Astrocyte switch to the hyperactive mode

Increasing pieces of evidence have suggested that astrocyte function has a strong influence on neuronal activity and plasticity, both in physiological and pathophysiological situations. In epilepsy, astrocytes have been shown to respond to epileptic neuronal seizures; however, whether they can act as a trigger for seizures has not been determined. Here, using the copper implantation method, spontaneous neuronal hyperactivity episodes were reliably induced during the week following implantation. With near 24‐h continuous recording for over 1 week of the local field potential with in vivo electrophysiology and astrocyte cytosolic Ca2+ with the fiber photometry method, spontaneous occurrences of seizure episodes were captured. Approximately 1 day after the implantation, isolated aberrant astrocyte Ca2+ events were often observed before they were accompanied by neuronal hyperactivity, suggesting the role of astrocytes in epileptogenesis. Within a single developed episode, astrocyte Ca2+ increase preceded the neuronal hyperactivity by ~20 s, suggesting that actions originating from astrocytes could be the trigger for the occurrence of epileptic seizures. Astrocyte‐specific stimulation by channelrhodopsin‐2 or deep‐brain direct current stimulation was capable of inducing neuronal hyperactivity. Injection of an astrocyte‐specific metabolic inhibitor, fluorocitrate, was able to significantly reduce the magnitude of spontaneously occurring neuronal hyperactivity. These results suggest that astrocytes have a role in triggering individual seizures and the reciprocal astrocyte‐neuron interactions likely amplify and exacerbate seizures. Therefore, future epilepsy treatment could be targeted at astrocytes to achieve epilepsy control.

. Several lines of evidence suggest astrocytes influence epilepsy both in humans and rodents (Bedner et al., 2015;Gómez-Gonzalo et al., 2010;Kang et al., 2005;Onodera et al., 2021;Tian et al., 2005).Upon seizure induction by kainic acid, astrocyte activity preceded seizures, and astrocytes were shown to have a role in controlling seizure propagation speed (Heuser et al., 2018).In partial epilepsy, aberrant neuronal hyperactivity propagates through the brain from a specific epileptic focus.In a series of human epilepsy studies, slow direct current (DC) potential change has been detected, which preceded epileptic discharges in selective electrodes close to the proposed epileptic focus (Ikeda et al., 2020).This DC shift has been proposed to be produced by astrocyte activity; thus, reactions emanating from the astrocytes could serve as a trigger for the aberrant neuronal hyperactivity.
Here, we expressed fluorescent resonance energy transfer (FRET) based Ca 2+ sensor protein, YC nano50 , specifically in astrocytes (Kanemaru et al., 2014) and detected the astrocyte activity using the fiber photometry method in the mouse hippocampus (Ikoma, Sasaki, et al., 2023).A mouse model for inducing relatively frequent spontaneously occurring epileptic hyperactivity was required to determine the sequence of neuronal and astrocytic activation time course.Electrical stimulation in the hippocampus can initiate epileptic seizures; however, the stimulation would inevitably activate the neurons directly.Pharmacological manipulations, such as the application of kainic acid, would also directly activate the neurons as well.Thus, the actual sequence of activation in naturally occurring spontaneous epileptic seizures could not be determined with these methods.Previous reports have shown that copper implantation elicits strong inflammation inside the brain leading to spontaneous seizures (Kendirli et al., 2014).With 24-h continuous recordings of astrocyte Ca 2+ , local field potential (LFP) fluctuations, and video monitoring, we were able to capture multiple occurrences of spontaneous neuronal hyperactivity episodes within a week after implantation of copper in the hippocampus.Interestingly, the astrocyte Ca 2+ increase preceded the neuronal hyperactivity by $20 s.Selective activation of astrocytes by (i) optogenetic activation and (ii) deep-brain direct current stimulation (DB-DCS) was attempted to show that astrocyte activity could be the cause of neuronal hyperactivity.Fluorocitrate has been shown to act specifically on astrocytes and inhibit their metabolic activity (Hirayama et al., 2015;Swanson & Graham, 1994).Thus, the effect of the application of fluorocitrate on copper-induced spontaneous epileptic neuronal hyperactivity was also studied.
Here, we demonstrated the presence of an astrocyte-driven switch of neuronal activity.Whether a similar mechanism is shared in the physiological transition of the mode of neuronal circuit activity remains to be sought.Epilepsy is a major disease affecting approximately 1% of the population.It is characterized by a large number of neurons firing synchronously and repetitively.Our result suggests that astrocytes act as the trigger for partial epilepsy and influence ictogenesis.Therefore, astrocytes could potentially serve as an alternative target for epilepsy treatment (Sano et al., 2021;Shimoda et al., 2022).

| Animals
This study was conducted in accordance with the recommendations of the Regulations for Animal Experiments and Related Activities at Tohoku University, and all experimental procedures were approved by the Institutional Animal Care and Use Committee of Tohoku University.We ensured to minimize animal suffering and reduce the number of animals used.
To express Ca 2+ sensor fluorescent protein YC nano50 or channelrhodopsin-2 (ChR2) specifically in astrocytes, we used the tetracycline-controlled transcriptional activation system.Mlc1-tTA mice were crossed with either tetO-YC nano50 or tetO-ChR2 to produce transgenic mice expressing the target protein specifically in astrocytes (Kanemaru et al., 2014;Tanaka et al., 2012).Adult male and female mice (8-30 weeks old) were used in this study.Mice were housed under standard laboratory conditions with food and water ad libitum.All mice used for experiments were fixed with 4% PFA and checked for the location of opto-rodes, copper, and cannulas.

| Electrode, optical fiber, and copper implantation details
A stainless-steel screw anchored over the cerebellum was used as the common ground.Tungsten electrodes (bare diameter 127 μm) with PFA coating (total diameter 178 μm; Cat #796500, AM-Systems, WA, USA) were used for the LFP recordings.The PFA coating was removed for approximately 0.3 mm from the tip of the electrodes.
Optical fibers (core diameter 400 μm, 0.39 NA; FT400UMT; Thorlabs) were hand polished and were used for the fiber photometry recordings.Handmade "opto-rode" was created by adhering the optical fiber and two tungsten electrodes with UV curing resin.In this study, copper implantation was used to induce epileptic hyperactivity.Pure copper line with insulation (total diameter 230 μm; Inter Medical, Tokyo, Japan) was used and only the cut-end was exposed.The exposed end was implanted into the brain parenchyma.Contact of the brain parenchyma with the copper presumably induced an inflammatory response leading to epileptic hyperactivity.For the DB-DCS, platinum electrodes (bare diameter 127 μm) with PFA coating (total diameter 203.2 μm; Cat #773000, AM-Systems, WA, USA) were used.The perfluoroalkoxy alkane (PFA) coating was removed for $0.3 mm from the tip.The use of platinum was essential for the stable DC flow.The platinum was also biologically inert and did not produce inflammation of the brain.
Electrophysiological recordings were amplified by a DAM50 extracellular amplifier (World Precision Instruments, FL, USA).The amplifier was set to the AC recording mode with a gain of 1000.
The signals were band-pass filtered in the range of 1 Hz-1 kHz.Other recording conditions are listed in the next section.

| Fiber photometry
Approximately 2 weeks were required for the fiber photometry recordings to become stabilized after the optical fiber implantation.As seizures occur under 1-week post copper implantation, we developed a two-term operation schedule.First, the opto-rode was implanted at the ventral dentate gyrus of the hippocampus (at 3.5 mm posterior, 2.7 mm lateral, and 1.92 mm deep from the bregma).This opto-rode and the stainless-steel screw over the cerebellum were fixed onto the mouse skull with dental cement.After more than 2 weeks of the first surgery, the second operation was performed.Copper was implanted into the dorsal dentate gyrus of the hippocampus (at 1.7 mm posterior, 0.65 mm lateral, and 2.0 mm deep from the bregma).Recordings started immediately after the second surgery.
For the fiber photometry recording, a custom-built optical block

