[1] 12w Activation of CysLT Receptors Induces Astrocyte Proliferation and Death After Oxygen-Glucose Deprivation

[1] 266w We recently found that 5-lipoxygenase (5-LOX) is activated to produce cysteinyl leukotrienes (CysLTs), and CysLTs may cause neuronal injury and astrocytosis through activation of CysLT 1 and CysLT 2 receptors in the brain after focal cerebral ischemia. However, the property of astrocyte responses to in vitro ischemic injury is not clear; whether 5-LOX, CysLTs, and their receptors are also involved in the responses of ischemic astrocytes remains unknown. In the present study, we performed oxygen-glucose deprivation (OGD) followed by recovery to induce ischemic-like injury in the cultured rat astrocytes. We found that 1-h OGD did not injure astrocytes (sub-lethal OGD) but induced astrocyte proliferation 48 and 72 h after recovery; whereas 4-h OGD moderately injured the cells (moderate OGD) and led to death 24-72 h after recovery. Inhibition of phospholipase A 2 and 5-LOX attenuated both the proliferation and death. Sub-lethal and moderate OGD enhanced the production of CysLTs that was inhibited by 5-LOX inhibitors. Sub-lethal OGD increased the expressions of CysLT 1 receptor mRNA and protein, while moderate OGD induced the expression of CysLT 2 receptor mRNA. Exogenously applied leukotriene D 4 (LTD 4 ) induced astrocyte proliferation at 1-10 nM and astrocyte death at 100-1,000 nM. The CysLT 1 receptor antagonist montelukast attenuated astrocyte proliferation, the CysLT 2 receptor antagonist BAY cysLT2 reversed astrocyte death, and the dual CysLT receptor antagonist BAY u9773 exhibited both effects. In addition, LTD 4 (100 nM) increased the expression of CysLT 2 receptor mRNA. Thus, in vitro ischemia activates astrocyte 5-LOX to produce CysLTs, and CysLTs result in CysLT 1 receptor-mediated proliferation and CysLT 2 receptor-mediated death. V

[1] 127w Astrocytes, the predominant cell type in the brain, are affected in, and contribute to, the cerebral ischemic injury (Panickar and Norenberg, 2005). They can carry out critical functions to influence the outcome of ischemic injury, such as ionic homeostasis, prevention of excitotoxicity, scavenging free radicals, provision of nutrients and growth factors, promotion of neovasculization, and support of synaptogenesis and neurogenesis (Chen-Roetling and Regan, 2006;Liu et al., 2006;Pan-ickar and Norenberg, 2005;Thoren et al., 2006). Generally, astrocytes are more resistant to ischemic injury than neurons. After focal cerebral ischemia in vivo, astrocytes are viable in the penumbral region and undergo reactive astrocytosis to finally form a glial scar in the late or chronic phase; while they are necrotic in the ischemic core (Panickar and Norenberg, 2005;Zhou et al., 2006).

[2] 166w Among the regulators of astrocyte function, the arachidonic acid (AA) cascade is one of the important bioactive systems (Wang et al., 2007), which is activated by P2X7 receptor agonists (Ballerini et al., 2005) and ionophore A23187 (Seregi et al., 1990) in astrocytes. AA is produced and released from the membrane phospholipids by a phosholipase A 2 (PLA 2 ) enzyme, and its subsequent metabolism to bioactive prostaglandins, thromboxanes, leukotrienes, and epoxy fatty acids by cyclooxygenase (COX), lipoxygenase and cytochrome P450 epoxygenase enzymes, coupled to specific terminal synthases (Bosetti, 2007). Of the AA metabolites, cysteinyl leukotrienes (CysLTs, namely LTC 4 , LTD 4 and LTE 4 ), the 5-lipoxygenase (5-LOX) metabolites of AA, play an important role in ischemic insults in the brain (Jatana et al., 2006;Zhou et al., 2006). The actions of CysLTs are mediated by their receptors, CysLT 1 and CysLT 2 receptors (Brink et al., 2003;Kanaoka and Boyce, 2004). Increasing evidence has indicated the roles of CysLTs in regulating astrocyte functions after cerebral ischemia.

