UPF 1069

Mild activation of poly(ADP-ribose) polymerase (PARP) is neuroprotective in rat hippocampal slice models of ischemic tolerance

Elisabetta Gerace, Tania Scartabelli, Laura Formentini, Elisa Landucci, Flavio Moroni, Alberto Chiarugi and Domenico E. Pellegrini-Giampietro
Department of Preclinical and Clinical Pharmacology, University of Florence, Viale G. Pieraccini 6, 50139 Florence, Italy Keywords: DHPG, neuroprotection, NMDA, organotypic slices, poly(ADP-ribose) polymerases, preconditioning

Abstract
Ischemic tolerance is a phenomenon in which exposure to a mild preconditioning stress results in resistance to a subsequent lethal ischemic insult. Here we investigated the role of poly(ADP-ribose) polymerase (PARP) in the development of ischemic tolerance by using organotypic rat hippocampal slices exposed to 30 min oxygen-glucose deprivation (OGD), which leads to selective injury of the CA1 subregion 24 h later. We developed models of pharmacological preconditioning by exposing slices to subtoxic concentrations of either N-methyl-d-aspartate (NMDA) or (S)-3,5-dihydroxyphenylglycine (DHPG) and then, 24 h later, to 30 min OGD. Under these conditions, we observed a significant reduction in OGD-induced CA1 damage. Exposure of slices to the PARP-1 and -2 inhibitors TIQ-A, PJ-34 and UPF 1069 during preconditioning prevented the development of OGD tolerance in a concentration-dependent manner. NMDA and DHPG preconditioning increased the activity of PARP, as detected by immunoblots using antibodies against the poly(ADP-ribose) polymer product, but was not associated with consumption of cellular NAD+ or ATP. Neuroprotection induced by preconditioning was also prevented by the caspase inhibitor Z-VAD-FMK. The modest but significant increase in caspase-3 ⁄ 7 induced by preconditioning, however, was not associated with PARP-1 cleavage, as occurred with staurosporine. Finally, TIQ-A prevented the activation of ERK1 ⁄ 2 and Akt induced by NMDA preconditioning, suggesting that the protective mechanism evoked by PARP requires activation of these prosurvival mediators. Our results suggest that preconditioning with appropriate pharmacological stimuli may promote neuroprotective mechanisms triggered by the sublethal activation of two otherwise deleterious executioners such as PARP and caspase-3 ⁄ 7.

Introduction

Ischemic tolerance is a cellular defense program in which exposure to a subtoxic preconditioning stimulus results in resistance to a subsequent lethal ischemic insult (Gidday, 2006; O’Duffy et al., 2007; Steiger & Hanggi, 2007; Stenzel-Poore et al., 2007; Obrenov- itch, 2008; Dirnagl et al., 2009). The preconditioning stimulus has the ability to regulate transductional and traslational pathways that generate an array of mechanisms resulting in inhibition of pro- grammed cell death or augmentation of cell survival processes. The endogenous mechanisms of increased neuronal resistance induced by preconditioning offer attractive targets for therapeutic strategies, but these processes are still not clearly understood.
Among the possible molecular mediators of ischemic tolerance, the pathways triggered by activation of ionotropic (iGlu) and metabo- tropic glutamate (mGlu) receptors have received particular attention. Subtoxic concentrations of N-methyl-d-aspartate (NMDA) have been used as preconditioning stimuli and shown to produce neuroprotection
through various mechanisms (Grabb & Choi, 1999; Raval et al., 2003; Soriano et al., 2006). Similarly, activation of group I mGlu receptors of the mGlu1 and mGlu5 subtypes with sublethal concentrations of (S)-3,5-dihydroxyphenylglycine (DHPG) has recently been demon- strated to represent an effective preconditioning stimulus that can attenuate the toxic effects of NMDA and oxygen-glucose deprivation (OGD) (Blaabjerg et al., 2003; Werner et al., 2007).
Poly(ADP-ribose) polymerases (PARPs) are a family of enzymes involved in DNA repair that have been proposed to play a key role in post-ischemic neuronal death ´(Szabo & Dawson, 1998; Pieper et al., 1999; Skaper, 2003; Chiarugi, 2005; Moroni, 2008). Upon activation, PARP-1 catalyzes the attachment of chains of poly(ADP-ribose) (PAR) from it substrate NAD+ to a variety of nuclear acceptor proteins. During cerebral ischemia, massive DNA damage induces hyperactivation of PARP-1 and marked depletion of cellular NAD+ and ATP, leading to energy failure and necrotic cell death (Ha &
Snyder, 1999; Moroni et al., 2001). A ‘transcriptional hypothesis’ has also been suggested for PARP-1, in which the formation of PAR

Correspondence: Dr D. E. Pellegrini-Giampietro, as above. E-mail: [email protected]
Received 6 December 2011, revised 8 March 2012, accepted 12 March 2012
regulates the balance between death and survival programs by modulating the gene expression of neuroactive proteins (Ziegler &
Oei, 2001; Chiarugi, 2002; Kraus & Lis, 2003). PARP-2 plays a

ª 2012 The Authors. European Journal of Neuroscience ª 2012 Federation of European Neuroscience Societies and Blackwell Publishing Ltd

complementary role in models of cerebral ischemia compared with PARP-1 (Moroni et al., 2009), but the contribution to cell death of other members of the PARP family appears to be negligible (Schreiber et al., 2006).
Because both NMDA (Eliasson et al., 1997; Lo et al., 1998; Mandir et al., 2000) and DHPG (Meli et al., 2005) can stimulate PARP activity, we sought to determine the role of this enzyme in the induction of ischemic tolerance in organotypic rat hippocampal slices exposed to OGD, an in vitro model of cerebral ischemia in use in our laboratory (Pellegrini-Giampietro et al., 1999b). To this aim, we developed a pharmacological preconditioning protocol using subtoxic concentrations of NMDA and further characterised a DHPG precon- ditioning paradigm we recently described in a previous report (Werner et al., 2007). Our results show that mild activation of PARP-1 following preconditioning with DHPG and NMDA may have a beneficial role in the development of ischemic tolerance.

Materials and methods
Experiments and animal use procedures were in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals (NIH Publications No. 80-23, revised 1996). The experi- mental protocols were approved by the Animal Care Committee of the Department of Pharmacology, University of Florence, in compliance with the European Convention for the Protection of Vertebrate Animals used for Experimental and Other Scientific Purposes (ETS no. 123) and the European Communities Council Directive of 24 November 1986 (86 ⁄ 609 ⁄ EEC). All efforts were made to minimise the number of animals used.

Materials
DHPG and staurosporine were purchased from Ascent Scientific (Weston-Super-Mare, UK). Glutamate, NMDA, N-methyl-N-nitro-N- nitrosoguanidine (MNNG) and propidium iodide (PI) were purchased from Sigma (St Louis, MO, USA). Z-VAD-FMK (carbonbenzoxy- valyl-alanyl-aspartyl-[O-methyl]-fuoromethylketone) was purchased from Promega (Madison, WI, USA). Thieno(2,3-c)isoquinolin-5-one (TIQ-A) was synthesised as described by Pellicciari et al. (2003), whereas N-(6-oxo-5,6-dihydro-phenanthridin-2-yl)-(N,N-dymethila- mino)acetamide hydrochloride (PJ-34) was purchased from Alexis Biochemicals (Vinci, FI, Italy). Tissue culture reagents were obtained from Gibco-BRL (San Giuliano Milanese, MI, Italy) and Sigma.

Preparation of rat organotypic hippocampal slice cultures Organotypic hippocampal slice cultures were prepared as previously
reported (Pellegrini-Giampietro et al., 1999a,b). Briefly, hippocampi were removed from the brains of 8–10-day-old Wistar rats (Harlan, MI, Italy), and transverse slices (420 lm) were prepared using a McIlwain tissue chopper in a sterile environment. Slices were first placed in Hanks’ balanced salt solution (supplemented with 5 mg ⁄ mL glucose and 3.75 lg ⁄ mL amphotericin B) and then transferred onto 30-mm-diameter semiporous membrane inserts (Millicell-CM PIC M 03050; Millipore, Italy), which were placed in six-well tissue culture plates containing 1.2 mL medium per well. The slice culture medium consisted of 50% Eagle’s minimal essential medium, 25% heat- inactivated horse serum, 25% Hanks’ balanced salt solution, 5 mg ⁄ mL glucose, 2 mm l-glutamine and 3.75 lg ⁄ mL amphotericin B. Slices were maintained at 37 ti C in an incubator in an atmosphere of humidified air and 5% CO2 and culture medium was changed twice

a week. Slices were kept in culture for 12–14 days and before experiments all slices were screened for viability by incubating them for 30 min with PI (5 lg ⁄ mL); slices displaying signs of neurode- generation were discarded from the study.

