Using a fluorescent viability assay, immunocytochemistry, patch-clamp recordings, and Ca2+ imaging analysis, we report that ouabain, a specific ligand of the Na+,K+-ATPase cardiac glycoside binding site, can prevent glutamate receptor agonist-induced apoptosis in cultured rat cortical neurons. In our model of excitotoxicity, a 240-min exposure to 30 μM N-methyl-d-aspartate (NMDA) or kainate caused apoptosis in ∼50% of neurons. These effects were accompanied by a significant decrease in the number of neurons that were immunopositive for the antiapoptotic peptide Bcl-2. Apoptotic injury was completely prevented when the agonists were applied together with 0.1 or 1 nM ouabain, resulting in a greater survival of neurons, and the percentage of neurons expressing Bcl-2 remained similar to those obtained without agonist treatments. In addition, subnanomolar concentrations of ouabain prevented the increase of spontaneous excitatory postsynaptic current's frequency and the intracellular Ca2+ overload induced by excitotoxic insults. Loading neurons with 1,2-bis(2-aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid or inhibition of the plasma membrane Na+,Ca2+-exchanger by 2-(2-(4-(4-nitrobenzyloxy)phenyl)ethyl)isothiourea methanesulfonate (KB-R7943) eliminated ouabain's effects on NMDA- or kainite-evoked enhancement of spontaneous synaptic activity. Our data suggest that during excitotoxic insults ouabain accelerates Ca2+ extrusion from neurons via the Na+,Ca2+ exchanger. Because intracellular Ca2+ accumulation caused by the activation of glutamate receptors and boosted synaptic activity represents a key factor in triggering neuronal apoptosis, up-regulation of Ca2+ extrusion abolishes its development. These antiapoptotic effects are independent of Na+,K+-ATPase ion transport function and are initiated by concentrations of ouabain that are within the range of an endogenous analog, suggesting a novel functional role for Na+,K+-ATPase in neuroprotection.
Ionotropic glutamate receptors (GluRs) are critically involved in physiological processes in the mammalian central nervous system (CNS), including generation of neuronal activity patterns (Iwasato et al., 2000) and learning and memory (Bliss and Collingridge, 1993; Tang et al., 1999). Functional deregulation of neuronal metabolism resulting from an overactivation of GluRs leads to neuronal death and underlies a variety of CNS disorders including stroke, neurodegenerative diseases, and spinal cord and brain injuries (Choi, 1988; Olney, 1994; Lipton, 1999). The prolonged presence of glutamate released from neurons and glial cells by nonquantal secretion (Rossi et al., 2000) and activation of GluRs have extensive consequences for neuron functioning. These consequences start with intracellular Ca2+ overload, imbalance of transmembrane ion gradients, and activation of various intracellular cascades and end with the destruction of the plasma membrane or nuclear apparatus of neurons (Choi, 1987, 1988; Olney, 1994; Green and Reed, 1998; Kidd, 1998). Massive cytoplasmic Ca2+ accumulation is thought to be one of the most important triggers of various cell death mechanisms, usually ending as apoptosis (Khodorov, 2004). Apoptosis, or programmed cell death, plays an enormous role in the development and formation of organs, as well as in the functioning of rapidly renewing tissues (Johnston, 1994) and is the key factor in neuronal pathogenesis and necrosis (Choi, 1988; Olney, 1994; Lipton, 1999; Khodorov, 2004).
Similar neuronal dysfunction and neurodegeneration are induced by micromolar concentrations of ouabain, a specific inhibitor of Na+,K+-ATPase (Xiao et al., 2002). Na+,K+-ATPase sets the cellular ion gradients for K+ and Na+ by active transport, thereby providing the driving force for membrane excitability and many other transporters and exchangers. It is now recognized that Na+,K+-ATPase, in addition to its primary role as an ion transporter, can function as a receptor signaling molecule. The extracellular loops of the catalytic α-subunit of Na+,K+-ATPase form a binding site that is the only known, highly specific receptor for ouabain and other cardiotonic steroids (Ogawa et al., 2009; Lingrel, 2010) and their circulating endogenous analogs (Blaustein, 1993; Schoner and Scheiner-Bobis, 2007; Bagrov and Shapiro, 2008). Upon binding of ouabain to this receptor, Na+,K+-ATPase interacts with neighboring membrane proteins to affect diverse cell functions such as protein synthesis, proliferation, cell differentiation, gene expression, regulation of intracellular Ca2+, contractile properties, synaptic efficacy, and neural differentiation (Xie and Askari, 2002; Krivoi et al., 2006; Aperia, 2007; Hazelwood et al., 2008; Desfrere et al., 2009; Li and Xie, 2009; Radzyukevich et al., 2009; Rose et al., 2009; Heiny et al., 2010). In addition, ouabain in nanomolar concentrations stimulates the reversed mode of plasma membrane Na+,Ca2+ exchange in snail neurons (Saghian et al., 1996).
