The Pharmacogenomics Journal (2004) 4, 336–344 & 2004 Nature Publishing Group All rights reserved 1470-269X/04 $30.00 www.nature.com/tpj ORIGINAL ARTICLE

Valproic acid inhibits activity and suppresses excitotoxicity-induced GAPDH nuclear accumulation and apoptotic death in neurons

H Kanai1,3 ABSTRACT 2 Valproic acid (VPA), used to treat bipolar mood disorder and seizures, also A Sawa inhibits histone deacetylase (HDAC). Here, we found that VPA and other 1 R-W Chen HDAC inhibitors, butyrate and , robustly protected mature P Leeds1 cerebellar granule cell cultures from excitotoxicity induced by SYM 2081 ((2S, D-M Chuang1 4R)-4-methylglutamate), an inhibitor of excitatory amino-acid transporters and an agonist of low-affinity kainate receptors. These neuroprotective effects 1Molecular Neurobiology Section, National required protracted treatment and were correlated with enhanced acetylated Institute of Mental Health, National Institutes of histone levels, indicating HDAC inhibition. SYM-induced excitotoxicity was 2 Health, Bethesda, MD, USA; Departments of blocked by MK-801 ((5R,10S)-( þ )-5-methyl-10,11-dihydro-5H-dibenzo[a,d]- Neuroscience, Psychiatry and Behavioural Science, Johns Hopkins University School of cyclohepten-5,10-imine hydrogen maleate), supporting that the toxicity was Medicine, Baltimore, MD, USA largely N-methyl-D-aspartate receptor dependent. SYM excitotoxicity had apoptotic characteristics and was prevented by a caspase inhibitor. SYM- Correspondence: induced apoptosis was associated with a rapid and robust nuclear accumula- D-M Chuang, Molecular Neurobiology tion of glyceraldehyde-3-phosphate dehydrogenase (GAPDH), a housekeep- Section, National Institute of Mental Health, National Institutes of Health, Bldg. 10, Rm. ing gene previously shown to be proapoptotic. VPA pretreatment suppressed 4C-206, 10 Center Dr MSC 1363, Bethesda, SYM 2081-induced GAPDH nuclear accumulation, concurrent with its MD 20892-1363, USA. neuroprotective effects. Chromatin immunoprecipitation (ChIP) revealed that Tel: þ 1 301 496 4915 GAPDH is copresent with acetylated histone H3, including Lys9-acetylated Fax: þ 1 301 480 9290 E-mail: [email protected] histone, and that VPA treatment caused a time-dependent decrease in the levels of nuclear GAPDH with a concomitant increase in acetylated histones in the ChIP complex. Our results strongly suggest that VPA protects neurons from excitotoxicity through inhibition of HDAC activity and that this protective effect may involve suppression of excitotoxicity-induced accumula- tion of GAPDH protein in the nucleus. The Pharmacogenomics Journal (2004) 4, 336–344. doi:10.1038/sj.tpj.6500269 Published online 3 August 2004

Keywords: ; excitotoxiciy; cerebellar granule cells; neuroprotection; GAPDH; histone deacetylase

3Present address: Department of Psychiatry, Shiga University of Medical Science, Seta- Tsukinowa Cho, Otsu, Shiga, 520-2192, Japan. INTRODUCTION Valproic acid (VPA, 2-propylpentanoic acid) has been used in the therapy of Received: 04 March 2004 seizures (for a review, see Johannessen1). It has also long been hypothesized that Revised: 27 May 2004 Accepted: 18 June 2004 the anticonvulsant action of VPA may be mediated through facilitation of the Published online 3 August 2004 neurotransmission of g-aminobutyric acid (GABA) (for a review, see Johannes- Valproate, an HDAC inhibitor, inhibits excitotoxicity H Kanai et al 337

