Selective, Reversible Caspase-3 Inhibitor Is Neuroprotective and Reveals Distinct Pathways of Cell Death after Neonatal Hypoxic-ischemic Brain Injury*

Hypoxia-ischemia (H-I) in the developing brain results in brain injury with prominent features of both apoptosis and necrosis. A peptide-based pan-caspase inhibitor is neuroprotective against neonatal H-I brain injury, suggesting a central role of caspases in brain injury. Because previously studied peptide-based caspase inhibitors are not potent and are only partially selective, the exact contribution of specific caspases and other proteases to injury after H-I is not clear. In this study, we explored the neuroprotective effects of a small, reversible caspase-3 inhibitor M826. M826 selectively and potently inhibited both caspase-3 enzymatic activity and apoptosis in cultured cells in vitro. In a rat model of neonatal H-I, M826 blocked caspase-3 activation and cleavage of its substrates, which begins 6 h and peaks 24 h after H-I. Although M826 significantly reduced DNA fragmentation and brain tissue loss, it did not prevent calpain activation in the cortex. This activation, which is associated with excitotoxic/necrotic cell injury, occurred within 30 min to 2 h after H-I even in the presence of M826. Similar to calpain activation, we found evidence of caspase-2 processing within 30 min to 2 h after H-I that was not affected by M826. Caspase-2 processing appeared to be secondary to calpain-mediated cleavage and was not associated with caspase-2 activation. These data suggest that caspase-3 specifically contributes to delayed cell death and brain injury after neonatal H-I and that calpain activation is associated with and likely a marker for the early component of excitotoxic/necrotic brain injury previously demonstrated in this model.

Hypoxic-ischemic (H-I) 1 encephalopathy in the prenatal and perinatal period is a major cause of morbidity and mortality and often results in cognitive impairment, seizures, and motor impairment leading to cerebral palsy (1,2). Many studies of neonatal H-I brain injury have utilized the well characterized Levine model in which unilateral carotid ligation is followed by exposure to hypoxia in postnatal day (P) 7 rats (3)(4)(5). This model of H-I results in a reproducible pattern of hemispheric injury ipsilateral, but not contralateral, to the carotid ligation (5)(6)(7). There are prominent features of both apoptosis and necrosis when this model is performed in neonatal rats and mice (1, 8 -11). Inhibition of caspases utilizing a pan-caspase inhibitor partially protects against brain injury after neonatal H-I injury in this model (12), and similar inhibitors have been shown to partially protect against ischemic injury in adult models (13)(14)(15)(16). Previously utilized peptide-based caspase inhibitors (e.g. Boc-D-fmk, z-VAD-fmk, z-DEVD-fmk) required relatively large doses in vivo for their protective effects, and at high concentrations, their effects are more likely to be less selective. Thus, although these studies suggest a role for caspases, the specific caspases and other proteases, which contribute to brain injury after neonatal H-I, have not been clarified.
Caspases are a family of cysteine aspartyl-specific proteases. They are mammalian homologues of CED-3, which is required for programmed cell death in the nematode Caenorhabditis elegans (17). To date, 13 mammalian caspases have been identified that share similarities in amino acid sequence, structure, and biochemical properties (18 -20). They are normally expressed as proenzymes  comprising an N-terminal prodomain, a large subunit (ϳ20 kDa), and a small subunit (ϳ10 kDa). Activation of caspases requires proteolytic processing between domains and formation of a heterodimer containing the large and small subunits. Each activated caspase recognizes distinct tetrapeptide motifs leading to the diversity of substrate specificity and intracellular function (21). The apoptotic initiators (caspase-2, -8, -9, and -10) containing large N-terminal prodomains generally act upstream of small prodomain effector caspases (caspase-3, -6, and -7).