| Deep-brain direct current stimulation
Methods for the astrocyte ChR2 stimulation, transcranial direct current stimulation (tDCS), and train pulse stimulation are detailed in the Supplementary Information (SI Materials and Methods).For the DB-DCS, a pair of platinum electrodes was used to obtain a stable DC with low resistance and minimal brain damage.Two platinum electrodes were adhered to the center optical fiber and the rotation of the opto-rode was adjusted so that the two electrodes were aligned to the coronal axis.A DC of 0.01 mA in amplitude was generated by an isolator (ISO-Flex, A.M.P.I., Jerusalem, Israel).Out of the two electrodes aligned to the coronal axis, the one close to the center was used as the anode and the lateral electrode was used as the cathode.
One mouse was used for this experiment.
To record the LFP change during DB-DCS stimulation, two tungsten electrodes were adhered to the guide cannula on the sagittal axis, perpendicular to the two platinum electrodes adhered to the guide cannula on the coronal axis.This opto-rode was implanted into the dorsal dentate gyrus of the hippocampus (at 2.06 mm posterior, 1.0 mm lateral, and 1.9 mm deep from the bregma).The stainless steel screw anchored to the skull over the cerebellum was used as the reference ground.
Drug delivery into the hippocampus was done using a custommade cannula.The guide cannula was made by cutting a 25 G needle (Terumo, Tokyo, Japan).The guide cannula was adhered to the optical fiber of the opto-rode.The dummy cannula and the internal cannula were made from a 32 G needle (React System, Osaka, Japan).The dummy cannula was inserted into the guide cannula, which was replaced with the internal cannula on the day of the experiment.
A detailed method for preparing fluorocitrate is described elsewhere (Hirayama et al., 2015).Fluorocitrate (1 μmol/L) in saline plus 0.1 mmol/L Na 2 HPO 4 was prepared.For the control experiments, the same volume of saline plus 0.1 mmol/L Na 2 HPO 4 was used.Experiments were performed 2 weeks after the surgery.The drug was delivered at an injection rate of 0.1 μL/min for 10 min.
Briefly, 1 h after the drug delivery, a DC pulse, 0.01 mA in amplitude, was applied between the platinum electrodes for 15 s in duration using ISO-Flex or NL800A Current Stimulus Isolator (Digitimer, SG6 9BL, UK).Out of the two platinum electrodes, the central electrode was used as the anode and the lateral electrode was used as the cathode.Four mice were used for the saline injection control experiment and five were used for the fluorocitrate injection experiment.

| Evaluating the effects of fluorocitrate on epileptic episodes
An opto-rode with two tungsten electrodes and a guide cannula adhered to the fiber was made for this experiment.The two-term operation schedule was used.For the first surgery, the opto-rode was implanted into the ventral dentate gyrus of the hippocampus (at 3.5 mm posterior, 2.7 mm lateral, and 1.92 mm deep from the bregma) and a stainless steel screw was anchored to the skull over the cerebellum for the reference ground.More than 2 weeks after the first surgery, the second surgery was performed where copper was implanted into the dorsal dentate gyrus of the hippocampus (at 1.7 mm posterior, 0.65 mm lateral, and 2.0 mm deep from the bregma).Recording started immediately after the second operation.We waited for the initial neuronal hyperactivity episode to occur, and the drug delivery was done after its detection.A 1 μmol/L fluorocitrate was injected at an injection rate of 0.1 μL/min for 10 min.After this initial drug delivery, the same drug delivery session was performed every 24 h for a total of 4 days.Four mice were used for saline control experiments and seven were used for fluorocitrate injection.

| Quantification and statistical analysis
All statistical tests were done with Origin Pro 9.0 software (OriginLab, MA, USA) or Excel (Microsoft, WA, USA).All population data are expressed as mean ± s.e.m.Statistical analyses were performed with unpaired t-test (two-sided), Pearson's chi-square test, one-way ANOVA, Tukey's test, and Kolmogorov-Smirnov test.

| Copper-induced spontaneous neuronal hyperactivity
Our aim in this study was to reveal the role of astrocyte activity in epileptic seizure episodes.For this purpose, a mouse model in which spontaneous seizure episodes occur relatively frequently was needed.
Copper implantation in the mouse hippocampus has been shown to reliably induce spontaneous seizures in a previous report (Kendirli et al., 2014).We found that spontaneous episodes of neuronal hyperactivity started as early as 2-3 days after the copper implantation and the occurrence of these episodes lasted for about a week.Following a latent period, a prolonged chronic phase of epilepsy would likely develop; however, it would take months per animal to study such a chronic phase.In this study, we focused on the acute phase to facilitate the cycle of experiments.Therefore, the results described in this article are mostly relevant to the acute symptomatic seizures occurring $1 week of brain insult, such as traumatic brain injury, rather than the seizures occurring in the chronic phase of epilepsy.
Transgenic mice expressing a Ca 2+ sensor, YC nano50 , specifically in astrocytes were used for this study (Figure S1A-C).Opto-rodes were created in-house using a center optical fiber attached with a pair of tungsten wire electrodes on its sides.We found that the copper implantation immediately induced spontaneous neuronal hyperactive episodes with a delay of only a couple of days.Therefore, to capture even these initial events, the opto-rode needed to be implanted into the dentate gyrus of the ventral hippocampus prior to the copper implantation.Approximately 2 weeks were spent after the opto-rode implantation surgery for the optical and electrical recordings to stabilize.Then, copper was implanted into the dentate gyrus of the dorsal hippocampus (Figure S1C).Optical and electrical recordings continued for the following five consecutive days.
Copper exposure to the brain induces strong inflammation in the surrounding tissue (Figure S1D).The severely damaged area around the point of insertion had no apparent optical fluorescence signal emitted from YC nano50 , thus, the cells were likely to be exceedingly unhealthy (Figure 1a).The seizure onset zone was assumed to be located at the interface between this unhealthy tissue and the surrounding tissue.The recording opto-rode was located at some distance from the copper insertion site (Figure S1C), as no fluctuation in the fluorescence signal was detected within this unhealthy area.At 30-72 h post copper implantation, the initial occurrence of spontaneous neuronal hyperactivity was observed in most mice.Brief subsidence was observed after this initial hyperactivity; however, a recurrence of hyperactivity occurred during 84-114 h post copper implantation.Briefly, 5.78 ± 1.24 neuronal hyperactivity episodes were recorded during the 5 days post copper insertion and all mice experienced at least one episode of hyperactivity (n = 9 mice) (Figure 1b).Although the mouse was tethered with optical fiber and electrical wire cables, relatively unrestricted movements were guaranteed.The 24-h near continuous recording allowed detection of most episodes, which only occurred several times per day at most.Neuronal hyperactivity during these episodes was apparent in the hippocampal LFP recordings; however, apparent epileptic behavior could not be observed through video monitoring.The ongoing animal behavior appeared to become interrupted upon the occurrence of a neuronal hyperactivity episode, similar to the behavioral appearance of an absence seizure episode in human epilepsy patients.Neuronal hyperactivity episodes could be roughly categorized into two types based on the characteristics of the waveforms of the LFP recordings.
The first type, in most cases, consisted of the following four components: (i) a high-frequency oscillatory (HFO) component at around 80-120 Hz, (ii) slow wave, (iii) general suppression, and (iv) strong repetitive burst activity (Figure 1c).The components were also not often discrete, and they overlapped with each other in time.Not all of the episodes categorized as the first type were equipped with the complete four components and the sequence of the components varied.The common characteristic of this type was that the waveforms were usually low in amplitude.This type of neuronal hyperactivity is normally classified as the low voltage fast (LVF) type (Avoli et al., 2016).In the second type of episode, neuronal hyperactivity started with repetitive and regular burst activity with large amplitudes (Figure 1d).Such burst activity persisted throughout the episode.This type of neuronal hyperactivity can be classified as the hypersynchronous (HYP) type.Both LVF and HYP types of neuronal hyperactivity episodes are typically observed in human temporal lobe epilepsy (Avoli et al., 2016).In our copper mouse model, the percentage of episodes belonging to the LVF type were 67.4% and those belonging to the HYP type were 32.6% out of 54 events detected.In both types, there were no apparent differences in the astrocyte activity accompanying the neuronal hyperactivity episodes.

