Oxygen-sensitive δ-Opioid Receptor-regulated Survival and Death Signals

The detrimental effect of severe hypoxia (SH) on neurons can be mitigated by hypoxic preconditioning (HPC), but the molecular mechanisms involved remain unclear, and an understanding of these may provide novel solutions for hypoxic/ischemic disorders (e.g. stroke). Here, we show that the δ-opioid receptor (DOR), an oxygen-sensitive membrane protein, mediates the HPC protection through specific signaling pathways. Although SH caused a decrease in DOR expression and neuronal injury, HPC induced an increase in DOR mRNA and protein levels and reversed the reduction in levels of the endogenous DOR peptide, leucine enkephalin, normally seen during SH, thus protecting the neurons from SH insult. The HPC-induced protection could be blocked by DOR antagonists. The DOR-mediated HPC protection depended on an increase in ERK and Bcl 2 activity, which counteracted the SH-induced increase in p38 MAPK activities and cytochrome c release. The cross-talk between ERK and p38 MAPKs displays a “yinyang” antagonism under the control of the DOR-G protein-protein kinase C pathway. Our findings demonstrate a novel mechanism of HPC neuroprotection (i.e. the intracellular up-regulation of DOR-regulated survival signals).

Neuronal death as a result of neuronal injury following hypoxic/ischemic insults, such as stroke, is an irreversible process that leads to long term neurological deficit. The prevention of neuronal injury is therefore critical in rescuing the brain from neurological disaster. However, clinical strategies that may help mitigate the effects of hypoxic/ischemic injury are still very limited.
One strategy that has been shown to provide effective protection from harmful stress is known as preconditioning. This involves transient, but sublethal, exposure to a stress, resulting in enhanced cellular resistance to subsequent severe stress. It was initially demonstrated in the heart (1) and subsequently found to work in other organ beds (2). The effects of preconditioning have been widely studied in the whole brain (3,4), as well as in vitro in brain slices (5,6) and neuronal cultures (7)(8)(9). Most studies to date have shown the beneficial effects of preconditioning on rescuing neurons from cell injury in response to subsequent severe insults. Hypoxic preconditioning (HPC), 1 for example, caused by lowering the oxygen content or by combined oxygen and glucose deprivation, protects against subsequent hypoxic injury (4,7,8,10). However, a recent study failed to show neuronal protection with HPC treatment (11), suggesting that the neuronal response to preconditioning may vary depending on neuronal situation and involve complex mechanisms. These molecular mechanisms remain unclear, especially with regard to intracellular signal transduction. In this study, we demonstrated that, in neurons in culture, HPCinduced neuroprotection is dependent on specific factors and that the effect is mediated by intracellular up-regulation of ␦-opioid receptor (DOR)-regulated survival signals, which suppress the increased activity of intracellular death signals during severe hypoxia (SH).
DOR is a guanine nucleotide-binding regulatory protein (G protein)-coupled receptor that is widely distributed in different mammalian cells (12). In addition to its role in the regulation of opiate-induced analgesia and opioid dependence (13), DOR may be an important mediator in the energy-conserving state of hibernation (14). Systemic administration of DOR agonists increases the survival time of mice exposed to subsequent hypoxic challenge (15,16). However, the targeted tissue and the underlying mechanisms are unknown, because the systemic distribution of DOR agonists may trigger cell protection in nonneuronal organs of the body (17,18). Recent data from our and other laboratories show that DOR activation is neuroprotective and predisposes neurons to survive under severe stress, such as glutamate-, hypoxia-, or serum deprivation-induced cell injury (19 -21). Moreover, in cortical neurons, we found that DOR inactivation by administration of a specific DOR antagonist causes cell death and further augments neuronal injury during SH, suggesting a crucial role of DOR in maintaining neuronal survival (20). Interestingly, we have found that DOR density is much higher in turtle brain than in rat brain (22) and that turtle neurons are much more tolerant than rat neurons of hypoxic stress (23). These observations prompted us to examine the possibility that DOR expression may be a critical determinant of neuronal survival during stress.
Based on our previous work, we tested the hypothesis that HPC up-regulates DOR expression and enhances DOR-initiated survival signals, thus protecting neurons from subsequent severe stress. Because it is impractical to analyze molecular mechanisms by administering multiple drugs to live animals, we performed this study on neurons in culture and specifically focused on cortical neurons, since we have previously shown that the cortex has the highest DOR density (22,24).

EXPERIMENTAL PROCEDURES
Primary Culture of Rat Cortical Neurons-Cortical neurons were isolated from embryonic Sprague-Dawley rats (embryonic days 16 and 17) as described previously (19,20). All animal treatments were performed according to the guidelines of the Animal Care Committee of Yale University School of Medicine, which is approved by the American Association for Accreditation of Laboratory Animal Care. In brief, after anesthetizing the pregnant dam with halothane, the embryonic rats were removed and immediately rinsed in neurobasal medium (Invitrogen). Under a dissecting microscope, the embryonic brain cortex was carefully removed, soaked in 0.1 M phosphate-buffered saline (PBS) (pH 7.4), containing 0.6% glucose, collected by centrifugation, and treated with trypsin (Invitrogen) for 15 min at 37°C. After neutralizing the trypsin with trypsin inhibitor (Invitrogen) for 5 min, a cell suspension was prepared by repeated passage through a pipette and filtration through an 80-m nylon mesh. It was then diluted with neurobasal medium containing B-27 (Invitrogen), 0.5 mM glutamine (Sigma), 25 M glutamate (Sigma), and 100 IU/ml penicillin and 100 mg/ml streptomycin (both from Invitrogen). The cells were counted by staining with trypan blue (Invitrogen), diluted to a density of 10 6 cells/ml in the same medium, and plated onto poly-D-lysine-precoated dishes. Cells were cultured at 37°C in 21% O 2 and 5% CO 2 in a humidified incubator for 8 days in vitro (DIV), at which time DOR protein expression reached a plateau (see "Results").