Whereas ouabain, as an endogenous agent, has been found in subnanomolar concentrations in rat blood plasma and cerebrospinal fluid (Blaustein, 1993; Dobretsov and Stimers, 2005; Schoner and Scheiner-Bobis, 2007; Bagrov and Shapiro, 2008), its functional relevance for CNS is not clearly understood. Previously, antiapoptotic action of low ouabain doses was described when kainic acid (KA) and ouabain were injected into the rat brain in vivo (Golden and Martin, 2006). Here, we investigate the pharmacological effects of ouabain in a wide range of concentrations (0.01 nM to 30 μM) in vitro on cortical neurons in primary culture under normal conditions and excitotoxic stress. Neuronal viability is also tested in the presence of digoxin, another highly specific ligand of Na+,K+-ATPase (Katz et al., 2010). We examine the possibility that Na+,K+-ATPase can interact with the signaling pathways involved in neuronal injury triggered by GluR (NMDAR, AMPAR, and KAR subtypes) overactivation and, moreover, may antagonize neurodegeneration. Using a fluorescent viability assay (FVA) with confocal microscopy and immunocytochemistry we demonstrate that ouabain at subnanomolar concentrations prevents rat cortical neuron apoptosis during GluR agonist-induced stress. Based on our patch-clamp and Ca2+ imaging experiments, this neuroprotective antiapoptotic effect of ouabain results from the up-regulation of Ca2+ extrusion mechanisms and involves the plasma membrane Na+,Ca2+ exchanger (NCX).
Materials and Methods
Preparation and Solutions.
Cell cultures were prepared as described previously (Antonov et al., 1998; Mironova et al., 2007). All procedures using animals were in accordance with recommendations of the Federation for Laboratory Animal Science Associations and approved by the local Institutional Animal Care and Use Committees. Wistar rats (provided by the Sechenov Institute's Animal Facility) 16 days pregnant (26 animals in this study) were sacrificed by CO2 inhalation. Fetuses (10–15) were removed, and their cerebral cortices were isolated, enzymatically dissociated, and used to prepare primary neuronal cultures. Cells were used for experiments after 7 to 15 days in culture (Mironova et al., 2007; Han and Stevens, 2009). Experiments were performed at room temperature (20–23°C).
Neuronal cultures were perfused by the indicated concentrations of drugs dissolved in the bathing solutions. The principal external bathing solution consisted of 140 mM NaCl, 2.8 mM KCl, 1.0 mM CaCl2, 1.0 mM MgCl2, and 10 mM HEPES. The content of the external solution slightly varied depending on the purpose of the experiments. In experiments with N-methyl-d-aspartate (NMDA), Mg2+ was omitted from the bathing solution, because it blocks the channels of NMDARs (Nowak et al., 1984). The pH of each external solution was adjusted to 7.4 with NaOH.
Recordings of integral cellular currents were done by using whole-cell configurations of the patch-clamp technique. In most of the experiments, pipettes were filled with a solution containing 9 mM NaCl, 17.5 mM KCl, 121.5 mM K-gluconate, 1 mM MgSO4, 10 mM HEPES, 0.2 mM EGTA, 2 mM MgATP, and 0.5 mM NaGTP (Han and Stevens, 2009). In some ramp experiments 0.5 μM tetrodotoxin was added to the external solution, and Cs+ intracellular pipette solution was used to increase noise resolution and stability of whole-cell recording within a wide range of membrane voltage by blocking K+ channels. Substitution of Cs+ for K+ in pipette solution does not affect either the kinetics or conductance of GluR channels (Antonov et al., 1995, 1998; Traynelis et al., 2010). The Cs+ pipette solution contained 120 mM CsF, 10 mM CsCl, 10 mM EGTA, and 10 mM HEPES. The pH was adjusted to 7.4.
Directly before the experiments, coverslips with neuronal cultures were placed in the bathing solution. To trigger necrotic cell injury and apoptosis 30 μM NMDA (30 μM glycine was present as a coagonist of NMDAR; Johnson and Ascher, 1987) or 30 μM KA was added to the bathing solution. When the effects of ouabain or digoxin were studied, the compounds were applied alone or in combination with the glutamate receptor agonists. Neuronal cultures were subjected to FVA and immunocytochemistry after 240 min of continuous presence of the compounds. In some experiments the effects of longer treatments (360 min) were studied.
Quantification of Cell Viability.
In FVA cell viability (Mironova et al., 2007) was determined from fluorescent images by quantification of fluorescence after staining all nuclei with acridine orange (AO) and staining dead cell nuclei with ethidium bromide (EB). First, cells were treated with 0.001% AO for 30 s. After complete washout of contaminating AO, cells were exposed to 0.002% EB for 30 s. This procedure was applied directly before every measurement. AO readily permeates the plasma membrane, stains the nuclei of neurons, and emits a green fluorescence in live neurons; however, it undergoes a red emission shift in apoptotic nuclei (orange spectrum) (Zelenin, 1966). EB (red fluorescence) is normally impermeable and enters the nuclei of necrotic cells when the plasma membrane is compromised. As a result, in fluorescent images the nuclei of live neurons, labeled with AO, looked green and the nuclei of injured neurons, labeled with EB, looked red. In the absence of colocalized pixels cell viability was estimated by the ratio of green pixels (the number) to the total number of fluorescent pixels (red plus green). If some neurons exhibited apoptotic transformations their nuclei looked yellow-orange, revealing colocalization of fluorescence in the green and red spectral regions. In this case, the fractions of live, apoptotic, and necrotic cells were calculated on the basis of correlation plot as the ratio of green, yellow-orange, and red pixels to the total number of fluorescent pixels (the sum of green, yellow-orange, and red), correspondingly.
The antiapoptotic peptide Bcl-2 was visualized immunochemically by using rabbit polyclonal antibodies and secondary antibodies conjugated with R-phycoerythrin on fixed cultures. Cells were fixed with 4% paraformaldehyde solution in phosphate-buffered saline (PBS) for 30 min. After fixation, cells were washed twice with PBS (15 min × 2). Before treatment with 2% bovine serum albumin, cells were incubated with 0.2% Triton X-100 for 15 min, washed with PBS, and exposed to primary antibodies for 12 h at 4°C. After washing to remove primary antibodies, fluorochrome-conjugated secondary antibodies were added. Reactions with secondary antibodies lasted for 40 min at room temperature (23°C). Before data were recorded, coverslips with antibody-bound preparations were pasted on slides with Moviol glue to prevent the fading of fluorochromes.