sen1). VPA is also known to be a potent teratogen in humans drogenase (GAPDH), a proapoptotic protein, in the excito- and rodent animal models. Structural and functional studies toxicity and VPA neuroprotection. of VPA and related compounds suggest that the teratogenic and anticonvulsant effects of VPA can be separated and MATERIALS AND METHODS therefore may be mediated by distinct mechanisms. In Materials addition, VPA has clinical efficacy for the treatment of VPA, , TSA, and MK-801 ((5R,10S)-( þ )-5- bipolar mood disorder (also referred to as manic depressive methyl-10,11-dihydro-5H-dibenzo[a,d]cyclohepten-5,10- illness). VPA and lithium are two primary drugs used for the imine hydrogen maleate) were purchased from Sigma/ treatment and prophylaxis of mania and depression, and Aldrich (St Louis, MO, USA). SYM 2081, CNQX (6-cyano-7- VPA was often found to be effective in patients resistant to nitroquinoxaline-2,3-dione), NBQX (1,2,3,4-tetrahydro-6- 2 lithium therapy (for a review, see Post et al ). The mechan- nitro-2,3-dioxo-benzo[f]quinoxaline-7-sulfonamide), and isms underlying the therapeutic effects of mood stabilizers GYKI 52466 (2,3-benzodiazepine) were products of Tocris are unclear. However, the requirement of protracted treat- Cookson Inc. (Ellisville, MO, USA). Ac-DEVD-CHO (Ac-Asp- ment for the beneficial effects to appear suggests that Glu-Val-Asp-CHO). Except where noted, other chemicals regulation of gene expression is a contributor to clinical were from Sigma and culturing materials were from Gibco efficacy. BRL/Invitrogen (Carlsbad, CA, USA). VPA was recently found to be a direct inhibitor of histone 3,4 deacetylase (HDAC) in clinically relevant concentrations. Preparation of CGC and drug treatment HDAC has been catalogued into three classes (for a review, CGC were prepared from 8-day-old Sprague–Dawley rat pups 5 see Gray and Ekstrom ). Class I HDACs consist of HDAC 1, 2, as described.16 Cells were seeded at a density of 2.0 Â 105 3, and 8, whereas Class II HDACs include HDAC 4, 5, 6, 7, 9, cells/cm2 onto 96-well plates precoated with poly-L-lysine and 10. Class III HDACs are structurally unrelated to the and cultured in Dulbecco’s modified Eagle’s medium human Class I and II HDACs, but consist of homologues of (DMEM) containing 10% FCS, 50 g/ml gentamicin, and the yeast Sir2 protein. Class I and II HDACs, but not Class III 25 mM KCl. Cytosine-D-arabinofuranoside (10 mM) was HDACs, are known to be inhibited by short-chain fatty acids added 24 h after plating to arrest the growth of non- (eg VPA and butyrate) and hydroxamic acid (eg trichostatin neuronal cells. For routine neuroprotective studies, VPA, A (trichostatin A (4,6-dimethyl-7-[p-dimethylaminophenyl]- butyrate, or TSA at indicated concentrations were added to 7-oxohepta-2,4-dieno-hydroxamic acid), TSA)). Accumulat- the cultures starting from 7 days in vitro (7-DIV), unless ing evidence supports the notion that hyperacetylation of otherwise indicated. SYM 2081 (SYM) was added to the histone induced by HDAC inhibitors causes antiprolifera- cultures on 13-DIV to trigger neuronal death. Cell viability tion and reversal of neoplastic characteristics through was determined 24 h later by 3-(4,5-dimethylthiazol-2-yl)- 6 induction of genes involved in cell cycle and morphology. 2,5-diphenyl tetrazolium bromide (MTT) assays (see below). Moreover, HDAC inhibitors were found to induce apoptosis characterized by typical morphological changes.7,8 Conver- Measurement of Neurotoxicity sely, HDAC inhibitors were reported to be neuroprotective Cell viability was determined by measuring the activity of via enhancing the acetylation of SP1, a cytoprotective cellular dehydrogenase that cleaves MTT in a colormetric transcription factor, and that this process is necessary for assay as described.16 Active dehydrogenase reduced MTT in the induction of antioxidant proteins.9 Further, an HDAC viable cells to form insoluble formazan, which was then inhibitor, phenylbutyrate, was found to prolong markedly dissolved in dimethylsulfoxide and quantified spectropho- life span in Drosophila, presumably through modulation of tometrically at 540 nm. Data represent the means7SEM of gene expression.10 % of control from three to four separate experiments. Emerging evidence suggests that VPA is neuroprotective as Statistical significance was determined using analysis of exemplified by inhibition of some forms of apoptosis (for a variance (ANOVA) unless noted elsewhere. A P-value of review, see Loy and Tariot11). VPA also protects against cell p0.05 was considered significant. death induced by the stress on endoplasmic reticulum in C6 glioma cells.12 In addition, VPA induces axonal remodeling Morphological and Immunohistochemical Studies and synapsin clustering, presumably through indirect For DNA staining with 4,6-diaminodiphenyl-2-phenylin- inhibition of glycogen synthase kinase-3 (GSK-3).13 The dole (DAPI), the cells were washed with phosphate-buffered present study was undertaken to investigate the neuropro- saline (PBS; pH 7.4) and fixed for 10 min in 4% paraformal- tective effects of VPA and its underlying mechanisms against dehyde in PBS. After fixation, cells were washed with PBS, excitotoxicity induced by SYM 2081 ((2S,4R)-4-methylglu- stained for 20 min with 10 mg/ml DAPI, and washed again tamate), an inhibitor of excitatory amino-acid transporters with PBS. Fluorescence was detected by UV light micro- (EAATs)14 and an agonist of low-affinity kainate receptors,15 scopy. in mature cerebellar granule cells (CGC). We also examined For GAPDH immunostaining, the cells were washed with whether the neuroprotective effects of VPA were mimicked warm PBS and fixed for 15 min in 4% paraformaldehyde in by inhibitors of HDAC and were associated with hyperace- PBS. After fixation, cells were washed with PBS and tylation of histone. In addition, we studied the role of the incubated with a purified polyclonal rabbit anti-GAPDH nuclear accumulation of glyceraldehyde-3-phosphate dehy- antibody at 1 : 5000 dilution (gifted by Dr. Akira Sawa) in PBS

www.nature.com/tpj Valproate, an HDAC inhibitor, inhibits excitotoxicity H Kanai et al 338

