An Inhibitor of p38 Mitogen-activated Protein Kinase Protects Neonatal Cardiac Myocytes from Ischemia*

Cellular ischemia results in activation of a number of kinases, including p38 mitogen-activated protein kinase (MAPK); however, it is not yet clear whether p38 MAPK activation plays a role in cellular damage or is part of a protective response against ischemia. We have developed a model to study ischemia in cultured neonatal rat cardiac myocytes. In this model, two distinct phases of p38 MAPK activation were observed during ischemia. The first phase began within 10 min and lasted less than 1 h, and the second began after 2 h and lasted throughout the ischemic period. Similar to previous studies using in vivo models, the nonspecific activator of p38 MAPK and c-Jun NH2-terminal kinase, anisomycin, protected cardiac myocytes from ischemic injury, decreasing the release of cytosolic lactate dehydrogenase by approximately 25%. We demonstrated, however, that a selective inhibitor of p38 MAPK, SB 203580, also protected cardiac myocytes against extended ischemia in a dose-dependent manner. The protective effect was seen even when the inhibitor was present during only the second, sustained phase of p38 MAPK activation. We found that ischemia induced apoptosis in neonatal rat cardiac myocytes and that SB 203580 reduced activation of caspase-3, a key event in apoptosis. These results suggest that p38 MAPK induces apoptosis during ischemia in cardiac myocytes and that selective inhibition of p38 MAPK could be developed as a potential therapy for ischemic heart disease.

The heart is subjected to episodes of ischemia followed by reperfusion in a number of situations, including angina, myocardial infarction, and cardiac surgery, and these stresses can result in cell injury and death. Part of the cellular response to ischemia/reperfusion is activation of several members of the mitogen-activated protein kinase (MAPK) 1 family. In many different cell types, p38 MAPK and c-Jun NH 2 -terminal kinase (JNK) family members are activated predominantly by cellular stresses or inflammatory signals, e.g. hyperosmolarity, chemical or heat stress, endotoxin, and cytokines (1)(2)(3)(4), whereas the extracellular signal-regulated kinases (ERKs) are activated by mitogenic stimuli (5).
In the isolated perfused rat heart, p38 MAPK is activated by global ischemia, and activation is maintained during reperfusion (6,7). In the same model, neither ERKs nor JNKs are activated by ischemia, whereas reperfusion after ischemia activates JNK (6 -8). Different studies have shown activation (8) or lack of activation (6) of ERKs on reperfusion, possibly the result of different assay methods. More recently, it was observed that although JNK1 (also termed JNK46) is not activated by ischemia, this stress results in translocation of JNK1 to the nucleus, where it is then phosphorylated and activated on reperfusion (9). Ischemia and reperfusion also activate members of the MAPK family in kidney and liver differentially (7,10,11). However it is not clear from these studies whether activation of these kinases is part of the protective response of the cell or if these signals mediate the cellular damage and death caused by ischemia or ischemia/reperfusion. Evidence suggests that myocardial ischemic cell death occurs by both apoptosis and necrosis (12,13). From the timing of p38 MAPK activation during ischemia and initiation of apoptosis, Yin et al. (7) speculate that activation of p38 MAPK initiates the signal for apoptotic cell death. Indeed, p38 MAPK activation has been implicated in mediating apoptosis in several cell types (14 -16). Recent studies in neonatal rat cardiac myocytes support a role for the ␣ isoform of p38 MAPK in mediating apoptosis; overexpression of activated MAPK kinase 3b, which phosphorylates and activates p38 MAPK, induces apoptosis that is increased by coexpression of p38␣ and is decreased by expression of a dominant negative form of this isoform (17).
In contrast, a separate study demonstrates that activation of p38 MAPK can prevent apoptosis in neonatal rat cardiac myocytes (18). Furthermore, others have proposed that p38 MAPK activation mediates a phenomenon termed preconditioning, which confers cardiac protection from ischemia. Preconditioning is a highly effective method of protecting the heart from ischemic damage by subjecting it to sublethal periods of ischemia before the prolonged ischemia (6,8,19,20). A protective function for p38 MAPK is supported by a recent study in which the role of p38 MAPK in preconditioning was examined in isolated rabbit cardiac myocytes (21). Pretreatment with anisomycin, an activator of p38 MAPK, protects isolated rabbit cardiac myocytes against ischemia-induced cell fragility, leading to the suggestion that p38 MAPK protects the heart against ischemia. The addition of SB 203580, a selective inhibitor of p38 MAPK (22,23), during a preconditioning treatment abolishes the protective effect of preconditioning, supporting the initial observation (21). Similar results, showing that SB 203580 inhibits the protection afforded by ischemic preconditioning against myocardial infarction, were obtained using isolated rat hearts (24).