Imaging and Image Processing.
Fluorescence images were captured by using a Leica SP5 MF scanning confocal microscope (inverted) (Leica Microsystems, Inc., Bannockburn, IL). Cultures were viewed with 20× (HCX APO CS 20×/0.70; Leica Microsystems, Inc.) or 63× (HCX APO CS 63×/1.4; Leica Microsystems, Inc.) immersion objectives. To resolve fine details an additional electronic zoom with a factor of 1.5 to 3.5 was used. Fluorochromes were excited with a 488-nm laser line. Images were captured in the green and red parts of the spectrum (for the FVA) and the red spectral region (for phycoerythrin). To improve signal-to-noise ratio six scans (512 × 512 pixel array) were averaged at each optical section. Some areas of neuronal culture contained glia, forming a thin, flat layer subjacent to neuronal bodies bulging out above. To exclude glia from images, we set confocal optical section (for 63× objective the thickness of optical section was 0.4 μm) to cross neuronal bodies only. Transmitted light images were also captured to verify the correct position of the focal plane by using morphological criteria.
For two-channel imaging of AO and EB emission, the emitted fluorescence was acquired at 500 to 560 nm (green region of spectrum for AO) and >600 nm (red region of spectrum for EB) and collected simultaneously by using separate photo multiplier tubes. Microscope settings were adjusted so that imaging conditions for both channels were kept equal and constant. The confocal images from both channels were merged by using standard Leica Microsystems, Inc. software and ImageJ software (http://rsbweb.nih.gov/ij/). Images were processed in ImageJ software by using a custom-written plug-in (http://sibarov.ru/index.php?slab = software). The noise threshold value was determined automatically by using the ISODATA algorithm implemented in ImageJ. To quantify colocalized and noncolocalized fluorescence correlation plots, which sort values of given pixels in the first image as the x-coordinate and values of corresponding pixels in the second image as the y-coordinate, images were generated for each measurement. On the resulting image noncolocalized pixels looked green and red and were attributed to live and necrotic neurons, respectively. Pixels with colocalized green and red fluorescence looked yellow-orange and were attributed to the nuclei of apoptotic neurons.
Patch-clamp recordings were carried out on the stage of an inverted microscope (Diaphot TMD; Nikon, Tokyo, Japan) with Hoffman modulation contrast optics. Patch pipettes (2–4 MΩ) were pulled from 1.5-mm (outer diameter) borosilicate standard wall capillaries with inner filament (Sutter Instrument Company, Novato, CA). Recordings were made by using a MultiClamp 700B amplifier (Molecular Devices, Sunnyvale, CA). Whole-cell currents were recorded at membrane voltage of −70 mV during bath perfusion of 30 μM NMDA + 30 μM glycine or 30 μM KA applied by using a multibarrel perfusion system. Continuous recordings of spontaneous excitatory postsynaptic currents (sEPSCs) were low-pass filtered at 1.5 kHz and digitized at 20 kHz with DigiData 1440A and pClamp 10 software (Molecular Devices). To analyze current-voltage (I-V) relationships, 2-s-long, voltage ramp (from −100 to + 30 mV) protocols were applied at the steady-state currents under control conditions and in the presence of GluR agonists. The I-V relationships of NMDA- or KA-induced conductance were then obtained by subtraction of the control ramp current from those in the presence of GluR agonists. Ramp recordings were low-pass filtered at 100 Hz and digitized at 1000 Hz.
Loading of AM Esters and Ca2+ Imaging.
Cells were loaded with BAPTA AM (2 and 5 μM), Fluo-3 AM (4 μM), and Fura-2 AM (10 and 20 μM) by using conventional protocols. In brief, neuronal cultures were incubated with the AM esters, and 0.02% Pluronic F-127 was added to the external solution for 45 min in the dark at 20 to 23°C. Then, the AM esters were washed out, and cells were incubated in the external solution for another 30 min. For Fluo-3 experiments coverslips with loaded cultures were placed in the perfusion chamber that was mounted on the stage of a Leica SP5 MF inverted microscope. Fluorescence was activated with 488-nm laser light. Images were captured every minute during 60-min experiments. For Fura-2 the perfusion chamber with neuronal culture was mounted on the stage of a Nikon TMS inverted epifluorescence microscope equipped with a 300-W xenon lamp (Intracellular Imaging Inc., Cincinnati, OH) and 30× dry objective (Nikon). Cells were visualized with a high-resolution digital black/white charge-coupled device camera (Cohu 4910; Cohu Electronics, Poway, CA), and [Ca2+]i was estimated by the 340/380 ration method, using a Kd value of 315 nM for 23°C. Data were analyzed with InCytIm2 (Intracellular Imaging and Photometry System, Cincinnati, OH) and Excel (Microsoft, Redmond, WA).
