Neuroscience 226 (2012) 10–20 TRANSCRIPTIONAL EXPRESSION PATTERNS TRIGGERED BY CHEMICALLY DISTINCT NEUROPROTECTIVE MOLECULES D. J. PAPPAS, a* P. A. GABATTO, a D. OKSENBERG, a axis plays a significant role in recovery from glutamate- a a b P. KHANKHANIAN, S. E. BARANZINI, L. GAN AND induced excitotoxicity. Ó 2012 IBRO. Published by Elsevier a J. R. OKSENBERG Ltd. All rights reserved. a Department of Neurology, University of California, San Francisco, CA, USA b Gladstone Institute of Neurological Disease, University of Key words: neuron, glutamate, NMDA, excitotoxicity, neuro- California, San Francisco, CA, USA protection, microarray. Abstract—Glutamate-mediated excitotoxicity has been pur- ported to underlie many neurodegenerative disorders. A INTRODUCTION subtype of glutamate receptors, namely N-methyl-D-aspar- Deregulation of glutamate homeostasis within the central tate (NMDA) receptors, has been recognized as potential tar- nervous system and the resultant excitotoxic injury has gets for neuroprotection. To increase our understanding of the mechanisms that underlie this neuroprotection, we been linked to the pathological mechanisms that ensue employed a mouse model of glutamate receptor-induced after cerebral ischemia and trauma and potentially excitotoxic injury. Primary cortical neurons derived from underlie complex neurodegenerative disorders (Sheldon postnatal day-0 CD-1 mice were cultured in the presence and Robinson, 2007; Dong et al., 2009). However, or absence of neuroprotective molecules and exposed to despite decades of research and promising data from NMDA. Following a recovery period, whole genome expres- both in vitro experiments and clinical trials, efforts on sion was measured by microarray analysis. We used a com- specific therapeutic targeting of ionotropic glutamate (N- bination of database and text mining, as well as systems methyl-D-aspartate (NMDA)–AMPA–Kainate) receptors modeling to identify signatures within the differentially have been largely unsuccessful (Villmann and Becker, expressed genes. While molecules differed in their mecha- 2007; Besancon et al., 2008). The clinical shortcomings nisms of action, we found significant overlap in the expres- sion of a core group of genes and pathways. Many of these of targeting NMDA receptors may be due to poor molecules have clear links to neuronal protection and sur- relevance of animal models or suboptimal design of vival, including ion channels, transporters, as well as signal- clinical trials (Hoyte et al., 2004). The disconnect may ing pathways including the mitogen-activated protein kinase also arise from an oversimplified standard model of (MAPK), the Toll-like receptor (TLR), and the hypoxic induc- excitotoxicity, which links cell death to a linear cascade of ible factor (HIF). Within the TLR pathway, we also discov- signaling events following receptor overstimulation ered a significant enrichment of interferon regulatory (Besancon et al., 2008). For example, NMDA receptors factor 7 (IRF7)-regulated genes. Knockdown of Irf7 by RNA (NMDA-R) may stimulate cell survival or cell death interference resulted in reduced survival following NMDA signals, depending on their subcellular localization. treatment. Given the prominent role that IRF7 plays in the Whereas extra-synaptic NMDA-R activation may transduction of type-I interferons (IFNs), we also tested whether type-I IFNs alone functioned as neuroprotective preferentially trigger cell death cascades, synaptic agents and found that type-I IFNs were sufficient to promote NMDA-R activation may promote neuroprotection, neuronal survival. Our data suggest that the TLR/IRF7/IFN (Hardingham and Bading, 2010). The release of axonal glutamate can be preceded by large Na+ influxes which have been suggested to be more detrimental than the *Corresponding author. Address: 675 Nelson Rising Lane, Room 2+ 240, San Francisco, CA 94158, USA. Tel: +1-415-502-7211; fax: ultimate Ca imbalance of the standard model +1-415-476-5229. (Besancon et al., 2008). Moreover, an expanded E-mail addresses: [email protected] (D. J. Pappas), gabatto- repertoire of glutamate and Ca2+ sensing receptors and [email protected] (P. A. Gabatto), [email protected] (D. transporters in the CNS continues to unfold (Villmann Oksenberg), [email protected] (P. Khankhanian), sebaran@c- gl.ucsf.edu (S. E. Baranzini), [email protected] (L. Gan), and Becker, 2007; Besancon et al., 2008; Trapp and [email protected] (J. R. Oksenberg). Stys, 2009). Neuroprotective agents may have multiple Abbreviations: CBP, cAMP-response element binding protein; CREB, mechanistic roles in neuroprotection. For example c-AMP response element binding protein; DEG, differentially expressed genes; DKS, domain knowledge scores; HIF, hypoxic Riluzole, an FDA approved therapeutic for the treatment inducible factor; IFN, interferon; IRF7, interferon regulatory factor 7; of amyotrophic lateral sclerosis (ALS), has been MAPK, mitogen-activated protein kinase; MCAO, middle cerebral proposed to act as an antagonist of both glutamate artery occlusion; MEF-2, myocyte enhancer factor 2; NMDA, N- receptors and glutamate transporters (Villmann and methyl-D-aspartate; NMDA-R, NMDA receptors; TF, transcription factors; TFBS, TF binding sites; TLR, Toll-like receptor; TRAF4, TNF Becker, 2007), in addition to a tetrodotoxin-sensitive receptor-associated factor 4. sodium channel blocker (Song et al., 1997), and a two- 0306-4522/12 $36.00 Ó 2012 IBRO. Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.neuroscience.2012.09.007 10 D. J. Pappas et al. / Neuroscience 226 (2012) 10–20 11 pore potassium channel agonist (Mathie and Veale, 2007). technical replicates were concentration) to determine Also, the standard model has been limited by a neuronal neuroprotective efficacy following a NMDA-induced excitotoxic centric view. However, astrocytes and oligodendrocytes shock. Molecule concentrations that resulted in the highest level of cell viability (Table 1) were used for subsequent RNA are critical players in glutamate regulation and express a extraction and microarray analysis. Of the 20 molecules used, similar complement of ionotropic and metabotropic 14 were classified as protective and six non-protective. The glutamate receptors that render them vulnerable to experimental design included single replicates for treatments excitotoxic injury (Bolton and Paul, 2006). with the 20 molecules and five biological replicates for non- Finally, while many pathogenic mechanisms of treatment/vehicle controls. For RNA isolation, culture neurons glutamate excitotoxicity and cell death pathways have were pre-treated for 1 h in NBAM+ media (NBAM with either media alone, vehicle, or molecule), followed by a 1-h incubation been well established, we still do not fully understand in excitotoxic media (EXM+, 120 mM NaCl, 5.3 mM KCL, the complexities and multiplicity of networks, pathways, 1.8 mM CaCl2,15mMD-glucose, 25 mM Tris, pH 7.4 and intracellular signaling cascades that promote supplemented with 10 lM glycine and 100 lM NMDA) neuroprotection and cell survival (Lau and Tymianski, containing the respective molecule additives as in the NBAM+. 2010). To increase our understanding of the intracellular Following incubation, neurons were washed with NBAM, and mechanisms of neuroprotection, the current study used incubated for an additional 16 h in the respective NBAM+, at genome-wide expression analysis followed by a multi- which time the cells were harvested for RNA isolation (see below). step analytical approach that included text and database mining, as well as biological systems analysis. By employing primary mouse cortical neurons exposed to RNA isolation, microarray analysis, and quantitative an excitotoxic insult of NMDA in the presence or RT-PCR absence of neuroprotective molecules, we were able to Total RNA was extracted using the RNeasy Mini Kit (Qiagen, identify expression profiles that may represent shared Valencia, CA, USA, 74104) according to the manufacturer’s signatures of neuroprotection. Interestingly, while recommended protocols. For microarray analysis, 4 lg of total diverging chemically and acting through different RNA was sent to the Keck Foundation Biotechnology Resource putative mechanisms of action, we found that these at Yale University (New Haven, CT, USA) and assayed using molecules converged at the level of whole-genome the MouseWG-6 v1.1 Expression Bead Chips (Illumina, San transcription. Namely, these signatures include mitogen- Diego, CA, USA). The data were deposited in the National activated protein kinase (MAPK) signaling, calcium ion Center for Biotechnology Information Gene Expression Omnibus (Barrett et al., 2011), GEO ID: GSE38701. RNA transport, and cellular adhesion, as well as pathways expression from single genes was assessed using TaqManÒ related to ischemic tolerance, such as the hypoxic Gene Expression assays (Applied Biosystems, Carlsbad, CA, inducible factor (HIF) and Toll-like receptor (TLR) USA, 4440040) on an ABI 7900HT Sequence Detection pathways. Activation of these pathways may underlie a System, following all manufacturers’ protocols. Assays were fundamental mechanism driving neuronal survival. performed in triplicate using 50 ng of RNA per reaction and reported using the DDCt comparative method as detailed in the ABI User Bulletin #2, using Gapdh as the sample calibrator and EXPERIMENTAL PROCEDURES the references (relative controls) indicated in respective figures. Primary cortical neuron generation Neuronal transfection and RNA