Co-expression of heat shock transcription factors 1 and 2 in rat retinal ganglion cells

Neuroscience Letters 405 (2006) 191–195

Co-expression of heat shock transcription factors 1 and 2 in rat retinal ganglion cells Jacky M.K. Kwong a,∗ , Maziar Lalezary b,c , Jessica K. Nguyen d , Christine Yang e , Anuj Khattar f , Natik Piri a , Sergey Mareninov a , Lynn K. Gordon a,g , Joseph Caprioli a a

Department of Ophthalmology, University of California Los Angeles, Los Angeles, CA, USA Department of Neuroscience, University of California Los Angeles, Los Angeles, CA, USA c School of Medicine, University of California San Diego, San Diego, CA, USA d Department of Bioengineering, University of California Los Angeles, Los Angeles, CA, USA e Department of Microbiology, Immunology and Molecular Genetics, University of California Los Angeles, Los Angeles, CA, USA f Department of Molecular, Cell, and Developmental Biology, University of California Los Angeles, Los Angeles, CA, USA g Ophthalmology Section, Greater Los Angeles VA Healthcare System, Los Angeles, CA, USA b

Received 17 May 2006; received in revised form 26 June 2006; accepted 28 June 2006

Abstract Heat shock protein (HSP) plays an important role in the maintenance of neuronal survival during harmful conditions. Previously, we reported that metabolic stress induces HSP72 in retinal ganglion cells (RGCs) and protects against excitotoxicity, hypoxia and experimental glaucoma. To understand heat shock protein transcriptional mechanisms, we examined the cellular expression of heat shock factors 1 (HSF1) and 2 (HSF2) in the unstressed adult rat retina. Western blotting, immunohistochemistry and RT-PCR showed that mRNA and protein of HSF1 and HSF2 were present in the rat retina and predominantly expressed in RGC layer cells. Western blotting of dissociated RGC suspensions harvested with Thy-1 immuno-labeled magnetic beads confirmed that RGCs expressed HSF1, HSF2 and HSP72. Our findings suggest that both heat shock transcription factors 1 and 2 are linked to the heat shock response in retinal ganglion cells. © 2006 Elsevier Ireland Ltd. All rights reserved. Keywords: Heat shock protein; Stress; Rat; Heat shock factor; Retina; Ganglion cell

The heat shock response is a well-conserved response to diverse environmental and physiological challenges, and results in the immediate induction of genes encoding molecular chaperones, known as heat shock proteins (HSPs) [14]. Families of HSPs are classified according to their molecular weights namely HSP100, HSP90, HSP70, HSP60, HSP40 and small HSP (approximately 20 kDa). These proteins function as chaperones under normal, developmental and stressful conditions [1], and they also play a role in disease pathological processes [7]. In the central nervous system, intracellular expression of HSP72 (an inducible form of the HSP70 family) has been demonstrated to protect neurons against heat shock, oxidative stress, apoptotic stimuli, excitotoxic insults, and ischemic-like conditions [4]. Neurons ∗

Correspondence to: Department of Ophthalmology, Jules Stein Eye Institute, David Geffen School of Medicine at University of California Los Angeles, Room B-146, 100 Stein Plaza, Los Angeles, CA 90095-7000, USA. Tel.: +1 310 206 7900; fax: +1 310 206 7773. E-mail address: [email protected] (J.M.K. Kwong). 0304-3940/$ – see front matter © 2006 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.neulet.2006.06.070

of transgenic mice expressing HSP72 [10,18] or those of rats injected with the herpes virus containing HSP72 genes also have been shown to be resistant to ischemia and seizures [25]. We previously observed that heat shock pre-conditioning increased the production of HSP72 in retinal ganglion cells (RGCs) in vitro and in vivo, and protected neurons against N-methyl-d-aspartatemediated excitotoxicity and glaucoma [3,9,16]. These findings suggest that regulation of HSP72 is essential to the protection of retinal neurons against noxious insults. In eukaryotic cells, the regulation of hsp genes requires the activation and translocation to the nucleus of a trans-regulatory protein, the heat shock factor (HSF), which recognizes modular sequence elements referred to as the heat shock element (HSE) located within the hsp gene promoters [13]. At least three HSFs (HSF1 to 3) have been isolated from the human, mouse and chicken genomes, while an additional factor, HSF4 has recently been described in human cells. HSF1 and HSF2 are believed to be the major members in HSF family and HSF3 is found in avian. The existence of multiple HSFs in vertebrates suggests