Rabbit polyclonal antibodies for Bcl-2 and secondary antibodies conjugated with R-phycoerythrin were purchased from Abcam Inc. (Cambridge, MA). Fura-2 AM was from Fluka (Buchs, Switzerland), Fluo-3 AM was from MoBiTec (Göttingen, Germany), and Pluronic F-127 was from Molecular Probes (Carlsbad, CA). Other compounds were from Sigma-Aldrich (St. Louis, MO). Stock solutions of 10 mM NMDA, 10 mM glycine, and 10 mM KA dissolved in distilled water were stored frozen and thawed on the day of use. Stock solutions with ouabain dissolved in distilled water and digoxin dissolved in ethanol at concentrations of 10 μM and 1 mM, respectively, were stored refrigerated. Stock solutions of 10 mM BAPTA AM, 10 mM Fura-2 AM, and 10 mM Fluo-3 AM dissolved in dimethyl sulfoxide were stored frozen. All drugs were diluted in the external solution to the indicated concentrations before use.
Quantitative data are expressed as means ± S.E.M. Student's two-tailed t tests, ANOVA, Tukey's, and Bonferroni multiple comparison methods were used for statistical analysis. Number of experiments is indicated by n throughout. The data were considered as significantly different based on a confidence level of 0.05.
Neuroprotective, Antiapoptotic Effects of Subnanomolar Ouabain in Excitotoxic Insults.
Experiments were performed on cultured cortical neurons by using a microscopy-based viability assay that rapidly detects and counts the proportion of live, necrotic, and apoptotic neurons (FVA) (Mironova et al., 2007). The neurons were dual-stained with acridine orange and ethidium bromide. A correlation plot of the intensity of green versus red fluorescence revealed the proportion of live, apoptotic, and necrotic cells.
Under control conditions, when neurons were perfused with the bathing solution, the majority of nuclei remained green (viable) for up to 240 min (Fig. 1A). The image correlation analysis (Fig. 1B) showed that the majority of cells mapped in the region of strong green fluorescence (live) and did not colocalize with the few cells showing low red fluorescence. This indicates that the majority of cells were vital with an intact plasma membrane. Only a small number of nuclei showed a shift in the acridine orange fluorescence to the red spectral region, indicating that these cells were starting to undergo apoptosis. It is noteworthy that viable cortical neurons showed no response to nanomolar concentrations of ouabain (Fig. 1, C and D). Treatment with 1 nM ouabain for 240 min did not induce any notable change in neuronal viability, as indicated both from the correlation analysis (Fig. 1C) and quantitative comparisons of the number of cells in each state (Fig. 1D). Therefore, in contrast to the effects of ouabain at concentrations exceeding 1 μM, which causes large changes in ion balance and is known to be neurotoxic (Xiao et al., 2002), cultured cortical neurons were indifferent to the action of 1 nM ouabain.
Sustained exposure of neurons to GluR agonists is a neurotoxic insult that triggers necrosis and apoptosis. When neurons were exposed to 30 μM NMDA for 240 min (30 μM NMDA was always applied in combination with 30 μM glycine, which is a coagonist of NMDARs; Johnson and Ascher, 1987), the treated neurons showed significantly fewer live cells (green) and more condensed apoptotic nuclei (orange) and necrotic (red) nuclei (Fig. 2A). A similar pattern of neurotoxicity occurred when neurons were exposed to 30 μM KA for 240 min (Fig. 2B). Therefore, cells exposed to chronic agonist exhibited significant apoptosis and necrosis, as expected. The green-red emission was highly colocalized (Fig. 2, C and D), indicating that apoptosis is the dominant mechanism of cell death in these neurons. Sustained exposure of neurons to either NMDA or KA caused a significant decrease in the number of viable neurons and a concomitant increase in the number of neurons undergoing cell apoptosis and cell death. These findings indicate that apoptosis plays a major role in the neurotoxicity produced by sustained exposure to GluR agonists.
Strikingly, this neurotoxicity can be prevented by including ouabain at 0.1 or 1 nM with the GluR agonists (Fig. 2, E and F). Neurons incubated with 30 μM NMDA and 0.1 nM ouabain showed a viability comparable with the control conditions (Fig. 2E), without significant apoptosis. Even longer treatment with 30 μM NMDA and 1 nM ouabain exceeding 360 min did not reveal the loss of neuronal viability. For instance, the obtained percentages of live, apoptotic, and necrotic neurons were 94 ± 1, 2 ± 1, and 4 ± 1% (n = 9), respectively, and were not significantly different from the control values (p > 0.05; ANOVA, post hoc Bonferroni test). A similar protective effect occurred when 30 μM KA and 0.1 nM ouabain were simultaneously applied to neurons (Fig. 2F). The corresponding correlation plots (Fig. 2, G and H) show almost no colocalized fluorescence in the green and red spectral regions, indicating that the viable cells were not undergoing apoptosis.
From the average data for NMDA (Fig. 2I) and KA (Fig. 2J), it is apparent that long-term activation of both NMDARs and AMPAR/KARs induces excitotoxicity, in which the dominant mechanism of cell rundown is apoptosis. However, the inclusion of 0.1 or 1 nM ouabain with GluR agonists significantly decreased the quantity of apoptotic neurons and nearly doubled the number of viable neurons. The number of necrotic cells remained small and unchanged in both conditions and was comparable with values obtained under the control conditions (Fig. 1D). It should be noted that 0.01 nM ouabain was ineffective at preventing apoptosis. This result suggests that ouabain at subnanomolar concentrations (0.1–1 nM) actually increases cell viability during excitotoxic insult, by preventing the development of apoptosis.