containing 0.1% Triton X-100 and 5% normal donkey serum as indicated above for 10 min on ice. Cell lysates were for 2 h at room temperature. Signal visualization was sonicated for a total time of 60 s at 10 s/interval to shear performed using donkey anti-rabbit secondary antibody chromatin. The supernatant fraction (200 ml) of 12 000 g was conjugated to rhodamin (1 : 200 dilution; Chemicon, CA, diluted 10-fold in ChIP dilution buffer containing the same USA). Fluorescence was detected by confocal laser-scanning protease inhibitors. To reduce nonspecific background, 80 ml microscopy together with differential interference contrast of salmon sperm DNA/protein A agarose slurry was added to imaging of the same field. the chromatin solution followed by incubation at 41C for 2 h. The beads were spun down and supernatant fractions Western Blotting were incubated with anti-acetylated histone H3 antibody at Cells cultured in 100 mm2 dishes were detached by scraping 41C for 12 h. The immunocomplex was collected by and sonicated 30 s in sampling buffer containing 10 mM centrifugation after 2 h incubation at 41C with 60 mlof Tris-HCl (pH 7.4), 150 mM NaCl, 1% Nonidet P-40, 1% salmon sperm DNA/protein A agarose slurry. The protein A sodium deoxycholate, 0.1% SDS, 5 mM EDTA, 1 mM phe- agarose/antibody/acetylated histone H3 complex was se- nylmethylsulfonyl fluoride (PMSF), 0.2 U/ml aprotinin, quentially washed with 1 ml of low-salt immunocomplex 50 mg/ml leupeptin, and 10 mg/ml pepstatin A. Proteins were wash buffer, high-salt immunocomplex wash buffer, LiCl separated by 10% SDS-PAGE and then transferred onto a immunocomplex wash buffer and 1 Â TE. The immunocom- PVDF membrane (Millipore, Bedford, MA, USA). Blots were plex was eluted by adding 25 ml SDS sample buffer contain- incubated with primary antibodies against acetylated his- ing 5 mM DTT and boiling with agitation for 5 min. An tone H3, acetylated histone H3 on lysine9 (Upstate aliquot of 20 ml was used for Western blotting as described in Biotechnology, Charlottesville, VA, USA), dimethylated the previous section and probed with anti-acetylated histone H3 on lysine9 (Upstate Biotechnology), or GAPDH histone H3 (Upstate Biotechnology), anti-acetylated histone (Gifted by Dr Akira Sawa), and then probed with a horse- H3 Lys9, and anti-GAPDH antibodies. radish peroxidase-conjugated secondary antibody. Detec- tion was performed by using enhanced chemiluminescence (ECL kit, Amersham Pharmacia Biotech., NJ, USA). RESULTS CGC cultured in DMEM were treated on 13-DIV with Nuclear Preparation 100 mM SYM 2081 (SYM), a purported glutamate transporter The procedures were essentially as described previously.16 blocker and glutamate receptor agonist, which induced CGC cultured in 100-mm dishes were gently rinsed twice more than 60% loss of cell viability after 24 h exposure with ice-cold PBS containing 1 mM PMSF, 1 mg/ml aprotinin, (Figure 1a). This SYM-induced cell death was completely and 1 mg/ml pepstatin A, and resuspended in 500 ml lysis blocked by the presence of MK-801 (an N-methyl-D- buffer (250 mM sucrose, 20 mM Tris-HCl, 10 mM KCl, 2 mM aspartate (NMDA) receptor antagonist), but was only

EDTA, 1.5 mM MgCl2, 1 mM dithiothreitol (DTT), 0.2 mg/ml slightly blocked by the copresence of NBQX (a non-NMDA leupeptin, 1 mg/ml pepstatin A, and 0.1 mM PMSF). After receptor antagonist) and GYKI 52466 (a kainate/AMPA (a- scraping into a Dounce homogenizer (Wheaton, NJ, USA) amino-3-hydroxy-5-methylisoxazole-4-propionic acid) re- and incubating for 30 min on ice, the cells were vigorously ceptor antagonist). On the other hand, Ac-DEVD-CHO (a homogenized by 60 strokes with a tight-fitting pestle. The preferential caspase-3 inhibitor) robustly suppressed SYM nuclei were isolated by discontinuous gradient centrifuga- toxicity, suggesting that this form of toxicity occurs through tion (800 g, 20 min). The supernatant was discarded, and the apoptotic mechanisms. DNA staining with DAPI revealed nuclear pellet was washed with lysis buffer and centrifuged that SYM treatment caused chromatin condensation and again. The nuclear pellet was resuspended by sonication in nuclear fragmentation in CGC compared with the untreated SDS-containing sampling buffer for Western blotting of control (Figure 1b). GAPDH protein. All steps were performed at 41C. Pretreatment of CGC with VPA for 6 days starting from 7- DIV induced a concentration-dependent suppression of cell Chromatin Immunoprecipitation (ChIP) Assay death induced by exposure to SYM for 24 h (Figure 2a). The To study interactions between DNA-binding protein and neuroprotective effect of VPA was significant at 0.2 mM and proteins of interest, ChIP assay in combination with enhanced progressively by increasing the drug concentra- Western blotting was performed essentially as described tion with a near-complete protection at 1.6 mM. Pretreat- previously17 using a ChIP assay Kit (Upstate Biotechnology) ment of CGC with VPA alone did not affect cell viability in with some modifications. CGC cultured on 10-cm dishes the concentration range of 0.2–1.6 mM. To examine were treated with 1% formaldehyde for 10 min at 371Cto whether VPA-induced neuroprotection may be related to crosslink histones to DNA. The medium was aspirated and its ability to inhibit HDAC, we tested the effects of two other crosslinked cells were washed twice with ice-cold PBS HDAC inhibitors, butyrate and TSA. Similar to the effects of containing protease inhibitors (1 mM PMSF, 1 mg/ml aproti- VPA, butyrate and TSA pretreatment for 6 days protected nin, and 1 mg/ml pepstatin A). Cells were scraped from the CGC from excitotoxicity in a concentration-dependent plates in PBS and centrifuged for 5 min at 700 g at 41C. The manner (Figure 2b and c). The effective concentration cell pellets were resuspended in 200 ml of SDS Lysis Buffer ranges were 0.125–1.0 mM for butyrate and 25–100 nM for (Upstate Biotechnology) containing the protease inhibitors TSA. We also examined the effects of and