Hypoxic Preconditioning and Severe Prolonged Hypoxia-On DIV 8, the cultured neurons were divided into different groups. For HPC treatment, the culture was placed in a 5% O 2 and 5% CO 2 environment in an incubator for 3, 6, or 9 h. For SH treatment, the cultures were exposed to 1% oxygen for 5 days, unless stated otherwise. For HPC plus SH treatment, cells were treated with HPC for the indicated time, followed by SH treatment 24 h later (10).
LDH Assay-The assay kit was obtained from Sigma (228-UV) and used as described previously (20). The advantage of this method is that it allows cell viability to be continuously monitored. In brief, a sample of the culture medium was centrifuged to remove detached cells, and then 100 l of the supernatant was mixed with 1 ml of prewarmed assay reagent at 30°C for 30 s, and the absorbance at 340 nm was read at 30-s intervals for 2 min on a Beckman spectrophotometer (DU-640; Fullerton, CA). The change in absorbance was then expressed as a concentration in units/liter as described in the kit.
Measurement of Cytoplasmic Histone-associated DNA Fragment-We used a cell death detection enzyme-linked immunosorbent assay kit (Roche Applied Science) to perform the photometric measurement of cytoplasmic histone-associated DNA fragment (mono-and polynucleosomes) after inducing cell death. Following the manufacturer's instructions, the neuronal cells were collected and lysed, and then, after centrifugation at 20,000 ϫ g for 10 min at 4°C, the supernatant (cytoplasmic fraction) was carefully transferred to another tube and diluted 10-fold. For a positive control, three independent neuronal cultures were treated for 24 h with 1, 3, or 10 M camptothecin (Tocris) to induce cell death. One hundred microliters of the diluted sample mixture was added to a microtiter plate precoated with anti-histone antibody, which was then incubated at room temperature for 2 h. After washing, peroxidase-conjugated anti-DNA antibodies were added for 1.5 h at room temperature to react with the DNA in the bound nucleosome. The amount of bound peroxidase was quantified photometrically with 2,2Ј-azino-di-3-benzthiazoline sulfonate at 405 nm after washing.
The extent of cell death was expressed as an enrichment factor calculated as the ratio of the absorbance of the sample (containing dying/dead neurons) to that of the corresponding control neurons. Estimation of Cell Viability-Neuronal viability was estimated using a staining kit (L-3224; Molecular Probes, Inc., Eugene, OR) as described previously (20). In brief, a mixture of calcein AM and ethidium homodimer-1 was added to the neurons. Live neurons take up the cellpermeant calcein AM and hydrolyze it with an intracellular esterase to generate a cleavage product of calcein with a green fluorescence. Ethidium homodimer-1 enters injured or dead neurons with damaged plasma membrane and binds to nucleic acids, giving an enhanced red fluorescence. The results were examined using a fluorescence microscopy system (Zeiss Axiovert 25, Sony Progressive 3CCD and camera adapter CMA-D2) with blue excitation at 488/515 nm and green excitation at 514/550 nm.
Immunoblots-After removing the culture medium, the neuronal cells were washed twice with PBS and then suspended in PBS containing a protease inhibitor mixture (1 M pepstatin, 200 M phenylmethylsulfonyl fluoride, 1 M leupeptin, and 1 mM EDTA; Roche Applied Science) followed by homogenization with a pellet pestle (Kontes, Vineland, NJ). All procedures were performed on ice as described previously (26). The mixture was lysed by freeze-thawing three times and then centrifuged at 620 ϫ g for 5 min to remove cellular debris, and samples of the supernatant were subjected to electrophoresis. For cytochrome c immunoblots to measure release from the mitochondria into the cytosol, the supernatant was further centrifuged at 15,000 ϫ g for 20 min to prepare the cytosolic fraction. The protein concentration was determined using a Bio-Rad D C protein assay kit (Hercules, CA).
Semiquantitative Reverse Transcription-PCR to Detect DOR mRNA Expression-Total cellular RNA in the cultured neurons was extracted using TRIzol reagent (Invitrogen), as described in the manufacturer's protocol. The RNA pellet was washed twice with 75% ethanol, dried, and resuspended in diethylpyrocarbonate-treated water. cDNA was synthesized at 42°C for 47 min using 5 g of the total RNA sample, 0.5 g of poly(dT) 15  DOR and glyceraldehyde 3-phosphate dehydrogenase cDNAs were amplified using specific primers designed as previously described (27). The sequences for DOR were 5Ј-GCT GTG CTC TCC ATT GAC TAC-3Ј and 5Ј-GAT GTC CAC CAG CGT CCA GAC-3Ј (transcript product of 480 bases), and those for glyceraldehyde-3-phosphate dehydrogenase were 5Ј-TTA GCA CCC CTG GCC AAG G-3Ј and 5Ј-CTT ACT CCT TGG AGG CCA TG-3Ј (transcript product of 535 bases). The reverse transcription product (2.5 l) was amplified with a mixture of primers and 2.5 units of REDTag DNA polymerase (Sigma) for 30 cycles under conditions of 1 min at 95°C for denaturation, 50 s at 58°C for annealing, and 1 min at 72°C for extension. The reaction products were electrophoresed through a 2% agarose gel and visualized by ethidium bromide staining. The densities of the bands were measured using an image analytic system (Alpha Innotech, San Leandro, CA). DOR mRNA levels were expressed as the ratio of the DOR and glyceraldehyde 3-phosphate dehydrogenase PCR products.
