AMP-activated protein kinase (AMPK)–induced preconditioning in primary cortical neurons involves activation of MCL-1

Authors

  • Ujval Anilkumar,

    1. Department of Physiology and Medical Physics, Centre for the Study of Neurological Disorders, Royal College of Surgeons in Ireland, Dublin, Ireland
    Search for more papers by this author
  • Petronela Weisová,

    1. Department of Physiology and Medical Physics, Centre for the Study of Neurological Disorders, Royal College of Surgeons in Ireland, Dublin, Ireland
    Current affiliation:
    1. Max F. Perutz Laboratories, University of Vienna, Dr. Bohr-Gasse 9, Austria
    Search for more papers by this author
  • Heiko Düssmann,

    1. Department of Physiology and Medical Physics, Centre for the Study of Neurological Disorders, Royal College of Surgeons in Ireland, Dublin, Ireland
    Search for more papers by this author
  • Caoimhín G. Concannon,

    1. Department of Physiology and Medical Physics, Centre for the Study of Neurological Disorders, Royal College of Surgeons in Ireland, Dublin, Ireland
    Search for more papers by this author
  • Hans-Georg König,

    1. Department of Physiology and Medical Physics, Centre for the Study of Neurological Disorders, Royal College of Surgeons in Ireland, Dublin, Ireland
    Search for more papers by this author
  • Jochen H. M. Prehn

    Corresponding author
    • Department of Physiology and Medical Physics, Centre for the Study of Neurological Disorders, Royal College of Surgeons in Ireland, Dublin, Ireland
    Search for more papers by this author

Address correspondence and reprint requests to Jochen H. M. Prehn, Department of Physiology and Medical Physics, Royal College of Surgeons in Ireland, 123 St Stephen's Green, Dublin 2, Ireland. E-mail: prehn@rcsi.ie

Abstract

Neuronal preconditioning is a phenomenon where a previous exposure to a sub-lethal stress stimulus increases the resistance of neurons towards a second, normally lethal stress stimulus. Activation of the energy stress sensor, AMP-activated protein kinase (AMPK) has been shown to contribute to the protective effects of ischaemic and mitochondrial uncoupling-induced preconditioning in neurons, however, the molecular basis of AMPK-mediated preconditioning has been less well characterized. We investigated the effect of AMPK preconditioning using 5-aminoimidazole-4-carboxamide riboside (AICAR) in a model of NMDA-mediated excitotoxic injury in primary mouse cortical neurons. Activation of AMPK with low concentrations of AICAR (0.1 mM for 2 h) induced a transient increase in AMPK phosphorylation, protecting neurons against NMDA-induced excitotoxicity. Analysing potential targets of AMPK activation, demonstrated a marked increase in mRNA expression and protein levels of the anti-apoptotic BCL-2 family protein myeloid cell leukaemia sequence 1 (MCL-1) in AICAR-preconditioned neurons. Interestingly, over-expression of MCL-1 protected neurons against NMDA-induced excitotoxicity while MCL-1 gene silencing abolished the effect of AICAR preconditioning. Monitored intracellular Ca2+ levels during NMDA excitation revealed that MCL-1 over-expressing neurons exhibited improved bioenergetics and markedly reduced Ca2+ elevations, suggesting a potential mechanism through which MCL-1 confers neuroprotection. This study identifies MCL-1 as a key effector of AMPK-induced preconditioning in neurons.

Abbreviations used
AICAR

5-aminoimidazole-4-carboxamide riboside

AMPK

AMP-activated protein kinase

BCL-2

B-cell lymphoma 2

CCD

charge coupled device

CNS

central nervous system

DIV

days in vitro

GLUT 3

glucose transporter 3

HEPES

N-[2-hydroxyethyl]piperazine-N'-[2-emanesulphonic acid]

MCL-1

myeloid cell leukaemia sequence 1

PBS

phosphate buffered saline

qPCR

quantitative PCR

shRNA

small hairpin RNA

siRNA

small interfering RNA

TMRM

tetramethylrhodamine methyl ester

Glutamate receptor overactivation has been postulated to play a major role in neuronal loss associated with ischaemic stroke, traumatic and epileptic brain injury (Choi and Rothman 1990). Neuronal death induced by glutamate receptor overactivation is crucially dependent on the magnitude of ionic imbalance, mitochondrial dysfunction, oxidative stress, glucose availability and ATP depletion (Choi 1987; Castilho et al. 1998; Delgado-Esteban et al. 2000). During excitotoxic injury neurons can undergo rapid necrosis, a poly(ADP-ribose) polymerase-dependent cell death pathway (parthanatos), or a more delayed form of caspase-independent, calpain-dependent apoptosis. Ultimately, which cell death pathways are activated depends on intrinsic neuronal properties as well as the intensity and duration of the excitotoxic insult (Ankarcrona et al. 1995; Lankiewicz et al. 2000; Yu et al. 2002; D'Orsi et al. 2012). Delayed, caspase-independent excitotoxic apoptosis differs from necrosis or parthanatos in that the initial ionic and energetic imbalance initially recovers. Nevertheless, neurons die subsequently as a result of mitochondrial permeabilization, a process that is controlled by B-cell lymphoma 2 (BCL-2) family proteins (Ward et al. 2007). BCL-2 family proteins either promote or inhibit apoptosis (Youle and Strasser 2008). They are classified into pro-apoptotic activators (BH-3 only proteins such as BIM, PUMA, BID, BAD, NOXA), pro-apoptotic effectors (BAX and BAK) and anti-apoptotic inhibitors (such as BCL-2, BCL-XL, MCL-1 and BCL-W) (Cheng et al. 2001; Wei et al. 2001; Zong et al. 2001).

