Deep brain stimulation of the anterior nuclei of the thalamus can alleviate seizure severity and induce hippocampal GABAergic neuronal changes in a pilocarpine-induced epileptic mouse brain

Abstract

Deep brain stimulation (DBS) of the anterior nucleus of the thalamus (ANT) has been widely used as an effective treatment for refractory temporal lobe epilepsy. Despite its promising clinical outcome, the exact mechanism of how ANT-DBS alleviates seizure severity has not been fully understood, especially at the cellular level. To assess effects of DBS, the present study examined electroencephalography (EEG) signals and locomotor behavior changes and conducted immunohistochemical analyses to examine changes in neuronal activity, number of neurons, and neurogenesis of inhibitory neurons in different hippocampal subregions. ANT-DBS alleviated seizure activity, abnormal locomotor behaviors, reduced theta-band, increased gamma-band EEG power in the interictal state, and increased the number of neurons in the dentate gyrus (DG). The number of parvalbumin- and somatostatin-expressing inhibitory neurons was recovered to the level in DG and CA1 of naïve mice. Notably, BrdU-positive inhibitory neurons were increased. In conclusion, ANT-DBS not only could reduce the number of seizures, but also could induce neuronal changes in the hippocampus, which is a key region involved in chronic epileptogenesis. Importantly, our results suggest that ANT-DBS may lead to hippocampal subregion-specific cellular recovery of GABAergic inhibitory neurons.

Introduction

Temporal lobe epilepsy (TLE) is the most common type of human focal epilepsy (Devinsky et al. 2018). It is characterized by recurrent, unprovoked focal seizures originating from the temporal lobe of the brain (Engel Jr 1996). Approximately 30% of TLE patients are resistant to antiepileptic drugs (AEDs). They continue to suffer from persistent seizures (Laxer et al. 2014). To treat drug-resistant epilepsy, neurostimulation therapy such as vagal nerve stimulation or deep brain stimulation (DBS) to a specific brain area has been widely used (Van Der Vlis et al. 2019). DBS has been proven to be highly effective in suppressing seizures in many clinical cases when it is applied to the centromedian nucleus of the thalamus, the hippocampus, and the anterior nucleus of the thalamus (ANT) (Fisher et al. 2010). In particular, high frequency (>100 Hz) DBS delivered to the ANT (ANT-DBS) has gained widespread attention since it not only can significantly reduce seizure frequency, but also can result in improved cognitive performance (Fisher et al. 2010; Kim et al. 2017). The ANT is well known to have widespread frontal and temporal cortical connections as well as heavy reciprocal connections with limbic circuits (Child and Benarroch 2013). Many animal experiments examining the epileptic brain have frequently revealed that regions influenced by ANT-DBS exhibit neuroprotective effects manifested as reduced excitotoxicity in epileptic neuronal networks (McKinnon et al. 2019).

Chronic epileptogenesis is known to be highly associated with drastic changes in neurons of the hippocampus, which is considered to be the main generator of temporal lobe epilepsy (Dudek and Sutula 2007). Neuronal injury and death in the hippocampus have been extensively studied both in experimental epilepsy models and clinical reports (Buckmaster and Jongen-Rêlo 1999; Buckmaster et al. 2002; Kapur 2003). In particular, GABAergic inhibitory interneurons, which play a crucial role in shaping epileptic activity (Soukupová et al. 2014; Wenzel et al. 2019; Lim et al. 2021), show decreased populations in many studies (Buckmaster and Dudek 1997; Peng et al. 2013; Hsiao et al. 2015). A few studies have reported that ANT-DBS can increase neurogenesis or mediate neuroprotective pathways, leading to recovery or prevention of hippocampal neuronal loss (Toda et al. 2008; Chen et al. 2017; Du et al. 2020). However, neuron-type specific changes have been rarely investigated regarding effects of ANT-DBS.

Moreover, hippocampal neurons are differentially distributed in hippocampal subregions such as CA1, CA3, and DG (Zhang et al. 2009), which have distinct reciprocal connections with the ANT as well as its associated brain regions (Child and Benarroch 2013). However, whether the ability of ANT-DBS to suppress seizure activity is due to distinct neuronal changes across different hippocampal subregions remains unclear.