Measurement of Leucine Enkephalin-The intraneuronal content of leucine enkephalin (LE) and the amount of LE release into the culture medium were determined using an enzyme-linked immunoassay kit from Phoenix Pharmaceuticals (Belmont, CA). The culture medium was collected by aspiration, and the neurons in the dish were scraped off after three washes with PBS. LE in both the medium and the cells was measured using the peptide extraction and assay procedures described by the kit manufacturer. First, 1% trifluoroacetic acid was added to the samples, which were then mixed thoroughly and centrifuged. The supernatant was applied to a Sep C18 column equilibrated with 60% acetonitrile in 1% trifluoroacetic acid, and LE was eluted from the column using the same solution, which was then removed by drying under nitrogen. The samples were reconstituted in 0.5 ml of assay buffer. Duplicate samples and standards were mixed with biotinylated anti-LE antibody in a 96-well plate. The plate was then incubated for 2.5 h at room temperature. After five washes with the buffer, streptavidin-horseradish peroxidase was added, and the plate was incubated for 1 h at room temperature. After color development using peroxidase substrate, the reaction was stopped by the addition of 2 N HCl, and the optical density of the wells was read at 450 nm. The values of LE are expressed as ng (mg of protein) Ϫ1 .
Statistics-All experiments were performed on 3-6 (n value in the text) independent batches of neuronal cultures prepared on separate days. The data throughout are expressed as the mean Ϯ S.E. Statistical analysis was performed using the Newman-Keuls test of analysis of variance for multiple comparisons. A significance level of 5% was used.

HPC-induced Neuroprotection Is Dependent on the Duration
of Preconditioning-Neuronal responses to hypoxia depend greatly on the duration of the stress (27). Since results are inconsistent regarding HPC-induced protection (10, 11), we asked whether the effects of HPC depend on the duration of treatment by subjecting cortical neurons to different durations of HPC. As shown in the upper panel of Fig. 1A, when neuronal cells were exposed to prolonged SH (1% oxygen), LDH leakage, a reliable index of neuronal injury (19,20), increased to 27 Ϯ 4, 199 Ϯ 15, and 353 Ϯ 34% after 1, 3, and 5 days of SH as compared with control levels at the same neuronal age (all p Ͻ 0.01 as compared with the control). In neurons preconditioned with mild hypoxia (5% oxygen) for 6 h (HPC6 in Fig. 1) before SH, injury was attenuated, as shown by a significant reduction in LDH release as compared with SH alone (102 Ϯ 13 and 106 Ϯ 18% over the control, respectively, after 3 and 5 days of SH, both p Ͻ 0.001 as compared with the SH group without HPC). No appreciable neuroprotection was seen with 3 h of HPC (HPC3) (LDH release increase to 165 Ϯ 21 and 296 Ϯ 42%, respectively, over control after 3 and 5 days of SH; p Ͼ 0.05 compared with SH alone). In addition, a longer period of treatment (HPC9) did not result in a further reduction in LDH release (p Ͼ 0.05 as compared with the 6 h HPC group). We therefore chose the 6-h HPC regimen for induction of neuroprotection in subsequent experiments.
DOR Function Is Critical for HPC-induced Neuroprotection-The DOR is highly expressed in the cortex (22, 24) and is

FIG. 1. DOR function is critical for HPC-induced neuroprotection.
A, effect of HPC on neuronal injury during SH. Upper panel, cells were left untreated or were subjected to SH for the time indicated on the x axis with or without prior treatment with HPC by exposure to 5% oxygen for 3, 6, or 9 h (indicated as HPC3, HPC6, and HPC9), and then LDH release was measured. Lower panel, cells were treated with 6 h of HPC or with HPC followed by SH for the indicated time. Where indicated, different concentrations of naltrindole, a DOR antagonist, were added before HPC treatment. Daggers indicate a statistically significant difference compared with the SH group (upper panel) or the HPC plus SH group (lower panel). Note that 6-h HPC treatment resulted in marked reduction of SH-induced LDH leakage, and this effect was completely blocked by naltrindole. B, effect of HPC on DNA degradation during SH. Nucleosome (a DNA degradation product) was measured quantitatively by enzyme-linked immunosorbent assay in controls and cells treated for 6 h with HPC and/or 5 days of SH; naltrindole was also added in some cases before HPC (n ϭ 5). The enrichment factor represents the increase in the amount of intracellular nucleosome. The daggers and asterisks show a statistically significant difference as compared with the control or HPC plus SH group, respectively. Note that HPC attenuated DNA degradation during SH, and this effect was blocked by naltrindole. C, HPC prevents the morphological changes in neurons caused by SH. Live and dead neurons were stained as described under "Experimental Procedures." The left and right series of photographs show, respectively, surviving and dead neurons in the same field. Note that neuronal death induced by 5 days of SH was attenuated by 6 h of HPC, and this effect was blocked by DOR inactivation (1 M naltrindole).
homogeneously expressed in cortical neurons in culture (data not shown). To examine the role of the DOR in HPC-mediated neuroprotection, a specific DOR blocker, naltrindole, was used to determine whether DOR inactivation affected HPC-induced protection. Under basal conditions and without HPC, treatment with 0.01 or 0.1 M naltrindole alone for 5 days did not result in neuronal toxicity, whereas treatment with a higher concentration (1.0 M) led to a significant increase in LDH release (98 Ϯ 12% over the control level, n ϭ 6, p Ͻ 0.05). As shown in the lower panel of Fig. 1A, 0.01-1.0 M naltrindole applied to HPC-treated neurons largely attenuated the HPCinduced protection. For example, with 0.1 M naltrindole, the level of LDH release increased by 178 Ϯ 12 or 318 Ϯ 39% over the control after 3 or 5 days of SH, respectively, which was similar to that of SH neurons without HPC treatment (p Ͼ 0.05 compared with SH alone).
