- Email: [email protected]
Development of an in vitro model of neuronal activity induced excitotoxicity using photoconductive stimulation
Cell Calcium 47 (2010) 441–448
Contents lists available at ScienceDirect
Cell Calcium journal homepage: www.elsevier.com/locate/ceca
Development of an in vitro model of neuronal activity induced excitotoxicity using photoconductive stimulation Evgeny Pavlov b , R. Carolina Gutiérrez a , Yuan Zhang a , Audrey C. Kertesz a , Johanna Hung a , Ferdinand Joseph Espina a , Michael A. Colicos a,∗ a b
Department of Physiology & Pharmacology and the Hotchkiss Brain Institute, University of Calgary, Calgary, AB, Canada T2N 4N1 Department of Physiology and Biophysics, Dalhousie University, Halifax, NS, Canada B3H 1X5
a r t i c l e
i n f o
Article history: Received 21 September 2009 Received in revised form 10 March 2010 Accepted 25 March 2010
Keywords: Cell death Neuron Excitotoxicity Cyclosporin A Mitochondria Permeability transition
a b s t r a c t Loss of the ability to regulate calcium is a central event leading to neuronal cell death during a wide range of pathological conditions including stroke and seizure. Here we present a new dissociated hippocampal cell culture model of acute electrical activity which incorporates the photoconductive stimulation of neuronal networks grown on silicon wafers. This technology allows precise modeling of user defined neuronal activity patterns, and the study of their effect on neuronal physiology. Here, seizure-like conditions were created by continuous stimulation, causing hundreds of neurons to fire synchronously at 50 Hz for 4 min. This stimulation protocol induced cell death as monitored by propidium iodide staining. The number of dead cells per stimulation region increased from 3.6 ± 2.1 preceding stimulation to 81 ± 21 30 min following stimulation. Excitotoxicity primarily affected excitatory rather than inhibitory neurons, and was preceded by an increase in intracellular calcium as well as changes in the mitochondrial membrane potential, as measured by a tetramethylrhodamine methyl ester (TMRM) assay. Cyclosporin A (CsA), a mitochondrial permeability transition pore (PTP) blocker, was effective in preventing cell death. We propose that photoconductive stimulation is a useful tool for investigating the pathogenesis of excitotoxicity in vitro. © 2010 Elsevier Ltd. All rights reserved.
1. Introduction Excessive intracellular calcium is a primary trigger of neuronal cell death in a variety of neuropathological conditions. The use of cell culture models which mimic conditions of calcium overload has been critical for detailed studies of the processes leading to such stress-induced cell death. For example, dissociated cultures are widely used to model neuronal glutamate toxicity following stroke. In the brain, stroke induces calcium deregulation through release of glutamate from damaged neurons which in turn activates NMDA receptors and causes uncontrolled calcium increase followed by massive neuronal death [1,2]. In culture models glutamate is added directly to the tissue culture mimicking the condition likely experienced by neurons in the brain [3], where glutamate causes permanent activation of NMDA receptors and calcium influx accompanied by the loss of neuronal firing. In contrast, during seizures neurons undergo massive intracellular calcium increase caused by constant excessive neuronal electrical
∗ Corresponding author at: Department of Physiology & Pharmacology, University of Calgary, 3330 Hospital Drive NW, Calgary, AB, Canada T2N 4N1. Tel.: +1 403 220 4594; fax: +1 403 283 7137. E-mail address: [email protected] (M.A. Colicos). 0143-4160/$ – see front matter © 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.ceca.2010.03.005
activity. Thus the key requirement in developing a culture model of seizure is the induction of continuous neuronal firing in these cells. Several in vitro cell culture paradigms have been used to investigate the pathophysiological mechanisms underlying excitotoxicity in the context of hippocampal cell death following seizure (for review see [4,3]). Global neuronal firing can be enhanced chemically by placing cells into solution with low magnesium [5,6], pilocarpine [7], kainic acid [8], or other drugs. Another approach to the development of a culture model for seizure activity has focused on the use of direct electrical stimulation of neuronal cultures, which would not require chemical intervention and induce higher and more controlled firing rates [9]. While multi-electrode arrays are a very useful tool for this, they have the disadvantage being costly, and it is impractical to simultaneously grow large numbers of parallel cultures using this technology. We have taken an alternate strategy based on photoconductive stimulation of neurons grown on silicon wafers. Multiple wafers can be generated from a single animal and maintained in standard 24 well culture plates, and stimulation can be performed on an upright fluorescence microscope. By targeting a specific region of the neuronal culture with light, the local conductivity of the underlying silicon increases. Then, by applying an electrical stimulus to the wafer, the majority of current passes through the illuminated region, specifically firing the targeted subset of neurons. Thus a highly
442
E. Pavlov et al. / Cell Calcium 47 (2010) 441–448
precise area of stimulation can be achieved; and also allows internal controls to be recorded in unstimulated regions of the same wafer. In this report we use high-frequency photoconductive stimulation of neurons to recreate the conditions which are experienced by neurons during acute electrical hyperactivity. We found that the pathological changes we observed closely resemble changes seen in neurons in the intact brain under neuropathological conditions. This includes a significant increase in calcium accumulation, which if not restored results in neuronal cell death. We also demonstrate how a biochemical mechanism contributing to this cell death (opening of the mitochondrial permeability transition pore) can be directly monitored, and therapeutically blocked. Additionally, we characterize the cell death in terms of the general neuronal cell types involved (inhibitory versus excitatory neurons), addressing a question of subtype susceptibility. We conclude that this in vitro model allows real-time dynamic study of neuronal physiology during extreme activity. As it does not involve chemical intervention, this model permits large-scale screening for therapeutic strategies for treatment of excitability related neuronal damage such as caused by seizures.
immersion objective. Images were captured using custom built video equipment based on a WATEC N105 CCD camera, and AMCAP (Microsoft) software, and processed using ImageJ (NIH). 2.4. Network activity analysis In order to determine whether self-sustaining activity could be triggered by brief high-frequency stimulation, neuronal firing over the entire field of view was monitored for at least 10 min following a 1 s, 50 Hz pulse. Images of triggered neuronal activity were made at 15 frames per second using cultures loaded with the calcium dye Fluo-4-AM. For staining, 2 g of Fluo-4 were added to cells on a wafer in 500 l of recording solution and incubated at 37 ◦ C for 30 min prior to recording. Individual neurons were tracked over time and their activity traces extracted by calculating the mean grey value in an ROI encompassing the cell body. These activity traces were analyzed by custom software to determine the firing times of each neuron in the field of view. All neurons during the 10 min recording were simultaneously analyzed for their spatial and temporal correlations using custom written routines in Matlab. In all experiments light intensity was kept to the minimum in order to avoid photobleaching.