[3] 210w In the rats with focal cerebral ischemia, we have found that 5-LOX in the brain was activated and the production of CysLTs was increased with two peaks at 3-24 h and 7 days after reperfusion following 30 min middle cerebral artery occlusion. The activated 5-LOX and the increased CysLTs tempo-spatially related to the astrocytosis in the penumbral region in late phase (7-14 days) (Zhou et al., 2006). In addition, we found that CysLT 1 and CysLT 2 receptor expression was increased in the reactive astrocytes in the penumbral region in rat brain with focal cerebral ischemia (Fang et al., 2006(Fang et al., , 2007)); the CysLT 1 receptor antagonist pranlukast attenuated the astrocytosis in late or chronic phase in the rats or mice with focal cerebral ischemia (Fang et al., 2006;Yu et al., 2005). These findings provide indirect evidence for the role of CysLTs in cerebral ischemic injury in vivo. Moreover, it has been reported that the cultured astrocytes produced and released CysLTs after 1-h oxygen-glucose deprivation (OGD)-induced in vitro ischemia; the released CysLTs might play an autocrine role in the induction of reactive astrocytosis mediated by the CysLT 1 receptor (Ciccarelli et al., 2004). However, the role of CysLT 2 receptor activation in ischemic astrocytes needs to be investigated.

[4] 143w In addition, astrocytes undergo different functional changes in response to different severities of ischemic injury. Milder OGD (1 h or shorter) usually induces neuronal protection or ischemic preconditioning, while more severe OGD (2 h or longer) leads to astrocyte death and increasing neuron vulnerability to ischemic injury (Chavez et al., 2006;Ciccarelli et al., 2007;Kim et al., 2007;Panickar and Norenberg, 2005;Zhao and Flavin, 2000). Therefore, the properties of astrocyte responses to different severities of OGD remain to be clarified. Also, it is unclear whether 5-LOX activation and its metabolites are involved in the responses of ischemic astrocytes, and which subtypes of CysLT receptors mediate these responses. Thus in this study we observed 5-LOX activation in astrocytes after different periods of OGD and recovery, and determined whether astrocyte proliferation or death is induced by 5-LOX activation and mediated by CysLT 1 and CysLT 2 receptors.

[1] 220w We determined the effects of exposure to various durations of OGD on astrocyte viability. The results showed that OGD time-dependently decreased astrocyte viability. The viability was not affected at 1 h after OGD, but gradually decreased by 10, 32, and 44% at 2, 4, and 6 h after OGD, respectively (Fig. 1). Then we treated the astrocytes with 1-and 4-h OGD and observed viability at 24, 48, and 72 h after recovery. Treatment with 1-h OGD increased the viability at 48 and 72 h after recovery (indicating astrocyte proliferation, Fig. 2A); however, 4-h OGD decreased the viability at 24-72 h after recovery (indicating astrocyte injury, Fig. 2B). In addition, apoptosis, and necrosis were induced by 4-h OGD and 48-h recovery, but not by 1-h OGD and 24-h recovery (Fig. 2C). The percentages of Hoechst 33342-positive apoptotic and PI-positive necrotic cells were 4.7% 6 0.9% and 0% in control astrocytes, 3.0% 6 1.0% and 0.2% 6 0.1% after 1-h OGD/48-h recovery, and 38.4% 6 1.2% and 11.0% 6 0.8% (P < 0.01 vs. control, n 5 4 for each group) after 4-h OGD/24-h recovery. On the basis of these results, we induced astrocyte proliferation by 1-h OGD and 48-h recovery (defined as sub-lethal OGD), and astrocyte death by 4-h OGD and 24-h recovery (defined as moderate OGD) in the following experiments.

[2] 97w To determine whether 5-LOX is involved in OGD/recovery-induced astrocyte proliferation and death, we observed the effects of leukotriene pathway enzyme inhibitors on both 5-LOX products (CysLTs) and astrocyte proliferation/death. The results showed that both astrocyte proliferation and death were attenuated by a PLA 2 inhibitor quinacrine (1 and 5 lM, Figs. 3A,B), 5-LOX inhibitors, caffeic acid (5 and 25 lM, Figs. 3E,F) and zileuton (1 and 5 lM, Figs. 3G,H), and a 5-LOXactivating protein inhibitor MK-886 (1 and 5 lM, Figs. 3I,J); however, these responses were not affected by a cyclooxygenase inhibitor indomethacin (1-25 lM, Figs. 3C,D).