Caspase-3, among effector caspases, has been implicated in neuronal apoptosis during normal brain development and in delayed neuronal cell death after brain injury in the developing and adult brain (9,(22)(23)(24)(25). Once activated, caspase-3 is directly responsible for proteolytic cleavages of a variety of fundamental proteins including cytoskeletal proteins, kinases, and DNArepair enzymes (26 -29). Caspase-3 activation can irreversibly commit cells to undergo morphological features of apoptosis including nuclear condensation and DNA fragmentation. At least two different initiator caspases, caspase 8 and 9, can be responsible for activation of caspase-3 through distinct cellular signaling pathways (30 -39). Because caspase-3 is known to be a major contributor to the apoptotic machinery in many cell types, development of selective and potent caspase-3 inhibitors has emerged as a therapeutic target.
We have utilized a new small, reversible inhibitor of caspase-3, M826, to determine the role of caspase-3 after neonatal H-I as well as to develop a compound that may have therapeutic potential. M826 demonstrated high selectivity and potency toward caspase-3 in recombinant enzyme-based as well as whole cell-based assays. It also blocked almost all caspase-3 activation and its substrate cleavage after neonatal H-I. Despite this, early excitotoxic/necrotic cell death associated with calpain activation and cleavage but not activation of caspase-2 was still present in caspase-3 inhibitor-treated animals. Our results suggest that caspase-3 contributes to delayed cell death and that early events associated with calpain activation may be involved in the rapidly occurring excitotoxic/necrotic component of cell death after neonatal H-I.

EXPERIMENTAL PROCEDURES
Enzymatic Caspase Inhibition Assay-M826 (molecular weight, 648), a selective, reversible caspase-3 inhibitor was synthesized and provided by Merck Frosst, Inc. To determine potency and selectivity of M826 against different caspases, enzymatic caspase assays were performed with active, recombinant caspase 1-10 as described previously (21,40). Briefly, caspase subunits were expressed in Escherichia coli, and each caspase was folded by rapid dilution from its corresponding denatured large and small subunits (40). After folding, active enzyme was separated from inactive subunit protein by ion-exchange chromatography. The activity of each enzyme was measured using fluorometric assays as described (40). To test the ability of M826 to inhibit apoptosis in a cell-based assay system, caspase-3 activation was induced by camptothecin (NT2) or etoposide (mouse cerebellar granule neurons (mCGNs) and mouse cortical neurons (mCORTs)) in the presence of various concentrations of M826. Briefly, mCGN cultures were obtained from 7-9-day-old CD-1 mouse pups, and mCORT cultures were obtained from E16-E18 rat fetuses. After trypsin treatment for mCGNs and papain treatment for mCORTs, tissue was triturated, and cells were plated in 96-well microplates coated with poly-D-lysine (VWR) (10 5 cells/well for mCGNs and 25,000 cells for mCORTs). mCGNs were grown in Eagle's minimal essential medium (Sigma) with 25 mM glucose, 10% fetal bovine serum (Sigma), 2 mM glutamine, 100 g/ml gentamicin for 6 days, and mCORTs were grown in neurobasal medium and B27 supplement (Invitrogen) for 4 -7 days. Etoposide was then added to a final concentration of 50 g/ml with and without different concentrations of M826. After 20 h, cells were harvested, cell lysates were prepared, and a cell death ELISA was performed with a commercially available kit per the manufacturer's instructions (Roche Molecular Biochemicals). For NT2 cells, 5000 cells were plated/well in 96-well plates. The following day, camptothecin 5 g/ml plus and minus different concentrations of M826 or other inhibitor were added for 5 h. Cells were then harvested, lysed, and analyzed as above in the dell death ELISA.
Animals and the Surgical Procedure-Newborn Sprague-Dawley rats (dam plus 10 pups per litter) were obtained from Sasco Breeders when the pups were 3-4 days of age. The pups were housed with their dam in the home cage under a 12:12-h light:dark cycle with food and water freely available throughout the study. The neonatal H-I brain injury was performed based on the Levine procedure (1, 4, 5) as described previously (9,12,42).
M826 was dissolved in phosphate-buffered saline containing 20% dimethyl sulfoxide. For in vivo studies, either vehicle or compound (3 or 30 nmol in 5 l of vehicle) was intracerebroventricular (ICV)-injected into the left hemisphere of P7 rats as described previously (12,42,43).