192

J.M.K. Kwong et al. / Neuroscience Letters 405 (2006) 191–195

that different HSFs mediate the response to various forms of physiological and environmental stimuli [11,17]. Recent reports indicated that the DNA binding activities of HSF1 and HSF2 were altered in cell-type- and stress-specific manner [15,19,22]. To implicate the induction of HSP as a therapeutic tool for optic neuropathies, it is fundamental to understanding the transcriptional pathways of HSP in retinal neurons and especially RGCs. Therefore, we performed (i) immunohistochemistry, RT-PCR and western blotting on unstressed rat retinas to examine the localization and expression of HSF1 and HSF2, and (ii) western blotting of dissociated RGC suspension harvested by Thy-1 antibody coated magnetic beads. All experiments were approved by the Animal Research Committee of the University of California, Los Angeles and were performed in compliance with the Association for Research in Ophthalmic and Vision Research. Consistent with our animal studies [5,8,16], albino male Wistar rats weighing 250–300 g were used. Adult rats were kept in the animal room for 1 week before experiments, and allowed free access to food and water. The animal room was lit with fluorescent lights (330 lux) turned on at 3 a.m. and off at 3 p.m., and was maintained at 21 ◦ C. To examine the localization of HSFs in the retinas, fluorescence immunohistochemistry was performed as previously described [8]. Animals were deeply anesthetized with intramuscular injections of 0.8 ml/kg of a cocktail containing ketamine, xylazine, and acepromazine, and transcardially perfused with ice cold 4% paraformaldehyde in 0.1 M phosphate buffer. The enucleated eyeballs were fixed for 4 h, incubated with 30% sucrose overnight at 4 ◦ C, and embedded in OCT compound (Sakura Finetec, Torrance, CA, USA). Ten-micrometer thick sections were obtained along the vertical meridian through the optic nerve head. After washing with PBS containing 0.1% Triton X-100, the sections were incubated with blocking serum solution, primary antibody against HSF1 (rat monoclonal at 1:150; Chemicon, Temecula, CA, USA) or HSF2 (rat monoclonal at 1:250; Lab Vision Corporation, Fremont, CA, USA) at 4 ◦ C overnight, and secondary antibody. Anti-biotin antibody conjugated with Cy-3 (Sigma, St. Louis, MO, USA) and 4 ,6diamidino-2-phenylindole, dihydrochloride (DAPI; Molecular Probes, Inc., Eugene, OR, USA) were used for fluorescent labeling and nuclear counter-staining respectively. Retinal sections were examined with a fluorescence microscope (Axioplan, Carl Zeiss, Oberkochen, Germany) and imaged. For negative controls, retinal sections were incubated with blocking solution by replacing the primary antibody or with another species secondary antibody by replacing the original secondary antibody. After euthanasia with inhalation of CO2 , the retinas were dissected immediately and the total retinal RNA was extracted with RNAzol B (Tel-Test, Friendswood, TX, USA) and purified with RNeasy MinElute Cleanup kit (Qiagen, Valencia, CA, USA). After quantification by spectrophotometry at 260 nm and integrity analysis with denaturing agarose gel electrophoresis (1% agarose, 2.2 M formaldehyde), retinal RNA was reverse transcribed to cDNA with SuperScript First-strand Synthesis System (Invitrogen, Carlsbad, CA, USA). The oligonucleotide pairs were as follows: HSF1, 5 -CTGGTGCACTACACGGCTCA-3 (1151–1170) and 5 -GTTGTGCTGGCTTGACC-

