at SciVerse ScienceDirect

Neuropharmacology 63 (2012) 873e882

Contents lists available

Neuropharmacology

journal homepage: www.elsevier .com/locate/neuropharm

Activation of Akt/FoxO and inactivation of MEK/ERK pathways contribute toinduction of neuroprotection against transient global cerebral ischemia bydelayed hypoxic postconditioning in adult rats

Lixuan Zhan a, Danfang Li a, Donghai Liang b, Baoxing Wu a,1, Pingping Zhu a,1, Yanmei Wang a,Weiwen Sun a, En Xu a,*

a Institute of Neurosciences and the Second Affiliated Hospital of Guangzhou Medical College, Key Laboratory of Neurogenetics and Channelopathies of Guangdong Provinceand the Ministry of Education of China, 250 Changgang Dong RD, Guangzhou 510260, ChinabCollege of Life Sciences, Peking University, Beijing 100871, China

a r t i c l e i n f o

Article history:Received 23 December 2011Received in revised form12 June 2012Accepted 15 June 2012

Keywords:Cerebral ischemiaHypoxiaNeuroprotectionPostconditioningSignal pathways

Abbreviations: IT, Ischemic tolerance; I/R, ischischemic postconditioning; HPC, hypoxic postconditideprivation; tGCI, transient global cerebral ischemiakinase; ERK, extracellular signal-regulated kinaseprotein kinase; MEK, mitogen-activated protein kinaskinase.* Corresponding author. Tel.: þ86 20 34153252; fax

E-mail address: enxu@163.net (E. Xu).1 These authors contributed equally to this work.

0028-3908/$ e see front matter � 2012 Elsevier Ltd.http://dx.doi.org/10.1016/j.neuropharm.2012.06.035

a b s t r a c t

Ischemic postconditioning, a series of mechanical interruptions of blood flow immediately after reper-fusion, has been described in brain studies. However, hypoxic postconditioning (HPC) has never beenreported in transient global cerebral ischemia (tGCI) adult rat model. The purpose of this study is toexplore the effects of neuroprotection by delayed HPC against tGCI in adult rats and investigateunderlying mechanisms involving the Akt/Forkhead transcription factor, class O (FoxO) and mitogen-activated protein kinase kinase (MEK)/extracellular signal-regulated kinase (ERK) pathways. Post-conditioning with 60e120 min hypoxia significantly reduced cell death in hippocampal CA1 subregionafter 10 min of tGCI. Postconditioning was effective only when applied 1e2 days after tGCI. Nevertheless,the combination of hypoxic preconditioning and postconditioning provided no additive protection.Additionally, postconditioning increased phosphorylation of Akt and FoxOs after tGCI. Inhibiting phos-phorylation of Akt and FoxOs with LY294002 suppressed the postconditioning-induced neuroprotection.In addition, postconditioning blocked the increase of MEK and ERK phosphorylation after tGCI. Inhibitingphosphorylation of MEK and ERK with U0126 attenuated neuronal damage after tGCI. These resultssuggest that delayed HPC exerts neuroprotection against tGCI-induced injury in adult rats. The activationof Akt/FoxO and inactivation of MEK/ERK pathways by postconditioning contributed to the induction ofneuroprotection against tGCI.

� 2012 Elsevier Ltd. All rights reserved.

1. Introduction

Sublethal ischemic or hypoxic preconditioning, a phenomenoncalled ischemic tolerance (IT), can reduce brain ischemia andreperfusion (I/R) injury when applied before cerebral ischemia(Kapinya, 2005; Zhan et al., 2010). However, preconditioning isclinically feasible only when brain events are predictable and

emia and reperfusion; IPC,oning; OGD, oxygen glucose; PI3K, phosphoinositide 3-; MAPK, mitogen-activatede kinase; JNK, Jun N-terminal

: þ86 20 34153595.

All rights reserved.

controllable. Hence, interventions after the onset of ischemia willhave broader clinical applications. Ischemic postconditioning (IPC),initially described by Zhao et al. (2003), refers to a series ofmechanical interruptions of bloodflowconducted immediately afterreperfusion. In subsequent studies, the IPC-induced neuroprotectionwas confirmed both in vivo (Gao et al., 2008a, 2008b; Lee et al., 2008;Pignataro et al., 2008; Wang et al., 2008; Zhao et al., 2006) andin vitro (Pignataro et al., 2008). Although IPC offers neuroprotectionagainst cerebral ischemia in animal models, its application may befairly limited due to the difficulties of achieving well-controlledinterruptions of reperfusion and the risks associated with thesemanipulations. On the other hand, moderate hypoxia, which doesnot cause neuronal death, has become an attractive method inanimal research for postconditioning and therefore may be safer tobe applied in clinical practice. Recently, the existence of delayedneuroprotection induced by hypoxic postconditioning (HPC) wasshown both in vivo, with a chronic intermittent hypoxia initiated 5 d

Fig. 1. Effects of hypoxic postconditioning on neuronal damage in CA1 subregion. (A) Photomicrograph of sections shows Cresyl violet staining of the hippocampus. The pictures onthe right are magnified from the square areas on the left. Sham group (a, b, n ¼ 11). tGCI 10 min group, histology was performed 7 days after tGCI (c, d, n ¼ 12). tGCI þ hypoxia group,histology was performed 7 days after reperfusion in rats subjected to tGCI and 1 d later 120 min of hypoxia (hypoxic postconditioning; e, f, n ¼ 11). tGCI þ hypoxia Re30d group,

L. Zhan et al. / Neuropharmacology 63 (2012) 873e882874

L. Zhan et al. / Neuropharmacology 63 (2012) 873e882 875

after focal ischemia, and on cultured neurons with hypoxia per-formed 14 h after oxygen glucose deprivation (OGD) (Leconte et al.,2009). Furthermore, we have previously shown that hypoxic pre-conditioning (8% O2) lasting for 30e120 min performed 1e4 daysbefore transient global cerebral ischemia (tGCI) in the adult ratsprotected the brain (Zhan et al., 2010). However, whether HPC exertsneuroprotective effects against global cerebral I/R injury is unknown.

The molecular mechanisms underlying neuroprotectioninduced by postconditioning are largely unclear. Recently, we havedemonstrated that the neuronal damage induced by tGCI mayassociate with the attenuation of phosphorylated Akt and Forkheadtranscription factors (FoxOs), the reduction of NaþeKþATPaseactivity, and the augmentation of apoptosis (Zhan et al., 2010,2011). Furthermore, the signaling pathways of phosphoinositide 3-kinase (PI3K)/Akt (Yano et al., 2001; Kitano et al., 2007) andextracellular signal-regulated kinase (ERK) 1/2 (Lee and Lo, 2003)may play important roles in regulating apoptosis after brainischemia. Therefore, the activation of protein kinases involved inthese two signal pathways was investigated in HPC.

