Address correspondence and reprint requests to Michelle M. Aarts, Dept of Biological Sciences, University of Toronto at Scarborough, 1265 Military Trail (SW525), Scarborough, ON M1C 1A4, Canada. E-mail:email@example.com
NMDA-type glutamate receptors mediate both trophic and excitotoxic signalling in CNS neurons. We have previously shown that blocking NMDAR- post-synaptic density-95 (PSD95) interactions provides significant protection from excitotoxicity and in vivo ischaemia; however, the mechanism of neuroprotection is unclear. Here, we report that blocking PSD-95 interactions with the Tat-NR2B9c peptide enhances a Ca2+ -dependent protective pathway converging on cAMP Response Element binding protein (CREB) activation. We provide evidence that Tat-NR2B9c neuroprotection from oxygen glucose deprivation and NMDA toxicity occurs in parallel with the activation of calmodulin kinase signalling and is dependent on a sustained phosphorylation of the CREB transcription factor and its activator CaMKIV. Tat-NR2B9c-dependent neuroprotection and CREB phosphorylation are blocked by coapplication of CaM kinase (KN93 and STO-609) or CREB (KG-501) inhibitors, and by siRNA knockdown of CaMKIV. These results are mirrored in vivo in a rat model of permanent focal ischaemia. Tat-NR2B9c application significantly reduces infarct size and causes a selective and sustained elevation in CaMKIV phosphorylation; effects which are blocked by coadministration of KN93. Thus, calcium-dependent nuclear signalling via CaMKIV and CREB is critical for neuroprotection via NMDAR-PSD95 blockade, both in vitro and in vivo. This study highlights the importance of maintaining neuronal function following ischaemic injury. Future stroke research should target neurotrophic and pro-survival signal pathways in the development of novel neuroprotective strategies.
Excitotoxicity is the pathological process implicated in CNS (central nervous system) trauma, stroke and neurodegenerative disease. Excitotoxicity via calcium-permeable glutamate receptors has been highlighted as a critical initiating event in ischaemia-induced brain damage. In experimental models, excitotoxicity can be blocked using glutamate receptor antagonists, such as the non-competitive N-methyl-d-aspartate type glutamate receptor (NMDAR) antagonist MK-801. However, anti-excitotoxic therapies (AET) have not yielded viable therapeutic options for stroke because of a lack of positive clinical outcomes (Davis et al. 2000). NMDAR blockade has also been shown to induce widespread apoptosis in developing neurons and exacerbate neurodegeneration in the adult CNS (Ikonomidou et al. 2000; Adams et al. 2004). These latter effects stem from the fact that NMDAR activation is necessary for pro-survival signalling in developing neurons and the regulation of CNS plasticity in the adult [for review see (Hardingham 2006)]. Indeed, glutamatergic communication and neuronal communication are necessary for positive therapeutic outcomes and long-term neuronal survival following injury (Ikonomidou and Turski 2002; Calabresi et al. 2003).
Contrary to NMDAR blockade, inhibition of specific post-synaptic NMDAR signalling can afford neuroprotection from ischaemia with little or no effect on NMDAR-mediated Ca2+ entry (Aarts et al. 2002). In particular, peptides targeting NMDAR-PSD95 interactions uncouple receptor-dependent toxicity and provide significant protection in rat models of stroke (Aarts et al. 2002; Cui et al. 2007; Sun et al. 2008). Tat-NR2B9c is a cell-permeable, 20-amino-acid peptide containing the PDZ [post-synaptic density-95 (PSD95), Drosophila disc large (Dlg1) and zonula occludens-1 (zo-1) proteins]-binding motif of the NMDAR NR2B subunit coupled to the HIV-1 TAT membrane transduction domain. Tat-NR2B9c binds the first and second PDZ domains of PSD95, thereby inhibiting PSD95 interaction with NMDAR receptors. When applied up to 3 h after stroke onset, Tat-NR2B9c dramatically reduces infarct volume (87% over controls) and restores long-term neurobehavioral and functional outcomes in rats, primates (Aarts et al. 2002; Sun et al. 2008; Cook et al. 2012), and humans (Hill et al. 2012). Tat-NR2B9c neuroprotection was initially attributed to reduced nitric oxide (NO) production, as neuronal nitric oxide synthase (nNOS) is linked to NMDARs via PSD95(Kornau et al. 1995). However, NO is also linked to PI3 kinase/Akt, mGluR and opioid receptor trophic signalling in central neurons (Brenman and Bredt 1997; Sanchez-Blazquez et al. 2010; Boix et al. 2011). NO plays an important role in synaptic modulation (Contestabile 2000) as well as learning and memory formation (Llansola et al. 2009). Notably, clinical stroke trials with NO inhibitors and free-radical scavengers have failed to produce therapeutic outcomes (Savitz and Fisher 2007; Ginsberg 2008). More recent studies show that Tat-NR2B9c has no influence on nNOS activity in medium spiny neurons, yet still afford protection against NMDA-induced toxicity (Fan et al. 2009) implying that other signal pathways are influenced by Tat-NR2B9c peptide. Thus, the mechanism of robust Tat-NR2B9c neuroprotection remains unclear and elucidating the mechanism responsible for neuroprotection may provide more specific therapeutic avenues for stroke research.