To further confirm the HPC-induced protection, we assessed DNA breakdown as an index of neuronal damage, since DNA degradation has been suggested to occur before plasma membrane breakdown (28). As shown in Fig. 1B, when quantitative Top, Western blot analysis for the DOR was performed on three independent cultures for each time point. Bottom, DOR signal normalized to that for actin and expressed as a density unit. Asterisks represent a significant difference compared with the DIV 4 cultures. Note that DOR expression was low at DIV 4 and reached a plateau at DIV 8. B, temporal change in DOR expression during SH. Cells were grown for 1, 3, or 5 days in normoxic conditions or under SH and then analyzed for DOR expression. Top, representative blots. Bottom, DOR signal normalized to that for actin and expressed as a density unit (n ϭ 3). The daggers represent a significant difference compared with the corresponding control culture. C, effect of HPC on DOR expression during SH. Cells were left untreated or were treated for 3 or 5 days with SH; others were treated for 3, 6, or 9 h with HPC and then grown under either normoxic conditions or SH for 3 or 5 days. Top and middle, representative DOR blots from two independent cultures for each of the sets of conditions shown by the horizontal line using 3 days (top) or 5 days (middle) of SH. Bottom, change in DOR expression relative to untreated day 5 control levels under various conditions using 5 days of SH (n ϭ 4). The asterisks and daggers represent significant differences as compared with the control group and corresponding HPC group without SH treatment, respectively. Note that 6 h (HPC6) or 9 h (HPC9), but not 3 h (HPC3), of HPC resulted in up-regulation of DOR expression. In neurons exposed to SH for 3 days, the HPC effect was essentially the same (data not shown). D, lack of effect of a DOR antagonist on DOR expression. Top, representative blots showing the effect of adding naltrindole before 6 h of HPC or 5 days of SH. Two independent samples for each set of treatment conditions are shown. Bottom, statistical data (n ϭ 4) with daggers indicating a significant difference from the control. Note that naltrindole (0.01-1 M) treatment did not affect DOR expression in either HPC-or SH-treated neurons. E, effect of HPC on DOR mRNA levels during SH. Cells were left untreated or were treated for 1-5 days with SH with and without a previous 6 h of HPC treatment, and then DOR mRNA was measured by reverse transcription-PCR. In the top and middle panels, the left lane shows a 100-bp DNA marker ladder. Glyceraldehyde-3-phosphate dehydrogenase was used as an internal control. Diethylpyrocarbonate water was used instead of primers in the no template control (NTC). The statistical results, shown at the bottom, show that HPC treatment attenuated the decrease in DOR mRNA caused by SH (n ϭ 3). The asterisks and daggers represent significant differences compared with control and SH group at the same age, respectively. enzyme-linked immunosorbent assay was used to measure the cytoplasmic levels of the DNA degradation product, the nucleosome, SH significantly increased nucleosome levels, and HPC attenuated this increase, consistent with the results of the LDH assay. HPC did not affect nucleosome levels in basal conditions under normoxia. The most striking observation is that DOR blocker (naltrindole) reversed the protective effect of HPC on the SH neurons, as evidenced by an increase in the level of nucleosome.
Morphological examination of cell viability provided additional evidence for the protective effect of HPC in our model. As shown in Fig. 1C, SH caused severe neuronal death and loss of connection between neurons, whereas few dead cells were seen in the normoxic control. HPC alone did not significantly affect neuronal viability but preserved the viability of neurons exposed to SH, and this effect was blocked by naltrindole.
DOR Expression Is Oxygen-sensitive and Is Up-regulated after HPC-We then asked whether DOR expression changed in response to SH with or without HPC. On Western blots, the molecular weight of the DOR band, which could be downregulated by DOR small interfering RNA (data not shown), was about 46 kDa, consistent with a previous report (29). Under base-line conditions, DOR expression reached a plateau after DIV 8 ( Fig. 2A). As shown in Fig. 2B, exposure to SH for 3 or 5 days was associated with a 24 Ϯ 4 or 51 Ϯ 8% decrease, respectively, in DOR levels (p Ͻ 0.01).
As shown in Fig. 2C, HPC for 6 or 9 h resulted in marked up-regulation of DOR expression in normoxic neurons (for 6 h HPC, a 40 Ϯ 9 and 54 Ϯ 7% increase, respectively, over the control at 5 days, p Ͻ 0.001). When HPC-treated neurons were exposed to SH, their DOR levels were much higher than that in cells treated with SH alone. For instance, DOR levels were 28 Ϯ 2 to 39 Ϯ 2% higher in neurons exposed to 6 or 9 h of HPC plus SH than in untreated controls. The up-regulation of DOR expression correlated well with the neuroprotective effect in cortical neurons with HPC treatment (Fig. 1A). In contrast, 3-h HPC treatment had no significant effect on DOR expression during normoxia or after SH, consistent with the lack of neuroprotection described above (Fig. 1A).
As shown in Fig. 2D, DOR expression in either HPC-or SH-treated cells was not affected by pretreatment with 0.01-1.0 M naltrindole, indicating that the blocking effect of naltrindole on HPC protection was due to an effect on DOR function, not receptor expression.
We then determined whether HPC-and/or SH-induced changes in DOR protein levels were associated with changes in DOR mRNA levels. Fig. 2E shows that, compared with agematched neurons under normoxia, DOR mRNA levels were significantly decreased after 3 or 5 days of SH to 46 Ϯ 11 and 24 Ϯ 8% of control levels, respectively (p Ͻ 0.05). HPC treatment partially reversed the SH-induced decrease in DOR mRNA levels to 72 Ϯ 16 and 62 Ϯ 11% of control levels after 3 and 5 days of SH, respectively (p Ͻ 0.01 compared with SH alone). These data suggest that HPC regulates DOR expression in cortical neurons not only at the protein level, but also at the mRNA level.