OGD and exposure to glutamate receptor agonists in rat organotypic hippocampal slices
Cultures were exposed to OGD as previously reported in detail (Pellegrini-Giampietro et al., 1999a,b). Briefly, the slices were subjected to OGD by exposing them to a serum- and glucose-free medium saturated with 95% N2 and 5% CO2. Following 30 min of incubation at 37 ti C in an airtight anoxic chamber equipped with an oxygen gas controller (BioSpherix, New York, NY, USA), the cultures were transferred to oxygenated serum-free medium (75% Eagle’s minimal essential medium, 25% Hank’s balanced salt solution; 2 mm l-glutamine and 3.75 lg ⁄ mL amphotericin B) containing 5 mg ⁄ mL glucose and returned to the incubator under normoxic conditions until neuronal injury was evaluated 24 h later.
Exposure to the ionotropic glutamate receptor agonist NMDA (0.1– 100 lm for 15, 30 or 60 min) or to the group I mGlu agonist DHPG (0.1–300 lm for 30 min) was carried out in the incubator using serum-free medium as previously described (Pellegrini-Giampietro et al., 1999a). Hippocampal slices were then cultured for an additional 24 h in a fresh serum-free medium and then evaluated for CA1 pyramidal cell injury. To achieve maximal neuronal injury, hippo- campal slices were exposed for 24 h to 10 mm glutamate in the incubator using serum-free medium.

Assessment of CA1 pyramidal cell injury
Cell injury was assessed using the fluorescent dye PI (5 lg ⁄ mL), a highly polar compound which enters the cell only if the membrane is damaged and becomes fluorescent upon binding to DNA. PI was added to the medium at the end of the 24-h recovery period following OGD or exposure to NMDA or DHPG. Thirty minutes later, fluorescence was viewed using an inverted fluorescence microscope (Olympus IX-50; Solent Scientific, Segensworth, UK) equipped with a xenon-arc lamp, a low-power objective (4 ·) and a rhodamine filter. Images were digitised using a video image obtained by a CCD camera (Diagnostic Instruments Inc., Sterling Heights, MI, USA) controlled by software (InCyt Im1ti ; Intracellular Imaging Inc., Cincinnati, OH, USA) and subsequently analysed using the Image-Pro Plus morphometric analysis software (Media Cybernetics, Silver Spring, MD, USA). To quantify cell death, the CA1 hippocampal subfield was identified and encompassed in a frame using the drawing function in the image software (ImageJ; NIH, Bethesda, MD, USA) and the optical density of PI fluorescence was detected. There was a linear correlation between CA1 PI fluorescence and the number of injured CA1 pyramidal cells as detected by morphological criteria (Pellegrini- Giampietro et al., 1999a).

Western blotting
Cultured slices were washed with cold 0.01 m phosphate-buffered saline, pH 7.4, and four slices per sample were gently transferred and dissolved in a tube containing 1% sodium dodecyl sulfate (SDS). Total protein levels were quantified using the Pierce (Rockford, IL, USA) BCA (bicinchoninic acid) Protein Assay. Lysates (20 lg per lane of protein) were resolved by electrophoresis on a 4–20% SDS-poly- acrylamide gel (Bio-Rad Laboratories, Hercules, CA, USA) and

transferred onto nitrocellulose membranes. Blots were blocked for 1 h at room temperature in 20 mm Tris-buffered saline, pH 7.6, 0.1% Tween 20 (TBS-T) containing 5% non-fat dry milk, and then incubated overnight at 4 tiC with a mouse monoclonal anti-PAR (10H) antibody (Alexis Biochemicals) or with a polyclonal rabbit antybody directed against PARP-1, phospho-ERK1 ⁄ 2 (Thr202 ⁄ Thr204), phospho-Akt (Ser473) or phospho-GSK3-b (Ser9) (all from Cell Signaling Technology, Beverly, MA, USA), all diluted 1 : 1000 in TBS-T containing either 5% non-fat dry milk (anti-PAR) or 5% bovine serum albumin (all other antibodies). The loading control anti- b-actin antibody was monoclonal from Sigma. Immunodetection was performed with secondary antibodies (1 : 2000 anti-mouse or anti- rabbit IgG from donkey; Amersham Biosciences, UK) conjugated to horseradish peroxidase in TBS-T containing 5% non-fat dry milk. Membranes were washed with TBS-T and then reactive bands were detected using chemiluminescence (ECLplus; Euroclone, Padova, Italy). Quantitative analysis was performed using the QuantityOne analysis software (Bio-Rad).

Measurement of NAD+ and ATP endogenous contents in organotypic hippocampal slices
NAD+ contents were quantified by means of an enzymatic cycling procedure as described by Shah et al. (1995), while ATP levels were measured using the ATPliteti Luminescence ATP Detection Assay System (Perkin-Elmer, Waltham, MA, USA).

Caspase-3 and -7 activities in organotypic hippocampal slices Caspase-3 and -7 activities were assessed using the Caspase-Gloti 3 ⁄ 7 Assay kit (Promega). Organotypic hippocampal slices were washed with cold 0.01 m phosphate-buffered saline, pH 7.4, and two slices per sample were gently transferred into an Eppendorf tube containing 110 lL Caspase-Gloti 3 ⁄ 7 reagent and maintained at 22 ti C for 1 h. Then, 100 lL lysate was transferred onto black multiwell plates and luminescence was measured using a TopCount-NXTti (Packard, Warrenville, IL, USA) microplate scintillation and luminescence counter.