TAG-3 (1467–1448) (GenBank accession number X83094). HSF2, 5 -GTAAGCTTGTCCGCCTGGAA-3 (1601–1620) and 5 -ATATGCCTAGTCAGCCAGCC-3 (1918–1889) (GenBank accession number NM031694). Amplification conditions were as follows: hot start of 2 min at 95 ◦ C; 30 cycles of denaturing (95 ◦ C for 30 s), annealing (60 ◦ C for 15 s), and extension (72 ◦ C for 30 s); and a final extension of 7 min at 72 ◦ C. The PCR products were separated by electrophoresis in a 2% agarose gel, visualized under UV light in the presence of ethidium bromide and photographed. Four freshly dissected retinas were homogenized in lysis buffer (10 mM Tris, pH 7.4, 150 mM NaCl, 1 mM EDTA) supplemented with protease and phosphatase inhibitor cocktails according to published procedures [4]. One microgram of protein from each sample was subjected to electrophoresis (Mini-Protean II system; Bio-Rad, Hercules, CA, USA) on 12% polyacrylamide gels and transferred to polyvinylidene fluoride membrane (Immobilon-P, Millipore Corporation; Bedford, MA, USA). After blocking with 5% non-fat powdered milk, the membrane was incubated with primary antibodies against HSF1 (1:5000), HSF2 (1:5000), HSP72 (1:5000; Stressgen Biotechnologies, Victoria, BC, Canada) or GFAP (1:10000; Sigma) overnight at 4 ◦ C and followed by incubation with peroxidaseconjugated secondary antibodies. The signals were visualized using an ECL plus Detection Kit (Amersham Biosciences; England). Two independent experiments were performed (n = 4 each). To confirm the expression of HSFs in RGCs, the procedures of RGC isolation using magnetic beads were adopted and modified [20,23]. Four freshly isolated adult rat retinas were dissociated in D-PBS without Ca2+ and Mg2+ containing 20 U/ml papain, 1 mM l-cystein and 0.005% DNase I (Worthington, Lakewood, NJ, USA) at 37 ◦ C for 30 min. The retinas were gently titrated with 1 mL pipette in a solution containing D-PBS, 0.15% trysin inhibitor, 0.15% BSA (Roche, Indianapolis, IN, USA) and 0.005% DNase I. Single cells were obtained after centrifugation and re-suspended in D-PBS containing 0.1% BSA. Cells were purified using paramagnetic beads with specific antibodies (Dynal, Oslo, Norway) in sequential steps. Macrophages and adherent cells were removed by attachment to CD11b/c monoclonal antibody (BD Pharmingen, San Diego, CA, USA) coated beads. Cells were then selected with magnetic beads coated with the Thy-1.1 monoclonal antibody (Chemicon, Temecula, CA, USA). After multiple washes removing the non-adherent cells, the attached Thy-1 positive cells, referred as RGCs, were released by incubation in DNase buffer and then transferred to Neurobasal A medium (Invitrogen, Carlsbad, CA, USA) containing 0.8% BSA. RGCs were centrifuged and resuspended in serum-free medium. Approximately 95% of cells were Thy-1 positive as estimated by immunocytochemistry (data not shown). Western blot analysis was performed on the cell suspension as described above. Two independent experiments were performed (n = 4 each). To visualize the localization of HSF1 and HSF2 in RGCs, retrograde labeling using Fluorogold (FG; Fluorochrome, Denver, CO, USA) were performed in rats as published elsewhere [6]. A Gelfoam (Upjohn, Kalamazoo, MI, USA) soaked with 6% Fluorogold was applied onto the surface

J.M.K. Kwong et al. / Neuroscience Letters 405 (2006) 191–195

193

Fig. 1. Immunohistochemistry of HSF1 and HSF2 in rat retina. Moderate to intense HSF1 immunoreactivity (arrows) was shown in cells (a, c) in the RGCL and was mainly present in cytoplasm as shown by counterstaining with DAPI (b, c). Moderate HSF1 immunoreactivity was also noted in IPL, OPL, and inner segments while strong immunoreactivity was detected in the outer segments of photoreceptor cells. Intense immunoreactivity of HSF2 was predominantly noted in the cells (d, f) in the RGCL and appeared to be cytoplasmic as counterstained by DAPI (e, f). RGCL, retinal ganglion cell layer; IPL, inner plexiform layer; INL, inner nuclear layer; ONL, outer nuclear layer; RPE, retinal pigment epithelium. Bar = 20 ␮m.