HPC Prevents the Decrease in Leucine Enkephalin Content Seen during Severe Hypoxia-Since naltrindole, a DOR antagonist, can increase hypoxia-induced injury in the absence of an exogenous DOR agonist (19), it was possible that it might block the binding of DOR to an endogenous DOR peptide in the cultures. We therefore determined whether there was endogenous release of a ␦-opioid peptide and whether HPC affected its release in addition to altering DOR expression. We measured leucine enkephalin, since it is the dominant ␦-opioid peptide in the brain (30). As shown in Fig. 3, after 3 and 5 days of SH, the intraneuronal content of leucine enkephalin was significantly decreased by 53 Ϯ 15 and 60 Ϯ 16%, respectively, as compared with age-matched control neurons (upper panel), and release of leucine enkephalin into the culture medium was attenuated by 36 Ϯ 10 and 41 Ϯ 12%, respectively (lower panel). HPC alone had no effect on leucine enkephalin levels in the neurons and culture medium during normoxia, but HPC prevented the SHinduced decrease in leucine enkephalin in neurons and restored leucine enkephalin release to the control level. These effects of HPC were not affected by treatment with 0.1 M naltrindole, indicating that the blocking effect of naltrindole on HPC protection is not associated with a change in endogenous DOR agonists.

DOR-mediated HPC Protection Is Dependent on Regulation of Bcl 2 Expression and Cytochrome c Release-
The importance of DOR in HPC-induced neuroprotection raised the interesting question of its intracellular basis. To address this, we examined whether HPC-induced DOR up-regulation affected the intracellular protective/deadly proteins, Bcl 2 and cytochrome c, since Bcl 2 is thought to protect neurons from stroke damage by preventing cytochrome c release from the mitochondria and the generation of subsequent signals (31). As shown in Fig. 4, A and B, Bcl 2 expression and cytochrome c release did not significantly change with age in normoxic neurons. However, a progressive decrease in Bcl 2 levels, which was associated with an increase in cytochrome c release, was seen in neurons exposed to SH. Treatment with HPC alone resulted in a significant increase in Bcl 2 expression under normoxia (131 Ϯ 18% in HPC versus 100% in control, p Ͻ 0.001) (Fig. 4C), whereas, in HPC plus SH neurons, Bcl 2 levels were much higher than in SH-treated neurons (79 Ϯ 9 versus 26 Ϯ 5% as compared with the control, p Ͻ 0.01).
HPC had no effect on cytochrome c release during normoxia but significantly reduced the SH-induced release of cytochrome c (92 Ϯ 14% in HPC ϩ SH cells and 246 Ϯ 29% in SH cells compared with control cells, p Ͻ 0.0001) (Fig. 4D).
The effects of HPC on Bcl 2 expression and cytochrome c release could be reversed by naltrindole in a dose-dependent manner (Fig. 4, E and F), suggesting that these SH-induced changes can be prevented by HPC-induced DOR up-regulation.
MAPKs Act Downstream of DOR Activation in HPC Signaling-Since MAPKs play an important role in transducing signals from the cell surface to the nucleus and have been shown to act as downstream signals in a DOR-transfected cell line (32), we examined whether changes in phosphorylated MAPKs were affected by DOR activation under HPC.
As shown in Fig. 5A, SH significantly increased the levels of phosphorylated p38 MAPK (pp38) to 144 Ϯ 15% (1-day SH), 175 Ϯ 21% (3-day SH), and 213 Ϯ 30% (5-day SH) of control levels (all p Ͻ 0.05 as compared with the control). After HPC, a small, but significant, decrease in pp38 levels was seen on day 1 in neurons during normoxia. Most importantly, in HPC plus SH neurons, the increase in pp38 seen in SH cells was blocked, and this effect was reversed by naltrindole.
Neither HPC nor SH had any appreciable effect on the levels of phosphorylated c-Jun N-terminal protein kinase (pJNK). After naltrindole treatment, a nonsignificant trend to an increase was seen in HPC plus SH neurons (Fig. 5B).
An increase in levels of phosphorylated extracellular signalregulated kinase (pERK) with age was seen in normoxic neurons (Fig. 5C). HPC induced a major increase in pERK levels in neurons during normoxia (173 Ϯ 25 and 322 Ϯ 38% of control levels at 3 and 5 days after HPC, p Ͻ 0.01). pERK levels decreased to 36 Ϯ 9 and 64 Ϯ 7% of control levels after 1 and 3 days of SH, respectively (p Ͻ 0.05). HPC prevented the SHinduced reduction in pERK levels, and this effect was abolished by naltrindole treatment, which also resulted in increased pp38 activity but had no effect on pJNK (Fig. 5D). These results show that HPC, acting via DOR function, prevents both the increase in pp38 and the decrease in pERK seen in SH neurons.
MAPKs Regulate Levels of Protective and Death Proteins in the Neuron-Since the HPC-induced changes in MAPKs, Bcl 2, and cytochrome c appeared to be under the control of DOR function, we then investigated whether MAPK signaling was linked to regulation of Bcl 2 expression and/or cytochrome c release in DOR-mediated HPC neuroprotection. Because phos-phorylation of MAPKs can be pharmacologically inhibited (25,33), we used several specific MAPK inhibitors to block MAPK phosphorylation.
As shown in Fig. 6A, HPC alone had no effect on pp38 and pJNK levels but significantly increased pERK levels. Treatment with U0126, an ERK upstream kinase inhibitor, blocked the HPC-induced increase in pERK; in addition, pp38 levels increased significantly in the same neurons, suggesting an interaction between these two MAPKs. Treatment with the p38 MAPK inhibitor, SB202190, and the JNK blocker, SP600125, reversed the SH-induced pp38 increase and pERK decrease with no significant effect on pJNK. In HPC plus SH neurons, an increase in pERK was observed, and this was blocked by ERK inhibition (U0126), again with an increase in pp38.
As shown in Fig. 6B, in HPC cells, treatment with U0126 also reduced the Bcl 2 increase and enhanced cytochrome c release, suggesting a functional role of pERK in regulating Bcl 2 and cytochrome c. In SH cells, the decrease in Bcl 2 expression and the increase in cytochrome c release were completely blocked by SB202190 plus SP600125. In HPC plus SH neurons, U0126 returned Bcl 2 levels and cytochrome c release to those seen in SH neurons.