Statistical analysis
Data are presented as means ± SEM of n experiments. Statistical significance of differences between PI fluorescence intensities or Western blot optical densities was analysed using one-way anova with a post hoc Tukey’s w test for multiple comparisons. All statistical calculations were performed using Graph-Pad Prism v. 5 for Windows (GraphPad Software, San Diego, CA, USA). A probability value (P) of < 0.05 was considered significant. Results Preconditioning with NMDA and DHPG in rat organotypic hippocampal slices As previously reported (Pellegrini-Giampietro et al., 1999b; Moroni et al., 2001; Werner et al., 2007; Scartabelli et al., 2008), maximal damage [145 ± 5 CA1 PI relative fluorescence units (RFU)] was achieved in this system by exposing organotypic hippocampal slices to 10 mm glutamate for 24 h (Fig. 1A). Exposure to OGD for periods ranging from 10 to 30 min led 24 h later to selective and time- dependent increases in the levels of PI fluorescence in the CA1 region (Fig. 1A). CA1 injury was modest but significant when slices were exposed to 20 min OGD and increased to 66 ± 3 and 83 ± 2% (F4,20 = 130, P < 0.0001, one-way anova) of the fluorescence intensity observed with 10 mm glutamate when slices were exposed, respectively, to 25 and 30 min OGD. Using both morphological and biochemical criteria, we have previously demonstrated that CA1 pyramidal cells in this model undergo a caspase-dependent, apoptotic- like neurodegeneration (Moroni et al., 2001) and that selective and delayed CA1 injury in rat organotypic slices in vitro reproduces the pattern observed following transient global ischemia in mammals in vivo (Pulsinelli et al., 1982). We have previously shown that exposure of organotypic hippocampal slices to iGlu and mGlu receptor agonists such as NMDA, a-amino-3- hydroxy-5-methyl-4-isoxazole (AMPA) or DHPG at relatively large concentrations or for prolonged periods of time results in extensive damage in all pyramidal cell layers (Werner et al., 2007; Scartabelli et al., 2008). Because tolerated doses of glutamate receptor agonists can be neuroprotective, we sought to develop pharmacological models of preconditoning by establishing which was the toxic threshold in our system for NMDA, which has been used as a preconditioning stimulus in various in vitro culture models (Grabb & Choi, 1999; Raval et al., 2003; Soriano et al., 2006), and for DHPG, a non-selective group I mGlu receptor agonist that we have already utilised as a preconditioning (Werner et al., 2007) and postconditioning (Scartabelli et al., 2008) agent. Figure 1B shows that, when added to the incubation medium for 60 min, the lowest concentration of NMDA that induced a significant increase (F5,24 = 55.90, P < 0.0001, one-way anova) in CA1 PI fluorescence 24 h later was 10 lm. When NMDA was incubated for 30 min, the toxic threshold for NMDA rose to 30 lm (F6,28 = 37.75, P < 0.0001, one-way anova), and when incubated for 15 min, toxicity was not observed up to 30 and 100 lm (F6,28 = 45.73, P < 0.0001, one- way anova). On the other hand, 30 min exposure to DHPG produced 24 h later a modest but significant degree of neurotoxicity (F5,12 = 273.1, P < 0.0001, one-way anova) in the CA1 region at both 100 and 300 lm, but not at lower concentrations (Fig. 1C). When we tested the neuroprotective potential of various concen- trations of NMDA that were below (1 and 3 lm for 30 and 60 min, 10 lm for 30 min) or slightly above (10 lm for 60 min) the toxic threshold, we observed that pre-exposure of hippocampal slices to 1 and 3 lm but not to 10 lm NMDA was able to reduce (F8,27 = 16.94, P < 0.0001, one-way anova) the CA1 injury induced 24 h later by a toxic challenge with 30 min OGD (Fig. 2). Because preconditioning with 3 lm NMDA for 60 min produced the highest degree of tolerance, we selected this paradigm for our subsequent experiments. Similarly, when the highest concentration of DHPG that was tolerated without producing a toxic response (i.e. 10 lm) was added to the incubation medium for 30 min, the CA1 injury induced 24 h later by 30 min exposure to OGD was reduced by 40 ± 7% (F4,10 = 26.88, P < 0.0001, one-way anova) (Fig. 3). Previous exposure to a slightly toxic concentration of DHPG (100 lm for 30 min) did not induce tolerance but rather exacerbated the subsequent neuronal death induced by OGD (Fig. 3B). Interestingly, both of these pharmacolog- ical preconditioning protocols (3 lm NMDA for 60 min and 10 lm DHPG for 30 min) were also able to reduce 24 h later the neuronal damage induced in organotypic hippocampal slices by a toxic exposure to 10 lm NMDA or AMPA for 24 h. NMDA toxicity was reduced (F2,9 = 5.101, P = 0.0330, one-way anova) by 37 ± 7% when slices were preconditioned with NMDA and by 35 ± 8% when they were preconditioned with DHPG, whereas AMPA toxicity was reduced (F2,9 = 24.39, P = 0.0002, one-way anova) by 27 ± 2% when slices were preconditioned with NMDA and by 33 ± 5% when they were preconditioned with DHPG. A 100 ** 80 ** 60 40 10 min OGD CA1 20 min OGD * CA3 DG 20 30 min OGD 10 mM glutamate 0 010 20 30 OGD duration (min) B 130 120 110 100 90 80 70 60 50 40 30 20 10 0 C 130 120 110 100 90 80 70 60 50 40 30 20 10 0 10 μM DHPG 300 μM DHPG ** * OGD 30 min 0.1 110 NMDA [μM] 100 OGD 30 min 0.1 1 1 100 30 min DHPG [μM] 1,000 Fig. 1. Neuronal death induced by OGD and glutamate receptor agonists in rat organotypic hippocampal slices. Cultured slices were exposed to OGD, DHPG or NMDA for the indicated period and 24 h later incubated with PI for fluorescence detection of its optical density in the CA1 region. (A) OGD caused increasing levels of CA1 injury when applied for 20–30 min. Data are expressed as percentage of maximal damage produced by 24 h exposure to 10 mm glutamate. Hippocampal slices on the right, photographed under fluorescence optics, display intense PI labeling in the CA1 subregion when exposed to increasing periods of OGD and labeling that extends also to the CA3 subregion and the dentate gyrus (DG) when exposed to 10 mm glutamate for 24 h. (B) Exposure to increasing concentrations of NMDA for 15, 30 or 60 min induced CA1 injury that was modest as compared with 30 min OGD but was significant vs. control at 10 lm (when applied for 60 min), 30 lm (for 15 and 30 min) and 100 lm (for 15 min). Data are expressed as CA1 PI fluorescence 24 h after exposure to OGD or NMDA. Hippocampal slices at top, photographed under fluorescence optics, display modest PI labeling in the CA1 subregion only when exposed to 10 lm NMDA for 60 min. (C) Exposure to increasing concentrations of DHPG for 30 min induced CA1 injury that was modest as compared with 30 min OGD but was significant vs. control at 100–300 lm. Data are expressed as CA1 PI fluorescence 24 h after exposure to OGD or DHPG. Hippocampal slices at top, photographed under fluorescence optics, display modest PI labeling in the CA1 subregion only when exposed to 300 lm DHPG for 30 min. Values represent the mean ± SEM of five (A) or four (B and C) experiments performed in triplicate. *P < 0.05 and **P < 0.01 vs. basal PI fluorescence (anova + Tukey’s w test). PARP is involved in the induction of pharmacological preconditioning To determine the role of PARP in the induction of pharmacological preconditioning, we used TIQ-A, a PARP-1 ⁄ PARP-2 inhibitor characterised in our laboratory (Chiarugi et al., 2003) (Fig. 4). Figure 4B shows that TIQ-A completely prevented the induction of tolerance to OGD induced by NMDA in a concentration-dependent manner (F8,22 = 2.798, P = 0.0266, one-way anova) when present in the incubation medium during exposure of slices to NMDA and the subsequent 24 h recovery period. Another nonselective PARP inhibitor, PJ-34 (Abdelkarim et al., 2001), also prevented the neuroprotective effects of NMDA preconditioning under the same experimental conditions (data not shown), and so did the compound UPF 1069 at concentrations that selectively inhibit PARP-2 (Moroni et al., 2009) (Fig. 4B). TIQ-A was also able to prevent the neuroprotection induced by preconditioning with DHPG A B Control 3 μM NMDA 60 min OGD NMDA preconditioning 3 μM 48 h OGD 30 min 24 h PI-OD PI-OD 125 100 75 50 25 0 NMDA OGD PI-OD CRL OGD 30 min 1 μM NMDA 3 μM NMDA 10 μM NMDA GLU 60 min 24 h 30 min 24 h + OGD + OGD + OGD Fig. 2. NMDA preconditioning in rat organotypic hippocampal slices. (A) Control: slice exposed to a subtoxic concentration of NMDA (3 lm for 60 min) and 48 h later incubated with PI for fluorescence detection of its optical density (PI-OD), displaying background levels of fluorescence. OGD: hippocampal slice exposed to 30 min OGD, displaying intense PI labeling in the CA1 subregion 24 h later. NMDA preconditioning: hippocampal slice exposed to 3 lm NMDA for 60 min 24 h prior to OGD, displaying reduced CA1 PI labeling as compared with OGD alone. (B) Quantitative analysis of CA1 PI fluorescence expressed as percentage of maximal damage produced by 24 h exposure to 10 mm glutamate. Preconditioning with 1–3 lm but not with 10 lm NMDA significantly reduced OGD injury 24 h later. Bars represent the mean ± SEM of five experiments. **P < 0.01 and *P < 0.05 vs. 30 min OGD (anova + Tukey’s w test). A Control 10 μM DHPG 30 min OGD DHPG preconditioning 10 μM DHPG 48 h OGD 30 min OGD 24 h PI-OD PI-OD PI-OD B 0 CRL OGD 10 μM 100 μM GLU 30 min DHPG DHPG 30 min 24 h 30 min 24 h + OGD + OGD Fig. 3. DHPG preconditioning in rat organotypic hippocampal slices. (A) Control: slice exposed to a subtoxic concentration of DHPG (10 lm for 30 min) and 48 h later incubated with PI for fluorescence detection of its optical density (PI-OD), displaying background levels of fluorescence. OGD: hippocampal slice exposed to 30 min OGD, displaying intense PI labeling in the CA1 subregion 24 h later. DHPG preconditioning: hippocampal slice exposed to 10 lm DHPG for 30 min 24 h prior to OGD, displaying reduced CA1 PI labeling as compared with OGD alone. (B) Quantitative analysis of CA1 PI fluorescence expressed as percentage of maximal damage produced by 24 h exposure to 10 mm glutamate. Preconditioning with 10 lm DHPG for 30 min reduced OGD injury whereas 30 min pre- exposure to 100 lm DHPG exacerbated OGD toxicity, as detected 24 h later. Bars represent the mean ± SEM of five experiments. **P < 0.01 and *P < 0.05 vs. 30 min OGD (anova + Tukey’s w test). (F3,8 = 14.04, P = 0.0015, one-way anova) (Fig. 4C) or with a sublethal period (10 min) of OGD (F3,18 = 5.055, P = 0.0103, one- way anova) (Fig. 4D), a protocol we have used and described in a previous report (Werner et al., 2007). To establish which was the critical period for the beneficial involvement of PARP in the development of tolerance to OGD, we added TIQ-A to the incubation medium only during the 60 min exposure to a preconditioning dose of NMDA (Fig. 5, ‘Pre’) or only during the subsequent 24-h recovery period (Fig. 5, ‘Post’). Our results show that the development of tolerance to OGD could be completely abolished (F4,18 = 3.536, P = 0.0269, one-way anova) only when PARP was inhibited during the exposure to the preconditioning stimulus and not at later time points (Fig. 5B). A PARP inhibitors PC OGD PI-OD 24 h 30 min 24 h B 125 100 75 50 25 0 C 125 100 75 50 25 D 125 100 75 50 25 OGD 0.1 1 10 30 100 0.1 1 μM 0 OGD 10 100 μM 0 OGD 10 100 μM 30 min 3 μM NMDA 60 min + OGD 30 min 10 μM DHPG 30 min + OGD 30 min OGD 10 min + OGD Fig. 4. PARP-1 and PARP-2 activity are required for the induction of pharmacological and ischemic preconditioning in rat organotypic hippocampal slices. (A) Schematic diagram showing that slices were incubated with PARP inhibitors 15 min before preconditioning (PC), during PC and during the subsequent 24-h recovery period. NMDA (3 lm for 60 min) and DHPG (10 lm for 30 min) preconditioning were performed as described in Figs 2 and 3, respectively. Ischemic preconditioning was induced by exposing the slices to a brief (10 min) sublethal OGD 24 h prior to the toxic 30-min OGD challenge. CA1 injury was assessed 24 h later by assessing the optical density of PI