A Caspase Active Site Probe Reveals High Fractional Inhibition Needed to Block DNA Fragmentation*

Apoptotic markers consist of either caspase substrate cleavage products or phenotypic changes that manifest themselves as a consequence of caspase-mediated substrate cleavage. We have shown recently that pharmacological inhibitors of caspase activity prevent the appearance of two such apoptotic manifestations, αII-spectrin cleavage and DNA fragmentation, but that blockade of the latter required a significantly higher concentration of inhibitor. We investigated this phenomenon through the use of a novel radiolabeled caspase inhibitor, [125I]M808, which acts as a caspase active site probe. [125I]M808 bound to active caspases irreversibly and with high sensitivity in apoptotic cell extracts, in tissue extracts from several commonly used animal models of cellular injury, and in living cells. Moreover, [125I]M808 detected active caspases in septic mice when injected intravenously. Using this caspase probe, an active site occupancy assay was developed and used to measure the fractional inhibition required to block apoptosis-induced DNA fragmentation. In thymocytes, occupancy of up to 40% of caspase active sites had no effect on DNA fragmentation, whereas inhibition of half of the DNA cleaving activity required between 65 and 75% of active site occupancy. These results suggest that a high and persistent fractional inhibition will be required for successful caspase inhibition-based therapies.

The deliberate removal of excess or damaged cells is a universal feature of multicellular organisms and proceeds by a biochemical suicide program known as apoptosis (reviewed in Refs. 1 and 2). Signals that activate the apoptotic cascade include DNA damage, anoxia, lack of survival factors, and cytotoxic drugs. The downstream executioners of the apoptotic death pathway belong to a family of cysteine proteases termed caspases. Caspases exist as pro-enzymes that are activated by proximity-induced conformational change and autoprocessing, or by proteolytic processing by either an upstream activator caspase or the serine protease granzyme B (3)(4)(5)(6). The resulting heterotetramer consists of two large and two small subunits and cleaves a limited set of protein substrates with the loosely conserved recognition sequence X-Glu-X-Asp. Cleavage occurs exclusively after the aspartic acid residue (7,8). Seven of the 12 currently identified caspases participate in apoptosis, whereas the others play a role in inflammation (reviewed by Refs. 2 and 8).
Caspase-3 is the major effector apoptotic caspase in many cell types (9) and proteolysis of its substrates contributes to the biochemical and morphological changes that define apoptosis (10). For example, caspase-mediated cleavage of ␣II-spectrin is believed to contribute to membrane blebbing (11,12), whereas cleavage of ICAD leads to the internucleosomal fragmentation of DNA that typifies the apoptotic phenotype (13)(14)(15). Because of the lethal consequences of unbridled caspase activity, several checkpoints both upstream and downstream of caspase activation ensure tight control over the apoptotic pathway (reviewed in Refs. 8 and 16). Under several circumstances, however, normal safeguards do not suffice and improper caspase activation may occur. Ischemic diseases (stroke, myocardial infarction, renal ischemia, and organ transplantation), alcohol-induced hepatitis, sepsis, and Alzheimer's and Huntington's diseases, are conditions with demonstrated caspase-3 activation and where excessive caspase activity may contribute to the pathology (reviewed in Refs. 17 and 18). Caspase inhibitors, either polyspecific or caspase-3-specific, have proven efficacious in animal models of brain, liver, and heart ischemia (19 -24), in animal models of sepsis (25), liver injury (26), and cell transplantation (27,28). Hence, major efforts are currently underway to develop caspase inhibitors for clinical use.
Although the end point for compound efficacy is whether or not it will improve symptoms in preclinical animals models and in patients, an important aspect to drug development is to assess the percentage of target occupied by the drug, and to correlate this value with the occupancy required for effective treatment. In vivo occupancy studies for cell surface receptors have been mostly applied to anti-psychotic drugs and are performed by positron emission tomography and single photon emission-computed tomography (reviewed in Refs. 29 and 30). For intracellular enzymes, no reports of whole cell active site occupancy by inhibitors have been published. Instead, surrogate markers of enzymatic activity are used to evaluate the potency of inhibitors. For caspases, direct and indirect markers, such as substrate cleavage, phosphatidylserine exposure, and DNA fragmentation have been used to follow apoptosis in vivo and in vitro. Recently, however, we have shown that not all markers give the same indication of caspase activity (31). We speculated that the inhibition of some apoptotic manifestations require a greater percentage of caspase blockade than other markers. Here we characterized a radiolabeled, irreversible caspase inhibitor and established a methodology that enables the quantitation of active caspases in cells and the occupancy of their active sites by reversible, competitive caspase inhibitors.