of both superior colliculi after removal of the overlying cortex for retrograde labeling of the RGCs. One week later, animals were transcardially perfused with paraformaldehyde for immunohistochemistry as described above. Immunohistochemical analysis showed that there was diffuse immunoreactivity of HSF1 over the unstressed retina in the adult rat. However, moderate to intense immunoreactivity of HSF1 was noted mainly in the cells of the RGC layer (Fig. 1a and c) and; the positive immunoreactivity appeared to be cytoplasmic as counter-stained with DAPI (Fig. 1b and c). In the photoreceptor layer, strong immnoreactivity of HSF1 was observed in the outer segment while moderate immunoreactivity was noted in the inner segment. In contrast, the positive immunoreactivity of HSF2 was predominantly noted in the cells of the RGC layer but no remarkable immunoreactivity was observed in other retinal layers (Fig. 1d and e). The positive immunoreactivity of HSF2 in the RGC layer was detected in the cytoplasm of the cells (Fig. 1d and f). Total retinal mRNA was extracted for RT-PCR analysis on mRNA of HSF1 and HSF2 genes. Images of autoradiography show the positive bands approximately at 300 bp, which correspond to the PCR products for HSF1 and HSF2 genes (Fig. 2). Fig. 3 shows the results of western blotting analysis of the whole retinal proteins and the fraction containing RGCs

isolated with the Thy-1 magnetic beads. There were positive immunoreactive bands corresponding to HSF1, HSF2, HSP72 and GFAP in the total retinal protein sample. However, only positive immunoreactive bands for HSF1, HSF2 and HSP72 were detected in the RGC fraction; no GFAP was detected in this sample, indicating that the sample did not contain glial cells. Immunohistochemistry of HSF1 and HSF2 on the retrogradely labeled retinal sections demonstrated that Fluoro-gold labeled RGCs were predominantly labeled by HSF1 and HSF2 (Fig. 4). Our previous in vitro studies demonstrated that sub-lethal hyperthermic and hypoxic stress enhanced the induction of

Fig. 2. Expression of HSF1 and HSF2 mRNAs in rat retina. Photograph shows the gel electrophoresis of the products of HSF1 and HSF2 after RT-PCR. Total rat retinal RNAs was used.

194

J.M.K. Kwong et al. / Neuroscience Letters 405 (2006) 191–195

Fig. 3. Immunoblot analyses of rat retinal extracts and isolated RGCs. Note the immunoreactive bands of HSF1, HSF2 and HSP72 in both retinal extract and isolated RGCs. Positive immunoreactivity of GFAP, a glial cell marker, was only detected in retinal extract but not in the fractions of isolated RGCs indicating that the sample did not consist of glial cells. R, whole retinal extract; RGC, isolated RGCs using Thy-1 immuno-labeled magnetic beads; GFAP, glial fibrillary acidic protein.

HSP72 and protected RGCs against later excitotoxic and anoxic insults [3]. In rat models of optic nerve injury, systemic stresses including administration of zinc, hyperthermia, and geranylgeranylacetone (GGA, a heat shock inducer) increased the expression of HSP72 in RGCs and prevented cell death [5,9,16]. However, the mechanism of the heat shock response in the adult rat retina is not yet understood. This study using western blotting and RT-PCR demonstrated that both HSF1 and HSF2 were expressed in the adult rat retina but were present in different spatial patterns. HSF2 was apparent only in the RGC layer but HSF1 was predominantly expressed in both the RGC layer and photoreceptor cells. The presence of HSF1 and HSF2 in RGCs was confirmed by western blotting on Thy-1 magnetic bead isolated cells. Consistent with other unstressed tissues, our immunohistochemical studies indicated that HSF1 and HSF2 are mainly expressed in the cytoplasm of RGCs. It is known that once exposed to stimuli, HSFs are activated and translocated into the nucleus and recognize HSE located within the hsp gene promoters to induce HSP synthesis.