[3] 127w Moreover, the content of CysLTs in the medium was significantly decreased during OGD treatment and increased after recovery. After 1-h OGD treatment, CysLTs decreased at the end of the OGD period, but increased from 0.5 to 1 h after recovery with a peak at 0.5 h (1.5-fold), and then returned to the initial baseline level 2 h after recovery (Fig. 4A). After 4-h OGD treatment, CysLTs also decreased and increased from 0.5 to 4 h after recovery with a peak at 1 h (2-fold), and then returned to the initial baseline level 6 h after recovery (Fig. 4B). Caffeic acid (5 lM), zileuton (5 lM), and MK-886 (1 lM) significantly reduced the baseline level and inhibited the peak production of CysLTs after 1-or 4-h OGD (Figs. 4C,D).

[4] 104w We determined whether OGD/recovery regulates the expressions of CysLT 1 and CysLT 2 receptor mRNAs in astrocytes as detected by RT-PCR. We found that the CysLT 1 receptor was moderately in control astrocytes, whilst the CysLT 2 receptor was very weakly expressed in these cells (Fig. 5). After 1-h OGD, the CysLT 1 receptor expression was increased from 24 to 72 h after recovery, while CysLT 2 receptor expression did not change (Figs. 5A,C). In contrast, after 4-h OGD, the CysLT 2 receptor expression was significantly increased from 24 to 72 h after recovery, while CysLT 1 receptor expression did not change (Figs. 5B,D).

[5] 98w Next, we determined the expression of CysLT 1 receptor protein by Western blot analysis, and found a similar change to that in its mRNA (Fig. 6). To confirm the specificity of a polyclonal rabbit anti-human CysLT 1 receptor antibody used in this experiment, we used the protein samples from cultured human umbilical vein EA.hy926 cells or human neuroblastoma SK-N-SH cells, and detected the same bands (ca. 43 kDa) (Fig. 6A). We also tested the specificity of a polyclonal rabbit anti-human CysLT 2 receptor antibody (Cayman Chemicals), but did not confirm its specificity for rat astrocytes (data not shown).

[6] 237w Then, we determined whether CysLT 1 and CysLT 2 receptors mediate OGD/recovery-induced astrocyte proliferation and death. We found astrocyte proliferation A,C,E,G,I) After 1-h OGD and 48-h recovery, astrocyte viability was significantly increased, which was attenuated by quinacrine (A), caffeic acid (E), zileuton (G), and MK-886 (I) but not by indomethacin (C). (B,D,F,H,J) After 4-h OGD and 24-h recovery, the viability was significantly reduced, which was attenuated by quinacrine (B), caffeic acid (F), zileuton (H), and MK-886 (J) but not by indomethacin (D). Data are expressed as mean 6 SD; n 5 8; *P < 0.05, **P < 0.01 compared with control (without OGD); # P < 0.05, ## P < 0.01 compared with OGD alone. Fig. 4. OGD/recovery-induced production of CysLTs in astrocytes and the effects of 5-LOX inhibitors. (A) After 1-h OGD, the production of CysLTs was significantly increased at 0.5 and 1 h of recovery with a peak at 0.5 h. (B) After 4-h OGD, the production of CysLTs was significantly increased at 0.5-4 h of recovery with a peak at 2 h. (C) The summarized data for the effects of caffeic acid, zileuton and MK-886 on the production CysLTs at 0.5 h of recovery following 1-h OGD, and (D) the effects at 1 h of recovery following 4-h OGD; mean 6 SD; n 5 4; *P < 0.05, **P < 0.01 compared with control (without OGD); ## P < 0.01 compared with OGD alone.

[7] 117w after 1-h OGD and 48-h recovery as the viability and cell number were increased. The proliferation was inhibited by the selective CysLT 1 receptor antagonist montelukast (1 and 5 lM, Figs. 7A,D) and the CysLT 1 /CysLT 2 receptor dual antagonist BAY u9773 (1 and 5 lM, Figs. 7B,E), but not by the CysLT 2 receptor antagonist BAY cysLT2 (1 and 5 lM, Figs. 7C,F). In contrast, the astrocyte death occurred after 4-h OGD and 24-h recovery as their viability and cell number were decreased. The death was attenuated by BAY cysLT2 (1 and 5 lM, Figs. 8C,F) and BAY u9773 (1 and 5 lM, Figs. 8B,E), but not by montelukast (1 and 5 lM, Figs. 8A,D).

[8] 97w As the morphological evidence, the density of GFAPpositive astrocytes was increased (astrogliosis) after 1-h OGD and 48-h recovery, which was inhibited by montelukast (1 lM) (Fig. 9A). The density was decreased after 4-h OGD and 24-h recovery, which was inhibited by BAY cysLT2 (1 lM) (Fig. 9B). Similarly, GFAP expression was increased after 1-h OGD and 48-h recovery, which was inhibited by montelukast (1 lM) and BAY u9773 (1 lM, Fig. 9C). GFAP expression was decreased after 4-h OGD and 24-h recovery, which was blocked by BAY cysLT2 (1 lM) and BAY u9773 (1 lM, Fig. 9D).