Caspase-2-like activity was determined using VDVAD-AFC as a caspase-2 substrate. Tissue lysate (10 l) was incubated in a 96-well plate with 90 l of assay buffer (10 mM Hepes, pH 7.4, 42 mM KCl, 5 mM MgCl 2 , 1 mM dithiothreitol, 10% Sucrose) containing 30 M VDVAD-AFC (Calbiochem). The emitted fluorescence was measured every 15 min for 60 min at 37°C at an excitation wavelength of 405 nm and an emission wavelength of 500 nm using a microplate fluorescence reader (Bio-Tek Instruments). VDVAD-AFC cleavage activity was obtained from the slope by plotting fluorescence units against time. Ac-AFC (Calbiochem) was used to obtain a standard curve, and the enzyme activity was calculated as the pmol of AFC/mg of protein/min.
Terminal dUTP Nick-end Labeling, Staining, and Cell Death ELISA-Forty-micron brain sections were mounted onto glass slides and dried. DNA fragmentation was detected using a terminal dUTP nick-end labeling apoptosis detection kit (Upstate Biotechnology) according to the manufacturer's instruction. To quantify cell death, DNA fragmentation was assayed using a cell death ELISA kit as described (45).
Assessment of Brain Damage Due to H-I-To assess regional area loss, 1 week after H-I, brain sections were prepared as described above, and damage due to H-I was determined by calculating the amount of surviving tissue in coronal sections as described previously (12,42). Briefly, coronal sections from the genu of the corpus callosum to the end of the dorsal hippocampus were stained with cresyl violet as previously described (46). The cross-sectional areas of the striatum, cortex, and hippocampus in each of eight equally spaced reference planes were photo-scanned, and the area of each brain region was calculated using SigmaScan Pro (Jandel Scientific Software). The sections utilized for quantification corresponded approximately to plates 12,15,17,20,23,28,31, and 34 in the rat brain atlas (47).
Statistics-Data are presented as the mean Ϯ S.E. and were analyzed by analysis of variance followed by Dunn's multiple comparison method with significance set at p Ͻ 0.05 unless otherwise stated.

Effects of M826 in Vitro-
The structure of M826 (molecular weight, 648) is shown in Fig. 1. Utilizing recombinant, activated caspases, we determined the potency and selectivity of M826 to different caspases (Table I). M826 inhibited caspase-3 activity at a median inhibitory concentration (IC 50 ) value of 5 nM. It also inhibited enzymatic activity of caspase-7 that has similar protein structure and substrate specificity to caspase-3 (21,48,49). We compared the potency of M826 to that of peptide caspase inhibitors in an enzyme assay system with brain tissue lysates after neonatal H-I that contained active caspase-3. M826 was 50 -1000-fold more potent than the peptide-based caspase-3 inhibitors, z-DEVD-fmk, z-VAD-fmk, and Boc-D-fmk (data not shown). We then tested the ability of M826 to inhibit apoptosis in a cell-based assay system in which caspase-3-dependent cell death occurs. M826 potently inhibited DNA fragmentation induced by camptothecin or etoposide in cell cultures with an IC 50 of 30 -120 nM (Table II). As a comparison, z-VAD-fmk had an IC 50 of 1 M for inhibiting DNA fragmentation induced by camptothecin in NT2 cells. Because we were unable to test the potency of M826 against recombinant activated caspase-9, we assessed whether M826 could block the formation of the large and small subunits of caspase-3 in NT2 cells treated with camptothecin. This cleavage is mediated by activated caspase-9. Despite the ability of M826 to potently inhibit caspase-3 activation, high concentrations of M826 (Ͼ1 M) did not block the cleavage of pro-caspase-3 between the large and small subunit (data not shown).