Among the anti-apoptotic BCL-2 family proteins, myeloid cell leukaemia sequence 1 (MCL-1) is distinctive because it represent the only anti-apoptotic BCL-2 family member with an embryonic lethal phenotype present in gene-deficient mice (Rinkenberger et al. 2000). MCL-1 is essential for neuronal development as loss of MCL-1 in neuronal progenitors results in neuronal apoptosis (Arbour et al. 2008). Moreover, conditional MCL-1 gene deletion in the adult forebrain leads to massive cortical apoptosis and autophagy activation, suggesting a crucial role for MCL-1 also in the adult CNS (Germain et al. 2011). Interestingly, a recent study suggested that MCL-1 resides in two distinct isoforms and in two different mitochondrial compartments: an outer mitochondrial membrane localized MCL-1 that antagonizes apoptosis by binding pro-apoptotic BCL-2 family members, and a mitochondrial matrix MCL-1 that increases mitochondrial membrane potential, respiration and ATP production and regulates mitochondrial morphology (Perciavalle et al. 2012).

Preconditioning is a phenomenon where a stressful, but not damaging stimulus activates an endogenous adaptive response to reduce the impact of subsequent, more severe stimuli. Preconditioning-induced neuroprotection with metabolic toxins or low-dose excitotoxic stimuli has been successfully employed in models of excitotoxic and ischaemic injury (Wiegand et al. 1999; Smith et al. 2009; Navon et al. 2012). We have recently shown that mitochondrial uncoupling induced preconditioning protects neurons against subsequent excitotoxic injury through the activation of the energy sensor AMP-activated protein kinase (AMPK) which mediates an improvement of cellular and mitochondrial bioenergetics (Weisova et al. 2012). While the mammalian kinase AMPK regulates many aspects of cellular energy status (Hawley et al. 1996; Stein et al. 2000), the molecular factors mediating AMPK-induced preconditioning in neurons are less well characterized. In this study, we investigate the molecular signalling pathways involved in AMPK-mediated preconditioning induced by the pharmacological AMPK activator, 5-aminoimidazole-4-carboxamide riboside (AICAR), and identify MCL-1 as a key effector of AMPK-induced preconditioning.

Materials and methods

Materials

Foetal bovine serum, horse serum, minimal essential medium (MEM), B27 supplemented Neurobasal media, Fluo-4 AM and tetramethylrhodamine methyl ester (TMRM) were from Invitrogen (Bio Sciences, Dublin, Ireland). AICAR was obtained from Cell Signalling (Hertfordshire, UK). Compound C (water soluble) was obtained from Calbiochem (Merck Biosciences, UK). Glutamate, glycine, MK-801 and all other chemicals were from Sigma-Aldrich (Wicklow, Ireland).

Preparation of primary mouse neocortical neurons and cell culture

Mouse neocortical neurons were prepared and cultured as described previously (Concannon et al. 2010) with minor modifications (see Supporting Methods Information).

NMDA toxicity and determination of neuronal injury

Cortical neurons cultured on 24-well plate for DIV 9–11 were excited with NMDA/glycine (100 μM/10 μM) for 5 min and washed twice in experimental buffer containing (in mM): 120 NaCl, 3.5 KCl, 0.4 KH2PO4, 20 HEPES, 5 NaHCO3, 1.2 Na2SO4, 1.2 CaCl2 and 15 glucose, pH 7.4, supplemented with high Mg2+ (1.2 mM). To assess neuronal injury resulting from NMDA excitation, neocortical neurons were stained live with Hoechst 33258 at final concentration of 1 μg/mL for 10 min at 37°C. After incubation, nuclear morphology was assessed using an Eclipse TE 300 inverted microscope (Nikon, Dusseldorf, Germany) and 20x NA 0.45 dry objective. Images were taken using a SPOT RT SE CCD camera (Diagnostic instruments, Sterling Heights, MI, USA) and the appropriate filter sets. For each time point and treatment, cells were scored for condensed nuclei as a marker of apoptotic morphology in three subfields of each culture and repeated in triplicate. Condensed nuclei were counted as dead and expressed as percentage of total population. Images were processed using NIH Image J (Wayne Rasband, National Institute of Health, Bethesda, MD, USA). Analysis of images was performed by an independent investigator without prior knowledge of the treatment condition of the culture.

Protein extraction and western blotting

The cell pellets were lysed in ice-cold radio immunoprecipitation assay buffer (25 mM Tris HCl, 150 mM NaCl, 1% NP40, 1% sodium deoxycholate and 0.1% sodium dodecyl sulfate). The cell lysates were spun at 300 g for 3 min and supernatants were used for western blotting. Resulting blots were blocked with 5% milk for an hour and probed with a rabbit polyclonal anti-phosphorylated-(Thr-172)-AMPK (Cell Signalling, 1 : 1000), a rabbit polyclonal total anti-AMPKα antibody (Cell Signalling, 1 : 1000), a rabbit polyclonal anti-MCL-1 (Rockland, Gilbertsville, PA, USA, 1 : 1000), a rabbit polyclonal anti-NR 2A (Cell Signalling, 1 : 1000), a rabbit polyclonal anti-NR2B (Abcam, Cambridge, UK, 1 : 500) or a mouse monoclonal anti-β-actin (Sigma, 1 : 5000). Species specific horseradish peroxidase-conjugated secondary antibodies (Pierce, Northumberland, UK, 1 : 10 000) were detected using Super-Signal West Pico Chemiluminescent Substrate (Pierce) and imaged using a FujiFilm LAS-3000 imaging system (FujiFilm, Sheffield, UK).

Flow cytometry-based quantification of glucose transporter 3 (GLUT 3) cell surface expression

The GLUT3 cell surface expression assays were performed as described previously (Weisova et al. 2009) with minor modifications. Briefly, neocortical neurons cultured on 24-well plates were harvested after appropriate treatment and washed with phosphate buffered saline (PBS). Neurons were fixed with 1% formalin for 20–25 min on ice and quenched using glycine (0.1 M) and incubated in an anti-GLUT 3 antibody (Millipore Biosciences Research Reagents, Billercia, MA, USA) at a concentration of 1 : 250 diluted in PBS plus 0.1% bovine serum albumin at 4°C overnight. Cells were washed twice with PBS and incubated with in an Alexa Fluor 488 goat anti-rabbit IgG (H+L) antibody (Invitrogen) diluted 1 : 250 for 1 h. Samples were acquired on a Partec CyFlow ML flow cytometer with a minimum of 10 000 events per sample and analysed using FloMax software (Partec, Munster, Germany).