Akt, a downstream factorof PI3K, functions critically in controllingthe balance between survival and apoptosis by regulating the phos-phorylation of its downstream components (Franke et al., 2003;Downward, 2004) including the mammalian members of the FoxOs.The FoxO family mainly consists of three members, namely FKHR(FoxO1), FKHRL1 (FoxO3a), and AFX (FoxO4) (Hoekman et al., 2006).Whenphosphorylated by Akt, FoxOwill be retained in the cytoplasm(Van Der Heide et al., 2004). Conversely, dephosphorylation of FoxOcauses itself and its target genes translocation from cytoplasm tonucleus (Gilley et al., 2003; Brunet et al., 1999). The activation of theAkt pathway contributes to postconditioning-mediated neuro-protection (Zhao, 2009). However, most previous studies assessedphosphorylation of Akt alonewithout focusing on its substrates, suchas FoxOs. The activationof FoxOs is involved in themechanismsof celldeath inducedbycerebral ischemia (Kawanoet al., 2002; Shiodaet al.,2007). Moreover, our prior study has shown that hypoxic pre-conditioning enhanced phosphorylation of FoxOwhichmaymediateITafter hypoxic preconditioning (Zhan et al., 2010). However,much isstill unknown about the roles of FoxO in the protective effect of HPC.

ERK1/2, one of the best-characterized members of the mitogen-activated protein kinase (MAPK) family, mediates a range of activ-ities frommetabolism, motility, and inflammation to cell death andsurvival. It is phosphorylated and activated through a three-tieredmitogen-activated protein kinase kinase (MEK) mode via cellsurface receptors stimulated by growth factors or cytokines. Acti-vation of ERK1/2 pathway was, at least in part, involved in theneuronal damage associated with excito-toxic injury modelsincluding cerebral ischemia (Alessandrini et al., 1999). However, theexact roles of ERK1/2 pathway in ischemic models are highlydebatable (Roux and Blenis, 2004; Wang et al., 2004). Besides,whether ERK1/2 is involved in the mechanisms of neuroprotectionbrought about by HPC has not been reported previously.

Therefore, the aim of the present study was to explore the effectof neuroprotection by delayed HPC against tGCI in adult rats.

histology was performed 30 days after tGCI with HPC (g, h, n ¼ 6). Hypoxia þ tGCI þ hypoxiahypoxic preconditioning 1 d before tGCI and 120 min of HPC 1 d after tGCI (g, h, n ¼ 8). Ssubregion after different durations of hypoxia. Rats were exposed to hypoxia 1 d after tGCI fand 6, respectively). (C) Numbers of surviving cell in CA1 subregion after different ischemia-h120 min (the number of rats in each group was 11, 5, and 6, respectively). (D) Numbers opreconditioning plus postconditioning or 30 days after postconditioning (the number of ratsgroup. (E) Effect of HPC on NeuN-positive neurons in CA1 subregion. Photomicrograph of secmagnified from the square areas on the left. Scale bar: a, c, e: 250 mm, b, d, f: 25 mm. A barsubregion. (F) Effect of HPC on FluoroeJade B positive neurons in CA1 subregion. Photomicroright are magnified from the square areas on the left. Scale bar: a, c, e: 100 mm, b, d, f: 25 mmCA1 subregion. Sham, sham-operated rats. Data are mean � S.D. *p < 0.05 vs. sham animallegend, the reader is referred to the web version of this article.)

Furthermore, we also investigated whether the activation of Akt/FoxO and inactivation of MEK/ERK pathways contributed to theinduction of neuroprotection in this type of postconditioning.

2. Materials and methods

Experiments were performed on adult male Wistar rats weighing 250 w 300 g(Southern Medical University, Guangdong, China). Rats were treated in accordancewith the Guide for the Care and Use of Laboratory Animals (NIH Publication No. 80-23, Revised 1996). The Guangzhou Medical College Committee on Use and Care ofAnimals closely monitored the experiments to ensure compliance with the NIHregulations. All efforts were made to reduce the number of animals used andminimize animal suffering.

2.1. Transient global ischemia

Transient global ischemia was induced by applying the four-vessel occlusionmethod (Pulsinelli and Brierley, 1979). Briefly, the animals were anesthetized withchloral hydrate (350 mg/kg, i.p.). Vertebral arteries were electrocauterized, andcommon carotid arteries were isolated. Forebrain ischemia was induced in awakerats 24 h after the surgery by occluding both common carotid arteries for 10 min.After occlusion, rats that had lost righting reflex within 1 min and whose pupilswere dilated were selected for experiments. Rectal temperature was maintained at37w 38 �C throughout the procedure. Sham-operated rats were performedwith thesame surgical procedures except that the arteries were not occluded.

2.2. Hypoxic postconditioning

Hypoxia was induced as described previously (Zhan et al., 2010). Briefly, ratswere placed in a sealed plastic chamber of 9000 cm3 through which air containing8% O2 and 92% N2 flowed continuously at room temperature. Total gas flow was200 mL/min and no more than 3 rats were placed in the chamber at any given time.

In the present study, 218 rats were used for experiment, four of which were deadduring tGCI,13 during reperfusion and 3 after intracerebroventricular catheterization.

Postconditioning experimental protocol is illustrated in Supplemental Fig. 1. Toevaluate the effects of different hypoxic duration (Supplemental Fig. 1A), rats wererandomly assigned to five groups consisting of tGCI group (Group 1) and tGCI, hypoxia-treated groups with 30, 60, 120, and 180 min hypoxia, respectively (Groups 2e5, per-formed 1 d after tGCI). To detect therapeutic time windows for HPC (SupplementalFig. 1B), 120 min hypoxia was performed at 1 d, 2 d or 3 d after tGCI (Groups 4, 6e7).To detect whether the combination of postconditioning with preconditioning exerteda synergistic effect, rats in Group 9 received both preconditioningwith 30min hypoxia1 d before tGCI and postconditioning with 120 min hypoxia conducted 1 d after tGCI(Supplemental Fig. 1C). Also, to determine whether HPC results in long-term neuro-protection, rats in Group 10 were subjected to postconditioning with 120 min hypoxia1 d after tGCI and then recovered for 30 d (Supplemental Fig. 1D).

