Neuroprotective effects of atorvastatin against ... - Wiley Online Library

Journal of Neurochemistry, 2005, 92, 1386–1398

doi:10.1111/j.1471-4159.2004.02980.x

Neuroprotective effects of atorvastatin against glutamate-induced excitotoxicity in primary cortical neurones Julian Bo¨sel,*,1 Florin Gandor,*,1 Christoph Harms,* Michael Synowitz, Ulrike Harms,à Pierre Chryso Djoufack,§ Dirk Megow,* Ulrich Dirnagl,* Heide Ho¨rtnagl,à Klaus B. Fink§ and Matthias Endres* *Klinik und Poliklinik fu¨r Neurologie, Charite´ – Universita¨tsmedizin Berlin, Berlin, Germany Zellula¨re Neurowissenschaften, Max-Delbru¨ck-Centrum fu¨r Molekulare Medizin, Berlin-Buch, Berlin, Germany àInstitut fu¨r Pharmakologie und Toxikologie, Charite´ – Universita¨tsmedizin Berlin, Berlin, Germany §Institut fu¨r Pharmakologie und Toxikologie, Universita¨tsklinikum Bonn, Bonn, Germany

Abstract Statins [3-hydroxy-3-methylglutaryl-coenzyme A (HMG-CoA) reductase inhibitors] exert cholesterol-independent pleiotropic effects that include anti-thrombotic, anti-inflammatory, and antioxidative properties. Here, we examined direct protective effects of atorvastatin on neurones in different cell damage models in vitro. Primary cortical neurones were pre-treated with atorvastatin and then exposed to (i) glutamate, (ii) oxygen– glucose deprivation or (iii) several apoptosis-inducing compounds. Atorvastatin significantly protected from glutamate-induced excitotoxicity as evidenced by propidium iodide staining, nuclear morphology, release of lactate dehydrogenase, and mitochondrial tetrazolium metabolism, but not from oxygen–glucose deprivation or apoptotic cell death. This anti-

excitototoxic effect was evident with 2–4 days pre-treatment but not with daily administration or shorter-term pre-treatment. The protective properties occurred independently of 3-hydroxy3-methylglutaryl-CoA reductase inhibition because co-treatment with mevalonate or other isoprenoids did not reverse or attenuate neuroprotection. Atorvastatin attenuated the glutamate-induced increase of intracellular calcium, which was associated with a modulation of NMDA receptor function. Taken together, atorvastatin exerts specific anti-excitotoxic effects independent of 3-hydroxy-3-methylglutaryl-CoA reductase inhibition, which has potential therapeutic implications. Keywords: excitotoxicity, 3-hydroxy-3-methylglutaryl-coenzyme A reductase inhibitors, neuroprotection, statins. J. Neurochem. (2005) 92, 1386–1398.

Statins [3-hydroxy-3-methylglutaryl-coenzyme A (HMGCoA) reductase inhibitors] are potent cholesterol-lowering drugs used for the treatment of hypercholesterolaemia (Stein 2002) and coronary heart disease (Rader 2003). In addition, statins reduce stroke incidence and may also reduce the risk of Alzheimer’s disease (Hess et al. 2000; Cucchiara and Kasner 2001; Heart Protection Study Collaborative Group 2002). Moreover, recent clinical and experimental trials have shown that pre-treatment with statins improves stroke outcome (Endres et al. 1998; Endres and Laufs 2004; Greisenegger et al. 2004; Marti-Fabregas et al. 2004). Interestingly, cholesterol is not an established risk factor for stroke (Hebert et al. 1995; Amarenco 2001). Likewise, experimental studies have demonstrated that many of the protective effects of statins are cholesterol-independent. Their underlying protective mechanisms include pleiotropic effects on endotheliumdependent vasodilation, atherosclerotic plaque stabilization,

anti-thrombotic and anti-inflammatory effects (Bo¨sel and Endres 2002; Liao 2002; Tsiara et al. 2003).

1386

Received July 6, 2004; revised manuscript received November 8, 2004; accepted November 9, 2004. Address correspondence and reprint requests to Priv.-Doz. Dr med. Matthias Endres, Klinik und Poliklinik fu¨r Neurologie, Universtita¨tsmedizin Berlin – Charite´, Campus Mitte, Schumannstr. 20/21, D-10117 Berlin, Germany. E-mail: [email protected] 1 JB and FG contributed equally to this study. Abbreviations used: AMPA, alpha-amino-3-hydroxy-5-methylisoxazole-4-proprionic acid; [Ca2+]i, intracellular calcium levels (semiquantitative); DIV, day in vitro; FPP, farnesylpyrophosphate; GFAP, glial fibrillary acid protein; GGPP, geranylgeranylpyrophosphate; HMG, 3-hydroxy-3-methylglutaryl; LDH, lactate dehydrogenase; MAP-2, microtubule-associated protein-2; MTT, mitochondrial tetrazolium transformation; NF-jB, nuclear factor-jB; OGD, oxygen–glucose deprivation; PI, propidium iodide; PSD-95, postsynaptic density protein.

2005 International Society for Neurochemistry, J. Neurochem. (2005) 92, 1386–1398

Direct neuroprotective effects of statins 1387

It is presently unclear, however, whether direct effects on neurones, independent of the vascular component, may contribute to the beneficial effects statin treatment has on stroke or neurodegenerative disease. Although Zacco et al. (2003) recently described protective effects of statins on cultured neurones, several other reports, including our own, demonstrate that statins at micro- and millimolar concentrations have neurotoxic effects in vitro, such as the induction of apoptosis (Michikawa and Yanagisawa 1999; Tanaka et al. 2000; Garcia-Roman et al. 2001) or inhibition of neurite outgrowth (Fan et al. 2002; Meske et al. 2003; Schulz et al. 2004). Here, we demonstrate that high nanomolar to low micromolar concentrations of atorvastatin – only when given several days before the insult – protected cortical neurones in culture from glutamate-excitotoxicity. Atorvastatin, however, did not protect from oxygen–glucose deprivation (OGD), an in vitro model of ischaemic cell death, or apoptosis induced by the DNA damaging compound camptothecin or the pan-kinase inhibitor staurosporine. The anti-excitotoxic effect was independent of HMGCoA reductase inhibition, involved attenuation of intracellular Ca2+-increase and was associated with reduced NMDA receptor activity.

Materials and methods Materials Dimethyl sulfoxide, high range molecular weight standard and enzyme standard for kinetic lactate dehydrogenase (LDH)-test, glutamate, staurosporine, camptothecin, MK-801, NMDA, alphaamino-3-hydroxy-5-methylisoxazole-4-proprionic acid (AMPA), Hoechst dye 33342, goat anti-mouse FITC and goat anti-rabbit TRITC antibodies, paraformaldehyde, normal goat serum, propidium iodide, nifedipine and mevastatin were obtained from Sigma (Deisenhofen, Germany). Mouse microtubule-associated protein-2 (MAP-2) antibody, rabbit glial fibrillary acid protein (GFAP) antibody and polyclonal rabbit anti-NMDAR2B were from Chemicon (Temecula, CA, USA); PSD-95 (postsynaptic density protein) antibody was from Upstate (#05-427; Biomol, Hamburg, Germany); secondary horseradish-peroxidase conjugated anti-rabbit antibody was from Bio-Rad (Hercules, CA, USA). Chemiluminescence western blotting substrate (Lumi-Light plus western blotting substrate) was from Roche (Mannheim, Germany); kainate was from Tocris Cookson Inc. (Ellisville, MO, USA), x-conotoxin GVIA (H-6615) was from BACHEM (Germany); x-agatoxin IVA was from Peptide Institute (Japan); polyvinylidene difluoride membrane was from Immobilon P Millipore (Bedford, MA, USA); Omniscript Reverse Transcriptase kit was from Qiagen (Hilden, Germany). Ethylcholine aziridinium (AF64A) was prepared from acetylethylcholine mustard (Research Biochemicals International, Natick, MA, USA) according to Fisher et al. (1982). Alexa Fluor 488 phalloidin, Fluo-4-AM and the tetrazolium salt 3[4,5-dimethylthiazol-2-yl]-2,5-diphenyltetrazolium bromide (MTT) were from Molecular Probes (Eugene, OR, USA). Atorvastatin was kindly provided by Go¨decke AG, Freiburg, Germany. Neurobasal