fluorescence (PI-OD) in this region. (B) Quantitative analysis of CA1 PI fluorescence expressed as percentage of 30-min OGD-induced CA1 toxicity, showing that the tolerance induced by NMDA was significantly reverted by the PARP-1 ⁄ PARP-2 inhibitor TIQ-A and by the PARP-2 inhibitor UPF 1069 in a dose-dependent manner. (C, D) Quantitative analysis showing that the tolerance induced by DHPG (C) and brief OGD (D) was significantly reverted by TIQ-A in a dose-dependent manner. Bars represent the mean ± SEM of four experiments. **P < 0.01 and *P < 0.05 vs. 30 min OGD (anova + Tukey’s w test). A Total TiQ-A Pre TiQ-A Post TiQ-A 3 μM NMDA 60 min 3 μM NMDA 60 min 3 μM NMDA 60 min 24 h 24 h 24 h OGD 30 min OGD 30 min OGD 30 min 24 h 24 h 24 h PI-OD PI-OD PI-OD B 125 100 75 50 25 0 OGD Total Pre Post 30 min 3 μM NMDA 60 min + OGD Fig. 5. Early involvement of PARP in the induction of NMDA preconditioning. (A) Schematic diagram showing that slices were incubated with 10 lm TIQ-A either 15 min before NMDA preconditioning, during PC and during the subsequent 24-h recovery period (Total), or 15 min before and during preconditioning (Pre), or only during the 24-h recovery period (Post). NMDA preconditioning (3 lm for 60 min) was performed as described in Fig. 3. CA1 injury was assessed 24 h after OGD by assessing the optical density of PI fluorescence (PI-OD) in this region. (B) Quantitative analysis of CA1 PI fluorescence expressed as percentage of 30-min OGD-induced CA1 toxicity, showing that the tolerance induced by NMDA was significantly reverted by TIQ-A when present in the incubation medium at early but not at later time points after preconditioning. Bars represent the mean ± SEM of four experiments. **P < 0.01 and *P < 0.05 vs. 30 min OGD (anova + Tukey’s w test). Activation of PARP and NAD+ ⁄ ATP consumption in preconditioned hippocampal slices To directly evaluate the activation of PARP under our experimental conditions, we examined the formation of its product, the PAR polymer, in organotypic hippocampal slices immediately after their exposure to preconditioning concentrations of NMDA and DHPG (Fig. 6). Western blot analysis with an anti-PAR antibody revealed that both preconditioning protocols induced an increase in PAR formation that was significant (F7,24 = 14.95, P < 0.0001, one-way anova) but relatively modest as compared with that evoked by the alkylating agent and PARP-1 activator (Cipriani et al., 2005) MNNG (100 lm for 5 min), and was completely prevented by the presence of 10 lm TIQ-A in the incubation medium. In our system, incubation of hippocampal slices with 100 lm MNNG for 5 min produced 24 h later a neurodegeneration of CA1 pyramidal cells that was prevented by 10 lm TIQ-A (data not shown). Interestingly, the remarkable activation of PARP induced by MNNG was associated with a dramatic decrease of cellular NAD+ (F2,6 = 24.99, P = 0.0012, one-way anova) and ATP (F2,6 = 13.97, P = 0.0055, one-way anova), levels that were evident up to 4 h after exposure to the alkylating agent, whereas NMDA and DHPG did not produce any significant reduction (Fig. 7). The reductions in NAD+ and ATP observed immediately after exposure to MNNG appeared to be PARP-dependent in that they were all prevented by the use of TIQ-A (Table 1). Activation of caspase-3 ⁄ 7 in the induction of pharmacological preconditioning CRL TIQ-A NMDA NMDA + TIQ-A DHPG DHPG + TIQ-A MNNG MNNG + TIQ-A Because caspase-3 activation has been shown to be essential for neuroprotection in preconditioning (Garnier et al., 2003; McLaughlin 250 148 98 72 PAR et al., 2003), we examined its role in our models by using the caspase inhibitor Z-VAD-FMK. Figure 8 shows that neuroprotection induced by both NMDA (F2,9 = 4.895, P = 0.0364, one-way anova) and DHPG (F2,9 = 8.708, P = 0.0079, one-way anova) preconditioning was prevented when 100 lm Z-VAD-FMK was added to the 55 NMDA 36 45 225 200 175 150 125 100 75 50 25 0 CRL TIQ-A NMDA NMDA DHPG DHPG MNNG MNNG β- actin A B 120 100 80 60 40 20 0 120 100 80 60 40 DHPG MNNG 0 1 2 3 4 h * * * + TIQ-A + TIQ-A + TIQ-A 20 Fig. 6. PARP activation in preconditioned hippocampal slices. Slices were 0 exposed to either the NMDA (3 lm for 60 min) or DHPG (10 lm for 30 min) preconditioning protocol and then immediately processed for Western blotting. 0 1 2 3 4 TIQ-A (10 lm) was added to the incubation medium 15 min before and during the preconditioning exposure. Top: representative Western blot using a monoclonal anti-PAR antibody to detect PARP activity in organotypic hippocampal slices. The numbers on the left indicate the position of the molecular mass markers (kDa). b-Actin was used as loading control. Bottom: quantitative analysis of the region denoted by brackets of immunoreactive bands, showing that PARP activity, as detected by the formation of PAR, was modestly increased after exposure to preconditioning concentrations of NMDA and DHPG, in a TIQ-A-dependent manner. Exposure of slices to 100 lm MNNG for 5 min induced a robust increase in PAR formation that was prevented by TIQ-A. Data are expressed as percentage of control PARP activity. Bars represent the mean ± SEM of three experiments. **P < 0.01 and *P < 0.05 vs. control (anova + Tukey’s w test). h Fig. 7. Limited NAD+ ⁄ ATP consumption in preconditioned hippocampal slices. Slices were exposed to 3 lm NMDA for 60 min, 10 lm DHPG for 30 min or 100 lm MNNG for 5 min and then assayed to determine NAD+ (A) or ATP (B) cellular levels immediately after, 2 h or 4 h after the treatment. Results are expressed as percentage of control NAD+ or ATP levels in untreated slices at the same time point. The preconditioning concentrations of NMDA and DHPG induced a slight, non-significant decrease of NAD+ and ATP levels as compared with the dramatic decrease induced by the alkylating agent MNNG. Bars represent the mean ± SEM of three experiments run in quadruplicate. **P < 0.01 and *P < 0.05 vs. control (anova + Tukey’s w test). Table 1. NAD+ ⁄ ATP consumption in preconditioned hippocampal slices is dependent on PARP activity known to induce apoptotic cell death in this system (Meli et al., 2004). Exposure of hippocampal slices to NMDA and DHPG preconditioning produced an increase in caspase-3 ⁄ 7 activity 1 h later that was modest Basal NAD+ ATP Absorbance % Counts per second % 0.32 ± 0.08 100 106 434 ± 25 551 100 as compared with that induced by staurosporine (F5,18 = 9.780, P = 0.0002, one-way anova) and was completely prevented by the presence of 100 lm Z-VAD-FMK in the incubation medium (Fig. 8D). 10 lm TIQ-A 60 min 0.36 ± 0.12 112 95 903 ± 35 256 90 3 lm NMDA 60 min 0.28 ± 0.07 86 80 611 ± 21 062 75 NMDA + TIQ-A 0.30 ± 0.07 93 116 007 ± 31 790 108 10 lm DHPG 30 min 0.24 ± 0.05 74 68 191 ± 15 791 64 DHPG + TIQ-A 5 min 0.37 ± 0.21 115 104 486 ± 30 906 98 100 lm MNNG 0.19 ± 0.04* 58 41 808 ± 13 017* 39 MNNG + TIQ-A 0.26 ± 0.11 80 104 572 ± 50 000 98 Slices were exposed to drugs as indicated and then assayed to determine NAD+ or ATP cellular levels immediately after the treatment. TIQ-A was added to the incubation medium 15 min prior to and during the period of incubation with the other agents. Results are expressed as absorbance for NAD+ and as counts per second for ATP, as well as percentage of basal NAD+ or ATP levels in untreated slices. The modest nonsignficant decrease in NAD+ and ATP levels induced by preconditioning with NMDA and DHPG, as well as the robust decrease in- duced by MNNG, were all prevented by TIQ-A. Values represent the mean ± SEM of at least three experiments run in triplicate. *P < 0.05 vs. basal (anova + Tukey’s w test). incubation medium during exposure of slices to the preconditioning agents and the subsequent 24 h recovery period. We then examined the activity of caspase-3 ⁄ 7 in preconditioned slices and following incubation with 10 lm staurosporine for 24 h, a condition that is Sublethal caspase activation has been shown to result in PARP-1 cleavage that leads to ischemic tolerance (Garnier et al., 2003). Hence, we examined whether caspase-3 ⁄ 7 activation could determine PARP-1 cleavage under our experimental conditions by using an antibody that detects endogenous levels of full-length PARP-1 (116 kDa), as well as the large fragment (89 kDa) of PARP-1 resulting from caspase cleavage (Fig. 9). Western blot analysis revealed that PARP-1 was cleaved only when organotypic slices were exposed to 10 lm staurosporine for 24 h (F9,20 = 6.729, P = 0.0002, one-way anova) or 24 h after exposure to OGD for 30 min, a type of insult that selectively affects CA1 pyramidal cells and not glial cells (Moroni et al., 2001), and not when they were exposed to either of the three preconditioning paradigms used in this study and processed immedi- ately after (Fig. 9A) or 24 h later (Fig. 9B). Downstream mediators of the neuroprotective effects of PARP Activation of the prosurvival ERK1 ⁄ 2 (Shamloo et al., 1999; Jones et al., 2004; Choi et al., 2006) and PI3K-Akt (Yano et al., 2001; Gao et al., 2010) signaling pathways has been implicated in the neuropro- A 100 μM Z-VAD-FMK PC OGD PI-OD 24 h 30 min 24 h B 125 100 75 50 25 0 OGD 30 min Z-VAD-FMK * 3 μM NMDA 60 min + OGD C 125 100 75 50 25 0 OGD 30 min Z-VAD-FMK * 10 μM DHPG 30 min + OGD D 175 150 125 100 75 50 25 0 3 μM NMDA 10 μM DHPG 60 min 30 min Fig. 8. Caspase activity is required for the induction of NMDA and DHPG preconditioning in rat organotypic hippocampal slices. (A) Schematic diagram showing that slices were incubated with the caspase inhibitor Z-VAD-FMK 15 min before preconditioning (PC), during PC and during the subsequent 24-h recovery period. NMDA preconditioning (3 lm for 60 min) and DHPG (10 lm for 30 min) were performed as described in Figs 2 and 3. CA1 injury was assessed 24 h later by assessing the optical density of PI fluorescence (PI-OD) in this region. (B, C) Quantitative analysis of CA1 PI fluorescence expressed as percentage of 30-min OGD- induced CA1 toxicity, showing that the tolerance induced by NMDA (B) and DHPG (C) was significantly reverted by 100 lm Z-VAD-FMK. Bars represent the mean ± SEM of four experiments. *P < 0.05 vs. 30-min OGD (anova + Tukey’s w test). (D) Slices were exposed to either the NMDA or the DHPG preconditioning protocol and caspase-3 ⁄ 7 activity was detected using the Caspase-Gloti3 ⁄ 7 Assay kit 1 h later. Z-VAD-FMK (100 lm) was added to the incubation medium 15 min before and during the preconditioning exposure. Caspase-3 ⁄ 7 activity was increased in preconditioned slices, but to a lesser extent as compared with incubation with 10 lm staurosporine for 24 h. Data are expressed as percentage of control caspase-3 ⁄ 7 activity. Bars represent the mean ± SEM of three experiments. **P < 0.01 and *P < 0.05 vs. control (anova + Tukey’s w test). A B 116 89 45 116 89 45 full length PARP-1 cleaved PARP-1 β-actin 225 200 175 150 125 100 75 50 25 0 ** 225 200 175 150 125 100 75 50 25 0 full length PARP-1 cleaved PARP-1 ** CRL NMDA DHPG OGD 10 min STAU OGD 30 min CRL NMDA DHPG OGD 10 min STAU Fig. 9. PARP-1 is not cleaved following pharmacological and ischemic preconditioning in rat organotypic hippocampal slices. Slices were exposed to either 3 lm NMDA for 60 min, 10 lm DHPG for 30 min or 10 min OGD and then lysed and processed for Western blotting immediately after (A) or 24 h later (B). Top: representative Western blot using a polyclonal rabbit antibody directed against both full-length PARP-1 (116 kDa) and the large fragment (89 kDa) resulting from caspase cleavage. b-Actin was used as loading control. Bottom: quantitative analysis of immunoreactive bands, showing that PARP-1 was not cleaved in our three preconditioning models as compared with incubation with 10 lm staurosporine for 24 h or 24 h after exposure to 30 min OGD. Bars represent the mean ± SEM of three experiments. **P < 0.01 and *P < 0.05 vs. control (anova + Tukey’s w test). tective manifestations in rodent models of ischemic preconditioning in vivo. To identify the pathways involved in mediating the protective effects of PARP in our preconditioning protocols in