| Astrocyte Ca 2+ increases in concert with neuronal hyperactivity
The copper method was applied to the transgenic mice which express a FRET-based Ca 2+ sensor specifically in astrocytes (Mlc1-tTA::tetO-YC nano50 ) (Kanemaru et al., 2014).Astrocyte Ca 2+ activity was monitored using the fiber photometry system along with the LFP recorded from the opto-rode implanted into the dentate gyrus of the ventral hippocampus.
Briefly, 415 nm excitation light was sent through the optical fiber, which primarily excited the CFP component of the YC nano50 protein and CFP emission (fCFP) was detected using a PMT.When Ca 2+ binds to YC nano50 , FRET occurs which results in fCFP decrease and YFP emission (fYFP) increase, which was detected using a second PMT.
First, the fluorescence changes of fYFP and fCFP relative to the baseline were calculated.Typically, the ratio of fYFP over fCFP is taken as a measure of cytosolic Ca 2+ concentration changes.However, as shown previously, fluorescence recordings using the single photon fiber photometry method are heavily influenced by the changes in the local brain blood volume (BBV) and the cytosolic pH (Ikoma, Sasaki, et al., 2023;Ikoma, Takahashi, et al., 2023).To account for the possi-  Discussions).The spectrogram was analyzed for 0-200 Hz with a 1 s window and 80% overlap.Day 1; a large astrocyte Ca 2+ increase was seen without apparent neuronal hyperactivity.Day 2; a HFO in the LFP accompanied the astrocyte Ca 2+ elevation.Burst activity was still not detected.Day 3; HFO was followed by a burst activity.(b) The peak amplitudes of the aberrant astrocytic Ca 2+ event detected after copper implantation were examined.Ca 2+ events not accompanied by neuronal hyperactivity, which typically occurred early on Day 1 after copper implantation (nonseizure) and those that were accompanied by neuronal hyperactivity, which typically occurred on subsequent days (seizure) were compared.A significant increase in the peak Ca 2+ amplitude was observed in the Seizure accompanied astrocytic Ca 2+ events (nonseizure 18.65 ± 2.50%, seizure 30.39 ± 3.34%, n = 5 mouse, Student's t-test, p < .05).(c) The onset of the astrocyte activity was determined.In this particular case, the astrocyte Ca 2+ (top) onset was shown to precede the neuronal hyperactivity (bottom) onset.(d) Astrocyte Ca 2+ dynamics were calculated either by the ratio method (fYFP/fCFP) or the difference method (fYFPÀdYFP).The astrocyte Ca 2+ onset time estimated from either method matched very well.(e) Evaluation of the ratio method and difference method.The time difference between the astrocyte Ca 2+ onset time and the neuronal hyperactivity onset time was calculated.The astrocyte Ca 2+ onset time was determined either by using the ratio method (fYFP/fCFP) or the difference method (fYFPÀdYFP).No significant difference was observed between the two methods (n = 22 episodes, Student's t-test, p = .939).Data are presented as mean ± s.e.m.CFP, cyan fluorescent protein; LFP, local field potential; YFP, yellow fluorescent protein.
concluded that, for the onset time determination, both methods were equally adequate and used either of them in later analysis (for further analysis, see SI Discussions).

| Aberrant astrocyte events precede neuronal hyperactivity
After copper implantation, we manually examined the recorded hippocampal LFP for the presence of neuronal hyperactivity.On Day 1 after copper implantation, characteristic neuronal hyperactivity episodes were still never detected in the LFP recordings.Interestingly, sizable astrocytic Ca 2+ elevations without accompanying neuronal hyperactivity could sometimes be encountered already on Day 1 (Figure 2a).
Such 'aberrant' astrocytic Ca 2+ events were never detected before copper implantation.The occurrence of these initial aberrant astrocytic Ca 2+ increase events could be the prelude to the expected occurrence of neuronal hyperactivity events in the following days.
After Day 2, neuronal hyperactivity episodes start to become detected.All detected neuronal hyperactivity episodes in the subsequent days were always accompanied by Ca 2+ increases in astrocytes (Figures 2a and 3a).
To investigate this initial Ca 2+ increase phenomenon more closely, we first needed to define 'aberrant' astrocytic Ca 2+ events.Although the fluorescence signals detected with the optical fiber inserted in the dorsal hippocampus fluctuated, events with a concomitant increase in fYFP and decrease in fCFP with peak ratio (fYFP/fCFP) in amplitudes more than 10% from the baseline were never encountered prior to the copper implantation (Figure S2A,B).These aberrant astrocytic Ca 2+ events were only encountered after copper implantation.
In 5 out of 12 mice examined, aberrant Ca 2+ events without accompanying neuronal hyperactivity were detected before the detection of the very first neuronal hyperactivity episode after copper implantation.In most experiments described in the following sections, copper was implanted to induce spontaneous neuronal hyperactivity episodes.In some of the experiments, pharmacological treatments followed.However, for examining the early onset and development of neuronal hyperactivity, mice used for experiments intended for other purposes could be included.For the analysis described above, mice from Figure 3c and both fluorocitrate and saline-treated mice in Day 2).The peak amplitude of the aberrant Ca 2+ events was significantly lower in the nonseizure type compared with the seizure type (18.65% ± 2.50% and 30.39% ± 3.34%, respectively, n = 5 mouse, Student's t-test, p < .05; Figure 2b).
These findings show that the occurrence of spontaneous aberrant astrocytic Ca 2+ elevation events precede the development of neuronal hyperactivity.Therefore, these unusually large astrocyte Ca 2+ fluctuations may promote epileptogenesis.We also showed that the occurrence of aberrant astrocytic Ca 2+ events does not require the occurrence of an apparent neuronal hyperactivity.However, the occurrence of neuronal hyperactivity seems to enhance the magnitude of the astrocytic Ca 2+ elevations.Therefore, aberrant astrocytic Ca 2+ elevations may not only promote neuronal hyperactivity but the astrocytic Ca 2+ could also be amplified due to the result of neuronto-astrocyte interaction.

| Astrocyte Ca 2+ increase precedes neuronal hyperactivity
The onset of the astrocytic Ca 2+ elevations upon epileptic episodes appeared to be much earlier than the onset of the neuronal hyperactivity (Figures 2a and 3a).To determine the order of astrocyte Ca 2+ elevation and neuronal hyperactivity, the onset times of both events needed to be evaluated.Astrocyte Ca 2+ onset could be determined objectively as shown above.The onset of the neuronal hyperactivity was rather difficult to identify.As mentioned earlier, the sequence of the characteristic LFP waveform (HFO, DC-shift, burst activity, etc.) was not always constant in the LVF type of neuronal hyperactivity episodes (Figure 1c).Therefore, setting a pre-set rule for the detection of the neuronal hyperactivity episode detection was difficult.Therefore, we decided to visually determine the neuronal hyperactivity onset using two human observers (Figures 3b and S3).The observers were blind to the waveform of the astrocyte Ca 2+ elevations, which always accompanied the neuronal hyperactivity episodes.The observers were instructed to observe only the LFP recordings and visually determine the onset where the characteristics of the LFP fluctuations deviated from the norm.Although the instruction was rather vague, the onset independently determined by the two observers matched well, (R 2 = 0.678 in 22 events, n = 4 mouse, R 2 was calculated for the model y = x).This shows that the detection method is reliable (Figure S3B).
The mean of the onset times determined by the two observers was set as the neuronal hyperactivity onset.The onset times of the astrocyte Ca 2+ elevation and the neuronal hyperactivity in each epileptic episode were compared and it was found that the astrocyte onset was significantly earlier than the neuronal onset (23.27 ± 8.34 s latency, n = 4 mouse, Student's t-test, p < .01, Figure 3c).In no case, does the neuronal hyperactivity onset precede the astrocyte Ca 2+ elevation onset.
Previous reports show that astrocyte activity change often synchronizes with changes in neuronal activity (Oliveira & Araque, 2022).
At the very early onset phase of each epileptic episode detected in our study, a robust astrocyte Ca 2+ elevation was observed but no apparent changes in the LFP waveform could be observed with human eye inspection.However, subtle changes in the LFP waveform in the frequency domain could have occurred at the time of astrocyte Ca 2+ elevation onset.To evaluate this possibility, the power spectrum of the LFP 5 s before and after astrocyte Ca 2+ elevation onset was compared.In every frequency band that was analyzed, no significant change was detected between the pre-and post-astrocyte Ca 2+ elevation onset (Figure 4).In comparison, during the visually detected neuronal hyperactivity period, a significant increase in the power of the LFP was detected in the 12-120 Hz range (Figure 4).These results suggest that the astrocyte Ca 2+ elevation in these seizure episodes is not the result of the neuronal hyperactivity occurrence.It is possible that astrocyte activity may trigger neuronal hyperactivity.