Most previous ischemia studies have investigated MAPK activation in whole heart, which contains a large proportion of non-myocyte cells, mainly fibroblasts and endothelial cells. In the present study, we used primary cultures of neonatal rat cardiac myocytes and confirmed that p38 MAPK is activated in a model of ischemia which uses a glucose-free hypoxic incubation. We report that activation of p38 MAPK occurred in two distinct phases and that inhibition of p38 MAPK during the second phase protected cardiac myocytes from ischemic injury. These results are consistent with the hypothesis that sustained p38 MAPK mediates ischemia-induced cell injury and death in neonatal rat cardiac myocytes.

EXPERIMENTAL PROCEDURES
Reagents-All antibodies were used according to manufacturers' protocols. Anisomycin (Sigma) was dissolved in dimethyl sulfoxide (Me 2 SO) at 5 mg/ml and used to give a final Me 2 SO concentration less than 0.01%. SB 203580 (Calbiochem) was dissolved in Me 2 SO at 10 mM and used to give final a Me 2 SO concentration less than 0.1%.
Culture of Ventricular Myocytes-Primary cultures of ventricular myocytes from 1-day-old Sprague-Dawley rats were performed by gentle, serial trypsinization, as described previously (25) with modifications (26). A preplating step was included to reduce the number of contaminating non-myocytes. Myocytes were plated at 800 cells/mm 2 in 35-or 60-mm dishes (Falcon). Myocytes represented 90 -95% of total adhering cells. Division of non-myocytes was prevented by the addition of 0.1 mM bromodeoxyuridine to medium for the first 4 days of culture. Cells were maintained at 37°C in a 1% CO 2 incubator in M-199 medium (Life Technologies, Inc.) containing 10% fetal bovine serum (Hyclone), 50 units/ml penicillin, and 80 M vitamin B 12 for the first 4 days. Vitamin C (80 M) was present from day 2. On day 4, myocytes were placed in defined (10 g/ml insulin, 10 g/ml transferrin, 80 M vitamin C, 50 units/ml penicillin, and 80 M vitamin B 12 ) M-199 medium. Myocytes exhibited a spontaneous contraction rate of approximately 250 -300 beats/min, and cultures with a slower contraction rate were not used. The higher rate of contraction versus that seen by Simpson and Savion (25) is partly the result of the increased density of cultures used here. Experiments were performed on days 5 and 6 of culture.
Induction of Ischemia-Ischemia was induced in a humidified 37°C incubator within an air-tight Plexiglas glove box (Anaerobic Systems) maintained with 0.2-0.5% O 2 , 1% CO 2 , and the balance N 2 . Medium (defined minimal essential medium and Hank's balanced salt solution without glucose) was equilibrated to low O 2 within the glove box for at least 90 min before commencing experiments. Inside the glove box, cells were washed twice with warm, preequilibrated medium before the addition of incubation medium (1.5 ml/35-mm dish). O 2 was measured using an electronic gas analyzer OXOR ® II or Fyrite ® (both from Bacharach).
Lactate Dehydrogenase (LDH) Assay-After ischemic or normoxic treatments, incubation medium was stored at 4°C, and the same volume of cold buffer (10 mM Tris-HCl, pH 7.4, 1 mM EDTA) was added to the cells. The cells were scraped and lysed by trituration. Lysates were centrifuged at 4°C at 16,000 ϫ g for 15 min, and the supernatant was stored at 4°C. LDH activity was measured from both medium (released LDH) and cell lysate (retained LDH) using a spectrophotometric assay (Sigma). Results were expressed as released LDH activity as a percent of total (released plus retained) LDH activity.
Analysis of the Phosphorylation State of p38 MAPK, ERK2, and JNK-After treatment, cells were placed on ice, and the incubation medium was aspirated and discarded. Cells were washed once with cold phosphate-buffered saline. Laemmli loading buffer at 2 ϫ concentration (2% SDS, 20% glycerol, 0.04 mg/ml bromphenol blue, 0.12 M Tris-HCl, pH 6.8, 0.28 M ␤-mercaptoethanol) was added (150 l/35-mm dish). Cells were scraped and lysed by trituration. Samples were frozen in dry ice/ethanol then transferred immediately to Ϫ80°C. Prior to electrophoresis, samples were heated to 95°C for 5 min. Electrophoresis was performed with approximately 20 g of protein/sample on 10% low ratio bisacrylamide (100:1, acrylamide:bisacrylamide). After Western blotting, filters were probed sequentially with dual phospho-p38 MAPK (Thr 180 Tyr 182 ), total p38 MAPK (both from New England Biolabs), or ERK2 antiserum, and immunoreactivity was detected by enhanced chemiluminescence. ERK2 antiserum (DC3; obtained from Dr. J. E. Ferrell, Stanford University) was raised against Xenopus ERK2 (27) and recognizes both non-phosphorylated and phosphorylated forms of ERK2. Antiserum was used at a dilution of 1/500. Blots were stripped by incubation in 62.5 mM Tris-HCl, pH 6.8, 100 mM ␤-mercaptoethanol, 2% SDS for 30 min at 50°C followed by two washes with phosphatebuffered saline and 0.05% Tween, then blocking. Activation of p38 MAPK requires phosphorylation on both Thr 180 and Tyr 182 , which is specifically recognized by the antibody used, and therefore activation is expressed as the ratio of dual phospho-p38 MAPK to total p38 MAPK immunoreactivity, which allows correction for differences in protein loading. Because dual phospho-p38 immunoreactivity was usually undetectable in control cells, results were normalized to the ratio of the 10-min ischemia sample.