Therefore, the objective of this study was to investigate whether neurons across different hippocampal subregions might be differentially affected by chronic epileptic conditions and ANT-DBS to understand the efficacy of ANT-DBS on epileptic brain activity. We explored how ANT-DBS could modulate seizure outcomes by examining electroencephalography (EEG) signals and cellular changes in different hippocampal subregions, including GABAergic inhibitory neuronal populations.

Materials and methods

Animal care

All experimental procedures were approved by the Institutional Animal Care and Use Committee (IACUC) of Sungkyunkwan University (SKKU) (SKKUIACUC16-12-3-2). All procedures were conducted in accordance with the Guide for the Care and Use of Laboratory Animals of the Animal Protection Law & the Laboratory Animal Act set by Korea Animal and Plant Quarantine Agency and Korea Ministry of Food and Drug Safety. Seven-week-old male C57bl/6 mice (n = 36, OrientBio Inc., South Korea) were used. Upon arrival, they underwent a 1-week adaptation period. These animals were housed in individual cages with ad libitum access to food and water. The environment was maintained on a 12-h light/dark cycle at a temperature of 23–24°C with a humidity of 50–60%.

Pilocarpine injections

Experimental procedures were conducted as shown in Fig. 1A. Mice were injected with scopolamine methyl nitrate (1 mg/kg, intraperitoneal; Sigma-Aldrich, USA) at 30 min prior to pilocarpine injection (270 mg/kg in saline, 100 μL, intraperitoneal; Sigma-Aldrich) to minimize peripheral effects of pilocarpine. Control mice were treated with an equivalent volume of saline. Within 10–30 min after pilocarpine injection, status epilepticus was induced. Using a modified Racine scale (Racine 1972; Cela et al. 2019), animals that exhibited unilateral forelimb clonus were scored as stage-3, those that showed bilateral forelimb clonus with rearing were scored as stage-4, those that exhibited rearing and falling were scored as stage-5, and those that showed wild running, jumping, and vocalizations were scored as stage-6. Mice that reached ≥stage-3 during a 2-h monitoring period after pilocarpine injection were selected for further experiments. Mice that received pilocarpine (PIL) injection showed an average score of 4.21 ± 0.65, whereas control mice (CTR) showed a score of 0 (Fig. 1B). We then injected diazepam (10 mg/kg, intraperitoneal; SAMJIN, South Korea) into mice to interrupt status epilepticus. All mice were supplied with 10% w/w glucose water for 3 consecutive days.

Open field test

To measure hyperactive behavior as one behavioral impairment resulting from TLE (Müller et al. 2009; Otsuka et al. 2016), we conducted open field test (OFT) for 2 groups of mice at 7 days after pilocarpine or saline injection. The experimenter was blinded to the treatment that mice had received. The OFT was performed during the daytime (2:00–4:00 P.M.). Mice were placed in an open square box (60 x 60 cm) for 5 min with a video recording and analysis system (Ethovision XT, Noldus, USA). Mice were then returned to their home cages. The PIL group showed heightened locomotor behavior compared to the CTR group (Fig. 1C–E).

Surgical and experimental procedures for EEG recording and ANT-DBS

After 1 week of OFT, which was 2 weeks after pilocarpine injection, mice were initially anesthetized with 4% isoflurane (HanaParm, South Korea) and maintained with 1.5–2% isoflurane after they were transferred to a stereotaxic frame (Kopf Instruments, USA). Screw electrodes (diameter: 0.5 mm; Worcester Polytechnic Institute, USA) for EEG recordings were positioned bilaterally on the motor cortex area [1 mm anterior from bregma; 1.5 mm lateral from midline]. A ground electrode was placed above the olfactory cortex (Fig. 1F). One week after the implantation, which was 3 weeks after pilocarpine injection, a video-supported EEG-monitoring system (CinePlex and OmniPlex, Plexon, USA) was used for simultaneous 12-h recording of awake EEG signals and movements of mice in their home cages. Animals exhibiting spontaneous recurrent seizures including focal and generalized seizures were selected for further experiments. On the next day, mice were anesthetized with 2% isoflurane and bipolar tungsten electrodes (MS303T/2-B/SPC; Plastics One Inc., USA) were implanted into anteroventral thalamic nuclei [0.85 mm posterior to bregma, 0.875 mm lateral from midline; 3 mm depth] (Fig. 1F). Two weeks after that, which was 5 weeks after pilocarpine injection, ANT-DBS was delivered for 12 h per day for 2 consecutive days (10 min ON/20 min OFF; amplitude, 2 V; frequency, 130 Hz; pulse width, 90 μs). EEG recording was performed on the 2^nd day and the 7^th day following the stimulation (Fig. 1F).