ATP levels

ATP levels were determined using a luciferase assay (Weisova et al. 2009) (see Supporting Methods Information).

Plasmids and transfection

Neocortical neurons (DIV 7) and NSC 34 cells were transfected using Lipofectamine 2000 (Invitrogen) or calcium-phosphate (Goetze et al. 2004). For inhibition of AMPK, cells were transfected with a vector expressing a siRNA targeting AMPK –α1/ α2 or a control sequence as previously described (Weisova et al. 2009). For over-expression of MCL-1, cells were transfected with vector expressing MCL-1 (MC200829; OriGene, Rockville, MD, USA). For inhibition of Mcl-1 cells were transfected with a vector expressing shRNA targeting MCL-1 (SCHLNG-Nm_008562; Sigma) or a scramble control vector (SHC001; Sigma). A plasmid with enhanced CFP (ECFP-C1; Clontech, Dublin, Ireland) was used to identify the transfected neurons for confocal microscopy experiments and a plasmid with enhanced GFP (eGFP-N1; Clontech) was used to allow the identification of transfected neurons for cell death assays. An average transfection efficiency of 26.2 ± 0.7% was achieved in our experiments. Cells were used for experiments 48 h after transfection.

Confocal microscopy

Primary neocortical neurons were loaded with TMRM (20 nM) and Fluo-4 AM (3 μM) for 30 min at 37°C in the dark in experimental buffer. The buffer on Willco dishes (Willco Wells B.V., Amsterdam, the Netherlands) with neurons was exchanged for buffer without Fluo-4 AM, covered with a thin layer of embryo tested mineral oil (Sigma, Ireland) to prevent evaporation, and placed on a stage of an LSM 710 Zeiss confocal microscope with a 63x NA 1.4 differential interference contrast objective and a thermostatically regulated chamber maintained at 37°C (Zeiss, Jena, Germany). During live cell imaging, multiple fields of view were addressed using the multiple time series macro in combination with the hardware autofocus. On stage cells were treated with 100 μM NMDA/10 μM glycine for 5 min. Images were captured every 60 s. To terminate NMDA receptor activation, MK 801 (10 μM) was added to neurons 5 min after excitation. CFP was excited at 405 nm and the emission collected in the range of 440–490 nm, Fluo-4 was excited at 488 nm and the emission was collected in the range of 500–550 nm and TMRM was excited at 543 nm and emission was collected in the range of 560–700 nm. Quantification of Fluo-4 fluorescence intensities is described in Supporting Methods Information. Images were processed using MetaMorph software version 7.5 (Molecular Devices, Berkshire, UK), and data were normalized to baseline response.

Gene expression analysis

Expression of MCL-1, AMPK-α1 and β-actin mRNA was determined by qPCR (see Supporting Methods Information).

Statistical analysis

Data are presented as mean ± SEM. One-way anova followed by Tukey's post hoc test was performed to determine statistical significance. p-values p ≤ 0.05 were considered to be statistically significant.

Results

Preconditioning with a low concentration of AICAR activates AMPK and improves cellular bioenergetics

AICAR is a cell permeable precursor of the monophosphate nucleotide ZMP (Sabina et al. 1985), which mimics the effects of AMP on AMPK activation (Sullivan et al. 1994). Previous studies have established that AMPK can be activated using AICAR (Sabina et al. 1985; Sullivan et al. 1994), however, prolonged treatments with high concentrations of AICAR (1–5 mM) have also shown to promote apoptosis in neurons and other cell types (Meisse et al. 2002; Kefas et al. 2003; Concannon et al. 2010) similar to over-expression of constitutively active mutants of AMPK (Meisse et al. 2002; Concannon et al. 2010). We therefore set out to establish the optimal concentration of AICAR required to activate AMPK in cultured neocortical neurons without inducing cell death per se. Continuous treatment of neuronal cultures with different concentration of AICAR (0.1, 0.25, 0.5, 1.0 and 2.5 mM) for 24 h resulted in a concentration dependent cell death as evidenced by nuclear pyknosis, with 0.1 mM AICAR showing no significant increase in toxicity over a 24-h time period (Fig. 1a). Maximal AMPK activation was observed after 2-h treatment with 0.1 mM AICAR (Fig. 1b and c), as determined by Western blot analysis using a specific antibody that detects activated, Thr172-phosphorylated AMPK (Hawley et al. 1996; Stein et al. 2000). This data suggested that treatment with AICAR at 0.1 mM activates AMPK in cultured neocortical neurons without any significant cytotoxic effects.

Figure 1.

Activation of AMP-activated protein kinase (AMPK) with a subtoxic concentration of 5-aminoimidazole-4-carboxamide riboside (AICAR) (0.1 mM). (a) Dose-response study in cortical neurons. Nuclei were stained with Hoechst 33358 (1 μg/mL) 24 h after AICAR addition. Controls were treated with vehicle. Neurons with uniformly stained nuclei were counted as viable and condensed nuclei counted as apoptotic. Data are from two different set of experiments performed in triplicate and were pooled for analysis. *p ≤ 0.05, difference between vehicle and 0.25 to 2.5 mM AICAR treatment. (b) Treatment with AICAR (0.1 mM) activates AMPK as analysed by western blotting. β-actin served as loading control. Similar responses were obtained in samples from a second separate experiment. (c) AMPK activation quantified by densitometry of western blotting experiments and normalized to total AMPK levels. *p ≤ 0.05, difference between vehicle and AICAR treated at 2 h and 4 h. Data are presented as means ± SEM (n = 4 treatments from two experiments).