2. Materials and methods 2.5. Immunohistochemistry 2.1. Dissociated neuronal co-cultures Cultures suitable for photoconductive stimulation were prepared as described [10]; briefly, hippocampi were dissected from P0 (newborn) Sprague–Dawley rat pups, and then neurons and glia were dissociated and plated together on pretreated Si wafers [11]. Cultures were maintained in Eagle’s Basal Media (BME, GIBCOInvitrogen) supplemented with B-27, penicillin, streptomycin and l-glutamine. Functional synapses formed after a few days in culture, and networks were left to develop for approximately 2 weeks, at which point robust spontaneous activity could be observed. For experiments that required construct expression, cultures were transfected using Lipofectamine 2000 (Invitrogen) the day before use. 2.2. Photoconductive stimulation-induced acute neuronal activity Photoconductive stimulation of rat hippocampal neuronal cultures grown on silicon wafers was performed as previously described [11]. A ∼3 mm diameter field of view (which encompasses hundreds of neurons) was illuminated with 434 nm light on an Olympus BX60WI microscope using a 10× water immersion objective. As a general parallel of activity occurring during status epilepticus (SE), 50 Hz stimulation was used and continuously applied to the neurons for 4 min. Although in current work we limited our experiments to only this “acute stimulation” protocol, any profile of frequencies, duration, cycles or intensity can be programmed. Running experiments with illumination but without current applied to fire the neurons resulted in no cell death above baseline levels, directly differentiating activity driven death from phototoxic effects. In addition, experiments were done using more “normal” paradigms of neuronal activity, 10–50 Hz stimulation for 10 s, repeated after an interstimulus interval of a few minutes. No changes in the mitochondrial morphology or the membrane potential indicated by TMRM labeling were detectable (data not shown). 2.3. Video imaging Live video imaging was performed on an Olympus BX60WI microscope with either a 100× 1.00 NA or a 60× 0.90 NA water
To correlate cell death with neuronal cell type, cultures were stimulated using the SE protocol (50 Hz, 4 min) and subsequently stained with propidium iodide (Sigma). Predominantly inhibitory neurons were identified using anti-GAD65 antibodies (Chemicon, AB5082) and by morphology. Immunocytochemistry was carried out as described previously [12]. Briefly, the cultures were fixed using 4% paraformaldehyde for 20 min. Antibody detection was carried out in a low permeabilization buffer (0.01% Triton X 100) using a 1:1000 dilution of the GAD primary antibody and 1:100 donkey anti-rabbit Cy5 antibody (Jackson Immunoresearch Laboratories) for visualization. 2.6. Mitochondrial membrane potential assay Mitochondrial membrane potential dissipation was monitored by measuring TMRM fluorescence as previously described [13]. In order to do this TMRM (20 nM, Molecular Probes) was added to the recording dish and allowed to equilibrate for 20 min prior to the beginning of measurement. Fluorescence intensity of TMRM during continuous stimulation was measured from regions enriched in mitochondria and from control cytoplasm regions and quantified using ImageJ (NIH). 2.7. Mitochondrial morphology changes Mitochondrial morphology was monitored by fluorescent microscopy using mitochondrially targeted green fluorescent protein (GFP). pAcGFP1-Mito cDNA (MTS-GFP, Clontech) encodes a fusion of a mitochondrial targeting sequence derived from the precursor of subunit VIII of human cytochrome C oxidase and GFP. Cultured cells were transfected with the DNA using PolyFect® Transfection Reagent (Qiagen) according to the manufacturer’s recommended protocol using 1 g of DNA per 2 ml of media. 2.8. Cell death assay Cell death was assayed using propidium iodide staining as described previously [13]. Propidium iodide was added to the recording solution at concentration of 20 M. Dead cells, identified from propidium iodide staining of the nucleus, were counted
E. Pavlov et al. / Cell Calcium 47 (2010) 441–448
443
Fig. 1. Induction of spontaneous self-sustaining activity in dissociated neuronal cultures by photoconductive stimulation. Neuronal activity was visualized by imaging calcium influx with Fluo-4. After a control period of 4 min, a 1 s, 50 Hz stimulation was delivered to the observed region of the culture, initiating a sustained cascade of rapid activity. (A) Sample frames from the time course, beginning at the start of the control period (t = 0). Stimulation occurs at t = 4, after which firing of individual neurons can be seen, as well as global depolarizations of underlying glial cells (more easily seen at this magnification) caused by the high level of overlying neuronal activity in the neurons. Glial depolarizations were not included in the data analysis. Scale bar = 50 m. (B) Spatial and temporal representation of firing events 1 min before and 2 min after stimulation of the observed region of the culture. Stimulation point is indicated by arrow. Firing events were isolated by peak detection, and translated to the 3D graph. Firing events are linked if they occur within a specific time of each other for visualization. Immediately after the 1 s stimulation extensive activity is initiated and sustained for at least several minutes. This effect was consistent, and quantified in n = 3 cases.
in the central field of the region undergoing stimulation and in 4 adjacent fields (10×). 2.9. Statistical analysis Statistical analysis was done using t-tests and the online statistics calculator from GraphPad Software, Inc. 3. Results 3.1. Spontaneous activity in neuronal cultures Cultures of primary neurons have been well documented to have complex connectivity and to express a wide variety of neuronal firing patterns, including robust spontaneous activity and bursting behaviors. To establish our culture methodology as appropriate
for the modeling of neuronal network behavior, we determined whether DIV14 dissociated neuronal cultures had sufficiently complex connectivity to allow self-sustaining activity in response to a brief (1 s) high-frequency (50 Hz) pulse. To induce the activity we used photoconductive stimulation, a technique that uses the change in conductivity of illuminated silicon to target a capacitive induction of neuronal firing [11]. This stimulation technique results in the induction of action potentials indistinguishable from spontaneous firing [14]. To monitor neuronal activity we visualized action potentials using the calcium indicator Fluo-4 (Fig. 1A). In an unperturbed DIV14 network, a low level of spontaneous activity was observed (Fig. 1B). Upon applying a 1 s, 50 Hz stimulation to a large population of neurons, a massive increase in activity was observed (Fig. 1B). This activity was completely self-sustaining, continuing for the extent of a 10 min recording. This finding is in agreement with previous reports using other initiators of such activity [4]. This
444
E. Pavlov et al. / Cell Calcium 47 (2010) 441–448
Fig. 2. Disruption of intracellular calcium homeostasis by a SE model leads to neuronal death. Neurons were stimulated at 50 Hz for 4 min. Each trace corresponds to Fluo-4 fluorescence from an individual cell body. Cell death was visualized by propidium iodide (PI), and cells were grouped according to their PI signal following stimulation. (A) Fluorescence from cells which were PI positive immediately after the end of the stimulation. Note the abrupt drop in Fluo-4 fluorescence which presumably indicates loss of the dye due to the cell membrane permeabilization. (B) Fluorescence from the cells which became PI positive 30 min after the end of stimulation. (C) Fluorescence from the cells which survived stimulation. Note that these cells maintained their ability to regulate intracellular calcium. (D) PI signal of cells before stimulation (left), PI staining immediately following 4 min of stimulation (center) and PI staining 30 min after stimulation (right). Scale bar is 20 m. (E) Number of dead cells per image over 3 time points, illustrating that both acute and protracted cell death can be observed (t-test, p = 0.002, n = 3).
Fig. 3. Mitochondrial shape change in response to stimulation. Neurons with labeled mitochondria were monitored continuously during a 4 min, 50 Hz stimulation. (A) t = 0 min, (B) t = 2 min, and (C) t = 4 min. Note the transition of the shape of mitochondria from elongated structures typical to normal neurons to a smaller, punctate form. This result was observed in every case (n = 12). Scale bar is 8 m. (D) Statistics were performed by measuring the length of mitochondria from 3 independent experiments, in each experiment maximal length of 100 individual mitochondria were measured and results combined together. Neurons become noticeably smaller upon stimulation. Bin size is 1.5 m, error bars are SD.
E. Pavlov et al. / Cell Calcium 47 (2010) 441–448
445
confirms that in our experiments the neuronal networks formed by dissociated cultured neurons are inherently capable of robust, self-sustaining activity. 3.2. Photoconductive stimulation driven activity induces an increase in intracellular calcium and cell death To assess the effects of excessive neuronal activity on intracellular calcium homeostasis, we induced continuous neuronal firing by photoconductive stimulation. A brief stimulation depolarizes neurons and results in the generation of action potentials and an accompanying increase in cytoplasmic calcium concentration through voltage-gated calcium and NMDA receptor channels. Under normal firing conditions induced by brief stimulation, calcium concentrations quickly decrease back to pre-stimulation levels without affecting cell viability [11,15,16]. Seizure conditions, however, are characterized by an uncontrolled increase in the duration and frequency of neuronal firing. To recreate these conditions in our experiments, we stressed the neuronal network using continuous photoconductive stimulation for 4 min, to ensure continuous neuronal activity in the entire network. We observed that continuous 50 Hz stimulation caused a rapid, substantial increase in intracellular Ca2+ concentration in some neurons, as detected by Fluo-4 fluorescence. Among these, three distinct subpopulations were observed (Fig. 2A–C). Cells in Fig. 2A–C were assigned to each category based on their resistance to the cell death as monitored by PI staining. Specifically cells at panel (A) were PI positive immediately after the end of stimulation. Cells at panel (B) were PI negative immediately after the end of stimulation but developed PI staining 30 min later, while cells at panel (C) were PI negative 30 min after stimulation, During the stimulation event the intracellular Ca2+ of some neurons dropped rapidly after a few minutes (Fig. 2A), while another group maintained elevated Ca2+ levels throughout the time course of stimulation (Fig. 2B). In addition, a subpopulation of neurons continued normal neuronal activity after the stimulation episode without losing Ca2+ homeostasis (Fig. 2C). We also quantified the extent of cell death both immediately after (4 min) and 30 min following stimulation (Fig. 2D). We observed that the stimulation protocol caused substantial neuronal cell death from 3.6 ± 2.1 dead cells per image before stimulation to 30 ± 6 cells after (p = 0.002, n = 3), as determined by PI staining (Fig. 2E). The number of PI positive cells further increased to 81 ± 21 cells per image (p = 0.02, n = 3) during the 30 min following stimulation (Fig. 2E, n = 13 independent experiments). The neurons that were PI positive immediately after stimulation belonged to the subpopulation characterized by a complete drop in Fluo-4 fluorescence during stimulation. This suggests that the drop in fluorescence was caused by membrane permeabilization and the release of the Ca2+ bound indicator (see Fig. 2A). On the other hand, cells, which did not display PI staining even 30 min after the end of stimulation were able to maintain normal calcium levels (Fig. 2C). This demonstrates that neuronal cell death caused by acute stimulation strongly correlates with disruption of calcium homeostasis. 3.3. Continuous stimulation results in loss of mitochondrial membrane potential and changes in mitochondrial morphology A change in mitochondrial morphology from elongated to spherical shape is a key event following glutamate-induced calcium toxicity [17–19]. To investigate whether a similar phenomenon occurs during extreme neuronal activity we monitored mitochondrial shape using targeted GFP protein (Fig. 3). We observed that constant neuronal activity initiated a change in mitochondrial shape, from elongated to spherical (n = 12 independent experiments), similar to previous observations in glutamate stress
Fig. 4. Loss of mitochondrial membrane potential in the neurons following SE. Mitochondrial membrane potential was monitored using TMRM, a potential sensitive dye. (A) Prior to stimulation, mitochondria show visible TMRM signal. (B) Following SE stimulation, TMRM fluorescence rapidly dissipated, indicating a loss of membrane potential. Scale bar is 8 m. (C) TMRM signal collected from a single mitochondrial and a single cytoplasmic region. A decline in mitochondrial fluorescence corresponds with an increase in cytoplasmic fluorescence, suggesting loss of mitochondrial membrane integrity. (D) The ratio of mitochondrial to cytoplasmic TMRM signal declines after stimulation (n = 7 cells in each condition). All ratios were significantly different (t-test, p < 0.002), error bars are SEM.