medium, supplement B27 and TRIZOL reagent were from Gibco/ BRL Life Technologies (Eggenstein, Germany); modified Eagle’s medium, phosphate-buffered saline, HEPES-buffer, trypsin/EDTA, penicillin/streptomycin, L-glutamine, collagen-G and poly L-lysin were from Biochrom (Berlin, Germany). Primary neuronal cell culture Primary neuronal cultures of cerebral cortex were obtained from rat embryos (E16–18) according to Brewer (1995) with modifications as described before (Harms et al. 2001). Briefly, after dissection of cerebral cortices and the dissociation procedure, cells were plated in 96-, 24- or 6-well plates (200 000 cells per cm2). The 96-well plates were used for Ca2+ measurements, 24-well plates for evaluation of cellular damage, immunohistochemistry, propidium iodide staining and electrophysiology, and 6-well plates for mRNA measurements and immunoprecipitation experiments. Wells were pre-treated by incubation with poly-L-lysin (0.5% w/v in phosphate-buffered saline) for 1 h at room temperature (24C), then rinsed with phosphate-buffered saline, followed by incubation with coating medium (dissociation medium with 0.03& w/v collagen G) for 1 h at 37C, then rinsed twice with phosphate-buffered saline, before seeding cells in starter medium. Cultures were kept at 36.5C and 5% CO2 and fed every 4 days with cultivating medium (starter medium without glutamate) by replacing half of the medium. The exact composition of the serum-free culture medium is given in Brewer et al. (1993). Experiments with serum-free cultures were carried out between day in vitro (DIV) 8 and 10. We have demonstrated that neuronal purity is higher than 90% until DIV 14 (less than 10% astrocytes until DIV 14 and less than 1% micoglia until DIV 28, see also Lautenschlager et al. 2000). Injury paradigms Cells were pre-treated by addition of atorvastatin to the culture medium in the given concentrations and for the given durations before being subjected to the following injury paradigms. Although we applied a pre-treatment with 1 lM of atorvastatin for 4 days in most of the cases, we chose a lower concentration (100 nM) of atorvastatin for receptor agonist experiments assessing calcium dynamics because we had observed a slightly increased vulnerability of cells loaded with Fluo4-AM in the previous glutamate experiments leading to a higher basal toxicity of atorvastatin. For assessment of ‘full kill’, Triton-X-100 at 0.5% was added 1 h prior to measurement of LDH. Glutamate (stock-solutions in H2O) was added to the culture medium at a concentration of 50 lM for 1 h and then removed. NMDA, kainate and AMPA were added to the culture medium in concentrations of 20, 100 and 500 lM for 20 min (calcium measurements) or 1 h (damage assessments) and then removed. OGD: medium was removed from cells and preserved. Cells were rinsed with phosphate-buffered saline, then kept in an oxygen-free chamber for 135 min in a balanced salt solution free of glucose at pO2 < 2 mmHg, thereafter the preserved cell culture medium was restored as described previously (Bruer et al. 1997). Camptothecin was dissolved in H2O and added to the culture medium at a final concentration of 10 lM for 48 h as described previously (Stefanis et al. 1999). Staurosporine was dissolved in dimethyl sulfoxide (10 mM stock solution) and diluted with phosphate-buffered saline to give the final concentrations of 100 or 300 nM in culture (dimethyl sulfoxide final concentration


1388 J. Bo¨sel et al.

0.003%). AF64A: freshly prepared AF64A at a final concentration of 40 or 80 lM or an equivalent amount of vehicle was added to cortical cultures as described previously (Lautenschlager et al. 2000). Control cultures received the equivalent concentration of the respective vehicle. Quantification of damage At given time points the condition of cells was determined morphologically by phase contrast microscopy and damage assessed by the measurement of LDH release into the medium indicating loss of membrane integrity (Koh and Choi 1987). In most experiments, damage assessment was complemented by mitochondrial tetrazolium (MTT) assay, an assay based on the cleavage of the yellow tetrazolium salt MTT to purple formazan by mitochondrial enzymes in metabolically active cells (Denizot and Lang 1986). MTT assay was performed 24 h after damage. To assess nuclear morphology in diverse models of cell death, primary cortical neurones were stained with Hoechst dye 33342 at 1 lg/mL (final concentration 1 : 2000) after fixation with 4% paraformaldehyde and exposure to 0.1% Triton X-100 as described elsewhere (Farinelli and Greene 1996). Immunocytochemistry Immunocytochemistry was performed as described previously (Harms et al. 2001). Cortical neurones were fixed for 15 min with 4% paraformaldehyde, permeabilized with 0.1% Triton-X-100 in phosphate-buffered saline for 8 min, rinsed twice with phosphatebuffered saline and blocked with 10% goat serum in phosphatebuffered saline for 30 min. Primary antibodies were incubated for 1 h at room temperature (MAP-2, GFAP, both 1 : 100 in 3% goat serum) and subsequently developed with secondary antibodies coupled with FITC (green) or TRITC (red). After mounting with Fluoromount, pictures were taken with a fluorescence microscope (Leica, Bensheim, Germany). Propidium iodide staining Propidium iodide (PI) staining of neuronal cell cultures was performed as described previously (Harms et al. 2004). Briefly, cortical neurones were incubated for 1 min with 0.02 mg/mL PI (stock solution 1 mg/mL, 1 : 50) in medium with gentle shaking and rinsed once with phosphate-buffered saline. Conditioned medium was reapplied and phase contrast and fluorescent pictures were taken immediately using an inverse fluorescence microscope with a digital camera (Leica). Cell counts Cell counts were performed from merged phase contrast micrographs and red fluorescent pictures were stained with PI. Neurones with dendrites that did not show any PI-positive signal in their nuclei were designated as intact, viable neurones. Mean intact neurones in three representative high power fields were counted. Electrophysiology Membrane currents of neurones cultured on glass cover-slips were measured with the patch-clamp technique in the whole cell configuration at room temperature. Current signals were amplified with conventional electronics (EPC-9 amplifier, HEKA Electronics, Lambrecht/Pfalz, Germany) filtered at 3 kHz and sampled at 5 kHz by an interface connected to a PC (IBM compatible), which

