Address correspondence and reprint requests to Takao Urabe, MD, Department of Neurology, Juntendo University School of Medicine, 2-1-1 Hongo, Bunkyo-ku, Tokyo 113–8421, Japan. E-mail: firstname.lastname@example.org
This study was designed to investigate the neuroprotective effect of intrinsic and extrinsic erythropoietin (EPO) against hypoxia/ischemia, and determine the optimal time-window with respect to the EPO-induced neuroprotection. Experiments were conducted using primary mixed neuronal/astrocytic cultures and neuron-rich cultures. Hypoxia (2%) induces hypoxia-inducible factor-1α (HIF-1α) activity followed by strong EPO expression in mixed cultures and weak expression in neuron-rich cultures as documented by both western blot and RT–PCR. Immunoreactive EPO was strongly detected in astrocytes, whereas EPOR was only detected in neurons. Neurons were significantly damaged in neuron-rich cultures but were distinctly rescued in mixed cultures. Application of recombinant human EPO (rhEPO) (0.1 U/mL) within 6 h before or after hypoxia significantly increased neuronal survival compared with no rhEPO treatment. Application of rhEPO after onset of reoxygenation achieved the maximal neuronal protection against ischemia/reperfusion injury (6 h hypoxia followed 24 h reoxygenation). Our results indicate that HIF-1α induces EPO gene released by astrocytes and acts as an essential mediator of neuroprotection, prove the protective role of intrinsic astrocytic-neuronal signaling pathway in hypoxic/ischemic injury and demonstrate an optimal therapeutic time-window of extrinsic rhEPO in ischemia/reperfusion injury in vitro. The results point to the potential beneficial effects of HIF-1α and EPO for the possible treatment of stroke.
However, there are no studies that compared the EPO expression levels and the neuroprotective effect between mixed neuronal/astrocytic cultures and neuron-rich cultures. In the present study, we determined EPO expression levels and its neuroprotective effect using both mixed neuronal/astrocytic cultures and neuron-rich cultures exposed to hypoxia and reoxygenation. We considered that the in vitro model of mixed culture might be more physiological than separate cultures for addressing the protective role of intrinsic astrocytic-neuronal signaling in hypoxic/ischemic injury. Furthermore, we compared neuronal survival in the presence or absence of antibodies against EPOR to examine whether neuroprotection is mediated by the intrinsic EPO and its cognate receptor. In this study, we investigated the neuroprotective effect of extrinsic recombinant human EPO (rhEPO) and determined the optimal ‘therapeutic time-window’ with respect to the EPO-induced neuroprotection in hypoxia/ischemia and ischemia/reperfusion injury. EPO was applied before hypoxic stimulation in previous studies (Sinor and Greenberg 2000; Siren et al. 2001b; Ruscher et al. 2002), but the present results demonstrated that not only pre-treatment and post-treatment applications of EPO, but its simultaneous application with the start of hypoxia and after the onset of reoxygenation represents a more relevant approach with respect to possible clinical application of this neuroprotective factor.
Materials and methods
Primary neuronal and mixed cultures and in vitro ischemia/reperfusion induction
Primary cultures of cerebral cortical cells were obtained from embryos (E 16–17) of Wistar rats. Cultures were performed as described previously (Ravati et al. 2001; Tanaka et al. 2002), with the following modifications: the cerebral cortex was dissected out and dissociated, then seeded at a density of 2.0 × 105 cells/cm2 on six-well dishes or two-well Chamber slides coated with 5% polyethylenimine (Sigma Chemical Co., St. Louis, MO, USA). The cultures were kept in a 37°C incubator in a humidified atmosphere containing 95% O2/5% CO2. After incubation in F12/MEM 10% fetal bovine serum for 24 h, the medium was changed to F12/MEM 5% calf serum, 5% horse serum.
For neuron-rich cultures, on day 4, cytosine arabinoside (10 µm) was added for 48 h to limit the growth of glia cells. For mixed neuronal/astrocytic cultures, the cells were incubated in araC-free medium. Studies were performed at in vitro day 8, the time at which the mixed cultures consisted of 45–50% neurons and 55–60% astrocytes, and the neuron-rich cultures consisted of 85–90% neurons and 10–15% astrocytes, as determined by immunofluorescence staining with the neuron-specific markers microtubule-associated protein-2 (MAP2; Chemicon International, Inc., Temecula, CA, USA) and the astrocyte-specific marker glial fibrillary acidic protein (GFAP; Dako Corporation, Carpinteria, CA, USA).