2.3. Drug injection

LY294002 (Cell Signaling Technology, Beverly, MA, USA. 50 mmol/L in 25%dimethylsulfoxide sulfoxide in 0.01 M phosphate buffer saline (PBS, pH 7.4)) (Endoet al., 2006; Zhan et al., 2010) or the vehicle (25% dimethylsulfoxide in PBS) wasinjected intracerebroventricularly 30 min before HPC (10 mL, i.c.v., bregma: 1.5 mmlateral, 0.8 mm posterior, 4.0 mm deep).

U0126 (Sigma, St. Louis, Mo., USA. 5 mmol/L in 25% dimethylsulfoxide sulfoxidein PBS) (Chan et al., 2009) or the vehicle was injected intracerebroventricularly 24 hafter tGCI (10 mL, i.c.v.).

2.4. Histology

Rats were perfused intracardially with normal saline, followed by 4% para-formaldehyde in PBS under anesthesia. The brainswere removed quickly and furtherfixed with the same solution at 4 �C overnight. Post-fixed brains were immersed in15%, 30% sucrose in the same fixative for cytoprotection and cut into 30-mm thick

group, histology was performed 7 days after reperfusion in rats received both 30 min ofcale bar: a, c, e, g, i: 250 mm, b, d, f, h, j: 25 mm. (B) Numbers of surviving cell in CA1or 30 min, 60 min, 120 min, or 180 min (the number of rats in each group was 11, 8, 11,ypoxia intervals. Rats were subjected to tGCI, 1 w 3 days later, hypoxia was induced forf surviving cell in CA1 subregion of rats received preconditioning, postconditioning,in each group was 9, 11, 8, and 6, respectively). Data are mean � S.D. *p < 0.05 vs. tGCItions show immunostaining of NeuN in the hippocampus. The pictures on the right aregraph presents quantitative analysis of immunoreactive cell counting of NeuN in CA1graph of sections show FluoroeJade B staining in the hippocampus. The pictures on the. A bar graph presents quantitative analysis of FluoroeJade B positive cell counting in

s, #p < 0.05 vs. tGCI group. (For interpretation of the references to color in this figure

Fig. 2. Effect of hypoxic postconditioning on phosphorylation of Akt and FoxOs in CA1 subregion. (A) Representative images of Immunohistochemistry of phospho-Akt (Ser473) andphospho-FKHR (Ser256) in Sham group, hypoxia groups, tGCI groups, and tGCI þ hypoxia groups. The pictures on the low right are magnified from the square areas on the left. (B)Quantitative analysis of immunoreactive cell counting of phospho-Akt and phospho-FoxOs in CA1 subregion. The positively immunoreactive cells were counted by total number offour non-repeated random fields (0.037 mm2 per field � 4 ¼ 0.148 mm2 total) in CA1 subregion, including stratum oriens, stratum pyramidale, stratum radiatum, stratumlacunosum-moleculare and alveus. Data are shown as mean � S.D. *p < 0.05 vs. sham animals and #p < 0.05 vs. tGCI groups at the same time point (n ¼ 6 in each group). (C)Representative images of Western blot of phospho-Akt and phospho-FoxOs, total Akt and total FoxOs after tGCI in rats with and without postconditioning. Relative optical densitiesof protein bands were calibrated with b-actin and normalized to those in Sham-operated rats. (D) Quantitative analyses of total Akt and FoxOs levels in CA1 subregion in rats withand without postconditioning. (E) Quantitative analyses of phospho-Akt and phospho-FoxOs levels in CA1 subregion in rats with and without postconditioning. Each bar representsthe mean � S.D. *p < 0.05 vs. sham animals and #p < 0.05 vs. tGCI groups at the same time point (n ¼ 3 in each group). Hypoxia groups, Sham-operated, hypoxia-treated groups;tGCI þ hypoxia, tGCI, hypoxia-treated groups.

L. Zhan et al. / Neuropharmacology 63 (2012) 873e882 877

slices using a cryotome (Thermo, Runcorn, Cheshire, UK). Sections selected from thedorsal hippocampus (between AP 4.8e5.8 mm, interaural or AP �3.3 to 3.4 mm,Bregma) were used for Nissl staining or Fluoro-Jade B staining.

Nissl staining was performed with 0.1% Cresyl violet (Sigma), according to thestandard procedure. Fluoro-Jade B staining was used to label degenerating cells(Schmued et al., 2005). Briefly, sections were immersed in 70% ethanol, washed withdistilled water, and treated with 0.06% potassium permanganate solution for 10min.Next, the sections were incubatedwith 0.004% Fluoro-Jade B (Millipore, BedfordMA,USA) in 0.1% acetic acid for 20 min at 25 �C, washed, and mounted with distreneplasticizer xylene (Sigma).

The sections from Nissl staining were examined under a light microscope(�660). Surviving cells showed well-stained Nissl bodies, whereas, damaged cellswere either swollen with the loss of stainable Nissl material or necrotic with frag-menting deeply staining dendrites. Meanwhile, Fluoro-Jade B stained images wereanalyzed with a fluorescence microscope (Leica Microsystems, Wetzlar, Hessen,Germany). Cell counts were conducted by densities as described previously (Wanget al., 2011). The cells in the CA1 pyramidal layer were quantitatively analyzedwithin three non-repeated rectangular areas of 0.037 mm2. Data were quantifiedbilaterally in sections from each brain and assessed double-blindedly. Also, foursections for each animal were evaluated.

2.5. Immunohistochemistry

Rats were sacrificed 0, 24 and 144 h after hypoxia or normoxia (n ¼ 6 in eachgroup). Single-lable, free-floating immunohistochemistry was performed asdescribed previously (Zhan et al., 2010). The primary antibodies used in the studiesinclude neuronal nuclei (NeuN,1:2000; Chemicon, Tenecula, CA, USA), phospho-Akt(Ser473) (1:200; Cell Signaling), phospho-FKHR (Ser256) (1:400), phospho-FKHRL1(Ser253) (1:400), phospho-AFX (Ser197) (1:400; Signalway Antibody, Pearland, TX,USA), phospho-MEK1/2 (Ser217/221) (1:400) and phospho-ERK1/2 (Thr202/Tyr204)(1:500; Cell Signaling). Immunopositive cells in which the reaction product waspresentwithin a clear and regular-shaped cytoplasmic borderwere quantified undera light microscope (�660). The total number of immunoreactive cells was countedby total number of four non-repeated random fields (0.037 mm2 perfield � 4 ¼ 0.148 mm2 total) in the CA1 subregions.