We compared caspase active site occupancy with DNA fragmentation, and demonstrate that inhibition of up to 40% of caspase activity has no effect on apoptosis-induced DNA fragmentation, suggesting that a high fractional inhibition of caspases is required for inhibition of DNA cleavage. The caspase active site probe was able to detect active caspases in vivo, however, its uneven distribution precluded determination of caspase active site occupancy. We therefore devised an ex vivo method that estimates the percentage of caspase that must be blocked to inhibit DNA cleavage in vivo, and applied the methodology to the rat cecal ligation and puncture (CLP) 1 sepsis model.
Animals-Female Sprague-Dawley rats (250 -300 g; Charles River) and ND4 mice (20 -25 g, Harlan Sprague-Dawley) were housed in a 12-h light/dark cycle with free access to food and water. All procedures were carried out under appropriate Animal Care Committee approval in strict accordance to Merck and Co. animal care policies.
Cecal Ligation and Perforation-Animals were anesthetized with 2.5% isoflurane and body temperature was maintained by use of a thermoregulated heated blanket. A midline incision was made in the abdominal wall of the animal and the cecum exteriorized. The cecum was ligated with a nylon (4-0) suture proximal to the ileocecal valve. In mice, perforation of the cecum was done using a 23-gauge needle passed through the distal portion of the cecum. The abdominal wall was sutured with polydioxane suture (4-0) and the skin was sutured with surgical glue. Rat femoral vein cannulation and CLP surgeries were done sequentially. First, femoral vein cannulation was performed by a small incision in the inguinal region and isolation of the femoral vein. A Silicone catheter (0.02 "ϫ 0.037," Lomir) connected to a polyurethane catheter (PU-C30, 3 French, 80 cm, Instech Solomon) was inserted into the vena cava, exteriorized at the nape of the neck, and clamped for the duration of the surgery. Cannulation was immediately followed by cecal ligation and perforation, using a 20-gauge cannula (Abbott Ireland) passed through the distal portion of the exteriorized cecum. Braided silk (size 0) was threaded through the cannula and secured in place to allow leakage of the cecal content into the peritoneum. Sham-operated animals had the cecum exteriorized but no ligation or puncture of the cecum. The abdominal wall was sutured with polydioxane suture (4-0) and the skin was sutured with clips. Immediately after surgery, all animals received 1 ml of 0.9% saline administered by subcutaneous injection. In rats where compound was delivered continuously, a bolus of vehicle or compound (M867) was administered via the intravenous catheter, which was then connected to a Medfusion 2010i Syringe pump (Medex Inc.) at a delivery rate of 2 ml/h/kg for 24 h.
Cell Culture-All cells were cultured in a humidified incubator at 37°C with 5% CO 2 . Jurkat cells (ATCC TIB-152) were maintained in RPMI 1640 supplemented with 10% fetal bovine serum, 2 mM L-glutamine, and antibiotics (penicillin and streptamycin). Rat thymocytes were cultured for 24 h at a density of 10 ϫ 10 6 cells/ml in CytoSF4 media (Kempt Technologies) supplemented with 2 mM L-glutamine and penicillin-streptamycin. NT2 cells (ATCC 3813555) were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, 2 mM L-glutamine, and antibiotics. Apoptosis was induced by addition of campothecin (3 g/ml) on serum-starved Jurkat cells suspended at a density of 1 million cells/ml in CytoSf4 media or by addition of camptothecin (5 g/ml) added to the culture media of NT2 cells that had reached 60 -80% confluence. Cultured rat thymocytes spontaneously initiated apoptosis under the culture conditions used. Typically, 30% of thymocytes exhibited subdiploid DNA content after 24 h of culture.
Thymocyte Cell Suspensions-Single-cell thymocyte suspensions were obtained by grinding fresh thymi in 50-m Medicon (Dako) and Medimachine (Dako) with 2 ml of ice-cold isolation buffer (phosphatebuffered saline, 2 mM glucose, 2 mM L-glutamine, 1% fetal bovine serum), for 2 15-s pulse. The suspension was filtered through a 50-m nylon mesh (BD Bioscience). Cells were centrifuged at 300 ϫ g for 10 min at 4°C, suspended in 10 ml of hypotonic lysis buffer (17 mM Tris-Cl, pH 7.5, 140 mM NH 4 Cl), and incubated 10 min at room temperature. Thymocytes were pelleted by centrifugation at 300 ϫ g for 10 min, suspended in 5-10 ml of isolation buffer without fetal bovine serum, and counted with hemocytometer and trypan blue staining.