paradigms (for review see [17,20]). This suggests that, in our experiments, a change in mitochondrial function could play a central role in neuronal death. Opening of the calcium-induced permeability transition pore (PTP) is known to be the central event in the process of necrotic cell death induced by calcium toxicity [21]. A large body of experimental data suggests that PTP opening is responsible for cell death in pathological conditions ranging from ischemia-reperfusion injury in heart to neuronal glutamate toxicity in the brain. Since the neuronal death in our experiments showed a strong correlation with an uncontrolled increase in calcium concentration and a dramatic
446
E. Pavlov et al. / Cell Calcium 47 (2010) 441–448
Fig. 5. Cyclosporin A protects cells from death and prevents calcium deregulation. To confirm that PTP opening is causal in the neuronal cell death observed, CsA (a PTP blocker) was added during an SE stimulation experiment. (A) Propidium iodide stained neurons 30 min after SE stimulation in the absence and (B) in the presence of CSA. Scale bar is 20 m. (C) The number of dead cells in a single field (∼0.64 mm) decreased significantly with CsA treatment (15 ± 5 compared to 60 ± 14 in control cultures (n = 3 cultures for each condition, t-test, p = 0.006), error bars are SEM). (D) Sample traces of Fluo-4 signal in neurons during SE stimulation in the presence of CsA. Few cells lost calcium homeostasis as indicated by most showing normal calcium oscillations.
change in mitochondrial morphology, we tested the hypothesis that neuronal death was caused by PTP activation. PTP opening leads to inner mitochondrial membrane depolarization and can be detected using the potential sensitive dye TMRM [13]. TMRM is a charged fluorescent dye which accumulates in intact mitochondria due to the presence of the negative potential in their inner membrane. Upon mitochondrial depolarization triggered by PTP opening, TMRM is released from mitochondria into the cytoplasm which leads to dissipation of the fluorescent signal. Fig. 4A and B depicts images of mitochondria loaded with TMRM before and after stimulation, and show that continuous stimulation leads to an abrupt decrease in fluorescent signal from mitochondrial regions (arrowheads), which suggests opening of the PTP (n = 9 independent experiments). At the whole-cell level, TMRM fluorescence from mitochondria progressively decreases with the duration of stimulation, and this process coincides with increase in cytoplasmic fluorescence due to release of TMRM from mitochondria (Fig. 4C and D). In these experiments, the ratio between fluorescence of mitochondrial to cytoplasmic regions progressively decreased from 0.97 ± 0.05 to 0.77 ± 0.12 (p = 0.002, n = 7) within 20 s following the beginning of neuronal firing and further decreased to 0.52 ± 0.1 (p = 0.001, n = 7) at 40 s, confirming mitochondrial depolarization. Importantly the time from the beginning of stimulation to onset of mitochondrial depolarization varied from experiment to experiment most likely corresponding to the variation in time required for mPTP to be activated. This variation is consistent with variation in time in onset of calcium deregulation (see Fig. 2A–C). 3.4. Prevention of stimulation-induced neuronal death To further confirm the link between calcium overload, PTP opening and cell death we compared the number of dead cells
when stimulation was done in the presence and in the absence of cyclosporin A (CsA)—a blocker of mitochondrial PTP [22]. In these experiments we first induced seizure-like activity (50 Hz for 4 min) in a region of neurons in the absence of CsA and estimated the number of dead cells 30 min after stimulation using PI staining. Next, CsA at a concentration of 5 M was added to the bath and another region of the same chip was stimulated in the same way followed by counting of PI positive cells. The use of the same chip ensured that stimulation conditions were identical in the presence and in the absence of CsA. As can be seen from Fig. 5A and B the number of dead cells in the presence of CsA (15 ± 5) was significantly lower than in control cells (60 ± 14) (p = 0.006, 3 independent experiments). Furthermore, in the presence of CsA, most of the cells demonstrated normal activity with typical reversible fluctuations of the cytoplasmic calcium concentrations (Fig. 5C). We conclude that extreme seizure-like conditions induce opening of the calcium-dependent PTP and thus cell death, which can be prevented by CsA treatment. 3.5. Inhibitory neurons are more resistant to seizure-induced cell death than excitatory neurons Seizure activity has been speculated to alter the balance between inhibitory and excitatory neurons in the brain (for review see [23]). One of the advantages of our in vitro model system is that we can directly assess changes in the inhibitory and excitatory balance following acute electrical stimulation. After performing the stimulation protocol in the presence of PI, we fixed the neurons and probed them with anti-GAD65 antibody (Fig. 6A) to identify inhibitory neurons, as well as with DAPI to determine the total cell population (Fig. 6B). We determined that while approximately 18 ± 3% of the total neuronal population died, the percentage of dead inhibitory neurons was only 7 ± 0.3% (statistically different at
E. Pavlov et al. / Cell Calcium 47 (2010) 441–448
447
4. Discussion Neuronal death caused by hyper-excitation and excessive neuronal activity is a common factor in several neuropathological conditions including traumatic head injury, stroke and seizures. Preventing the acute loss of brain cells under these conditions is an important goal in developing therapeutic strategies to limit the risk of permanent brain damage. Here we introduce a new technology designed to reproduce the physiological stress that occurs in neurons during acute electrical activity similar to that seen under these conditions, and demonstrate its effectiveness in characterizing critical cell death processes. The main advantage of the present model is that it allows the study of cell damage caused exclusively by electrical stimulation without the use of any chemical stimulation. By using a silicon-based interface with the neuronal cultures, we can accurately specify any combination or progression of neuronal activity while concurrently monitoring the electrophysiological and biochemical status of the neurons before, during and after the firing event. In this report we use our model to recreate the neuronal duress caused by excessive electrical activity, using a generalized activity profile similar to that which is observed during seizures. It is important to note that while our current study focused on a simple 4 min, 50 Hz stimulation protocol, by using photoconductive stimulation, one can apply any stimulation pattern or combination of patterns the network. This flexibility allows researchers complete and precise control over the electrical activity they wish to investigate. This flexibility currently cannot be achieved either in existing culture nor whole-animal models of seizure. For example, low magnesium based culture models of seizure can only create conditions of continuous firing at relatively low frequency (10 Hz) resulting in substantial cell death only after hours of activity. Whole-animal models, on the other hand, where seizure activity is induced by chemicals such as pilocarpine cause excessive neuronal activity leading to massive amounts of cell death, a situation which is not paralleled in most human seizure. Also, while techniques such as field stimulation could provide a similar global level of activity, the lack of spatial resolution (to prohibit the firing of a subset of neurons) precludes the use of internal controls. Our experiments demonstrate the effectiveness of this model for the study of seizures in three critical areas:
Fig. 6. SE stimulation differentially affects inhibitory and excitatory neuron cell death. Following stimulation, cultures were fixed and inhibitory neurons identified using an anti-GAD-65 antibody. Immunocytochemistry protocol was optimized to detect cell bodies of inhibitory cells. (A) Representative image of the GAD-65 immunoreactivity in cell bodies of DIV 10 cells. (B) Triple label of inhibitory neurons using GAD-65 (green), DAPI (blue) to provide the total number of cells in the cultures, and PI (red) to detect dead cells. (C) GAD-65 neurons were less vulnerable to cell death than the general neuronal population (n = 16 regions from 3 different experiments, t-test, p = 0.003). Scale bars A = 10 , B = 100 .
p = 0.003, Student’s t-test) of the total number of inhibitory neurons (Fig. 6C), taken from 3 independent experiments using cells from 3 different culture rounds. Percentages were calculated from a total population of 4215 cells, 278 of which were inhibitory. This suggests that excitatory neurons have a 2–3-fold higher rate of cell death than inhibitory neurons following acute electrical stimulation.