2.6. Western blotting

Rats were sacrificed 0 and 24 h after hypoxia or normoxia (n ¼ 3 in each group).The CA1 subregion protein extraction was performed as described previously (Yanoet al., 2001). Protein concentration was determined by using BCA method, as rec-ommended by the manufacturer (Beyotime, Jiangsu, China). To apply Westernblotting analysis, 50 mg proteins of each sample were separated by sodium dodecylsulfate-polyacrylamide gel electrophoresis (SDS-PAGE) using 10% acrylamid gels,and then transferred to PVDF membranes (Millipore). Western blotting analyseswere performed as described previously (Endo et al., 2006). Primary antibodiesincluded phospho-Akt (1:1000), phospho-FKHR (1:1000), phospho-FKHRL1(1:4000), phospho-AFX (1:4000), phospho-MEK1/2 (1:1000), and phospho-ERK1/2 (1:1000). For normalization, the membrane was stripped and reprobed withantibodies against Akt (1:8000), FKHR (1:1000), FKHRL1 (1:1000), AFX (1:1000),MEK (1:1000), ERK1/2 (1:1000), and b-actin (1:1000). All these antibodies werepurchased from Cell Signaling Technology, except for phospho-FKHR, phospho-FKHRL1 and phospho-AFX from Signalway Antibody. Densitometric analysis for thequantification of the bands was performed using image analysis software (QuantityOne, Bio-Rad Laboratories, Inc. Hercules, CA, USA). Relative optical densities ofprotein bands were calibrated with b-actin and normalized to those in Sham-operated rats.

2.7. Data analyses

Statistical analyses were performed with the Statistical Package for SocialSciences for Windows, version 11.5 (SPSS, Inc, Chicago, Illinois, USA). Measurementdata were summarized by mean � S.D. One-way analyses of variance (ANOVA),followed by Tukey’s H.S.D. post hoc test and two-way ANOVA, followed by Student-Newman-Keuls post hoc test, were used in this study. p < 0.05 was consideredstatistically significant.

3. Results

3.1. Effects of hypoxic postconditioning on neuronal damage in thehippocampal CA1 subregion

Pyramidal cells in the hippocampal CA1 region that underwentSham operation showed well-stained Nissl bodies (Fig. 1A, a and b).Whereas, damaged cells either swollen with loss of stainable Nisslmaterial or necrotic with fragmenting deeply staining dendrites

remarkably increased 7 days after 10 min of tGCI (Fig. 1A, c and d).Compared with Sham-operated group, 10 min of tGCI destroyed79.69% of CA1 neurons.

The protective effect of HPC depends on the duration of hypoxiaand the time intervals between ischemia and hypoxia. Comparedwith tGCI group, exposing animals to hypoxia 2 days after ischemiafor 60 min or 120 min offered protection against the neuronaldamage by tGCI (p < 0.05). In addition, 120 min of hypoxia provedto be more effective in reducing cell damage. However, 30 min or180 min of hypoxia exerted no protective effect (Fig. 1B). Todetermine the optimal ischemia-hypoxic interval, 120 min ofhypoxia was selected as the duration of post-treatment for thesubsequent study. The maximal neuroprotection was induced with1-day interval between ischemia and hypoxia, and 2-day intervalalso led to statistically significant effect (Fig. 1C).

To compare the protective effects of HPC with those of hypoxicpreconditioning described in a previous report (Zhan et al., 2010),rats were exposed to 30 min of hypoxic preconditioning, followed24 h later by 10 min of tGCI, reperfused for 24 h and then subjectedto an additional 120 min of HPC. The degrees of neuroprotectionoffered by hypoxic preconditioning and postconditioning werevirtually identical (Fig. 1D). In addition, there was no potentiationwhen pre- and postconditioning were combined, indicating thatboth strategies are not additive (Fig. 1A, i, j and D).

To determine whether HPC provided long-term protection, celldamage was accessed 30 days after reperfusion in HPC rats. Theassessment showed more surviving cells in CA1 than that of tGCIgroup (Fig. 1A, g, h and D).

The NeuN-positive CA1 pyramidal neurons in Sham-operatedanimals were located as a layer of 3e5 cells in width, and noneuronal reduction was observed. However, a marked reduction ofNeuN-positive neurons occurred 7 days after reperfusion in tGCIgroup. Interestingly, compared with tGCI group, considerableincrease of NeuN-positive neurons in CA1 was observed in HPCgroup (120 min of hypoxia 1 day after tGCI) (Fig. 1E).

FluoroeJade B positive cells in CA1 layer of hippocampus werenot observed in Sham-operated animals (Fig. 1F, a and b). However,a marked increase of FluoroeJade B positive neurons occurred 7days after tGCI (Fig. 1F, c and d). Moreover, the number ofFluoroeJade B positive cells in CA1 was less in HPC group than thatin the tGCI group (Fig. 1F, e and f).

3.2. Effect of hypoxic postconditioning on phosphorylation of Aktand FoxOs

The phosphorylation of Akt and FoxOs after tGCI with orwithout HPC was investigated. The immunohistochemistry studyshowed the distribution of phospho-Akt positive cells in hippo-campus was mainly in the pyramidal cell layers (Fig. 2A, aed).Also, immunostaining of phospho-Akt in CA1 of tGCI animalsnotably decreased after reperfusion. In contrast, a remarkableincrease was displayed in Sham-operated, hypoxia-treated rats.Noteworthily, HPC significantly increased phospho-Akt positivecells in tGCI, hypoxia-treated rats. Phospho-FoxOs positive cells inhippocampus were distributed mainly in the pyramidal cell layers(Fig. 2A, a0ed0). Immunostaining of phospho-FKHR and phospho-FKHRL1 in tGCI animals decreased at 24 and 144 h post-normoxia, and phospho-AFX decreased at 0, 24 and 144 h.However, immunoreactivity of phospho-FKHR and phospho-FKHRL1 in Sham-operated, hypoxia-treated groups increasedsignificantly after hypoxia. Remarkably, tGCI, hypoxia-treated ratsshowed an increase in immunostaining of phospho-FKHR andphospho-AFX 24 h after hypoxia, and phospho-FKHRL1 increasedimmediately after hypoxia. The quantitative analysis of immuno-histochemistry was shown in Fig. 2B.