Phosphorylation of JNK was assayed as above, except that 50 g of cell lysate protein was electrophoresed, and filters were probed with anti-active JNK (Promega).
Assay of p38 MAPK Activity-The p38 MAP kinase assay kit from New England Biolabs was used with a few modifications. Sodium orthovanadate (2 mM), 1 mM phenylmethylsulfonyl fluoride, 25 g/ml aprotinin, 25 g/ml leupeptin, and 20 g/ml soybean trypsin inhibitor were added fresh to lysis buffer, and the cells were lysed by trituration. p38 MAPK was immunoprecipitated, its catalytic activity determined using the in vitro kinase assay to phosphorylate recombinant activating transcription factor-2 (ATF2), and the reaction mixture separated by SDS-PAGE. Western blots were probed with the ATF2 antibody provided in the kit (specific for phospho-Thr 71 ), and immunoreactivity was detected by enhanced chemiluminescence. Filters were stripped as above and reprobed with anti-p38 MAPK. The ratio of phospho-ATF2 to p38 MAPK immunoreactivity was determined for each sample, and then results were expressed as fold activation over control.
Immunoprecipitation of Dual Phosphorylated p38 MAPK-Cardiac myocytes (one 100-mm dish/treatment) were treated, and then the incubation medium was aspirated and discarded. Cells were washed once with cold phosphate-buffered saline and then scraped into 800 l of lysis buffer (10 mM Tris-HCl, pH 7.4, 1 mM EDTA, 1 mM EGTA, 1% Triton X-100, 2 mM sodium orthovanadate, 1 mM phenylmethylsulfonyl fluoride, 25 g/ml aprotinin, 25 g/ml leupeptin, and 20 g/ml soybean trypsin inhibitor) and lysed by trituration. Samples were extracted on ice for 15 min, and then cell debris was removed by centrifugation at 15,000 ϫ g for 10 min. A sample of supernatant was retained for electrophoresis. 20 l of dual phospho-p38 MAPK (Thr 180 Tyr 182 ) antibody (New England Biolabs) was added to the remainder of the supernatant, and samples were rotated overnight at 4°C. 60 l of a 50% slurry of protein A-Sepharose CL-4B (Amersham Pharmacia Biotech) was added, and the samples were rotated for 2 h at 4°C. The immunoprecipitates were washed four times with phosphate-buffered saline, electrophoresed on 10% low ratio bisacrylamide SDS-PAGE, then subjected to Western blot analysis with anti-p38 MAPK, anti-dual phospho-p38 MAPK (Thr 180 Tyr 182 ), or anti-p38␣ (Santa Cruz Biotechnology).
Densitometric Analysis-Autoradiographs were scanned using an ArcusII flatbed scanner (AGFA) with FotoLookPS 2.07.2, and the band density was analyzed by NIH Image TM .
ATF2 Phosphorylation-Preparation of nuclear proteins from cardiac myocytes was performed exactly as described by Clerk and Sugden (28). Approximately 50 g of nuclear protein was electrophoresed on 8% SDS-PAGE, and Western blot analysis was performed. ATF2 was detected using phosphorylation state-independent antibody from Santa Cruz Biotechnology.
Cell Viability Assay-Cell death from ischemic or normoxic incubations was assessed using two dyes that distinguish between live and dead cells. Calcein acetoxymethyl ester (calcein AM; Molecular Probes) and propidium iodide (PI) were added to the incubation medium at final concentrations of 2 M and 1 g/ml, respectively, and dishes incubated at 37°C for 15 min (ischemic samples were maintained under ischemic conditions during this incubation). Cells were viewed using a Zeiss microscope and a 40 ϫ objective and were scored as live (green cytosolic fluorescence from calcein AM) or dead (red nuclear fluorescence from PI).
Isolation of DNA and Agarose Gel Electrophoresis-After ischemic or normoxic incubation, cells were scraped into incubation medium to allow retention of any floating cells and were harvested by centrifugation. DNA was prepared by standard techniques (29). Identical amounts of DNA (2 g) were electrophoresed through 1.8% agarose and DNA visualized on a UV transilluminator.
Detection of CPP32 Immunoreactivity-Samples were treated exactly as for analysis of p38 MAPK immunoreactivity except that samples were submitted to 12% SDS-PAGE, and filters were probed with anti-CPP32 (H-277) from Santa Cruz Biotechnology.