vitro, we used phospho-specific antibodies to measure the relative levels of the phosphorylated, active forms of ERK1 ⁄ 2 and Akt and of the phosphorylated, inactive form of GSK3-b immediately after exposure to 3 lm NMDA for 60 min. Figures 10A and B show that the NMDA preconditioning protocol induced a significant increase in the phos- phorylation of ERK1 ⁄ 2 (F2,9 = 11.32, P = 0.0035, one-way anova) and Akt (F2,9 = 4.972, P = 0.0351, one-way anova) that was completely prevented by the PARP inhibitor TIQ-A (10 lm). The phosphorylation of GSK3-b was not altered (F2,9 = 0.02156, P = 0.9787, one-way anova) by preconditioning, nor by the use of TIQ-A (Fig. 10C). Discussion A variety of preconditioning stimuli, including brief OGD, hypoxia, oxidative stress and the use of diverse pharmacological agents, have been demonstrated to induce ischemic tolerance in cortical or hippocampal slices in vitro (for reviews see Gidday, 2006; Dirnagl et al., 2009). Along this line, the results of our study show that it is possible to evoke a neuroprotective preconditioning response against a successive 30 min OGD insult by pre-exposing organotypic hippo- campal slices 24 h earlier to concentrations of the iGlu receptor agonist NMDA (3 lm for 60 min) or of the group I mGlu receptor agonist DHPG (10 lm for 30 min) that were below their toxic threshold. Previous work has demonstrated that preconditioning with NMDA and DHPG may activate multiple neuroprotective pathways in organotypic hippocampal slices. For example, the protein kinase C e isozyme was shown to be necessary for the induction of tolerance by ischemic and NMDA-mediated preconditioning (Raval et al., 2003). On the other hand, neuroprotection against NMDA toxicity induced by sublethal doses of DHPG in the same preparation requires activation of both mGlu1 and mGlu5 receptors (Werner et al., 2007) and upregulation of Rab5b, a small GTPase associated with NMDA receptor endocytosis (Arnett et al., 2004). Our results showing that NMDA preconditioning induces phosphorylation, and therefore activation, of ERK1 ⁄ 2 and Akt, but not GSK3-b, are in agreement with previous studies reporting that the ERK1 ⁄ 2 and Akt prosurvival pathways contribute to the development of ischemic tolerance in vivo (Shamloo et al., 1999; Yano et al., 2001; Choi et al., 2006) and that the protective effects of Akt in preconditoned rats are not mediated by its downstream effector GSK3-b (Gao et al., 2010). PARP-1 activation is known to represent a major cause of neuronal death in excitotoxicity, oxidative stress and cerebral ischemia (Cosi et al., 1994; Eliasson et al., 1997; Szabo´ & Dawson, 1998; Pieper et al., 1999) but its role in models of ischemic tolerance has not yet been clarified. In an astrocyte–neuron coculture model, ‘chemical’ ischemic preconditioning leads to mild caspase-3 activation and PARP-1 cleavage, suggesting a cause–effect relationship between inhibition of PARP-1-mediated cell death pathways and increased resistance to a subsequent lethal OGD challenge (Garnier et al., 2003). Caspase-3 activation and PARP-1 cleavage were also observed in a rat model of ischemic preconditioning in vivo (Lee et al., 2008) but, in a more recent report, caspase-3 and PARP-1 appear not to be required for the development of ischemic tolerance in a mouse model of focal ischemia (Faraco et al., 2010). Together, however, these findings are somewhat at odds with the results of a previous study showing that ischemic preconditioning in the rat heart in vivo attenuates activation of caspases and the subsequent cleavage of PARP-1 induced by ischemia ⁄ reperfu- sion (Piot et al., 1999), suggesting that PARP-1 activity may be beneficial under these conditions. Moreover, PARP-1 appears to ameliorate hippocampal cellular recovery following sublethal transient A 44 42 45 pERK1/2 β-actin B 64 45 pAKTSer473 β-actin C 43 45 pGSK3-βSer9 β-actin 10 μM TIQ-A 10 μM TIQ-A 10 μM TIQ-A 150 175 ** 150 125 * 150 125 125 100 100 100 75 75 75 50 50 50 25 0 CRL 3 μM NMDA 60 min 25 0 CRL 3 μM NMDA 60 min 25 0 CRL 3 μM NMDA 60 min Fig. 10. The PARP inhibitor TIQ-A prevents the increase in ERK1 ⁄ 2 and Akt phosphorylation induced by NMDA preconditioning. Slices were exposed to the NMDA (3 lm for 60 min) preconditioning protocol and then immediately processed for Western blotting. TIQ-A (10 lm) was added to the incubation medium 15 min before and during the preconditioning exposure. Top: representative Western blots using polyclonal rabbit antibodies directed against phospho-ERK1 ⁄ 2 (Thr202 ⁄ Thr204) (A), phospho-Akt (Ser473) (B) or phospho-GSK3-b (Ser9) (C). The numbers on the left indicate the estimated molecular mass (kDa). b-Actin was used as loading control. Bottom: quantitative analysis of immunoreactive bands, showing that NMDA preconditioning induced a significant increase in the phosphorylation of ERK1 ⁄ 2 and Akt, but not GSK3-b, that was completely prevented by the PARP inhibitor TIQ-A (10 lm). Data are expressed as percentage of control phosphorylation in control untreated slices. Bars represent the mean ± SEM of three experiments. **P < 0.01 and *P < 0.05 vs. control (anova + Tukey’s w test). global ischemia in rats (Nagayama et al., 2000) and to protect neurons against apoptosis induced by oxidative stress (Diaz-Hernandez et al., 2007), suggesting that depending on the degree and nature of the insult, and especially in conditions that more closely resemble the mild or sublethal progressive stress of endogenous preconditioning and perhaps chronic neurodegeneration, PARP-1 activation may play a neuropro- tective role. In our study, a relatively selective inhibitor of PARP-2 was also able to prevent the development of OGD tolerance induced by NMDA preconditioning. Because PARP-2 appears to be as detrimental as PARP-1 in a model of necrotic post-ischemic cell death and is beneficial in a model of apoptotic OGD in which PARP-1 inhibitors have no effect (Moroni et al., 2009), it appears that PARP-1 and PARP-2 may play similar or differential roles depending on the severity of the insult of the particular experimental model that is used. A crucial factor that appears to determine whether PARP-1 mediates neuronal death or neuroprotection is the depletion of intracellular NAD+. In cultured neurons, nuclear PARP-1 utilises the cytosolic pool of intracellular NAD+ to synthesise PAR polymers (Alano et al., 2007), resulting in energy failure by either consumption of ATP for NAD+ resynthesis (Zhang et al., 1994) or impairment of the NAD+-dependent steps of glycolysis (Ying et al., 2003). Although initial studies had shown that inhibition of PARP activity in cells can be protective or toxic depending on the levels of cellular NAD+ content (Coppola et al., 1995), PARP-1-induced apoptosis-inducing factor release and cell death can be triggered by the direct action of PAR on mitochondria in a manner that appears to be independent of NAD+ depletion (Andrabi et al., 2006; Yu et al., 2006). More recent studies have demonstrated that NAD+ depletion is both necessary and sufficient for the glycolytic inhibition, mitochondrial failure, apoptosis-inducing factor translocation and cell death induced by PARP-1 in astrocytes (Alano et al., 2004) and neurons (Alano, 2009). In accordance with this view, PARP-1 activation is responsible for cell death under conditions associated with drastic depletion of NAD+ and ATP (Endres et al., 1997; Tokime et al., 1998; Diaz-Hernandez et al., 2007), but plays a neuroprotective role in models of sublethal global ischemia (Nagayama et al., 2000), progressive neurodegeneration (Diaz-Hernandez et al., 2007) and pharmacological preconditioning (this study) in which the cellular NAD+ and ATP contents are unchanged. Caspase-3 and -7 are executioners of apoptosis (Riedl & Shi, 2004) that are known to cleave and inactivate PARP-1 (Lazebnik et al., 1994; Le et al., 2002), possibly in an effort to preserve ATP and provide the energy required for the apoptotic active process. Caspase- 3 can also be activated by mechanisms that do not lead to cell death, such as low levels of oxidative stress and excitotoxicity (Bonfoco et al., 1995) or during the development of ischemic tolerance in neurons (Garnier et al., 2003; McLaughlin et al., 2003; Lee et al., 2008). In some of these circumstances, the modest activation of caspase-3 induced by ischemic preconditioning is still able to cleave and inactivate PARP-1 (Garnier et al., 2003; Lee et al., 2008), and this mechanism has been proposed to confer resistance to neurons against subsequent insults, including ischemia, that would otherwise induce PARP-1-mediated cell death. By contrast, our DHPG and NMDA preconditioning protocols induced a similar modest increase in the activity of caspase-3 and -7, as compared with what observed with staurosporine or 30 min OGD, but were unable to produce a significant and detectable cleavage of PARP-1. Hence, it appears that in our system PARP-1 cleavage occurs only following the substantial activation of caspase-3 ⁄ 7 induced by conditions that produce apoptotic cell death, that is, incubation with 10 lm staurosporine for 24 h (Meli et al., 2004) or 30 min OGD (Moroni et al., 2001), whereas following DHPG and NMDA preconditioning caspase-3 ⁄ 7 is only modestly activated and PARP-1 is not cleaved or cleaved in a negligible manner, but rather appears to be necessary for the development of tolerance to subsequent insults in neurons. Our data suggest that appropriate pharmacological or ischemic preconditioning stimuli may promote a neuroprotective mechanism that requires the sublethal activation of both caspase-3 ⁄ 7 and PARPs (PARP-1 and PARP-2). The importance of these signaling pathways in the development of ischemic tolerance is highlighted by the obser- vation that both the caspase inhibitor Z-VAD-FMK and the PARP inhibitors TIQ-A, PJ34 and UPF 1069 prevented the neuroprotective effects induced by NMDA and DHPG preconditioning. Although we cannot rule out that the mild activation of caspase-3 ⁄ 7 may produce a partial but non-detectable cleavage of PARP-1, caspase-3 ⁄ 7 and PARPs are more likely to confer resistance to neurons via independent albeit complementary mechanisms. Caspase-3 has been demonstrated to provide neuroprotection in preconditioning models by mechanisms other than the cleavage of PARP-1, such as the increased synthesis of heat-shock protein 70 (McLaughlin et al., 2003). On the other hand, despite the established role of PARP in post-ischemic cell death, several reports have demonstrated that PARP activation in physiology or in conditions of mild cellular stress may be beneficial via multiple mechanisms that may support the survival of neurons by inducing the expression of anti-apoptotic genes (Bhakar et al., 2002; Chiarugi &
Moskowitz, 2003). In this study, we show that downstream mediators such as ERK1 ⁄ 2 and Akt, which are crucial prosurvival signaling pathways in models of preconditioning (Shamloo et al., 1999; Yano et al., 2001; Choi et al., 2006; Gao et al., 2010), are also regulated by PARP activity. PARP-1 activation has indeed been shown to dramatically amplify ERK-signals (Cohen-Armon et al., 2007) and, under conditions of mild ATP consumption, to stimulate AMP- activated protein kinase (Huang & Shen, 2009; Li et al., 2010), which induces activation of Akt (Leclerc et al., 2010).
In conclusion, our results suggest that under conditions of sublethal cellular stress, such as those evoked by NMDA and DHPG preconditioning in our in vitro models of ischemic tolerance, mild activation of caspase-3 ⁄ 7 and PARP may produce a neuroprotective response. Because PARP inhibitors have entered the stage of clinical testing for the treatment of cancer (Horvath & Szabo, 2007; Liang &
Tan, 2010) and are currently under investigation as potential therapeutic agents in stroke and neurotrauma (Komjati et al., 2005; Moroni & Chiarugi, 2009), our results indicate that caution should be exercised when considering the use of PARP inhibitors in chronic neurodegenerative diseases.

