Transient incubation of cultured hippocampal neurons in the absence of magnesium induces rhythmic and synchronized epileptiform-like activity

Cell culture models are important tools to study epileptogenesis mechanisms. The aim of this work was to characterize the spontaneous and synchronized rhythmic activity developed by cultured hippocampal neurons after transient incubation in zero Mg2+ to model Status Epilepticus. Cultured hippocampal neurons were transiently incubated with a Mg2+-free solution and the activity of neuronal networks was evaluated using single cell calcium imaging and whole-cell current clamp recordings. Here we report the development of synchronized and spontaneous [Ca2+]i transients in cultured hippocampal neurons immediately after transient incubation in a Mg2+-free solution. Spontaneous and synchronous [Ca2+]i oscillations were observed when the cells were then incubated in the presence of Mg2+. Functional studies also showed that transient incubation in Mg2+-free medium induces neuronal rhythmic burst activity that was prevented by antagonists of glutamate receptors. In conclusion, we report the development of epileptiform-like activity, characterized by spontaneous and synchronized discharges, in cultured hippocampal neurons transiently incubated in the absence of Mg2+. This model will allow studying synaptic alterations contributing to the hyperexcitability that underlies the development of seizures and will be useful in pharmacological studies for testing new drugs for the treatment of epilepsy.

activity recorded in the latter model becomes resistant to benzodiazepines after prolonged periods 3 . Therefore, neuronal cell cultures and hippocampal slices (both acute preparations and organotypic cultures) are considered valuable tools to study the cellular and molecular alterations in synaptic connectivity and in plasticity mechanisms associated with diseases of the brain, including epilepsy.
Most studies using neuronal cultures as a model to study epileptogenesis are mainly focused on the analysis of the response during incubation in a Mg 2+ -free salt solution, which enhances neuronal activity and may resemble the Status Epilepticus (SE) period (e.g. 6,14,15 ). In the present work we analyzed the alterations in the physiology of cultured hippocampal neurons incubated transiently in the absence of Mg 2+ ([Mg 2+ ] 0 ), and then returned to a [Mg 2+ ]-containing salt solution to model the period after SE. The data obtained by single cell calcium imaging analysis and whole-cell current clamp recordings showed that after induction of epileptogenesis, when neurons were returned to a solution containing magnesium, they were prone to develop spontaneous, recurrent and synchronous activity. This type of activity is comparable with the neuronal activity observed in the chronic phase of epilepsy, both in animal models of the disease and in human patients. Overall, the results of this study show that transient incubation of cultured hippocampal neurons in [Mg 2+ ] 0 can be used as a model of epileptogenesis. This model will be useful to investigate the synaptic mechanisms associated with epileptogenesis and to test new drugs for the treatment of epilepsy in different stages of the disease. 2+ ] 0 medium induces epileptiform activity. To validate the transient incubation in [Mg 2+ ] 0 medium as an experimental strategy to model Status Epilepticus in vitro we analyzed the frequency of action potential firing in high density primary cultures of rat hippocampal neurons, using whole-cell current clamp electrophysiology. Neuronal activity was also measured in hippocampal neurons incubated in control salt solution. An approximately threefold increase in the frequency of action potentials was observed in neurons incubated in [Mg 2+ ] 0 medium (Fig. 1A,B,G), confirming that this experimental condition induces the development of epileptiform discharges.

Incubation of hippocampal neurons in [Mg
In additional experiments we characterized the role of AMPA (AMPAR) and NMDA (NMDAR) receptors in the firing of action potentials in hippocampal neurons incubated in the absence of Mg 2+ . Hippocampal neurons were incubated in [Mg 2+ ] 0 medium for 15 min and afterwards whole-cell current clamp recordings were performed in the same medium and in the presence or absence of the following inhibitors: APV (50 μM; NMDAR antagonist) and/or CNQX (20 μM; AMPAR antagonist), and TTX (500 nM; blocker of voltage-gated Na + channels). The increase in the frequency of firing of action potentials in [Mg 2+ ] 0 medium when compared with the control (Fig. 1A,B,G) was abolished upon incubation with APV, CNQX or with the two drugs together ( Fig. 1C-E,G). As expected, TTX completely blocked the firing activity recorded in cells maintained in the absence of Mg 2+ (Fig. 1F,G).

