer segment PDE, peptide maps appear unrelated to the other two forms (7-12% co-migration). In contrast, peptide maps of a 36,000-dalton proteolysis product of calmodulin-sensitive PDE are highly related to the peptide maps of a rat brain calmodulin-sensitive holoenzyme (73% co-migration). These results suggest that the multiple PDE forms are distinct molecular entities.     

Regulation of lens connexin 45.6 by apoptotic protease,caspase-3

Gap junctions are important in maintaining lens homeostasis. Here we report that connexin 45.6 (Cx45.6) was partially truncated to a 46 kDa fragment during chicken lens development. This specific truncation initiated during embryonic days and the truncated fragment accumulated towards the later developmental stages. When membranes of the embryonic lens were subjected to caspase-3 treatment, the 46 kDa fragment of Cx45.6 was reproduced, suggesting apoptotic protease caspase-3 is a potential protease involved. The COOH-terminus of Cx45.6 in GST-fusion protein was also cleaved by caspase-3, confirming that Cx45.6 is a direct substrate of caspase-3. Induction of apoptosis in lens primary cultures regenerated the 46 kDa fragment and this cleavage was blocked by a caspase-3 inhibitor. Alteration of amino acid residue Asp364 or Glu367 to Ala prevented Cx45.6 from cleavage by caspase-3, suggesting that the cleavage site of Cx45.6 is likely to be between Glu367 and Gly361. Phosphorylation of Ser363, a known substrate for casein kinase II (CKII) in vivo, inhibited the cleavage of Cx45.6 by caspase-3. Thus, this study demonstrates that a lens connexin can be a direct target of caspase-3 and the cleavage by caspase-3 leads to the development-associated truncation of Cx45.6. Finally, caspase-3 mediated truncation can be modulated by the specific connexin phosphorylation.     

Cleavage at the caspase-6 site is required for neuronal dysfunction and degeneration due to mutant huntingtin

Cleavage of huntingtin (htt) has been characterized in vitro, and accumulation of caspase cleavage fragments represents an early pathological change in brains of Huntington's disease (HD) patients. However, the relationship between htt proteolysis and the pathogenesis of HD is unknown. To determine whether caspase cleavage of htt is a key event in the neuronal dysfunction and selective neurodegeneration in HD, we generated YAC mice expressing caspase-3- and caspase-6-resistant mutant htt. Mice expressing mutant htt, resistant to cleavage by caspase-6 but not caspase-3, maintain normal neuronal function and do not develop striatal neurodegeneration. Furthermore, caspase-6-resistant mutant htt mice are protected against neurotoxicity induced by multiple stressors including NMDA, quinolinic acid (QA), and staurosporine. These results are consistent with proteolysis of htt at the caspase-6 cleavage site being an important event in mediating neuronal dysfunction and neurodegeneration and highlight the significant role of htt proteolysis and excitotoxicity in HD.     

Caspase-3 mediated cleavage of HsRad51 at an unconventional site.

The human recombinase HsRad51 is cleaved during apoptosis. We have earlier observed cleavage of the 41-kDa full-length protein into a 33-kDa product in apoptotic Jurkat cells and in in vitro translated HsRad51 after treatment with activated S-100 extract. In this study, site-directed mutagenesis was used for mapping of the cleavage site to AQVD274 downward arrow G, which does not correspond to a conventional caspase cleavage site. The absence of HsRad51 cleavage in staurosporine-treated apoptotic MCF-7 cells, which lack caspase-3, indicates that caspase-3 is essential for HsRad51 cleavage in vivo. Cleavage into the 33-kDa fragment was generated by recombinant caspase-3 and -7 in in vitro translated wild type HsRad51, but not in the HsRad51 AQVE274 downward arrow G mutant. Similarly, HsRad51 of Jurkat cell extracts was cleaved into the 33-kDa product by recombinant caspase-3, whereas caspase-7 failed to cleave endogenous HsRad51. The cleavage of in vitro translated wild type and AQVE274 downward arrow G mutant HsRad51 as well as of endogenous HsRad51 also gave rise to a smaller fragment, which corresponds in size to a recently reported DVLD187 downward arrow N HsRad51 cleavage product. In Jurkat cell extracts, the AQVD274 downward arrow G and DVLD187 downward arrow N cleavage products of HsRad51 appeared at equal concentrations of caspase-3. Moreover both fragments were generated by induction of apoptosis in MDA-MB 157 cells with staurosporine and in Jurkat cells with camptothecin. Thus, two sites in the HsRad51 sequence are targets for caspase cleavage both in vitro and in vivo.     

Caspase-3-dependent cleavage of Bcl-2 promotes release of cytochrome c.

Caspases are cysteine proteases that mediate apoptosis by proteolysis of specific substrates. Although many caspase substrates have been identified, for most substrates the physiologic caspase(s) required for cleavage is unknown. The Bcl-2 protein, which inhibits apoptosis, is cleaved at Asp-34 by caspases during apoptosis and by recombinant caspase-3 in vitro. In the present study, we show that endogenous caspase-3 is a physiologic caspase for Bcl-2. Apoptotic extracts from 293 cells cleave Bcl-2 but not Bax, even though Bax is cleaved to an 18-kDa fragment in SK-NSH cells treated with ionizing radiation. In contrast to Bcl-2, cleavage of Bax was only partially blocked by caspase inhibitors. Inhibitor profiles indicate that Bax may be cleaved by more than one type of noncaspase protease. Immunodepletion of caspase-3 from 293 extracts abolished cleavage of Bcl-2 and caspase-7, whereas immunodepletion of caspase-7 had no effect on Bcl-2 cleavage. Furthermore, MCF-7 cells, which lack caspase-3 expression, do not cleave Bcl-2 following staurosporine-induced cell death. However, transient transfection of caspase-3 into MCF-7 cells restores Bcl-2 cleavage after staurosporine treatment. These results demonstrate that in these models of apoptosis, specific cleavage of Bcl-2 requires activation of caspase-3. When the pro-apoptotic caspase cleavage fragment of Bcl-2 is transfected into baby hamster kidney cells, it localizes to mitochondria and causes the release of cytochrome c into the cytosol. Therefore, caspase-3-dependent cleavage of Bcl-2 appears to promote further caspase activation as part of a positive feedback loop for executing the cell.