Statistical Analysis-Data were compared using Student's t test for observations between two samples with unequal variance, with onetailed distribution. A p value of less than 0.05 was considered significant.

Activation of MAPKs and Stress-activated Kinases by
Ischemia-Activation of MAPKs in primary neonatal rat cardiac myocytes in response to simulated ischemia was investigated.

p38 MAPK Mediates Cardiac Myocyte Cell Death from Ischemia
The model used combines two properties of ischemia: decreased energy source, because incubations are performed in the absence of glucose, and hypoxia, with oxygen levels between 0.2 and 0.5%. p38 MAPK activation was estimated by Western blot analysis using an antibody that specifically recognizes the dual phosphorylated (on residues Thr 180 and Tyr 182 ), active form of the enzyme. Antibody recognizing p38 MAPK regardless of its phosphorylation state was used to normalize for differences in protein loading. This antibody is specific for the ␣ isoform of p38 MAPK, and the level detected remained constant throughout ischemia. Dual phosphorylation of p38 MAPK was observed within 10 min of ischemia, remained maximal until 30 min, then decreased but remained above basal until 180 min of ischemia (Fig. 1, A and B). At later time points, dual phospho-FIG. 1. Ischemia results in phosphorylation and activation of p38 MAPK. Panel A, cardiac myocytes were incubated in ischemic conditions for 10 -420 min or were harvested without any treatment (control). Cell lysates were prepared as described under "Experimental Procedures," and approximately 30 g of protein was separated by 10% SDS-low bis ratio PAGE. The same filter was probed sequentially with antibodies against p38 MAPK or phospho-p38 MAPK. This result is representative of between three and six experiments for different time points (for graphic representation of combined results, see panel B). Also shown is a 15-min incubation with 0.2 M anisomycin or 10 min with 100 nM 4␤-phorbol 12-myristate 13-acetate treatment. Note that in other experiments, anisomycin resulted in higher phosphorylation of p38 MAPK than seen here, giving about the same level as 10 min of ischemia. Panel B, the ratio of phospho-p38 MAPK to p38 MAPK immunoreactivity was determined from Western blot analysis as represented in panel A. The ratios were then normalized to 1 for the 10-min ischemia sample. The data represent between three and six experiments, each performed with a different preparation of cells and are plotted as the mean (ϮS.E.). All ischemia time points are significantly different from normoxia (p Ͻ 0.05). Panel C, cardiac myocytes were fed with fresh medium and then incubated in normoxic conditions for 10 -420 min, ischemic conditions for 10 min, or were harvested without any treatment (control). Samples were then treated as described for panel A. Panel D, the ratio of phospho-p38 MAPK to p38 MAPK immunoreactivity was determined from Western blot analysis as represented in panel C. The ratios were then normalized to 1 for the 10-min ischemia sample such that panels B and D have the same scale. Data represent between three and five experiments, each performed with a different preparation of cells, and are plotted as the mean (ϮS.E.). Panel E, cardiac myocytes underwent an ischemic incubation for 20 min or were treated with 0.2 M anisomycin for 30 min or were harvested without treatment (control). Cells were lysed and p38 MAPK activity measured as described under "Experimental Procedures." The same Western blot was probed sequentially with anti-phospho-ATF2 and anti-p38 MAPK. The ratio of activation was calculated as phospho-ATF2:p38 MAPK immunoreactivity and was normalized to 1 for the control sample. This result is representative of three experiments, each with a different preparation of cells. rylation increased again with a peak at 240 min and remained high for 420 min (Fig. 1, A and B).
We and others (30) have observed that when cardiac myocytes are fed with fresh medium, they cease contracting for a period of time. In our study, cells stopped contracting for approximately 20 min, then spontaneous contraction recovered gradually to a normal rate within 60 -90 min. Because this corresponds with the timing of the first phase of p38 MAPK activation, and a change of medium is required to induce ischemic conditions, we examined p38 phosphorylation levels after changing incubation medium and maintaining cells under normoxic conditions. Transient phosphorylation of p38 MAPK was observed after simply feeding fresh medium (Fig. 1, C and  D). Cardiac myocytes can be maintained healthily in culture for up to 8 days with multiple changes of medium. Although this does not show that the initial ischemic p38 MAPK activation and that induced by changing medium are equivalent, these results do demonstrate that transient activation of p38 MAPK can occur without long term harmful effects to cardiac myocytes.
To confirm that dual phosphorylation of p38 MAPK during ischemia truly reflected activation, p38 MAPK was immunoprecipitated and used in an in vitro kinase assay with recombinant ATF2 as a substrate. We observed more than a 6-fold increase in the phosphorylation of ATF2 over basal after 20 min of ischemia (Fig. 1E) and more than a 4-fold increase after 25 or 30 min of ischemia (n ϭ 1, data not shown). A similar activation ratio was obtained by incubating cells with anisomycin, an activator of p38 MAPK (Fig. 1E). Thus, p38 MAPK is indeed activated during ischemia.