controlled the amplifier output. The acquisition and the analysis of the data were performed with WinTIDA software (HEKA). Capacitive currents and series resistance compensation were performed. Recording pipettes were fabricated from borosilicate glass capillaries (Hilgenberg, Malsfeld, Germany). Pipette resistance was 4 ± 8 MW. Series resistances were typically 16 ± 33 MW and series compensation was set to 80%. The pipette solution for patch-clamp recordings contained the following (in mM): 130 KCl, 2 MgCl2, 0.5 CaCl2, 5 EGTA, and 10 mM HEPES. The pH was adjusted with KOH to 7.4. To obtain an estimate of the resting potential, we noted the membrane potential at the time of establishing the whole cell configuration. Cell membrane potential was held at )60 mV. NMDA was applied by changing the superfusion in the bath. The flow rate was 20 mL/min and a complete exchange of solutions in the bath (volume 200 lL) waschieved in < s. mRNA isolation and reverse transcription–polymerase chain reaction Cells were harvested and total RNA was isolated from cells using TRIZOL reagent. First-strand cDNA was synthetized from total RNA with the Omniscript Reverse Transcriptase kit and amplified by PCR. Primer sequences, PCR conditions and amplicon length were as follows: NR2A: 5¢-TTATTGGGAGATGTCCCTCG-3¢ (sense) and 5¢-CACGTCTATTGCTGCAGGAA-3¢ (antisense); 30 s at 94C, 30 s at 57C, and 60 s at 72C; 42 cycles and 225 bp; NR2B: 5¢-ATCAGTGCTTGCTTCACGG-3¢ (sense) and 5¢-GGGTTGGACTGGTTCCCTAT-3¢ (antisense); 30 s at 94C, 30 s at 57C, and 60 s at 72C; 35 cycles and 182 bp; NR2C: 5¢-CAGCCCAGACAGCATGTCT-3¢ (sense) and 5¢-ACCCCACTGTCCCTGTAGC-3¢ (antisense); 30 s at 94C, 30 s at 57C, and 60 s at 72C; 41 cycles and 179 bp; NR2D: 5¢-CGATGGCGTCTGGAATGG-3¢ (sense) and 5¢-CTGGCAAGAAAGATGACCGC-3¢ (antisense); 30 s at 94C, 30 s at 57C, and 60 s at 72C; 42 cycles and 485 bp (Freeman et al. 1998). The same cDNA was used for b-actin amplification (5¢-ATGGATGACGATATCGCT-3¢ (sense) and 5¢-ATGAGGTAGTCTGTCAGG T-3¢ (antisense); 45 s at 94C, 30 s at 56C, and 60 s at 72C; 23 cycles and 570 bp (Zhang et al. 2002) to confirm that equal amounts of RNA were reversely transcribed. For semiquantitative analysis, PCR cycles were chosen within the exponential phase of the cDNA amplification. Equal amounts of RT–PCR products were separated on 2% agarose gel. Immunoprecipitation Cultures were incubated with 0.5 mL lysis buffer per well (150 mM NaCl, 10 mM Tris-HCl, 1 mM EDTA, 1 mM EGTA, 1% Triton X-100, Complete mini Protease Inhibitor Roche, 1 mM phenylmethylsulphonyl fluoride, and 50 mM sodium orthovanadate; pH 7.4) for 30 min with agitation at 4C. After lysis, cells were scraped, transferred into 1.5-mL tubes, sonicated, and frozen at )80C. After thawing, homogenates were centrifuged at 100 000 g for 30 min at 4C and pellets were solubilized under non-denaturing conditions according to Sans et al. (2000). The supernatant was incubated with monoclonal anti-PSD-95 antibody (Upstate 05-427) pre-coupled with protein A agarose overnight. The immunoprecipitates were briefly centrifuged at 2000 g and washed three times in Tris buffer (50 mM Tris-HCl, pH 7.4, 0.1% Triton X-100; 1% sodium deoxycholate), and separated by sodium dodecyl sulfate– polyacrylamide gel electrophoresis.



Sodium dodecyl sulphate–polyacrylamide gel electrophoresis and western blotting For each sample 25 lg total protein was loaded on a 5% sodium dodecyl sulfate–polyacrylamide gel electrophoresis gel. After electrophoresis, protein was transferred to a polyvinylidene difluoride membrane. Blots were blocked overnight with 5% non-fat dry milk in Tris-buffered saline with Tween 20 (50 mM Tris, pH 7.5, 150 mM NaCl, 0.05% Tween 20) and probed with polyclonal rabbit anti-NR2B at 1 : 2500 in 5% non-fat milk/Tris-buffered saline with Tween 20. Immunoblots were processed with secondary horseradish peroxidase conjugated anti-rabbit antibody. For signal detection we used Lumi-Light plus western blotting substrate (Roche) and a Roche LumiImager. Measurement of intracellular Ca2+ Intracellular free Ca2+-levels were assessed semiquantitatively according to Minta et al. (1989) and Grynkiewicz et al. (1985) by loading cells with 5 lM of the cell-permeant fluorescent calciumindicator Fluo-4 AM for 45 min at room temperature, then rinsing cells once with phosphate-buffered saline and subjecting them to fresh medium before the experiment. The fluorescent signal was measured at indicated time points using a multi-well fluorescence plate reader (CytoFluor II, PerSeptive Biosystems, Framingham, MA, USA). Data are presented as difference in relative fluorescence units (RFUs) between indicated treatments and controls or percentage of relative fluorescence of controls. Measurement of intracellular cholesterol Atorvastatin-treated cells were harvested from 6-well plates by washing out medium twice with phosphate-buffered saline, scraping off cells carefully and collecting material in small volumes of phosphate-buffered saline. Samples were mixed 1 : 1 with lysis buffer (phosphate-buffered saline + 2.5 mM cholic acid + 0.05% Triton X-100), sonicated and heated for 5 min to 95C to denaturate enzymes, which could interfere with the cholesterol assay. Total cholesterol per well was measured after enzymatic hydrolysis of cholesterol esters by coupled cholesterol oxidase and peroxidase reaction followed by assessment of Amplex Red fluorescence in a CytoFluor fluorescence reader. Data analysis Data are presented as mean ± SD and were pooled from three or more independent experiments (i.e. obtained from three independent preparations of neuronal cultures), unless stated otherwise. For statistical analysis one-way ANOVA followed by Tukey’s post hoc test was applied in most of the cases. In the rare cases of a failed normality test, ANOVA on ranks and a subsequent Dunn’s test were performed. For electrophysiological experiments, a Wilcoxon rank test was performed. p < 0.05 was considered statistically significant.

Results

Atorvastatin protects from glutamate excitotoxicity Primary cortical neurones were pre-treated with atorvastatin (1 lM) for 96 h and then subjected to 50 lM glutamate at DIV 10. Immunostaining with neurone- and glia-specific antibodies demonstrated that primary cultures contained less

than 10% glia (Fig. 1a). The majority of neurones treated with glutamate showed a loss of MAP-2 positive dendrites, which was not visible after atorvastatin pre-treatment (Fig. 1a). Phase contrast microscopy images merged with propidium iodide stainings (Fig. 1b) and cell counting (Fig. 1c) demonstrated that atorvastatin pre-treatment significantly protected cortical neurones from glutamate-excitotoxicity. To further determine dose- and time-dependence, neuronal cultures were pre-treated with different concentrations of atorvastatin for 24, 48, 72, or 96 h. The release of LDH into the culture medium (reflecting loss of membrane integrity) and MTT metabolism (reflecting mitochondrial function) were measured 24 h after glutamate exposure. Figure 2 demonstrates that significant protection was observed with 100 nM and 1 lM atorvastatin (Figs 2a and b) and pre-treatment periods of 48–96 h but not 24 h (Figs 2c and d). Figure 2(e) presents a single exemplary experiment in which maximal LDH release (‘full kill’) was induced by Triton X-100 and absolute LDH values are given. Similarly, protection was also observed with another HMG-CoA reductase inhibitor, mevastatin (Fig. 2f). Concentrations of 10 lM atorvastatin or higher were neurotoxic, as reported previously (Schulz et al. 2004), and accumulation of the statin by daily repeated application did not enhance protection. Pre-treatment periods longer than 96 h (i.e. 120 and 144 h) did not further augment the protective effect. Differential effects on excitotoxicity versus apoptosis and oxygen glucose deprivation Next, we tested whether the protective effects of atorvastatin are restricted to excitotoxicity induced by glutamate or whether they extend to apoptotic and/or ischaemic cell death. Neurones were pre-treated with 1 lM atorvastatin for 96 h (according to the optimal treatment regimen, see Fig. 2) and were exposed to (i) 50 lM glutamate, (ii) 135 min OGD as an in vitro model for ischaemic cell death, or (iii) 10 lM camptothecin, which is a well-characterized inducer of apoptosis (all injuries at DIV 10). Figure 3 demonstrates that atorvastatin protected from glutamate-induced cell death but not from OGD or camptothecin-induced apoptosis. In contrast, there was even a significant enhancement of camptothecin-induced apoptosis in neurones treated with atorvastastin compared to vehicle-controls (Fig. 3). Similarly, atorvastatin did not protect from apoptotic cell death induced by the pan-kinase inhibitor staurosporine (100 or 300 nM for 24 and 48 h) and enhanced (by 1.4-fold) apoptotic cell death caused by the cholinergic toxin ethylcholine aziridinium (AF64A) (40 or 80 lM for 24 and 48 h), well-characterized inducers of apoptotic cell death (Stefanis et al. 1999; Harms et al. 2001, 2004), data not shown. Together, atorvastatin pre-treatment selectively protected neurones from excitotoxicity but not from OGD, and it even significantly augmented apoptotic cell death.