In the unstressed rat retina, we observed that HSF1 was present in most retinal layers but expressed in higher basal levels in RGCs and photoreceptor cells while HSF2 was predominantly expressed in RGCs. Barbe and co-workers [2,21] demonstrated that whole body hyperthermia increased the synthesis of HSP72 protein and mRNA transcripts in photoreceptors and other retinal layers and subsequently preserved photoreceptors against light injury. Similarly, our laboratory demonstrated that there was increased production of HSP72 in RGCs after exposed to hyperthermia, zinc and GGA, and that HSP72 enhanced RGC survival against excitotoxicity [9] and glaucomatous damage in the rat [5,16]. Since the production of HSP72 depends on the activation of HSF1 during stress, we speculate that the basal level of HSF1 may correlate with the synthesis of HSP72. However, it was reported that the basal level of HSF1 may not simply relate to HSP72 production, and possibly higher levels of HSF1 may lower the threshold for the heat shock response in the presence of larger amounts of HSC70 (constitutive form) and HSP90 [22] or repress other non-heat shock genes. Xiao et al. [24] showed that the basal levels of HSPs were not affected by the lack of HSF1 in transgenic mice. HSF2 has been considered as an orphan member of the HSF family and found to be abundantly expressed and active for DNA binding in embryonic or differentiating cells such as mouse stem cells, embryonic carcinoma cells, and during hemin-mediated differentiation of human K562 erythroleukemia cells. In most adult tissues, HSF2 does not show significant HSE-binding activity [12]. However, we demonstrated that both HSF1 and HSF2 were expressed in relatively high levels in RGCs of adult retinas suggesting that HSF1 and HSF2 may play a physiological role in unstressed condition. Therefore, it will be interesting to further investigate the distribution and the roles of the constitutive levels of HSF1 and HSF2 in unstressed RGCs and study how HSFs regulate HSP72 synthesis after exposure to stress. Our study confirms that HSF1 and HSF2 are expressed in RGCs in adult rat retina. The existence of multiple HSFs in RGCs suggests that different HSFs mediate the response to different stimuli. In order to use heat shock response as a therapeutic tool in the future, further experiments should be performed to examine how the signaling pathway controls activation of HSFs.

Fig. 4. Immunohistochemistry of HSF1 and HSF2 in Fluoro-gold-labeled RGCs. Positive HSF1 immunoreactivity (arrows) was shown in RGCs (a, c) which were retrogradely labeled by Fluoro-gold (b, c). Some HSF1 immuno-positive cells were not labeled by Fluoro-gold (arrowheads). Positive HSF2 immunoreactivity (d, f; arrows) was predominantly noted in Fluoro-gold labeled RGCs (e, f). Bar = 20 ␮m.

J.M.K. Kwong et al. / Neuroscience Letters 405 (2006) 191–195

Acknowledgements This work was supported by The Gerald Oppenheimer Family Foundation Center for the Prevention of Eye Disease, Los Angeles, CA (J.M.K. Kwong), Research to Prevent Blindness Lew R. Wasserman Merit Award and Physician-Scientist Award, New York, NY (J. Caprioli). References [1] V.R. Agash, F.-U. Hartl, Roles of molecular chaperones in cytoplasmic protein folding, Semin. Cell Dev. Biol. 11 (2000) 15–25. [2] M.F. Barbe, M. Tytell, D.J. Grower, W.J. Welch, Hyperthermia protects against light damage in the rat retina, Science 241 (1988) 1817–1820. [3] J. Caprioli, S. Kitano, J.E. Morgan, Hyperthermia and hypoxia increase tolerance of retinal ganglion cells to anoxia and excitotoxicity, Invest. Ophthalmol. Vis. Sci. 37 (1996) 2376–2381. [4] K. Chiu, T.T. Lam, W.W.Y. Li, J. Caprioli, J.M.K. Kwong, Calpain and N-methyl-D-aspartate (NMDA)-induced excitotoxicity in rat retinas, Brain Res. 1046 (2005) 207–215. [5] Y. Ishii, J.M.K. Kwong, J. Caprioli, Retinal ganglion cell protection with geranylgeranylacetone, a heat shock protein inducer, in a rat glaucoma model, Invest. Ophthalmol. Vis. Sci. 44 (2003) 1982–1992. [6] J.-Z. Ji, W. Elyaman, H.K. Yip, V.W.H. Lee, L.-W. Yick, J. Hugon, K.F. So, CNTF promotes survival of retinal ganglion cells after induction of ocular hypertension in rats: the possible involvement of STAT3 pathway, Eur. J. Neurosci. 19 (2004) 265–272. [7] C. Jolly, R.I. Morimoto, Role of the heat shock response and molecular chaperones in oncogenesis and cell death, J. Natl. Cancer Inst. 92 (2000) 1564–1572. [8] J.M.K. Kwong, J. Caprioli, Expression of phosphorylated c-Jun Nterminal protein kinase (JNK) in experimental glaucoma in rats, Exp. Eye Res. 82 (2006) 576–582. [9] J.M.K. Kwong, T.T. Lam, J. Caprioli, Hyperthermic pre-conditioning protects retinal neurons from N-methyl-D-aspartate (NMDA)-induced apoptosis in rat, Brain Res. 970 (2003) 119–130. [10] J.E. Lee, M.A. Yenari, G.H. Sun, L. Xu, M.R. Emond, D. Cheng, G.K. Steinberg, R.G. Giffard, Differential neuroprotection from human heat shock protein 70 overexpression in vitro and in vivo models of ischemia and ischemia-like conditions, Exp. Neurol. 170 (2001) 129–139. [11] C.J. Marcuccilli, S.K. Mathur, R.I. Morimoto, R.J. Miller, Regulatory differences in the stress response of hippocampal neurons and glial cells after heat shock, J. Neurosci. 16 (1996) 478–485.