In vitro hypoxia/ischemia (0, 0.5, 3, 6, 12, 24 h) was induced by placing the cultures into a modular hypoxic incubator (Model-9200) bubbled with a 2% O2/5% CO2/93% N2 gas mixture. Reoxygenation was induced by returning the cultures to the normoxia incubator for 24 h. The medium was replaced with serum-free medium (F12/MEM) before hypoxia or normoxia.
Immunocytochemical staining of erythropoietin and erythropoietin receptor
After every same period (0–24 h) of normoxia and hypoxia as described above, cultures were fixed in 4% paraformaldehyde, blocked with Block ACE (Yukijirushi Co., Sapporo, Japan), and then incubated with primary rabbit polyclonal antibody against EPO (1 : 500) or EPOR (1 : 500) (Santa Cruz Biotechnology, Santa Cruz, CA, USA) overnight at 4°C. Biotinylated horse anti-rabbit antibody was used as a secondary antibody (1 : 100). ABC Elite kit (Vector Laboratories, Burlingame, CA, USA) was applied to detect EPO and EPOR with 3,3′-diaminobenzidine (Vector Laboratories). Absence of primary antibodies was used as an initial negative control.
Double immunofluorescence staining of erythropoietin and glial fibrillary acidic protein or microtubule-associated protein 2
After 6-h hypoxia, mixed cultures and neuron-rich cultures were fixed and pre-treated as described above. After incubation overnight with the primary EPO antibody (1 : 500), the astrocyte-specific marker GFAP (1 : 100, 1 h) and the neuron-specific marker MAP2 (1 : 200, 1 h) diluted in 2% bovine serum albumin and 0.5% Triton X-100 were applied and the cultures were then incubated with Texas red-conjugated secondary antibody (1 : 500, Vector Laboratories) and fluorescein isothiocyanate (FITC) green-coupled secondary antibody (1 : 100, Vector Laboratories). Hoechst 33342 was used to stain the nucleus.
After hypoxia, total RNA was extracted from cultured cells by using Isogen (Nippon Gene, Tokyo, Japan). DNA-free total RNA (4 µg per sample) was reverse-transcribed into first-strand cDNA using the Reverse Transcription System (Stratagene, La Jolla, CA, USA). The RT product (2 µL) was then amplified by PCR. The primer sequences were as follows: EPO, 5′-TCCTTGCTACTGATTCCTCTCTGG-3′ (forward), and 5′-AAGTATCCGCTGTGAGTGTTCG-3′ (reverse, BLAST, accession number D10763, expected PCR product, 449 bp); EPOR, 5′-GGTAACTTCCAGCTATGGCT-3′ (forward) and 5′-CTAAGCTCCTGTGCCCTCGG-3′ (reverse, D13566, 413 bp product); HIF-1α, 5′-TGCTTGGTGCTGATTTGTGA-3′ (forward) and 5′-GGTCAGATGATCAGAGTCCA-3′ (reverse, AF057308, 209-bp product); β-actin: 5′-AGAAGAGCTATGAGCTGCCTGACG-3′ (forward) and 5′-TACTTGCGCTCAGGAGGAGCAATG-3′ (reverse, BC063166, 301 bp product). The primers were synthesized by Qiagen (Tokyo, Japan). The optimal conditions for amplification were experimentally determined according to previously established procedures (Zhu et al. 1999). The parameters for PCR amplification were 95°C for 1 min, 60°C for 1 min, and 72°C for 1 min for 35 cycles, and a final incubation at 72°C for 7 min. PCR products were separated on 2.5% agarose gels, visualized by ethidium bromide staining, and quantified using Scan Image software (Scion Corporation, Frederick, MD, USA).