The Pharmacogenomics Journal Valproate, an HDAC inhibitor, inhibits excitotoxicity H Kanai et al 339

Figure 1 SYM 2081-induced neurotoxicity of CGC. (a) Pharmaco- logical properties of SYM-induced toxicity. CGC on 13-DIV were l treated with vehicle or 100 M SYM. When used, indicated drugs Figure 2 VPA and two other HDAC inhibitors, butyrate and TSA, were added 5 min prior to SYM, and cell viability was determined suppress SYM-induced excitotoxicity in mature CGC. Cells on 7-DIV 24 h after SYM addition by MTT assay. The concentrations of SYM, were pretreated with indicated concentrations of drugs prior to Ac-DEVD-CHO, MK-801, NBQX, and GYKI were 100, 200, 10, 100, exposure to 100 lM SYM on 13-DIV for 24 h. Data are means7SEM and 50 lM, respectively. Data are means7SEM of % of vehicle wwww of % of cell viability of vehicle control. *Po0.05, **Po0.01, control from three independent cultures. Po0.001 compared ***Po0.001 compared with the group of SYM alone. with the basal control; **Po0.01, ***Po0.001 compared with the group of SYM alone. (b) DAPI staining of nuclei in untreated and SYM-treated cells. CGC on 13-DIV were treated with vehicle or 100 lM SYM for 24 h, and DAPI staining for DNA in the nuclei was performed as described in the Materials and methods. Arrows show citotoxicity in CGC (data not shown). To ascertain that nuclei with condensed chromatin and fragmentation, as indicated HDAC was inhibited in CGC treated with TSA or VPA, cells by bright fluorescence, in SYM-treated cultures, while arrowheads were harvested on DIV-13 after treatment with TSA (100 nM) l show intact nuclei. Scale bar: 10 m. or VPA (1.6 mM) for 2–6 days. The results showed that both TSA and VPA markedly enhanced the levels of acetylated histone H3 on Lys9 (Figure 4a). These effects were time 2-methyl-2-pentenoic acid, two VPA analogs inactive in dependent with a more robust effect for VPA than for TSA. blocking HDAC activity.3 Neither drug in the concentration In contrast, levels of methylated histone H3, total histone range of 0.2–1.6 mM was found to be effective in protecting H3 protein, and total GAPDH protein were unaffected by CGC from SYM-induced excitotoxicity following a 6-day either TSA or VPA. pretreatment period starting from 7-DIV (data not shown). Adding VPA on 7-DIV and harvesting cells on 9-, 11-, and VPA (1.6 mM)-induced neuroprotection against SYM toxi- 13-DIV also resulted in a time-dependent increase in the city required protracted treatment (Figure 3a). Pretreatment levels of acetylated histone H3 compared with respective for one day was ineffective, while 3–7 days of pretreatment controls (Figure 4b). In contrast, basal levels of nuclear produced a significant and progressive increase in neuro- GAPDH protein were not changed by VPA compared with protection. Similar time-dependent neuroprotective effects respective time points of the control. Previous studies have were observed when cells were pretreated with butyrate shown that nuclear translocation of GAPDH is involved in (1 mM) or TSA (100 nM) (Figure 3b and c). VPA, butyrate, apoptosis induced by a variety of insults in neurons and and TSA also time and dose dependently protected against non-neuronal cells.16,18–20 However, the role of GAPDH in glutamate (100 mM)-induced, NMDA receptor-mediated ex- excitotoxicity-related cell death has not been reported.

www.nature.com/tpj Valproate, an HDAC inhibitor, inhibits excitotoxicity H Kanai et al 340

Figure 4 VPA and TSA selectively increase the levels of acetylated histone H3 protein. (a) CGC were treated with TSA (100 nM) or VPA (1.6 mM) for 2, 4, or 6 days, and all cultures were harvested on 13- DIV for Western blotting of acetylated histone H3 on Lys9, dimethylated histone H3 on Lys9, total histone H3 protein, and Figure 3 VPA, butyrate, and TSA protect CGC from SYM excito- total GAPDH protein. (b) Cells were harvested at indicated DIV for toxicity in a time-dependent manner. VPA (1.6 mM), butyrate Western blotting of GAPDH and acetylated histone H3 protein in (1 mM), or TSA (100 nM) was added to CGC cultures at different the nuclear fraction. When used, VPA (1.6 mM) was added to the DIV to achieve different days of pre-exposure time. All cells were cultures on 7-DIV and cells were harvested on 9-, 11-, or 13-DIV. treated with SYM (100 lM) on 13-DIV and cell viability was measured 24 h later. Data are means7SEM of % of cell viability of vehicle control. **Po0.01, ***Po0.001 compared with the group of SYM alone. nuclear GAPDH interacts with DNA and a protein complex containing acetylated histone, we performed ChIP assays using 13-DIV CGC fixed with formaldehyde. The ChIP Therefore, we first performed immunohistochemical stain- complexes isolated with a polyclonal antibody preparation ing with a GAPDH-specific antibody to show that GAPDH that recognized acetylated histone H3 of approximately was accumulated in the nuclei of CGC stimulated with SYM 17 kDa were then subjected to SDS-PAGE and immuno- for 8 h and that this process was associated with cell blotted with anti-GAPDH, anti-total-acetylated histone H3, shrinkage (Figure 5A). VPA pretreatment for 6 days blocked and anti-Lys9-acetylated histone H3 antibodies. The results SYM-induced GAPDH nuclear accumulation with little or no showed that GAPDH was present in the ChIP complex effect on basal levels of nuclear GAPDH. We then measured containing acetylated histone H3, including Lys9-acetylated the levels of GAPDH protein by Western blotting in nuclear histone H3 (Figure 6). Moreover, VPA treatment rapidly and preparations from CGC treated with SYM with or without robustly decreased levels of GAPDH in the ChIP complex VPA pretreatment. SYM induced a marked increase in with a concomitant increase in the levels of acetylated nuclear GAPDH levels 8 h after stimulation (Figure 5B). This histone H3. GAPDH nuclear accumulation was associated with a decrease in GAPDH protein in the cytosolic (10 000 g supernatant) fraction, thus suggesting nuclear translocation DISCUSSION (data not shown). VPA pretreatment blocked SYM-induced The present study demonstrates that long-term VPA treat- increase in nuclear GAPDH protein. To determine whether ment has neuroprotective effects against SYM-induced