| Astrocyte activity precedes neuronal hyperactivity at the seizure onset zone
In the above experiments, the recording opto-rode was placed at a small distance from the copper implantation site.We assume that the seizure onset zone is at or very near to the copper placement.Thus, both the neuronal hyperactivity and the astrocyte Ca 2+ elevation could initiate at the copper implantation site (dentate gyrus of the dorsal hippocampus) and propagate to the ventral area where the opto-rode was placed.If the propagation of the astrocyte Ca 2+ elevation is much faster than that of the neuronal hyperactivity, then, even if both events are initiated simultaneously at the seizure onset zone, astrocyte Ca 2+ elevation may appear to occur earlier than the neuronal hyperactivity with the recording from the ventral opto-rode.This situation is unlikely as Ca 2+ waves in the astrocyte network typically have been reported to travel $10 times slower (Kuga et al., 2011) than the propagation speed of neuronal hyperactivity observed in human or mouse seizures (Trevelyan et al., 2007).That being said, whether the astrocyte Ca 2+ elevation always precedes neuronal hyperactivity at different distances from the copper implantation site was examined.
We inserted two opto-rodes into the mouse hippocampus for this experiment (Figure 5a).For this dual fiber experiment, one PMT was F I G U R E 4 No significant neuronal activity change before and after astrocyte Ca 2+ increase onset.The power spectrum of the LFP before (1) and after (2) astrocyte Ca 2+ onset, and (3) during neuronal hyperactivity was compared.Significant difference between (1) and ( 3), ( 2) and (3) was found but no significant difference between (1) and ( 2) was observed in all frequency bands (data from 22 episodes from n = 4 mice, one-way ANOVA with Tukey's post-hoc test, *p < .05,***p < .0001).Data are presented as mean ± s.e.m.
used respectively for each fiber recording.As introduced above, there was no significant difference in the astrocyte Ca 2+ onset time determined from the ratio (fYFP/fCFP) or the difference (fYFPÀdYFP) method (Figure 2d).Therefore, we used the difference method (fYFPÀdYFP) for estimating the astrocyte Ca 2+ onset time for these dual recordings.Opto-rodes were implanted into the central (D 1 ; 1.2 mm from the copper implantation site) and ventral (D 2 ; 2.8 mm) dentate gyrus of the hippocampus (Figure 5a) and the onset of astrocyte Ca 2+ elevation and neuronal hyperactivity were determined in the recordings from both opto-rodes.As expected, astrocyte Ca 2+ onset and neuronal hyperactivity onset were, earlier at D 1 compared with D 2 (astrocyte: 1.99 ± 0.44 s, neuron: 8.45 ± 6.89 s), probably because D 1 was closer to the copper implantation site compared with D 2 (Figure 5c).
Next, the time difference between the astrocyte Ca 2+ and the neuronal hyperactivity onsets at D 1 and D 2 were compared.In all five mice recorded, astrocyte Ca 2+ onset always preceded the neuronal hyperactivity onset at both D 1 and D 2 locations (Figure 5d).In one mouse, neuronal hyperactivity was largely delayed (mean 42.07 s) compared with the astrocyte Ca 2+ onset specifically at the D 2 location.If we assume a linear speed of neuronal hyperactivity propagation for this case, an exceptionally slow propagation speed would be estimated (0.04 mm/s).It is likely that, in this particular case, a significant delay mechanism that affects only the propagation of neuronal hyperactivity was set in place between D 1 and D 2 .However, even including this case, the time difference between the astrocyte Ca 2+ and neuronal hyperactivity onsets was not statistically different between D 1 and D 2 , which shows that astrocyte Ca 2+ onset always precedes neuronal hyperactivity onset irrespective of the distance from the seizure onset zone (Figure 5d).
For the example recording shown in Figure 5b, assuming that propagation speed is constant and the seizure onset zone is at the copper implantation site, the astrocyte Ca 2+ and neuronal hyperactivity onsets at the seizure onset zone can be estimated.In this particular case, the astrocyte Ca 2+ onset was estimated to precede neuronal hyperactivity onset at the seizure onset zone by 5.24 s.As mentioned above, the recording from one animal had an unexceptionally delayed neuronal hyperactivity at D 2 .This data were statistically shown as an outlier (Smirnov-Grubs test, p < .01),and thus this data were excluded from further analysis.In recordings from the remaining four animals, astrocyte Ca 2+ and neuronal hyperactivity onsets at the copper implantation site were estimated (Figure 5e).As the propagation speed was estimated to be similar in astrocytes and neurons, the estimated onset difference at the copper implantation site still indicates that astrocyte Ca 2+ increase occurs earlier than the neuronal hyperactivity onset even at the presumed seizure onset zone.

| Optogenetic stimulation of astrocytes causes neuronal hyperactivity
Whether astrocyte activity can causally evoke neuronal hyperactivity was next examined.Transgenic mice expressing ChR2, a nonselective cation channel, specifically in astrocytes were used (Mlc1-tTA::tetO-ChR2) (Tanaka et al., 2012).A glass fiber was implanted into the dentate gyrus of the ventral hippocampus of these mice (Figure 6a).In all three mice that were tested, 5 s of 470 nm photostimulation led to neuronal hyperactivity (Figure 6b; Shimoda et al., 2022).This result shows that astrocyte activation alone can cause neuronal hyperactivity in vivo.The signal cascade that leads from astrocyte ChR2 photoactivation to neuronal hyperactivity was not identified.
Previous studies in the cerebellum and hippocampus have shown that glutamate is released from astrocytes in response to astrocyte ChR2 photoactivation (Sasaki et al., 2012;Shen et al., 2017).It is possible that similar mechanisms operate by the astrocyte ChR2 photoactivation in vivo; however, whether the same astrocyte-to-neuron mechanisms operate in the copper-induced spontaneous epileptic episodes is undetermined.Neuronal hyperactivity evoked after the astrocyte ChR2 photostimulation pulse mostly resembled the high amplitude repetitive burst activity observed in spontaneous LVF neuronal hyperactivity episodes induced by copper implantation; however, the full four featured components of a typical LVF epileptic episode could not be identified.