(1) Studies of intracellular processes during acute electrical activity. Excessive accumulation of Ca2+ inside neurons during seizure is the major cause of toxicity leading to cell death (see [24] for review). Although influx of calcium is known to create favorable conditions for opening of the mitochondrial PTP, it is not clear whether PTP opening is a cause or a consequence of cell death. Here we demonstrate that prolonged, seizure-like stimulation induces substantial Ca2+ accumulation, followed by PTP opening leading to cell death, and that this can be prevented by CsA, a blocker of PTP [25]. We suggest that in the case of acute neuronal electrical activity, the major event which causes cell death is permeability transition pore (PTP) opening induced by massive calcium accumulation inside mitochondria. Our observation is consistent with recent report on the potential involvement of mPTP in neuronal death caused by exposure of mice to kainic acid [26]. To our knowledge, the present study is the first direct demonstration at the level of primary culture that prolonged neuronal firing is sufficient to induce neuronal cell death through a PTP-dependent mechanism. We propose that the most likely scenario of the cell death seen in our experiments is the following: (I) continuous stimulation leads to repetitive activation of voltage-gated Ca2+ channels; (II) opening of voltage-gated and NMDA receptor calcium channels causes substantial Ca2+ uptake, which dramatically increases its concentration in the cytoplasm; (III) the increase in cytoplasmic
448
E. Pavlov et al. / Cell Calcium 47 (2010) 441–448
Ca2+ leads to its massive Ca2+ accumulation by mitochondria, causing PTP opening. PTP opening directly leads to disruption of mitochondrial function and cell death. (2) Studies of various subpopulations of neuronal cells. One of the critical questions which can be addressed is the relative susceptibility of different types of neurons to seizure-induced death. Here we were able to demonstrate that inhibitory neurons are more resistant to seizure-induced cell death compared to the total neuronal population. (3) A simple tool for drug screening. As an example we show that cell death can be prevented by CsA. Similarly, other drug candidates can be applied and tested neuroprotective effects. Identified candidates can later be used for further testing in animal models. In summary, the silicon-based excitotoxicity paradigm presented in this paper is an advanced method to assist the study of pathological events in neurons under user defined activity conditions while simultaneously monitoring neuronal physiology. As a highly scalable in vitro model, it allows studies to be performed which open unique possibilities to screen for drugs which will reduce neuronal damage. Moreover, as a culture based paradigm, unique genetic or environmental alterations are possible, allowing investigations such as the role of channelopathies or viral infections in seizure susceptibility. Using this model we also demonstrate the direct link between activity induced cell death and mitochondrial PTP opening, and importantly we demonstrate a methodology for preventing this cell death. We also demonstrate that the balance between excitatory and inhibitory neurons is altered following seizure, which is an important factor when considering the long term ramifications of seizure events on the systems level. Acknowledgements We thank R. Flynn for providing comments on the manuscript. Grants: This work was supported by CIHR grant #69-6113 to R. French and EP, HSFA support to EP, CIHR and AIF grants to M.C., M.C. is an AHFMR Scholar and an HSF Investigator. References [1] S.M. Rothman, J.W. Olney, Glutamate and the pathophysiology of hypoxic–ischemic brain damage, Ann. Neurol. 19 (1986) 105–111. [2] D.W. Choi, Glutamate neurotoxicity and diseases of the nervous system, Neuron 1 (1988) 623–634. [3] D.W. Choi, M. Maulucci-Gedde, A.R. Kriegstein, Glutamate neurotoxicity in cortical cell culture, J. Neurosci. 7 (1987) 357–368. [4] L.S. Deshpande, J.K. Lou, A. Mian, R.E. Blair, S. Sombati, E. Attkisson, R.J. Delorenzo, Time course and mechanism of hippocampal neuronal death in an in vitro model of status epilepticus: role of NMDA receptor activation and NMDA dependent calcium entry, Eur. J. Pharmacol. 583 (2008) 73–83.
[5] S. Sombati, R.J. Delorenzo, Recurrent spontaneous seizure activity in hippocampal neuronal networks in culture, J. Neurophysiol. 73 (1995) 1706– 1711. [6] R.J. DeLorenzo, S. Pal, S. Sombati, Prolonged activation of the N-methyl-Daspartate receptor-Ca2+ transduction pathway causes spontaneous recurrent epileptiform discharges in hippocampal neurons in culture, Proc. Natl. Acad. Sci. U.S.A. 95 (1998) 14482–14487. [7] G. Curia, D. Longo, G. Biagini, R.S. Jones, M. Avoli, The pilocarpine model of temporal lobe epilepsy, J. Neurosci. Methods 172 (2008) 143–157. [8] U. Madsen, T.B. Stensbol, P. Krogsgaard-Larsen, Inhibitors of AMPA and kainate receptors, Curr. Med. Chem. 8 (2001) 1291–1301. [9] D.A. Wagenaar, R. Madhavan, J. Pine, S.M. Potter, Controlling bursting in cortical cultures with closed-loop multi-electrode stimulation, J. Neurosci. 25 (2005) 680–688. [10] M.A. Colicos, B.E. Collins, M.J. Sailor, Y. Goda, Remodeling of synaptic actin induced by photoconductive stimulation, Cell 107 (2001) 605–616. [11] Y. Goda, M.A. Colicos, Photoconductive stimulation of neurons cultured on silicon wafers, Nat. Prot. 1 (2006) 461–467. [12] H. Khosravani, C. Altier, G.W. Zamponi, M.A. Colicos, The Arg473Cysneuroligin-1 mutation modulates NMDA mediated synaptic transmission and receptor distribution in hippocampal neurons, FEBS Lett. 579 (2005) 6587– 6594. [13] A.Y. Abramov, C. Fraley, C.T. Diao, R. Winkfein, M.A. Colicos, M.R. Duchen, R.J. French, E. Pavlov, Targeted polyphosphatase expression alters mitochondrial metabolism and inhibits calcium-dependent cell death, Proc. Natl. Acad. Sci. U.S.A. 104 (2007) 18091–18096. [14] J. Hung, M.A. Colicos, Astrocytic Ca(2+) waves guide CNS growth cones to remote regions of neuronal activity, PLoS One 3 (11) (2008) e3692, doi:10.1371. [15] I. Schoen, P. Fromherz, The mechanism of extracellular stimulation of nerve cells on an electrolyte-oxide-semiconductor capacitor, Biophys. J. 92 (2007) 1096–1111. [16] A. Starovoytov, J. Choi, H.S. Seung, Light-directed electrical stimulation of neurons cultured on silicon wafers, J. Neurophysiol. 93 (2005) 1090–1098. [17] G.L. Rintoul, A.J. Filiano, J.B. Brocard, G.J. Kress, I.J. Reynolds, Glutamate decreases mitochondrial size and movement in primary forebrain neurons, J. Neurosci. 23 (2003) 7881–7888. [18] I.J. Reynolds, G.L. Rintoul, Mitochondrial stop and go: signals that regulate organelle movement, Sci. STKE (2004) 2004. [19] T. Brustovetsky, V. Li, N. Brustovetsky, Stimulation of glutamate receptors in cultured hippocampal neurons causes Ca2+ -dependent mitochondrial contraction, Cell Calcium 46 (2009) 18–29. [20] K. Okamoto, J.M. Shaw, Mitochondrial morphology and dynamics in yeast and multicellular eukaryotes, Annu. Rev. Genet. 39 (2005) 503–536. [21] P. Bernardi, A. Krauskopf, E. Basso, V. Petronilli, E. Blachly-Dyson, F. Di Lisa, M.A. Forte, The mitochondrial permeability transition from in vitro artifact to disease target, FEBS J. 273 (2006) 2077–2099. [22] M. Crompton, H. Ellinger, A. Costi, Inhibition by cyclosporin A of a Ca2+ dependent pore in heart mitochondria activated by inorganic phosphate and oxidative stress, Biochem. J. 255 (1988) 357–360. [23] D.A. McCormick, D. Contreras, On the cellular and network bases of epileptic seizures, Annu. Rev. Physiol. 63 (2001) 815–846. [24] A.L. Nieminen, T.G. Petrie, J.J. Lemasters, W.R. Selman, Cyclosporin A delays mitochondrial depolarization induced by N-methyl-D-aspartate in cortical neurons: evidence of the mitochondrial permeability transition, Neuroscience 75 (1996) 993–997. [25] A.P. Halestrap, A.M. Davidson, Inhibition of Ca2(+)-induced large-amplitude swelling of liver and heart mitochondria by cyclosporin is probably caused by the inhibitor binding to mitochondrial-matrix peptidyl-prolyl cis-trans isomerase and preventing it interacting with the adenin, Biochem. J. 268 (1990) 153–160. [26] J. Lee, S.E. Schriner, D.C. Wallace, Adenine nucleotide translocator 1 deficiency increases resistance of mouse brain and neurons to excitotoxic insults, Biochim. Biophys. Acta 1787 (2009) 364–370.