Extractions of nuclear and cytosolic proteins and western blotting
The cultures were first homogenized by IGEPAL CA-630 buffer containing 1% IGEPAL CA-630 (Sigma), 10 mm Tris-HCl, 10 mm NaCl and 3 mm MgCl2, and then lyzed by CelLytic cell Lysis/Extraction Regent (Sigma) with freshly added 1% protease inhibitor to obtain cytosolic and nuclear extracts. After protein determination, each sample (10 µg/lane) was then subjected to 10% sodium dodecyl sulfate–polyacrylamide gel electrophoresis. Protein bands were transferred to nitrocellulose membranes and incubated overnight at 4°C with polyclonal antibody against EPO (1 : 1000), EPOR (1 : 1000), HIF-1α (1 : 1000) and a monoclonal antibody against α-tubulin (1 : 1000). The signals were obtained by binding of a secondary anti-rabbit HRP-conjugated antibody and were visualized in the linear range by chemiluminescence (ECL plus kit, Amersham Biosciences, Arlington Heights, IL, USA). For quantitative evaluation, the immunoreactive bands were subjected to densitometry analysis using ScanImage software. The experiments were repeated three times with similar results.
Assessment of neuronal survival and detection of hypoxic-induced neuronal death
The neuroprotective effect of intrinsic and extrinsic EPO was investigated by assessment of neuronal survival. Non-viable cells were determined by staining with the dye propidium iodide (PI, 5 µg/mL), which cannot cross the membrane of viable cells but readily enters and stains non-viable cells exhibiting red fluorescent nucleus. Fluorescein diacetate (FDA, 5 µg/mL) staining was used to determine viable cells with regular-sized green fluorescent cell bodies. The mean percentage of viable neurons was determined by counting eight randomly selected non-overlapping fields using confocal laser microscopy. Each experiment was performed six times independently with different cultures.
To detect the morphology characteristic of neuronal death generated by hypoxia and determine if the assessment of neuronal survival by PI and FDA staining is accurate, we also verified neuronal survival by double-staining with the neuron-specific markers MAP2 (1 : 200, 1 h) and the fluorescent DNA-binding dye, Hoechst 33342 (10 µg/mL, 5min). Cells containing large nuclei with uniformly stained chromatin were considered viable, and cells with condensed or fragmented nuclei were considered apoptotic (Tanaka et al. 2002). The number of viable or apoptotic cells among MAP2-positive cells was counted and the percentage of viable neuron cells was calculated. The neuronal survival was similar as detected by PI and FDA staining.
Induction of neuroprotection by intrinsic erythropoietin after hypoxia/ischemia
Neuronal survival was assessed and compared between mixed cultures and neuron-rich cultures after every period of hypoxia (0, 3, 6, 12, 24 h). To test whether neuroprotection is mediated by the specific interaction of EPO with its cognate receptor, we co-applied an antibody against EPOR (the same as that used for immunocytochemical experiments, 2.5 µg/mL) during every period of hypoxia in parallel experiments and assessed the effect on neuronal survival.
Recombinant human erythropoietin stimulation of neurons
Neuron-rich cultures (8 DIV) were pre-treated with rhEPO (Kirin, Tokyo) at final concentrations of 0.001–10 U/mL, respectively, for 0, 1, 3, 6, 12 or 24 h under normoxic condition before hypoxia (24 h). For post-treatment, the cultures were treated with rhEPO (0.1 U/mL) at 1, 3, 6 and 12 h after the introduction of hypoxia or after onset of reoxygenation. Neuronal survival was assessed as described above after a total of 24 h hypoxia or 6 h hypoxia followed 24 h reoxygenation.
All data are presented as mean ± SD. Statistical comparisons of neuronal survival were made by anova test when more than two groups were involved. The data from RT–PCR and immunoblot were statistically assessed by the two-tailed Mann–Whitney U-test. A p-value < 0.05 was considered to indicate statistical significance. The statview J5.0 statistical software package was used for these analyzes.