Transient incubation in epileptogenic conditions induces synchronous neuronal activity and the development of spontaneous [Ca 2+
] i oscillations. An episode of continuous seizure activity is sufficient to induce temporal lobe epilepsy (TLE) in diverse mammalian species 16 , and accordingly the occurrence of de novo SE is thought to contribute to development of TLE in humans 17   The results were normalized for the mean frequency of action potential detected under control conditions. The results represent the fold change (mean ± SEM) of at least 3 independent experiments performed in distinct preparations. *p < 0.05, **p < 0.01, ***p < 0.001; one-way ANOVA followed by Bonferroni test. showed an initial increase in Fluo-4 fluorescence of ~128% (Fig. 3H). This initial response was followed by a slow decrease towards a plateau at ~80% above the resting Fluo-4 fluorescence (at 5-30 min; Fig. 3H), and the [Ca 2+ ] i further decreased upon incubation of the cells in control salt solution (Fig. 3H). In comparison, the cells that did not recover from the incubation period in [Mg 2+ ] 0 medium ( Fig. 3E and F) showed a similar (~114%) but sustained increase in Fluo-4 fluorescence when incubated under conditions that model SE (Fig. 3I), and the effects were maintained even after incubation with control salt solution (Fig. 3I)    ] i response to the incubation under the latter conditions (after the initial rapid increase) correlated well with the pattern of response when the epileptogenic stimulus was removed. Although the two protocols used to induce epileptiform-like activity, for 15 min and 30 min, induced spontaneous [Ca 2+ ] i transients ( Fig. 5A-C), there were clear differences in the results obtained. Thus, cells that underwent incubation for 30 min displayed a 56% shorter latency to develop spontaneous calcium oscillations when compared with cells that were incubated for only 15 min (Fig. 5D). The frequency of spontaneous calcium transients was 58% higher in cells that were subjected to 30 min of [Mg 2+ ] 0 medium ( Fig. 5E) but the events displayed a shorter duration (39% difference) under these conditions (Fig. 5F). These results indicate that the longer period of incubation in the absence of Mg 2+ induces more frequent but shorter-lasting spontaneous calcium transients. Interestingly, about 82% of the cells that were exposed to [Mg 2+ ] 0 medium for 15 min recovered the [Ca 2+ ] i to basal levels, while only about 46% of the cells showed the same type of behaviour upon a 30 min incubation (Fig. 5G). This suggests that the cells subject to the latter conditions were less efficient in maintaining the [Ca 2+ ] i homeostasis.

Characterization of the [Ca 2+ ] i transients induced by 15 minutes of incubation under epileptogenic conditions. In order to determine whether different incubation periods in [Mg
To further investigate the mechanisms underlying spontaneous [Ca 2+ ] i oscillations in hippocampal neurons subjected to transient incubation in [Mg 2+ ] 0 medium for 30 min, the cells were further incubated in control salt solution, to allow the development of calcium transients, before perfusion with tetrodotoxin (TTX), a blocker of voltage-gated Na + channels. Figure 5H shows that addition of TTX (1 µM) to the medium after the third spontaneous calcium transient completely abolished [Ca 2+ ] i oscillations, indicating that this phenomenon is dependent on neuronal activity.  www.nature.com/scientificreports/ Control experiments showed that the pattern of activity recorded in hippocampal neurons maintained always in control salt solution (with 2 mM Mg 2+ ) was stable and low throughout the recording period (Fig. 6A,E). Burst analysis using the Poisson Surprise method 18 showed that during the 25 min of incubation in control salt solution after transient perfusion with [Mg 2+ ] 0 medium there was an increase of rhythmic bursts when compared with the control condition (Fig. 6A,B,F).

Characterization of rhythmic electrical activity in hippocampal neurons transiently incubated
In additional experiments, we performed a pharmacological characterization of the bursts of action potentials in hippocampal neurons subjected to a transient incubation in the absence of Mg 2+ , followed by incubation in a solution containing a physiological [Mg 2+ ]. The inhibitors of ionotropic glutamate receptors, APV and CNQX, decreased bursting activity, but the effect was statistically significant only in the experiments performed in the presence of the non-NMDA receptor antagonist (Fig. 6C,D). Both the average number of bursts (Fig. 6E) and the intraburst frequency of events (Fig. 6F) were significantly reduced by CNQX.