[9] 107w To confirm the relevant receptor subtypes in astrocyte proliferation or death, we observed the effect of 48-h exposure to LTD 4 , an agonist for both CysLT 1 and CysLT 2 receptors. We found that LTD 4 did not affect the number of astrocytes at 0.01 and 0.1 nM, increased the number at 1 and 10 nM, and reduced the number at 100 and 1,000 nM (Fig. 10A). Furthermore, LTD 4 (1 nM)-induced astrocyte proliferation was blocked by montelukast (1 lM) and BAY u9773 (1 lM); while LTD 4 (100 nM)induced astrocyte death was blocked by BAY cysLT2 (1 lM) and BAY u9773 (1 lM, Fig. 10B).

[10] 60w Additionally, we observed whether LTD 4 regulates the expression of CysLT 1 and CysLT 2 receptors. LTD 4 (1 and 100 nM) did not change the expression of CysLT 1 receptor mRNA (Figs. 11A,B) and protein (data not shown). However, LTD 4 at 100 nM (not at 1 nM) significantly increased the expression of CysLT 2 receptor mRNA (Figs. 11C,D).

[1] 255w In the present study, we found that OGD-induced ischemia activates 5-LOX in cultured rat astrocytes to produce CysLTs and that these CysLTs result in either astrocyte proliferation or death in a manner dependent on the severity of the OGD treatment. Interestingly, sublethal OGD (1 h) induces the CysLT 1 receptor-mediated Fig. 6. CysLT 1 receptor protein expression in astrocytes after OGD/ recovery. Western blot analysis shows that the expression of CysLT 1 receptor increased from 24-h recovery following 1-h OGD (A), but the expression did not change after 4-h OGD and recovery (B). The specificity of a polyclonal rabbit anti-human CysLT 1 antibody used in rat astrocytes were confirmed by using the samples from cultured human umbilical vein EA.hy926 cells (EA) or human neuroblastoma SK-N-SH cells (SK). Lane 1, OGD for 1 h or 4 h without recovery; lane 2, OGD and 24-h recovery; lane 3, OGD and 48-h recovery; lane 4, OGD and 72-h recovery. Data are expressed as mean 6 SD; n 5 4; **P < 0.01 compared with control (without OGD). astrocyte proliferation; while moderate OGD (4 h) induces the CysLT 2 receptor-mediated astrocyte death. These findings have been supported by 5-LOX enzymatic activation and inhibition, CysLT 1 and CysLT 2 receptor expression and antagonism, and by stimulation of CysLT receptors by LTD 4 . In addition to confirming CysLT 1 receptor signaling in astrocytosis (Ciccarelli et al., 2004), our study reveals differential responses to varying severities of OGD treatment and highlights the role of CysLT 2 receptor activation in ischemic astrocytes.

[2] 231w Our first goal was to determine 5-LOX activation in astrocytes after OGD treatment. The results showed a major role of 5-LOX in the responses of ischemic astrocytes. The pharmacological evidence is that OGDinduced astrocyte proliferation and death were attenuated (Fig. 3) by the PLA 2 inhibitor (quinacrine), the 5-LOX inhibitors (caffeic acid and zileuton) and the 5-LOX activating protein inhibitor (MK-886), but not the cyclo-oxygenase inhibitor (indomethacin). Consistent with our observations, 5-LOX activation has been reported in astrocytes (Ciccarelli et al., 2004) and neurons (Ge et al., 2006) after OGD treatment, and has been shown to mediate reversed glutamate uptake-mediated astrocyte death (Re et al., 2006). Our results also support our recent finding that 5-LOX expression was increased in the proliferated reactive astrocytes in the penumbral region in late phase (7-14 days) after focal cerebral ischemia in rats (Zhou et al., 2006). In contrast, it has been reported that COX does not mediate reversed glutamate uptake-mediated astrocyte death (Re et al., 2006), and its subtype inducible COX-2 is expressed in the neurons but not in the astrocytes of mixed neuron/astrocyte cultures after shorter OGD treatment (ischemic preconditioning) (Kim et al., 2007). Together with our present findings, it appears that the COX pathway does not play a role in OGD-induced astrocyte proliferation and death, while 5-LOX activation resulting in CysLT production and CysLT receptor signaling plays a crucial role in these cells.