Acknowledgements
This work was supported by the University of Florence, the Italian Ministry of University and Research (MIUR, PRIN 2006 and 2008 projects), the Ente Cassa di Risparmio di Firenze (Florence, Italy) and the Compagnia di San Paolo (Turin, Italy).

Abbreviations
DHPG, (S)-3,5-dihydroxyphenylglycine; iGlu, ionotropic glutamate; mGlu, metabotropic glutamate; MNNG, N-methyl-N-nitro-N-nitrosoguanidine; NMDA, N-methyl-d-aspartate; OGD, oxygen-glucose deprivation; PAR, poly(ADP-ribose); PARP, poly(ADP-ribose) polymerase; PI, propidium iodide; PJ-34, N-(6-oxo-5,6-dihydro-phenanthridin-2-yl)-(N,N-dymethilamino)acetam- ide hydrochloride; TIQ-A, thieno(2,3-c)isoquinolin-5-one; Z-VAD-FMK, car- bonbenzoxy-valyl-alanyl-aspartyl-[O-methyl]-fuoromethylketone.

References
Abdelkarim, G.E., Gertz, K., Harms, C., Katchanov, J., Dirnagl, U., Szabo, C.
& Endres, M. (2001) Protective effects of PJ34, a novel, potent inhibitor of poly(ADP-ribose) polymerase (PARP) in in vitro and in vivo models of stroke. Int. J. Mol. Med., 7, 255–260.
Alano, C.C. (2009) NAD(+) depletion mediates PARP-1 cytotoxicity. J. Neurochem., 108, 39.
Alano, C.C., Ying, W.H. & Swanson, R.A. (2004) Poly(ADP-ribose) polymerase-1-mediated cell death in astrocytes requires NAD(+) depletion and mitochondrial permeability transition. J. Biol. Chem., 279, 18895– 18902.
Alano, C.C., Tran, A., Tao, R., Ying, W.H., Karliner, J.S. & Swanson, R.A. (2007) Differences among cell types in NAD(+) compartmentalization: a comparison of neurons, astrocytes, and cardiac myocytes. J. Neurosci. Res., 85, 3378–3385.
Andrabi, S.A., Kim, N.S., Yu, S.W., Wang, H., Koh, D.W., Sasaki, M., Klaus, J.A., Otsuka, T., Zhang, Z., Koehler, R.C., Hurn, P.D., Poirier, G.G., Dawson, V.L. & Dawson, T.M. (2006) Poly(ADP-ribose) (PAR) polymer is a death signal. Proc. Natl. Acad. Sci. USA, 103, 18308–18313.
Arnett, A.L.H., Bayazitov, I., Blaabjerg, M., Fang, L.W., Zimmer, J. & Baskys, A. (2004) Antisense oligonucleotide against GTPase Rab5b inhibits metabotropic agonist DHPG-induced neuroprotection. Brain Res., 1028, 59–65.
Bhakar, A.L., Tannis, L.L., Zeindler, C., Russo, M.P., Jobin, C., Park, D.S., MacPherson, S. & Barker, P.A. (2002) Constitutive nuclear factor-kappa B activity is required for central neuron survival. J. Neurosci., 22, 8466–8475.
Blaabjerg, M., Fang, L.W., Zimmer, J. & Baskys, A. (2003) Neuroprotection against NMDA excitotoxicity by group I metabotropic glutamate receptors is associated with reduction of NMDA stimulated currents. Exp. Neurol., 183, 573–580.
Bonfoco, E., Krainc, D., Ankarcrona, M., Nicotera, P. & Lipton, S.A. (1995) Apoptosis and necrosis: two distinct events induced, respectively, by mild and intense insults with N-methyl-d-aspartate or nitric oxid ⁄ superoxide in cortical cell cultures. Proc. Natl. Acad. Sci. USA, 92, 7162–7166.
Chiarugi, A. (2002) Poly(ADP-ribose) polymerase: killer or conspirator? The ‘suicide hypothesis’ revisited. Trends Pharmacol. Sci., 23, 122–129.
Chiarugi, A. (2005) Poly(ADP-ribosyl)ation and stroke. Pharmacol. Res., 52, 15–24.
Chiarugi, A. & Moskowitz, M.A. (2003) Poly(ADP-ribose) polymerase-1 activity promotes NF-jB-driven transcription and microglial activation: implication for neurodegenerative disorders. J. Neurochem., 85, 306–317.
Chiarugi, A., Meli, E., Calvani, M., Picca, R., Baronti, R., Camaioni, E., Costantino, G., Marinozzi, M., Pellegrini-Giampietro, D.E., Pellicciari, R. &
Moroni, F. (2003) Novel isoquinolinone-derived inhibitors of poly(ADP- ribose) polymerase-1: pharmacological characterization and neuroprotective effects in an in vitro model of cerebral ischemia. J. Pharmacol. Exp. Ther., 305, 943–949.
Choi, J.S., Kim, H.Y., Cha, J.H. & Lee, M.Y. (2006) Ischemic preconditioning- induced activation of ERK1 ⁄ 2 in the rat hippocampus. Neurosci. Lett., 409, 187–191.
Cipriani, G., Rapizzi, E., Vannacci, A., Rizzuto, R., Moroni, F. & Chiarugi, A. (2005) Nuclear poly(ADP-ribose) polymerase-1 rapidly triggers mitochon- drial dysfunction. J. Biol. Chem., 280, 17227–17234.
Cohen-Armon, M., Visochek, L., Rozensal, D., Kalal, A., Geistrikh, I., Klein, R., Bendetz-Nezer, S., Yao, Z. & Seger, R. (2007) DNA-independent PARP- 1 activation by phosphorylated ERK2 increases EIk1 activity: a link to histone acetylation. Mol. Cell, 25, 297–308.
Coppola, S., Nosseri, C., Maresca, V. & Ghibelli, L. (1995) Different basal NAD levels determine opposite effects of poly(ADP-ribosyl)polymerase inhibitors on H2O2-induced apoptosis. Exp. Cell Res., 221, 462–469.
Cosi, C., Suzuki, H., Milani, D., Facci, L., Menegazzi, M., Vantini, G., Kanai, Y. & Skaper, S.D. (1994) Poly(ADP-ribose) polymerase: early involvement in glutamate-induced neurotoxicity in cultured cerebellar granule cells. J. Neurosci. Res., 39, 38–46.
Diaz-Hernandez, J.I., Moncada, S., Bolanos, J.P. & Almeida, A. (2007) Poly(ADP-ribose) polymerase-1 protects neurons against apoptosis induced by oxidative stress. Cell Death Differ., 14, 1211–1221.
Dirnagl, U., Becker, K. & Meisel, A. (2009) Preconditioning and tolerance against cerebral ischaemia: from experimental strategies to clinical use. Lancet Neurol., 8, 398–412.
Eliasson, M.J.L., Sampei, K., Mandir, A.S., Hurn, P.D., Traystman, R.J., Bao, J., Pieper, A., Wang, Z.-Q., Dawson, T.M., Snyder, S.H. & Dawson, V.L.
(1997)Poly(ADP-ribose) polymerase gene disruption renders mice resistant to cerebral ischemia. Nat. Med., 3, 1089–1095.