Fig. 3. Effect of hypoxic postconditioning on phosphorylation of MEK1/2 and ERK1/2 in CA1 subregion. (A) Changes in phospho-MEK1/2-positive cells after tGCI in rats with andwithout postconditioning. Representative immunoreactive cells of phospho-MEK1/2 are shown. (B) Changes in phospho-MEK1/2-positive cells after tGCI in rats with and without

L. Zhan et al. / Neuropharmacology 63 (2012) 873e882878

L. Zhan et al. / Neuropharmacology 63 (2012) 873e882 879

As shown in Fig. 2C and D of Western blot analysis, the changesof phospho-Akt and phospho-FoxOs resembled those resultsdescribed above. In tGCI groups, levels of phospho-Akt andphospho-FoxOs decreased at 0 and 24 h post-normoxia. Hypoxiasignificantly increased phospho-Akt and phospho-FoxOs levelscompared with Sham-operated group 24 h after hypoxia andincreased phospho-Akt and phospho-FKHRL1 at 0 h. In tGCI,hypoxia-treated groups, levels of phospho-Akt, phospho-FKHRand phospho-AFX increased after hypoxia, and the level ofphospho-FKHRL1 increased immediately and then returned toSham level 24 h after hypoxia. Two-way ANOVA revealed animportant effect of HPC on levels of phospho-Akt and phospho-FoxOs in CA1 of tGCI rats. However, no significant change in thelevels of total Akt and FoxOs was observed in ischemic brains withor without HPC.

3.3. Effect of hypoxic postconditioning on phosphorylation ofMEK1/2 and ERK1/2

As shown in Fig. 3A, immunostaining of phospho-MEK1/2 inCA1 of tGCI groups remarkably increased at 0 and 24 h and thenreturned to Sham level at 144 h. However, hypoxia reduced theimmunoreactivity of phospho-MEK1/2 in Sham-operated rats. Inaddition, HPC eliminated the changes in the immunoreactivity ofphospho-MEK1/2 observed after tGCI, and immunoreactivity ofphospho-MEK1/2 in tGCI, hypoxia-treated groups did not changesignificantly for at least 144 h after hypoxia. As shown in Fig. 3B,immunostaining of phospho-ERK1/2 in tGCI groups markedlyincreased after reperfusion. However, hypoxia reduced the immu-noreactivity of phospho-ERK1/2 in Sham-operated rats. In tGCI,hypoxia-treated groups, immunoreactivity of phospho-ERK1/2retained at the Sham level.

As shown in Fig. 3C and D of Western blot analysis, the changesof phospho-MEK1/2 and phospho-ERK1/2 resembled those resultsdescribed above. In tGCI groups, phospho-MEK1/2 level dramati-cally increased to 253.62 � 20.45% and 266.10 � 22.82% at 0 and24 h, respectively. Phospho-ERK1/2 level sharply increased to299.70� 22.58% and 289.16� 9.22%. In contrast, in Sham-operated,hypoxia-treated groups, phospho-MEK1/2 level decreased to53.82 � 2.73% and 40.59 � 5.55% at 0 and 24 h, respectively.Phospho-ERK1/2 level decreased to 26.56 � 12.37% and41.03 � 4.88%. In tGCI, hypoxia-treated groups, neither phospho-MEK1/2 nor phospho-ERK1/2 was changed significantlycompared with Sham-operated group except for a transientincrease of phospho-ERK1/2 level immediately after hypoxia. Thetotal levels of MEK1/2 and ERK1/2 did not change significantly inischemic brains with or without HPC.

3.4. Inhibition of Akt/FoxOs and MEK/ERK pathways exhibiteddifferent effect to the protection from hypoxic postconditioning

To further confirm the role of Akt/FoxOs pathway in HPC,LY294002, a PI3K inhibitor, was administrated 30 min before HPC.The injection of LY294002 reduced phospho-Akt and phospho-FoxOs 24 h after hypoxia in tGCI rats that received HPC (n ¼ 3,Fig. 4A). Furthermore, compared with tGCI, hypoxia-treated group,infusion of vehicle did not attenuate the neuroprotective effect of

postconditioning. Representative immunoreactive cells of phospho-ERK1/2 with and withoanimals and #p < 0.05 vs. tGCI groups at the same time point (n ¼ 6 in each group). (C)postconditioning. Representative protein bands for phospho-MEK1/2, total MEK1/2 and b-acand without postconditioning. Representative protein bands for phospho-ERK1/2, total ERKwith b-actin and normalized to those in Sham rats. Each bar represents the mean � S.D. *p <

each group).

postconditioning (n ¼ 6, Fig. 4C, g and h, D; Fig. 4E, g and h, F).Notably, the injection of LY294002 increased the neuronal damagein CA1 of tGCI, hypoxia-treated rats (n¼ 6, Fig. 4C, i and j, D; Fig. 4E,i and j, F). These data indicated that HPC significantly reversed thedown-regulation of phospho-Akt and phospho-FoxOs induced bytGCI, and its role was completely reversed by pretreatment withLY294002.

Similarly, U0126, an MEK inhibitor, was administrated 24 hafter tGCI or 30 min before HPC to further confirm the role of MEK/ERK pathway. Treatment with U0126 after tGCI dramaticallydecreased phospho-MEK1/2 and phospho-ERK1/2 levels in tGCIgroup (Fig. 4B). In addition, the injection of U0126 before HPCsignificantly attenuated phospho-MEK1/2 and phospho-ERK1/2levels in tGCI, hypoxia-treated rats. Intriguingly, the injection ofU0126 strongly blocked neuronal damage induced by tGCI (n ¼ 6,Fig. 4C, k and l, D; Fig. 4E, k and l, F), but did not increase thenumber of surviving cells in tGCI, hypoxia-treated group (n ¼ 6,Fig. 4D and F). These data indicated that inhibition of up-regulation of phospho-MEK1/2 and phospho-ERK1/2 significantlyattenuated neuronal damage, and inhibition of MEK/ERK signalpathway may contribute to the protective effect of HPC. However,the effects of HPC and inhibition of MEK/ERK signal pathway werenot additive.

4. Discussion

Our study demonstrates that HPC exerts neuroprotectionagainst tGCI-induced neuronal injury as manifested by increasednumber of surviving cells and NeuN-positive cells and decreasednumber of FluoroeJade B positive neurons in CA1 subregion ofadult rats. The degree of neuroprotection after HPC was equivalentto that of hypoxic preconditioning in the same tGCI model.Nevertheless, HPC combining with preconditioning provided noadditive protection.