Apoptosis Detection-DNA fragmentation was quantified by subdiploid DNA content, which was determined with flow cytometry. Briefly, 1.5 ϫ 10 6 thymocytes or 5 ϫ 10 5 Jurkat cells were fixed for 30 min on ice in 80% EtOH. The cells were pelleted by centrifugation at 300 ϫ g for 10 min, suspended in phosphate-buffered saline, and incubated on ice for 45 min. Pelleted cells were suspended in 500 l of PI staining solution (phosphate-buffered saline, 0.1% Triton X-100, 50 g/ml RNase A (Roche), 25 g/ml propidium iodide (Sigma)). Flow cytometry was performed with a FACS Calibur instrument (BD Bioscience) on 20,000 events per sample, each sample was prepared in duplicate. Very small cell debris was electronically gated out based on forward light scatter. A comparison of DNA fragmentation assessed by subdiploid DNA content and cytoplasmic nucleosomes has been performed elsewhere (31). Both methods yielded identical results.
Tissues, Cell Extracts, and Caspase Activity-All tissues were either frozen or processed within 20 min of their removal. Rat tissues and cultured cells were lysed in ice-cold TNE lysis buffer (50 mM Tris-Cl, pH 7.5, 1% Nonidet P-40, 2 mM EDTA) supplemented with 5 mM dithiothreitol and Complete TM protease inhibitor mixture (Roche). Thymi from mice injected with [ 125 I]M808 were homogenized in lysis buffer supplemented with 25 M M808. Nuclear debris was removed by centrifugation at 13,000 rpm for 10 min and the soluble fraction was quantitated by Bradford assay (Bio-Rad). Caspase-3-like activity was determined with the fluorogenic substrate Ac-DEVD-AMC as described previously (33), with a concentration of Ac-DEVD-AMC (10 M) equivalent to its K m on caspase-3 (34). According to the equation, IC 50 ϭ K i (1 ϩ S/K m ), which is applicable to reversible competitive inhibitors, the IC 50 ϭ 2 K i under the conditions used. K i is the inhibitory constant. The K i values for M791 and M826 were determined on purified recombinant human caspase-3 at a final concentration of 0.15 nM in ICE III buffer (50 mM Hepes-KOH, pH 7.0, 0.1% CHAPS, 10% sucrose, 2 mM EDTA), supplemented with 5 mM dithiothreitol and preincubated for 15 min with the indicated concentrations of M791 or M826. Recombinant human caspase-3 was a gift from Nancy Thornberry and was purified from Escherichia coli expressed subunits as described in Ref. 34.
[ 125 I]M808 Labeling and SDS-PAGE-Active caspases in tissue or cell extracts were labeled by incubating 60 -80 g of lysate with 1.25 nM [ 125 I]M808 for 15 min at 25°C. Labeling was stopped by addition of Laemmli buffer (2% SDS, 0.35 M ␤-mercaptoethanol, 50 mM Tris-Cl, pH 6.8, 10% glycerol, 0.05% bromophenol blue) and heating at 95°C for 5 min. The K i values for M791 and M826 were determined with [ 125 I]M808 and purified recombinant human caspase-3 at a final concentration of 0.15 nM in ICE III buffer, supplemented with 5 mM dithiothreitol. Caspase-3 and the inhibitors were incubated 15 min with the indicated concentrations of M791 or M826 before the addition of the probe. [ 125 I]M808 was added to a final concentration of 0.25 nM for 5 min before stopping the reaction by boiling in 1ϫ Laemmli buffer.
Caspase labeling in whole cells was performed as described below. Serum-starved Jurkat cells (0.5 ϫ 10 6 ) were plated in CytoSF4 with camptothecin (3 g/ml) and the indicated concentrations of caspase inhibitors for 4 h. The cells were then centrifuged at 350 ϫ g for 10 min and then suspended in CytoSF4 containing 12.5 nM [ 125 I]M808 and the indicated concentrations of caspase inhibitor for a 1-h incubation period at 37°C. Labeling was stopped by washing the cells with 1 ml of cold phosphate-buffered saline supplemented with 25 M M808. The cells were pelleted by centrifugation and lysed in TNE buffer supplemented with 25 M M808 and protease inhibitors. Rat thymocytes were treated the same way except that 3 million cells per assay were used. Labeling with 12.5 nM [ 125 I]M808 was performed for 15 min at 37°C.
In vivo labeling of active caspases was performed by intravenous injection (jugular vein) of 125 l of a 0.43 M solution of [ 125 I]M808 (1125 Ci/mmol) in 10% PEG200. Assuming an average blood volume of 1.25 ml per mouse, the expected initial blood concentration of [ 125 I]M808 is 43 nM. Animals were euthanized 45 min after the [ 125 I]M808 injection. Thymi were recovered and processed for protein extraction as described above. Denatured protein extracts in 1ϫ Laemmli buffer were resolved through an 18% polyacrylamide SDS gel (Invitrogen) at 30 mA in Tris glycine buffer. All samples labeled with [ 125 I]M808 were migrated until the dye front reached 0.5 cm above the end of the gel. All material 1 cm above the dye front was cut out to remove free [ 125 I]M808. The gels were fixed for 45 min in a 40% methanol, 10% acetic acid solution and dried for 90 min at 80°C under vacuum before exposure on MS Kodak film with MS intensifying screens (Kodak).