M826 Blocks Caspase-3 Activation and Is Neuroprotective in Vivo-To evaluate the effects of the selective caspase-3 inhibitor, M826, on caspase-3 activation and its downstream effects, we first determined the concentration of M826 present in the brain after direct ICV administration. M826 (30 nmol) was injected ICV into the left hemisphere of P7 rats, and concentrations of the compound were analyzed ( Fig. 2A). M826 was detectable bilaterally in the cortex with a similar time course profile. The brain concentration 24 h after ICV injection remained greater than 1 M. To test the ability of M826 to inhibit caspase-3 activity after H-I, P7 rats were ICV-injected with either vehicle or M826 immediately before exposure to hypoxia (2 h after the carotid ligation), and then caspase-3 activity in the brain was assessed 24 h after H-I. Consistent with previous results (9, 12), caspase-3 activity markedly increased unilaterally in the hippocampus and cortex as assessed 24 h after H-I (Fig. 2B). A single ICV injection of M826 completely blocked H-I-induced caspase-3 activation. When given immediately after H-I, M826 (3 or 30 nmol) also blocked most caspase-3 activation at 24 and 48 h post-H-I (Fig. 2C).
To test whether caspase-3 inhibition by M826 results in neuroprotection against neonatal H-I brain injury, different measures of cell death were assessed (Fig. 3). We first utilized an ELISA assay that quantitatively detects DNA fragmentation (45). M826 significantly reduced chromosomal DNA cleavage by ϳ30% as compared with vehicle-treated animals (Fig.  3A). We next investigated whether other indicators of tissue injury were affected by M826. Consistent with the ELISA assay, there was a clear decrease in the number of terminal dUTP nick-end labeling-positive cells in M826 versus vehicle-treated animals (Fig. 3B, upper panels). Similarly to previous results (12,42), neonatal H-I resulted in significant brain tissue loss in the striatum, hippocampus, and cortex when assessed 1 week after H-I (Fig. 3, C and D). M826 significantly reduced this tissue loss by ϳ30 -35% as compared with the vehicle-treated group (Fig. 3, C and D). Thus, in both assays, assessing cell and tissue injury at 24 h as well as 1 week post-H-I, M826 had consistent and significant effects. The protective effects are similar to our prior results with the non-selective, non-potent pan-caspase inhibitor Boc-D-fmk (12).
M826 Blocks Caspase-3 but Not Calpain Activation-To assess the mechanism of M826 effects, we tested whether M826 selectively inhibited caspase-3 activation and its substrate cleavage by biochemical and immunohistochemical analysis. After neonatal H-I, activation of both caspase-3 and calpain was detectable in the cortex ipsilateral to the ligation (Fig. 4). In vehicle-treated animals, both p30 and p17 cleavage products of caspase-3 corresponding to prodomain-less caspase-3 in the presence and absence of the small subunit, respectively, were detectable in the cortex ipsilateral to the ligation 24 h after H-I. M826 inhibited production of the active form of caspase-3, the p17 fragment, but it did not influence production of the p30 fragment. Importantly, this compound inhibited cleavage of the caspase-3 substrate poly(ADP-ribose) polymerase as well as generation of the caspase-3-specific p120 proteolytic fragment of ␣-spectrin in a dose-dependent manner (Fig. 4). The Ca 2ϩdependent cysteine protease, calpain, has also been implicated in both necrotic and apoptotic cell death (24,50,51), in particular during neurodegeneration (52). Calpain activation after H-I was detected by cleavage of ␣-spectrin into p150 and p145 fragments as well as degradation of the cyclin-dependent kinase (cdk5) activator, p35 (Fig. 4). In contrast to its effect on caspase-3, M826 did not influence calpain activation after H-I, as it had no effect on calpain-mediated cleavage of either ␣-spectrin or p35 (Fig. 4). We further assessed caspase-3 activation and its substrate cleavage by immunofluorescent microscopy (Fig. 5). In agreement with our previous findings (9), active caspase-3 immunoreactive (IR) cells were present in all layers of the cortex as well as in pyramidal neurons of the hippocampus of vehicle-treated rats. Consistent with these results, caspase-3-cleaved spectrin-IR cells were present in the same pattern as the caspase-3-IR cells (Fig. 5). Calpain-cleaved spectrin-IR cells also appeared in the same brain regions as caspase-3-IR cells (Fig. 5), although double labeling revealed virtually no overlap in cellular populations (data not shown). A   single ICV injection of M826 abolished virtually all active caspase-3 IR cells as well as caspase-3-cleaved spectrin-IR cells after H-I (Fig. 5). In contrast, M826 had no effect on the appearance of cells with calpain-cleaved spectrin IR cells (Fig.  5). Thus, both biochemically and at the cellular level, M826 blocked most if not all activation of the caspase-3 pathway, yet activation of calpain still occurred. Evidence for a Different Time Course of Calpain and Caspase-3 Activation after H-I-To investigate the time course and potential relationship between caspase-3 and calpain acti-vation, we examined the activation of these 2 proteases after neonatal H-I. As we have previously described (9,12), activated caspase-3, as marked by the appearance of the p17 fragment of caspase-3 by Western blot, did not appear clearly until 6 h post-H-I ipsilateral to carotid ligation with a peak at 24 h (Fig.  6A). Consistent with this finding, an increase in the caspase-3 cleavage product, p120 spectrin, increased with a matching time course (Fig. 6A). In contrast to the delayed caspase-3 activation, evidence of calpain activation was seen as early as 30 min after H-I as marked by an increase in the p150/145

FIG. 2. M826 blocks caspase-3 activity after neonatal H-I.