To determine if ischemia-induced phosphorylation is unique to p38 MAPK, we examined phosphorylation of other MAPKs. Phosphorylation of ERK2 (p42 MAPK) was estimated using a gel electrophoresis mobility shift assay with an antibody that detects both inactive (non-phosphorylated) and active (dual phosphorylated) ERK2. The reduced mobility form of ERK2, indicating phosphorylation as seen with 4␤-phorbol 12-myristate 13-acetate treatment, was not observed at any time during prolonged ischemia ( Fig. 2A). Similarly, probing with antiactive JNK, which detects the dual phosphorylated active forms of both JNK1 and JNK2, showed little or no activation of JNK in 10 -240 min of ischemia (Fig. 2B) and no activation in 300, 360, or 420 min of ischemia (n ϭ 1, data not shown).
Taken together, our results indicate that changing medium transiently activates p38 MAPK, whereas ischemia results in a transient, then sustained activation of p38 MAPK, but not ERK2 or JNK, in primary cultures of neonatal rat cardiac myocytes; these results are consistent with previous studies performed in whole heart (6 -8).
Significance of p38 MAPK Activation in Ischemia-Release of the cytosolic enzyme LDH, caused by cell membrane leakage, was used to assess cell damage resulting from ischemia. Under the conditions used in this study, 7-9 h of ischemic incubation resulted in the release of between 45 and 60% of cellular LDH, FIG. 3. Anisomycin protects cardiac myocytes from ischemic injury. Cardiac myocytes were pretreated with 0.2 M anisomycin or 0.1% Me 2 SO vehicle, then incubated in ischemia for 7-9 h in the presence (n ϭ 2, duplicates) or absence (n ϭ 3, duplicates) of 0.2 M anisomycin, as indicated. LDH activity from medium and cell lysate was measured as described under "Experimental Procedures." Results are expressed as LDH released into the medium as a percent of the total LDH activity and then normalized to 100% for vehicle-treated samples. Data are plotted as the mean (ϮS.E.). *p Ͻ 0.05; **p Ͻ 0.025 versus vehicle-treated sample. The two right columns are not significantly different from each other.

FIG. 2. Ischemia does not result in phosphorylation of ERK2 or JNKs.
Panel A, as for Fig. 1A, except that filters were probed with ERK2 antiserum. 4␤-Phorbol 12-myristate 13-acetate and anisomycin treatment are as described for Fig. 1A. Panel B, as for Fig. 1A, except that 50 g of protein was loaded, and the filter was probed with anti-active JNK. As a positive control, cells were treated with 0.2 M anisomycin for 30 min. This filter was coincubated with anti-p38 MAPK, showing equal protein loading. This result is representative of three experiments, except that in one experiment a small amount of active JNK was detected in the 30-min ischemia sample; however, this represented only approximately 10% of active JNK seen with anisomycin treatment.
p38 MAPK Mediates Cardiac Myocyte Cell Death from Ischemia whereas 7-9 h of normoxic incubation resulted in release of less than 6% of total LDH. To determine if the activation of p38 MAPK protects myocytes from ischemic stress or mediates the damage from ischemia, we first used anisomycin. Anisomycin is a protein synthesis inhibitor that activates MAPK family members and has been shown to protect myocytes from ischemia (21). This result has been used to implicate p38 MAPK in the protective mechanism (21). To confirm that anisomycin was protective in our model, cells were pretreated with anisomycin and then subjected to ischemia, either in the presence or absence of additional anisomycin. Fig. 3 demonstrates that anisomycin protected myocytes from ischemia-induced injury; LDH release was reduced significantly when anisomycin was present either during the pretreatment only or during both the pretreatment and the prolonged ischemia.
Anisomycin is a nonspecific reagent, and therefore these data do not prove that p38 MAPK activation is the mechanism by which anisomycin protects myocytes. More direct evidence for this would require inhibition of the anisomycin-induced protection by a selective p38 MAPK inhibitor. Therefore, we next used an inhibitor of p38 MAPK, SB 203580, which has been shown to inhibit p38 MAPK selectively over other MAPK family members and several other kinases (23,31). Surprisingly, we found that the presence of SB 203580 during ischemia resulted in a significant dose-dependent decrease in LDH release from myocytes (Fig. 4A). These data indicate that inhibition of p38 MAPK during ischemia protects myocytes from ischemic damage, whereas previous results have suggested that activation of p38 MAPK protects myocytes.