Preconditioning neuroprotection, the process whereby a sublethal insult confers protection against a subsequent and normally lethal exposure, also affords robust neuroprotection. Preconditioning has been linked to increased CREB (cAMP response element–binding protein) phosphorylation (Meller et al. 2005) and CREB-dependent transcription [for review see (Kitagawa 2007)]. CREB phosphorylation at Ser133 regulates neuronal plasticity and memory formation and is required for neuronal survival (glutamate- and Ca2+ -dependent survival) during development (Ao et al. 2006; Balazs 2006). In CNS neurons, CREB phosphorylation is induced by synaptic activation of NMDARs and lies downstream of Ca2+/Calmodulin (CaM)-dependent protein kinase (CaMK) activation (Deisseroth et al. 1996). Both CaMKII and CaMKIV can regulate CREB activity; however, CaMKIV is specifically linked to trophic CREB activation and transcription (Bok et al. 2007). CaMKIV, a nuclear serine/threonine kinase, phosphorylates both CREB (at Ser133) and its transcription partner CBP (CREB-binding protein) (Impey et al. 2002), thereby activating trophic gene transcription (Bito et al. 1996; Impey et al. 2002; Feliciano and Edelman 2009). CREB may also be regulated by PKA, MAP-family kinases (ERK1/2, p38-MAPK), and JNK3; however, these kinases can disrupt CREB–CBP binding by phosphorylating Ser142 and Ser143 of CREB as well as activate c-fos/junB and NFAT (nuclear factor of activated T cells) transcription factors, which result in inflammatory gene transcription and neurotoxicity (Kornhauser et al. 2002; Lopez de et al. 2007). Finally, it should be noted that transient, cytoplasmic kinase activation is not sufficient for CREB-dependent transcription; prolonged CREB/CBP phosphorylation requires the addition of nuclear Ca2+ signals to promote trophic transcription (Deisseroth et al. 1996). Together, these works indicate that both CaMKIV and CREB activation are important for neuronal survival; however, the role of NMDAR signalling in regulating CaMKIV or CREB during neuronal injury is unclear.
We hypothesized that the robust neuroprotection provided by uncoupling PSD95 from NMDARs requires enhanced activation of CaM kinase signalling to the nucleus and subsequent CREB activation. Thus, we used in vitro and in vivo models of ischaemic neuronal injury to determine the contribution of CaM kinase signalling and pCREB to Tat-NR2B9c mediated neuroprotection. Tat-NR2B9c treatment results in transient activation of CaMKII and sustained NMDAR-dependent phosphorylation of CaMKIV and CREB, which is not afforded by NMDAR blockade. CaMKIV activation is required for the therapeutic effects of Tat-NR2B9c as CaMK blockade abrogates CREB phosphorylation and more importantly, neuroprotection both in vitro and in vivo. Our results show that Tat-NR2B9c enhances CaM kinase IV and CREB activation downstream of NMDARs and indicate that maintenance of trophic glutamatergic signalling may be critical to the ability of an antistroke therapy to afford neuroprotection.
Material and methods
Rat primary neuronal culture
All cell culture reagents including minimal essential media (MEM), Neurobasal, B27 supplement, glutamine and foetal bovine serum (FBS) were purchased from Invitrogen (Burlington, ON, CAN). Drugs and peptides were used at the following concentrations: Tat-NR2B9c or Tat-AA (control) peptide (200 nM, APTC, Hospital for Sick Children, Toronto); MK-801 (10 μM, Sigma Aldrich, Oakville, ON, CAN); KN93 and KN92 (inactive KN93 control; 10 μM; Calbiochem, Billerica, MA, USA); CaM kinase inhibitor STO-609 (10 μM; Calbiochem); CREB inhibitor KG501 (2-naphthol-AS-E-phosphate; 10 μM Sigma). Primary cortical neurons were isolated from E18 Sprague–Dawley rat embryos as previously described (Aarts et al. 2002). Briefly, whole cortices were dissected from E18 embryos and placed in ice-cold MEM + 25 mM glucose. Pooled cortices were digested with trypsin-EDTA for 7 min at 37°C, washed once in plating media (PM) (MEM with 10% FBS, 25 μM glutamate, 25 mM glucose and 25 mM dextrose) and then dissociated by tituration. Cell suspensions were passed through a 0.75-μm cell strainer, counted and seeded in poly-l-lysine-coated plates at a density of 1.5 × 105 cells/cm2. Cultures were maintained at 37°C in a humidified 5% CO2 atmosphere. After 2 h, PM was exchanged for growth media (Neurobasal containing 1% FBS, 1% glutamine and B-27 supplement) and glial growth was inhibited at DIV 4 by addition of 10 μM 5-fluoro-deoxyuridine (Sigma). Half the media was changed every 3 days and cells were used experimentally on DIV 12.
In vitro: oxygen glucose deprivation (OGD)
OGD in DIV12 cortical cultures was achieved via incubation in a 37°C anaerobic chamber (Plas Labs; Lansing, MI, USA) containing 5% CO2, 10% H2 and 85% N2 (< 0.01% O2). Cells were washed three times in degassed, glucose-free HEPES Buffered Solution (HBS) and maintained in anoxic glucose-free HBS containing CNQX (25 μM, Sigma) and nimodipine (2 nM, Sigma) for 1 h. OGD was terminated via washing with normal HBS containing CNQX and nimodipine (HCN) and normoxic incubation. Propidium iodide (PI) uptake was assessed immediately post insult and 20 h later as described below. HBS was comprised of 25 mM HEPES acid, 140 mM NaCl, 33 mM glucose, 5.4 mM KCl, 1.3 mM CaCl2 and 0.5 μM glycine (all Sigma), with pH 7.35 and osmolarity 320–330 mOsm. Glucose-free HBS was made by substituting glucose with NMDG (Sigma). Cultures were treated with vehicle (as control) or combinations of drugs and peptides as indicated. Drugs were administered 30 min before and during OGD.