A, pharmacokinetics of M826 in P7 rats. P7 rats received an ICV injection of M826 (30 nmol), and the concentrations of the compound in brain were determined by mass spectroscopy (n ϭ 4). B, P7 rats underwent unilateral carotid artery ligation and exposure to hypoxia for 2.5 h. Vehicle or M826 (30 nmol) was ICV-injected before hypoxia. Brain tissue was then prepared 24 h after H-I and assayed for DEVD-AMC cleavage activity. Contra refers to contralateral, and Ipsi refers to ipsilateral to carotid ligation. C, vehicle (Veh) or M826 (3, 30 nmol) was ICV-injected immediately after H-I. Brain tissue was prepared 24 and 48 h after H-I and assayed for DEVD-AMC cleavage activity. Data represent the mean Ϯ S.E. *, p Ͻ 0.05 compared with contralateral hemispheres. #, p Ͻ 0.05 compared with the vehicle-treated ipsilateral hemisphere at 24 h. Data were analyzed by analysis of variance followed by Dunn's multiple comparison method. fragments of calpain-cleaved spectrin (Fig. 6A).
Evidence That Caspase-2 Is Cleaved but Not Activated after H-I-Because caspase-3 activation occurs in a delayed fashion, we wondered whether activation of other caspases such as initiator caspases might be activated at early times, 0 -6 h after H-I. In our previous studies, we found no evidence for caspase-1 activation in this model (12). Although we have found a small number of cells with caspase-8 activation in the P7 mouse after neonatal H-I (39) by biochemical assay and immunostaining in the P7 rat, we found no evidence for caspase-8 activation in the cortex or hippocampus (data not shown). It has been reported that the initiator/effector caspase, caspase-2, mediates cell death after certain stimuli (53,54) and, in some paradigms, is thought to contribute to cell death via a caspase-3-independent pathway (55)(56)(57). Thus, we explored the possible involvement of caspase-2 in brain injury after neonatal H-I. Concomitant with evidence of calpain activation, a proteolytic fragment of procaspase-2 (p33) was noted in the cortex ipsilateral but not contralateral to carotid ligation as early as 30 min post-H-I (Fig. 6A). Despite the appearance of the p33 fragment, there was no evidence for the formation of the activated form of caspase-2, the p12 fragment at any time after H-I. To determine whether there was any other evidence for caspase-2 activation after H-I, we also looked for the appearance of cleavage of the caspase-2 substrate VDVAD-AMC in cortical lysates. A base-line level of VDVAD-AMC activity was found in cortical tissues after H-I; however, the activity in the cortex ipsilateral to carotid ligation was not increased after H-I (Fig. 6B). To prove that VDVAD-AMC activity was a valid marker for caspase-2 activation, cortical tissue from P7 rats was spiked with recombinant activated caspase-2. This resulted in a substantial increase in VDVAD-AMC cleavage activity (Fig. 6B). Thus, although caspase-2 cleavage occurred rapidly after H-I, as for caspase-1 and -8, there was no evidence for significant caspase-2 activation.