Fig. 1 Atorvastatin protects cortical neurones from glutamate-induced cell death. Primary cortical rat neurones were pre-treated with 1 lM atorvastatin (Atorva) or vehicle (Control) for 4 days. Neurones were exposed to 50 lM glutamate (Glu) for 1 h. Following fixation, cells were stained with the neuronal marker microtubule-associated protein2 (green) or astroglial marker glial fibrillary acid protein (red) 24 h after treatment with glutamate. Bar, 20 lm (a). Phase contrast images were merged with propidium iodide (PI) fluorescence micrographs 24 h after glutamate exposure. Few PI-positive cells in control cultures relate to

the culture preparation. Glutamate-induced cell death is characterized by expanded or ruptured cytoplasms and shrunken nuclei with a strong PI-positive DNA-signal. Bar, 20 lm (b). Intact (i.e. living) neurones were counted and bar graphs represent mean numbers of intact neurones counted in three representative high power fields. Total cell numbers were 128 ± 4 per high power field and did not differ significantly between different experiments, **p < 0.001 vs. glutamate (c). Atorva, atorvastatin; Con, control; Glu, glutamate.

Anti-excitotoxic effects of atorvastatin are independent of 3-hydroxy-3-methylglutaryl-coenzyme A reductase inhibition Most of the biological effects of statins are mediated by inhibition of HMG-CoA reductase. To investigate whether the anti-excitotoxic properties of atorvastatin pre-treatment relate to HMG-CoA reductase inhibition, we co-treated neuronal cultures with mevalonate, which is the product of the HMG-CoA reductase reaction, as well as other downstream isoprenoids of the mevalonate cholesterol pathway, i.e. geranylgeranylpyrophosphate (GGPP), farnesylpyrophosphate (FPP), squalene, and LDL-cholesterol. Four-day pre-treatment with 1 lM atorvastatin significantly reduced total intracellular cholesterol levels from 23.53 to 12.66 mg/mL, reflecting successful cell entry and intracellular activity. However, Fig. 4 demonstrates that protection from glutamate excitotoxicity could not at all be

reversed or attenuated by co-treatment with various concentrations of mevalonate, the product of the HMGCoA reductase reaction. Moreover, several modifications of the mevalonate treatment protocol, i.e. daily repeated application or addition just before the insult did also not attenuate neuroprotection by atorvastatin, i.e. (i) administration of 1, 10, 100, 200, 1000, or 10 000 lM; (ii) hydrolysed vs. non-hydrolysed form; (iii) single vs. constant administration 4 days before insult or daily repeated administration of the given concentrations (data not shown). Similarly, co-treatment with isoprenoid intermediates, GGPP, FPP (at 0.001, 0.01, 0.1, 1, 5, 10, 50, 100 lM and single vs. repeated administrations), squalene (at 1–50 lM) or LDL-cholesterol itself (100 lg/mL) did also not bypass the anti-excitotoxic effect of atorvastatin pre-treatment (see Schulz et al. 2004 for detailed description of the treatment protocols).



effects of 1 lM atorvastatin in cortical neuronal cultures (Fig. 4). Taken together, our results indicate that the neuroprotective effects of low-dose atorvastatin are independent of HMG-CoA-reductase inhibition. Atorvastatin pre-treatment reduces glutamate-induced intracellular calcium increase To investigate the potential underlying mechanism of atorvastatin’s anti-excitotoxic properties, we measured intracellular calcium levels ([Ca2+]i) after exposure to 50 lM glutamate, as intracellular calcium overload is a prominent feature of excitotoxicity. Fluorometric [Ca2+]i measurements 20 min after 50 lM glutamate demonstrated significantly lower [Ca2+]i in neurones pre-treated with 1 lM atorvastatin for 96 h (Fig. 5a). Figure 5(b) visualizes the differences in [Ca2+]i dynamics in vehicle- vs. atorvastatin-treated neurones. The dose- and time-dependent effects of atorvastatin pre-treatment on [Ca2+]i (Figs 5c and d) closely paralleled the dynamics observed in the neuroprotection experiments (Figs 1 and 2). Likewise, reduction of [Ca2+]i overload by atorvastatin pre-treatment was also not attenuated by mevalonate co-treatment (Fig. 5e). In summary, the neuroprotective effects of atorvastatin correlate with a reduction of [Ca2+]i increase. Fig. 2 Neuroprotection by atorvastatin and mevastatin is dose- and time-dependent. Primary cortical rat neurones were pre-treated with atorvastatin (Ato) or mevastastin (Meva) at different concentrations and time intervals. Neurones were exposed to 50 lM glutamate (Glu) for 1 h. Cellular damage was quantified by release of lactate dehydrogenase (LDH) into the culture medium (a, c, e and f) or by reduction of MTT signal reflecting mitochondrial metabolism compromise (b and d). (a), (b) and (f) show dose-dependent protection after 4 days pretreatment with atorvastastin or mevastatin. (c) and (d) show timedependent protection after 1 lM atorvastatin. (e) displays absolute LDH values (U/mL) and an additional bar graph showing Triton-X 100induced maximal LDH release (‘full kill’). Mean + SD (percentage of control); *p < 0.05; **p < 0.005. Data are pooled from three independent experiments with n ¼ 20–24 wells (a and c) or with n ¼ 8 wells (b, d, e and f). Absolute LDH values (U/mL) of untreated controls: 84 ± 28 (a), 62 ± 20 (b), 77 ± 8 (e); 79 ± 8 (f); for MTT (OD): 853 ± 51 (b), 697 ± 23 (d).

We have previously demonstrated that atorvastatin at concentrations of 10 lM and higher induces cell death in primary cortical neurones by a GGPP-dependent pathway (Schulz et al. 2004). To rule out that the mevalonate cotreatment protocol was insufficient to bypass HMG-CoA reductase inhibition and to increase intracellular mevalonate levels in the presence of atorvastatin, we performed mevalonate co-administration experiments to reverse the toxic effects of 20 lM atorvastatin within the identical cell culture preparation and experimental setup. Mevalonate dosedependently reversed the toxic effect of 20 lM atorvastatin; however, it completely failed to attenuate the anti-excitotoxic

Effects of atorvastatin pre-treatment on NMDA, AMPA and kainate-induced injury Next, we differentiated whether NMDA, AMPA and/or kainate receptors are involved in the mechanisms underlying the protective effects of atorvastatin on glutamate-mediated cell death. Therefore, we pre-treated neurones with atorvastatin or vehicle for 4 days, then exposed them to increasing concentrations (20, 100, 500 lM) of NMDA, AMPA, or kainate, respectively, and measured [Ca2+]i along with markers of cell death. To avoid interference with voltagedependent Ca2+-channels, voltage-dependent Ca2+-channel blockers nifedipine (1 lM), x-conotoxin GVIA (0.1 lM) and x-agatoxin IVA (0.2 lM) were co-administered. Pre-treatment with atorvastatin caused a highly significant reduction of NMDA-induced LDH-release of approximately 50%, particularly in higher agonist concentrations, whereas reductions of AMPA- and kainate-induced LDH-release were less prominent (Fig. 6a). Moreover, NMDA-induced decrease in MTT metabolism was significantly attenuated by atorvastatin pre-treatment (approximately 40% decrease in non-treated vs. approximately 20% decrease in treated cells), whereas that induced by AMPA or kainate was unaffected (Fig. 6b). Similarly, NMDA-induced [Ca2+]i increase was significantly reduced by atorvastatin pre-treatment, whereas that induced by AMPA or kainate was not. When MK-801 (10 lM), an NMDA receptor antagonist, was administered 30 min prior to AMPA or kainate exposure, cellular damage was blunted in vehicle- and atorvastatin-pre-treated cells, and [Ca2+]i increases were not attenuated by atorvastastin pre-treatment