Endres, M., Wang, Z.-Q., Namura, S., Waeber, C. & Moskowitz, M.A. (1997) Ischemic brain injury is mediated by the activation of poly(ADP-ribose) polymerase. J. Cereb. Blood Flow Metab., 17, 1143–1151.
Faraco, G., Blasi, F., Min, W., Wang, Z.-Q., Moroni, F. & Chiarugi, A. (2010) Brain ischemic preconditioning does not require PARP-1. Stroke, 41, 181– 183.
Gao, F., Hu, X.Y., Xie, X.J., Xu, Q.Y., Wang, Y.P., Liu, X.B., Xiang, M.X., Sun, Y. & Wang, J.A. (2010) Heat shock protein 90 protects rat mesenchymal stem cells against hypoxia and serum deprivation-induced apoptosis via the PI3K ⁄ Akt and ERK1 ⁄ 2 pathways. J. Zhejiang Univ. Sci., 11, 608–617.
Garnier, P., Ying, W.H. & Swanson, R.A. (2003) Ischemic preconditioning by caspase cleavage of poly(ADP-ribose) polymerase-1. J. Neurosci., 23, 7967– 7973.
Gidday, J.M. (2006) Cerebral preconditioning and ischaemic tolerance. Nat. Rev. Neurosci., 7, 437–448.
Grabb, M.C. & Choi, D.W. (1999) Ischemic tolerance in murine cortical cell culture: critical role for NMDA receptors. J. Neurosci., 19, 1657–1662.
Ha, H.C. & Snyder, S.H. (1999) Poly(ADP-ribose) polymerase is a mediator of necrotic cell death by ATP depletion. Proc. Natl. Acad. Sci. USA, 96, 13978– 13982.
Horvath, E.M. & Szabo, C. (2007) Poly(ADP-ribose) polymerase as a drug target for cardiovascular disease and cancer: an update. Drug News Perspect., 20, 171–181.
Huang, Q. & Shen, H.M. (2009) To die or to live: the dual role of poly(ADP- ribose) polymerase-1 in autophagy and necrosis under oxidative stress and DNA damage. Autophagy, 5, 273–276.
Jones, M.K., Tsugawa, K., Tarnawski, A.S. & Baatar, D. (2004) Dual actions of nitric oxide on angiogenesis: possible roles of PKC, ERK, and AP-1. Biochem. Biophys. Res. Commun., 318, 520–528.
Komjati, K., Besson, V.C. & Szabo, C. (2005) Poly(ADP-ribose) polymerase inhibitors as potential therapeutic agents in stroke and neurotrauma. Curr. Drug Targets CNS Neurol. Disord., 4, 179–194.
Kraus, W.L. & Lis, J.T. (2003) PARP goes transcription. Cell, 113, 677–683. Lazebnik, Y.A., Kaufmann, S.H., Desnoyers, S., Poirier, G.G. & Earnshaw,
W.C. (1994) Cleavage of poly(ADP-ribose) polymerase by a proteinase with properties like ICE. Nature, 371, 346–347.
Le, D.A., Wu, Y.Q., Huang, Z.H., Matsushita, K., Plesnila, N., Augustinack, J.C., Hyman, B.T., Yuan, J.Y., Kuida, K., Flavell, R.A. & Moskowitz, M.A.
(2002)Caspase activation and neuroprotection in caspase-3-deficient mice after in vivo cerebral ischemia and in vitro oxygen glucose deprivation. Proc. Natl. Acad. Sci. USA, 99, 15188–15193.
Leclerc, G.M., Leclerc, G.J., Fu, G. & Barredo, J.C. (2010) AMPK-induced activation of Akt by AICAR is mediated by IGF-1R dependent and independent mechanisms in acute lymphoblastic leukemia. J. Mol. Signal., 5, 15.
Lee, T.H., Yang, J.T., Ko, Y.S., Kato, H., Itoyama, Y. & Kogure, K. (2008) Influence of ischemic preconditioning on levels of nerve growth factor, brain-derived neurotrophic factor and their high-affinity receptors in hippocampus following forebrain ischemia. Brain Res., 1187, 1–11.
Li, J., Benashski, S.E., Siegel, C., Liu, F.D. & McCullough, L.D. (2010) Adenosine monophosphate activated protein kinase inhibition is protective in both sexes after experimental stroke. Neurosci. Lett., 482, 62–65.
Liang, H.Y. & Tan, A.R. (2010) Iniparib, a PARP1 inhibitor for the potential treatment of cancer, including triple-negative breast cancer. IDrugs, 13, 646– 656.
Lo, E.H., Bosque-Hamilton, P. & Meng, W. (1998) Inhibition of poly(ADP- ribose) polymerase: reduction of ischemic injury and attenuation of N-methyl-D-aspartate-induced neurotransmitter dysregulation. Stroke, 29, 830–836.
Mandir, A.S., Poitras, M.F., Berliner, A.R., Herring, W.J., Guastella, D.B., Feldman, A., Poirier, G.G., Wang, Z.Q., Dawson, T.M. & Dawson, V.L. (2000) NMDA but not non-NMDA excitotoxicity is mediated by poly(ADP- ribose) polymerase. J. Neurosci., 20, 8005–8011.
McLaughlin, B., Hartnett, K.A., Erhardt, J.A., Legos, J.J., White, R.F., Barone, F.C. & Aizenman, E. (2003) Caspase 3 activation is essential for neuroprotection in preconditioning. Proc. Natl. Acad. Sci. USA, 100, 715– 720.
Meli, E., Pangallo, M., Picca, R., Baronti, R., Moroni, F. & Pellegrini- Giampietro, D.E. (2004) Differential role of poly(ADP-ribose) polymerase-1 in apoptotic and necrotic neuronal death induced by mild or intense NMDA exposure in vitro. Mol. Cell. Neurosci., 25, 172–180.
Meli, E., Baronti, R., Pangallo, M., Picca, R., Moroni, F. & Pellegrini- Giampietro, D.E. (2005) Group I metabotropic glutamate receptors stimulate the activity of poly(ADP-ribose) polymerase in mammalian mGlu1-