| DB-DCS activates astrocytes and induces neuronal hyperactivity
The tDCS has been reported to exclusively activate astrocytes and induce intracellular Ca 2+ increases (Monai et al., 2016).The effect of tDCS was evaluated in an awake astrocytic YC nano50 expressing transgenic mouse.An opto-rode was placed in the dentate gyrus of the dorsal hippocampus.Briefly, 0.1 mA DC was passed between the frontal skull (anode) and cerebellum (cathode) for 60 s (Figure 7a).
An increase in the astrocyte Ca 2+ was evident from the YC nano50 fluorescence recording.However, the amplitude of the Ca 2+ increase was far smaller than that observed in spontaneous epileptic episodes (Figure 7c,left).
In order to evoke a larger amplitude astrocytic Ca 2+ increase, a pair of platinum wire electrodes were implanted into the dentate gyrus of the rostral hippocampus (Figure 7a in amplitude reliably evoked a large astrocytic Ca 2+ increase (Figure 7c,right).No apparent damage to the tissue was observed by this stimulation (Figure 7b).LFP changes at the hippocampus in response to DB-DCS were recorded using a pair of tungsten wire electrodes placed perpendicular to the pair of platinum stimulating electrodes (Figure 8a,c).Briefly, 15 s of 0.01 mA DB-DCS evoked a large Ca 2+ increase, which was shortly followed by a neuronal hyperactivity episode (Figure 8d; left).The waveform characteristics of this episode evoked by the DB-DCS highly resembled the spontaneous LVF neuronal hyperactivity episodes caused by copper implantation.
The DB-DCS-evoked episode was equipped with all the featured components of the copper-induced spontaneous episodes (Figure S4); with the (i) high-frequency component around 80-120 Hz, the (ii) DCshift slow wave, the (iii) general suppression, and the (iv) high amplitude repetitive burst activity (Figure 8e).
We showed that DB-DCS did activate astrocytes and induced neuronal hyperactivity.This result suggests that astrocyte activity is causally linked to the induction of neuronal hyperactivity.However, it is possible that DB-DCS directly stimulates neurons, and induces neuronal hyperactivity on its own, without the involvement of the actions from the astrocytes.To address whether astrocytes are truly causally related to neuronal hyperactivity, an astrocyte-specific metabolic inhibitor, fluorocitrate was used.Fluorocitrate is reported to be specifically consumed by the astrocytes and replaces the endogenous citrate.As fluorocitrate cannot be metabolized, the introduction of this molecule results in the jamming of the tricarboxylic acid (TCA) cycle, and, thus, any astrocyte activity requiring ATP energy will become silenced (Hirayama et al., 2015;Swanson & Graham, 1994).Fluorocitrate was injected into the parenchyma of the hippocampus at a concentration of 1 μmol/L, at a speed of 0.1 μL/min for 10 min.The 1 h after injection, 15 s of 0.01 mA DB-DCS was applied (Figure 8b).DB-DCS elicited neuronal hyperactivity in 100% of the trial in salineinjected control, whereas the occurrence of neuronal hyperactivity evoked by DB-DCS was significantly lowered by fluorocitrate injection to 28.6% (Figure 8d,f, right).Interestingly, DB-DCS evoked peak Ca 2+ increase in astrocytes was not significantly affected by the fluorocitrate injection (Figure 8g).The effect on the exact time course of the Ca 2+ elevations could not be evaluated as the DB-DCS evoked Ca 2+ transients varied considerably between animals.These results suggest that, although, with DB-DCS, astrocytes are sufficiently activated, astrocyte action onto neurons is inhibited with fluorocitrate treatment, and neuronal hyperactivity induction is suppressed.The mechanism of fluorocitrate effect on seizure suppression is discussed in detail in the SI Discussions.
Metabolic support from astrocytes to neurons may be essential for the repetitive firing of action potentials, and the application of fluorocitrate may have inhibited those activities.Alternatively, fluorocitrate could directly act on neurons and inhibit their ability to produce hyperactivity.In either case, suppression of neuronal action potential firing or inhibition of neuronal excitatory synaptic transmission may occur as a consequence of fluorocitrate application.To address this possibility, a train of short electrical pulses 50 Hz (1 ms stimulation, 19 ms interval) was delivered to the hippocampus to directly stimulate neurons and elicit neuronal hyperactivity.Typical epileptic discharges were evoked following the train pulse stimulation, which is often referred to as "after discharges."After 50 min of intracerebroventricular injection of fluorocitrate (at 1 mmol/L concentration at a speed of 0.1 μL/min for 10 min), the same intensity train pulse stimulation was delivered.Even with fluorocitrate application, the train pulse stimulation was able to evoke neuronal hyperactivity (Figure S5B).
This shows that neurons have the potential to produce hyperactivity even in the presence of fluorocitrate; however, DB-DCS failed to evoke neuronal hyperactivity.Therefore, DB-DCS likely provokes the astrocyte's excitatory action on the neurons and this astrocyte-to-neuron signaling pathway is inhibited by fluorocitrate.It should also be mentioned that the duration of neuronal hyperactivity after the train stimulation tended to be longer after fluorocitrate application compared with the saline control, although the difference was not statistically significant (Figure S5C).Blocking the astrocyte function with fluorocitrate has been shown to create a brain environment susceptible to epileptic stimuli (Willoughby et al., 2003).Therefore, astrocytes likely have the dual function of both promoting and suppressing neuronal hyperactivity via separate mechanisms (Patel et al., 2019).

| Astrocyte inhibition suppresses spontaneously occurring neuronal hyperactivity
The effect of fluorocitrate on spontaneously occurring epileptic neuronal hyperactivity was studied using the copper implantation model (Figure 9a).After the copper implantation, we waited for the first occurrence of a neuronal hyperactivity episode.Shortly after the detection of the first episode, fluorocitrate was injected into the hippocampus parenchyma (at a concentration of 1 μmol/L at a speed of 0.1 μL/min for 10 min) every 24 h for 4 days (Figure 9b).The frequency of the occurrence of the typical astrocyte Ca 2+ elevation event did not significantly change with fluorocitrate injection compared to the saline control (Figure 9c).The total occurrence of the seizure episodes as detected by apparent neuronal hyperactivity in the LFP recordings also did not change (Figure 9d).However, on close examination of the LFP recordings, the magnitude of the seizure episodes seemed to become different with fluorocitrate application (Figure 10a).
The magnitude of the neuronal hyperactivity upon a seizure episode was quantified by summing the area over the five times SD of the baseline LFP.We termed this value as the hyperactive power (HP; Figure S6).HP for the first epileptic episode (HP 1st ) was calculated and HP for all the subsequent epileptic episodes (HP subsequent ) was divided by HP 1st (Figure 10b).A significant decrease in the HP ratio (HP subsequent /HP 1st ) was observed with fluorocitrate application compared with control (Figure 10c).However, it should be mentioned that the HP ratio increased largely in control.This could reflect a general ten- Overall, our findings suggest that astrocyte has a critical role in the creation and amplification of spontaneous neuronal hyperactivity.