Fig. 3 Atorvastatin exerts both anti-excitotoxic and pro-apoptotic effects. Primary cortical rat cultures were pre-treated with 1 lM atorvastatin (Atorva) or vehicle for 4 days and subjected to 50 lM glutamate, 10 lM camptothecin (Campto), 135 min combined oxygen– glucose deprivation (OGD). Damage was assessed by changes in nuclear morphology (Hoechst-staining), lactate dehydrogenase (LDH)-release, and MTT-conversion. Untreated control cells display the typical ellipsoid, smoothly and regularly limited morphology of healthy neuronal nuclei (Control). In the glutamate model, protection is reflected by a reduction of the number of small, condensed, excitotoxically damaged nuclei in the atorvastatin-treated cultures as compared to the vehicle-treated ones. Induction of apoptosis by camptothecin (Campto) increases the number of heterogeneously

stained, irregularly lined nuclei, which is slightly enhanced by atorvastatin. In the mixed apoptotic/excitotoxic OGD-model (OGD) atorvastatin pre-treatment decreases the number of condensed, excitotoxically destroyed nuclei but at the same time increases the number of heterogeneously structured, apoptotic nuclei. This qualitative outcome is reflected quantitatively by bar graphs representing the respective LDH and MTT data in the lower panel as mean + SD (percentage of control); *p < 0.05; **p < 0.005. Data were pooled from three independent experiments with n ¼ 24–32 wells. Absolute values of untreated controls in glutamate, camptothecin and OGD experiments were: LDH (U/mL): 31 ± 6, 60 ± 15, and 87 ± 9; MTT (OD): 850 ± 45, 703 ± 31, and 1225 ± 181.

(data not shown). These results suggest that modulation of NMDA-receptor activity plays an important role for the neuroprotective effects of atorvastatin.

4 days and NMDA receptor function was investigated by whole-cell patch-clamp electrophysiological recordings. Figure 7 demonstrates that treatment with atorvastatin moderately but significantly decreased NMDA(100 lM)-induced whole-cell currents compared to vehicle-treated control cells. This indicates that the electrophysiological properties of the NMDA receptors are modified by atorvastatin pre-treatment and contribute to the reduction in [Ca2+]i increase.

NMDA-induced whole-cell currents are attenuated by atorvastatin pre-treatment To analyse NMDA receptor modulation by atorvastatin, rat cortical neurones were pre-treated with 1 lM atorvastatin for



Fig. 4 Anti-excitotoxic effects of atorvastatin are independent of HMG-CoA reductase inhibition. Primary cortical rat cultures were pretreated with atorvastatin (Atorva) or vehicle plus mevalonate (Mev) or vehicle for 4 days and then subjected to 50 lM glutamate (Glu) at DIV 10. Damage was assessed by lactate dehydrogenase (LDH)release. As shown on the left side of the diagram, atorvastastin at 1 lM protected from glutamate-induced cell death, which was not prevented by co-incubation with mevalonate. As shown on the right side, in the same experimental setup but without glutamate insult, 20 lM atorvastatin conferred cell death, which was significantly attenuated by mevalonate. Mean + SD (percentage of controls); *p < 0.05; **p < 0.005. Data represent three independent experiments with n ¼ 20 wells. Absolute value for untreated control for LDH (U/mL): 64 ± 11.

Atorvastatin pre-treatment does not affect mRNA expression of NMDA receptor subunits NR2A, 2B, 2C, or 2D To further elucidate the underlying mechanisms, mRNA expression of NR2 subunits 2A, 2B, 2C, and 2D (also called grin2a, 2b, 2c, and 2d, respectively) after pre-treatment of rat cortical neurones with 1 lM atorvastatin for 4 days was determined. We did not detect any significant differences in

Fig. 5 Atorvastatin pre-treatment protects from glutamate-induced Ca2+i increase. Primary cortical rat cultures loaded with the intracellular calcium-indicator Fluo-4AM were pre-treated with 1 lM atorvastatin (Ato) or vehicle (veh) for 4 days and then exposed to 50 lM glutamate (Glu). Ca2+i levels measured 20 min after glutamate exposure were signifcantly lower in atorvastastin pre-treated neurones (a), which is visualized in an exemplary experiment with calcium measurements every minute (b). Protection from Ca2+i overload was dependent on the concentration (c) and time (d) of atorvastatin pretreatment. Co-administration of mevalonate (Mev) did not reverse atorvastatin’s effects on Ca2+i (e). Mean + SD (percentage of controls), *p < 0.05; **p < 0.005. Data were pooled from four (a), three (c), and two (d, e) independent experiments with n ¼ 23, 18, 12 wells, respectively. Data for the exemplary (b) is from one experiment with n ¼ 6 wells. Absolute values of untreated controls (RFUs): 290 ± 46 (a), 352 ± 33 (c), 243 ± 28 (d), 300 ± 44 (e).



Fig. 6 Differentiation of glutamatergic receptors regulated by atorvastatin pre-treatment. Primary cortical rat cultures that were left unloaded or were loaded with the intracellular calcium indicator Fluo4AM were pre-treated with 100 nM atorvastatin (Atorva, grey bars) or vehicle (black bars) for 4 days and not stimulated (Controls, SD only) or stimulated by NMDA, AMPA or kainate (all at 20, 100 and 500 lM) for 20 min (c) or 1 h (a, b) under the simultaneous blockade of voltagedependent Ca2+-channels by 0.1 lM x-conotoxin GVIA, 0.2 lM x-agatoxin IVA and 1 lM nifedipine. Increase of lactate dehydrogenase (LDH)-reIease (a) and reduction of MTT signal (b) were meas-

ured after 1 h agonist exposure and 24 h after washing-out. Increase of Ca2+i levels (c) was measured 20 min after stimulation. Mean + SD (percentage of controls); *p < 0.05; **p < 0.005. Data were pooled from three to four (AMPA LDH exp. 2) independent experiments with n ¼ 12–26 wells. Absolute values of untreated controls for LDH (U/ mL): 61 ± 16 [(a) NMDA], 76 ± 15 [(a) AMPA], 89 ± 19 [(a) kainate]; for MTT signal (OD): 780 ± 156 [(b) NMDA], 778 ± 159 [(b) AMPA], 780 ± 157 [(b) kainate]; for Ca2+i (RFU): 347 ± 19 [(c) NMDA], 298 ± 24 [(c) AMPA] and 292 ± 18 [(c) kainate].

atorvastatin-pre-treated cells compared to untreated controls (Fig. 8a). Moreover, immunoprecipitation of the NR2B subunit protein with PSD-95 did not demonstrate any difference in membrane-bound NR2B subunits between atorvastatin- vs. vehicle-treated cells (Fig. 8b). Taken together, these results suggest that atorvastatin-induced reduction in NMDA receptor function is not a direct result of altered NR2 subunit expression or surface expression.

of a class effect. We demonstrate that the anti-excitotoxic effects relate – at least in part – to modulations in NMDA receptor activity and [Ca2+]i dynamics. The anti-excitotoxic effects of atorvastatin could not be reversed by mevalonate or isoprenoid co-treatment, hence were apparently independent of HMG-CoA reductase inhibition. In contrast, atorvastatin did not protect from OGD and even exaggerated apoptotic cell death induced by several toxins. Most of the pleiotropic effects of statins are mediated by HMG-CoA reductase inhibition via reduction of intracellular levels of mevalonate and other isoprenoid intermediates (Cucchiara and Kasner 2001; Liao 2002), although exceptions have been described (Weitz-Schmidt 2002). Surprisingly, the protective anti-excitotoxic effects of atorvastatin pre-treatment could not be reversed by mevalonate co-treatment, which indicates that it is independent of HMG-CoA reductase inhibition. To avoid the possibility that the mevalonate treatment paradigm was insufficient to increase intra-

Discussion

Here, we demonstrate that the inhibitor of HMG-CoA reductase, atorvastatin, protects cortical neurones from excitotoxic cell death. This effect was evident at high nanomolar to low micromolar concentrations and only when neurones were pre-treated for at least 48 h. Mevastatin, another HMG-CoA reductase inhibitor of the statin class, also protected from glutamate-mediated cell death, indicative



(a)

(b)

Fig. 7 Atorvastatin pre-treatment attenuates NMDA induced wholecell currents. Rat cortical cultures were exposed to atorvastatin for 4 days and NMDA receptor function was investigated via whole-cell patch-clamp electrophysiological recordings. Pre-treatment with atorvastatin (Atorva, grey bar) significantly decreased 100 lM NMDAinduced whole-cell currents in comparison to untreated control cells (Vehicle, white bar). Mean + SD [l(nA)]; *p ¼ 0.05; n ¼ 10).