Expression of hypoxia-inducible factor-1α, erythropoietin and erythropoietin receptor on both mixed neuronal/astrocytic cultures and neuron-rich cultures in response to in vitro ischemia/reperfusion
Immunoblot analysis and RT–PCR showed that hypoxia/ischemia up-regulated the expression of HIF-1α as early as 30-min exposure to hypoxia and the expression level was still high at 6 h but reduced at 24 h in mixed neuronal/astrocytic cultures. The target gene of HIF-1α, EPO, was undetectable until at least 3-h exposure to hypoxia and lasted at least 24 h during hypoxia. Re-oxygenation significantly degraded the expression of HIF-1α and EPO protein in parallel with mRNA level (Figs 1 and 2).
Importantly, HIF-1α as well as EPO mRNA and protein levels were significantly higher in mixed cultures of neurons and astrocytes compared with those of neuron-rich cultures containing only a limited number of astrocytes at all times of exposure to hypoxia (p < 0.01–0.0001) (Figs 1b,c and 2b,c). EPOR expression was also up-regulated after hypoxia, especially at hypoxia 6 h. Furthermore, in contrast to EPO, EPOR mRNA and protein levels in mixed neuronal/astrocytic cultures were lower than those of neuron-rich cultures at almost all periods of exposure to hypoxia (Figs 1d and 2d).
Hypoxic-induced accumulation of HIF-1α protein was detected in both the cytoplasm and nuclei. On the other hand, EPO and EPOR were strongly detected in the cytoplasm but were almost undetectable in the nuclei (Fig. 2a).
Immunolocalization of erythropoietin and erythropoietin receptor in culture cells after hypoxia/ischemia
Immunocytochemistry staining showed that exposure to hypoxia (3–24 h) increased EPO immunoreactivity in astrocytes, especially in the cytoplasm. Immunoreactive EPO was also observed in neurons, though with weaker staining. On the other hand, EPOR immunoreactivity was not detected in astrocytes, whereas neuronal cell bodies were strongly positive for EPOR. The expression of EPO in astrocytes and neurons was also verified by double immunofluorescence labeling with GFAP and MAP2. Representative images (hypoxia for 6 h) are shown in Fig. 3.
Morphological characteristics of hypoxic-induced neuronal death
Figure 4 shows representative examples of neuron-rich culture cells under normoxia (Figs 4a–c) and hypoxia for 6 h (Figs 4d–f). After hypoxia, apoptotic neuronal cells were identified by their condensed and fragmented nuclei.
Protective effect of intrinsic erythropoietin–erythropoietin receptor interaction on neurons cultured with astrocytes
For hypoxic stress experiments, the medium was replaced with serum-free medium before hypoxia. This may be already a stress by itself. Therefore, to be able to determine the specific effect of hypoxia in terms of death and neuroprotection, we performed appropriate normoxia controls in serum-free medium for each time points. Although neuronal survival tended to diminish in serum-free medium, it was not statistically significant, even at 24 h.
In neuron-rich cultures, exposure to 3 h hypoxia tended to reduce neuronal survival from 83.3 ± 4.8% to 78.4 ± 2.1% but the change was not statistically significant. However, exposure to 6, 12 and 24 h hypoxia significantly and time-dependently reduced neuronal survival (p < 0.001 or 0.0001, Fig. 5a). In contrast, neuronal survival was not significantly reduced after 0–24 h hypoxia in mixed cultures, i.e. neurons were significantly rescued from hypoxic injury when cultured together with astrocytes (p > 0.05 vs. normoxia, Fig. 5b). Furthermore, in mixed neuronal/astrocytic cultures, co-application of anti-EPOR antibody (2.5 µg/mL) during 6–24 h hypoxia significantly reduced the protective effect of the astrocyte–neuron interaction compared with same period of hypoxia without the antibody (p < 0.001, Fig. 5b), but application of the antibody did not significantly alter neuronal survival compared with hypoxia/ischemia alone in neuron-rich cultures (Fig. 5a). Application of the antibody alone in cultures at normoxia did not alter neuronal viability. These results indicate that the intrinsic neuroprotection is mediated by the specific interaction of EPO with its cognate receptor.