To verify that the observed ouabain effects were mediated by Na+,K+-ATPase, digoxin, another highly specific ligand of the Na+,K+-ATPase cardiotonic steroids binding site (Katz et al., 2010), was also tested by using FVA. The incubation of neurons with 30 μM NMDA and 1 nM digoxin for 240 min did not cause apoptosis. The average proportions of neurons found in these experiments were 73 ± 7, 18 ± 3, and 9 ± 6% (n = 9) for live, apoptotic, and necrotic cells, respectively. These data did not differ significantly from the values obtained in experiments with 30 μM NMDA and 1 nM ouabain (p > 0.05; ANOVA, post hoc Bonferroni test). Therefore, 1 nM digoxin had similar antiapoptotic effects as 1 nM ouabain, suggesting that high-affinity binding of cardiotonic steroids to the Na+,K+-ATPase was indispensible for their antiapoptotic action.
Apoptosis induced by the activation of GluRs develops largely because of mitochondrial dysfunction (Green and Reed, 1998), although NMDARs and AMPAR/KARs trigger different apoptotic cascades (Wang et al., 2004). The endogenous antiapoptotic protein Bcl-2 is an important regulator of mitochondrial function and energy metabolism and is involved in many vital cell processes. A decrease in Bcl-2 levels is routinely used as a marker of apoptosis (Adams and Cory, 1998). Because ouabain in our experiments selectively inhibited apoptosis, we further tested whether the expression level of Bcl-2 changed during NMDA- or KA- induced neurodegeneration. In control conditions, the majority of cortical neurons showed a high level of Bcl-2 immunostaining (Fig. 3, A and B). Subjecting these neurons to excitotoxic stress produced by 30 μM KA (Fig. 3C) or 30 μM NMDA (Fig. 3D) caused a significant decrease in Bcl-2 levels. However, coincubation of the GluR agonists with 0.1 or 1 nM ouabain completely prevented this effect (Fig. 3, E and F). The quantity of Bcl-2-positive neurons obtained after NMDA or KA alone was significantly lower than control values (Fig. 3G), but was retained at near control values after combined application of NMDA or KA with ouabain (Fig. 3H). As in the case of FVA, experiments with 0.01 nM ouabain did not cause any change in the expression of Bcl-2. Thus, the measurement of Bcl-2 levels agreed well with the results obtained with FVA.
Concentration Dependence of Self-Neurotoxic Effect of Ouabain.
We also tested the neurotoxic effects of ouabain in a wide range of concentrations from 1 nM to 30 μM by using FVA. At 1 nM and below ouabain did not induce any changes in cell viability (Figs. 1 and 4). Sustained exposure (240 min) of neurons to 10 nM ouabain and above caused a significant decrease in the number of viable neurons (Fig. 4). In contrast to the effects of GluR agonists, the ouabain-induced cell rundown developed basically by necrosis, but not apoptosis. Our data are consistent with previous observations that neurons express the α3-isoform of Na+,K+-ATPase whose enzymatic activity is already affected by 10 nM ouabain (Richards et al., 2007) and the mechanism of neuronal death corresponds to necrosis (Xiao et al., 2002).
Regulation of Intracellular Ca2+ Concentration by Ouabain.
It is generally accepted that chronic neuronal depolarization followed by the free cytosolic Ca2+ concentration ([Ca2+]i) increase initiates neuronal apoptosis during excitotoxic stress (Choi, 1988; Olney, 1994; Lipton, 1999; Khodorov, 2004). To investigate the possible mechanisms of neuroprotective, antiapoptotic action of subnanomolar ouabain concentrations, we undertook experiments in which whole-cell currents were recorded during long-lasting 30 μM NMDA or 30 μM KA applications and further studied the effects of subnanomolar ouabain by directly measuring intracellular Ca2+ signals using Fluo-3 or Fura-2. Both agonists caused the generation by neurons of inward currents that reveal a steady state (direct currents) as long as agonists were present (Fig. 5). Applications of 30 μM NMDA to neurons loaded with Fluo-3 (Fig. 6A) or Fura-2 (Fig. 6B) induced sustained intracellular Ca2+ responses lasting as long as NMDA was present. On average somatic free [Ca2+]i under the control condition did not exceed 50 nM, whereas in the presence of NMDA it could reach 1 μM in our experiments. When 30 μM NMDA and 1 nM ouabain were applied simultaneously in experiments with either Fluo-3 (Fig. 6C) or Fura-2 (Fig. 6D) the Ca2+ response increased, but then declined in time gradually to almost the control value of [Ca2+]i. The dynamics of NMDA-induced [Ca2+]i increase in the absence and presence of 1 nM ouabain are compared in Fig. 6E. Obviously, 1 nM ouabain somehow stimulated intracellular Ca2+ buffering or extrusion of Ca2+ from neurons, resulting in decreased [Ca2+]i. Perhaps, this process prevents an induction of apoptosis during excitoxic insults.
Thus, Na+,K+-ATPase is involved in the regulation of [Ca2+]i during excitotoxic stress, suggesting that Ca2+ handling determines its neuroprotective antiapoptotic function.
Na+,K+-ATPase Is Involved in the Regulation of Spontaneous Synaptic Activity through Intracellular Ca2+-Dependent Mechanisms.
In addition to the direct currents, GluR agonists induced considerable increases in sEPSC frequency, which were stably maintained at high levels in the presence of agonists (Figs. 5 and 7A). We were surprised to find that, when applied on top of 30 μM NMDA or 30 μM KA effects, 1 nM ouabain affected sEPSC with a delay of approximately 1 min (Fig. 7A), so the value of the sEPSC frequency obtained either in NMDA (approximately 5 per s; Fig. 7B) or KA (approximately 1.2 per s; Fig. 7C) decreased to the value obtained under control conditions (approximately 0.3 per s; Fig. 7, B and C). Neither direct current amplitudes at −70 mV, nor I-V relationships of NMDA-activated (Fig. 7D) and KA-activated (Fig. 7E) integral currents were affected by 1 nM ouabain, suggesting the lack of its effect on NMDAR and AMPAR/KAR kinetics and conductance.