cellular mevalonate levels during HMG-CoA reductase inhibition we carefully used several concentrations and modes of administration. Moreover, we took advantage of an internal positive control within the same experimental setup. We had previously demonstrated that high concentrations (i.e. > 10 lM) of atorvastatin confer neurotoxicity, and that this effect can be reversed by mevalonate and GGPP but not FPP (Schulz et al. 2004). In fact, identical modes of mevalonate administration reversed the toxic effects of 20 lM atorvastatin, whereas they did not at all attenuate or bypass the protective effects of 1 lM on glutamate-induced cell death in the same neuronal preparation under identical experimental conditions. Similarly, administration of other intermediates of the mevalonate pathway (i.e. GGPP, FPP, squalene) and LDL cholesterol itself, using established modes of administration (Schulz et al. 2004), had also no effect on atorvastatinmediated protection from glutamate-induced cell death. Atorvastatin-treatment conferred a substantial reduction of glutamate-induced [Ca2+]i increase. In fact, [Ca2+]i overload is considered the main perpetrator of excitotoxic cell death, although it does not play a prominent role in apoptotic cell death (Choi 1995). We therefore infer that [Ca2+]i dynamics

Fig. 8 Effect of atorvastatin pre-treatment on NMDA receptor subunit expression. Primary cortical rat cultures were pre-treated with 1 lM atorvastatin (Atorva) or vehicle for 4 days and expression of NMDAR subunits assessed. RT–PCR products of NR2A, 2B, 2C, and 2D mRNA were isolated, first-strand cDNA was synthetized by reverse transcription and amplified using specific primers for rat NR2A, 2B, 2C and 2D and b-actin as housekeeping gene. PCR was performed on four different preparations of atorvastatin-treated or vehicle-treated cells, respectively. PCR products were separated on a 2% agarose gel, stained with ethidium bromide and PCR products from two different cell preparations are shown, displaying no significant difference between the groups (a). Western blots of NR2B subunits were performed in PSD-95-immunoprecipitated plasma membranes of atorvastatin vs. vehicle-treated neuronal cell cultures. Representative blots out of three preparations show no significant difference (b).

are mechanistically involved in the anti-excitotoxic properties of atorvastatin. Moreover, our results suggest that atorvastatin’s effect on [Ca2+]i, LDH release and MTT loss relate to modulation of NMDA receptor activity. In contrast, the role of AMPA or kainate receptors is less clear. The observation that atorvastatin also protected from AMPA and kainate-induced damage may relate to secondary activation of NMDA receptors. In fact, in the presence of an NMDAreceptor antagonist, cellular damage induced by AMPA or kainate was blunted and modest Ca2+ influxes were not attenuated by atorvastatin. Our patch-clamp experiments support the notion that the NMDA receptor function is attenuated by atorvastatin-treatment as NMDA-induced whole cell currents were reduced, which is very similar to the findings obtained with simvastatin pre-treatment by Zacco et al. (2003). It should be noted, however, that decreases in calcium influx were fairly large in atorvastatin pre-treated neurones, whereas there were only subtle changes in whole cell NMDA currents (see also Zacco et al. 2003). Possibly, this could reflect the disparity in receptors activa-



ted: somal receptors only in whole cell current experiments vs. all receptors in the calcium indicator experiments. Neuroprotection by atorvastatin was achieved only after pre-treatment for 2 days or longer. Therefore, it is unlikely that direct pharmacologic interactions of atorvastatin with glutamate receptors account for the protective effects. Rather, indirect effects mediated by changes in gene expression, post-translational mechanisms, or receptor subunit composition may be involved. The question whether atorvastatin’s neuroprotective effect relates to transcriptional vs. translational mechanisms remains to be addressed in future experiments. Presently, it is unclear whether the observed effects are the consequence of NMDA receptor inactivation, internalization, translocation of the receptors at the synapse, or altered subunit aggregation (for review see Bickler and Buck 1998). Particularly, the down-regulation of NR2B expression would be neuroprotective because NMDA receptor assembly with NR2B subunits would result in higher Ca2+ conductivity than assembly with NR2A or C. However, mRNA expression of NMDA receptor subunits NR2A-D was not different between atorvastatin- vs. vehicle-treated cultures. Therefore, we additionally performed immunoprecipitation of NR2B with PSD-95, an NMDA receptor cell surface membrane anchoring protein. Western blot analysis of this PSD-95-immunoprecipitated receptor subunit did not show differences between atorvastatin- vs. vehicle- treated cells. It should be noted that NMDA receptors are heterotetrameric and we did not look at NR1 and NR2A subunits. Moreover, PSD-95-immunoprecipitation experiments give only an assessment of the receptors found at the synapse but not of total receptors. However, synaptic NMDA receptors are distributed in much higher density than extrasynaptic NMDA receptors (i.e. 3/lm2 extrasynaptic vs. 10 000/lm2 synaptic receptors) in cultured hippocampal neurones (Cottrell et al. 2000). Other aspects of NMDA receptor function remain to be investigated. Potential HMG-CoA reductase-independent mechanisms of protection could involve post-translational effects on small G proteins or nuclear factor-jB (NF-jB). For example, Ras GTPase is up-regulated by statins partially independent of HMG-CoA reductase inhibition (Rombouts et al. 2003), and this inhibits NMDA receptor function by negatively regulating Src phosphorylation of subunit NR2A (Thornton et al. 2003). NMDA recepor subunit phosphorylation has been shown to affect its distribution and thus receptor function (Tingley et al. 1997). Another interesting candidate is the transcription factor NF-jB, which is activated by atorvastatin (also independently of HMG-CoA reductase inhibition) in human endothelial cells (Wagner et al. 2000). Importantly, NF-jB decreases glutamate-induced NMDA currents in neurones (Furukawa and Mattson 1998) and is involved in a negative feedbackregulation of the NR1 subunit of the NMDA receptor (Mao et al. 2002). In order to screen for potential changes in gene expression patterns by atorvastatin treatment, we performed

quantitative reverse-transcriptase polymerase chain reaction measurements in a number of candidate genes known to be involved in the regulation of neuronal death and survival (cfos; neuronal nitric oxide synthase nNOS; superoxide dismutase SOD), which did not yield significant differences after four-day atorvastatin pre-treatment (1 lM) compared to vehicle-controls (Endres, Stagliano et al. unpublished data). Statins have beneficial effects on a number of neurological disorders, including stroke, Alzheimer’s disease and multiple sclerosis (Jick et al. 2000; Kwak et al. 2000; White et al. 2000; Wolozin et al. 2000). Although there is good clinical and experimental evidence for systemic protective effects, such as anti-inflammatory, anti-thrombotic, vascular, or immunomodulatory mechanisms, direct neuroprotective effects have not been explored in much detail (Liao 2002; Endres and Laufs 2004). Simons et al. demonstrated that cholesterol depletion by lovastastin in lM concentrations inhibits the generation of b-amyloid in hippocampal neurones (Simons et al. 1998) and that high doses of simvastatin reduce cholesterol intermediates and metabolites in the corticospinal fluid (CSF) in humans (Simons et al. 2002). Similarly, hypercholesterolaemic patients treated with statins have lower CSF cholesterol concentrations (Fassbender et al. 2002). Recently, Zacco and colleagues demonstrated that a number of different statins protected neurones from NDMAinduced excitotoxicity in vitro (Zacco et al. 2003). In fact, the dose-range and pre-treatment regimes that yielded optimal neuroprotective effects were comparable to our study. In contrast to our finding, however, statin pretreatment conferred a reduction of NMDA receptor-mediated currents but did not result in decreased [Ca2+]I, and the antiexcitotoxic effects were significantly attenuated by both mevalonate and cholesterol (Zacco et al. 2003). One explanation for these obvious discrepancies relates to differences in the experimental setup: We employed serum-free near-pure neuronal cultures with glia accounting for less than 10% (Lautenschlager et al. 2000; also demonstrated by GFAPstaining in Fig. 1), whereas Zacco et al. (2003) used a bilayer glia-neurone culture system with serum-containing medium. Differences in the experimental model may have important effects, as astrocytes have ionotropic glutamate receptors as well and are known to exert modulation of neuroprotective effects by various mechanisms including glutamate transport and cycling (Parpura et al. 1994; Vernadakis 1996; Sonnewald et al. 2002). Previous studies demonstrated that statins exert proapoptotic effects at micromolar concentrations in rat cortical neurones and neuroblasts, and that these effects are mediated by HMG-CoA reductase inhibition (Tanaka et al. 2000; Garcia-Roman et al. 2001). We have recently demonstrated that atorvastatin at 20 lM is toxic in newly prepared primary embryonal cortical cultures, inhibits neurite outgrowth in concentration above 8 lM and that both effects depend on the inhibition of the isoprenoid intermediate GGPP (Schulz et al.