Recombinant human erythropoietin time and concentration dependently protects neurons from hypoxia/ischemia and ischemia/reperfusion injury in neuron-rich cultures
Concentrations of rhEPO at 0.1–1.0 U/mL achieved the maximal neuronal survival (66.8 ± 5.3%), compared with cultures exposed to hypoxia/ischemia only (32.3 ± 4.5%, p < 0.0001). However, rhEPO at concentrations lower than 0.001 U/mL or higher than 10 U/mL did not improve neuronal survival during hypoxia (p > 0.05, Fig. 6a). In experiments where rhEPO was applied before exposure to hypoxia, a significant neuronal survival was achieved when rhEPO was administered closest to exposure to hypoxia (0, 1, 3, 6 h before hypoxia, p < 0.001–0.0001 vs. hypoxia alone, Fig. 6b). However, application of rhEPO at more than 24 h before hypoxia reduced the efficacy of neuroprotection by rhEPO (p = 0.3651 vs. hypoxia alone; p < 0.0001 vs. 0–1 h pre-treatment). Application of rhEPO within 6 h after exposure to hypoxia showed a significant protective effect (p < 0.001–0.0001). In contrast, application of rhEPO at 12 h after hypoxia did not increase neuronal survival (p =0.6366 vs. hypoxia alone, Fig. 6c). Finally, administration of rhEPO at the start of hypoxia achieved the maximal beneficial effect on neuronal survival (65.6 ± 3.3%) against hypoxia/ischemia injury.
We also tested the effects of rhEPO on the ischemia/reperfusion injury at different time points. Application of rhEPO improved neuronal survival to a range of 34.4 ± 3.1% to 63.9 ± 5.2% compared with 6 h hypoxia followed 24 h reoxygenation alone (H6R24, 25.7 ± 4.4%, p < 0.001–0.0001, Fig. 6d). Time-series studies showed that neuronal survival significantly increased in cultures of neurons treated only with rhEPO after the onset of reoxygenation (63.9 ± 5.2%) and at the start of hypoxia (54.4 ± 5.7%), but to a lesser protective extent in cultures exposed to 6 h of hypoxia pre-treated with rhEPO (34.4 ± 3.1%, Fig. 6d).
The cellular response to hypoxia/ischemia is mainly controlled by HIF-1α. The HIF-1 target gene EPO has been described as neuroprotectant (Digicaylioglu and Lipton 2001; Prass et al. 2003; Jones and Bergeron 2004). The present study showed the magnitude and time course of the induction of HIF-1α and that EPO mRNA expression level paralleled protein up-regulation after hypoxia and the dose–response effects of endogenous EPO on neuronal survival by astrocytic-neuronal signaling pathway in mixed neuronal/astrocytic cultures compared with neuron-rich cultures in which the number of astrocytes was limited by cytosine arabinoside. Most previous in vitro studies used only neuron-rich cultures; however, the mixed culture of neurons and astrocytes might be more physiological for addressing the protective role of intrinsic astrocytic-neuronal signaling in ischemic/hypoxic injury. Furthermore, we also investigated the neuroprotective effect of extrinsic rhEPO and determined the optimal post-hypoxia time-window with respect to the EPO-induced neuroprotection using neuron-rich cultures.
In this study, RT–PCR and immunoblot analysis showed that in response to hypoxia/ischemia, HIF-1α as well as subsequent EPO mRNA and protein levels were significantly higher in mixed cultures compared with neuron-rich cultures after hypoxia. Consistent with the expression level of EPO, neurons cultured together with astrocytes were distinctly rescued from hypoxia/ischemia injury but were significantly damaged in neuron-rich cultures after the same periods of hypoxia (6, 12, 24 h). These findings provide support to previous studies, which provided strong evidence that astrocytes protect cultured neurons from degeneration induced by anoxia (Vibulsreth et al. 1987) or oxygen glucose deprivation (Ruscher et al. 2002). In addition, we demonstrated that the intrinsic neuroprotection is mediated by the intrinsic EPO with its cognate receptor because the neuroprotection could be blocked by anti-EPOR antibody in mixed neuronal/astrocytic cultures during 6–24 h hypoxia. In contrast, co-application of anti-EPOR antibody did not significantly alter neuronal survival compared with hypoxia alone. These results suggest that astrocytes, but not neurons, release sufficient amounts of EPO for paracrine neuroprotection. Although both astrocytes and neurons expressed EPO, we do not exclude the possibility that EPO immunostaining of neurons could be attributed to astrocytic release of EPO, which subsequently bound to neuronal EPOR.