Western Blotting-Proteins were resolved in an 18% SDS-polyacrylamide gel and transferred onto a 45-m nitrocellulose membrane (Invitrogen) in Tris glycine, 20% methanol buffer at 40 V for 2 h. Nonspecific protein binding was minimized by incubating the membranes for 1 h in blocking buffer (5% nonfat milk, TBS, 0.1% Tween 20). Caspase-3 antisera MF R280 and MF467 were raised in rabbits against recombinant large subunits (p17) of human and rat caspase-3, respectively. Both antisera were used at a 2000-fold dilution in blocking buffer and incubated for 1 h at 25°C. Membranes were washed in TBS, 0.1% Tween 20 and incubated for 45 min with horseradish peroxidase-coupled anti-rabbit IgG antibody (Amersham Biosciences) diluted 5000fold in blocking buffer. The enhanced chemiluminescence reaction was performed with Supersignal West Femto chemiluminescent reagent (Pierce) and exposed to Hyperfilm ECL (Amersham Biosciences). Densitometry on 125 I-exposed or chemiluminescence-exposed film was obtained using a Bio-Rad GS800 instrument and QuantityOne software (Bio-Rad). A pixel saturation calibration curve was done before each scan, with saturated pixels appearing in red. Only signals that were well below the apparatus pixel saturation limit were quantitated. Background densitometric values corresponding to the identical area size of each quantitated bands were subtracted. In some experiments, an [ 125 I]M808-labeled caspase-3 standard curve was performed to establish the linearity of the signal. A 20-fold dynamic range with a single film exposure time was achieved.
IC 50 Calculations and Statistical Analysis-Curve fitting was obtained using a sigmoidal Hill 4-parameter equation using SigmaPlot 8.0 software. Statistical analysis was performed by standard one-way analysis of variance. Where applicable, data were log-scaled so that underlying assumptions of equal variance and normality were better satisfied. All comparisons were deemed statistically significant at the 0.05 level.

RESULTS
[ 125 I]M808 Specificity and Sensitivity in Vitro-[ 125 I]M808 is an iodinated trifluoromethylketone caspase inhibitor predicted to bind covalently and irreversibly to the active site cysteine located in the large caspase subunit (Fig. 1). The inhibitory profile of M808 against purified recombinant caspases indicates a slight preference (2-fold) for caspase-3 over caspase-8, and a 10-fold preference over caspase-7 (Table I).
We tested the specificity of [ 125 I]M808 for active caspases in crude protein extracts. Equal amounts rat tissue homogenates were either incubated with [ 125 I]M808 or pretreated with granzyme-B to activate caspases prior to the addition of [ 125 I]M808. In parallel, caspase activity in granzyme-B-treated extracts was measured by cleavage of the fluorogenic substrate Ac-DEVD-AMC. The large subunit of fully processed caspase-3 migrated at 17 kDa (p17). Partially processed caspase-3 retains part (p19) or all (p20) of the pro-domain (34), but is as active as the p17 form. 2 Removal of the prodomain is self-catalyzed (33). As expected, the p17 subunit of purified recombinant human caspase-3 covalently bound [ 125 I]M808 ( Fig. 2A, first lane). All granzyme-B-treated rat tissues contained an [ 125 I]M808-labeled 19-kDa protein, whereas liver and thymus showed an additional polypeptide migrating at 20 kDa (second through fifth lanes). The identity of the radiolabeled proteins as caspases was ascertained by several means. Radiolabeled proteins migrating at this position were not observed when granzyme-B was omitted (data not shown). Moreover, preincubation of the extract with the caspase-3-specific inhibitor M791 abolished (sixth through eighth lanes) or greatly reduced (ninth lane) the appearance of the radiolabeled proteins. Caspase activity, as measured by Ac-DEVD-AMC cleavage, correlated with the amount of radiolabeled p19/p20 proteins (Fig. 2B).
[ 125 I]M808 recognized polypeptides migrating at 17, 19, and 20 kDa in extracts of cells treated with the apoptotic inducers camptothecin or anti-Fas antibody. Labeling of a 70-kDa protein was also observed in NT2 cell extracts, but not Jurkat (Fig.  2C). The p17-, p19-, and p20-radiolabeled proteins were absent if the apoptotic inducer was omitted, or if M791 had been added in the extract, whereas labeling of the 70-kDa polypeptide was not affected. Taken together, these results demonstrate that [ 125 I]M808 binds to caspases, including caspase-3, in a complex protein mixture.