In vitro: NMDA excitotoxicity
All cell manipulations were carried out in HCN. To trigger NMDAR-dependent excitotoxic cell death, neurons were washed two times in HBS, and incubated in HCN and NMDA (Sigma, concentrations as indicated). After 1 h, cells were returned to fresh HCN containing 10 μg/mL PI (Molecular Probes; Invitrogen; Burlington, ON, CAN). Cell death was assessed 20 h later as described below. Cells were pre-treated for 30 min with peptides and/or drugs as indicated.
In vitro: cell death assessment
Neuronal cell death was assessed via uptake of the fluorescent intercalating agent PI [see (Lau et al. 2007)]. Immediately following a 1-h exposure to either OGD or NMDA, cells were placed in HCN containing 10 μg/mL PI, and baseline levels of PI fluorescence were quantified using a BMG Fluor-optima fluorescent microplate reader. Cells were returned to the incubator and reassessed for PI uptake 20 h later. Fluorescent units measured at 20 h post OGD were normalized to the untreated control conditions for each experiment. Data are presented as the normalized fraction of total cell death occurring under non-treated control conditions 20 h post OGD.
In vitro: RNA interference
To support our findings, using pharmacological inhibitors, we performed CaMKIV knockdown using RNA interference. CaMKIV-specific siRNA (sc-72193) and nonsense control siRNA (sc-37007) were purchased from Santa Cruz Biotechnology (Santa Cruz, CA, USA) and transfection optimized in DIV 10 cortical neurons using Lipofectamine 2000 (Invitrogen). Transfection efficiency was confirmed using an FITC-tagged control siRNA (Fig. 5g) and knockdown of CaMKIV was confirmed by western blot 48 h after transfection, and toxicity assessed by PI uptake. A final siRNA dose of 125 nmol was chosen as yielding > 70% knockdown of CaMKIV and no residual toxicity to cortical neurons (as assessed by PI uptake) at 48 h post transfection. Cortical cultures, transfected on DIV 10 CaMKIV siRNA, were subjected to 1-h OGD injury and cell death assessment on DIV 12 as described above.
In vivo: permanent three pial vessel occlusion (3PVO) rat stroke model
Small, cortical infarcts were produced using the 3PVO rat stroke model as previously described (Forder et al. 2005; Sun et al. 2008). Male Sprague–Dawley rats (Charles River Laboratories, Sherbrooke, Quebec, QC, Canada) 280–320 g were used, where n = 8 per group. Excluding 12 h prior to surgery, animals had free access to standard laboratory chow both before and after surgery. Experimental procedures followed guidelines established by the Canadian Council for Animal Care and were approved by the University of Toronto Local Animal Care Committee. All efforts were made to minimize animal suffering, to minimize the number of animals used and to utilize alternatives to in vivo techniques as available. Rats were anaesthetized via 0.5 mL/kg intramuscular injection of ketamine (100 mg/kg), acepromazine (2 mg/kg) and xylazine (5 mg/kg) (all CDMV, St. Hyacinthe, QC, Canada), and supplemented with one-third the initial dose as required. Core body temperature was maintained at 37°C via a heating pad regulated by an anal temperature probe. The skull was exposed via a midline incision and a 6 × 4-mm cranial window was created above the right somatosensory cortex (2 mm caudal and 5 mm lateral to bregma) while keeping dura intact. Tat-NR2B9c or Tat-AA peptide (3 nmol/g) or saline was infused via tail vein injection 15 min prior to cauterization. Five microlitres of either KN93 (5 μg) or vehicle was applied to the vessel cauterization site 10 min prior to stroke. Small boluses (10–20 μL) of the vital dye patent blue violet (10 mM, Sigma in normal saline) were injected into the tail vein to visually demonstrate transit through cortical surface vessels. Three arteriolar branches of the middle cerebral artery (MCA) within the barrel cortex region were selected and electrically cauterized through the dura. Bolus injections of the vital dye were repeated to confirm that transit through the cauterized arterioles was blocked. The skull flap was then replaced and the scalp sutured. Each rat was returned to its individual cage under a heating lamp to maintain body temperature until full recovery. Breathing and recovery rates were monitored for up to 4 h following surgery. Animals were killed at 0, 15, 60 or 180 min post stroke for protein extraction, or 24 h post stroke to assess infarct size. Drug administration and infarct analysis were carried out using a double-blind protocol. Sham animals were exposed to all surgical treatments except cauterization of the pial vessels.
Triphenyltetrazolium chloride (TTC), a marker of mitochondrial function and cell viability, was used to assess infarct size, as previously described (Bederson et al. 1986). Animals were killed 24 h post stroke. The brains were quickly removed, sectioned into 2-mm coronal slices and incubated in saline containing 2% TTC (Sigma) for 15 min at 37°C. Brain sections were scanned and digital images obtained. In all cases, the 4th coronal section was quantified and infarct size presented as the percentage of the total hemisphere area remaining unstained (unstained/red areas, see Fig. 3a).
Primary cortical cultures subject to OGD or 40 μM NMDA were collected at the indicated times (0, 1, 2, 4 or 6 h post OGD). Cells were lysed for 10 min in ice-cold Cell-lytic buffer (Sigma) containing 20 μM Na Fluoride, 20 μM Na vanadate, phosphatase inhibitor cocktail II (Sigma) and protease inhibitors (Complete protease inhibitor cocktail, Roche, Mississauga, ON, CAN). Cell lysates were centrifuged at 10 000 g for 10 min to remove debris and immediately processed in Laemmli buffer at 90°C for 10 min. In vivo cortical tissue extracts from the infarct region and complimentary contralateral hemisphere regions were processed in a similar manner. Tissue was homogenized in chilled tubes on ice, using ice-cold Tissue-lytic buffer (Sigma) containing protease and phosphatase cocktail inhibitors, and processed as above. All protein samples were stored at −80°C until protein quantification by Bradford assay and subsequent western blot analysis.