Discussion
Neuronal cultures and brain slices are essential tools in the study of the cellular and molecular mechanisms underlying epileptogenesis since they allow studying alterations in synaptic connectivity in response to conditions of enhanced excitatory activity. Furthermore, they allow testing the effects of anti-epileptic drugs. One of the most common stimuli used to induce in vitro SE-like activity is the incubation of cultured neurons or brain slices in a Mg 2+ -free solution which enhances the activity of neuronal networks 3,5-12 . In this work we found that transient incubation of cultured hippocampal neurons under these conditions induces spontaneous and synchronized bursts of neuronal activity coupled to oscillations of the [Ca 2+ ] i . Since these bursts of activity take place after the trigger has been removed, they may be considered an in vitro mimetic of seizure activity following SE although with a distinct time scale. The pattern of [Ca 2+ ] i and electrophysiology responses (including spiking activity) observed in cultured hippocampal neurons subjected transiently to conditions that generate epileptiform-like activity, further validates the [Mg 2+ ] 0 model as an experimental strategy to study epileptogenesis in vitro. Furthermore, the periods of SE-like activity tested are appropriate not only to study the cellular and molecular alterations during this stage of the disease, but also to investigate long-term changes induced by SE and the development of chronic epilepsy.
Incubation of cultured hippocampal neurons in [Mg 2+ ] 0 medium enhanced the frequency of action potentials and increased rapidly the [Ca 2+ ] i as determined with the fluorescent indicator Fluo-4. The increased bursting activity under the same conditions was previously reported in hippocampal and cortical slices 9,10,12 , hippocampal organotypic cultures 11 and in hippocampal neuronal cultures 6,8 . In addition, we found that the network bursting observed in Mg 2+ -free solution was decreased in the presence of inhibitors of NMDA and non-NMDA (possibly AMPA) receptors, in accordance with previous observations 6 . The latter study also showed an increase in the probability of neurotransmitter release from nerve endings under the same conditions. The role of NMDA receptors in the increased excitability is likely to result from the effect of Mg 2+ as a blocker of the receptor channel from the outside of the cell at a resting membrane potential 3,19 . AMPA and NMDA receptor inhibitors also prevent the increased neuronal activity during Status Epilepticus 6,20 , similar to the results obtained in this work.
Importantly, after incubation of hippocampal neurons under conditions that model SE in vitro, around ~50% of the cells developed spontaneous [Ca 2+ ] i transients. This spontaneous activity occurred during a period when the cells were no longer incubated in [Mg 2+ ] 0 medium, under conditions that mimic the physiological extracellular [Mg 2+ ]. When hippocampal neurons were maintained in control salt solution throughout the entire experiment, no spontaneous oscillations were observed, indicating that this type of behaviour was not due to an artefact of the experimental settings, being instead evoked by the epileptogenic conditions. In addition, the spontaneous [Ca 2+ ] i oscillations were synchronous between the cells exposed to the SE stimulus and was blocked by TTX, an inhibitor of voltage-gated sodium channels, indicating that this process is dependent of neuronal activity. Accordingly, bursts of action potentials were also recorded in cultured hippocampal neurons in physiological [Mg 2+ ] after a period of incubation in the absence of the cation, with a frequency similar to the [Ca 2+ ] i oscillations. The increased electrical activity recorded in hippocampal neurons after a period that model SE in vitro was particularly sensitive to inhibition by the non-NMDA receptor antagonist CNQX, which affected the number of bursts and the intraburst activity. Since the pattern of neuronal activity after transient incubation in the absence of Mg 2+ was also sensitive to APV, an inhibitor of NMDA receptors, the effect of CNQX may be partly due to a decrease in AMPA receptor-mediated depolarization of the membrane which is required for activation of NMDA receptors 19 .
It is important to note that the frequency of action potentials decreased when hippocampal neurons were incubated in control salt solution after a period of exposure to conditions that model SE in vitro. This difference is probably due to the fact that after incubation in [Mg 2+ ] 0 the bursting activity alternates with interburst periods characterized by a lower frequency of AP firing. In the present work we also found (i) a significant reduction in bursting activity in the presence of CNQX or APV, and (ii) a greater decrease in frequency of APs in the presence of CNQX. These results contrast with those obtained in a previously reported study using hippocampal neurons exposed to [Mg 2+ ] 0 for 3 h and further incubated in culture medium supplemented with horse serum for 2 days. Under the latter conditions there was a reduction in spiking activity to approximately control levels in the presence of 25 μM APV, with no burst activity. Furthermore, although burst activity was observed in the presence of 10 μM CNQX it showed a shorter duration 20 . The latter response may arise from neuronal damage, as suggested by our results showing apoptotic death of hippocampal neurons after incubation in the absence of Mg 2+ for periods longer than 30 min. However, if this is not the case, the differences between the two sets of data may indicate that the mechanisms underlying the epileptiform discharges change along the time of incubation in Mg 2+ -containing medium. These changes may result from adaptative processes present in