[ 125 I]M808 Detects Caspase Activity in Animal Models of Apoptotic Injury-Using purified recombinant human caspase-3, the sensitivity threshold of [ 125 I]M808 was established at 0.5 fmol (0.02 ng; Fig. 2D). The intensity of the radiolabeled signal was directly proportional to the amount of caspase-3, with a 20-fold dynamic range per film exposure (Fig. 2E). With different exposure times, a linear range of 0.5 to 100 fmol of caspase-3 was achievable (data not shown). We wanted to know if this sensitivity would be sufficient to detect active caspases in pharmacological models of caspase activation. To that end, [ 125 I]M808 was incubated in tissue extracts from three animal models of ischemia and one animal model of sepsis. Hypoxia-ischemia (H-I) in neonatal rat brain leads to caspase-3 activation in the injured, ipsilateral side, but not the contralateral side (22). Incubation of rat brain extracts with [ 125 I]M808 resulted in the labeling of a 17-kDa protein in ipsilateral but not contralateral extracts (Fig. 2F, lanes 1 and 2). In a rat myocardial infarct model where caspase activation has been documented (36), a 17-kDa protein was detected in the injured (ventral left ventricle) side but not the control uninjured side (right ventricle; Fig. 2F, lanes 3 and 4). This signal could be competed by addition of M791 to the extract prior to labeling. Additional nonspecific proteins were also labeled by [ 125 I]M808 in both injured and non-injured tissues. [ 125 I]M808 failed to label any specific protein in the rat middle cerebral artery occlusion stroke model (data not shown). In the rat CLP model for sepsis, caspase-3 cleavage is evident from the abundance of caspase-3 p17 in the thymus of septic rats (31). Strong [ 125 I]M808 labeling of a 17-kDa protein was observed in septic but not in sham operated rat thymi (Fig. 2F,  lanes 5-9). Thus, [ 125 I]M808 recognizes active caspases in several 2 S. Roy, unpublished data.  types of ischemia-injured tissues and in the thymus from septic animals.
[ 125 I]M808 Measures Caspase Activity in Vitro-We have shown that [ 125 I]M808 is a sensitive probe that detects active caspases in apoptotic cells or in injured tissues. Next, we examined whether [ 125 I]M808 could be used to measure the fraction of active caspase-3 remaining in the presence of a reversible active site inhibitor. Purified recombinant caspase-3 was preincubated with increasing amounts of the reversible, com-petitive, and caspase-3-selective inhibitors, M791 or M826 (22,25). Half of the reaction was used for the Ac-DEVD-AMC cleavage rate assay, whereas the other half was incubated with [ 125 I]M808 before resolving on SDS-PAGE. The K i value for M791 measured by Ac-DEVD-AMC cleavage was 0.74 nM (Fig.  3A). Remarkably, the K i value for M791, when enzyme occupancy was measured with [ 125 I]M808, was virtually identical at 0.73 nM (Fig. 3, C and E). Similarly, the K i values for M826, measured by Ac-DEVD-AMC and [ 125 I]M808 labeling, were  (Fig. 3, B, D, and F) [ 125 I]M808 was added to the culture media for the last incubation hour. M791, which is a reversible and membrane-permeable caspase-3-selective inhibitor, blocked cell apoptosis as measured by DNA fragmentation with an IC 50 of 4.8 M (Fig.  4A). Western blot analysis of cell lysates using a caspase-3 antibody showed that M791 prevented the autoprocessing of p20 caspase-3 into p19 and p17, but did not block the overall formation of p20, as expected for a caspase-3 selective inhibitor (Fig. 4B). [ 125 I]M808 labeled all three forms of caspase-3 (Fig.  4C). Typically, 30% of the total pool of active caspases were labeled under the conditions used (data not shown). The occupancy of active caspase-3 by M791 was evident from the decrease in the sum of radiolabeled p17/p19/p20 intensity as the concentration of M791 increased (Fig. 4C). A concentration of 2.1 M M791 was required to block 50% of the caspase active sites, whereas 50% inhibition of DNA fragmentation necessitated 4.8 M (Fig. 4, A and D). This suggests that a large fractional inhibition of caspase-3 is required to inhibit apoptosis.