Western blotting and antibodies
In vivo and in vitro protein extracts in Laemmli buffer were separated by size using 10% sodium dodecyl sulfate–polyacrylamide gel electrophoresis and transferred to nitrocellulose membranes. Membranes were blocked in 5% non-fat milk or 2% bovine serum albumin in TBS + 0.1% Tween-20 before overnight incubation at 4°C with the indicated primary antibodies: rabbit anti-phospho-CREB, rabbit anti-CREB and rabbit anti-CaMKIV (Cell Signalling Technology, Inc. Danvers, MA, USA), rabbit anti-phospho-CaMKIV (Santa Cruz Biotechnology,) and rabbit anti-phospho-CaMKII and rabbit anti-CaMKII antibodies (AbCam Antibodies, Cambridge, MS, USA). Equal loading was confirmed in each experiment by blotting for β-actin (mouse anti-β-actin, Sigma). Blots were probed first with anti-phospho antibodies, stripped with Reblot Plus strong stripping agent (Millipore, Billerica, MA), blocked and probed with the corresponding antibody for total protein. Primary antibodies were detected with goat anti-rabbit horseradish peroxidase-conjugated secondary (Sigma) and chemiluminescent substrate (Pierce; Thermo Scientific, Rockford, IL, USA). Bands of interest were quantified by densitometry using the NIH Image J analysis software (U.S. National Institutes of Health, Bethesda, MD, available at http://rsb.info.nih.gov/nih-image/). Graphical results are expressed as the ratio of phosphorylated to total protein.
Data are presented as the mean ± SEM. Statistical comparisons were made by either one- or two-way analysis of variance (anova). Where significant differences were observed, post-hoc Tukey's test was performed for multiple comparisons. Statistical significance was inferred at p < 0.05, where for all cases *p < 0.05, **p < 0.01 and ***p < 0.001. Symbols: * is used to denote differences between different treatment groups, whereas †p < 0.05 or ‡p < 0.01 denotes a significant difference within a treatment group.
Uncoupling NMDAR-PSD95 protects cortical cultures and prolongs CREB phosphorylation in OGD
Cell death in DIV 12 rat cortical neurons was assessed by PI uptake following a 1-h OGD insult as previously described (Lau et al. 2007). Glutamatergic signals were confined to NMDA receptors by the addition of the AMPA receptor antagonist CNQX and the calcium channel blocker nimodipine. Application of the non-competitive NMDAR antagonist MK-801 (10 μM) significantly reduced cell death, confirming the central role of NMDAR activation in OGD-induced neuronal death (Fig. 1a). Administration of Tat-NR2B9c (200 nM) also significantly reduced cell death (Fig. 1a) confirming that uncoupling NMDAR-PSD95 interactions is effective in reducing OGD-induced neuronal death. Unlike MK-801, Tat-NR2B9c allows NMDAR function to remain intact and provides more robust long-term neuroprotection in vivo than NMDAR antagonists (Aarts et al. 2002; Sun et al. 2008; Martel et al. 2009). Trophic signalling via NMDAR-dependent CREB activation may be critical for neuronal survival post injury (Papadia et al. 2005). We therefore determined whether disrupting PSD95 interactions or NMDAR blockade alters CREB activation. Phosphorylation of CREB at serine-133 is essential for transcriptional activity, thus, we assessed the effect of Tat-NR9B9c and MK-801 treatment on Ser133 pCREB levels following OGD. DIV 12 rat cortical cultures were subjected to OGD for 1 h in the presence of CNQX and nimodipine, with or without Tat-NR2B9c (200 nM) or MK-801 (10 μM), and proteins were harvested at 0, 1, 2, 4 and 6 h post insult. While total CREB protein levels remained unchanged, pCREB levels were significantly reduced after 1 h of OGD treatment, remaining undetectable through the 6-h time point (Fig. 1b and c). Interestingly, in Tat-NR2B9c-treated cultures, pCREB levels remained significantly elevated for up to 4 h after OGD (see Fig. 1b and c), suggesting that the peptide maintains trophic signalling to the nucleus. This result is consistent with studies showing CREB activation up to 3 h is sufficient for trophic gene transcription and survival (Lee et al. 2005), whereas excitotoxic NMDAR activity results in rapid loss of CREB phosphorylation (Hardingham et al. 2002). Multiple protein phosphatases including PP1 (protein phosphatase 1) and PP2A are responsible for CREB inactivation (Lonze and Ginty 2002). CREB phosphatases prevent chronic CREB activation, linked to aberrant plasticity (Lee et al. 2005), and probably account for the loss in CREB phosphorylation seen at 6 h in all treatment groups. Unlike Tat-NR2B9c, MK-801 application failed to support signalling to CREB, as evidenced by the loss of CREB phosphorylation after 1-h OGD (Fig. 1b and c). It is important to note that although MK-801 can prevent acute experimental excitotoxicity in vitro, it is ineffective in animal models of stroke after 1 h of ischaemia (Ginsberg 2008) or in prolonged anoxia (Aarts et al. 2003). Recent work also shows that MK801 treatment blocks NMDA-dependent transcription (Soriano et al. 2008). This result indicates that Tat-NR2B9c may afford neuroprotection by maintaining or prolonging an NMDAR-dependent trophic signal pathway that would be blocked by classical AET.