[3] 89w In support of the involvement of CysLTs, we found 5-LOX was enzymatically activated to produce CysLTs after OGD/recovery. CysLT production was higher and longer lasting (4 h) after moderate OGD treatment than after sub-lethal OGD treatment (Fig. 4). It is noteworthy that the production of CysLTs decreased during OGD. This decrease might relate to consumption of intracellular glutathione, which is required for the synthesis of CysLTs as reported (Ciccarelli et al., 2004;Dringen and Hamprecht, 1998). Recovered from OGD, astrocytes might be able to synthesize glutathione and then produce CysLTs.

[4] 148w The most important finding in our present study is the definition of distinct roles for the CysLT 1 and CysLT 2 receptors in OGD/recovery-induced astrocyte responses, i.e. the CysLT 1 receptor mediates sub-lethal OGD-induced astrocyte proliferation and the CysLT 2 receptor mediates moderate OGD-induced astrocyte death. This is supported by increased CysLT 1 receptor expression after sub-lethal OGD treatment (Figs. 5A,B) but not after moderate OGD treatment, and CysLT 1 receptor antagonist blockade of sub-lethal OGD-induced astrocyte proliferation (Fig. 7) and GFAP expression (Figs. 9A,B). These results are consistent with those reported in the cultured astrocytes, in which CysLT 1 receptor moderately exists in the cells and mediates LTD 4 (0.3-3 nM)-induced proliferation (Ciccarelli et al., 2004). Also, CysLT 1 receptor expression is enhanced in the proliferated reactive astrocytes in the penumbral region in late phase (7-14 days) after focal cerebral ischemia in rats (Fang et al., 2006).

[5] 169w Another novel aspect of our research is the role of the CysLT 2 receptor in ischemic astrocytes. In contrast to CysLT 1 receptor mRNA expression, the CysLT 2 receptor mRNA was much lower in normal astrocytes and was induced after moderate OGD treatment rather than sublethal OGD treatment (Figs. 5C,D). In agreement with this finding, we recently reported that CysLT 2 receptor mRNA was increased in the penumbral region, an area of astrocyte proliferation, in late phase (7-14 days) after focal cerebral ischemia in rats (Fang et al., 2007). In addition, our pharmacological data in the present study shows that selective CysLT 2 receptor antagonism blocked moderate OGD-induced astrocyte death (Fig. 8) and reduced GFAP expression (Figs. 9C,D). We recently found that transfection with CysLT 2 , not CysLT 1 , receptor increased OGD-induced PC12 cell death that was attenuated by the dual CysLT 1 /CysLT 2 receptor antagonist, BAY u9773 (Sheng et al., 2006), suggesting a role of this receptor subtype in ischemic astrocyte and neural cell death.

[6] 130w Taking together, the present findings clearly indicate the distinct roles of CysLT 1 and CysLT 2 receptors in astrocyte responses to sub-lethal or moderate OGD. One of the possible reasons may be the differential expression pattern of the receptors. Because astrocytes produce and release a number of active factors like growth factors and cytokines (Panickar and Norenberg, 2005;Wang et al., 2007), the expression of CysLT receptors in astrocytes may be regulated by a variety of different mediators in varying temporal patterns. In inflammatory cells and other cells in peripheral tissues, interleukin (IL)-1b, IL-4, IL-5, IL-13, interferon-g, and transforming growth factor-b up-regulate CysLT 1 receptor expression; while IL-4 and interferon-g up-regulate CysLT 2 receptor expression (Amrani et al., 2001;Capra et al., 2007;Early et al., 2007;Espinosa et al., 2003;Woszczek et al., 2007).

[7] 238w We also found that exogenously applied LTD 4 induced CysLT 1 receptor-mediated astrocyte proliferation at lower concentrations (1 and 10 nM), but induced CysLT 2 receptor-mediated astrocyte death at higher concentrations (100-1,000 nM) (Fig. 10). These differential effects may be partially due to the differing affinity of LTD 4 for the CysLT 1 and CysLT 2 receptors, where LTD 4 has been shown to be about 10-fold more potent on the CysLT 1 receptor than the CysLT 2 receptor (Capra, 2004;Heise et al., 2000;Lynch et al., 1999). In addition, it has been shown in a number of systems that CysLT 1 and CysLT 2 receptors mediate different responses (Sheng et al., 2006;Wang et al., 2006). In the bleomycin-induced lung fibrosis model, the CysLT 1 and CysLT 2 receptor knockout mice show opposite effects, i.e. fibrosis is enhanced in the CysLT 1 receptor knockout (Beller et al., 2004a) and decreased in the CysLT 2 receptor knockout mouse (Beller et al., 2004b). Moreover, LTD 4 at 100 nM (not at 1 nM) up-regulates the expression of CysLT 2 receptor mRNA (Figs. 11C,D). However, LTD 4 at both 1 and 100 nM did not regulate the expression of CysLT 1 receptor mRNA and protein. These findings suggest that CysLT 2 receptor un-regulation may be partly involved in the LTD 4 (100 nM)-induced astrocyte death; whereas CysLT 1 receptor-mediated astrocyte proliferation may be result from other mechanisms rather than the receptor up-regulation.