Apoptosis and caspase-3 activation has been documented in thymocytes and splenocytes of septic rats and mice (31,37,38). Caspase-3 selective inhibitors such as M791 and M867, together with the absolute requirement for ICAD cleavage to allow DNA fragmentation (39), suggest that thymocyte apoptosis in vitro and during sepsis is mediated by caspase-3, and possibly caspase-7. M867 completely blocks apoptosis in vivo in thymi of septic rats, but with strikingly different efficacy, depending upon whether ␣II-spectrin cleavage or DNA fragmentation where used to assess apoptosis (31). Similar results were also obtained with cultured rat thymocytes, where 3-fold more M867 was required to inhibit 50% of the DNA fragmentation than ␣II-spectrin cleavage (31). In light of these results, we compared the potency of M867 at inhibiting DNA fragmentation with caspase enzyme occupancy, in cultured rat thymocytes. Cleaved caspase-3 and DNA fragmentation were quantified in parallel by Western blotting and flow cytometric determination of subdiploid cells. Shown in Fig. 5 is a repre-sentative example of five independent experiments. Western blot showed a minimal impact of M867 on caspase-3 cleavage except at the highest inhibitor concentration, as expected for a caspase-3 selective compound, given that its activation is presumed to be mediated by upstream caspases (Fig. 5A). M867 inhibited DNA fragmentation in cultured rat thymocytes with an IC 50 of 0.24 M (Fig. 5C). Incubation of thymocytes for 15 min with [ 125 I]M808 resulted in labeling of ϳ5% of the total active caspases (data not shown). A dose-dependent decrease in the intensity of [ 125 I]M808-labeled p17 was observed with thymocytes incubated with M867, with an IC 50 of 0.14 M (Fig.  5B). We calculated the actual active site occupancy in thymocytes based on the [ 125 I]M808 p17 signal corresponding to the p17 amount measured by Western blot, in the absence of M867 (Fig. 5, A and B, first and second lanes). The predicted [ 125 I]M808 p17 signal intensity was then calculated based on Western blot densitometric values for all cells exposed to M867. The ratio of the measured [ 125 I]M808 p17 densitometry and predicted [ 125 I]M808 value, multiplied by 100, corresponds to the percentage of free, unoccupied caspase active sites. Table II outlines these calculations and the percentage of free and occupied active sites calculated from the experiment shown in Fig. 5. Occupancy of up to 40% of active caspases by M867 did not result in any diminution of DNA fragmentation. A plot of percent caspase active site occupancy versus M867 concentration indicates that an M867 concentration of 0.14 M blocks 50% of the caspase active sites in dying thymocytes. At this FIG. 4. Inhibition of DNA fragmentation in apoptotic Jurkat cells requires greater than 50% caspase occupancy. A, percentage of residual DNA fragmentation activity and corresponding IC 50 value in apoptotic Jurkat cells treated with increasing amounts of M791. DNA fragmentation here and in all subsequent figures was measured by flow cytometry and was equated to the percentage of subdiploid cells. 100% DNA fragmentation activity was arbitrarily set at the percentage of subdiploid cells present in the absence of caspase inhibitor. B, caspase-3 Western blot on camptothecin-treated Jurkat cell extracts. Unprocessed (p32), fully (p17) and partially processed (p19 and p20) caspase-3 polypeptides are indicated by arrows. The concentration of M791 present is shown below each lane. C, [ 125 I]M808-labeled p17, p19, and p20 caspase subunits in healthy and apoptotic Jurkat cells exposed to increasing amounts of M791. The p17, p19, and p20 radiolabeled polypeptides are indicated by arrows. No labeling of full-length, unprocessed p32 caspase was observed. D, plot of [ 125 I]M808-labeled p17/p19/p20 densitometry against M791 concentration. The IC 50 determined by caspase active site occupancy is 2.1 M.
concentration, only 15% of DNA fragmentation was inhibited (Fig. 5C). The average IC 50 for M867 determined by DNA fragmentation in five independent experiments was 0.27 Ϯ 0.04 M. The same experiments yielded an average IC 50 of 0.13 Ϯ 0.02 M when caspase activity was measured with [ 125 I]M808. The apparent 2-fold decrease in M867 potency with respect to DNA fragmentation is statistically significant (p ϭ 0.003). In contrast, the concentration of M867 required to reduce ␣II-spectrin cleavage by half was not statistically different from the dose needed to achieve 50% occupancy of caspase active sites (data not shown). Other compounds tested in Jurkat cells and rat thymocytes showed that a 50% inhibition of DNA fragmentation required between 62 and 77% of caspase active sites to be blocked (Table III).