2004). Here, we demonstrate that atorvastatin protects from glumate-induced cell death at high nanomolar and low micromolar concentrations. Protection was limited to excitotoxic cell death; atorvastatin augmented apoptotic cell death induced by the DNA damaging compound camptothecin, the pan-kinase inhibitor staurosporine or the cholinotoxin AF64A. The latter corresponds to findings by Tanaka et al. (2000) and Garcia-Roman et al. (2001). Moreover, atorvastatin pre-treatment had no apparent protective or toxic effect in a model of in vitro ischaemia, i.e. OGD. Interestingly, both glutamate and apoptotic pathways are implicated in the propagation of cell death after OGD (Gwag et al. 1995). Therefore, it could be argued that the anti-excitotoxic vs. proapoptotic properties of atorvastatin balance out with no apparent net effect on final outcome after OGD. In conclusion, atorvastatin exerts direct anti-excitotoxic effects in neurones and this is independent of HMG-CoA reductase inhibition. The protective effects correlate to a reduction in NMDA-receptor mediated [Ca2+]i increase and cannot be observed in models of neuronal apoptosis or ischaemic cell death. Further careful in vitro and in vivo studies should address whether statin treatment exerts net benefit when in direct contact with neurones, as the dose and time window for protection described herein is narrow and limited to excitotoxicity. As several statins penetrate the blood–brain barrier (Hamelin and Turgeon 1998), the understanding of direct anti-excitotoxic versus pro-apoptotic effects is of imminent importance for the increasing clinical usage of these drugs. Acknowledgements This study was supported by grants from the Deutsche Forschungsgemeinschaft (En343/6 and En343/7 to ME, GRK to FG, Fi600/7 to KBF), VolkswagenStiftung (ME), BMBF (to ME and UD), and Hermann and Lilly Schilling Foundation (to UD).

Potential conflicts of interest: We have previously received grant support from AstraZeneca. References Amarenco P. (2001) Hypercholesterolemia, lipid-lowering agents, and the risk for brain infarction. Neurology 57, S35–S44. Bickler P. E. and Buck L. T. (1998) Adaptations of vertebrate neurons to hypoxia and anoxia: maintaining critical Ca2+ concentrations. J. Exp. Biol. 201, 1141–1152. Bo¨sel J. and Endres M. (2002) Clinical use of statins. Akt. Neurol. 29, 254–261. Brewer G. J. (1995) Serum-free B27/neurobasal medium supports differentiated growth of neurons from the striatum, substantia nigra, septum, cerebral cortex, cerebellum and dentate gyrus. J. Neurosci. Res. 42, 674–683. Brewer G. J., Torricelli J. R., Evege E. K. and Price P. J. (1993) Optimized survival of hippocampal neurons in B-27supplemented neurobasalTM, a new serum-free medium combination. J. Neurosi. Res. 35, 567–576.

Bruer U., Weih M. K., Isaev N. K., Meisel A., Ruscher K., Bergk A., Trendelenburg G., Wiegand F., Victorov I. V. and Dirnagl U. (1997) Induction of tolerance in rat cortical neurons: hypoxic preconditioning. FEBS Lett. 414, 117–121. Choi D. W. (1995) Calcium: still center-stage in hypoxic-ischemic neuronal death. Trends. Neurosci. 18, 58–60. Cottrell J. R., Dube G. R., Egles C. and Liu G. (2000) Distribution, density, and clustering of functional glutamate receptors before and after synaptogenesis in hippocampal neurons. J. Neurophysiol. 84, 1573–1587. Cucchiara B. and Kasner S. E. (2001) Use of statins in CNS disorders. J. Neurol. Sci. 187, 81–89. Denizot F. and Lang R. (1986) Rapid colorimetric assay for cell growth and survival. Modifications to the tetrazolium dye procedure giving improved sensitivity and reliability. J. Immunol. Meth. 22, 271–277. Endres M. and Laufs U. (2004) Effects of statins on endothelium and signalling mechanisms. Stroke 35 (11:suppl. 1), 2708–2711. Endres M., Laufs U., Huang Z., Nakamura T., Huang P. L., Moskowitz M. A. and Liao J. K. (1998) Stroke protection by 3-hydroxy-3methylglutaryl (HMG)-CoA reductase inhibitors mediated by endothelial nitric oxide synthase. Proc. Natl Acad. Sci. USA 95, 8880–8885. Fan Q. W. YuW., Gong J. S., Zou K., Sawamura N., Senda T., Yanagisawa K. and Michikawa M. (2002) Cholesterol-dependent modulation of dendrite outgrowth and microtubule stability in cultured neurons. J. Neurochem. 80, 178–190. Farinelli S. E. and Greene L. A. (1996) Cell cycle blockers mimosine, ciclopirox and deferoxamine prevent the death of PC12 cells and postmitotic sympathetic neurons after removal of trophic support. J. Neurosci. 16, 1150–1162. Fassbender K., Stroick M., Bertsch T. et al. (2002) Effects of statins on human cerebral cholesterol metabolism and secretion of Alzheimer amyloid peptide. Neurology 59, 1257–1258. Fisher A., Mantione C. R., Abraham D. J. and Hanin I. (1982) Long-term central cholinergic hypofunction in mice by ethylcholine aziridinium ion (AF64A) in vivo. J. Pharmacol. Exp. Ther. 222, 140–145. Freeman T. C., Dixon A. K., Campbell E. A., Tait T. M., Richardson P. J., Rice K. M., Maslen G. L., Metcalfe A. D., Streuli C. H. and Bentley D. R. (1998) Expression mapping of mouse genes. MGI Direct Data Submission, MGI Accession ID: 1199209. Furukawa K. and Mattson M. P. (1998) The transcription factor NFkappaB mediates increases in calcium currents and decreases in NMDA- and AMPA/kainate-induced currents induced by Tumor Necrosis Factor-alpha in hippocampal neurons. J. Neurochem. 70, 1876–1886. Garcia-Roman N., Alvarez A. M., Toro M. J., Montes A. and Lorenzo M. J. (2001) Lovastatin induces apoptosis of spontaneously immortalized rat brain neuroblasts: involvement of nonsterol isoprenoid biosynthesis inhibition. Mol. Cell. Neurosci. 17, 329–341. Greisenegger S., Mu¨llner M., Tentschert S., Lang W. and Lalouschek W. (2004) Effects of pretreatment with statins on the severity of acute ischemic cerebrovascular events. J. Neurol. Sci. 221, 5–10. Grynkiewicz G., Poenie M. and Tsien R. Y. (1985) A new generation of Ca2+ indicators with greatly improved fluorescence properties. J. Biol. Chem. 260, 3440–3450. Gwag B. J., Lobner D., Koh J. Y. and Choi D. W. (1995) Blockade of glutamate receptors unmasks neuronal apoptosis after oxygenglucose deprivation in vitro. Neuroscience 68, 615–619. Hamelin B. A. and Turgeon J. (1998) Hydrophilicity/lipophilicity: relevance for the pharmacology and clinical effects of HMG-CoA reductase inhibitors. Trends. Pharmacol. Sci. 19, 26–37. Harms C., Lautenschlager M., Bergk A., Katchanov J., Kapinya K., Herwig U., Megow D., Dirnagl U., Weber J. R. and Ho¨rtnagl H. (2001) Differential mechanisms of neuroprotection by 17-beta-