These data suggest that the increase of spontaneous synaptic activity in the neuronal network induced by GluR agonists may contribute in excitotoxicity via strengthening neuronal depolarization by endogenous glutamate. Ouabain at subnanomolar concentrations is able to recover sEPSC frequency to the control level.
In our experiments ouabain in the concentrations under study did not affect integral currents through NMDARs or ANPAR/KARs. We, therefore, verified the most prominent explanation of GluR agonists and ouabain effects on sEPSC frequency, suggesting that by interacting with presynaptic NMDARs and Na+,K+-ATPase these compounds are involved in the regulation of presynaptic [Ca2+]i. Loading of neurons with BAPTA (2 and 5 μM), a chelator of Ca2+, to increase the capacity of intracellular Ca2+ buffering systems, eliminated both the NMDA effect (Fig. 7, F and G) and the ouabain effect on sEPSC frequency (Fig. 7G).
Na+,K+-ATPase as a Signal Transducer Targeting the Plasma Membrane Na+,Ca2+ Exchanger.
A large body of evidence has accumulated suggesting that Na+,K+-ATPase molecules are tightly packed with other integral proteins in functional clusters in the cell plasma membranes of different tissues. This provides direct molecule interplay and functional interaction between Na+,K+-ATPase and neighboring proteins (Xie and Askari, 2002; Li and Xie, 2009). To look for the recipient of Na+,K+-ATPase regulatory action in neurons we focused on the plasma membrane NCX, because it has been shown that these two molecules are anchored, forming a functional complex in the plasma membrane of cardiomyocytes (Xie and Askari, 2002; Aperia, 2007).
Experiments were performed in which the effects of 30 μM NMDA and 1 nM ouabain were studied when 10 μM 2-(2-(4-(4-nitrobenzyloxy)phenyl)ethyl)isothiourea methanesulfonate (KB-R7943) was present in the bathing solution. It is known that KB-R7943 at concentrations approximately 100 nM blocks the NCX in a reverse mode of transport (Iwamoto et al., 1996; Breder et al., 2000), whereas in concentrations used here both the forward and reverse transport were affected (Kimura et al., 1999; Breder et al., 2000). Inhibition of the NCX did not cause considerable changes in direct current amplitudes (Fig. 8A), but it did induce significant increase of sEPSC frequency (Fig. 8B; p < 0.00033, ANOVA, post hoc Tukey's test; n = 12). Application of 30 μM NMDA caused a tremendous increase of sEPSC frequency, as in the experiments without KB-R7943 pretreatment (Fig. 8). In contrast to the ouabain effects obtained with active NCX, when the NCX was inhibited by KB-R7943, 1 nM ouabain failed to affect sEPSC frequency: it remained at the same high value as recorded in the presence of 30 μM NMDA (Fig. 8).
In Ca2+ imaging experiments with Fura-2, combined application of 30 μM NMDA and 1 nM ouabain caused temporal elevation of [Ca2+]i, but not Ca2+ overload (Fig. 6E). Adding 10 μM KB-R7943 on the top of ouabain effect resulted in an uncompensated rise of [Ca2+]i in neurons (Fig. 9A). Averaged data are shown in Fig. 9B. This observation demonstrates that the NCX under these particular conditions operates in a forward mode (removing Ca2+ from the cell) and is critically involved in intracellular Ca2+ regulation by ouabain.
Our experiments disclose the antiapoptotic and neuroprotective effects of ultralow concentrations of ouabain and digoxin, which are manifested in the maintenance of the viability of the vast majority of cortical neurons during neurotoxic stress induced by long-lasting activation of NMDARs or AMPAR/KARs. It is well established that the GluR antagonists (2R)-amino-5-phosphonopentanoate, a selective antagonist of NMDARs, and 6-cyano-7-nitroquinoxaline-2,3-dione, a selective antagonist of AMPA/KARs (Traynelis et al., 2010), exhibit neuroprotective effects by blocking receptors that trigger excitotoxicity, thereby preventing the development of both necrosis and apoptosis (Choi, 1988; Olney, 1994; Lipton, 1999; Mironova et al., 2007). In contrast, subnanomolar concentrations of ouabain or digoxin in our experiments inhibited only apoptosis (Fig. 2). Normal expression levels of the antiapoptotic protein Bcl-2 were found in the presence of ouabain compared with decreased levels observed after NMDA- or KA-induced neurotoxic stress (Fig. 3). This observation suggests that ouabain and digoxin, which interacts with the highly conserved cardiotonic receptor of Na+,K+-ATPase, can somehow stimulate antiapoptotic intracellular pathways. Because ouabain and digoxin inhibit Na+,K+-ATPase of rats at concentrations (Fig. 4) that significantly exceed (Sweadner, 1989; Xiao et al., 2002; Richards et al., 2007; Katz et al., 2010) those that are antiapoptotic (0.01–1 nM), one may speculate that their neuroprotective effect could be realized because of a signaling function of the cardiotonic receptor on Na+,K+-ATPase. The evaluation of the high-affinity binding state for ouabain with the equilibrium dissociation constant of approximately 1 nM in the crystal structure of Na+,K+-ATPase (Ogawa et al., 2009) supports this assumption.