Because caspase-2 processing occurred with the same time course as calpain activation after H-I, we investigated whether calpain might be involved in this cleavage. First, to rule out a role for caspase-3, we examined cortical tissue from H-I-treated rats that were given an ICV injection of either vehicle or M826.  (30 nmol) was ICV-injected immediately after H-I. One week later at P14, animals were sacrificed, and brain tissues were stained with cresyl violet. C, examples of brain injury seen in representative animals from each treatment are shown. D, regional area loss from the striatum, hippocampus, and cortex of each group was assessed as described under "Experimental Procedures." Data represent the mean Ϯ S.E. *, p Ͻ 0.05 compared with the vehicle-treated group. Groups from vehicle versus M826 treated groups were analyzed by t test.
M826 decreased the formation of the p17 fragment of activated caspase-3 in a dose-dependent fashion; however, it had no effect on the appearance of the p33 fragment of caspase-2 (Fig.   7A). To determine whether the p33 fragment could be the product of a calpain-mediated cleavage, P7 cortical tissue lysates were incubated in the presence or absence of calcium, EGTA, calpain, and a calpain inhibitor. We found that the appearance of the p33 fragment of caspase-2 required the presence of calcium and that calpain augmented pro-caspase-2 cleavage (Fig. 7B). Though calpain enhanced production of the p33 fragment, formation of the activated form of caspase-2, as detected by the appearance of the p12 fragment by Western blot or an increase in VDVAD-AMC cleavage activity, was not observed (Fig. 7B). Thus, cleavage of pro-caspase-2 to the p33  5. M826 blocks caspase-3 but not calpain activation as assessed by immunofluorescent staining. P7 rats underwent unilateral carotid artery ligation and exposure to hypoxia for 2.5 h. Vehicle or M826 was ICV-injected before hypoxia (n ϭ 10/group). Twenty-four hours later, animals were sacrificed, then brain sections were cut. Tissue sections from vehicle (Veh) or M826-treated rats were immunofluorescence-labeled with the following antibodies: anti-active caspase-3 antibody (Merck Frosst), anti-casp-3-cleaved spectrin p120 antibody (Merck Frosst), and anti-calpain-cleaved spectrin antibody Ab38. Representative results demonstrate that M826 blocked the appearance of almost all activated caspase-3 as well as caspase-3 cleaved ␣-spectrin-IR. M826 had no effect on calpain-cleaved ␣-spectrin-IR. Scale bars, 250 m. fragment occurs rapidly after H-I, and its cleavage is consistent with it resulting from calpain activation. Despite this, there was no evidence that activated caspase-2 was formed or was involved in H-I-induced caspase activation.

DISCUSSION
The present study demonstrates three major findings. First, we show that the small, reversible caspase-3 inhibitor, M826, selectively and potently inhibits caspase-3 enzymatic activity and caspase-3 activation in a cell-based assay system. Second, in an in vivo model of neonatal brain injury, a single ICV administration of M826 blocks almost all caspase-3 activation and cleavage of its substrates, resulting in significant neuroprotection against H-I-induced brain injury. Last, despite blockade of delayed caspase-3 activation, early excitotoxic/ necrotic cell death associated with calpain activation still occurs after neonatal H-I. We find that early calpain activation and caspase-2 processing without its activation precedes caspase-3 activation by hours and is not dependent on caspase-3. These data suggest that although caspase-3 contributes to delayed-neuronal cell death, excitotoxic/necrotic cell injury associated with molecular markers of calpain activation is likely to be involved in the non-caspase-3-depend-ent early component of cell death after neonatal H-I.