Previous studies have shown that AMPK activation is able to improve survival and cellular bioenergetics in primary neurons in response to glucose withdrawal or ATP depletion (Culmsee et al. 2001; Weisova et al. 2012). We therefore next investigated whether a 2-h exposure to 0.1 mM AICAR followed by washout improved cellular bioenergetics. A treatment with AICAR (0.1 mM) for 2 h resulted in a robust activation of AMPK (Fig. 2a and b) that was maintained for up to 24 h after washout (Fig. 2a and b). We have recently shown that activation of AMPK regulates glucose transporter 3 (GLUT 3) surface expression leading to neuronal tolerance during glutamate excitotoxicity (Weisova et al. 2009). Flow cytometry analysis demonstrated increased cell surface expression of GLUT 3 in neurons treated with 0.1 mM AICAR for 2 h, which remained elevated for up to 24 h after washout of AICAR (Fig. 2c). Similarly, the ATP levels of neurons treated with 0.1 mM AICAR for 2 h were significantly elevated up to 24 h post-washout (Fig. 2d). Exposure to the protonophore FCCP which dissipates ATP production in mitochondria served as an internal negative control (Fig. 2d).

Figure 2.

5-aminoimidazole-4-carboxamide riboside (AICAR) (0.1 mM) up-regulates AMP-activated protein kinase (AMPK) and stimulates neuronal metabolism in murine cortical neurons. (a) Model of AICAR preconditioning. Neurons were treated with 0.1 mM AICAR for 2 h, the medium was exchanged, and cells were recovered for various time periods up to 48 h after washout (Upper panel). Western blot analysis of AMPK activation status in cultured cortical neurons treated with AICAR (0.1 mM) for 2 h and washout for the indicated time periods (Lower panel). Similar responses were obtained in samples from three separate experiments. (b) Densitometry analysis showing activated AMPK normalized to total AMPK activity and expressed relative to vehicle treated controls. *p ≤ 0.05, difference between vehicle and AICAR treated neuronal cultures. Data are presented as means ± SEM (n = 6 treatments from three experiments). (c) Quantification of GLUT 3 surface expression by flow cytometry. Data are represented as mean ± SEM. *p ≤ 0.05, differences between vehicle and AICAR treated neurons. Data are from two different set of experiments performed in triplicate and were pooled for analysis. (d) Cortical neurons were treated with AICAR and their ATP content measured (pmol ATP/mg protein) at indicated time points. n = 2 experiments in triplicate; *p ≤ 0.05, difference between vehicle and AICAR treated neurons. Data reported as mean ± SEM.

Preconditioning with a low concentration of AICAR protects cortical neurons against NMDA induced excitotoxic cell death

Previous studies have shown that AMPK activation prior to a toxic challenge is able to protect neurons against metabolic, excitotoxic and ischaemic injury (Culmsee et al. 2001; Weisova et al. 2009, 2012). We therefore next investigated whether ‘preconditioning’ of neurons induced by a 2-h exposure to 0.1 mM AICAR followed by washout protected primary cortical neurons in a well-characterized model of NMDA-induced excitotoxic injury (D'Orsi et al. 2012). Neurons were pre-treated with AICAR for 2 h, followed by washout and recovery for varying time periods up to 48 h before being challenged with the NMDA stimulus (100 μM NMDA/10 μM glycine). As shown in Fig. 3, AICAR pre-treatment followed by washout and recovery for 24 h offered a significant protection against NMDA excitotoxicity in neurons exposed to NMDA for 5 min, as assessed by quantification of pyknotic nuclei using Hoechst 33258 staining 24 h after the NMDA exposure (Fig. 3a and b). No significant protection against NMDA excitotoxicity was observed after a 48 h washout period. These results suggested that activation of AMPK with low AICAR concentration (0.1 mM for 2 h) effectively preconditioned cortical neurons against NMDA excitotoxicity.

Figure 3.

5-aminoimidazole-4-carboxamide riboside (AICAR) pre-treatment protects cortical neurons against NMDA-mediated excitotoxicity. (a) Representative images of Hoechst-stained neurons treated with vehicle or AICAR, or exposed to NMDA/glycine (100 μM/10 μM) for 5 min in the absence or presence of AICAR preconditioning. Images were taken 24 h post-NMDA excitation. Scale bar, 10 μm. (b) Quantification of cell death. Neuronal cultures were treated with AICAR (0.1 mM) for 2 h and subsequently exposed to NMDA at 0 h, 6 h, 24 h and 48 h post-AICAR washout time points. Nuclei were stained with Hoechst and uniformly stained nuclei counted as viable and condensed nuclei scored as dead. Data are presented as mean ± SEM. *p ≤ 0.05, difference between NMDA treated and AICAR washout treated with NMDA (n = 9 treatments performed on neurons from three different experiments).

Inhibition of AMPK activation abolishes AICAR preconditioning

To demonstrate that AICAR preconditioning was mediated by an increased AMPK activity, we next employed the pharmacological AMPK inhibitor Compound C (Dasgupta and Milbrandt 2007). Compound C was added 30 min before and during the AICAR treatment. Inhibition of AICAR-induced AMPK activation was evident in cortical neurons as determined by western blotting (Fig. 4a). Inhibition of AMPK activity with Compound C also resulted in decreased GLUT 3 cell surface expression and ATP availability after 2 h of AICAR treatment (0.1 mM) (Fig. 4b and c). In addition, no significant neuroprotection against NMDA excitotoxicity was observed in neuronal cultures pre-treated with AICAR and Compound C for 2 h compared with cultures pre-treated with AICAR only for 2 h (Fig. 4d). Treatment with Compound C alone did not alter cell viability (data not shown).

Figure 4.