As mentioned previously, activation of p38 MAPK was shown to occur in two phases in these cells (Fig. 1, A and B). To determine if inhibition of p38 MAPK during only one, or both phases, was necessary for the protection seen with SB 203580, the p38 MAPK inhibitor was added at different times during ischemia, and the effect on injury was examined. Adding the inhibitor to cells 45 min after the start of ischemia (which is after the first peak of p38 MAPK phosphorylation, Fig. 1B) gave a level of protection not significantly different from that seen when the inhibitor was present from the start of ischemia (Fig. 4B). SB 203580 is a reversible inhibitor; isolated p38 MAPK can be washed free of inhibitor and be fully active (23,31). Therefore, to examine inhibition during the first phase only, cells were incubated with SB 203580 and then washed free of inhibitor after 45 min of ischemia. Using this protocol, we found that the presence of the inhibitor during only the first 45 min resulted in cell damage similar to that seen in vehicletreated cells (Fig. 4B). Thus, SB 203580 induced protection of myocytes from ischemia when present during only the second sustained phase of p38 MAPK activation, but not when present during only the first, transient phase of p38 MAPK activation. This suggests that sustained p38 MAPK activation results in cell damage.
LDH release is a commonly used marker for cell damage, but to confirm that decreased LDH release reflected a protection FIG. 4. p38 MAPK inhibitor protects neonatal cardiac myocytes from ischemic injury. Panel A, cardiac myocytes were incubated in ischemia or normoxia in the presence of 2, 5, or 10 M SB 203580 or 0.1% Me 2 SO vehicle. SB 203580 or Me 2 SO was added at the start of the incubation. After 7-9 h, LDH activity was measured from medium and cell lysate as described under "Experimental Procedures." Results were expressed as LDH released into the medium as a percent of the total LDH activity and then normalized to 100% for vehicletreated ischemia samples. The number of experiments, performed in duplicate, is indicated above the bars. *p Ͻ 0.05, **p Ͻ 0.0001 versus vehicle-treated sample. Panel B, as in panel A, except that for the left three bars, 10 M SB 203580 (SB) was added 45 min after the start of ischemic incubation or at the start of ischemia for comparison. For the right panel, 10 M SB 203580 or 0.1% Me 2 SO vehicle was added at the start of ischemia. After 45 min, the incubation medium was removed from all dishes, and the cells were washed twice with fresh medium, preequilibrated to hypoxic conditions. Cells were then incubated with vehicle or 10 M SB 203580 for the remainder of ischemia, as indicated. For both panels, data represent the mean (ϮS.E.) from three experiments performed in duplicate. *p Ͻ 0.001 versus vehicle-treated sample. ns indicates that the difference is not significant. Panel C, cardiac myocytes were incubated in ischemia or normoxia in the presence of 10 M SB 203580 or 0.1% Me 2 SO vehicle for 7-8 h, as indicated. Cells were stained with calcein AM and PI and then scored as live (green cytosolic fluorescence) or dead (red nuclear fluorescence). Data represent the percent of viable cells and are plotted as the mean (ϮS.E.) from three experiments. In each experiment, more than 800 cells were counted for each treatment. *p Ͻ 0.05.

p38 MAPK Mediates Cardiac Myocyte Cell Death from Ischemia
from cell death, we used a cell viability assay. Live cells are distinguished by the conversion of calcein AM to fluorescent calcein by intracellular esterases, but the intact membrane excludes PI. Dead cells are distinguished by entry of PI through damaged membrane and fluorescence of PI on binding to nucleic acid. We demonstrated that coincubation of myocytes with 10 M SB 203580 during ischemia significantly decreased cell death at two time points (Fig. 4C). It is important to note that cell death is delayed but not prevented; after 8 h of ischemia, cell death is reduced significantly by the presence of the inhibitor, but it is increased from basal cell death under normoxic conditions. Thus, inhibition of p38 MAPK delays cell injury and death resulting from ischemia.
Because it appeared that the consequence of the first and second phases of p38 MAPK activation differed, we attempted to determine whether the same or different isoforms of p38 MAPK are activated during these two different phases. By immunoprecipitation of dual phospho-p38 MAPK (Thr 180 Tyr 182 ; this antibody detects both activated ␣ and ␤ isoforms of p38 MAPK) at different times during ischemia, we determined that p38␣ is activated in both phases during ischemia (Fig. 5). p38␤ was only weakly detected in cell lysates, and therefore we could not determine if p38␤ was also present in any of the immunoprecipitates.
Ischemia Results in Apoptosis in Neonatal Rat Cardiac Myocytes-Staining with PI as described above in the cell viability assay allows visualization of nuclei and could potentially be used to distinguish morphologically between apoptotic and necrotic cells. Apoptotic cells generally show condensed, fragmented nuclei, whereas necrotic cells have normal nuclei. However, after ischemia in this cell type, many nuclei appeared slightly condensed but not fragmented and therefore were difficult to classify clearly as apoptotic or necrotic.