transfected cells and in cortical cell cultures. Neuropharmacology, 49(suppl. 1), 80–88.
Moroni, F. (2008) Poly(ADP-ribose)polymerase 1 (PARP-1) and postischemic brain damage. Curr. Opin. Pharmacol., 8, 96–103.
Moroni, F. & Chiarugi, A. (2009) Post-ischemic brain damage: targeting PARP- 1 within the ischemic neurovascular units as a realistic avenue to stroke treatment. FEBS J., 276, 36–45.
Moroni, F., Meli, E., Peruginelli, F., Chiarugi, A., Cozzi, A., Picca, R., Romagnoli, P., Pellicciari, R. & Pellegrini-Giampietro, D.E. (2001) Poly(ADP-ribose) polymerase inhibitors attenuate necrotic but not apoptotic neuronal death in experimental models of cerebral ischemia. Cell Death Differ., 8, 921–932.
Moroni, F., Formentini, L., Gerace, E., Camaioni, E., Pellegrini-Giampietro, D.E., Chiarugi, A. & Pellicciari, R. (2009) Selective PARP-2 inhibitors increase apoptosis in hippocampal slices but protect cortical cells in models of post-ichaemic brain damage. Br. J. Pharmacol., 157, 854–862.
Nagayama, T., Lan, J., Henshall, D.C., Chen, D.X., O’Horo, C., Simon, R.P. &
Chen, J. (2000) Induction of oxidative DNA damage in the peri-infarct region after permanent focal cerebral ischemia. J. Neurochem., 75, 1716– 1728.
Obrenovitch, T.P. (2008) Molecular physiology of preconditioning-induced brain tolerance to ischemia. Physiol. Rev., 88, 211–247.
O’Duffy, A.E., Bordelon, Y.M. & McLaughlin, B. (2007) Killer proteases and little strokes – how the things that do not kill you make you stronger. J. Cereb. Blood Flow Metab., 27, 655–668.
Pellegrini-Giampietro, D.E., Cozzi, A., Peruginelli, F., Leonardi, P., Meli, E., Pellicciari, R. & Moroni, F. (1999a) 1-Aminoindan-1,5-dicarboxylic acid and (S)-(+)-2-(3¢-carboxybicyclo[1.1.1]pentyl)-glycine, two mGlu1 receptor-pre- ferring antagonists, reduce neuronal death in in vitro and in vivo models of cerebral ischemia. Eur. J. Neurosci., 11, 3637–3647.
Pellegrini-Giampietro, D.E., Peruginelli, F., Meli, E., Cozzi, A., Albani Torregrossa, S., Pellicciari, R. & Moroni, F. (1999b) Protection with metabotropic glutamate 1 receptor antagonists in models of ischemic neuronal death: time-course and mechanisms. Neuropharmacology, 38, 1607–1619.
Pellicciari, R., Camaioni, E., Costantino, G., Marinozzi, M., Macchiarulo, A., Moroni, F. & Natalini, B. (2003) Towards new neuroprotective agents: design and synthesis of 4H-thieno[2,3-c] isoquinolin-one derivatives as potent PARP-1 inhibitors. Farmaco, 58, 851–858.
Pieper, A.A., Verma, A., Zhang, J. & Snyder, S.H. (1999) Poly(ADP-ribose) polymerase, nitric oxide and cell death. Trends Pharmacol. Sci., 20, 171– 181.
Piot, C.A., Martini, J.F., Bui, S.K. & Wolfe, C.L. (1999) Ischemic precondi- tioning attenuates ischemia ⁄ reperfusion-induced activation of caspases and subsequent cleavage of poly(ADP-ribose) polymerase in rat hearts in vivo. Cardiovasc. Res., 44, 536–542.
Pulsinelli, W.A., Brierley, J.B. & Plum, F. (1982) Temporal profile of neuronal damage in a model of transient forebrain ischemia. Ann. Neurol., 11, 491–498.
Raval, A.P., Dave, K.R., Mochly-Rosen, D., Sick, T.J. & Perez-Pinzon, M.A.
(2003)Epsilon PKC is required for the induction of tolerance by ischemic and NMDA-mediated preconditioning in the organotypic hippocampal slice. J. Neurosci., 23, 384–391.
Riedl, S.J. & Shi, Y.G. (2004) Molecular mechanisms of caspase regulation during apoptosis. Nat. Rev. Mol. Cell Biol., 5, 897–907.
Scartabelli, T., Gerace, E., Landucci, E., Moroni, F. & Pellegrini-Giampietro, D.E. (2008) Postconditioning with DHPG reduces CA1 pyramidal cell injury in rat organotypic hippocampal slices exposed to oxygen-glucose depriva- tion: a role for the PI3K-Akt signaling pathway. Neuropharmacology, 55, 509–516.
Schreiber, V., Dantzer, F., Ame, J.C. & de Murcia, G. (2006) Poly(ADP- ribose): novel functions for an old molecule. Nat. Rev. Mol. Cell Biol., 7, 517–528.
Shah, G.M., Poirier, D., Duchaine, C., Brochu, G., Desnoyers, S., Lagueux, J., Verreault, A., Hoflack, J.C., Kirkland, J.B. & Poirier, G.G. (1995) Methods for biochemical study of poly(ADP-ribose) metabolism in vitro and in vivo. Anal. Biochem., 227, 1–13.
Shamloo, M., Rytter, A. & Wieloch, T. (1999) Activation of the extracellular signal-regulated protein kinase cascade in the hippocampal CA1 region in a rat model of global cerebral ischemic preconditioning. Neuroscience, 93, 81– 88.
Skaper, S.D. (2003) Poly(ADP-ribose) polymerase-1 in acute neuronal death and inflammation: a strategy for neuroprotection. Ann. N. Y. Acad. Sci., 993, 217–228.
Soriano, F.X., Papadia, S., Hofmann, F., Hardingham, N.R., Bading, H. &
Hardingham, G.E. (2006) Preconditioning doses of NMDA promote

neuroprotection by enhancing neuronal excitability. J. Neurosci., 26, 4509– 4518.
Steiger, H.J. & Hanggi, D. (2007) Ischaemic preconditioning of the brain, mechanisms and applications. Acta Neurochir., 149, 1–10.
Stenzel-Poore, M.P., Stevens, S.L., King, J.S. & Simon, R.P. (2007) Preconditioning reprograms the response to ischemic injury and primes the emergence of unique endogenous neuroprotective phenotypes: a speculative synthesis. Stroke, 38, 680–685.
Szabo´, C. & Dawson, V.L. (1998) Role of poly(ADP-ribose) synthetase in inflammationandischaemia-reperfusion. TrendsPharmacol.Sci., 19,287–298.
Tokime, T., Nozaki, K., Sugino, T., Kikuchi, H., Hashimoto, N. & Ueda, K.
(1998)Enhanced poly(ADP-ribosyl)ation after focal ischemia in rat brain. J. Cereb. Blood Flow Metab., 18, 991–997.
Werner, C.G., Scartabelli, T., Pancani, T., Landucci, E., Moroni, F. &
Pellegrini-Giampietro, D.E. (2007) Differential role of mGlu1 and mGlu5 receptors in rat hippocampal slice models of ischemic tolerance. Eur. J. Neurosci., 25, 3597–3604.

Yano, S., Morioka, M., Fukunaga, K., Kawano, Y., Hara, T., Kai, Y., Hamada, K., Miyamoto, E. & Ushio, Y. (2001) Activation of Akt ⁄ protein kinase B contributes to induction of ischemic tolerance in the CA1 subfield of gerbil hippocampus. J. Cereb. Blood Flow Metab., 21, 351–360.
Ying, W.H., Garnier, P. & Swanson, R.A. (2003) NAD(+) repletion prevents PARP-1-induced glycolytic blockade and cell death in cultured mouse astrocytes. Biochem. Biophys. Res. Commun., 308, 809–813.
Yu, S.W., Andrabi, S.A., Wang, H., Kim, N.S., Poirier, G.G., Dawson, T.M. &
Dawson, V.L. (2006) Apoptosis-inducing factor mediates poly(ADP-ribose) (PAR) polymer-induced cell death. Proc. Natl. Acad. Sci. USA, 103, 18314– 18319.UPF 1069
Zhang, J., Dawson, V.L., Dawson, T.M. & Snyder, S.H. (1994) Nitric oxide activation of poly(ADP-ribose) synthetase in neurotoxicity. Science, 263, 687–689.
Ziegler, M. & Oei, S.L. (2001) A cellular survival switch: poly(ADP- ribosyl)ation stimulates DNA repair and silences transcription. BioEssays, 23, 543–548.