Continuous degradation of HIF-1α protein by oxygen in normoxic cells is prevented during hypoxia, leading to stabilization and activation of HIF-1, which translocates to the nucleus and binds to a conserved sequence (5′-RCGTG-3′) near the 5′ end of the hypoxia-response enhancer of the EPO gene and up-regulates EPO gene transcription (Semenza 2000; Sharp and Bernaudin 2004). In the present study, we found that HIF-1α was up-regulated as early as 30 min after hypoxia and EPO expression subsequently increased at least after 3 h hypoxia. HIF-1α was detected not only in the cytoplasm but also in nuclei of cultured cortical cells, suggesting the translocation of HIF-1α from the cytoplasm to the nucleus in response to hypoxia/ischemia. In contrast to HIF-1α, EPO and EPOR were only expressed in the cytoplasm, as demonstrated by western blot and immunocytochemistry staining. In addition, our results showed that reoxygenation significantly degraded HIF-1α activation, and reduced subsequent transcription of the target gene EPO and the expression of EPOR.
We demonstrated that neuroprotection by rhEPO was time- and concentration-dependent. Interestingly, significant protection with EPO was achieved only in a limited concentration range (0.1–1.0 U/mL) that was also dependent on the temporal exposure of EPO. This concentration range is similar to other injury paradigms in vitro (Chong et al. 2003) and in vivo models (Grasso et al. 2002). In addition, a significant neuronal survival was achieved with administration periods closest to hypoxia exposure and within a 6-h period after the onset of hypoxia before the induction of cellular mechanisms destining a cell to die. In contrast, application times that occur after the induction of specific signal transduction pathways of cell injury appear to render EPO ineffective as a cytoprotectant. This window of opportunity for protection by EPO most likely coincides with the progressive induction of secondary cellular pathways during this 6-h time span, such as cytochrome c release and cysteine protease induction in the cerebral cortex (Lin and Maiese 2001).
For ischemia/reperfusion injury, administration of EPO immediately after the onset of reoxygenation achieved the greatest neuronal protection in contrast to the pre-administration before hypoxia. The time point can be supported by a recent study that demonstrated that treatment with EPO after the onset of reperfusion promotes the greatest protection to the myocardial structure and preserves cardiac function during ischemia/reperfusion (Lipsic et al. 2004). The time point of treatment with EPO, as in the present study, represents a more relevant approach with respect to possible clinical applications of this neuroprotective factor.
Previous studies identified several key pathways by EPO that were critical for protection against neuronal injury and apoptosis (Digicaylioglu and Lipton 2001; Ravati et al. 2001; Chong et al. 2003). The prevention of neuronal apoptosis and microglial phagocytosis by EPO following its binding to the EPOR can occur through cellular pathways that involve enhanced Akt1 activity, Bad phosphorylation, and the maintenance of mitochondrial membrane stability. Alternatively, EPO may act directly upon cytochrome c, caspase 8, caspase 3, or caspase 1 to promote neuronal survival during hypoxic insults (Chong et al. 2003). Although we could not define further the cellular mechanisms underlying neuroprotection by EPO, our detection of neuronal apoptosis generated by hypoxia and the neuroprotection of EPO are consistent with the hypothesis that EPO acts in the CNS primarily as a direct protective factor in neurons via activation of anti-apoptotic pathways (Marti 2004), and we also proved the protective role of astrocytic-neuronal signaling pathway in hypoxic injury.
In conclusion, we present experimental evidence for the functional role of intrinsic EPO in hypoxia injury in vitro, demonstrated the beneficial effects of an optimal therapeutic time-window of extrinsic rhEPO in neuroprotection against in vitro ischemia/reperfusion injury. Extrapolation of these results clinically point to the possible therapeutic application for the treatment of stroke patients. Imitation of brain intrinsic protective mechanisms may be another novel strategy to future successful approaches to provide neuroprotection against hypoxia/ischemia.
This study was supported in part by a High Technology Research Center grant from the Ministry of Education, Culture, Sports, Science and Technology, Japan.
The rhEPO was obtained from KIRIN company, Tokyo, Japan.