The Pharmacogenomics Journal Valproate, an HDAC inhibitor, inhibits excitotoxicity H Kanai et al 341

Figure 6 GAPDH is copresent with acetylated histone and decreased by VPA treatment in the ChIP complex. CGC cultured on 10-cm dishes were treated with 1.6 mM VPA on 7, 9, or 11-DIV and treated with 1% formaldehyde on 13-DIV to crosslink histones to DNA. The ChIP assay was then performed using anti-acetylated histone H3 antibody to obtain the protein A agarose/antibody/ acetylated histone H3 complex as described in the Materials and Methods. Western blotting was then performed to probe for total acetylated histone H3 (a), acetylated histone H3 Lys9 (b), and GAPDH (c) in the ChIP complex. The upper panels show the immunoblots and the lower panels show their quantified results expressed as the means7SEM of % of untreated (vehicle) control from three independent experiments. Statistically significant posi- tive correlation was observed in protein levels of total acetylated Figure 5 VPA delays the accumulation of GAPDH in the nuclei of histone H3 and acetylated histone H3 on lysine 9 (Po0.01), while a CGC following SYM treatment. (A) Immunohistochemistry of significant negative correlation was observed in GAPDH protein GAPDH. Cells on 13-DIV were either untreated (a, d, and g) or levels (Po0.01). treated with SYM (100 lM) for 8 h without (b, e, and h) or with (c, f, and i) VPA (1.6 mM) pretreatment for 6 days. Shown are differential interference (DIC) images (a, b, and c), immunohistochemistry of GAPDH (d, e, and f) and their respective superimposed images (g, H3 protein induced by these two drugs in CGC. These h, and i). Note that SYM-induced excitotoxicity was associated with results strongly suggest that VPA-induced neuroprotection cell shrinkage and nuclear accumulation of GAPDH. These events involves inhibition of HDAC activity. HDACs include three were prevented by VPA pretreatment. Arrows in (e): cells with major classes and more than 10 different subtypes. The nuclear GAPDH accumulation. Scale bar, 10 lm. (b) Cells on 13-DIV were treated with SYM (100 lM) for 8 h without or with pretreat- subtypes of HDACs involved in the neuroprotective effects ment of VPA (1.6 mM) for 6 days. Nuclear GAPDH protein levels of VPA, butyrate, and TSA shown in this study require future were then determined by Western blotting. Upper panel: Western investigation. SYM-induced excitotoxicity of CGC showed blots; lower panel: quantified results expressed as the means7SEM cell shrinkage, chromatin condensation, and nuclear frag- of % of vehicle control at 0 time. *Po0.05 compared with VPA at 0 mentation, and was blocked by a caspase-3-prefering time. inhibitor. These observations suggest that the death of CGC is due, at least in part, to apoptosis. SYM-induced excitotoxicity in CGC was completely blocked by an NMDA receptor antagonist, MK-801, reminiscent with the finding excitotoxicity in CGC. This VPA-induced neuroprotection that glutamate or domoic acid-induced Ca2 þ influx and occurred at therapeutic concentrations used for the treat- toxicity in CGC is preferentially blocked by NMDA receptor ment of bipolar mood disorder and seizures and was blockers.21,22 Our results suggest that SYM-induced apopto- mimicked by other HDAC inhibitors such as butyrate and sis is due to the accumulation of glutamate resulting from TSA. The inhibition of HDAC by VPA and TSA was verified inhibition of EAATs14 or enhanced glutamate release due to by the time-dependent accumulations of acetylated histone the activation of kainate receptors,15,22 thus in turn

www.nature.com/tpj Valproate, an HDAC inhibitor, inhibits excitotoxicity H Kanai et al 342