To provide some clues in favor of the mechanism of the ouabain antiapoptotic effects, whole-cell patch-clamp records during long-lasting NMDA or KA presence in the chamber were performed. These conditions were similar to our excitotoxic insult experiments. When NMDA or KA was applied, neurons generated inward direct currents through the channels of activated NMDARs and AMPA/KARs, respectively (Fig. 5). Cultured cortical neurons express mRNA encoding the NR1, NR2A, and NR2B subunits of NMDARs (Zhong et al., 1994). Because the channels of NR1/NR2A and NR1/NR2B subunit compositions are highly permeable for Ca2+ (Traynelis et al., 2010), some fraction of direct currents is determined by the Ca2+ entry into neurons, resulting in an immediate, continued elevation of [Ca2+]i (Fig. 6, A and B) (MacDermott et al., 1986). In the case of AMPA/KARs the intracellular Ca2+ signal has a more complex nature and is determined by expression of the GluR2 subunit of AMPARs, which rules the Ca2+ permeability of their channels (Burnashev et al., 1992; Traynelis et al., 2010). Cultured cortical neurons reveal a variety of intracellular Ca2+ response kinetics after KA applications (Abushik et al., 2011). The intracellular Ca2+ signal and further delayed intracellular Ca2+ deregulation (Khodorov, 2004) is thought to trigger neuronal apoptosis. Considering the high Ca2+ permeability of NMDARs and GluR2 lacking AMPARs, the most plausible way to antagonize apoptosis would be an influence on channel open probability, kinetics, or conductance, which would eliminate Ca2+ entry in the cytoplasm. GluR antagonists (Khodorov, 2004; Mironova et al., 2007; Traynelis et al., 2010) and channel blockers, (+)-5-methyl-10,11-dihydro5H-di-benzo[a,d]cyclohepten-5,10-imine (MK-801), ketamine, memantine, etc. (Church et al., 1988; Antonov et al., 1995, 1998; Lipton, 1999; Traynelis et al., 2010) represent examples of such an influence. In our experiments no effects of ouabain at antiapoptotic concentrations (0.1 and 1 nM) either on the amplitude (Fig. 7A) or I-V relationship of direct currents transmitted through open NMDARs (Fig. 7D) or AMPA/KARs (Fig. 7E) were found. The lack of effects may suggest that the target of ouabain regulation is located downstream of the Ca2+ entry into neurons.
A growing body of evidence is accumulating showing that NMDARs, AMPARs, and KARs are expressed in presynaptic terminals, as well as in postsynaptic membranes (for review see Pinheiro and Mulle, 2008). These autoreceptors play a role in synaptic plasticity, providing either a potentiation or depression of EPSC and spontaneous transmitter release in different brain structures. Several forms of synaptic plasticity were also demonstrated in primary cultures of rat cortical neurons (Han and Stevens, 2009). In addition to direct currents in the presence of NMDA and KA, a tremendous increase of sEPSC frequency (Figs. 5 and 7, A-C) was observed. To keep the conditions of experiments similar to those in the excitotoxicity study we did not add tetrodotoxin, a blocker of voltage-gated Na+ channels, and bicuculline, an inhibitor of ligand-gated γ-aminobutyric acid type A receptors, in the bathing solution. Synaptic currents recorded under these particular conditions had different origins. Obviously, some giant sEPSCs appeared that, perhaps, represented EPSCs evoked by presynaptic neuron spike firing enforced by chronic neuronal depolarization in the network. sEPSCs of smaller amplitude may have represented miniature EPSCs. Some contribution of inhibitory currents was also possible. Overall, the antiapoptotic ouabain concentrations applied on top of the agonist effects abolished the increase of sEPSC frequency (Fig. 7A), which returned to the control value (Fig. 7, B and C). Clearly, loading neurons with BAPTA, a chelator of Ca2+, to extend the intracellular Ca2+ buffering capacity abolished both the NMDA-induced sEPSC discharge and the compensatory effects of ouabain (Fig. 7, F and G). This may suggest that the NMDA-induced sEPSC discharge is caused by the accumulation of Ca2+ in presynaptic boutons, whereas ouabain at antiapoptotic concentrations somehow up-regulates intracellular Ca2+ clearance processes. Therefore, the binding of ouabain to Na+,K+-ATPase may contribute to the regulation of presynaptic [Ca2+]i. Direct measurements of intracellular Ca2+ dynamics and concentration supported the conclusion drawn from the experiments described above. Whereas applications of NMDA to neurons caused the immediate, continuous elevation of [Ca2+]i (Fig. 6, A and B), combined applications of NMDA with 1 nM ouabain induced a transient rise of [Ca2+]i to a maximum reaching 1 μM in some neurons, which then declined gradually in time (Fig. 6, C and D). In 10 min the [Ca2+]i recovered to the control values (Fig. 6E). These results are consistent with the data obtained by using the patch-clamp technique and support the assumption that Na+,K+-ATPase is involved in the regulation of [Ca2+]i. Clearly, voltage-gated Ca2+ channels as an alternative to GluR way of Ca2+ entry (AWCE) could play a role in ouabain action. However, we did not find any effects of antiapoptotic ouabain concentrations on neuronal I-V relationships under the control conditions (data not shown). Therefore, ouabain-induced Ca2+ clearance most likely is determined by up-regulation of Ca2+ extrusion rather than inhibition of Ca2+ entry.