[8] 162w Currently, little is known how the intracellular signaling downstream to CysLT receptors changes in astrocytes with OGD treatment. However, LTD 4 activates the extracellular signal regulated kinases 1 and 2 (ERK1/2)/ mitogen-activated protein kinase (MAPK) pathway, which relates to CysLT 1 receptor-mediated astrocyte proliferation (Ciccarelli et al., 2004). Consistently, activated ERK1/2 is also associated with Na 1 /H 1 exchange activation (Kintner et al., 2005) and activation of A1 adenosine and mGlu3 metabotropic glutamate receptors (Ciccarelli et al., 2007) in astrocytes after OGD treatment. Therefore, ERK1/2 may be one of the signaling pathways for CysLT 1 receptor-mediated astrocyte proliferation after OGD treatment. Whereas, the signaling downstream to CysLT 2 receptor is not well investigated although CysLT 2 receptor responses in human endothelial cells has been reported to be mediated by the activation of phospholipase C and inositol-1,4,5-triphosphate (Woszczek et al., 2007). Thus, it remains to clarify the molecular mechanisms underlying the expression and responses of the CysLT 1 and CysLT 2 receptors.

[9] 116w In summary, we have shown that the 5-LOX pathway is activated to produce CysLTs in OGD-induced ischemia in cultured rat astrocytes. The CysLTs activated either CysLT 1 receptor-mediated astrocyte proliferation or CysLT 2 receptor-mediated death in a manner dependent on the severity of the OGD treatment. These findings might explain the acute astrocyte death in the ischemic core and the late or chronic astrocytosis in the penumbral region in the brain after focal cerebral ischemia in vivo (Fang et al., 2006(Fang et al., ,2007;;Yu et al., 2005;Zhou et al., 2006). The discovery of distinct roles of CysLT 1 and CysLT 2 receptor activation in ischemic astrocytes may lead to different therapeutic approaches for ischemic brain injury.

[1] 223w Primary cultured astrocytes were prepared from brains of neonatal rats born within 24 h as described previously (Ballerini et al., 2005;Ciccarelli et al., 2004). Briefly, the cerebral cortices were digested with 0.25% trypsin for 20 min at 37°C, and then the dissociated cells were seeded onto poly-L-lysine-coated 75 cm 2 flasks (BD falcon, USA) at a density of 2 3 10 6 cells/cm 2 . Cells were cultured in high glucose DMEM (Gibco, Grand Island, USA) supplemented with 10% fetal bovine serum, 2 mM glutamine, 100 units/mL penicillin and 100 lg/mL streptomycin. The cultures were maintained at 37°C in a humidified atmosphere (5% CO 2 and 95% air) and the medium was renewed every 3 days until confluence. On 11-14th day, the confluent cultures were shaken overnight to minimize microglia contamination. Then the adherent cells were trypsinized and replated at a density of 3 3 10 5 cells/mL onto 6-well or 96-well plates. More than 95% of the cultured cells were astrocytes as identified by immunofluorescent staining for glial fibrillary acidic protein (GFAP). Animal care was carried out in accordance with the National Institute of Health Guide for the Care and Use of Laboratory Ani-mals. To obtain synchronized cultures, the serum concentration was reduced to 0.5% for 72 h, starting from 24 h after cell seeding. Experiments were performed 4 days after replating.

[2] 100w Astrocytes seeded on cover slips were fixed by cold methanol for 5 min and incubated in 10% normal goat serum for 2 h to block nonspecific binding of IgG. Then the cells were reacted with mouse monoclonal antibody against GFAP (1:500, Chemicon, USA) at 4°C overnight. After washed in PBS, astrocytes were reacted with FITC-conjugated goat anti-mouse antibody (1:200, Chemicon) for 2 h at room temperature. Finally, the stained cells were observed with a fluorescent microscopy (Olympus BX51, Japan). Control slips treated with normal goat serum instead of the secondary antibody did not show the positive immunostaining (data not shown).