Previous studies have documented both apoptosis and necrosis induced by myocardial infarction (12,13). To examine if our model of ischemia induces apoptosis in cardiac myocytes, cells were subjected to a DNA laddering assay to assess the extent of DNA fragmentation. Identical amounts of DNA, from cells treated under either normoxic or ischemic conditions for 8 h, were visualized after agarose gel electrophoresis. In control samples incubated under normoxic conditions, a low level of basal DNA fragmentation was observed (Fig. 6A). However, in the ischemic sample, there was a decrease in high molecular weight DNA and a corresponding increase in low molecular weight DNA when compared with the normoxic sample (Fig.  6A). The low molecular weight DNA both in normoxia and ischemia showed the hallmark intranucleosomal laddering of apoptosis. Therefore, although it was not possible to quantitate apoptosis versus necrosis, we clearly demonstrated that ischemia induces apoptosis in neonatal cardiac myocytes. These data do not rule out the possibility that some cells undergo necrosis during ischemia.
SB 203580 Inhibits Activation of Caspase-3-To confirm that inhibition of p38 MAPK by SB 203580 delayed apoptosis, we examined the activation state of caspase-3 (also known as CPP32, Yama, or apopain). This enzyme has been identified as a key protease during the early stages of apoptosis and is activated by degradation of the 32-kDa proenzyme to approximately 17-and 12-kDa subunits that heterodimerize to give active enzyme (32). Using Western blot analysis, we determined that caspase-3 degradation is delayed in cardiac myocytes treated with SB 203580 compared with vehicle-treated cells (Fig. 6B). Therefore, apoptosis is delayed by inhibition of FIG. 5. p38␣ is phosphorylated during both phases of ischemia. Cardiac myocytes underwent ischemic incubation for 30, 240, or 300 min or were harvested without any treatment (0). Dual phospho-p38 MAPK was immunoprecipitated (IP), and approximately 5% of cell lysate before immunoprecipitation and all of the immunoprecipitated material were electrophoresed on 10% SDS-low bis ratio PAGE as described under "Experimental Procedures." The filter was probed with anti-p38␣ (Santa Cruz Biotechnology; SC), anti-p38 (from New England Biolabs; NEB), which is also specific for p38␣, or anti-dual phospho-p38 MAPK. This result is representative of two experiments performed with different preparations of cells. SB 203580 Blocks Phosphorylation of p38 MAPK Substrate-To confirm that SB 203580 inhibits p38 MAPK in this system, we examined phosphorylation of an endogenous p38 MAPK substrate, ATF2. p38 MAPK phosphorylates ATF2 on Thr 69 and Thr 71 , resulting in an electrophoretic mobility shift (3,28). Phosphorylation of these residues is essential for increased transcriptional activity of ATF2 (33). ATF2 is also phosphorylated on these residues by JNK (33), but, because JNK is not activated by ischemia in these cells (Fig. 2B), p38 MAPK is expected to be a major ATF2 kinase under ischemic conditions.
In nuclear extracts from control cells, maintained under normoxic conditions, the majority of ATF2 migrates at 69 kDa (Fig. 7, band 1), with a minor amount migrating with a reduced electrophoretic mobility (Fig. 7, band 2). Incubation under ischemic conditions resulted in ATF2 phosphorylation seen by an increase in intensity of band 2 and the appearance of an additional reduced mobility band (Fig. 7, band 3). When 10 M SB 203580 was present during the ischemic incubation, these phosphorylation events were partially inhibited; band 3 is no longer detectable, and bands 1 and 2 are increased in intensity compared with ischemia only. The partial inhibition of ATF2 phosphorylation may indicate incomplete inhibition of p38 MAPK or that another ATF2 kinase, which is not inhibited by SB 203580, is activated by ischemia. These data demonstrate that the p38 MAPK substrate, ATF2, is phosphorylated during ischemia and that the phosphorylation events are sensitive to inhibition by SB 203580. DISCUSSION In response to ischemia, cells activate biochemical pathways that allow adaptation to this stressful environment. However, on prolonged ischemia, these protective mechanisms may not be sufficient to maintain normal cellular function, and cell injury and death follow. Therefore, signal transduction pathways activated during ischemia may be either protective or part of the signal that leads to cell death. In this study, we demonstrated that ischemia induces both a transient and sustained activation of p38 MAPK in neonatal rat cardiac myocytes. In contrast, ERK2 and JNK are not activated during ischemia. By using the specific p38 MAPK inhibitor SB 203580, we demonstrated that sustained activation of p38 MAPK is deleterious to the cells and at least partly mediates apoptosis.