The above results suggest that HPC and SH trigger the phosphorylation of different MAPKs as a signaling mechanism to regulate levels of intracellular protective or death proteins, thereby rescuing the cell or resulting in cell death. To strengthen these conclusions, we measured LDH release to evaluate changes in cell viability associated with changes in MAPK activity. LDH release was not significantly affected in HPC neurons treated with MAPK inhibitors at the concentrations used, suggesting that these concentrations were not harmful to neurons within the experimental period (Fig. 6C). SH exposure increased LDH release, and this was partially  4). The daggers represent significant differences as compared with the control of the same age, and the asterisks show significant differences as compared with the 1-day SH group. C and D, effect of HPC on Bcl 2 expression (C) and cytochrome c release (D) during SH. The conditions used were 6 h of HPC and 5 days of SH. The horizontal lines indicate three independent samples using different culture conditions. The statistical analysis is shown in the bottom panels. The daggers and asterisks represent significant differences as compared with the control and SH group (n ϭ 3), respectively. Note that HPC prevented the decrease in Bcl 2 and increase in cytochrome c release in neurons subjected to 5 days of SH. E and F, effect of DOR inactivation on HPC-induced changes in Bcl 2 expression (E) and cytochrome c release (F) in SH. The conditions used were the same as in C and D, except that 0.01, 0.1, or 1 M naltrindole was added to some samples before HPC. The horizontal lines indicate two independent samples for each of the culture conditions. The statistical analysis is shown in the bottom panels. The daggers and asterisks represent significant differences (n ϭ 4) as compared with the control and SH group, respectively. Note that the effect of HPC on Bcl 2 expression and cytochrome c release was dose-dependently reversed by naltrindole. reversed by treatment with SB202190 plus SP600125. These data indicate an important role of p38 MAPK phosphorylation in neuronal injury during SH. In HPC plus SH cells, the SHenhanced LDH leakage was reduced, and this effect was attenuated by U0126, suggesting that ERK activation is essential for HPC-induced neuroprotection.
The G Protein-PKC Pathway Acts Upstream of MAPKs in HPC Signaling-Since DOR belongs to the G protein-coupled receptor family and PKC or PKA activity have been suggested to be involved as downstream signals of DOR-mediated G protein function (34,35), this raised the question of whether HPCmediated DOR activation may regulate downstream effectors via the G protein-PKC/PKA pathway. To test this, we examined the effect of G protein and protein kinase inhibitors on the regulation of MAPK, Bcl 2, and cytochrome c release in HPCtreated neurons.
In HPC-treated cells without SH, pertussis toxin decreased pERK and Bcl 2 levels and increased pp38 levels and cytochrome c release (Fig. 7A), showing an involvement of pertussis toxin-sensitive G proteins in the regulation of survival signals. Similar or greater effects were seen after PKC inhibition using calphostin C. In sharp contrast, no effect was seen after treatment with a potent PKA inhibitor, H89.
Interestingly, in HPC plus SH neurons (Fig. 7B), pertussis toxin had no effect on the HPC-induced changes in pERK, Bcl 2, pp38, and cytochrome c release, suggesting an involvement of pertussis toxin-insensitive G protein in the HPC neuroprotection against SH. Calphostin C treatment resulted in an increase in pp38 levels and cytochrome c release and a decrease in pERK and Bcl 2 expression, whereas PKA inhibition had no effect on any variable. These data suggest that, for HPC protection from SH insult, PKC activation is specifically required Temporal changes in levels of phosphorylated p38 MAPK (A), JNK (B), and ERK (C) were determined in untreated neurons or neurons subjected to 6 h of HPC (1, 3, and 5 indicate the day of culture after HPC) and/or 1, 3, or 5 days of SH; 0.1 M naltrindole was also added to some cultures before HPC treatment. Total MAPKs were stained to measure the phosphorylated MAPKs. The upper panels show representative blots from three independent samples for the different culture conditions, and the bottom panels show the statistical analysis. The daggers in A-C indicate significant differences as compared with the control group at the same age (n ϭ 3), whereas those in D (n ϭ 3) show significant differences as compared with the corresponding HPC group. Note that HPC up-regulated pERK and inhibited the SH-induced increase in pp38; these effects were blocked by treatment with naltrindole. In contrast, pJNK was not affected by HPC or SH.
for the regulation of ERK and p38 MAPK signals and for Bcl 2 expression and cytochrome c release.

DISCUSSION
This study is the first to demonstrate that 1) HPC-induced neuroprotection against SH is dependent on DOR function, 2) HPC up-regulates DOR expression and overcomes the decrease in DOR expression seen during SH, 3) HPC reverses the decrease in leucine enkephalin release seen during SH, 4) DORmediated neuroprotection in HPC neurons relies on G protein and PKC activities, and 5) in HPC neuroprotection, DORtransduced signals enhance pERK-Bcl 2 activity but suppress those of pp38-cytochrome c, displaying a "yin-yang" pattern. These results support a specific mechanism underlying preconditioning-induced neuroprotection, especially regarding the central role of the DOR, an oxygen-sensitive protein, which enhances the intracellular activity of the G protein-PKC-pERK-Bcl 2 pathway and suppresses the pp38 and cytochrome c death signals (Fig. 8).
The Regulation of DOR Expression Is Critical in HPC-mediated Neuroprotection-Three independent assays to detect neuronal injury (membrane integrity, DNA breakdown, and morphological pathology) showed that HPC largely attenuated the neuronal injury induced by subsequent SH. DOR inactivation totally abolished the HPC-mediated protective effect, indicating a critical role of the DOR in neuronal preconditioning. Moreover, we found that DOR was an oxygen-sensitive protein, since DOR expression varied greatly, depending on the duration and extent of hypoxia. Besides down-regulation after prolonged and severe hypoxia, DOR expression was up-regulated by mild hypoxia (i.e. 5% oxygen, as shown in this work). This conclusion is supported by previous studies. For example, NG108-15 cells show increased DOR expression after treat-ment with metallic cations, a condition that mimics hypoxia (36), whereas long term hypoxia (37) or ischemia (38) causes decreased DOR expression in the brain.