Accumulating evidence has demonstrated that there are prominent biochemical and morphological features of apoptosis as well as necrosis after brain injury in the neonatal model of H-I used in these studies (1, 3, 8 -11). The specific proteases that contribute to apoptotic cell death are beginning to be defined in different injury models. Caspase-3 activation has been well characterized in different cell death paradigms, and its activation often marks the commitment to apoptosis (18 -20). Our previous studies suggest that caspases contribute to brain injury based upon the neuroprotective effects of a pancaspase inhibitor Boc-D-fmk in this model (12). However, the exact contribution of the caspase-3 pathway to neonatal brain injury is not clear since Boc-D-fmk, like z-VAD-fmk and z-DEVD-fmk, is a peptide-mimetic caspase inhibitor that lacks selectivity and potency. Prior studies suggest that caspase-3 may be a major contributor to apoptotic cell death after neonatal H-I since there is marked activation of caspase-3 (9, 12, 24, 58 -60). In this study, utilizing a caspase-3-selective inhibitor, similar neuroprotection was observed with M826 as we previously observed with Boc-D-fmk, suggesting that caspase-3 is the major caspase contributing to delayed caspase-dependent cell death after neonatal H-I. Utilizing an ICV injection of 30 nmol of M826, we were able to block most but not all downstream cleavage of caspase-3 substrates such as ␣-spectrin. It is conceivable that even greater protection than we observed could be achieved with complete blockade of cleavage of all caspase-3 substrates. In other words, the efficacy noted (as measured by the DNA fragmentation and tissue loss endpoints) may be limited by the extent of inhibition achieved at the doses used. More pronounced effects may require a much higher fractional inhibition of the active caspase-3 pool in neonatal H-I brain. Despite a high level of caspase-3 inhibition as measured by spectrin p120 formation, any residual active caspase-3 that remains uninhibited could be sufficient to mediate the continued cleavage of target substrates. An alternative and equally viable possibility is that the non-inhibitable component of cell death observed is caspase-3-independent.
Analysis of the three-dimensional x-ray crystal structure of active caspase-3 allowed us to develop a small molecular, caspase-3 inhibitor. In this study, an enzyme assay with recombinant caspases reveals that the caspase-3 inhibitor M826 exhibits high selectivity and potency against caspase-3 and caspase-7 over other recombinant caspases tested (caspase-1, -2, -4, -5, -6, -8, and -10) ( Table I). The highly homologous caspase-3 and caspase-7 belong to group II caspases that share substrate specificity for DEXD (X refers to any amino acid). Although caspase-7, concomitantly with caspase-3, has been implicated in CD95/Fas-induced apoptosis in non-neuronal cells (61,62), its role in neuronal injury has yet to be determined. It, however, is unlikely that caspase-7 activation is involved in neuronal cell death because caspase-7 mRNA expression is barely detectable in brain, whereas it is present at higher levels in other tissues (63). A recent study suggests that caspase-7 mediates Fas-induced hepatocyte apoptosis through a compensatory mechanism in caspase-3-deficient mice (56). In agreement with previous findings in adult mice (63), we observed that caspase-7 protein in the neonatal (P7) brain was barely detectable as compared with the level found in Jurkat T cells (61) used as a positive control (data not shown). Moreover, an ICV injection of M826 with or without H-I injury did not alter the caspase-7 protein levels. Furthermore, caspase-3 expression is extremely high in the neonatal brain, and caspase-3 Ϫ/Ϫ mice have markedly decreased cell death in the developing brain (22,59). Thus, our data strongly suggest that caspase-3 is the major Group II down-FIG. 7. Caspase-2 processing after H-I is calpain-, but not caspase-3-dependent. A, P7 rats underwent unilateral carotid artery ligation and exposure to hypoxia for 2.5 h. Vehicle or M826 was ICVinjected before hypoxia. Cortical tissue lysates were subjected to SDS-PAGE and immunoblotting to determine caspase-2 processing and caspase-3 activation (n ϭ 4). B, cortical tissue lysates from sham control were incubated with Ca 2ϩ (1 mM), calpain I (0.7 g/l), EGTA (5 mM), or calpain inhibitor III (100 M) at 37°C for 30 min. The incubated samples (lanes 1-6) and activated caspase-2 (lane 7, 15 ng; lane 8, 75 ng) were subjected to VDVAD-AFC cleavage assay (VDVADase activity) and immunoblotting with anti-caspase-2 antibody. Data are representative of three similar experiments. stream caspase involved in apoptosis after neonatal brain injury.