Contents lists available at ScienceDirect
Cell Calcium journal homepage: www.elsevier.com/locate/ceca
Development of an in vitro model of neuronal activity induced excitotoxicity using photoconductive stimulation Evgeny Pavlov b , R. Carolina Gutiérrez a , Yuan Zhang a , Audrey C. Kertesz a , Johanna Hung a , Ferdinand Joseph Espina a , Michael A. Colicos a,∗ a b
Department of Physiology & Pharmacology and the Hotchkiss Brain Institute, University of Calgary, Calgary, AB, Canada T2N 4N1 Department of Physiology and Biophysics, Dalhousie University, Halifax, NS, Canada B3H 1X5
a r t i c l e
i n f o
Article history: Received 21 September 2009 Received in revised form 10 March 2010 Accepted 25 March 2010
Keywords: Cell death Neuron Excitotoxicity Cyclosporin A Mitochondria Permeability transition
a b s t r a c t Loss of the ability to regulate calcium is a central event leading to neuronal cell death during a wide range of pathological conditions including stroke and seizure. Here we present a new dissociated hippocampal cell culture model of acute electrical activity which incorporates the photoconductive stimulation of neuronal networks grown on silicon wafers. This technology allows precise modeling of user defined neuronal activity patterns, and the study of their effect on neuronal physiology. Here, seizure-like conditions were created by continuous stimulation, causing hundreds of neurons to fire synchronously at 50 Hz for 4 min. This stimulation protocol induced cell death as monitored by propidium iodide staining. The number of dead cells per stimulation region increased from 3.6 ± 2.1 preceding stimulation to 81 ± 21 30 min following stimulation. Excitotoxicity primarily affected excitatory rather than inhibitory neurons, and was preceded by an increase in intracellular calcium as well as changes in the mitochondrial membrane potential, as measured by a tetramethylrhodamine methyl ester (TMRM) assay. Cyclosporin A (CsA), a mitochondrial permeability transition pore (PTP) blocker, was effective in preventing cell death. We propose that photoconductive stimulation is a useful tool for investigating the pathogenesis of excitotoxicity in vitro. © 2010 Elsevier Ltd. All rights reserved.
1. Introduction Excessive intracellular calcium is a primary trigger of neuronal cell death in a variety of neuropathological conditions. The use of cell culture models which mimic conditions of calcium overload has been critical for detailed studies of the processes leading to such stress-induced cell death. For example, dissociated cultures are widely used to model neuronal glutamate toxicity following stroke. In the brain, stroke induces calcium deregulation through release of glutamate from damaged neurons which in turn activates NMDA receptors and causes uncontrolled calcium increase followed by massive neuronal death [1,2]. In culture models glutamate is added directly to the tissue culture mimicking the condition likely experienced by neurons in the brain [3], where glutamate causes permanent activation of NMDA receptors and calcium influx accompanied by the loss of neuronal firing. In contrast, during seizures neurons undergo massive intracellular calcium increase caused by constant excessive neuronal electrical
∗ Corresponding author at: Department of Physiology & Pharmacology, University of Calgary, 3330 Hospital Drive NW, Calgary, AB, Canada T2N 4N1. Tel.: +1 403 220 4594; fax: +1 403 283 7137. E-mail address: [email protected] (M.A. Colicos). 0143-4160/$ – see front matter © 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.ceca.2010.03.005
activity. Thus the key requirement in developing a culture model of seizure is the induction of continuous neuronal firing in these cells. Several in vitro cell culture paradigms have been used to investigate the pathophysiological mechanisms underlying excitotoxicity in the context of hippocampal cell death following seizure (for review see [4,3]). Global neuronal firing can be enhanced chemically by placing cells into solution with low magnesium [5,6], pilocarpine [7], kainic acid [8], or other drugs. Another approach to the development of a culture model for seizure activity has focused on the use of direct electrical stimulation of neuronal cultures, which would not require chemical intervention and induce higher and more controlled firing rates [9]. While multi-electrode arrays are a very useful tool for this, they have the disadvantage being costly, and it is impractical to simultaneously grow large numbers of parallel cultures using this technology. We have taken an alternate strategy based on photoconductive stimulation of neurons grown on silicon wafers. Multiple wafers can be generated from a single animal and maintained in standard 24 well culture plates, and stimulation can be performed on an upright fluorescence microscope. By targeting a specific region of the neuronal culture with light, the local conductivity of the underlying silicon increases. Then, by applying an electrical stimulus to the wafer, the majority of current passes through the illuminated region, specifically firing the targeted subset of neurons. Thus a highly
442
E. Pavlov et al. / Cell Calcium 47 (2010) 441–448
precise area of stimulation can be achieved; and also allows internal controls to be recorded in unstimulated regions of the same wafer. In this report we use high-frequency photoconductive stimulation of neurons to recreate the conditions which are experienced by neurons during acute electrical hyperactivity. We found that the pathological changes we observed closely resemble changes seen in neurons in the intact brain under neuropathological conditions. This includes a significant increase in calcium accumulation, which if not restored results in neuronal cell death. We also demonstrate how a biochemical mechanism contributing to this cell death (opening of the mitochondrial permeability transition pore) can be directly monitored, and therapeutically blocked. Additionally, we characterize the cell death in terms of the general neuronal cell types involved (inhibitory versus excitatory neurons), addressing a question of subtype susceptibility. We conclude that this in vitro model allows real-time dynamic study of neuronal physiology during extreme activity. As it does not involve chemical intervention, this model permits large-scale screening for therapeutic strategies for treatment of excitability related neuronal damage such as caused by seizures.
immersion objective. Images were captured using custom built video equipment based on a WATEC N105 CCD camera, and AMCAP (Microsoft) software, and processed using ImageJ (NIH). 2.4. Network activity analysis In order to determine whether self-sustaining activity could be triggered by brief high-frequency stimulation, neuronal firing over the entire field of view was monitored for at least 10 min following a 1 s, 50 Hz pulse. Images of triggered neuronal activity were made at 15 frames per second using cultures loaded with the calcium dye Fluo-4-AM. For staining, 2 g of Fluo-4 were added to cells on a wafer in 500 l of recording solution and incubated at 37 ◦ C for 30 min prior to recording. Individual neurons were tracked over time and their activity traces extracted by calculating the mean grey value in an ROI encompassing the cell body. These activity traces were analyzed by custom software to determine the firing times of each neuron in the field of view. All neurons during the 10 min recording were simultaneously analyzed for their spatial and temporal correlations using custom written routines in Matlab. In all experiments light intensity was kept to the minimum in order to avoid photobleaching.