[3] 219w At the end of OGD/recovery or LTD 4 treatment , astrocytes were washed twice with ice-cold PBS and then lysed at 4°C in Cell and Tissue Protein Extraction Solution, containing 1 mM pepstatin, 2 mM leupeptin, 80 mM aprotinin, 1 mM phenylmethylsulfonyl fluoride (Kangchen Biotechnology, Shanghai, China). The homogenates were centrifuged at 12,000g at 4°C for 30 min, and the supernatant was used. Protein concentrations were determined by the Bio-Rad protein assay (Bio-Rad Lab, Hercules, CA). Protein samples (60 lg) were separated by 12.5% SDS-polyacrylamide gels and then transferred to nitrocellulose membranes. After blocked with 10% fat free milk, the membranes were incubated with a mouse monoclonal antibody against GFAP (1:2,000, Chemicon), a polyclonal rabbit anti-human CysLT 1 receptor antibody (1:800, Cayman Chemicals, USA), and a mouse monoclonal antibody against glyceraldehyde-3phosphate dehydrogenase (GAPDH, 1:4,000, Kangchen, China) at 4°C overnight. After repeated wash, the membranes were reacted with a horseradish-conjugated goat anti-mouse antibody for 2 h. Then the membranes were reacted with enhanced chemiluminescence reagents (Kangchen, China) and exposed on an X-ray film. The optical densities of GFAP (50 kDa), CysLT 1 receptor (43 kDa) and GAPDH (36 kDa) bands on the X-ray film were quantitatively analyzed with a laser densitometer (Ultro Scan XL, Pharmacia LKB, Sweden). The results of expression are expressed as GFAP/GAPDH or CysLT 1 R/GAPDH ratio.

[4] 59w To confirm the specificity of the polyclonal rabbit antihuman CysLT 1 receptor antibody, we also used the protein samples from the cultured human umbilical vein EA.hy926 cells (a gift from Dr. Cora-Jean S. Edgell, University of North Carolina) and human neuroblastoma SK-N-SH cells (purchased from Institute of Cell Biology, Chinese Academy of Sciences, Shanghai, China) in Western blotting analysis.

[5] 35w Data are expressed as means 6 SD. Significance of differences was analyzed by one-way ANOVA followed by Dunnett's Multiple Comparison test (SPSS 10.0 for windows, 1999, SPSS, USA). P < 0.05 was considered statistically significant.

[1] 134w OGD was performed according to the reported methods (Ge et al., 2006;Song et al., 2004). Briefly, astrocytes were rinsed twice and incubated in Earle's solution without glucose (concentrations in mM: NaCl, 117; KCl, 5.3; CaCl 2, 1.8; NaHCO 3, 26; MgSO 4 , 0.8; NaH 2 PO 4 , 1.0). Then the cultures were moved into an anaerobic chamber filled with 95% N 2 and 5% CO 2 at 37°C. This procedure decreased pO 2 in the solution from 154.0 6 6.1 to 26.0 6 2.0 mmHg (mean 6 SD, n 5 5). Whereas normoxia controls were similarly washed and incubated with Earle's solution containing 5.6 mM glucose. At the end of OGD, the cultures were returned to the normal culture condition and the medium was replaced with culture medium for the indicated time.

[2] 173w To determine the involvement of 5-LOX and CysLT receptors, we used following inhibitors and antagonists: quinacrine (0.2-5 lM, a nonselective PLA 2 inhibitor), indomethacin (1-25 lM, a nonselective cyclooxygenase inhibitor), MK-886 (0.2-5 lM, a 5-LOX activating protein inhibitor), caffeic acid (1-25 lM, a 5-LOX inhibitor with antioxidant activity, Sigma-Adrich, USA), zileuton (0.2-5 lM, a 5-LOX inhibitor, Gaomeng Pharmaceutical Co., Beijing, China), montelukast (0.2-5 lM, a selective CysLT 1 receptor antagonist, Merck Pharmaceutical, USA), BAY u9773 (0.2-5 lM, a dual antagonist for CysLT 1 and CysLT 2 receptors, Sigma-Adrich) and BAY cysLT2 (0.2-5 lM, a selective CysLT 2 receptor antagonist [Example 11 in patent WO 2004/052389 A1] with IC 50 s in membrane binding assays of 35 nM versus the human CysLT 2 receptor and >1,000 nM versus the human CysLT 1 receptor, a kind gift of Dr. T. Jon Seiders of Amira Pharmaceuticals, USA). The agents were continuously applied from 30 min before OGD to the end of recovery. Control cells underwent the same procedure except they were not subjected to OGD treatment.