The algorithm of postconditioning (i.e., duration and cyclenumber) influences the effectiveness of neuroprotection, hencedoseeresponse and time-course studies were designed. Ourresults showed that 60e120 min of hypoxia significantly reducedcell damage after tGCI, while 30 min or 180 min of hypoxia didnot exert neuroprotection. These results suggest that HPC shouldbe sufficiently sensitive to initiate neuroprotection, but shouldnot be overly sensitive to avoid cell damage. This phenomenoncan be attributed to the short periods of hypoxia (<60 min) thatmakes it insensitive to alter protein expression responsible forneuroprotection after tGCI, whereas sustained hypoxia enhancescell death which can offset the hypoxia-induced protectiveeffects.

To date, early brain ischemic postconditioning has been repor-ted in focal ischemia (Gao et al., 2008b; Pignataro et al., 2008), anddelayed postconditioning has been described in both focal (Renet al., 2008) and global ischemia (Burda et al., 2006). Also,Leconte et al. (2009) reported HPC performed 5 days after focalcerebral ischemia can induce neuroprotection. Based on theseparadigms, the therapeutic time window of HPC was assessed. Theresults showed that hypoxia performed 1e2 days after tGCIinduced neuroprotection, while neuroprotection is insignificant ifthe intervention is delayed for 3 days. This phenomenon can be

ut postconditioning are shown. Data are shown as mean � S.D. *p < 0.05 vs. vs. shamChanges in phospho-MEK1/2 and total MEK1/2 after tGCI in rats with and withouttin are shown. (D) Changes in phospho-ERK1/2 and total ERK1/2 after tGCI in rats with1/2 and b-actin are shown. Relative optical densities of protein bands were calibrated0.05 vs. sham animals and #p < 0.05 vs. tGCI groups at the same time point (n ¼ 3 in

Fig. 4. Effect of inhibitions on phosphorylation of protein kinases and neuronal damage in CA1 subregion after tGCI with or without postconditioning. (A) PI3K inhibitor, LY294002inhibited phospho-Akt and phospho-FoxOs at 24 h in tGCI þ hypoxia animals. (B) MEK1/2 inhibitor, U0126 inhibited phospho-MEK1/2 and phospho-ERK1/2 at 24 h both in tGCI and

L. Zhan et al. / Neuropharmacology 63 (2012) 873e882880

L. Zhan et al. / Neuropharmacology 63 (2012) 873e882 881

ascribed to the delayed neuronal death in CA1 subregion after tGCI,which is confirmed by Sugawara et al. (1999)’s study that celldamage first appeared 2 days after ischemia. Also, our studyshowed that the majority of CA1 pyramidal cells underwentneuronal loss 3 days after ischemia. Therefore, HPC should beapplied 1e2 days before neuronal loss occurs after ischemia. In themost current studies, postconditioning was conducted from a fewseconds to several minutes after reperfusion. In reality, sucha narrow therapeutic time window will create a major obstacle toclinical translation. Therefore, delayed HPC prolongs the time-window for neuroprotection application in tGCI.

Another important message conveyed by our study was that theneuroprotection induced by HPC were still significant after a 30-day reperfusion, suggesting that HPC actually prevented ratherthan delayed neuronal injury after tGCI. Namely, HPC results inpermanent neuroprotection.

A cumulative protective effect of preconditioning plus post-conditioning has been noted in the heart studies (Yellon andHausenloy, 2005). We therefore explored this possible effect bycombining maximally effective hypoxic preconditioning parameter(Zhan et al., 2010) with the most effective HPC parameter from thisstudy. Although both pre- and postconditioning were protective, inagreement with previous brain studies (Pignataro et al., 2008; Gaoet al., 2008a), a cumulative effect was unfortunately not shown inour study. These results suggest that preconditioning and post-conditioning may share some similar mechanisms that induceprotection.

Little is known about the protective mechanisms mediatingpostconditioning against cerebral ischemia. Multiple intracellularmechanisms including the activation of protein kinase B, MAPK,adenosine receptors, KATP channels and Naþ/Kþ-ATPase have beenproposed to be involved in the preconditioning-induced neuro-protection (Xiong et al., 2003; Zhan et al., 2010, 2011; Zheng andZuo, 2004). Postconditioning in the brain may also involve theactivation of the family of protein kinases called reperfusion injurysalvage kinases (Gao et al., 2008b; Pignataro et al., 2008). Amongthem, Akt and ERK have been shown to regulate IPC-inducedneuroprotection.

Akt signaling pathway supports cell survival after cerebralischemia (Zhao et al., 2005). Our study showed that phospho-Aktincreased in CA1 of postconditioned rats without affecting totalAkt expression compared with the rats which underwent I/Rprocedure alone. Additionally, LY294002 attenuated the neuro-protection induced by HPC. Rapid IPC increased phospho-Akt, andAkt inhibition partially blocked the protective effect of post-conditioning (Gao et al., 2008b; Pignataro et al., 2008). Theseresults suggest a role of Akt in mediating postconditioning-induced neuroprotection. Although specific targets of Akt havenot been shown in postconditioning, the potential effectorsinclude glycoden synthase kinase 3 beta, cAMP-responsiveelement binding protein and FoxOs. Our previous study showedthe involvement of FoxOs in hypoxic preconditioning neuro-protection (Zhan et al., 2010). Thus, we further explored whetherHPC mediates neuroprotection through the same substrates. Wefound that phospho-FoxOs including phospho-FKHR, phospho-FKHRL1 and phospho-AFX increased after postconditioning,

tGCI þ hypoxia animals. Each bar represents the mean � S.D. *p < 0.05 vs. sham animals and(C) Photomicrograph of sections showing Cresyl violet staining of the hippocampus. (D) Quaimmunostaining of NeuN of the hippocampus. (F) Quantification of NeuN-positive neuronsgroup (e, f, n ¼ 11); tGCI þ vehicle þ hypoxia group, tGCI with vehicle infusion before HPC (g(i, j, n ¼ 6); tGCI þ U0126 group, infusion with U0126 24 h after tGCI (k, l, n ¼ 6). Scale bar: asham animals and #p < 0.05 vs. tGCI group and &p < 0.05 vs. tGCI þ hypoxia group. LY, LY2referred to the web version of this article.)

which is consistent with the changes of phospho-Akt. Further,LY294002 prevented phosphorylation of FoxOs after post-conditioning, followed by intensified neuronal damage. Theseresults suggest that the activation of Akt leads to the inactivationof FoxOs which may mediate neuroprotection after HPC. None-theless, the precise mechanism underlying the inactivation ofFoxOs leading to neuroprotection in this model is still unclear.Phospho-FoxOs, an inactive form of FoxOs, increased and totalFoxOs were not changed. Therefore, the dephosphorylation ofFoxOs, an active form of FoxOs, decreased after postconditioning.The reduction of dephosphorylation of FoxOs may down-regulateits target genes, and decrease cellular apoptosis afterpostconditioning.