2. Materials and methods 2.5. Immunohistochemistry 2.1. Dissociated neuronal co-cultures Cultures suitable for photoconductive stimulation were prepared as described [10]; briefly, hippocampi were dissected from P0 (newborn) Sprague–Dawley rat pups, and then neurons and glia were dissociated and plated together on pretreated Si wafers [11]. Cultures were maintained in Eagle’s Basal Media (BME, GIBCOInvitrogen) supplemented with B-27, penicillin, streptomycin and l-glutamine. Functional synapses formed after a few days in culture, and networks were left to develop for approximately 2 weeks, at which point robust spontaneous activity could be observed. For experiments that required construct expression, cultures were transfected using Lipofectamine 2000 (Invitrogen) the day before use. 2.2. Photoconductive stimulation-induced acute neuronal activity Photoconductive stimulation of rat hippocampal neuronal cultures grown on silicon wafers was performed as previously described [11]. A ∼3 mm diameter field of view (which encompasses hundreds of neurons) was illuminated with 434 nm light on an Olympus BX60WI microscope using a 10× water immersion objective. As a general parallel of activity occurring during status epilepticus (SE), 50 Hz stimulation was used and continuously applied to the neurons for 4 min. Although in current work we limited our experiments to only this “acute stimulation” protocol, any profile of frequencies, duration, cycles or intensity can be programmed. Running experiments with illumination but without current applied to fire the neurons resulted in no cell death above baseline levels, directly differentiating activity driven death from phototoxic effects. In addition, experiments were done using more “normal” paradigms of neuronal activity, 10–50 Hz stimulation for 10 s, repeated after an interstimulus interval of a few minutes. No changes in the mitochondrial morphology or the membrane potential indicated by TMRM labeling were detectable (data not shown). 2.3. Video imaging Live video imaging was performed on an Olympus BX60WI microscope with either a 100× 1.00 NA or a 60× 0.90 NA water
To correlate cell death with neuronal cell type, cultures were stimulated using the SE protocol (50 Hz, 4 min) and subsequently stained with propidium iodide (Sigma). Predominantly inhibitory neurons were identified using anti-GAD65 antibodies (Chemicon, AB5082) and by morphology. Immunocytochemistry was carried out as described previously [12]. Briefly, the cultures were fixed using 4% paraformaldehyde for 20 min. Antibody detection was carried out in a low permeabilization buffer (0.01% Triton X 100) using a 1:1000 dilution of the GAD primary antibody and 1:100 donkey anti-rabbit Cy5 antibody (Jackson Immunoresearch Laboratories) for visualization. 2.6. Mitochondrial membrane potential assay Mitochondrial membrane potential dissipation was monitored by measuring TMRM fluorescence as previously described [13]. In order to do this TMRM (20 nM, Molecular Probes) was added to the recording dish and allowed to equilibrate for 20 min prior to the beginning of measurement. Fluorescence intensity of TMRM during continuous stimulation was measured from regions enriched in mitochondria and from control cytoplasm regions and quantified using ImageJ (NIH). 2.7. Mitochondrial morphology changes Mitochondrial morphology was monitored by fluorescent microscopy using mitochondrially targeted green fluorescent protein (GFP). pAcGFP1-Mito cDNA (MTS-GFP, Clontech) encodes a fusion of a mitochondrial targeting sequence derived from the precursor of subunit VIII of human cytochrome C oxidase and GFP. Cultured cells were transfected with the DNA using PolyFect® Transfection Reagent (Qiagen) according to the manufacturer’s recommended protocol using 1 g of DNA per 2 ml of media. 2.8. Cell death assay Cell death was assayed using propidium iodide staining as described previously [13]. Propidium iodide was added to the recording solution at concentration of 20 M. Dead cells, identified from propidium iodide staining of the nucleus, were counted
E. Pavlov et al. / Cell Calcium 47 (2010) 441–448
443
Fig. 1. Induction of spontaneous self-sustaining activity in dissociated neuronal cultures by photoconductive stimulation. Neuronal activity was visualized by imaging calcium influx with Fluo-4. After a control period of 4 min, a 1 s, 50 Hz stimulation was delivered to the observed region of the culture, initiating a sustained cascade of rapid activity. (A) Sample frames from the time course, beginning at the start of the control period (t = 0). Stimulation occurs at t = 4, after which firing of individual neurons can be seen, as well as global depolarizations of underlying glial cells (more easily seen at this magnification) caused by the high level of overlying neuronal activity in the neurons. Glial depolarizations were not included in the data analysis. Scale bar = 50 m. (B) Spatial and temporal representation of firing events 1 min before and 2 min after stimulation of the observed region of the culture. Stimulation point is indicated by arrow. Firing events were isolated by peak detection, and translated to the 3D graph. Firing events are linked if they occur within a specific time of each other for visualization. Immediately after the 1 s stimulation extensive activity is initiated and sustained for at least several minutes. This effect was consistent, and quantified in n = 3 cases.
in the central field of the region undergoing stimulation and in 4 adjacent fields (10×). 2.9. Statistical analysis Statistical analysis was done using t-tests and the online statistics calculator from GraphPad Software, Inc. 3. Results 3.1. Spontaneous activity in neuronal cultures Cultures of primary neurons have been well documented to have complex connectivity and to express a wide variety of neuronal firing patterns, including robust spontaneous activity and bursting behaviors. To establish our culture methodology as appropriate
for the modeling of neuronal network behavior, we determined whether DIV14 dissociated neuronal cultures had sufficiently complex connectivity to allow self-sustaining activity in response to a brief (1 s) high-frequency (50 Hz) pulse. To induce the activity we used photoconductive stimulation, a technique that uses the change in conductivity of illuminated silicon to target a capacitive induction of neuronal firing [11]. This stimulation technique results in the induction of action potentials indistinguishable from spontaneous firing [14]. To monitor neuronal activity we visualized action potentials using the calcium indicator Fluo-4 (Fig. 1A). In an unperturbed DIV14 network, a low level of spontaneous activity was observed (Fig. 1B). Upon applying a 1 s, 50 Hz stimulation to a large population of neurons, a massive increase in activity was observed (Fig. 1B). This activity was completely self-sustaining, continuing for the extent of a 10 min recording. This finding is in agreement with previous reports using other initiators of such activity [4]. This
444
E. Pavlov et al. / Cell Calcium 47 (2010) 441–448
Fig. 2. Disruption of intracellular calcium homeostasis by a SE model leads to neuronal death. Neurons were stimulated at 50 Hz for 4 min. Each trace corresponds to Fluo-4 fluorescence from an individual cell body. Cell death was visualized by propidium iodide (PI), and cells were grouped according to their PI signal following stimulation. (A) Fluorescence from cells which were PI positive immediately after the end of the stimulation. Note the abrupt drop in Fluo-4 fluorescence which presumably indicates loss of the dye due to the cell membrane permeabilization. (B) Fluorescence from the cells which became PI positive 30 min after the end of stimulation. (C) Fluorescence from the cells which survived stimulation. Note that these cells maintained their ability to regulate intracellular calcium. (D) PI signal of cells before stimulation (left), PI staining immediately following 4 min of stimulation (center) and PI staining 30 min after stimulation (right). Scale bar is 20 m. (E) Number of dead cells per image over 3 time points, illustrating that both acute and protracted cell death can be observed (t-test, p = 0.002, n = 3).
Fig. 3. Mitochondrial shape change in response to stimulation. Neurons with labeled mitochondria were monitored continuously during a 4 min, 50 Hz stimulation. (A) t = 0 min, (B) t = 2 min, and (C) t = 4 min. Note the transition of the shape of mitochondria from elongated structures typical to normal neurons to a smaller, punctate form. This result was observed in every case (n = 12). Scale bar is 8 m. (D) Statistics were performed by measuring the length of mitochondria from 3 independent experiments, in each experiment maximal length of 100 individual mitochondria were measured and results combined together. Neurons become noticeably smaller upon stimulation. Bin size is 1.5 m, error bars are SD.
E. Pavlov et al. / Cell Calcium 47 (2010) 441–448
445
confirms that in our experiments the neuronal networks formed by dissociated cultured neurons are inherently capable of robust, self-sustaining activity. 3.2. Photoconductive stimulation driven activity induces an increase in intracellular calcium and cell death To assess the effects of excessive neuronal activity on intracellular calcium homeostasis, we induced continuous neuronal firing by photoconductive stimulation. A brief stimulation depolarizes neurons and results in the generation of action potentials and an accompanying increase in cytoplasmic calcium concentration through voltage-gated calcium and NMDA receptor channels. Under normal firing conditions induced by brief stimulation, calcium concentrations quickly decrease back to pre-stimulation levels without affecting cell viability [11,15,16]. Seizure conditions, however, are characterized by an uncontrolled increase in the duration and frequency of neuronal firing. To recreate these conditions in our experiments, we stressed the neuronal network using continuous photoconductive stimulation for 4 min, to ensure continuous neuronal activity in the entire network. We observed that continuous 50 Hz stimulation caused a rapid, substantial increase in intracellular Ca2+ concentration in some neurons, as detected by Fluo-4 fluorescence. Among these, three distinct subpopulations were observed (Fig. 2A–C). Cells in Fig. 2A–C were assigned to each category based on their resistance to the cell death as monitored by PI staining. Specifically cells at panel (A) were PI positive immediately after the end of stimulation. Cells at panel (B) were PI negative immediately after the end of stimulation but developed PI staining 30 min later, while cells at panel (C) were PI negative 30 min after stimulation, During the stimulation event the intracellular Ca2+ of some neurons dropped rapidly after a few minutes (Fig. 2A), while another group maintained elevated Ca2+ levels throughout the time course of stimulation (Fig. 2B). In addition, a subpopulation of neurons continued normal neuronal activity after the stimulation episode without losing Ca2+ homeostasis (Fig. 2C). We also quantified the extent of cell death both immediately after (4 min) and 30 min following stimulation (Fig. 2D). We observed that the stimulation protocol caused substantial neuronal cell death from 3.6 ± 2.1 dead cells per image before stimulation to 30 ± 6 cells after (p = 0.002, n = 3), as determined by PI staining (Fig. 2E). The number of PI positive cells further increased to 81 ± 21 cells per image (p = 0.02, n = 3) during the 30 min following stimulation (Fig. 2E, n = 13 independent experiments). The neurons that were PI positive immediately after stimulation belonged to the subpopulation characterized by a complete drop in Fluo-4 fluorescence during stimulation. This suggests that the drop in fluorescence was caused by membrane permeabilization and the release of the Ca2+ bound indicator (see Fig. 2A). On the other hand, cells, which did not display PI staining even 30 min after the end of stimulation were able to maintain normal calcium levels (Fig. 2C). This demonstrates that neuronal cell death caused by acute stimulation strongly correlates with disruption of calcium homeostasis. 3.3. Continuous stimulation results in loss of mitochondrial membrane potential and changes in mitochondrial morphology A change in mitochondrial morphology from elongated to spherical shape is a key event following glutamate-induced calcium toxicity [17–19]. To investigate whether a similar phenomenon occurs during extreme neuronal activity we monitored mitochondrial shape using targeted GFP protein (Fig. 3). We observed that constant neuronal activity initiated a change in mitochondrial shape, from elongated to spherical (n = 12 independent experiments), similar to previous observations in glutamate stress
Fig. 4. Loss of mitochondrial membrane potential in the neurons following SE. Mitochondrial membrane potential was monitored using TMRM, a potential sensitive dye. (A) Prior to stimulation, mitochondria show visible TMRM signal. (B) Following SE stimulation, TMRM fluorescence rapidly dissipated, indicating a loss of membrane potential. Scale bar is 8 m. (C) TMRM signal collected from a single mitochondrial and a single cytoplasmic region. A decline in mitochondrial fluorescence corresponds with an increase in cytoplasmic fluorescence, suggesting loss of mitochondrial membrane integrity. (D) The ratio of mitochondrial to cytoplasmic TMRM signal declines after stimulation (n = 7 cells in each condition). All ratios were significantly different (t-test, p < 0.002), error bars are SEM.