[ 125 I]M808 Detects Active Caspases in Vivo-Next, we determined whether [ 125 I]M808 detects active caspases in vivo. Cecal ligation and perforation in mice results in peritonitis and activation of caspases in tissues such as the spleen, thymus, and the gut (38,40,41). Peritonitis was induced for a little over 23 h by CLP surgery in mice, and [ 125 I]M808 was injected intravenously 45 min before euthanasia (Fig. 6A). Western blot on thymi protein extracts revealed the presence of p17 caspase-3 in all CLP, but not sham-operated animals (Fig. 6B). Similarly, a radiolabeled p17 protein was seen only in CLP but not in sham animals (Fig. 6C). Thus, [ 125 I]M808 detected active caspases in vivo. However, in contrast to protein extracts and whole cell labeling, the intensity of the p17 signal was not proportional to the amount of p17 caspase-3 present. This was also true for thymus proteins nonspecifically labeled with [ 125 I]M808. This discrepancy may be due to the variable accessibility of the probe to the injured thymus during sepsis, as this condition is known to cause organ hypoperfusion. As such, even though [ 125 I]M808 detected active caspases in vivo, its usefulness is limited to systems where tissues are equally perfused in all animals.
Fractional Inhibition of Caspases in an ex Vivo Assay-Because in vivo injection of [ 125 I]M808 did not allow caspase  Table II. C, percentage of DNA fragmentation activity measured by flow cytometry and percent of caspase active site occupancy in apoptotic rat thymocyte-treated M867. 100% DNA fragmentation activity was arbitrarily set at the percentage of subdiploid cells present in the absence of caspase inhibitor. The calculations for the determination of percent caspase occupancy are outlined in Table II and are based on densitometric values obtained from panels A and B. Half of the caspase active sites were occupied when the concentration of M867 was 0.14 M. fractional inhibition determination, we tested whether active site occupancy could be obtained by labeling cells ex vivo. Rats were subjected to CLP surgery and infused continuously with either vehicle or M867 at 1 or 4 mg/kg/h. The doses of M867 chosen were expected to have either no effect or partial inhibitory effects on thymocyte DNA fragmentation (31). Thymi were collected 24 h post-surgery and single cell suspensions were prepared for DNA fragmentation analysis and whole cell ex vivo [ 125 I]M808 labeling. Relative caspase-3 amounts were also quantified by Western blotting. As expected, CLP surgery resulted in a significant increase in the caspase-3 p17 fragment content of thymocytes. Infusion of M867 at either dose did not significantly affect p17 levels (Fig. 7A). M867 at 4 mg/kg/h reduced DNA fragmentation by 70%, whereas 1 mg/kg/h failed to inhibit (Fig. 7C). The slight increase in DNA cleavage observed by treatment with 1 mg/kg/h while statistically significant here, was not observed in other experiments (31). Active caspases were labeled in suspended thymocytes obtained from all treatment groups (Fig. 7B). There was a positive correlation between the amount of caspase-3 p17 fragment as measured by Western blotting, and the amount of [ 125 I]M808 p17 polypeptide (r ϭ 0.92; Fig. 7D). This positive correlation was used to calculate the percentage of caspase active sites occupied by M867, much in the same manner as what was outlined in Table   II. A mean fractional inhibition of 23 and 62% was found in the 1 and 4 mg/kg/h groups, respectively (Fig. 7E). Thus, caspase fractional inhibition measured in an animal model of cellular injury correlates well with the amount of M867 dosed and the percentage of DNA fragmentation inhibition achieved. DISCUSSION We describe here a novel assay that enables the determination of fractional inhibition of an intracellular enzyme, caspase-3, by a reversible active site inhibitor. We have shown the caspase probe [ 125 I]M808 to be a highly sensitive tool capable of detecting active caspases in several animal models of apoptotic injury. No labeling of unprocessed, full-length caspases was observed with this probe. Partial labeling of the total pool of free caspases with [ 125 I]M808 accurately reflects the amount of caspase activity remaining, both with purified enzyme and in living cells. The fact that fractional inhibition determination was feasible on a short-lived enzyme such as caspase-3 suggests that this method will also be applicable to other enzymes. Using this method, we show that high fractional inhibition of caspases is required to block DNA fragmentation in cultured cells.