Inhibition of AMP-activated protein kinase (AMPK) activation with Compound C abolishes the effect of 5-aminoimidazole-4-carboxamide riboside (AICAR) preconditioning on neuronal bioenergetics and NMDA toxicity. (a) Western blot analysis of cortical neurons treated with AICAR (0.1 mM for 2 h) only or co-treated with Compound C (10 μM) and harvested immediately afterwards. Similar results were obtained in two separate experiments. (b) Quantification of median fluorescence intensity (arbitrary units) by flow cytometry for GLUT 3 surface expression. *p ≤ 0.05, difference between neurons treated with AICAR (0.1 mM for 2 h) and those co-treated with Compound C (10 μM). Data are from three separate experiments performed in triplicate and were pooled for analysis. (c) Cortical neurons were treated with AICAR (0.1 mM for 2 h) only or co-treated with Compound C (10 μM) and ATP content (pmol/mg protein) was measured by luminescence assay. Data shown represent mean ± SEM. *p ≤ 0.05, difference between neurons treated with AICAR and co-treated with Compound C. Data are from two different set of experiments performed in triplicate and were pooled for analysis. (d) Cortical neurons treated with AICAR alone or co-treated with Compound C and exposed to NMDA for 5 min and assayed over 24 h. Nuclei were stained with Hoechst, with uniformly stained nuclei counted as viable and condensed nuclei scored as dead. Data are presented as mean ± SEM. *p ≤ 0.05, compared with AICAR co-treated with Compound C and exposed to NMDA and NMDA only treatment. Data are from two different set of experiments performed in triplicate and were pooled for analysis.

To rule out potential off-target effects of pharmacological inhibition of AMPK with Compound C, we also investigated the effect of AMPK inhibition using gene silencing. Transfection of cortical neurons with a previously described siRNA targeting AMPKα1/2 (Weisova et al. 2009) (Fig. 5a) showed a significant increase in NMDA-induced cell death in response to AICAR preconditioning when compared with control siRNA transfected neurons (Fig. 5b). In agreement with previous studies showing pro-apoptotic activities of prolonged AMPK activation during excitotoxic injury (Concannon et al. 2010; Davila et al. 2012) there was a significant decrease in excitotoxic cell death in AMPK siRNA transfected neurons that were not preconditioned with AICAR when compared with neurons transfected with control siRNA (Fig. 5b). In summary, these results suggest that AMPK activation was pivotal in mediating the effects of AICAR preconditioning.

Figure 5.

AMP-activated protein kinase (AMPK) inhibition using gene silencing attenuates AMPK mediated neuroprotection against NMDA-induced excitotoxicity. (a) Densitometry analysis showing AMPK levels in NSC 34 cells transfected with Control siRNA or AMPK siRNA (upper panel). Data shown represents mean ± SEM from n = 2 experiments. Western blot analysis of NSC 34 cells transfected with Control siRNA or AMPK siRNA. AMPK levels were assessed 48 h after transfection by western blotting. β-actin served as loading control (lower panel). Similar results were obtained in a separate experiment. (b) Cortical neurons were transfected with siRNA vector targeting AMPK α1/α2 or a Control siRNA. Condensed nuclei in the GFP positive cells were considered as dead (n = 120–148 cells per time point quantified). *p ≤ 0.05, compared with Control siRNA transfected neurons exposed to NMDA. #p ≤ 0.05, difference between Control siRNA and AMPK siRNA transfected neurons treated with 5-aminoimidazole-4-carboxamide riboside (AICAR) and exposed to NMDA. NS, no significance. Experiments were performed on neurons from three separate platings. Data presented as mean ± SEM.

AICAR preconditioning increases mRNA and protein levels of the anti-apoptotic BCL-2 family member MCL-1

To assess the role of increased gene transcription in mediating AICAR preconditioning-induced protection against NMDA toxicity, we analysed the mRNA expression of several genes involved in the regulation of cellular bioenergetics, mitochondrial function and in the control of neuronal apoptosis, including mtTFA, PGC-1α, BCL-w, survivin and MCL-1 by quantitative PCR (qPCR). While mRNA levels of mtTFA, PGC-1α, BCL-w and survivin did not change significantly (data not shown); we detected increased mRNA levels of the anti-apoptotic BCL-2 family protein, MCL-1. A significant increase in MCL-1 mRNA was sustained up to 6 h after washout of AICAR (Fig. 6a). Western blot analysis also revealed a significant increase in the protein levels of MCL-1 in response to AICAR preconditioning, with similar kinetics (Fig. 6b). Furthermore, the AICAR-induced increase in MCL-1 mRNA levels was abolished in neurons transfected with AMPK siRNA compared with control siRNA transfected neurons (Fig. 6c and d). In addition, inhibition of AMPK activity using Compound C abolished AICAR-induced activation of MCL-1 (Figure S1). These results suggest that AICAR preconditioning up-regulates the anti-apoptotic BCL-2 family protein MCL-1 in an AMPK-dependent manner.

Figure 6.

5-aminoimidazole-4-carboxamide riboside (AICAR) preconditioning increases mRNA and protein levels of myeloid cell leukaemia sequence 1 (MCL-1) at the time point of maximal protection against NMDA excitotoxicity. (a) Real time qPCR analysis of MCL-1 mRNA levels in cortical neurons treated with AICAR (0.1 mM for 2 h) with washout time points indicated. Expression levels were normalized to β-actin mRNA and expressed relative to vehicle treated control. Data shown represents mean ± SEM from n = 3 experiments. *p ≤ 0.05, difference between vehicle treated and AICAR treated cultures. (b) Western blot and densitometric analysis of MCL-1 protein levels in cultured cortical neurons treated on DIV 9–11 with 0.1 mM AICAR for 2 h and washout as indicated. For quantification MCL-1 levels were normalized to β–actin. Data presented as mean ± SEM. *p ≤ 0.05, difference between vehicle and AICAR treated neurons from three separate experiments. (c) Real time qPCR analysis of MCL-1 expression in cortical neurons transfected with control siRNA or AMP-activated protein kinase (AMPK) siRNA and treated with vehicle or AICAR (0.1 mM for 2 h). Expression levels were normalized to β-actin and expressed relative to vehicle treated control. Data shown represents mean ± SEM from n = 3 experiments. *p ≤ 0.05, difference between vehicle treated and AICAR treated cultures in control siRNA transfected neurons. (d) Real time qPCR analysis of AMPK-α1 expression in cortical neurons transfected with control siRNA or AMPK siRNA. Expression levels were normalized to β-actin. Data shown represents mean ± SEM from n = 3 experiments. *p ≤ 0.05, difference compared with cortical neurons transfected with control siRNA.