The acute effect of preconditioning-induced protection is independent of protein synthesis and mediated by posttranslational modification, whereas the delayed effects of preconditioning require new protein synthesis (39). In this work, HPC neuroprotection and increased DOR expression were both first apparent at 3 days after HPC, suggesting induction of delayed long term protection.
HPC prevented the decrease in DOR mRNA levels normally produced by SH. However, there is a discrepancy between the levels of mRNA and protein. HPC partially reversed the SHinduced decrease in DOR mRNA levels but substantially increased the amount of DOR protein compared with control, suggesting that the HPC-regulated expression of DOR mainly occurs at the posttranscriptional level, possibly due to an increased efficiency for DOR translation and/or an increased stability of DOR under HPC.
It is interesting to note that although HPC increased DOR expression in SH neurons, it only returned leucine enkephalin release to the control level (Fig. 3). As is well known, a decrease in receptor agonists may reduce receptor-mediated intracellular activity. Without an increase in receptor density, however, receptor-mediated activity does not necessarily increase with the amount of agonist. The number of receptors may be a major rate-limiting factor in terms of the increase in receptor-mediated activity. Our data from this study suggest that up-regulation of DOR expression is more important for enhancing DOR-mediated neuroprotection once the release of DOR peptides reaches a "normal" level. The importance of the receptors, rather than ␦-opioid peptides, is further documented by the FIG. 6. MAPKs regulate protective and death proteins in HPC neurons. The conditions used were 6 h of HPC and 5 days of SH. After the addition of MAPK inhibitors (SB202190, SP600125, or U0126 for p38 MAPK, JNK, or ERK inhibition, respectively) to the medium (ϩ or Ϫ indicates with or without inhibitor), the cells underwent the indicated treatments, and then MAPK phosphorylation (A, n ϭ 3), Bcl 2 expression and cytochrome c release (B, n ϭ 3), and LDH leakage (C, n ϭ 6) were measured. The daggers and asterisks indicate a significant difference from the control group and SH group, respectively. Note that the HPC-mediated increases in pERK and Bcl 2 could be inhibited by U0126. Also, ERK inhibition induced an increase in pp38 and cytochrome c release (A and B) and simultaneously abolished the ability of HPC to protect cells from SH, as indicated by LDH leakage (C). SB202190 plus SP600125 blocked the SH-induced increase in pp38 and led to an increase in pERK (A) and a decrease in SH-induced neuronal injury (C). results of DOR inactivation (i.e. HPC neuroprotection was blocked by naltrindole, without any significant change in the release of leucine enkephalin).
Although lacking an inactive form as a pharmacological control, naltrindole is a highly selective non-peptide DOR blocker and antagonizes both the endogenous and exogenous actions of enkephalin in various preparations in vivo and in vitro (12). In the same culture model, we have shown that administration of naltrindole specifically blocks the protective effect of [D-Ala 2 ,D-Leu 5 ]enkephalin, a DOR agonist, in glutamate-or hypoxiainduced neuronal injury (19,20). Moreover, in the present study, naltrindole treatment blocked HPC-induced neuroprotection but did not influence the up-regulation of DOR expression (Fig. 2D) or the restored release of leucine enkephalin (Fig. 3).
DOR Up-regulates ERK and Bcl 2 Signals during HPC-Activation of intracellular ERK has been suggested to induce neuroprotection by preventing neuronal death in stroke (40,41). For example, blockade of ERK function induces cell death in various models of neuronal injury (21,42). However, there is also evidence that ERK blockade protects neurons from stress (43). Our data strongly support the neuroprotective role of ERK in cortical neurons. This is based on the findings that 1) HPC increased pERK levels and reduced neuronal injury and 2) ERK blockade reversed HPC protection and increased death signals in neurons.
Since there is evidence for signal coupling of DOR to ERK activation (31,44), ERK activity may be a downstream signal of DOR activation in HPC neurons. Indeed, our data show that DOR inactivation abolished both the ERK up-regulation and neuroprotection induced by HPC.
Bcl 2 can promote neuronal tolerance to severe stress (9,45). In this study, we found that HPC increased neuronal Bcl 2 expression. Interestingly, there was an association between pERK and Bcl 2 in terms of their up-regulation after HPC. The increase in Bcl 2 expression in HPC-treated neurons was blocked by U0126, an inhibitor of mitogen-activated protein kinase/extracellular signal-regulated kinase kinase, the kinase upstream of ERK, suggesting that the up-regulation of Bcl 2 is dependent on ERK activity. In addition, HPC-induced neuroprotection was abolished by U0126 treatment. Thus, it is very likely that ERK is an upstream regulator of Bcl 2 and transmits critical signals for HPC-induced neuroprotection. This is supported by recent reports showing that ERK may act as a physiological Bcl 2 kinase (46) to prevent cell death by cytotoxic stimuli (21).
Since blocking of DOR by naltrindole in HPC-treated neurons eliminated both the up-regulation of the pERK and Bcl 2 signals and the neuroprotective effect, we believe that DORmediated HPC neuroprotection is dependent on the pERK-Bcl 2 signaling pathway.