activating NMDA receptors to trigger cell death. However, complex than in the nuclear preparation of unstimulated the lack of kainite-induced death of CGC speaks against the CGC (Figure 6 vs Figure 5b) suggests that only a small latter possibility.23 The role of EAATs in SYM-induced fraction of nuclear GAPDH is present with histone proteins apoptosis is further supported by the observations that in the ChIP complex. Interestingly, it was recently reported EAAT inhibitors exacerbate endogenous glutamate-induced that the histone H2B promoter interacts with a GAPDH- excitotoxicity through NMDA receptor activation,24 and containing transcriptional coactivator complex,31 indicat- that SYM is one of the most potent inhibitors of EAATs,14 ing that nuclear GAPDH may have an important function in which are expressed in CGC neurons.25,26 gene transcription. GAPDH also selectively interacts with Our pioneering studies show that GAPDH overexpression the polyglutamate region of huntingtin,32 a Huntington’s and nuclear translocation is involved in apoptosis of several disease protein that accumulates in the nuclei as the cell types subjected to various insults.16,18–20,27 This role of pathophysiology progresses. Thus, it seems possible that GAPDH in apoptosis is supported by these studies, which GAPDH may also function as a chaperone in mediating demonstrate that GAPDH antisense oligonucleotides sup- nuclear translocation of proapoptotic proteins. In a trans- press GAPDH overexpression and nuclear translocation and genic mouse model of Huntington’s disease, we found that protect cells from apoptotic insults. In addition, CGP 3466, the expression and nuclear levels of GAPDH were markedly a deprenyl derivative, which binds GAPDH, also blocks enhanced in several brain areas affected by this disease.33 Of growth factor deprivation-induced GAPDH nuclear translo- further interest, HDAC inhibitors have been reported to cation and apoptosis.28 Our present results provide the first reduce polyglutamine-induced toxicity and slow neurode- evidence that GAPDH nuclear translocation is associated generation in a Drosophila model of Huntington’s disease.34 with excitotoxicity and that VPA neuroprotection is accom- Inhibition of HDAC produces hyperacetylation of histone panied by suppression of SYM-induced nuclear transloca- tails, causing chromatin relaxation due to disruption of tion. Precise mechanisms underlying GAPDH nuclear histone–DNA and histone–histone interactions (for a review, translocation-mediated apoptosis are unclear, but could be see Zhang et al35). As a consequence, gene expression can be related to a decrease in GAPDH-associated functions such as activated and sometimes downregulated.10 VPA-induced glycolytic and uracil DNA glycosylase activities following hyperacetylation of histone proteins has also been shown nuclear translocation,27 or due to interaction with nuclear in mouse brain studies in vivo.36,37 The requirement of proteins and alterations in nuclear function in a gain-of- protracted pretreatment for HDAC inhibitors to elicit toxicity manner.20,29 One such nuclear protein that can neuroprotective effects in our studies suggests an involve- possibly interact with GAPDH is NcoR (A Sawa et al. ment of changes in gene expression. VPA has been reported unpublished results). to enhance the expression of cytoprotective proteins such as Our results using ChIP assays, which detect interactions Bcl-2,38 GRP (glucose-regulated protein),39 heat-shock pro- among chromatin-associated proteins, revealed that nuclear tein 70,40,41 and brain-derived neurotrophic factor.42 It GAPDH was copresent with acetylated histone H3 and that remains to be investigated whether VPA-induced upregula- long-term VPA caused a decrease in GAPDH in the ChIP tion of these proteins is triggered by HDAC inhibition and complex with a concurrent increase in acetylated histone. It whether induction of any of these proteins contributes to is currently unknown whether GAPDH directly binds VPA neuroprotection against SYM neurotoxicity. histone H3 or GAPDH and histone H3 are located in close vicinity on chromatin and therefore are crosslinked to a ACKNOWLEDGEMENTS common fragment of DNA and co-immunoprecipitated. In We thank Lori Christ in our Section of NIMH, NIH, as well as Yang- any event, our results robustly demonstrate that there is an Ja Lee-Wickner and Mariusz Karbowski of NINDS for their technical inverse relationship between the levels of histone H3 assistance and advice. acetylation and GAPDH nuclear accumulation in VPA- DUALITY OF INTEREST treated neurons, raising the possibility that enhanced None declared. histone acetylation may weaken the association of GAPDH with chromatin. It is conceivable that the free nuclear GAPDH not bound to chromatin will be exported to the ABBREVIATIONS cytoplasm. This is based on a recent report that GAPDH nuclear translocation is reversible and a novel exportin1- Ac-DEVD-CHO Ac-Asp-Glu-Val-Asp-CHO AMPA a-amino-3-hydroxy-5-methylisoxazole-4-propionic acid dependent nuclear export signal (NES) comprising 13 amino CGC cerebellar granule cells acids of the C-terminal domain of GAPDH has been CNQX 6-cyano-7-nitroquinoxaline-2,3-dione identified.30 Truncation or mutation of this NES abrogates DAPI 4,6-diaminodiphenyl-2-phenylindole exportin1 binding and causes nuclear accumulation of DIV day in vitro DMEM Dulbecco’s modified Eagle’s medium GAPDH. It remains to be investigated whether the acetyla- DTT dithiothreitol tion of other proteins such as the cell survival transcription GABA c-aminobutyric acid factor SP1 is enhanced by VPA under our experimental GAPDH glyceraldehyde-3-phosphate dehydrogenase conditions and, if so, whether their acetylation plays a role GSK glycogen synthase kinase GYKI 52466 2,3-benzodiazepine in the neuroprotection against excitotoxicity. The much greater decrease in GAPDH elicited by VPA in the ChIP

The Pharmacogenomics Journal Valproate, an HDAC inhibitor, inhibits excitotoxicity H Kanai et al 343