The MAPK pathways, including ERK1/2, c-Jun N-terminalkinase (JNK), and p38 pathways, are also closely related toischemic injury. However, the roles of ERK1/2 involved in neuro-protection are controversial (Sawe et al., 2008). Our study foundthat, after reperfusion, phospho-MEK and phospho-ERK levelsincreased, which were not accompanied by a rise in total MEK andERK. Thus, phosphorylation rather than new protein synthesisquickly regulates MEK and ERK activity after ischemia. In addition,HPC decreases phospho-MEK and phospho-ERK levels afterischemia. These results suggest that the reduction of MEK and ERKactivation seems responsible for the neuroprotection induced byHPC. We further confirmed that U0126, the inhibitor, blocked theincreases in MEK and ERK phosphorylation and attenuated braininjury after tGCI. Our study suggests that the inhibition of ERKpathway may contribute to the protective effect of HPC. Thereduction of MEK and ERK phosphorylation induced by HPC maydecrease oxygen radicals and proinflammatory cytokines, resultingin neuroprotection (Xing et al., 2008; Zhao et al., 2006). Furtherexperiments will be required to investigate this interestingpossibility.

In summary, our study demonstrates a protective effect ofdelayed HPC against tGCI-induced injury in adult rats. Also, theactivation of Akt/FoxO pathway, by promotion of Akt and FoxOsphosphorylation, is, at least partly responsible for the neuro-protective effect of HPC. In addition, the protective effect of HPCmay correlate with the inhibition of MEK1/2 and ERK1/2phosphorylation.

Acknowledgments

This work was supported by Natural Science Foundation ofGuangdong, China (Project 8151018201000035) and the Project ofScience and Technology Program of Guangzhou Bureau of Educa-tion, China (No. 08A002). Our sincere thanks go to Mr. Peifeng DU(Institute for Standardization of Nuclear Industry) for editing thispaper.

Appendix A. Supplementary material

Supplementary material associated with this article can befound, in the online version, at http://dx.doi.org/10.1016/j.neuropharm.2012.06.035.

#p < 0.05 vs. tGCI group and &p < 0.05 vs. tGCI þ hypoxia group (n ¼ 3 in each group).ntification of surviving cell in CA1 subregion. (E) Photomicrograph of sections showingin CA1 subregion. Sham group (a, b, n ¼ 11); tGCI group (c, d, n ¼ 12); tGCI þ hypoxia, h, n ¼ 6); tGCI þ LY294002 þ hypoxia group, tGCI with LY294002 infusion before HPC, c, e, g, i: 250 mm, b, d, f, h, j: 25 mm. Each bar represents the mean � S.D. *p < 0.05 vs.94002. (For interpretation of the references to color in this figure legend, the reader is

L. Zhan et al. / Neuropharmacology 63 (2012) 873e882882

References

Alessandrini, A., Namura, S., Moskowitz, M.A., Bonventre, J.V., 1999. MEK1 proteinkinase inhibition protects against damage resulting from focal cerebralischemia. Proc. Natl. Acad. Sci. U. S. A. 96, 12866e12869.

Brunet, A., Bonni, A., Zigmond, M.J., Lin, M.Z., Juo, P., Hu, L.S., et al., 1999. Aktpromotes cell survival by phosphorylating and inhibiting a Forkhead tran-scription factor. Cell 96, 857e868.

Burda, J., Danielisová, V., Némethová, M., Gottlieb, M., Matiasová, M.,Domotáková, I., et al., 2006. Delayed postconditioning initiates additivemechanism necessary for survival of selectively vulnerable neurons aftertransient ischemia in rat brain. Cell Mol. Neurobiol. 26, 1141e1151.

Chan, S.H., Wu, C.A., Wu, K.L., Ho, Y.H., Chang, A.Y., Chan, J.Y., 2009. Transcriptionalupregulation of mitochondrial uncoupling protein 2 protects against oxidativestress-associated neurogenic hypertension. Circ. Res. 105, 886e896.

Downward, J., 2004. PI3-kinase, Akt and cell survival. Semin. Cell Dev. Biol. 15,177e182.

Endo, H., Nito, C., Kamada, H., Nishi, T., Chan, P.H., 2006. Activation of the Akt/GSK3bsignaling pathway mediates survival of vulnerable hippocampal neurons aftertransient global cerebral ischemia in rats. J. Cereb. Blood Flow Metab. 26,1479e1489.

Franke, T.F., Hornik, C.P., Segev, L., Shostak, G.A., Sugimoto, C., 2003. PI3K/Akt andapoptosis: size matters. Oncogene 22, 8983e8998.

Gao, X., Ren, C., Zhao, H., 2008a. Protective effects of ischemic postconditioningcompared with gradual reperfusion or preconditioning. J. Neurosci. Res. 86,2505e2511.

Gao, X., Zhang, H., Takahashi, T., Hsieh, J., Liao, J., Steinberg, G.K., et al., 2008b. TheAkt signaling pathway contributes to postconditioning’s protection againststroke; the protection is associated with the MAPK and PKC pathways.J. Neurochem. 105, 943e955.

Gilley, J., Coffer, P.J., Ham, J., 2003. FOXO transcription factors directly activate bimgene expression and promote apoptosis in sympathetic neurons. J. Cell Biol. 162,613e622.

Hoekman, M.F., Jacobs, F.M., Smidt, M.P., Burbach, J.P., 2006. Spatial and temporalexpression of FoxO transcription factors in the developing and adult murinebrain. Gene Expr. Patterns 6, 134e140.

Kapinya, K.J., 2005. Ischemic tolerance in the brain. Acta Physiol. Hung. 92, 67e92.Kawano, T., Morioka, M., Yano, S., Hamada, J., Ushio, Y., Miyamoto, E., et al., 2002.

Decreased Akt activity is associated with activation of Forkhead transcriptionfactor after transient forebrain ischemia in gerbil hippocampus. J. Cereb. BloodFlow Metab. 22, 926e934.

Kitano, H., Young, J.M., Cheng, J., Wang, L., Hum, P.D., Murphy, S.J., 2007. Gender-specific response to isoflurane preconditioning in focal cerebral ischemia.J. Cereb. Blood Flow Metab. 27, 1377e1386.