MCL-1 is neuroprotective against NMDA toxicity

Previous studies demonstrated that over-expression of the anti-apoptotic proteins BCL-2 or BCL-XL can protect primary neurons against glutamate receptor-mediated excitotoxic injury (Wang et al. 2004; Dietz et al. 2007). We sought to determine whether MCL-1 exerts a similar neuroprotective activity in our setting of NMDA excitotoxicity. Neurons were transfected with a mammalian expression vector expressing MCL-1 and co-transfected with an EGFP expressing plasmid, or only transfected with the EGFP vector (controls). After 48 h cells were preconditioned with AICAR or treated with vehicle for 2 h, and then subjected to NMDA toxicity. Cell death of EGFP-positive cells was analysed 24 h after NMDA toxicity by evaluating nuclear pyknosis using Hoechst 33258 as well as cell shrinkage as morphological criteria (Fig. 7a and b). Analysis of EGFP-positive neurons that were not pre-conditioned with AICAR indicated that transfection with MCL-1 significantly reduced cell death compared with EGFP-transfected neurons (Fig. 7c) There was no significant difference in protection achieved with AICAR preconditioning, MCL-1 over-expression or combined AICAR preconditioning and Mcl-1 over-expression (Fig. 7c).

Figure 7.

Myeloid cell leukaemia sequence 1 (MCL-1) plays a critical role in 5-aminoimidazole-4-carboxamide riboside (AICAR) induced neuroprotection following NMDA induced excitotoxicity in cortical neurons. (a, b) Representative images of empty vector and Mcl-1 vector transfected cortical neurons. Cortical neurons were vehicle treated and sham-exposed (controls Vehicle), pre-treated with AICAR (0.1 mM for 2 h) and subsequently exposed to NMDA, or pre-treated with vehicle and then exposed to NMDA. GFP and Hoechst 33358 (1 μg/mL) fluorescence images were acquired to identify transfected cells and quantify nuclear apoptosis. GFP positive cells with condensed nucleus were considered as dead. Scale bar, 5 μm. (c) Quantification of the effect of MCL-1 over-expression (OE) on cell survival after NMDA toxicity. Neurons were vehicle treated and sham-exposed, pre-treated with AICAR (0.1 mM for 2 h) and subsequently exposed to NMDA, or pre-treated with vehicle and then exposed to NMDA. A total of n = 88–168 cells per treatment were quantified in neurons from three different platings. *p ≤ 0.05, compared with NMDA exposed, empty vector transfected controls. NS, no significance, compared between MCL-1 OE neurons exposed to NMDA treated with or without AICAR. (d) Effect of MCL-1 gene silencing. Top: NSC34 cells were transfected with shRNA against MCL-1 or scramble shRNA. Reduced expression levels of MCL-1 after shRNA expression compared with scramble were assessed by western blotting 48 h after transfection. β-actin served as control. Bottom: Quantification of neuronal survival. Neurons transfected with either MCL-1 shRNA or control shRNA were vehicle treated and sham-exposed, pre-treated with AICAR (0.1 mM for 2 h) and subsequently exposed to NMDA, or pre-treated with vehicle and then exposed to NMDA. A total of n = 98–126 cells per treatment condition were quantified. Experiments were performed on cultures from two independent platings. *p ≤ 0.05, differences between MCL-1 shRNA and Scramble shRNA transfected neurons. #p ≤ 0.05, compared with NMDA exposed Scramble shRNA transfected neurons with or without AICAR preconditioning.

We next sought to determine whether AICAR preconditioning still exerted a protective activity in neurons silenced for MCL-1 expression. A shRNA construct targeting MCL-1 was successfully tested in murine NSC-34 cells and demonstrated strongly reduced MCL-1 protein levels after 48 h of transfection when compared with scrambled vector transfected cells (Fig. 7d). Transfection of the shRNA targeting MCL-1 in cortical neurons induced significant cell death in control cultures when compared with neurons transfected with a scrambled vector (Fig. 7d), in line with previous findings demonstrating that MCL-1 is necessary to maintain survival in differentiated neurons (Germain et al. 2011). In addition, a significant increase in cell death was evident in NMDA-exposed neurons transfected with shRNA targeting MCL-1 when compared with scrambled vector transfected neurons. Of note, AICAR preconditioning failed to provide any protection in NMDA-exposed neurons transfected with MCL-1 shRNA when compared with non-preconditioned neurons transfected with either MCL-1 or scrambled shRNA (Fig. 7d). As expected, AICAR preconditioned neurons transfected with scrambled shRNA showed significant neuroprotection compared with non-preconditioned neurons. Collectively, these data suggest that MCL-1 is neuroprotective, and that MCL-1 gene silencing is dominant over the protective effect of AICAR preconditioning.