BrdU injection

After 5-bromo-2′-deoxyuridine (BrdU; Sigma) was dissolved in 0.1 M phosphate-buffered saline (PBS) at a concentration of 10 mg/ml, it was heated at 60°C for 10 min. The first BrdU injection (50 mg/kg, i.p.) was performed right after the second ANT-DBS delivery was finished (Fig. 1A). The BrdU solution was then additionally injected 3 times per day (8 h interval) for 3 consecutive days (Wojtowicz and Kee 2006; Stone et al. 2011).

Immunohistochemical staining

Mice at 6 weeks after pilocarpine injection were transcardially perfused with saline and 4% paraformaldehyde (PFA) using a peristaltic pump (P-1500; Harvard apparatus, USA). Their brains were extracted and postfixed in 4% PFA for 1 day. These brains were then immersed in 30% sucrose in PBS with 0.1% sodium azide for 3 days. They were then frozen in tissue freezing medium (FSC22; Leica Biosystems, Germany) and coronally sectioned to a thickness of 40 μm using a cryostat (CM1950; Leica Biosystems). Either PV & c-fos, SOM & c-fos, PV & BrdU, or SOM & BrdU was used to stain coronal brain slices. For each group, three 40 μm-thick coronal sections at 1.74–2.14 mm posterior to bregma in each animal were chosen. Primary antibodies used were goat anti-c-fos (Abcam, USA., 1:40), rabbit anti-neuronal nuclei (NeuN) (Abcam, 1:400), rabbit anti-PV (Abcam, 1:200), mouse anti-BrdU (Sigma-Aldrich, USA, 1:200), and rabbit anti-SOM (Peninsula, USA, 1:1000). Secondary antibodies used were anti-goat IgG Alexa555 (Molecular probes, USA, 1:350) and anti-rabbit IgG Alexa488 or Alexa568 (Molecular probes, 1:200). Cell nuclei were stained with 4′,6-diamindino-2-phenylindole (DAPI) (Sigma-Aldrich, 1:10,000). These stained brain slices were imaged using a confocal laser microscope (TCS SP8; Leica Microsystems) with a 20x objective lens (numerical aperture = 0.75).

Data analysis

To detect seizures, we mainly used amplitudes of EEG signals and behavioral seizures captured by concurrent video recording. For all EEG datasets, we first applied high-pass filter (>0.5 Hz). EEG signals over 10 mV artifacts were excluded (Supplementary Fig. S1). Among EEG signals over three standard deviations (SD) of baseline which exhibited no behavioral seizure (389.6 ± 15.8 μV on average), seizure-like EEG activity was determined if individual spikes lasted 30–200 ms and the interval between spikes was 80–800 ms (Jaseja and Jaseja 2012; Sharma et al. 2018) as shown in Supplementary Figures S1D and S1E. Detected seizures by these criteria were generally well-matched with behavioral seizures of Racine stage ≥3 in the video. If interval between spikes was >800 ms (Sharma et al. 2018) (Supplementary Fig. S2A and B), they were considered interictal spikes. For EEG analysis of interictal brain waves, 12 epochs of 10-min period without interictal spikes and ongoing seizure activities were randomly selected. In the same epochs, movements of animals in their home-cages were quantified. In the ANT-DBS group, movements within 5-min immediately after each stimulation were excluded to avoid electrical artifacts of the stimulus and to reject direct effects on EEG signals by the stimulus. Time-course power spectral density (PSD) was calculated by applying a multitaper transformation (sliding window: 1 s, bin: 100 ms). Interictal spikes were sorted out to eliminate effects of interictal spikes on brain waves to compare seizure-free brain waves. PSDs were summed in 5 distinct frequency ranges: 1–4 Hz (δ band), 4–8 Hz (θ band), 8–12 Hz (α band), 12–30 Hz (β band), and 30–50 Hz (γ band). Confocal imaging data were analyzed using Fiji (ImageJ, USA) and Imaris (Bitplane, UK). More details about EEG analysis are described in Supplementary Material. For histological analysis, cells overlaid with DAPI⁺ signals were individually counted in CA1, CA2, CA3, and DG of 10-μm z-stack images (Supplementary Fig. S3). Cells with cell bodies less than 10 μm were excluded. Numbers of cells counted in three brain sections were averaged. The method to divide subregions and other details are described in Supplementary Material.