[3] 90w In another series, leukotriene D 4 (LTD 4 , Sigma-Adrich, USA), an agonist of cysteinyl leukotriene receptors, was used as a stimulus instead of OGD. LTD 4 at 0.01-1,000 nM was added into the culture media, and astrocyte number was counted 48 h after LTD 4 treatment to determine its concentration-dependent effect. Thereafter, we treated the astrocytes with LTD 4 at 1 or 100 nM in the absence or presence of montelukast, BAY u9773 and BAY cysLT2 (applied 30 min before LTD 4 ) to determine the relevant receptor subtypes.

[4] 120w At the end of recovery, 3-(4,5-dimehythiazol-2-yl)-2,5diphenyltetrazolium bromide (MTT, Sigma-Adrich) was added to each well to reach a final concentration 0.5 mg/mL. After incubation for 4 h at 37°C, the medium was removed and 100 lL dimethyl sulfoxide was added to each well for 10 min. The plate was read at 490 nm using a microplate reader (Elx800, Bio-Tek Instrument, USA). Results were reported as the percentages of control. In another series, at the end of recovery, the cells were trypsinized, resuspended in the medium and mixed with 0.4% trypan blue (1:1, Sigma-Adrich). The trypan blue negative cells were counted using a haemocytometer. The proportion of the survival cells with negative staining was reported as a percentage of total cell counts.

[5] 98w For detection of cell death, astrocytes grown on the coverslips were stained with 10 lg/mL of Hoechst 33342 and 10 lg/mL of propidium iodide (PI, Sigma-Adrich) for 10 min at 37°C. Then, the cells were photographed under a fluorescent microscope (Olympus BX51, Japan). The apoptotic cells were determined as condensed or fragmented nuclei with strong bright Hoechst 33342 staining, and the necrotic cells as condensed nuclei with red PI staining. At least 10,000 cells were counted in at least three separate fields for each coverslip, and the apoptotic or necrotic cells were reported as percentages of total cells.

[6] 97w During and after OGD treatment, medium samples (200 lL) were removed and prepared according to the previously reported method (Ciccarelli et al., 2004;Ge et al., 2006). The CysLTs produced (LTC 4 , LTD 4 , and LTE 4 ) were measured by a commercial CysLT EIA kit (Cayman Chemical, MI) according to the manufacturer's instructions, and calculated as pg/mg protein. To determine the inhibition of 5-LOX enzymatic activation by its inhibitors, MK-886 (1 lM), caffeic acid (5 lM) and zileuton (5 lM) were added 30 min before the treatments and maintained until the end of the experiments.

[7] 65w At the end of OGD/recovery or LTD 4 treatment, total RNA was extracted from the cultured astrocytes using Trizol reagents (Gibco BRL) according to the manufacturer's instructions. The cDNA synthesis and PCR reactions were performed as reported previously (Fang et al., 2006(Fang et al., , 2007)). The primer sequences for rat CysLT 1 and CysLT 2 receptor as well as rat b-actin were the following.

[8] 57w CysLT 1 receptor forward: 5 0 -TCT CCG TTG TGG GTT TCT-3 0 and reverse: 5 0 -TAT AAG GCA TAG GTG GTG-3 0 (production size 214 bp); b-actin forward: 5 0 -TAC AAC CTC CTT GCA GCT CC-3 0 and reverse: 5 0 -GGA TCT TCA TGA GGT AGT CAG TC-3 0 (production size 620 bp).

[9] 58w CysLT 2 receptor forward: 5 0 -AGC GTT AGG AGT GCC TGG AT-3 0 and reverse: 5 0 -CAA GTG GAT GGT CCG AAG TG-3 0 (production size 520 bp); b-actin forward: 5 0 -AAC CCT AAG GCC AAC CGT GAA-3 0 and reverse: 5 0 -TCA TGA GGT AGT CTG TCA GGT-3 0 (production size 285 bp).

[10] 42w PCR products were separated by 2% agarose gel electrophoresis and visualized by ethidium bromide staining. The optical density of each band was measured by UVP gel analysis system (Bio-Rad, Richmond, CA). The results of semi-quantitative measures were expressed as CysLT receptor/b-actin ratio.