| DISCUSSION
One may believe that even the onset of the astrocyte Ca 2+ elevation preceding a seizure episode is triggered somehow by the neuronal input.However, one could also imagine a fluctuation of a substance of a nonneuronal origin in the local brain environment such as an inflammatory signal, cytokine (Nikolic et al., 2018;Sanz & Garcia-Gimeno, 2020).When the concentration of the slowly fluctuating and spatially diffuse signal exceeds a certain threshold, a rapid and recursive increase in the astrocyte Ca 2+ may be initiated.Development of the cytosolic Ca 2+ imaging technology has revealed to us the mesmerizing spontaneous Ca 2+ dynamics in astrocytes in physiological and pathophysiological conditions over the past number of decades (Hirase et al., 2004); however, the application of tetrodotoxin, for example, has largely failed to significantly alter these dynamics (Agarwal et al., 2017).Loose connections between neuronal and astrocytic activities exist; however, the presence of a totally independent rule that governs astrocyte activity has been suggested.In our study, we show no evidence of neuronal activity change before and after the 5-s time window from the astrocyte Ca 2+ onset.In addition, the propagation speed of the astrocyte Ca 2+ within the hippocampus F I G U R E 9 Effect of fluorocitrate injection on the spontaneous neuronal hyperactivity episodes induced by copper implantation.(a) Overview of the position of the copper implantation (at bregma À1.7, 0.65, and 2.0 mm) and the opto-rode-cannula complex placement (at bregma À3.5, 2.7, and 2.5 mm).(b) The paradigm used to study the effect of FC injection on the spontaneous neuronal hyperactivity induced by copper (Cu) implantation.After the Cu implantation, we waited for the first neuronal hyperactivity to occur.After detection of the first neuronal hyperactivity by local field potential examination, FC (1 μmol/L) was injected into the hippocampus parenchyma at a speed of 0.1 μL/min for 10 min.This injection was repeated every 24 h for 4 days.(c) The mean occurrence of the aberrant astrocyte Ca 2+ increase events was counted every 6 h after the first occurrence of the neuronal hyperactivity episode as detected by the LFP examination.No apparent difference was observed between the FC injected and the saline-injected control.(d) Neuronal hyperactivity was always accompanied by an aberrant astrocyte Ca 2+ increase.The total number of neuronal hyperactivity episodes was compared after saline or FC injection.No significant difference was observed (saline, 11.67 ± 2.40, n = 3 mice; FC, 8.40 ± 2.89, n = 5 mice; Student's t-test, p = .177).(e) Aberrant astrocyte Ca 2+ increase episodes were mostly accompanied by neuronal hyperactivity, which was discernable from the LFP recordings, in saline injected control animals.Among the control mice, in only one mouse, only $10% of the aberrant astrocyte Ca 2+ increase episodes were not accompanied by neuronal hyperactivity.In the other two mice, all aberrant astrocyte Ca 2+ increase episodes were accompanied by neuronal hyperactivity.However, among the FC-treated mice, in one mouse, as much as 87% of the aberrant astrocyte Ca 2+ increase episodes were not accompanied by neuronal hyperactivity.The percentage of the occurrence of aberrant astrocyte Ca 2+ increase failing to induce neuronal hyperactivity tended to be higher in the FC injection condition, although no significance was found with the statistics applied (saline, 3.03 ± 3.03%, n = 3 mice; FC, 35.83% ± 15.92%, n = 5 mice; Student's t-test, p = .175).FC, fluorocitrate.
was incredibly fast.This may be accomplished by the presence of the background inflammatory signal and all astrocytes in vicinity of the copper implantation site are ready to go at any moment by the slightest local environmental perturbance (see SI Discussions).
The idea that astrocytes possess the potential to initiate neuronal hyperactivity has long been sought (Beppu et al., 2014;Kang et al., 2005;Tian et al., 2005).In our recordings, astrocyte Ca 2+ elevation onset preceded neuronal hyperactivity onset (but see also  et al., 2023;Sasaki et al., 2012), and in the cerebellum, ChR2 photoactivated astrocyte acidification led to glutamate release via anion channels composed of LRRC8A and resulted in neuronal excitation as well as enhanced plasticity (Beppu et al., 2014(Beppu et al., , 2021;;Kanaya et al., 2023).ChR2 photoactivation in astrocytes has also been shown to activate and increase Ca 2+ levels in astrocytes (Chen et al., 2013;Tan et al., 2017;Yang et al., 2015).Either acidification or Ca 2+ increase could have induced the release of excitatory gliotransmitters or ions via channel-mediated release or exocytosis (Hamilton & Attwell, 2010), which may have led to the neuronal hyperexcitation that was observed here.
The neuronal hyperactivity that was induced following prominent Whether a similar signal cascade that has been evoked by astrocytic ChR2 photoactivation or DB-DCS occurs in copper-induced spontaneous epileptic episodes remains to be sought.We showed that the application of fluorocitrate reduced the magnitude neuronal hyperactivity.Therefore, in addition to triggering each epileptic episode, astrocyte-to-neuronal interaction may also amplify the neuronal hyperactivity.As neuronal activity has also been shown to increase astrocyte Ca 2+ , reciprocal neuron-astrocyte interaction likely takes part in the exacerbation of neuronal hyperactivity within each epileptic episode.
Spontaneous astrocyte Ca 2+ transients with abnormally large amplitudes were observed as early as $1 day after the copper implantation.These initial astrocyte Ca 2+ transients were often still not accompanied by neuronal hyperactivity.This suggests that the initial waves of astrocyte Ca 2+ propagations may promote the plasticity of the neuronal circuits and provide a basis for the epileptogenesis to subsequently generate (Sano et al., 2021;Tavassoli et al., 2022).In each subsequent epileptic episode, the Ca 2+ waves may also provide a basal shift in neuronal excitability, which paves a highway for neuronal epileptic hyperactivity to follow (Heuser et al., 2018).It is likely that astrocyte Ca 2+ elevation results in the gliotransmitter release, which facilitates neuronal hyperactivity as well as epileptogenic plasticity.Such candidates of gliotransmitter are glutamate and D-serine.
Similar mechanisms may operate in normal physiological situations for the volume control of the general neuronal excitability, which may switch the mode of information processing, and for meta-plasticity control (Asano et al., 2023;Tan et al., 2024).
In human studies, DC shifts have been recorded with electrodes placed on the cortical surface (Amzica & Steriade, 2000).It has been suggested that these DC shifts correlate with astrocyte activity.In epileptic patients, these DC shifts have been shown to precede HFO or ictal onset (Kanazawa et al., 2015).Our results using a rodent epilepsy model show that astrocyte Ca 2+ elevation precedes neuronal hyperactivity.Therefore, it is possible that astrocyte activity could switch the brain to the hyperactive mode in humans as well.It has to be pointed out that we also observed a DC shift-like slow wave in the LFP recorded with the hippocampal electrode upon spontaneous epileptic episodes in rodents.However, these slow waves occurred later than the astrocyte Ca 2+ elevation onsets.In fact, the slow waves did not always precede HFO or other waveforms reflecting epileptic neuronal hyperactivity.It is possible that the slow waves that we recorded reflect astrocytes' response to neuronal hyperactivity.The DC shifts that were recorded with the cortical surface electrodes in human epilepsy patients were qualitatively different from the slow waves recorded with the hippocampal electrodes in rodents.Use of the DC rather than the AC amplifiers may be required to capture the DC shifts associated with the initial astrocyte activity that precedes neuronal hyperactivity.Whether DC shifts recorded with DC amplifiers can be found in parallel with astrocyte Ca 2+ increase in our rodent model remains to be sought.In any case, we show that astrocyte Ca 2+ elevation preceded any alteration in the electrical LFP properties in our experimental recording conditions.Thus, the astrocyte Ca 2+ was shown to be a sensitive predictor of the spontaneously occurring neuronal hyperactive episode.
If astrocyte activity takes part in the initiation of epileptic neuronal hyperactivation in individual epileptic episodes in human patients, the future therapeutic strategy of epilepsy may be directed at controlling the astrocyte activity.As the role of astrocytes in epileptogenesis has also been suggested, suppressing the astrocyte activity may lead to a treatment for preventing further exacerbation of epilepsy.We also showed that astrocyte activity could precede a seizure event as much as 20 s from the neuronal hyperactivity onset.If new technology allows us to monitor the astrocyte activity, the occurrence of an epileptic episode could be predicted, which could alarm the patient for safety precautions to the upcoming seizure (see SI Discussions).
series from Hamamatsu Photonics (Shizuoka, Japan) was used.Two wavelength excitation LEDs were used; 420 nm (M420F2) and 505 nm LED light sources (M505F3; Thorlabs, NJ, USA).Bandpass filters FF01-420/10 (Semrock, IL, USA) and ET510/20m (Chroma; VT, USA) were used to ensure the wavelength for excitation of cyan fluorescent protein (CFP) and yellow fluorescent protein (YFP), respectively.A dichroic mirror (FF458-Di02; Semrock) was used to combine the two excitation lights.The next mirror (FF444/521/608-Di01; Semrock) was used to send both excitation lights to the optical fiber connected to the animal while allowing the emission of both CFP and YFP of the YC nano50 to pass through for detection.The excitation power at the tip of the bare optical fiber inserted into the brain was approximately 0.2 and 0.01 mW for 420 and 505 nm light, respectively.The two emission wavelengths were split with another dichroic mirror (Di02-R514, Semrock), and CFP and YFP were detected through respective bandpass filters (ET473/24 m, Chroma and ET537/29 m, Chroma).Two photomultiplier (PMT) sensor modules (H7422-40, C8137-02, A7423; Hamamatsu Photonics) with a current-to-voltage amplifier (C7319; Hamamatsu Photonics) were used to detect the emitted light.The excitation power of the LEDs and the driving voltage of the PMTs was adjusted so that the output from the PMTs was approximately 2 V for all excitation and emission combinations.The adjustment was done on the first recording of each mouse and the power of the LEDs and the driving voltage were fixed for the rest of the experiment.The 420 nm LED light was illuminated for 20 ms in duration and, after an interval of 20 ms, the 505 nm LED light was illuminated for 20 ms.After an interval of 440 ms, the sequence of 420 and 505 nm LED light illumination started again.The voltage output of the fluorescence emission detected with the PMTs was recorded at 2 kHz sampling frequency and the waveform was averaged for the 20-ms duration of the excitation light exposure.Therefore, the actual sampling of the fluorescence emission was every 500 ms for each PMT, which gave a sampling frequency of 2 Hz.This recording strategy was applied to all other photometry recordings.The double fiber photometry recording condition is described in the Supplementary Information (SI Materials and Methods).The PMT output signals and the electrophysiological signals were digitized using the A/D converter, TUSB-1612ADSM-S2Z (Turtle Industry, Ibaraki, Japan), and recorded using the software, LaBDAQ5-TL (Matsuyama Advance, Ehime, Japan) at a sampling frequency of 2 kHz.Near 24-h, continuous recordings were made.Fifty minutes of the hour was recorded and the rest of the hour was spent automatically saving the recorded data.Five mice were used for this experiment.
Characteristics of neuronal hyperactivity by copper insert.(a) Left; a fluorescent image of the copper implantation site with 488 nm excitation.Apparent damage to the tissue and loss of fluorescent YC nano50 signal was observed.Right; a fluorescent image of the optorode implantation site.YC nano50 expressing cells were sparsely observed.(b) Left; The mean number of neuronal hyperactivity episodes identified with the LFP recordings in the hippocampus after copper insertion was counted (bin size, 6 h).The bar graph is the total number of episodes detected within each bin from n = 9 animals (Y-axis label on the left).The total number of neuronal hyperactivity episodes during the 120 h from the copper insertion per animal is shown as a bar graph on the right (5.78 ± 1.24 episodes, n = 9 mice).Data are presented as mean ± s.e.m.(c) A characteristic LVF neuronal hyperactivity episode example induced by copper insertion.The LVF seizure consisted of four components, HFO, slow wave, suppression, and burst components.The spectrogram was analyzed for 0-120 Hz with a 1 s window and 80% overlap.Lower row figures; left; an enlarged view of a segment of the raw data of the HFO (top) and the HPF data of the same segment at 50 Hz (bottom).During HFO occurrence, high power at frequencies around 80-120 Hz is visible in the spectrogram (middle row).Center; an enlarged view of a segment of the raw data of the slow wave (top) and the LPF data of the same segment at 10 Hz (bottom).Suppression of LFP usually occurs after HFO and DC shift.Right; an enlarged view of a segment of the burst activity component.(d) A characteristic HYP neuronal hyperactivity episode.The spectrogram was analyzed for 50-500 Hz with a 1 s window and 80% overlap.Strong repetitive burst activity suddenly occurred and lasted until the end of neuronal hyperactivity.HFO, high-frequency oscillation; HPF, high-pass filtered; HYP, hypersynchronous; LFP, local field potential; LPF, low pass filtered; LVF, low voltage fast.
ble BBV and pH changes, in addition to the 415 nm, 505 nm excitation light was alternately delivered to selectively excite YFP, and the direct emission of YFP was measured and the fluorescence change relative to the baseline was calculated (dYFP).The fluctuation of this recording would not be influenced by cytosolic Ca 2+ changes and should represent background changes affected by fluctuations in factors such as BBV and pH.If we assume that BBV and pH fluctuation affect fYFP and dYFP equally, then the difference between fYFP and dYFP could also yield a Ca 2+ change readout(Asano et al., 2023;Tan et al., 2024).In our recordings, we focused on the onset time of Ca 2+ increase.Careful inspection of the raw data traces showed that dYFP fluctuations near the onset of the Ca 2+ increase were minimal (Figures2a, 3a and S2C).The Ca 2+ signal fluctuations were calculated with either the ratio method (fYFP/fCFP) or the difference method (fYFPÀdYFP).To objectively evaluate the onset of the astrocyte activity, the baseline of the representative Ca 2+ signals was set at approximately 50 s before an apparent increase in the Ca 2+ and an average value was calculated within a window of 10 s.The whole graph was represented as a percentage change from that baseline.The SD and mean of the baseline were calculated, and the SD was multiplied three to five times and added to the mean.The multiplication factor was selected arbitrarily based on the level of baseline noise and the detected signal increase amplitude for each mouse.A line was drawn at three to five times SD plus mean and the first time point where the astrocyte Ca 2+ signal crosses this line was determined.A linear line was fitted to the astrocyte Ca 2+ signal waveform between À2 and +2 s from this time point.The time point at which the fitted line crossed with the baseline mean was determined as the astrocyte Ca 2+ signal onset time (Figure2c).The Ca 2+ increase onset time was determined using either the ratio (fYFP/fCFP) or the difference (fYFPÀdYFP) methods of Ca 2+ signal calculation.The onset times calculated using either of the methods were not significantly different (Figure2d,e).Thus, we fYFP/fCFP fY F P /f C F s e i z u r e S e i z u r e * F I G U R E 2 Fluorescent signal change observed with neuronal hyperactivity.(a) Example of fluorescent signal and LFP recorded from one mouse after copper implantation.fYFP, dYFP, fCFP stands for FRET YFP, direct YFP and FRET CFP, respectively.The ratio was calculated by (fYFP/fCFP) and the difference was calculated by (fYFPÀdYFP).Both should ideally indicate astrocyte Ca 2+ change (for further information see SI