In vitro protection from NMDA toxicity is abrogated by coapplication of the CaMK inhibitors
CaM-dependent kinases have been linked to cell survival following injury or trophic withdrawal. The CaM kinase family member CaMKIV is of particular interest given its reported role in neuronal development, communication and survival (Bok et al. 2007). Localized to the nucleus, CaMKIV directly regulates CREB-dependent transcription, by phosphorylating CREB at serine-133 and also the required CREB coactivator CBP. Given our finding that Tat-NR2B9c maintains CREB in its active, phosphorylated state, we sought to determine whether CaM kinase signalling might be involved in the observed ischaemic neuroprotection. DIV 12 rat cortical cultures were pre-treated for 30 min with combinations of Tat-NR2B9c, KN93, KN92 (an inactive KN93 analogue), STO-609 or MK-801 in HBS containing CNQX and nimodipine. NMDA was then added at the indicated concentrations for 1 h to induce excitotoxicity. Application of Tat-NR2B9c and MK-801 significantly reduced NMDA-dependent excitotoxic cell death (Fig. 2), while coapplication of either the CaMK inhibitor KN93 or the CaMKK inhibitor STO-609 abolished the protective effects of Tat-NR2B9c (Fig. 2), implicating a role for the CaMK cascade in Tat-NR2B9c-induced protection. CaMKK modulates CaMKIV activity in response to Ca2+ signalling and synaptic activation (Enslen et al. 1995). Thus, the blockade of Tat-NR2B9c-mediated neuroprotection by KN93 and STO-609 indicates that sustained activation of CaMKIV rather than CaMKII is required for neuronal survival from excitotoxicity.
In vivo neuroprotection is blocked by inhibiting CaM kinase
In vivo administration of Tat-NR2B9c has been shown to provide significant long-term histological and functional neuroprotection in rat stroke models of cerebral ischaemia. Unlike NMDAR antagonists, Tat-NR9B9c is effective when administered up to 3 h after stroke onset (Aarts et al. 2002; Sun et al. 2008). Our present findings confirm that in vivo administration of Tat-NR2B9c (3 nmol/g) significantly reduces infarct size following 3PVO stroke in rats compared with vehicle or Tat-AA peptide controls, as assessed via TTC stain (Fig. 3a). Strikingly, coadministration of the CaMK inhibitor KN93 abolished the neuroprotective effects of Tat-NR2B9c (Fig. 3b). KN93 alone caused a small but non-significant increase in infarct size over control stroke (Fig. 3). These results suggest a central role for CaM kinase activity in the neuroprotection afforded by uncoupling NMDAR-PSD95 interactions.
Uncoupling NMDAR-PSD95 interactions prolongs CaMKIV phosphorylation during ischaemia
Although KN93 is widely used as an experimental inhibitor of CaMKIV, it was originally developed as a CaMKII antagonist (Enslen et al. 1994). For this reason, it was important to confirm whether TatNR2B9c-induced neuroprotection occurs as a result of altered CaMKII or CaMKIV activity. We assessed phosphorylation-dependent activation of CaMKIV and CaMKII via western blot analysis of brain tissue from rats subjected to 3PVO stroke. Cortical brain tissue was harvested from sham animals or the peri-infarct region at 0, 15 min, 1 h and 3 h post stroke from rats treated with Tat-AA, Tat-NR2B9c and/or KN93. Baseline CaMKIV phosphorylation was observed in untreated sham animals (Fig. 4a) and this phosphorylation rapidly and significantly decreased following stroke onset in Tat-AA-treated rats (Fig. 4a; †p < 0.05). Conversely, administration of Tat-NR2B9c caused prolonged maintenance of CaMKIV phosphorylation with significantly higher levels of activated pCaMKIV at both the 15-min and 3-h time points (as compared with Tat-AA *p < 0.05) (Fig. 4a). The effect of Tat-NR2B9c on pCaMKIV was blocked by the coapplication of KN93 (Tat-NR2B vs. Tat-NR2B + KN93; **p < 0.01). Administration of KN93 alone also abrogated CaMKIV phosphorylation at each time point examined (as compared to sham; ‡p < 0.01). Brain tissue lysates were analysed in parallel for pCaMKII and tCaMKII. Unlike CaMKIV, there was no significant reduction in pCaMKII levels following initial ischaemia (15 and 60 min) in Tat-AA-treated animals; however, pCaMKII levels were significantly attenuated 3 h after stroke onset (†p < 0.05), no effect of Tat-NR2B9c on pCaMKII in vivo (p > 0.05) (Fig. 4b). CaMKII can be robustly activated by a variety of mechanisms including voltage-gated Ca2+ channels (VGCC) (Wheeler et al. 2008) and autophosphorylation (Hudmon and Schulman 2002). Autophosphorylation in particular can confer Ca2+/CaM independence to CaMKII activity (Bayer et al. 2001; Hudmon and Schulman 2002; Vest et al. 2010) and this form of activation may then be resistant to NMDAR or Tat-NR2B9c-dependent mechanisms. The selective and prolonged activation of CaMKIV, but not CaMKII, and the blockade of neuroprotection by KN93, suggest a specific potentiation of the CaMKIV pathway by TatNR2B9c, and highlight a possible role for sustained CaMKIV activation in mediating Tat-NR2B9c-dependent protection from ischaemia.