17 Wells J, Farnham PJ. Characterizing transcription factor binding sites HDAC histone deacetylase using formaldehyde crosslinking and immunoprecipitation. Methods MK-801 (5R,10S)-( þ )-5-methyl-10,11-dihydro-5H-diben- 2002; 26: 48–56. zo[a,d]cyclohepten-5,10-imine hydrogen maleate 18 Ishitani R, Chuang D-M. Glyceraldehyde-3-phosphate dehydrogenase MTT 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium antisense oligodeoxynucleotides protect against cytosine arabinonu- bromide cleoside-induced apoptosis in cultured cerebellar neurons. Proc Natl NBQX 1,2,3,4-tetrahydro-6-nitro-2,3-dioxo-benzo[f ]quinoxa- Acad Sci USA 1996; 93: 9937–9941. line-7-sulfonamide 19 Saunders PA, Chalecka-Franaszek E, Chuang D-M. Subcellular distribu- NMDA N-methyl-D-aspartate tion of glyceraldehyde-3-phosphate dehydrogenase in cerebellar PMSF phenylmethylsulfonyl fluoride granule cells undergoing cytosine arabinoside-induced apoptosis. J TSA trichostatin A (4,6-dimethyl-7-[p-dimethylaminophe- Neurochem 1997; 69: 1820–1828. nyl]-7-oxohepta-2,4-dieno-hydroxamic acid) 20 Sawa A, Khan AA, Hester LD, Snyder SH. Glyceraldehyde-3-phosphate VPA valproic acid dehydrogenase: nuclear translocation participates in neuronal and nonneuronal cell death. Proc Natl Acad Sci USA 1997; 94: 11669–11674. 21 Nonaka S, Hough C, Chuang D-M. Chronic lithium treatment robustly protects CNS neurons against excitotoxicity by inhibiting NMDA receptor-mediated calcium influx. Proc Natl Acad Sci USA 1998; 95: 2642–2647. 22 Berman FW, Lepage KT, Murray TF. Domoic acid neurotoxicity in REFERENCES cultured cerebellar granule neurons is controlled preferentially by the 1 Johannessen CU. Mechanisms of action of valproate: a commentatory. NMDA receptor Ca(2+) influx pathway. Brain Res 2002; 924: 20–29. Neurochem Int 2000; 37: 103–110. 23 Ha BK, Vicini S, Rogers RC, Bresnahan JC, Buray RW, Beattie MS. 2 Post RM, Speer AM, Obrocea GV, Leverich GS. Acute and prophylactic Kainate-induced excitotoxicity is dependent upon extracellular potas- effects of anticonvulsants in bipolar depression. Clin Neurosci Res 2002; sium concentrations that regulate the activity of AMPA/KA type 2: 208–251. glutamate receptors. J Neurochem 2002; 83: 934–945. 3 Phiel CJ, Zhang F, Huang EY, Guenther MG, Lazar MA, Klein PS. Histone 24 Velasco I, Tapia R, Massieu L. Inhibition of glutamate uptake induces deacetylase is a direct target of valproic acid, a potent anticonvulsant, mood stabilizer, and teratogen. J Biol Chem 2001; 276: 36734–36741. progressive accumulation of extracellular glutamate and neuronal 4Go¨ttlicher M, Minucci S, Zhu P, Kramer OH, Schimpf A, Giavara S et al. damage in rat cortical cultures. J Neurosci Res 1998; 44: 551–561. Valproic acid defines a novel class of HDAC inhibitors inducing 25 Chaudhry FA, Lehre KP, Campagne MV, Ottersen OP, Danbolt NC, differentiation of transformed cells. EMBO J 2001; 20: 6969–6978. Stormmathisen J. Glutamate transporters in glial plasma-membranes— 5 Gray SG, Ekstrom TJ. The human histone deacetylase family. Exp Cell Res highly differentiated localizations revealed by quantitative ultrastructur- 2001; 262: 75–83. al immunocytochemistry. Neuron 1995; 15: 711–720. 6 Han JW, Ahn SH, Park SH, Wang SY, Bae GU, Seo DW et al. Apicidin, a 26 Mafra RA, Leao RM, Beirao PSL, Cruz JS. Electrophysiological evidence histone deacetylase inhibitor, inhibits proliferation of tumor cells for glial-subtype glutamate transporter functional expression in rat via induction of p21WAF1/Cip1 and gelsolin. Cancer Res 2000; 60: cerebellar granule neurons. Braz J Med Biol Res 2003; 36: 951–957. 6068–6074. 27 Saunders PA, Chen RW, Chuang D-M. Nuclear translocation of 7 Medina V, Edmonds B, Young GP, James R, Appleton S, Zalewski PD. glyceraldehyde-3-phosphate dehydrogenase isoforms during neuronal Induction of caspase-3 protease activity and apoptosis by butyrate and apoptosis. J Neurochem 1999; 72: 925–932. trichostatin A (inhibitors of histone deacetylase): dependence on 28 Carlile GW, Chalmers-Redman RME, Tatton NA, Pong A, Borden KE, protein synthesis and synergy with a mitochondrial/cytochrome c- Tatton WG. Reduced apoptosis after nerve growth factor and serum dependent pathway. Cancer Res 1997; 57: 3697–3707. withdrawal: conversion of tetrameric glyceraldehyde-3-phosphate 8 Kwon SH, Ahn SH, Kim YK, Bae GU, Yoon JW, Hong S et al. Apicidin, a dehydrogenase to a dimer. Mol Pharmacol 2000; 57:2–12. histone deacetylase inhibitor, induces apoptosis and Fas/Fas ligand 29 Krynetski EY, Kryneskaia NF, Gallo AE, Murti KG, Evans WE. A novel expression in human acute promyelocytic leukemia cells. J Biol Chem protein complex distinct from mismatch repair binds thioguanylated 2002; 277: 2073–2080. DNA. Mol Pharmacol 2001; 59: 367–374. 9 Ryu H, Lee J, Olofsson BA, Mwidau A, Deodoglu A, Escudero M et al. 30 Brown VM, Krynetski EY, Krynetskaia NF, Grieger D, Mukatira ST, Murti Histone deacetylase inhibitors prevent oxidative neuronal death KG et al. A novel CRM1-mediated nuclear export signal governs nuclear independent of expanded polyglutamine repeats via an Sp1-dependent accumulation of glyceraldehyde-3-phosphate dehydrogenase following pathway. Proc Natl Acad Sci USA 2003; 100: 4281–4286. genotoxic stress. J Biol Chem 2004; 279: 5984–5992. 10 Kang HL, Benzer S, Min KT. Life extension in Drosophila by feeding a 31 Zheng L, Roeder RG, Luo Y. S phase activation of the histone H2B drug. Proc Natl Acad Sci USA 2002; 99: 838–843. promoter by OCA-S, a coactivator complex that contains GAPDH as a 11 Loy R, Tariot PN. Neuroprotective properties of valproate—potential key component. Cell 2003; 114: 255–266. benefit for AD and tauopathies. J Mol Neurosci 2002; 19: 303–307. 32 Burke JR, Enghild JJ, Martin ME, Jou YS, Myers RM, Roses AD et al. 12 Bown CD, Wang JF, Young L. Increased expression of endoplasmic Huntingtin and DRPLA proteins selectively interact with the enzyme reticulum stress proteins following chronic valproate treatment of rat C6 GAPDH. Nat Med 1996; 2: 347–350. glioma cells. Neuropharmacology 2000; 39: 2162–2169. 33 Senatorov VV, Charles V, Reddy PH, Tagle DA, Chuang D-M. 13 Hall AC, Brennan A, Goold RG, Cleverley K, Lucas FR, Gordon-Weeks PR Overexpression and nuclear accumulation of glyceraldehyde-3-phos- et al. Valproate regulates GSK-3-mediated axonal remodeling and synapsin I clustering in developing neurons. Mol Cell Neurosci 2002; 20: phate dehydrogenase in a transgenic mouse model of Huntington’s 257–270. disease. Mol Cell Neurosci 2003; 22: 285–297. 14 Vandenberg RJ, Mitrovic AD, Chebib M, Balcar VJ, Johnston GAR. 34 Steffan JS, Bodai L, Pallos J, Poelman M, McCampbell A, Apostol BL et al. Contrasting modes of action of methylglutamate derivatives on the Histone deacetylase inhibitors arrest polyglutamine-dependent neuro- excitatory amino acid transporters, EAAT1 and EAAT2. Mol Pharmacol degeneration in Drosophila. Nature 2001; 413: 739–743. 1997; 51: 809–815. 35 Zhang CL, Mckinsey TA, Olson EN. Association of class II histone 15 Donevan SD, Beg A, Gunther JM, Twyman RE. The methylglutamate, deacetylases with heterochromatin protein 1: potential role for histone SYM 2081, is a potent and highly selective agonist at kainate receptors. methylation in control of muscle differentiation. Mol Cell Biol 2002; 22: J Pharmacol Exp Ther 1998; 285: 539–545. 7302–7312. 16 Chen RW, Saunders PA, Wei H, Li Z, Seth P, Chuang D-M. Involvement 36 Tremolizzo L, Carboni G, Ruzicka WB, Mitchell CP, Sugaya I, Tueting P of glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and p53 in et al. An epigenetic mouse model for molecular and behavioral neuronal apoptosis: evidence that GAPDH is upregulated by p53. J neuropathologies related to schizophrenia vulnerability. Proc Natl Acad Neurosci 1999; 19: 9654–9662. Sci USA 2002; 99: 17095–17100.