Figure
Figure9cwere used (note that fluorocitrate or saline was treated only after the detection of the first neuronal hyperactivity episode).After the occurrence of the first neuronal hyperactivity episode, all neuronal hyperactivity episodes were accompanied by aberrant Ca 2+ increase in astrocytes.On the other hand, aberrant Ca 2+ increase without accompanying neuronal hyperactivity was almost never encountered after the occurrence of the first neuronal hyperactivity episode.Only 1 in total of 89 aberrant Ca 2+ events lacked accompanying neuronal hyperactivity after the occurrence of the first neuronal hyperactivity episode (n = 7 mice from Figures3c and 9csaline control).The aberrant Ca 2+ events can be categorized into those that are not accompanied by neuronal hyperactivity (nonseizure type, which typically occurs early after copper insertion) and those that are accompanied Astrocyte Ca 2+ precedes neuronal hyperactivity.(a) An example of a spontaneously occurring neuronal hyperactivity episode induced by copper implantation.fYFP, dYFP, and fCFP stands for FRET YFP, direct YFP, and FRET CFP, respectively.Fluorescence of fYFP, fCFP, and dYFP were recorded with the fiber photometry.Ca 2+ concentration dynamics were estimated by taking the ratio between fYFP and fCFP (fYFP/fCFP).Only a small fluctuation of the dYFP was evident especially in the early onset of the fluorescence signal, suggesting that a nearly pure Ca 2+ increase has occurred.LFP was recorded from the hippocampal electrode (bottom).A strong astrocyte Ca 2+ elevation was seen preceding neuronal hyperactivity.(b) Example LFP traces of LVF and HYP neuronal hyperactivity episodes.Traces are aligned at astrocyte Ca 2+ increase onset time.The neuronal hyperactivity onset, determined by two observers, is indicated with a triangle and eyeball.In both types of neuronal hyperactivity episodes, astrocyte onset preceded neuronal hyperactivity onset.(c) The time difference between the astrocyte Ca 2+ onset and the neuronal hyperactivity onset was measured and averaged across multiple episodes in individual animals (23.27 ± 8.34 s; n = 4 mice).Data are presented as mean ± s.e.m.CFP, cyan fluorescent protein; HYP, hypersynchronous; LFP, local field potential; LVF, low voltage fast; YFP, yellow fluorescent protein.
Astrocyte activity precedes neuronal activity at recording sites with different distances from the copper implantation site.(a) Left; Copper (Cu) and the two opto-rode (fiber and tungsten [Tg] electrode complex) (D 1 and D 2 ) insertion locations in the hippocampus (3D representation using the 3D Brain Explorer, Allen Mouse Brain Connectivity Atlas, https://connectivity.brain-map.org/3d-viewer;Wang et al., 2020), Right; Fiber photometry settings for both fibers implanted in D 1 (Left) and D 2 (right).(b) Astrocyte Ca 2+ traces (calculated by fYFPÀdYFP) at D 1 and D 2 (top two traces) and local field potential traces at D 1 and D 2 (bottom two traces) in a single hyperactivity episode.Both astrocyte Ca 2+ increase and neuronal hyperactivity occurred earlier at D 1 compared to D 2 .As D 1 is closer to D 2 to the copper implantation site, this time difference between D 1 and D 2 suggests that the copper implantation site (Cu) is the seizure onset zone and both astrocyte and neuronal hyperactivity propagate from Cu. From the time difference at D 1 and D 2 , the astrocyte Ca 2+ onset at Cu and neuronal hyperactivity onset at Cu could be estimated.At all locations including Cu, D 1 , and D 2 , astrocyte Ca 2+ onset preceded neuronal hyperactivity onset.(c) The onset time difference between D 1 and D 2 was calculated for both astrocyte activity and neuronal hyperactivity.Astrocyte activity was delayed on average 1.99 ± 0.44 s and neuronal hyperactivity was delayed on average 8.45 ± 6.89 s at D 2 compared to D 1 (n = 5 mice, for each animal, 5, 9, 4, 9, and 3 episodes were recorded).(d) The time difference between the astrocyte Ca 2+ onset time (determined using the difference method) and the neuronal hyperactivity onset time at both D 1 and D 2 was calculated.In one mouse, an exceptionally long onset time difference was observed at D 2 (data shown as a red-filled circle).This data could be identified as an outlier (Smirnov-Grubbs test, p < .01).However, even with this data included, the time difference detected at D 1 and D 2 was not statistically different (D 1 , 7.18 ± 1.88 s; D 2 , 14.84 ± 7.06 s; n = 5 mice, Student's t-test, p = .32).This shows that astrocyte Ca 2+ increase consistently precedes neuronal hyperactivity onset irrespective of the distance from the copper implantation site.(e) Left; Data from n = 5 episodes from a single animal are shown.The onset time of astrocyte Ca 2+ elevation at the D 1 location is set to zero (t Ca-D1 ) and all data are plotted relative to t Ca-D1 .For example, there are n = 5 data points for the neuronal hyperactivity onset time relative to the t Ca-D1 at the D 1 location shown with open square symbols.The average and the s.e.m. are shown overlaid with error bars.Similarly, there is the same number of data points for the astrocyte Ca 2+ onset at D 2 (open circles) and for the neuronal hyperactivity onset time at D 2 (open squares; all relative to the t Ca-D1 ).A straight line was connected for the average data at D 1 and D 2 for astrocyte Ca 2+ (red line) and neuronal hyperactivity (blue line), respectively, and extrapolated to the Cu location.The time difference between the astrocyte Ca 2+ and the neuronal hyperactivity was calculated for the Cu, D 1 , and D 2 locations.Right; The average time differences between the astrocyte Ca 2+ onset and the neuronal hyperactivity onset at Cu, D 1 , and D 2 locations were calculated for each animal and the average and the s.e.m. are shown as bar graphs and error bars (Cu: 6.21 ± 2.52 s, D1: 7.03 ± 2.43 s, D2: 8.04 ± 2.40 s).The time difference at Cu was estimated by linearly fitting the time difference change between D 1 and D 2 .The one animal, which was analyzed as an outlier, was not included in this analysis.PMT, photomultiplier tube.