DOR Down-regulates p38 MAPK and Cytochrome c Signals during HPC-Increased activity of p38 MAPK plays a critical role in cell death in neurons under stressful conditions (47). FIG. 7. The G protein-PKC pathway acts upstream of MAPKs in HPC signaling. After treatment of a G protein blocker (pertussis toxin), a PKC inhibitor (calphostin C), or a PKA inhibitor (H89) (ϩ or Ϫ indicates with or without blocker/inhibitor), the cells were subjected to 6 h of HPC and 5 days of normoxia (A) or SH (B), and then phosphorylated MAPKs, Bcl 2 expression, and cytochrome c release were measured (n ϭ 3). The dagger indicates a significant difference from the corresponding group without drug treatment. Note that pertussis toxin or calphostin C blocked the HPC-induced changes in pp38, pERK, Bcl 2, and cytochrome c release. In contrast, the PKA inhibitor, H89, had no appreciable effect on either HPC or HPC plus SH neurons.

FIG. 8.
Scheme showing the pathways that mediate HPC-induced neuroprotection. HPC prevents the decrease in DOR mRNA and protein levels and the LE release seen during SH. HPC enhances DOR function to promote neuroprotection through LE release and the G protein (G)-PKC-pERK signal cascade, which, in turn, increases levels of the cellular protective protein, Bcl 2. This pathway may antagonize the p38 MAPK-mediated cytochrome c release from mitochondria seen during SH.
Inhibition of p38 MAPK confers neuroprotection in vitro against excitotoxic exposure (48) and reduces acute ischemic injury in vivo (49). In the present study, SH caused a significant increase in p38 MAPK activity and serious injury to the neurons, whereas a p38 MAPK inhibitor reduced the increase in pp38 levels and reduced SH injury. Since HPC suppressed the increase in p38 MAPK activity and reduced neuronal injury and since DOR inactivation reversed this effect, we conclude that DOR activation suppresses p38 MAPK signals in HPCtreated neurons. As regards how p38 MAPK induces cell death during SH, we found that cytochrome c was a major downstream effector (i.e. an increase in pp38 levels led to an increase in cytochrome c release). In mitochondria-based cell death, Bcl 2 inhibits pore formation on the mitochondria and eliminates cytochrome c release (50). Once released, cytochrome c activates other downstream proteases, leading to cell death (31). Several lines of evidence have suggested that pp38-induced cytochrome c release may play an important role in cell death (51,52). In this study, HPC markedly reduced the SH-induced increase in cytochrome c release, and this effect was blocked by DOR inactivation, indicating an inhibitory role of the DOR on cytochrome c release.
ERK and p38 MAPK Interact with Each Other during SH or HPC Treatment-In sharp contrast to the DOR-mediated upregulation of pERK and down-regulation of pp38 in HPC neurons, there was no appreciable change in JNK activity. At the downstream level, we also found that the changes in Bcl 2 and cytochrome c were a response to DOR-mediated HPC protection. For example, although the effect of Bax on mitochondria is inhibited by Bcl 2 and is involved in cytochrome c release in neurons (50), we did not see any correlation between the DOR-MAPK pathway and Bax activity in HPC protection (data not shown).
An interesting finding in this work was the "yin-yang" change in the activities of ERK and p38 MAPK in terms of their role in SH injury and HPC protection. First, increased pp38 and decreased pERK levels were observed in our neuronal model after prolonged exposure to SH, and these changes were prevented by HPC treatment. Furthermore, the ERK inhibitor, U0126, not only reduced ERK activity but also increased p38 MAPK activity, whereas the opposite was true for p38 MAPK inhibition. Taken together, these results suggest that there may be cross-talk between the ERK and p38 MAPK signals. The pp38-cytochrome c signals play a predominant role during SH treatment, leading to neuronal injury/death, whereas HPC increases the activity of the DOR-pERK-Bcl 2 pathway and decreases that of the SH-pp38-cytochrome c pathway, thus protecting neurons from injury.
G Protein-PKC Transduces Signals from the DOR to MAPKs-This study suggests that DOR-mediated changes in pERK-Bcl 2 signaling are very likely linked to both pertussis toxin-sensitive and -insensitive G proteins, depending on the neuronal condition. This possibility is supported by the results of a previous study showing that, in neuron-like cells or cell lines expressing cloned DOR, ERK can be activated by signaling cascades that are initiated by G protein after DOR activation and that naltrindole abolishes this effect (53). Interestingly, blockade of G protein also induced "yin-yang" changes in pERK and pp38 in HPC neurons in the absence of SH (i.e. down-regulation of pERK-Bcl 2 activity and up-regulation of pp38-cytochrome c signals). These results further demonstrate the "yin-yang" control of the dynamic balance between the ERK and p38 MAPK pathways in the HPC effect. Another interesting observation is that HPC could regulate protective/deadly signals through either pertussis toxin-sensitive or -insensitive G proteins, depending on neuronal environment (severe hy-poxia or normoxia). Although DOR coupling to G protein might function mainly in a pertussis toxin-insensitive manner (54), our results suggest that DOR-G protein coupling may switch between pertussis toxin-sensitive and -insensitive G proteins under different conditions.
Since PKC is involved in DOR signaling (35,55), we examined whether it was involved in HPC protection and found that PKC was activated by HPC-mediated DOR signaling, which resulted in positive regulation of pERK-Bcl 2 and negative regulation of pp38 and cytochrome c release. This regulation is specific to HPC neurons, since PKA inhibition had no significant effect on HPC-DOR signaling. Thus, PKC plays a specific role between DOR and MAPKs in HPC neuroprotection.
In summary, expression of the DOR, an oxygen-sensitive protein, in cortical neurons is up-regulated by mild hypoxic treatment, which mitigates the neuronal damage caused by subsequent SH. The possible mechanism of DOR-mediated HPC protection is dependent on "yin-yang" changes in the activities of ERK and p38 MAPK, thus regulating protective and death signals in the neurons, especially Bcl 2 and cytochrome c release. DOR signals are transduced to MAPKs by the G protein-PKC pathway. As illustrated in Fig. 8, our results suggest that the DOR-G protein-PKC-pERK-Bcl 2 pathway is one of the most important neuroprotective mechanisms in cortical neurons and can be activated by preconditioning to counteract stress-induced death signals (i.e. pp38-cytochrome c).