AICAR preconditioning and MCL-1 over-expression reduce cytosolic Ca2+ overloading during acute NMDA excitation

Interestingly, both AMPK activation as well as MCL-1 over-expression has been shown to improve mitochondrial bioenergetics (Dasgupta and Milbrandt 2007; Yu and Yang 2010; Perciavalle et al. 2012). As NMDA-induced Ca2+ overloading is critically dependent on the bioenergetic ability of neurons to restore ion homeostasis (Budd and Nicholls 1998; Ward et al. 2000, 2007; Weisova et al. 2009), we addressed the question whether AICAR preconditioning or MCL-1 over-expression is able to reduce NMDA-induced Ca2+ elevations. To address this question, we performed time-lapse confocal imaging in cortical neurons co-loaded with Fluo-4 AM and TMRM to determine the individual Ca2+ and mitochondrial membrane potential responses to an acute NMDA challenge. Analysis of neurons transfected with a plasmid over-expressing MCL-1 revealed significantly reduced cytosolic Ca2+ levels in response to NMDA compared with control transfected cells (Fig. 8a–c). Quantification of individual Ca2+ responses revealed a marked decrease in cytosolic Ca2+ influx in neurons with MCL-1 over-expression in comparison with control transfected neurons (Fig. 8c). Although MCL-1 over-expressing cells had a higher baseline TMRM fluorescence (Fig. 8d), no significant difference in the absolute depolarisation of ∆ψm was evident during NMDA excitation in neurons over-expressing MCL-1 compared with control-transfected neurons (Fig. 8e). Similar to MCL-1 over-expression, AICAR preconditioning significantly decreased cytosolic Ca2+ levels in response to NMDA excitation. However, there was no significant difference observed between AICAR-preconditioned neurons and neurons over-expressing MCL-1 or AICAR-preconditioned neurons over-expressing MCL-1 (Fig. 8c). The reduction in cytosolic Ca2+ levels in response to NMDA excitation could be attributed to decreased NMDA receptor expression. However, we found no significant reduction in expression levels of the NMDA receptor subunit 2A and 2B in AICAR preconditioned neurons (Figure S2).

Figure 8.

Myeloid cell leukaemia sequence 1 (MCL-1) inhibits neuronal Ca2+ overloading during NMDA excitation. (a, b) Representative traces from cortical neurons transfected with a MCL-1 over-expressing vector or control empty vector for 48 h. Neurons were loaded with tetramethylrhodamine methyl ester (TMRM) and Fluo-4 AM for 30 min at 37°C before being mounted on the stage of LSM710 and treated with 100 μM NMDA for 5 min. (c) Quantification of Fluo-4 fluorescence (area under the curve) in neurons transfected with MCL-1 over-expressing or empty vector. Neurons were pre-treated with 5-aminoimidazole-4-carboxamide riboside (AICAR) or vehicle. Data were obtained from six separate experiments from three different platings (n = 19–25 neurons per treatment quantified). *p ≤ 0.05, compared with empty vector transfected neurons treated with NMDA only. NS, no significance. (d) Quantification of baseline TMRM fluorescence in MCL-1 over-expressing or empty vector-transfected neurons. *p ≤ 0.05, compared with empty vector transfected neurons. Two sample t-test assuming equal variances was used to assess the statistical significance. (e) Quantification of NMDA-induced depolarization by TMRM fluorescence measurements in neurons transfected with MCL-1 or empty vector. NS, no significance, two sample t-test assuming equal variances was used to assess the statistical significance.

To determine the role of endogenous MCL-1 in mediating the effects of AICAR on NMDA-induced Ca2+ overloading, we transfected neurons with scramble shRNA or shRNA targeting MCL-1. AICAR-preconditioned neurons transfected with scramble shRNA and exposed to NMDA showed a significant decrease in cytosolic Ca2+ influx compared with scramble shRNA-transfected, NMDA-exposed controls (Fig. 9a). Of note, gene silencing of MCL-1 abolished the effect of AICAR on NMDA-induced Ca2+ increases (Fig. 9a). Interestingly, no significant differences in cytosolic Ca2+ responses were observed between scramble and MCL-1 shRNA-transfected, NMDA-exposed neurons, suggesting that MCL-1 requires AMPK activation to mediate its activity (Fig. 9a). We also observed no significant differences between scramble and MCL-1 shRNA transfected neurons in the baseline TMRM fluorescence or in the absolute depolarization of ∆ψm during NMDA excitation (Fig. 9b and c). Collectively, these data suggest that MCL-1 mediates the effect of AICAR preconditioning on neuronal Ca2+ homeostasis in response to NMDA excitation.

Figure 9.

Myeloid cell leukaemia sequence 1 (MCL-1) gene silencing abolishes the effect of 5-aminoimidazole-4-carboxamide riboside (AICAR) pre-treatment on NMDA-induced neuronal Ca2+ overloading. (a) Quantification of Fluo-4 fluorescence (area under the curve) in neurons transfected with shRNA targeting MCL-1 or Scramble shRNA. Neurons were pre-treated with AICAR or vehicle. Data were obtained from three separate experiments from two different platings (n = 9–15 neurons per treatment quantified). *p ≤ 0.05, compared with scramble shRNA transfected neurons treated with NMDA only. #p ≤ 0.05, compared with AICAR-preconditioned neurons transfected with shRNA targeting MCL-1. (b) Quantification of baseline tetramethylrhodamine methyl ester (TMRM) fluorescence in shRNA targeting MCL-1 or scramble shRNA transfected neurons. NS, no significance. Two sample t-test assuming equal variances was used to assess the statistical significance. (c) Quantification of NMDA-induced depolarization by TMRM fluorescence measurements in neurons transfected with shRNA targeting MCL-1 or scramble shRNA. NS, no significance, two sample t-test assuming equal variances was used to assess the statistical significance.

Discussion

In this study we describe a potent preconditioning effect resulting from a transient, 2-h exposure to the AMPK activator, AICAR, in a model of NMDA excitotoxicity in mouse cortical neurons. Interestingly, the preconditioning effect of the transient AICAR exposure was maintained for up to 24 h after washout of AICAR, suggesting that AICAR preconditioning likely induced long-lasting alterations in neurons. In line with this observation, we demonstrate that AICAR preconditioning was associated with increased MCL-1 mRNA and protein levels. Furthermore, we demonstrate that MCL-1 exerts potent protective effects against NMDA-induced Ca2+ overloading and cell death, and that MCL-1 modulation is dominant over the effect of AICAR preconditioning. These experiments therefore identify the anti-apoptotic protein MCL-1 as a key effector of AMPK-mediated preconditioning in neurons.