neurons that allow adjusting their excitability depending on the activity of neuronal networks 21 ] 0 developed calcium oscillations at a later point, when incubated in control salt solution. Although it is not clear why these cells showed a distinct behaviour, several hypothesis can be raised: (i) silent cells may be different since this culture is known to contain distinct types of neurons (e.g. glutamatergic and GABAergic neurons) 25,26 ; (ii) cells that did not develop [Ca 2+ ] i oscillations may lack the minimum number of synaptic contacts required for synchronization of the calcium responses; (iii) the expression of different Ca 2+ buffering mechanisms and/ or surface receptors for glutamate may account for responses to transient incubation in [Mg 2+ ] 0 . It is important to point out that the density of the neuronal culture is a crucial factor for this model. Indeed, a previous study showed that, despite manifesting increased EPSC frequency, cells from low-density cultures did not exhibit action potential bursting when maintained in zero magnesium conditions 6 .
Several studies have shown that the epileptogenic period is characterized by an alteration of [Ca 2+ ] i homeostatic mechanisms (reviewed in 27 ), but the intracellular calcium dynamics in SE is still poorly characterized. The increase in the [Ca 2+ ] i in SE may be coupled to activation of genes coding for growth factors, such as brainderived neurotrophic factor (BDNF) and fibroblast growth factor (FGF), and to changes in the expression of cytoskeletal proteins and in glutamate receptors 28,29 . Together, these alterations are thought to contribute to the development of epileptic circuits 30 . In accordance with their key role in epilepsy, voltage-dependent Ca 2+ channels, together with Ca 2+ -binding proteins, have been reported to be involved in all stages of the pathogenesis of epilepsy 31,32 . The key role of Ca 2+ in epileptogenesis suggests that the [Ca 2+ ] i rise, with a consequent activation of downstream signalling mechanisms, may be important in the induction of the bursts of synaptic activity and [Ca 2+ ] i oscillations after transient incubation of hippocampal neurons in [Mg 2+ ] o medium. This effect was more remarkable when longer periods in the absence of Mg 2+ were tested, and may explain the shorter lag-phase for initiation of the [Ca 2+ ] i transients upon removal of the buffer that induces SE-like activity, as well as their increased frequency. The distinct profiles of the spontaneous [Ca 2+ ] i oscillations depending on the duration of incubation in the absence of Mg 2+ may be due to differential activation of Ca 2+ -dependent signalling mechanisms during this period. The distinct duration of sustained Ca 2+ signals during the period of incubation in [Mg 2+ ] o medium may be important in determining which transcription factors are preferentially activated 33,34 . Also, the Ca 2+ -and calmodulin-dependent protein kinase II can undergo inactivation by autophosphorylation on threonine 305/306 after prolonged activity, with an impact on downstream mechanisms such as gene expression [35][36][37] .
The Fluo-4 calcium imaging in single cells also showed that the spontaneous [Ca 2+ ] i oscillations recorded after incubation of hippocampal neurons in [Mg 2+ ] o were synchronized. This synchronized activity may be explained based on the numerous synaptic connections established by hippocampal neurons in primary cultures (e.g. 38 ) and resemble the coordinated hyperactivity of a population of glutamatergic neurons that underlie seizures [39][40][41] . The excessive activity of neuronal networks may also arise from a deficient neuronal inhibition due to an insufficient GABA A receptor mediated neurotransmission 42,43,47 .
The results discussed above showed that incubation of hippocampal neurons under conditions that model SE induces a massive influx of Ca 2+ downstream of NMDA receptor activation. As expected, the longer the cells were exposed to SE the higher was cell death due to excitotoxic mechanisms resulting from an [Ca 2+ ] i overload. Excitotoxicity is characterized as a deleterious effect resulting from an excessive or prolonged activation of glutamate receptors by excitatory signals. This induces a multitude of deleterious effects, such as impairment of intracellular calcium homeostasis, compromised organelle functions, increase in nitric oxide and free radical production, persistent activation of proteases and kinases, increase in expression of pro-death transcription factors and immediate early genes 44 . In particular, NMDA receptors play a fundamental role in the development of excitoxicity due to their high Ca 2+ permeability 45,48 .
In summary, the work presented here shows that transient incubation of hippocampal neurons in [Mg 2+ ] 0 medium leads to the development of spontaneous bursts of activity after returning them to a salt solution containing Mg 2+ . This can be observed by recording action potentials in single cells with whole-cell current clamp electrophysiology and by measuring the [Ca 2+ ] i in populations of neurons. Whether similar responses can be induced by transient depolarization of cultured neurons with KCl, or other stimuli, remains to be investigated. These can be excellent tools to elucidate the molecular mechanisms underlying epileptogenesis at an early stage of the disease and to test new drugs for epilepsy.

Methods
Cultures of hippocampal neurons. Cultures of hippocampal neurons with a density of 8.0 × 10 4 cells/ cm 2 were prepared from Wistar rat embryos (E18-E19) 46 . Animals were obtained from the CNC animal facility, and the project was approved by the institutional Animal Ethics Committee (ORBEA) as well as by the the National authorities (Direcção-Geral de Alimentação e Veterinária) (References 0421/000/000/2013 and 0421/000/000/2020