At least four members of the p38 MAPK family have been identified, and it is probable that different isoforms have specific physiological functions (1, 22, 34 -37). The isoforms ␣ (also termed p38, CSBP, or RK) and ␤ are both expressed in heart tissue. Although these isoforms share approximately 74% sequence identity, they have been suggested to have opposing functions in cardiac myocytes (17). Expression of an activated mutant of MAPK kinase 3b, an upstream activator of both ␣ and ␤ isoforms of p38 MAPK, results both in a hypertrophic response and to apoptosis in neonatal rat cardiac myocytes (17). By using coexpression of individual activated p38 MAPK isoforms or of dominant negative inhibitory fragments, the same study showed that apoptosis appears to be mediated by the ␣ isoform, whereas the hypertrophic response is mediated by the ␤ isoform. In addition, suppression of p38␤ using the dominant negative fragment results in an increase in cell death, suggesting that this isoform can function to promote cell survival. In another study, overexpression of MAPK kinase 6, a different selective activator of p38 MAPK, protects neonatal cardiac myocytes against apoptosis (18). Thus, in cardiac myocytes it appears that p38␤ protects against, whereas p38␣ promotes, apoptotic cell death.
In our study, we observed that transient activation of p38␣ does not have a deleterious effect on the cell, whereas sustained activation of p38␣ induces apoptosis. However, we could not rule out activation of p38␤ or other p38 MAPK isoforms during one or both phases of p38 MAPK activation seen during ischemia. Therefore, it is possible that the two phases differ not only in duration, but also in the balance of isoforms activated. A short period of ischemia is not detrimental to cardiac myocytes and paradoxically protects against subsequent prolonged ischemia, as discussed in the Introduction. In models similar to that used in our study, preconditioning neonatal rat cardiac myocytes with 25 or 30 min of ischemic incubation followed by 30 min of normoxic recovery was shown to protect cultured neonatal cardiac myocytes from ischemia (30,38). Thus, from the data presented here, preconditioning would be expected to result in only the first phase of p38␣ activation, and previous studies have shown that p38 MAPK is necessary for the protective effect of preconditioning (21,24). It is intriguing that p38 MAPK appears to play a role both in protection of myocytes against injury and in mediating cellular apoptosis, and we propose that transient versus sustained activation of p38 MAPK, possibly in combination with activation of different isoforms, determines these different cellular effects. The most likely explanation is that the transient p38 MAPK activation that occurs on initiation of ischemia represents an adaptive response of the cell. Cardiac myocytes clearly can adapt to ischemic stress as shown by the fact that they can be protected by preconditioning. This hypothesis provides an explanation for the apparent discrepancy between this study, where SB 203580 protects against ischemia, and the previous studies in which SB 203580 inhibits preconditioning protection (21,24).
Differential cellular effects of transient versus prolonged activation have been demonstrated previously for MAPK family members, for ERK2 in primary rat hepatocytes (39), and for JNK in rat mesangial cells (40). In addition, transient p38 MAPK activation is not sufficient to induce neuronal differentiation in rat pheochromocytoma PC12 cells, whereas sustained activation is sufficient (41). Interestingly, a transient activation of the ERK/MAPK pathway is additionally required to allow diffentiation (41). In a similar way, it is likely that other proteins activated or inactivated at different times during ischemia regulate or act in combination with p38 MAPK. FIG. 7. SB 203580 partially reverses ischemia-induced phosphorylation of ATF2. A nuclear fraction (50 g of protein) from cardiac myocytes incubated under normoxia (C), ischemia (I), or ischemia plus 10 M SB 203580 (IϩSB) for 60 min was electrophoresed on 8% SDS-PAGE and detected with anti-ATF2. An upward mobility shift (to bands 2 and 3) was used as an indicator of ATF2 phosphorylation (3,28). This result is representative of three independent experiments.

p38 MAPK Mediates Cardiac Myocyte Cell Death from Ischemia
For example, the prolonged phase of p38 MAPK activation may be caused by inactivation of specific regulatory phosphatases or, alternatively, degradation of an antiapoptotic protein may be required before p38 MAPK-mediated apoptosis can proceed. These possibilities can now be explored.
Although adult and neonatal cardiac myocytes can differ in their responses, it is unlikely that the protection seen here with SB 203580 would not translate to adult cells. A study performed in isolated adult rat hearts demonstrates that the presence of SB 203580 during ischemia preserves cardiac function during ischemia and improves postischemic recovery of cardiac function (42). Therefore, it is predicted that SB 203580, or a novel selective inhibitor of p38␣, will prove useful to protect against damage from ischemic episodes in adult animals, both by decreasing cell death and by improving cardiac function.
We cannot rule out the possibility that SB 203580 targets another kinase, in addition to p38 MAPK, which mediates cell death. Selectivity has been demonstrated against other kinases (22,23), but it has recently been reported that SB 203580 inhibits cardiac JNK2-related isoforms, albeit with a higher IC 50 than for p38 MAPK (43). It is unlikely that inhibition of JNK explains the protection reported here because we and others (6,7) have shown that JNK2 is not activated by ischemia.
In conclusion, our results strongly support a role for sustained p38 MAPK activation in mediating apoptosis induced by ischemia. This study used an experimental design that does not require overexpression of the proteins involved but rather inhibition of endogenous kinase. In addition, we suggest that transient activation can have very different cellular consequences from sustained activation of p38 MAPK in cardiac myocytes.