estradiol in apoptotic versus necrotic neurodegeneration. J. Neurosci. 21, 2600–2609. Harms C., Bo¨sel J., Lautenschlager M., Harms U., Braun J. S., Ho¨rtnagl H., Dirnagl U., Kwiatkowski D. J., Fink K. and Endres M. (2004) Neuronal gelsolin prevents apoptosis by enhancing actin depolymerization. Mol. Cell. Neurosci. 25, 69–82. Heart Protection Study Collaborative Group (2002) MRC/BHF heart protection study of cholesterol lowering with simvastatin in 20 536 high-risk individuals: a randomized placebo-controlled trial. Lancet 360, 7–22. Hebert P. R., Gaziano J. M. and Hennekens C. H. (1995) An overview of trials of cholesterol lowering and risk of stroke. Arch. Intern. Med. 155, 50–55. Hess D. C., Demchuk A. M., Brass L. M. and Yatsu F. M. (2000) HMGCoA reductase inhibitors (statins). A promising approach to stroke prevention. Neurology 54, 790–796. Jick H., Zornberg G. L., Jick S. S., Seshadri S. and Drachmann D. A. (2000) Statins and the risk of dementia. Lancet 356, 1627–1631. Koh J. Y. and Choi D. W. (1987) Quantitative determination of glutamate mediated cortical neuronal injury in cell culture by lactate dehydrogenase efflux assay. J. Neurosci. Meth. 20, 83–90. Kwak B., Mulhaupt F., Myit S. and Mach F. (2000) Statins as a newly recognized type of immunomodulator. Nat. Med. 6, 1399–1402. Lautenschlager M., Onufriev M. V., Gulyaeva N. V. et al. (2000) Role of nitric oxide in the ethylcholine aziridinium model of delayed apoptotic neurodegeneration in vivo and in vitro. Neuroscience 97, 383–393. Liao J. K. (2002) Beyond lipid lowering: the role of statins in vascular protection. Int. J. Cardiol. 86, 5–18. Mao X., Moerman A. M. and Barger S. W. (2002) Neuronal kappaBbinding factors consist of Sp1-related proteins. J. Biol. Chem. 277, 44 911–44 919. Marti-Fabregas J., Gomis M., Arboix A. et al. (2004) Favorable outcome of ischemic stroke in patients pretreated with statins. Stroke 35, 1117–1123. Meske V., Albert F., Richter D., Schwarze J. and Ohm T. G. (2003) Blockade of HMG-CoA reductase activity causes changes in microtubule-stabilizing protein tau via suppression of geranylgeranylpyrophosphate formation: implications for Alzheimer’s disease. Eur. J. Neurosci. 17, 93–102. Michikawa M. and Yanagisawa K. (1999) Inhibition of cholesterol production but not of nonsterol isoprenoid products induces neuronal cell death. J. Neurochem. 72, 2278–2285. Minta A., Kao J. P. and Tsien R. Y. (1989) Fluorescent indicators for cytosolic calcium based on rhodamine and fluorescein chromophores. J. Biol. Chem. 264, 8171–8178. Parpura V., Basarsky T. A., Liu F., Jeftinija K., Jeftinija S. and Haydon P. G. (1994) Glutamate-mediated astrocyte-neuron signalling. Nature 369, 744–748. Rader D. J. (2003) Therapy to reduce the risk of coronary heart disease. Clin. Cardiol. 26, 2–8. Rombouts K., Kisanga E., Hellemans K., Wielant A., Schuppan D. and Geerts A. (2003) Effects of HMG-CoA reductase inhibitors on proliferation and protein synthesis by rat hepatic stellate cells. J. Hepatol. 38, 546–572. Sans N., Petralia R. S., Wang Y. X., Blahos J. 2nd, Hell J. W. and Wenthold R. J. (2000) A developmental change in NMDA receptor-associated proteins at hippocampal synapses. J. Neurosci. 20, 1260–1271.

Schulz J. G., Bo¨sel J., Stoeckel M., Megow D., Dirnagl U. and Endres M. (2004) HMG-CoA reductase inhibition causes neurite loss by interfering with geranylgeranylpyrophosphate synthesis. J. Neurochem. 89, 24–32. Simons M., Keller P., De Strooper B., Beyreuther K., Dotti C. G. and Simons K. (1998) Cholesterol depletion inhibits the generation of b-amyloid in hippocampal neurons. Proc. Natl Acad. Sci. USA 95, 6460–6464. Simons M., Schwarzler F., Lutjohann D., von Bergmann K., Beyreuther K., Dichgans J., Wormstall H., Hartmann T. and Schulz J. B. (2002) Treatment with simvastatin in normocholesterolemic patients with Alzheimer’s disease: a 26-week randomised, placebocontroplled, double-blind trial. Ann. Neurol. 52, 346–350. Sonnewald U., Qu H. and Aschner M. (2002) Pharmacology and toxicology of astrocyte-neuron glutamate transport and cycling. J. Pharmacol. Exp. Ther. 301, 1–6. Stefanis L., Park D. S., Friedman W. J. and Greene L. A. (1999) Caspase-dependent and -independent death of camptothecin-treated embryonic cortical neurons. J. Neurosci. 19, 6235–6247. Stein A. (2002) Management of dyslipidemia. Am. Heart. J. 144, 543– 550. Tanaka T., Tatsuno I., Uchida D. et al. (2000) Geranylgeranyl-pyrophosphate, an isoprenoid of mevalonate cascade, is a critcal compound for rat primary cultured cortical neurons to protect the cell death induced by 3-hydroxy-3-methylglutaryl-CoA inhibition. J. Neurosci. 20, 2852–2859. Thornton C., Yaka R., Dinh S. and Ron D. (2003) H-Ras modulates N-Methyl-D-aspartate function via inhibition of Src tyrosine kinase activity. J. Biol. Chem. 278, 23823–23829. Tingley W. G., Ehlers M. D., Kameyama K., Doherty C., Ptak J. B., Riley C. T. and Huganir R. L. (1997) Characterization of the protein kinase A and protein kinase C phosphorylation of the N-methyl-D-aspartate receptor NR1 subunit using phosphorylation site-specific antibodies. J. Biol. Chem. 272, 5157–5166. Tsiara S., Elisaf M. and Mikhailidis D. P. (2003) Early vascular benefits of statin therapy. Curr. Med. Res. Opin. 19, 540–556. Vernadakis A. (1996) Glia-neuron intercommunications and synaptic plasticity. Prog. Neurobiol. 49, 185–214. Wagner A. H., Gebauer M., Gu¨ldenzoph B. and Hecker M. (2000) 3-Hydroxy-3-methylglutaryl coenzyme A reductase-independent inhibition of CD40 expression by atorvastatin in human endothelial cells. Thromb. Vasc. Biol. 22, 1784–1789. Weitz-Schmidt G. (2002) Statins as anti-inflammatory agents. Trends Pharmacol. Sci. 23, 482–486. White H. D., Simes R. J., Anderson N. E. et al. for the LIPID Study Group (2000) Pravastatin therapy and the risk of stroke. N. Eng. J. Med. 343, 317–326. Wolozin B., Kellman W., Rousseau P., Celesia G. G. and Siegel G. (2000) Decreased prevalence of Alzheimer disease associated with 3-hydroxy-3-methylglutaryl coenzyme A reductase inhibitors. Arch. Neurol. 57, 1439–1443. Zacco A., Togo J., Spence K., Ellis A., Lloyd D., Furlong S. and Piser T. (2003) 3-Hydroxy-3-methylglutaryl coenzyme A reductase inhibitors protect cortical neurons from excitotoxicity. J. Neurosci. 23, 11 104–11 111. Zhang J., Shi F., Badour K., Deng Y., McGavin M. K. H. and Siminovitch K. A. (2002) WASp verprolin homology, cofilin homology, and acidic region domain-mediated actin polymerisation is required for T cell development. Proc. Natl Acad. Sci. 99, 2240–2245.