Because caspase-3 activation is delayed after neonatal H-I, we sought to determine whether activation of other proteases might contribute to the early excitotoxic/necrotic cell injury seen in this model (64). We first examined whether calpain activation occurred at time points when excitotoxic neuronal swelling without apoptotic changes are present (60,64). Calpains are cytosolic cysteine proteases that have been implicated in necrotic neuronal death after ischemic and excitotoxic neuronal injury (8,50). Two major calpains, mu-calpain (calpain I) and m-calpain (calpain II), are ubiquitously present as proenzymes and are activated by elevated levels of intracellular Ca 2ϩ and autocleavage of both large and small subunits. Despite the different substrate preference, calpains and caspases share a number of common substrates including cytoskeletal and signal transduction proteins and transcription factors (50). One of common substrates, non-erythroid ␣-spectrin, has been well characterized, and it can be cleaved into p150 and p145 fragments by calpains and a p120 fragment by caspase-3 (44,50). Therefore, specific ␣-spectrin cleavage products can serve as dual molecular markers for both calpain activation and caspase-3 activation. Utilizing antibodies specific to the calpain-cleaved p150 fragment and the caspase-3cleaved p120 fragment, we found that M826 selectively blocked the caspase-3 pathway without affecting calpain activity after H-I. Furthermore, calpain activation occurred very rapidly after H-I; however, caspase-3 activation was delayed for at least 6 h. Although our studies demonstrate that caspase-3 activation is not required for calpain activation, it is possible that calpain activation may contribute to cell death both independently of caspase-3 and/or upstream of caspase-3 activation. Although calpains have been implicated in excitotoxic/necrotic brain injury, recent studies also suggest the role of calpains in certain settings of apoptosis via the interaction with the caspase cascades. For example, pro-apoptotic proteins Bid and Bax can be cleaved by calpains to generate active forms, which promote apoptotic cell death (65)(66)(67). Cross-talk between two cysteine proteases are further supported by recent findings that calpain and caspases are concomitantly activated after oxygen-glucose deprivation (OGD) in septo-hippocampal cultures (68) and that calpains directly activate caspase-12 after OGD in glial cells (69). The possible contribution of both calpain and caspase activation to different pathways of death, perhaps in different cells, is a study showing that a calpain inhibitor (calpain inhibitor III) and a pan-caspase inhibitor (z-VAD-fmk) synergistically protect hippocampal neurons from ischemic injury (41). Of note, however, z-VAD-fmk, which is thought to be a pan-caspase inhibitor, also inhibits calpain at a lower IC 50 than that of caspase-3 activity (24). A recent study in neonatal H-I suggests that calpain and caspase-3 activation occurs in some of the same cells (24). Further studies with selective caspase-3 and calpain inhibitors in neonatal H-I models are necessary to address how both protease cascades interact in neonatal brain injury.
In this study, we asked whether the activation of the initiator caspase, caspase-2, might precede delayed caspase-3 activation and be in a position to contribute to cell injury. We found that there was cleavage of procaspase-2 and the appearance of a p33 form. However, this cleavage was not associated with the appearance of the p12, activated form of caspase-2 nor was there an increase in caspase-2 activity in brain tissue lysates. The fact that caspase-2 cleavage occurred with the same time course as calpain activation and that the addition of calpain to P7 cortical tissue could increase formation of the p33 form of caspase-2 suggests that caspase-2 cleavage is a result of cal-pain activation after neonatal H-I. Although studies with either caspase-2 knockout mice or selective caspase-2 inhibitors will be required to rule out a role for caspase-2, our current data suggest that it is unlikely to contribute to brain injury after neonatal H-I.
In summary, we developed a non-peptide caspase-3 inhibitor that exhibits potency and selectivity for caspase-3. An ICV injection of M826 blocks caspase-3 activation and is neuroprotective against neonatal H-I-induced brain injury. The selective inhibition of caspase-3 reveals that this caspase contributes to a significant component of brain injury after H-I. In addition, early activation of excitotoxic/necrotic cell injury, which precedes caspase-3 activation and is associated with calpain activation, appears to independently contribute to neonatal H-Iinduced brain injury. Future approaches to block both calpain and caspase-3 activation may provide additive neuroprotection against neonatal H-I.