Enhanced activation of CaMKIV signalling and pCREB by Tat-NR2B9c occurs downstream of NMDA receptors
Our data indicate that anoxic and ischaemic neuroprotection against experimental stroke requires activation of pCaMKIV leading to maintenance of pCREB levels. To further confirm that CaMKIV- and CREB-dependent protection is activated via NMDARs, we treated DIV 12 cortical neurons with 40 μM NMDA in the presence of CNQX and nimodipine, with or without Tat-AA, Tat-NR2B9c, KN93, KN92 or STO-609. STO-609 was used to confirm the involvement of CaMKIV in the observed protection as STO-609 specifically inhibits the CaMKIV activator CaMKK, while having no influence on CaMKII activation. CaMKII and CaMKIV phosphorylation levels were assessed 0, 30, 60, 120 and 240 min following NMDA stimulation (Fig. 5). NMDA application caused an initial increase in pCaMKIV (30 min), which decayed with time (Fig. 5a and b). Application of Tat-NR2B9c caused sustained CaMKIV phosphorylation for up to 4 h after NMDA application, and as predicted, coapplication of KN93 or STO-609 (but not KN92) abrogated this CaMKIV phosphorylation (Fig. 5a and b). Tat-NR2B also caused a small enhancement in pCaMKII levels; however, this effect was only significant at 30 min post NMDA administration and did not persist at longer time points (Fig. 5c and d), indicating that the peptide does not cause a sustained increase in overall CaMKII activation. CaMKII plays an important role in the initial signal events associated with NMDAR activation (Wu et al. 2001), but it appears that CaMKII activation is transient (within the first 30 min of this experiment) and is not significantly altered by Tat-NR2B9c treatment. We further confirmed the specific role of CaMKIV in Tat-NR2B9c neuroprotection by knockdown of CaMKIV using siRNA. DIV 10 Cortical cultures were transfected with CaMKIV or control siRNA, 48 h prior to a 1-h OGD injury ± Tat-NR2B9c treatment. Cell death was assessed at 20 h following OGD by PI uptake (Fig. 5e). As shown earlier, Tat-NR2B9c provides significant neuroprotection from OGD injury (p < 0.01). CaMKIV siRNA, but not control siRNA, significantly blocks this Tat-NR2B9c-mediated neuroprotection (p < 0.01). CaMKIV knockdown in cortical cultures was confirmed by western blot (Fig. 5f). Overall, Tat-NR9B9c treatment causes a robust and sustained activation of CaMKIV phosphorylation and neuroprotection, which is blocked by CaMKIV knockdown, suggesting that CaMKIV has a central role in Tat-NR2B9c-dependent neuroprotection.
To confirm that CREB is the downstream target of Tat-NR2B9c-dependent neuroprotection, we examined pCREB levels in cortical neurons subjected to 1 h OGD or 40 μM NMDA with or without Tat-NR2B9c and CaMK inhibitors. As in Fig. 1, we found that OGD caused a transient spike in CREB phosphorylation at 15 min, probably because of early activation of Ca2+ channels in anoxia (Fig. 6a). CREB phosphorylation had decayed by 1-h OGD in control cultures, however, it was significantly elevated at 1 h (p < 0.001) and 4 h (p < 0.01) after OGD by treatment with Tat-NR2B9c compared with controls. Both the early activation of pCREB and the sustained phosphorylation induced by Tat-NR2B9c were blocked by application of KN93 (but not KN92) and the CaMKK inhibitor STO-609 (Fig. 6a and b). The similar level of pCREB inhibition caused by KN93 and STO-609 in OGD further supports a role for CaMKIV-dependent CREB activation as the signal pathway modified by Tat-NR9B9c treatment.
Finally, the role of CREB activation in Tat-NR2B9c-dependent neuroprotection was examined in the presence of 10 μM KG-501. KG-501 is a cell-permeable naphthamide compound that blocks the binding of CREB to CBP, and effectively inhibits CREB-mediated gene transcription (Best et al. 2004). DIV 12 rat cortical cultures were pre-treated for 30 min with Tat-NR2B9c (200 nM) with or without 10 μM KG501 followed by application of 10, 40 or 100 μM NMDA for 1 h in HCN. As seen in earlier experiments, Tat-NR2B9c provides significant neuroprotection from excitotoxicity (p < 0.01), which is blocked by KG-501 (Fig. 6e). Importantly, blocking CREB with KG-501 does not enhance cell death at either lethal or sublethal doses of NMDA. Thus, Tat-NR2B9c-mediated neuroprotection requires CaMK- dependent signal transduction, converging on sustained CaMKIV and CREB activation.
The failure of AET and other therapeutics in clinical stroke treatment lead researchers to question the feasibility of neuroprotection. New insights into the cellular events responsible for neuronal death and a better understanding of both the toxic and trophic roles of excitatory neurotransmission is opening up new avenues for therapeutic research. The presence of protective signalling cascades downstream of NMDAR activation, such as enhanced antioxidant defences, suppression of pro-apoptotic signalling and maintenance of trophic signal events (Papadia et al. 2005; Balazs 2006) have helped to explain some of the observed disadvantages of overt NMDAR blockade (Ikonomidou and Turski 2002). Selectively uncoupling NMDARs from specific intracellular signals represents an exciting new approach to prevent both injury- and disease-induced neuronal damage. We present novel evidence that the NMDAR-dependent activation of CREB, mediated by CaMKIV signalling, is critical for ischaemic neuroprotection by the PSD95 inhibitor, Tat-NR2B9c.