F
,b).Passing DC locally between these two platinum electrodes (DB-DCS) with only 0I G U R E 6 Astrocyte-specific ChR2 light stimulation evokes neuronal hyperactivity.(a) Mlc1-tTA::tetO-ChR2 mouse which expresses ChR2 specifically in astrocytes was used to activate astrocytes by light stimulation.An opto-rode (fiber and tungsten [Tg] electrode complex) was implanted into the ventral dentate gyrus of the hippocampus to record the local field potential of the hippocampus by light stimulation.(b) 5 s of 0.5 mW blue light stimulation in the hippocampus evoked neuronal hyperactivity in all three mice tested.ChR2, channelrhodopsin-2.
Direct current stimulation stimulates astrocyte Ca 2+ increase.(a) Mlc1-tTA::tetO-YC nano50 mouse was used to evaluate astrocyte Ca 2+ change by direct current stimulation.An overview of the mouse skull (upper panel).For the tDCS, anode and cathode placed on the skull were used.For the DB-DCS, an opto-rode accompanied by a pair of platinum electrodes (Pt) was used.The opto-rode was implanted into the dorsal dentate gyrus of the hippocampus.(b) A fluorescent image of the brain section at the opto-rode insertion site of the YC nano50 expressing mouse.DB-DCS seemingly did no apparent damage to the brain tissue.(c) Comparison of the effect of the tDCS and the DB-DCS.A 60 s of tDCS at 0.1 mA or DB-DCS at 0.01 mA was delivered.fYFP, dYFP, fCFP stands for FRET YFP, direct YFP and FRET CFP respectively.The traces are an average of the three trials tested in one mouse.CFP, cyan fluorescent protein; DB-DCS, deep-brain direct current stimulation; tDCS, transcranial direct current stimulation; YFP, yellow fluorescent protein.
dency for the subsequent neuronal discharges associated with the epileptic episode to become larger as the exacerbation of epilepsy occurs with repeated occurrences of epileptic episodes.With the application of fluorocitrate, this epileptogenic tendency became less and the HP ratio tended to be close to 1 and a significant difference was observed U R E 8 DB-DCS evokes neuronal hyperactivity.(a) An overview of the mouse skull for experiments using DB-DCS with LFP and fiber photometry recordings.The location of the cannula placement for the use of drug injection into the brain parenchyma is also shown.An opto-rode with platinum stimulation electrodes (Pt), tungsten recording electrodes (Tg), and an injection cannula was inserted into the dorsal dentate gyrus of the hippocampus.(b) Schema of the drug (fluorocitrate; FC) injection and the DB-DCS delivery time course.First, 1 μmol/L fluorocitrate was injected at 0.1 μL/min speed for 10 min.For control, the control solution was injected instead.One hour after the start of the injection, 0.01 mA DB-DCS for 15 s was delivered.(c) A photo of the opto-rode and cannula complex.(d) DB-DCS stimulates astrocytes to evoke neuronal hyperactivity.DB-DCS evoked a large astrocyte Ca 2+ elevation, which preceded the neuronal hyperactivity (left).The spectrogram was analyzed for 0-200 Hz with a 1 s window and 80% overlap.With the injection of fluorocitrate, DB-DCS still evoked a sizable Ca 2+ elevation in astrocytes; however, neuronal discharge was not evoked (right).(e) Characteristics of neuronal hyperactivity evoked by DB-DCS.Much like the low voltage fast activity in copper-induced neuronal hyperactivity episodes, it had the four components of a typical LFP waveform with HFO, slow wave, suppression, and burst components.(f) Neuronal hyperactivity occurrence rate by DB-DCS was significantly decreased in fluorocitrate treated mouse compared to control (control vs. fluorocitrate was 100% [n = 4] to 28.6% [n = 7], p < .05,Pearson's chi-square test).(g) No apparent change in peak astrocyte Ca 2+ was detected by fluorocitrate treatment (control vs. fluorocitrate was 17.22% [n = 4] to 12.96% [n = 7], p = .543,Student's t-test).Data are presented as mean ± s.e.m.DB-DCS, deep-brain direct current stimulation; HFO, high-frequency oscillation; HPF, high-pass filtered; LPF, low-pass filtered; LFP, local field potential.comparedwith control (Figure10c, right).As shown in Figure10a, a large astrocytic Ca 2+ elevation often associated with an epileptic discharge was observed in control as well as in the presence of fluorocitrate.However, in the presence of fluorocitrate, Ca 2+ transients not accompanied by typical neuronal hyperactivity were often observed, although the rate of encounter of such no neuronal hyperactivity episodes was not statistically significant (Figure9e; see SI Discussions).

Figure
FigureS7), which provides evidence that astrocytes switch the mode of neuronal activity; however, the relationship between astrocyte and neuronal activities is still a correlative one.To show a causal relationship, specific perturbation of astrocyte activity was necessary and we accomplished this either with the astrocyte-specific ChR2 photoactivation or with the DB-DCS.The mechanisms of astrocytic ChR2 photoactivation leading to neuronal hyperactivity are not identified in the

astrocyte
Ca 2+ increase by DB-DCS was similar to the LVF type of copper-induced spontaneous epileptic episodes.This neuronal hyperactivity was fully suppressed by fluorocitrate even though the peak Ca 2+ response induced by DB-DCS was not significantly suppressed (Figure8d-g).These suggest that the gliotransmitter release mechanisms downstream of the astrocyte Ca 2+ elevation required energetic support from adenosine triphosphate (ATP) production by the TCA cycle in astrocytes.It may be that ATP energy is required for the creation of cytosolic or vesicular ion and gliotransmitter balance(Karus et al., 2015; see also SI Discussions).Astrocyte inhibition suppresses spontaneously occurring neuronal hyperactivity.(a) Raw traces of Ca 2+ elevation episodes pre-and post-fluorocitrate injection.fYFP, dYFP, fCFP, and LFP stands for FRET YFP, direct YFP, FRET CFP, and local field potential, respectively.The spectrogram was analyzed for 0-120 Hz with a 1 s window and 80% overlap.An apparent subsidence in neuronal hyperactivity is evident in the post-fluorocitrate LFP trace.(b) The first large astrocyte Ca 2+ elevation episode after copper implantation was captured and the corresponding LFP response in the hippocampus was collected (top trace, left).The hyperactive power was calculated from the LFP trace (HP 1st ).After the detection of such a Ca 2+ event, either saline (top) or fluorocitrate (bottom) was injected.The LFP trace accompanying a subsequent episode of another large astrocyte Ca 2+ event is shown on the right and HP was calculated for this episode (HP subsequent ).In control saline injection sessions (top), the subsequent episodes had usually large HP subsequent to HP 1st , because the epileptic hyperactivity tends to develop with each occurrence of episodes.However, in fluorocitrate injection sessions (bottom), the subsequent episodes often were of the same magnitude or smaller.(c) Left: Cumulative frequency histogram of the HP subsequent /HP 1st for control (black) and fluorocitrate (red) injection conditions.A significant decrease in the neuronal hyperactivity magnitude was observed ( p < .005,Kolmogorov-Smirnov test).Note that in control experiments, most episodes tend to escalate in magnitude.The escalation of magnitude is greatly suppressed by the fluorocitrate treatment, in some mice, almost no neuronal hyperactivity was seen.Right: Bar graph of the mean ratio for each mouse.The suppression of neuronal hyperactivity was significant (control vs. fluorocitrate is 4.81 ± 1.56 (n = 3) to 1.38 ± 0.61 (n = 5), p < .05,Student's t-test).CFP, cyan fluorescent protein; LFP, local field potential; YFP, yellow fluorescent protein.