In neurons, Ca2+ extrusion is operated by the plasma membrane Ca2+ pump and Na+,Ca2+ exchangers. The plasma membrane Ca2+ pump has high Ca2+ affinity but low transport capacity, whereas the NCX has a low affinity, but a higher capacity to transport Ca2+ (Bano et al., 2005). Inhibition of Ca2+ efflux from cells by the NCX is sufficient to cause sustained intracellular Ca2+ elevation and the demise of neurons. The expression of the NCX prevented Ca2+ overload and rescued neurons from excitotoxic death (Bano et al., 2005). Treatment of cortical neurons with a specific inhibitor of the NCX, KB-R7943 (Iwamoto et al., 1996; Breder et al., 2000), in our experiments prevented the compensatory effects of ouabain, which lost the ability to decrease [Ca2+]i (Fig. 9) and the sEPSC frequency in the NMDA-induced sEPSC discharge (Fig. 8). Known KB-R7943 side effects (partial inhibition NMDAR and L-type Ca2+ channels; Brustovetsky et al., 2011) should oppose intracellular Ca2+ accumulation that was not observed in our experiments. Therefore, this observation could be interpreted in a way that the NCX is the most likely candidate as a molecular target for Na+,K+-ATPase signal regulation. This functional interaction of the ouabain liganded Na+,K+-ATPase with the NCX somehow enforces Ca2+ extrusion and protects neurons from Ca2+ overload. Our conclusion is consistent with a previous study on snail neurons, which demonstrated a stimulation of the plasma membrane Na+,Ca2+ exchange by nanomolar concentrations of ouabain (Saghian et al., 1996). In those experimental conditions, however, the reversed mode of transport was up-regulated, causing intracellular Ca2+ accumulation. Presumably, these effects are induced by nondirect ouabain action on the NCX and may be secondary to a rise of intracellular cAMP (Saghian et al., 1996). Because Na+,K+-ATPase is the only known, highly specific receptor for ouabain and other cardiotonic steroids (Ogawa et al., 2009; Lingrel, 2010), it is unlikely that ouabain directly interacts with the NCX. In our experiments another specific ligand of the Na+,K+-ATPase cardiotonic steroid binding site, digoxin, reveals antiapoptotic action as well as ouabian. This corroborates our assumption that it is Na+,K+-ATPase that is a primary target triggering neuroprotection.
The interpretation of our data is illustrated in Fig. 10. Under normal conditions the capacity of intracellular Ca2+ buffering systems and Ca2+ extrusion by the NCX are sufficient to compensate Ca2+ that enter neurons through the AWCE representing different types of voltage-gated Ca2+ channels and pumps (Fig. 10A). The activation of autosynaptic and postsynaptic GluRs (in Fig. 10 only NMDARs are shown in presynaptic terminals for simplicity) causes depolarization and additional Ca2+ entry through the channels of NMDARs and AWCE, resulting in Ca2+ overload (Fig. 10B). In addition, the accumulation of free Ca2+ in presynaptic boutons elevates the probability of spontaneous vesicular transmitter release increasing sEPSC frequency. The occupation by ouabain of its binding site on Na+,K+-ATPase is followed by an acceleration of Ca2+ extrusion by the NCX (Fig. 10C). This prevents Ca2+ accumulation in cytoplasm. Inhibition of the NCX eliminates Ca2+ extrusion from neurons resulting in further Ca2+ overload, which makes ouabain binding ineffective (Fig. 10D). Whether those functional interactions between Na+,K+-ATPase and the NCX include direct molecular interactions remains to be elucidated.
The range of ouabain antiapoptotic concentrations corresponds well with the endogenous ouabain level, which varies from 0.1 to 0.74 nM in rat blood plasma and cerebrospinal fluid (Dobretsov and Stimers, 2005). Similar antiapoptotic effects of low ouabain doses have been shown to be associated with enhanced production of Bcl-2 in another neurodegeneration model when KA and ouabain were injected in the brain in vivo (Golden and Martin, 2006). Both findings provide corroborating evidence for the physiological relevance of endogenous ouabain. Thus, the data presented here demonstrate a novel function of Na+,K+-ATPase as a neuroprotective molecule that might be triggered by binding of endogenous ouabain or its analogs to a highly conserved cardiotonic/ouabain receptor site.
Participated in research design: Krivoi and Antonov.
Conducted experiments: Sibarov, Bolshakov, Abushik, and Antonov.
Performed data analysis: Sibarov, Bolshakov, Abushik, and Antonov.
Wrote or contributed to the writing of the manuscript: Krivoi and Antonov.
We thank Dr. J. W. Johnson and Dr. J. A. Heiny for reading the manuscript and providing critical suggestions.
This work was supported by the Russian Foundation for Basic Research [Grants 08-04-00423, 11-04-00397 (to S.M.A.); 10-04-00970 (to I.I.K.)]; the Russian Federation Ministry of Education and Science [Contract 8476] (to IEPhB RAS); and Saint-Petersburg State University [Research Grant 1.37. 118.2011] (to I.I.K.).
Article, publication date, and citation information can be found at http://jpet.aspetjournals.org.
- ionotropic glutamate receptor
- acetoxymethyl ester
- analysis of variance
- alternative to GluR ways of Ca2+ entry
- 2-amino-3-(5-methyl-3-oxo-1,2-oxazol-4-yl)propanoic acid receptor
- acridine orange
- 1,2-bis(2-aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid
- intracellular calcium concentration
- central nervous system
- ethidium bromide
- excitatory postsynaptic current
- spontaneous EPSC
- fluorescent viability assay
- kainic acid
- KA receptor
- 2-(2-(4-(4-nitrobenzyloxy)phenyl)ethyl)isothiourea methanesulfonate
- Na+/Ca2+ exchanger
- NMDA receptor
- ouabain molecule
- phosphate-buffered saline.
- Received July 11, 2012.
- Accepted August 24, 2012.
- Copyright © 2012 by The American Society for Pharmacology and Experimental Therapeutics