NMDA receptors are linked to a network of scaffold and signal transduction proteins implicated in acute synaptic signalling and neurotoxicity to long-term plasticity and neuronal survival. Physiological NMDAR activation is necessary for trophic signalling events including phosphorylation of Akt and CREB (Ciani et al. 2002) and the activation of CaMKII and CaMKIV signalling (Bayer et al. 2001; Impey et al. 2002; Dick and Bading 2010). However, NMDAR overactivation is linked to Ca2+-dependent excitotoxic events including free-radical formation and kinase activation (Nakamura and Lipton 2010). This dichotomy of neurotrophic versus neurotoxic signalling has contributed to our inability to effectively use NMDA antagonists as neuroprotective therapies. The clinical failure of AET prompted the design of the Tat-NR2B9c peptide, which uncouples PSD95 and its associated signal proteins from the NR2B subunit, while allowing NMDAR-dependent calcium influx to continue unaffected (Aarts et al. 2002). Tat-NR2B9c significantly reduces anoxic neuronal death in both in vitro and in vivo models of stroke (Aarts et al. 2002; Sun et al. 2008) without interfering with NMDAR-dependent plasticity or pro-survival signalling to CREB (Soriano et al. 2008). Although protection was initially linked to decreased NO production, Tat-NR2B9c provides robust and long-term neuroprotection in vivo that is not paralleled by NOS inhibition alone (Ginsberg 2008), and in vitro protection can be afforded in the absence of any alteration in nNOS activity (Fan et al. 2009). Rather, CREB-dependent transcription of trophic genes can provide robust neuroprotection (Lee et al. 2005). Altogether these findings suggest that Tat-NR2B9c protects neurons by maintaining NMDAR-dependent trophic signalling, and with this in mind, we investigated the requirement of the calmodulin–dependent kinase cascade and CREB in the neuroprotection afforded by Tat-NR2B9c.
Neuronal death via oxygen glucose deprivation or NMDA excitotoxicity can be blocked by both MK-801 and Tat-NR2B9c (Figs 1and2); however, only Tat-NR2B9c treatment results in prolonged phosphorylation of CREB, indicating that uncoupling PSD95 interactions maintains normal, trophic signalling during the injury process. Our observation that Tat-NR2B9c application maintains CREB in its active state post OGD (Fig. 1b and c) led us to investigate whether CaMKIV, the most important trophic regulator of CREB (Lonze and Ginty 2002), is involved in NMDAR-dependent protective signalling. Our findings indicate that Tat-NR2B9c neuroprotection in vitro and in vivo requires CaMKIV (Figs 2, 3and6) and that Tat-NR2B9c enhances CaMKIV phosphorylation after injury (Figs 4and5). Finally, Tat-NR2B9c prolongs CREB activation following injury (Figs 1and6), and neuroprotection from excitotoxicity requires CREB activation (Fig. 6). Together, these results indicate that CaMKIV-dependent signalling and CREB activation are required for neuroprotection.
CNS neurons display a rapid decay in CREB phosphorylation during injury (< 15 min), an event associated with subsequent death (Walton and Dragunow 2000). Tat-NR2B9c prolongs CREB phosphorylation following OGD and peptide-mediated protection is reduced by the coapplication of CaMK inhibitor KN93, implicating CaM kinase signalling as a critical component of neuroprotection. As KN93 can block both CaMKIV and CaMKII activity (Enslen et al. 1994), we assessed the phosphorylation of both proteins in our injury models and further confirmed the critical role for CaMKIV in neuroprotection using CaMKIV siRNA and the CaMKK inhibitor STO-609 (Figs 5and6). As shown in Fig. 4, in vivo application of Tat-NR2B9c causes a specific maintenance of CaMKIV but not CaMKII phosphorylation indicating that CaMKIV signalling is selectively enhanced in Tat-NR2B9c-mediated neuroprotection. We compared infarct size in rats treated with Tat-NR2B9c with or without concomitant administration of KN93. In agreement with our in vitro cell death assays, the application of KN93 abolished the in vivo protective capacity of Tat-NR2B9c (see Fig. 3). Moreover, western blot analysis shows that Tat-NR2B9c prolonged the phosphorylation of CaMKIV but not CaMKII (Figs 4and6) in both in vitro and in vivo injury models, and that CaMKIV phosphorylation was blocked by KN93. These findings indicate a critical role for the nuclear CaMKIV-CREB pathway in Tat-NR2B9c-mediated neuroprotection. The lack of significant effect of Tat-NR2B9c or KN93 on CaMKII phosphorylation in vivo may be attributed to the relative abundance of CaMKII protein in central neurons or its activation by non-NMDAR channels or signalling events during stroke (Wheeler et al. 2008). CaMKII has multiple Ca2+/CaM-dependent and independent (autophosphorylated) activation states (Merrill et al. 2005; Buard et al. 2010) that may be regulated independent of NMDARs. Indeed, VGCCs have been shown to bind directly to CaMKII enabling the channels to play a primary role in CaMKII activation (Wheeler et al. 2008). Thus, ischaemic injury activates multiple Ca2+ channels that may contribute to the lack of effect of Tat-NR2B9c on CaMKII activation in vivo.