Leconte, C., Tixier, E., Freret, T., Toutain, J., Saulnier, R., Boulouard, M., et al., 2009.Delayed hypoxic postconditioning protects against cerebral ischemia in themouse. Stroke 40, 3349e3355.

Lee, J.J., Li, L., Jung, H.H., Zuo, Z., 2008. Postconditioning with isoflurane reducedischemia-induced brain injury in rats. Anesthesiology 108, 1055e1062.

Lee, S.R., Lo, E.H., 2003. Interactions between p38 mitogen-activated protein kinaseand caspase-3 in cerebral endothelial cell death after hypoxia-reoxygenation.Stroke 34, 2704e2709.

Pignataro, G., Meller, R., Inoue, K., Ordonez, A.N., Ashley, M.D., Xiong, Z., et al., 2008.In vivo and in vitro characterization of a novel neuroprotective strategy forstroke: ischemic postconditioning. J. Cereb. Blood Flow Metab. 28, 232e241.

Pulsinelli, W.A., Brierley, J.B., 1979. A new model of bilateral hemispheric ischemiain the unanesthetized rat. Stroke 10, 267e272.

Ren, C., Gao, X., Niu, G., Yan, Z., Chen, X., Zhao, H., 2008. Delayed postconditioningprotects against focal ischemic brain injury in rats. PLoS One 3, e3851.

Roux, P.P., Blenis, J., 2004. ERK and p38 MAPK-activated protein kinases: a family ofprotein kinases with diverse biological functions. Microbiol. Mol. Biol. Rev. 68,320e344.

Sawe, N., Steinberg, G., Zhao, H., 2008. Dual roles of the MAPK/ERK1/2 cell signalingpathway after stroke. J. Neurosci. Res. 86, 1659e1669.

Schmued, L.C., Stowers, C.C., Scallet, A.C., Xu, L., 2005. Fluoro-Jade C results in ultrahigh resolution and contrast labeling of degenerating neurons. Brain Res. 1035,24e31.

Shioda, N., Han, F., Moriguchi, S., Fukunaga, K., 2007. Constitutively active calci-neurin mediates delayed neuronal death through Fas-ligand expression viaactivation of NFAT and FKHR transcriptional activities in mouse brain ischemia.J. Neurochem. 102, 1506e1517.

Sugawara, T., Fujimura, M., Morita-Fujimura, Y., Kawase, M., Chan, P.H., 1999.Mitochondrial release of cytochrome c corresponds to the selective vulnera-bility of hippocampal CA1 neurons in rats after transient global cerebralischemia. J. Neurosci. 19 (RC39), 1e6.

Van Der Heide, L.P., Hoekman, M.F., Smidt, M.P., 2004. The ins and outs of FoxOshuttling: mechanisms of FoxO translocation and transcriptional regulation.Biochem. J. 380, 297e309.

Wang, J.Y., Shen, J., Gao, Q., Ye, Z.G., Yang, S.Y., Liang, H.W., et al., 2008. Ischemicpostconditioning protects against global cerebral ischemia/reperfusion-inducedinjury in rats. Stroke 39, 983e990.

Wang, Z.Q., Wu, D.C., Huang, F.P., Yang, G.Y., 2004. Inhibition of MEK/ERK1/2pathway reduces pro-inflammatory cytokine interleukin-1 expression in focalcerebral ischemia. Brain Res. 996, 55e66.

Wang, Y., Zhan, L., Zeng, W., Li, K., Sun, W., Xu, Z.C., et al., 2011. The effect of GABA onthe hypoxia-induced increase of epilepsy susceptibility in neonate rat. Neuro-chem. Res. 36, 2409e2416.

Xing, B.Z., Chen, H., Zhang, M., Zhao, D.M., Jiang, R., Hiu, X.H., et al., 2008.Ischemic post-conditioning protects brain and reduces inflammation ina rat model of focal cerebral ischemia/reperfusion. J. Neurochem. 105,1737e1745.

Xiong, L., Zheng, Y., Wu, M., Hou, L., Zhu, Z., Zhang, X., et al., 2003. Preconditioningwith isoflurane produces dose-dependent neuroprotection via activation ofadenosine triphosphate-regulated potassium channels after focal cerebralischemia in rats. Anesth. Analg. 96, 233e237.

Yano, S., Morioka, M., Fukunaga, K., Kawano, T., Hara, T., Kai, Y., et al., 2001. Acti-vation of Akt/protein kinase B contributes to induction of ischemic tolerance inthe ca1 subfield of gerbil hippocampus. J. Cereb. Blood Flow Metab. 21,351e360.

Yellon, D.M., Hausenloy, D.J., 2005. Realizing the clinical potential of ischemicpreconditioning and postconditioning. Nat. Clin. Pract. Cardiovasc. Med. 2,568e575.

Zhan, L., Wang, T., Li, W., Xu, Z.C., Sun, W., Xu, E., 2010. Activation of Akt/FoxOsignaling pathway contributes to induction of neuroprotection against transientglobal ischemia by hypoxic preconditioning in adult rats. J. Neurochem. 114,897e908.

Zhan, L., Peng, W., Sun, W., Xu, E., 2011. Hypoxic preconditioning induces neuro-protection against transient global ischemia in adult rats via preserving theactivity of Naþ/Kþ-ATPase. Neurochem. Int. 59, 65e72.

Zhao, Z.Q., Corvera, J.S., Halkos, M.E., Kerendi, F., Wang, N.P., Guyton, R.A., et al.,2003. Inhibition of myocardial injury by ischemic postconditioning duringreperfusion: comparison with ischemic preconditioning. Am. J. Physiol. HeartCirc. Physiol. 285, H579eH588.

Zhao, H., Shimohata, T., Wang, J.Q., Sun, G., Schaal, D.W., Sapolsky, R.M., et al., 2005.Akt contributes to neuroprotection by hypothermia against cerebral ischemia inrats. J. Neurosci. 25, 9794e9806.

Zhao, H., Sapolsky, R.M., Steinberg, G.K., 2006. Interrupting reperfusion as a stroketherapy: ischemic postconditioning reduces infarct size after focal ischemia inrats. J. Cereb. Blood Flow Metab. 26, 1114e1121.

Zhao, H., 2009. Ischemic postconditioning as a novel avenue to protect against braininjury after stroke. J. Cereb. Blood Flow Metab. 29, 873e885.

Zheng, S., Zuo, Z., 2004. Isoflurane preconditioning induces neuroprotection againstischemia via activation of p38 mitogen-activated protein kinase. Mol. Phar-macol. 65, 1172e1180.