www.nature.com/tpj Valproate, an HDAC inhibitor, inhibits excitotoxicity H Kanai et al 344

37 Yildirim E, Zhang Z, Uz T, Chen C-Q, Manev R, Manev H. Valproate cortical neurons from spontaneous cell death: a role of histone administration to mice increases histone acetylation and 5-lipoxygenase deacetylase inhibition. FEBS Lett 2003; 542: 74–78. content in the hippocampus. Neurosci Lett 2003; 345: 141–143. 41 Ren M, Leng Y, Jeong MR, Leeds PR, Chuang D-M. Valproic acid reduces 38 Chen G, Zeng WZ, Yuan PX, Huang LD, Jiang YM, Zhao ZH et al. The brain damage induced by transient focal cerebral ischemia in rats: mood-stabilizing agents lithium and valproate robustly increase the potential roles of histone deacetylase inhibition and heat shock protein levels of the neuroprotective protein bcl-2 in the CNS. J Neurochem induction. J Neurochem 2004; 89: 1358–1367. 1999; 72: 879–882. 42 Fukumoto T, Morinobu S, Okamoto Y, Kagaya A, Yamawaki S. Chronic 39 Bown CD, Wang JF, Chen B, Young LT. Regulation of ER stress proteins lithium treatment increases the expression of brain-derived neuro- by valproate: therapeutic implications. Bipolar Disord 2002; 4: 145–151. 40 Jeong MR, Hashimoto R, Senatorov VV, Fujimaki K, Ren M, Lee MS et al. trophic factor in the rat brain. Psychopharmacology (Berl) 2001; 158: Valproic acid, a mood stabilizer and anticonvulsant, protects rat cerebral 100–106.

The Pharmacogenomics Journal