We show clear evidence that application of Tat-NR2B9c induces a selective potentiation of CaMKIV phosphorylation and maintenance of CREB activity post injury both in vitro and in vivo. A model for enhanced nuclear Ca2+/CaM/CaMKIV signalling facilitated by Tat-NR2B9c is shown in Fig. 7. PSD95 binds a large number of enzymes and scaffold proteins including nNOS and A-kinase anchor proteins (AKAPs). AKAPs such as AKAP79 are known to link PSD95 and neuronal channels to enzymes such as calcineurin/PP2B and PP1. This PSD95-dependent anchoring brings Ca2+/CaM-activated enzymes (nNOS, PP1 and calcineurin/PP2B) into close association with NMDA receptors (Lonze and Ginty 2002; Jurado et al. 2010; Le et al. 2011). Each of these enzymes in turn has the potential to negatively regulate Ca2+/CaM/CaMK signalling and CREB activation. We propose that uncoupling PSD95 from NMDARs (i) prevents the activation of negative regulators of Ca2+/CaM signalling and CREB, and (ii) removes competition for Ca2+/CaM binding, allowing prolonged nuclear Ca2+/CaM signalling downstream of NMDARs. The capacity of nuclear CaMKIV to phosphorylate both CREB and CBP (Impey et al. 2002) supports a primary role for CaMKIV in mediating the long-term protective effects of Tat-NR2B9c. CaMKIV can be activated directly by CaM or via phosphorylation by CaMKK. Although our in vitro data also show that Tat-NR2B9c causes a transient increase in CaMKII phosphorylation, there is no prolonged activation of this kinase that might indicate a role in neuroprotection. Rather, this transient event may be indicative of enhanced Ca2+/CaM-dependent signalling downstream of NMDARs. Indeed, Gardoni et al. (2006) showed that CaMKII and PSD95 may compete for binding to NMDAR NR2 subunits (Gardoni et al. 2006). Although CaMKII cannot be ruled out as an early signal event enhanced by Tat-NR2B9c, specific activation of CaMKIV requires CaM and CaMKK. It is important to note that while transfection of either constitutively active CaMKII or CaMKIV has been shown to promote neuronal survival (Hansen et al. 2003), only CaMKIV-mediated protection has been shown to be dependent on CREB activation (Bok et al. 2007). The CaMKK inhibitor STO-609 causes similar inactivation of CaMKIV and CREB to KN93 and both inhibitors block Tat-NR2B9c-dependent neuroprotection. CREB activation is implicated in neuronal survival in multiple models of preconditioning neuroprotection (Lee et al. 2005, 2009; Papadia et al. 2005, 2005; Lin et al. 2009; Sakamoto et al. 2011) and neuroprotection can be abrogated by expression of dominant negative CREB (Gao et al. 2004; Lee et al. 2005). It has also been reported that pCREB levels are sustained in neurons that survive ischaemia (Walton et al. 1996), while those destined to die had rapid CREB dephosphorylation (Walton and Dragunow 2000). Finally, work from independent groups indicates that only prolonged CREB phosphorylation (> 30 min) downstream of glutamate receptor activation promotes transcription (Deisseroth et al. 1996; Impey et al. 2002). Accordingly, we observed a rapid attenuation of CaMKIV and CREB phosphorylation in cortical neurons following NMDA, OGD and ischaemic injury. Tat-NR9B9c treatment resulted in prolonged CREB phosphorylation for as long as 4 h after injury onset. Interestingly, this agrees with the finding that Tat-NR2B9c application caused a small yet significant potentiating effect on CRE-dependent reporter gene expression (Martel et al. 2009) downstream of NMDAR activation. Thus, Tat-NR2B9c may provide significant neuroprotection by maintaining trophic, CREB-dependent transcription during ischaemia.
Our findings support previously published work on CREB-dependent neuroprotection and also provide the first demonstrated link between NMDAR-dependent CaMKIV signalling and ischaemic neuroprotection. It has been known for some time that synaptic activity and glutamatergic signalling can activate CREB; however, there are limited studies mapping the activation pathway from the membrane to the nucleus. Far fewer studies have drawn a link between NMDAR activation and CaMKIV regulation of pCREB (Impey et al. 2002; Feliciano and Edelman 2009). There is also little direct data available relating CaMKIV function to ischaemic or anoxic injury; however, there is a clear role for CaMKIV in survival of cultured neurons from trophic withdrawal. Constitutively active CaMKIV, but not CaMKII, rescues cerebellar granular neurons from KCl deprivation-induced apoptosis (See et al. 2001) as well as the survival of spiral ganglion neurons (Hansen et al. 2003) and chicken motor neurons in the absence of trophic support(Perez-Garcia et al. 2008), while dominant negative CaMKIV blocks the survival promoting effects of 30 mM KCl. It should be noted that CaMKIV regulates a number of downstream transcription and signalling targets, which most recently includes phosphorylation of salt-induced kinase (SIK2), a negative regulator of CREB activity in neurons (Sasaki et al. 2011). Sasaki et al. showed that CaMKIV-phosphorylated sik2 is targeted for degradation and that sik2−/− mice show decreased ischaemic neuronal injury compared with wild type. Thus, while our data show a clear link between CaMKIV, CREB activation and neuroprotection in experimental models of ischaemic injury, we cannot discount the possible neuroprotective effect of prolonged CaMKIV activation on distinct, non-CREB targets.
Overall, we provide novel evidence that, unlike NMDAR blockade, Tat-NR2B9c selectively enhances neuroprotective signalling via CaMKIV and CREB activation and that this CaMK-dependent signalling is required for neuroprotection. These findings may help to explain why stroke therapies aimed at simply blocking excitotoxic events have met with failure in clinical trials. This study also highlights the importance of maintaining trophic signalling and neuronal function in ischaemic neuroprotection, and provides new targets for neuroprotective research.
This work was funded by the Canadian Stroke Network and the Canada Research Chairs Program (grants to MA), a Canadian Institutes of Health Research Fellowship to KB and a Natural Sciences and Engineering Research Council graduate scholarship to RB. The authors declare that they have no conflict of interests.