paradigms (for review see [17,20]). This suggests that, in our experiments, a change in mitochondrial function could play a central role in neuronal death. Opening of the calcium-induced permeability transition pore (PTP) is known to be the central event in the process of necrotic cell death induced by calcium toxicity [21]. A large body of experimental data suggests that PTP opening is responsible for cell death in pathological conditions ranging from ischemia-reperfusion injury in heart to neuronal glutamate toxicity in the brain. Since the neuronal death in our experiments showed a strong correlation with an uncontrolled increase in calcium concentration and a dramatic
446
E. Pavlov et al. / Cell Calcium 47 (2010) 441–448
Fig. 5. Cyclosporin A protects cells from death and prevents calcium deregulation. To confirm that PTP opening is causal in the neuronal cell death observed, CsA (a PTP blocker) was added during an SE stimulation experiment. (A) Propidium iodide stained neurons 30 min after SE stimulation in the absence and (B) in the presence of CSA. Scale bar is 20 m. (C) The number of dead cells in a single field (∼0.64 mm) decreased significantly with CsA treatment (15 ± 5 compared to 60 ± 14 in control cultures (n = 3 cultures for each condition, t-test, p = 0.006), error bars are SEM). (D) Sample traces of Fluo-4 signal in neurons during SE stimulation in the presence of CsA. Few cells lost calcium homeostasis as indicated by most showing normal calcium oscillations.
change in mitochondrial morphology, we tested the hypothesis that neuronal death was caused by PTP activation. PTP opening leads to inner mitochondrial membrane depolarization and can be detected using the potential sensitive dye TMRM [13]. TMRM is a charged fluorescent dye which accumulates in intact mitochondria due to the presence of the negative potential in their inner membrane. Upon mitochondrial depolarization triggered by PTP opening, TMRM is released from mitochondria into the cytoplasm which leads to dissipation of the fluorescent signal. Fig. 4A and B depicts images of mitochondria loaded with TMRM before and after stimulation, and show that continuous stimulation leads to an abrupt decrease in fluorescent signal from mitochondrial regions (arrowheads), which suggests opening of the PTP (n = 9 independent experiments). At the whole-cell level, TMRM fluorescence from mitochondria progressively decreases with the duration of stimulation, and this process coincides with increase in cytoplasmic fluorescence due to release of TMRM from mitochondria (Fig. 4C and D). In these experiments, the ratio between fluorescence of mitochondrial to cytoplasmic regions progressively decreased from 0.97 ± 0.05 to 0.77 ± 0.12 (p = 0.002, n = 7) within 20 s following the beginning of neuronal firing and further decreased to 0.52 ± 0.1 (p = 0.001, n = 7) at 40 s, confirming mitochondrial depolarization. Importantly the time from the beginning of stimulation to onset of mitochondrial depolarization varied from experiment to experiment most likely corresponding to the variation in time required for mPTP to be activated. This variation is consistent with variation in time in onset of calcium deregulation (see Fig. 2A–C). 3.4. Prevention of stimulation-induced neuronal death To further confirm the link between calcium overload, PTP opening and cell death we compared the number of dead cells
when stimulation was done in the presence and in the absence of cyclosporin A (CsA)—a blocker of mitochondrial PTP [22]. In these experiments we first induced seizure-like activity (50 Hz for 4 min) in a region of neurons in the absence of CsA and estimated the number of dead cells 30 min after stimulation using PI staining. Next, CsA at a concentration of 5 M was added to the bath and another region of the same chip was stimulated in the same way followed by counting of PI positive cells. The use of the same chip ensured that stimulation conditions were identical in the presence and in the absence of CsA. As can be seen from Fig. 5A and B the number of dead cells in the presence of CsA (15 ± 5) was significantly lower than in control cells (60 ± 14) (p = 0.006, 3 independent experiments). Furthermore, in the presence of CsA, most of the cells demonstrated normal activity with typical reversible fluctuations of the cytoplasmic calcium concentrations (Fig. 5C). We conclude that extreme seizure-like conditions induce opening of the calcium-dependent PTP and thus cell death, which can be prevented by CsA treatment. 3.5. Inhibitory neurons are more resistant to seizure-induced cell death than excitatory neurons Seizure activity has been speculated to alter the balance between inhibitory and excitatory neurons in the brain (for review see [23]). One of the advantages of our in vitro model system is that we can directly assess changes in the inhibitory and excitatory balance following acute electrical stimulation. After performing the stimulation protocol in the presence of PI, we fixed the neurons and probed them with anti-GAD65 antibody (Fig. 6A) to identify inhibitory neurons, as well as with DAPI to determine the total cell population (Fig. 6B). We determined that while approximately 18 ± 3% of the total neuronal population died, the percentage of dead inhibitory neurons was only 7 ± 0.3% (statistically different at
E. Pavlov et al. / Cell Calcium 47 (2010) 441–448
447
4. Discussion Neuronal death caused by hyper-excitation and excessive neuronal activity is a common factor in several neuropathological conditions including traumatic head injury, stroke and seizures. Preventing the acute loss of brain cells under these conditions is an important goal in developing therapeutic strategies to limit the risk of permanent brain damage. Here we introduce a new technology designed to reproduce the physiological stress that occurs in neurons during acute electrical activity similar to that seen under these conditions, and demonstrate its effectiveness in characterizing critical cell death processes. The main advantage of the present model is that it allows the study of cell damage caused exclusively by electrical stimulation without the use of any chemical stimulation. By using a silicon-based interface with the neuronal cultures, we can accurately specify any combination or progression of neuronal activity while concurrently monitoring the electrophysiological and biochemical status of the neurons before, during and after the firing event. In this report we use our model to recreate the neuronal duress caused by excessive electrical activity, using a generalized activity profile similar to that which is observed during seizures. It is important to note that while our current study focused on a simple 4 min, 50 Hz stimulation protocol, by using photoconductive stimulation, one can apply any stimulation pattern or combination of patterns the network. This flexibility allows researchers complete and precise control over the electrical activity they wish to investigate. This flexibility currently cannot be achieved either in existing culture nor whole-animal models of seizure. For example, low magnesium based culture models of seizure can only create conditions of continuous firing at relatively low frequency (10 Hz) resulting in substantial cell death only after hours of activity. Whole-animal models, on the other hand, where seizure activity is induced by chemicals such as pilocarpine cause excessive neuronal activity leading to massive amounts of cell death, a situation which is not paralleled in most human seizure. Also, while techniques such as field stimulation could provide a similar global level of activity, the lack of spatial resolution (to prohibit the firing of a subset of neurons) precludes the use of internal controls. Our experiments demonstrate the effectiveness of this model for the study of seizures in three critical areas:
Fig. 6. SE stimulation differentially affects inhibitory and excitatory neuron cell death. Following stimulation, cultures were fixed and inhibitory neurons identified using an anti-GAD-65 antibody. Immunocytochemistry protocol was optimized to detect cell bodies of inhibitory cells. (A) Representative image of the GAD-65 immunoreactivity in cell bodies of DIV 10 cells. (B) Triple label of inhibitory neurons using GAD-65 (green), DAPI (blue) to provide the total number of cells in the cultures, and PI (red) to detect dead cells. (C) GAD-65 neurons were less vulnerable to cell death than the general neuronal population (n = 16 regions from 3 different experiments, t-test, p = 0.003). Scale bars A = 10 , B = 100 .
p = 0.003, Student’s t-test) of the total number of inhibitory neurons (Fig. 6C), taken from 3 independent experiments using cells from 3 different culture rounds. Percentages were calculated from a total population of 4215 cells, 278 of which were inhibitory. This suggests that excitatory neurons have a 2–3-fold higher rate of cell death than inhibitory neurons following acute electrical stimulation.