195

[12] A. Mathew, S.K. Mathur, C. Jolly, S.G. Fox, S. Kim, R.I. Morimoto, Stress-specific activation and repression of heat shock factors 1 and 2, Mol. Cell Biol. 21 (2001) 7163–7171. [13] R.I. Morimoto, Cells in stress: transcriptional activation of heat shock genes, Science 259 (1993) 1409–1410. [14] R.I. Morimoto, M.G. Santoro, Stress-inducible responses and heat shock proteins: new pharmacologic targets for cytoprotection, Nat. Biotechnol. 16 (1998) 833–838. [15] A.J. Morrison, S.J. Rush, I.R. Brown, Heat shock transcription factors and the hsp70 induction response in brain and kidney of the hyperthermic rat during postnatal development, J. Neurochem. 75 (2000) 363–372. [16] K.H. Park, F. Cozier, O.C. Ong, J. Caprioli, Induction of heat shock protein 72 protects retinal ganglion cells in a rat glaucoma model, Invest. Ophthalmol. Vis. Sci. 42 (2001) 1522–1530. [17] L. Pirkkala, P. Nyk¨anen, L. Sistonen, Roles of the heat shock transcription factors in regulation of the heat shock response and beyond, FASEB J. 15 (2001) 1118–1131. [18] J.C. Plumier, A.M. Krueger, R.W. Currie, D. Kontoyiannis, G. Kollias, G.N. Pagoulatos, Transgenic mice expressing the human inducible Hsp70 have hippocampal neurons resistant to ischemic injury, Cell Stress Chaperones 2 (1997) 162–167. [19] L. Sistonen, K.D. Sarge, R.I. Morimoto, Human heat shock factors 1 and 2 are differentially activated and can synergistically induce hsp70 gene transcription, Mol. Cell Biol. 14 (1994) 2087–2099. [20] G. Tezel, M.B. Wax, Increased production of tumor necrosis factor-␣ glial cells exposed simulated ischemia or elevated hydrostatic pressure induces apoptosis in co-cultured retinal ganglion cells, J. Neurosci. 20 (2000) 8693–8700. [21] M. Tytell, M.F. Barbe, I.R. Brown, Induction of heat shock (stress) protein 70 and its mRNA in the normal and light-damaged rat retina after whole body hyperthermia, J. Neurosci. Res. 38 (1994) 19–31. [22] S.K. Van Why, S. Kim, J. Geibel, F.A. Seebach, M. Kashgarian, N.J. Siegel, Threshold for cellular disruption and activation of the stress response in renal epithelia, Am. J. Physiol. 277 (1999) F227–F234. [23] D.-Y. Wu, J.-Q. Zheng, M.A. McDonald, B. Chang, J.L. Twiss, PKC isozymes in the enhanced regrowth of retinal neuritis after optic nerve injury, Invest. Ophthalmol. Vis. Sci. 44 (2003) 2783–2790. [24] X. Xiao, X. Zuo, A.A. Davis, D.R. McMillan, B.B. Curry, J.A. Richardson, I.J. Benjamin, HSF1 is required for extra-embryonic development, postnatal growth and protection during inflammatory response in mice, EMBO J. 18 (1999) 5943–5952. [25] M.A. Yenari, S.L. Fink, G.H. Sun, L.K. Chang, M.K. Patel, D.M. Kunis, D. Onley, D.Y. Ho, R.M. Sapolsky, G.K. Steinberg, Gene therapy with HSP72 is neuroprotective in rat models of stroke and epilsy, Ann. Neurol. 44 (1998) 584–591.