Preconditioning has been extensively investigated in both in vitro and in vivo models of neuronal injury, and has been shown to protect neurons against excitotoxic, ischaemic and oxidative injury (Kitagawa et al. 1990; Chen et al. 1996; Grabb and Choi 1999; Navon et al. 2012). Several factors have been associated with the protective effect of preconditioning: these include the activation of potassium channels (Smith et al. 2009), maintenance of intracellular Ca2+ homeostasis (Bickler et al. 2010), preservation of mitochondrial membrane integrity, membrane potential, membrane fluidity as well as prevention of mitochondrial swelling (Zhang et al. 2003), activation of cyclic AMP response element-binding protein (Meller et al. 2005) and reduced activation of caspase 3, poly (ADP-ribose) polymerase, or nitric oxide synthase (Gidday et al. 1999; Garnier et al. 2003; McLaughlin et al. 2003). We have recently identified a key role for the ancient energy sensor AMPK during preconditioning in neurons (Weisova et al. 2012). AMPK is expressed at high levels in neurons of the central nervous system (Turnley et al. 1999), and has previously been described to have potent neuroprotective effects (Culmsee et al. 2001). While AMPK is known to have numerous acute effects on mitochondrial function and cellular bioenergetics (Bergeron et al. 2001; Zong et al. 2002; Ojuka 2004), key effectors of AMPK that may modulate long-term alterations in neurons have been less frequently described. We here identify the anti-apoptotic protein MCL-1 as a potential effector of AMPK that mediates some of AMPK's preconditioning activities. Because MCL-1 mRNA and protein were found to be up-regulated up to 6 h after washout of AICAR, but neuroprotection was evident for up to 24 h after washout, other long-term alterations may also contribute to AMPK-induced preconditioning. In this context, AMPK has previously been shown to protect endothelial cells from hypoxia and oxygen glucose deprivation by up-regulating the anti-apoptotic proteins Bcl-2 and survivin (Liu et al. 2010). Our finding that MCL-1 is up-regulated at the mRNA and protein level is interesting as MCL-1 is subject to rapid protein turnover because of proteasomal degradation (Yang et al. 1995; Akgul et al. 2000; Nijhawan et al. 2003), and considering that AMPK activation inhibits cap-dependent protein translation (Kimura et al. 2003; Liu et al. 2006). The mechanisms responsible for the MCL-1 mRNA up-regulation in response to AICAR therefore warrant further investigation, and may include effects of AMPK on mRNA stability or proteasomal degradation as shown in other systems previously (Yun et al. 2005; Viana et al. 2008).

MCL-1 has previously been suggested to be a key regulator of neuronal survival by inhibiting apoptosis and by inhibiting a potentially toxic overactivation of autophagy (Germain et al. 2011). MCL-1 has also been shown to be essential for neuronal development, as deletion of MCL-1 induced apoptosis in neuronal progenitors and newly committed neurons as they commence their migration away from the ventricular zone (Arbour et al. 2008). Recently, MCL-1 has been shown to regulate the survival of adult neuronal precursors cells, with conditional deletion or over-expression of MCL-1 increasing or decreasing neuronal apoptosis by two fold respectively (Malone et al. 2012). However, the effect of MCL-1 may extend beyond the regulation of the mitochondrial apoptosis pathway. Indeed, a recent study has demonstrated that MCL-1 exists in two forms that regulate apparently different aspects of mitochondrial function: MCL-1OM (40 and 38 kDa) and MCL-1Matrix (36 kDa). MCL-1OM resides on the mitochondrial outer membrane and antagonizes pro-apoptotic proteins to inhibit mitochondrial outer membrane permeabilization. In contrast, MCL-1Matrix localizes to the matrix and maintains normal inner mitochondrial membrane structure, and increases mitochondrial membrane potential and bioenergetics (Perciavalle et al. 2012) Of note, we noted a significant increase in the protein levels of both MCL-1 variants in response to AICAR preconditioning, suggesting that these are jointly up-regulated by AMPK. The finding that AMPK also up-regulated the protein levels of MCL-1Matrix is particularly interesting considering the ancient function of AMPK in the maintenance of cellular bioenergetics. Improved mitochondrial function and bioenergetics are known to facilitate the extrusion of Ca2+ during Ca2+ overloading, by activating plasma membrane Ca2+-ATPases and Na+/Ca2+ exchangers (Kiedrowski and Costa 1995; White and Reynolds 1995). We observed significantly reduced cytosolic Ca2+ levels in response to NMDA in MCL-1 over-expressing and AICAR-preconditioned cells. It is therefore likely that neuroprotection induced by either MCL-1 over-expression or AICAR preconditioning is largely because of a significantly reduced neuronal Ca2+ overload during the initial period of NMDA excitation. Our finding that MCL-1 over-expression itself reduces cytosolic Ca2+ overloading may further indicate that MCL-1 is indeed an evolutionary conserved regulator of mitochondrial physiology, similar to other BCL-2 family proteins and apoptosis-associated proteins for which such ‘day-time’ roles have been increasingly described (Kilbride and Prehn 2012).

Acknowledgements

This study was supported through the National Biophotonics and Imaging Platform, Ireland, funded by the Irish Government's programme for research in Third level Institutions, Cycle 4, and through a grant from Science Foundation Ireland (08/IN.1/B1949) to J.H.M.P. The authors declare no conflict of interest.

Authorship credits

U.A., P.W. and J.H.M.P. designed research; U.A., P.W., C.G.C. and H.G.K. performed research; U.A. H.D. and J.H.M.P. analysed data; U.A. and J.H.M.P. wrote the article.

Ancillary