(1) Studies of intracellular processes during acute electrical activity. Excessive accumulation of Ca2+ inside neurons during seizure is the major cause of toxicity leading to cell death (see [24] for review). Although influx of calcium is known to create favorable conditions for opening of the mitochondrial PTP, it is not clear whether PTP opening is a cause or a consequence of cell death. Here we demonstrate that prolonged, seizure-like stimulation induces substantial Ca2+ accumulation, followed by PTP opening leading to cell death, and that this can be prevented by CsA, a blocker of PTP [25]. We suggest that in the case of acute neuronal electrical activity, the major event which causes cell death is permeability transition pore (PTP) opening induced by massive calcium accumulation inside mitochondria. Our observation is consistent with recent report on the potential involvement of mPTP in neuronal death caused by exposure of mice to kainic acid [26]. To our knowledge, the present study is the first direct demonstration at the level of primary culture that prolonged neuronal firing is sufficient to induce neuronal cell death through a PTP-dependent mechanism. We propose that the most likely scenario of the cell death seen in our experiments is the following: (I) continuous stimulation leads to repetitive activation of voltage-gated Ca2+ channels; (II) opening of voltage-gated and NMDA receptor calcium channels causes substantial Ca2+ uptake, which dramatically increases its concentration in the cytoplasm; (III) the increase in cytoplasmic
448
E. Pavlov et al. / Cell Calcium 47 (2010) 441–448
Ca2+ leads to its massive Ca2+ accumulation by mitochondria, causing PTP opening. PTP opening directly leads to disruption of mitochondrial function and cell death. (2) Studies of various subpopulations of neuronal cells. One of the critical questions which can be addressed is the relative susceptibility of different types of neurons to seizure-induced death. Here we were able to demonstrate that inhibitory neurons are more resistant to seizure-induced cell death compared to the total neuronal population. (3) A simple tool for drug screening. As an example we show that cell death can be prevented by CsA. Similarly, other drug candidates can be applied and tested neuroprotective effects. Identified candidates can later be used for further testing in animal models. In summary, the silicon-based excitotoxicity paradigm presented in this paper is an advanced method to assist the study of pathological events in neurons under user defined activity conditions while simultaneously monitoring neuronal physiology. As a highly scalable in vitro model, it allows studies to be performed which open unique possibilities to screen for drugs which will reduce neuronal damage. Moreover, as a culture based paradigm, unique genetic or environmental alterations are possible, allowing investigations such as the role of channelopathies or viral infections in seizure susceptibility. Using this model we also demonstrate the direct link between activity induced cell death and mitochondrial PTP opening, and importantly we demonstrate a methodology for preventing this cell death. We also demonstrate that the balance between excitatory and inhibitory neurons is altered following seizure, which is an important factor when considering the long term ramifications of seizure events on the systems level. Acknowledgements We thank R. Flynn for providing comments on the manuscript. Grants: This work was supported by CIHR grant #69-6113 to R. French and EP, HSFA support to EP, CIHR and AIF grants to M.C., M.C. is an AHFMR Scholar and an HSF Investigator. References [1] S.M. Rothman, J.W. Olney, Glutamate and the pathophysiology of hypoxic–ischemic brain damage, Ann. Neurol. 19 (1986) 105–111. [2] D.W. Choi, Glutamate neurotoxicity and diseases of the nervous system, Neuron 1 (1988) 623–634. [3] D.W. Choi, M. Maulucci-Gedde, A.R. Kriegstein, Glutamate neurotoxicity in cortical cell culture, J. Neurosci. 7 (1987) 357–368. [4] L.S. Deshpande, J.K. Lou, A. Mian, R.E. Blair, S. Sombati, E. Attkisson, R.J. Delorenzo, Time course and mechanism of hippocampal neuronal death in an in vitro model of status epilepticus: role of NMDA receptor activation and NMDA dependent calcium entry, Eur. J. Pharmacol. 583 (2008) 73–83.
[5] S. Sombati, R.J. Delorenzo, Recurrent spontaneous seizure activity in hippocampal neuronal networks in culture, J. Neurophysiol. 73 (1995) 1706– 1711. [6] R.J. DeLorenzo, S. Pal, S. Sombati, Prolonged activation of the N-methyl-Daspartate receptor-Ca2+ transduction pathway causes spontaneous recurrent epileptiform discharges in hippocampal neurons in culture, Proc. Natl. Acad. Sci. U.S.A. 95 (1998) 14482–14487. [7] G. Curia, D. Longo, G. Biagini, R.S. Jones, M. Avoli, The pilocarpine model of temporal lobe epilepsy, J. Neurosci. Methods 172 (2008) 143–157. [8] U. Madsen, T.B. Stensbol, P. Krogsgaard-Larsen, Inhibitors of AMPA and kainate receptors, Curr. Med. Chem. 8 (2001) 1291–1301. [9] D.A. Wagenaar, R. Madhavan, J. Pine, S.M. Potter, Controlling bursting in cortical cultures with closed-loop multi-electrode stimulation, J. Neurosci. 25 (2005) 680–688. [10] M.A. Colicos, B.E. Collins, M.J. Sailor, Y. Goda, Remodeling of synaptic actin induced by photoconductive stimulation, Cell 107 (2001) 605–616. [11] Y. Goda, M.A. Colicos, Photoconductive stimulation of neurons cultured on silicon wafers, Nat. Prot. 1 (2006) 461–467. [12] H. Khosravani, C. Altier, G.W. Zamponi, M.A. Colicos, The Arg473Cysneuroligin-1 mutation modulates NMDA mediated synaptic transmission and receptor distribution in hippocampal neurons, FEBS Lett. 579 (2005) 6587– 6594. [13] A.Y. Abramov, C. Fraley, C.T. Diao, R. Winkfein, M.A. Colicos, M.R. Duchen, R.J. French, E. Pavlov, Targeted polyphosphatase expression alters mitochondrial metabolism and inhibits calcium-dependent cell death, Proc. Natl. Acad. Sci. U.S.A. 104 (2007) 18091–18096. [14] J. Hung, M.A. Colicos, Astrocytic Ca(2+) waves guide CNS growth cones to remote regions of neuronal activity, PLoS One 3 (11) (2008) e3692, doi:10.1371. [15] I. Schoen, P. Fromherz, The mechanism of extracellular stimulation of nerve cells on an electrolyte-oxide-semiconductor capacitor, Biophys. J. 92 (2007) 1096–1111. [16] A. Starovoytov, J. Choi, H.S. Seung, Light-directed electrical stimulation of neurons cultured on silicon wafers, J. Neurophysiol. 93 (2005) 1090–1098. [17] G.L. Rintoul, A.J. Filiano, J.B. Brocard, G.J. Kress, I.J. Reynolds, Glutamate decreases mitochondrial size and movement in primary forebrain neurons, J. Neurosci. 23 (2003) 7881–7888. [18] I.J. Reynolds, G.L. Rintoul, Mitochondrial stop and go: signals that regulate organelle movement, Sci. STKE (2004) 2004. [19] T. Brustovetsky, V. Li, N. Brustovetsky, Stimulation of glutamate receptors in cultured hippocampal neurons causes Ca2+ -dependent mitochondrial contraction, Cell Calcium 46 (2009) 18–29. [20] K. Okamoto, J.M. Shaw, Mitochondrial morphology and dynamics in yeast and multicellular eukaryotes, Annu. Rev. Genet. 39 (2005) 503–536. [21] P. Bernardi, A. Krauskopf, E. Basso, V. Petronilli, E. Blachly-Dyson, F. Di Lisa, M.A. Forte, The mitochondrial permeability transition from in vitro artifact to disease target, FEBS J. 273 (2006) 2077–2099. [22] M. Crompton, H. Ellinger, A. Costi, Inhibition by cyclosporin A of a Ca2+ dependent pore in heart mitochondria activated by inorganic phosphate and oxidative stress, Biochem. J. 255 (1988) 357–360. [23] D.A. McCormick, D. Contreras, On the cellular and network bases of epileptic seizures, Annu. Rev. Physiol. 63 (2001) 815–846. [24] A.L. Nieminen, T.G. Petrie, J.J. Lemasters, W.R. Selman, Cyclosporin A delays mitochondrial depolarization induced by N-methyl-D-aspartate in cortical neurons: evidence of the mitochondrial permeability transition, Neuroscience 75 (1996) 993–997. [25] A.P. Halestrap, A.M. Davidson, Inhibition of Ca2(+)-induced large-amplitude swelling of liver and heart mitochondria by cyclosporin is probably caused by the inhibitor binding to mitochondrial-matrix peptidyl-prolyl cis-trans isomerase and preventing it interacting with the adenin, Biochem. J. 268 (1990) 153–160. [26] J. Lee, S.E. Schriner, D.C. Wallace, Adenine nucleotide translocator 1 deficiency increases resistance of mouse brain and neurons to excitotoxic insults, Biochim. Biophys. Acta 1787 (2009) 364–370.