The determination of caspase occupancy described in this work makes use of intact cells as opposed to lysed cells or . Arrows indicate full-length pro-caspase-3 (p32) and the processed caspase large subunit (p17). C, in vivo [ 125 I]M808-labeled protein in thymi extracts from septic mice. The arrow points to the radiolabeled caspase large subunit. Several radiolabeled nonspecific polypeptides were present, both in CLP and sham operated mice. homogenized tissues that had been exposed to M867. In theory, it should have been possible to measure fractional inhibition by M867 with direct exposure of lysates to [ 125 I]M808. However, we found that cells or tissues incubated with M867 or M826, and subsequently lysed and exposed briefly to [ 125 I]M808, exhibited far greater enzyme occupancy relative to the IC 50 determined by DNA fragmentation (data not shown). We speculate that much of M867 and M826 are membrane-associated in live cells and freed upon lysis, thus gaining access to free caspase active sites. This is a compound-specific effect because caspase active site occupancy by M791 is the same whether determined on living or lysed cells. 3 Although the [ 125 I]M808 caspase active site probe has proven itself extremely useful for ex vivo fractional inhibition determination, the methodology suffers from a few limitations. Occupancy of caspase active sites by a reversible inhibitor must be determined rapidly to minimize dissociation of the reversible inhibitor, and will be artificially low if the drug has a short dissociation half-life. For M826 and M867, the dissociation half-life with purified enzyme is 60 min at 25°C (data not shown). This limitation did not pose a problem in cultured cells because occupancy was determined while M826, M867, and M791 were still present in the culture media. We have determined caspase occupancy on cells first incubated with these drugs, then washed to remove free compound and incubated with the active site probe. With all manipulations except probe incubation carried out on ice, compound dissociation was kept at a minimum, and caspase active site occupancy was only slightly decreased by up to 3 h incubation on ice (1.5-fold increase in IC 50 ; data not shown). Hence, [ 125 I]M808 permits reasonable estimates of fractional inhibition by a reversible inhibitor if all procedures are carried out quickly, even if the compound is no longer present in solution. This is what was achieved with ex vivo thymocyte labeling of septic rats dosed with M867.
One of the major obstacles to fractional inhibition determination with [ 125 I]M808 is its limitation to systems where the large subunit of cleaved caspases can be detected by Western blotting. Additionally, if an inhibitor is prone to membrane association, such as is likely the case with M867, occupancy must be determined in tissues where intact cells can be rapidly isolated. In this respect, CLP-induced sepsis is a good model to study in vivo fractional inhibition because large amounts of p17 are generated in the thymus, and thymocyte cell suspensions can be rapidly made. The ex vivo caspase active site occupancy determined in septic animals dosed with M867 is probably a slight underestimation of the actual value, because of M867 dissociation during tissue processing time. A more accurate determination of fractional inhibition might be achievable by directly labeling the tissue extract, as opposed to whole cells.  7. [ 125 I]M808 can be used ex vivo to determine caspase active site occupancy in septic rats dosed with M867. Rats underwent either CLP (n ϭ 19) or sham surgery (n ϭ 2) and were dosed by continuous intravenous infusion of vehicle (n ϭ 5), M867 at 1 mg/kg/h (n ϭ 7), or M867 at 4 mg/kg/h (n ϭ 7). Sham operated animals were dosed with vehicle. Thymocytes were recovered 24 h following surgery, and analyzed for DNA fragmentation by flow cytometry as described under "Experimental Procedures." Caspase-3 cleavage products were detected by Western blotting (A). In parallel, thymocytes were labeled with [ 125 I]M808 ex vivo (B). Arrows point to pro-caspase-3 (p32) or the p17 large subunit. A nonspecific (n.s.) [ 125 I]M808-labeled protein is detected in all extracts including sham animals, but does not co-migrate with caspase-3 p32. C, average percent subdiploid thymocytes present in vehicle and drug-treated animal groups. D, correlation between the p17 caspase-3 subunit measured by Western blotting and [ 125 I]M808-labeled p17 caspase densitometry in vehicle-treated CLP animals. C3 refers to caspase-3. E, average percent caspase active site occupancy by M867 in vehicle and drug-treated (1 and 4 mg/kg/h M867) animals. Values were calculated from the slope obtained in panel C. Error bars in C and E represent Ϯ S.E. caspase inhibitor that does not accumulate in membranes, such as M791, would need to be used to compare occupancy measured by ex vivo labeling, and occupancy measured by direct labeling of tissue extracts. Unfortunately, M791 was not sufficiently efficacious in rat septicemia to test this hypothesis.
The [ 125 I]M808 caspase active site probe also has potential utility in clinical caspase inhibitor development for sepsis. Although apoptotic cells are known to be rapidly cleared from the circulation, apoptotic peripheral blood monocytes and lymphocytes have been detected in septic patients or in patients undergoing chemotherapy (42)(43)(44). It would be interesting to test whether [ 125 I]M808 can detect active caspases in whole blood, and eventually, as caspase inhibitors reach clinical trials, determine occupancy ex vivo in blood of patients.
In summary, we have measured caspase active site occupancy by potent, reversible inhibitors using a radiolabeled active site probe. Our data indicates that a fractional inhibition between 65 and 75% is required to block DNA fragmentation by 50%, and explains in part why a 3-4-fold higher concentration of M867 is needed to oppose DNA cleavage compared with ␣II-spectrin cleavage. As prevention of DNA fragmentation is most likely sine qua non for cell survival, our findings suggest that a high and persistent blockage of caspase active sites will be needed for therapeutic benefit.