Study design and statistics

All statistical analyses were performed with IBM SPSS v.19 (USA). Shapiro–Wilk normality test was conducted for all data sets. Accordingly, we used either an independent t-test or the Mann–Whitney U test to examine differences between two independent samples [CTR vs. PIL groups]. We used either a paired t-test or the Wilcoxon signed-rank test when comparing 2 dependent samples (before DBS vs. after DBS). To analyze more than 2 groups of histology data, the Kruskal–Wallis test and Friedman test were used with Bonferroni post hoc analysis. Likewise, we calculated either Pearson’s or Spearman’s correlation coefficient to examine a linear relationship between 2 variables. Linear regression models were fitted using ordinary least squares. Data are expressed as mean ± SD. Throughout figures, ^*, ^**, and ^*** indicate statistical significance of P < 0.05, P < 0.01, and P < 0.001, respectively.

Results

ANT-DBS affects locomotion, epileptic seizures, and cortical EEG power in the interictal state

To investigate effects of ANT-DBS on epileptic brain activity, cortical EEG signals were measured via video-supported EEG recording before and after ANT-DBS in motor cortex area. We counted the number of seizures before and after ANT-DBS and analyzed the power spectra of seizure-free EEG signals. For spectral analysis, we chose time epochs that had no ongoing seizures or interictal spikes. In general, there was a significant peak in EEG power in the low frequency range between 3.4–5.8 Hz and 7.8–8.7 Hz (theta-band; Fig. 2A and B) in epileptic mice. However, such peak was not observed in control mice (Supplementary Fig. S4). Interestingly, ANT-DBS significantly decreased theta-band EEG power. In the 10 min ON-20 min OFF cycle of ANT-DBS, the decreased theta power was maintained nearly until the end of the OFF cycle (Supplementary Fig. S5). On the other hand, ANT-DBS increased EEG power in the high frequency range between 41.2 and 50 Hz (gamma-band; Fig. 2B). Decreased theta power and increased gamma power were apparent in bandpass filtered EEG signals for different frequency bands (Fig. 2C). ANT-DBS apparently affected interictal brain oscillations in theta power, resulting in a reduction from 3.95 ± 0.39 × 10⁻² mV²/Hz to 2.41 ± 0.43 × 10⁻² mV²/Hz and an increase in gamma power from 3.22 ± 0.73 × 10⁻³ mV²/Hz to 5.48 ± 0.94 × 10⁻³ mV²/Hz, which was similar to the power of control animals (Fig. 2D). These effects lasted for up to 1 week, at which point mice were sacrificed for brain extraction (Supplementary Fig. S6 and Fig. 2D).

We further analyzed behavioral characteristics of mice in home-cages that were simultaneously recorded for EEG signals. Compared to naïve mice, before receiving ANT-DBS, epileptic mice showed hyperactive locomotive behaviors in their home-cages (Fig. 2E–G). In the group of animals that received ANT-DBS, the hyperactive locomotion was significantly reduced compared to that in the pre-ANT-DBS group (Fig. 2E–G). In particular, the post-ANT-DBS group exhibited significantly lower distance traveled at a lower velocity than the pre-ANT-DBS group. Locomotive behaviors of mice in the post-ANT-DBS group were comparable to those in the naïve group. Coupled with these locomotor behavior changes, in the ANT-DBS group, the number of chronic seizures under post-ANT-DBS condition was significantly decreased compared to that under the pre-ANT-DBS condition. The duration of individual seizures was also decreased (Fig. 2H and I). ANT-DBS also reduced the number of interictal spikes (Supplementary Fig. S7). These effects persisted for 1 week (Fig. 2F–I).

We then investigated relationships among changes in epileptic seizures, interictal brain activity in theta and gamma oscillations, and behaviors. In the chronic epileptic group before ANT-DBS, heightened theta power was highly correlated with the number of seizures (Fig. 3A) and the distance traveled in home-cages (Fig. 3B). In contrast, gamma power showed a negative correlation with the number of seizures (Fig. 3C) and locomotor behaviors in home-cages (Fig. 3D). In other words, greater increases in theta power and greater decreases in gamma power were associated with more severe seizures in chronic epileptic animals. Moreover, theta power and gamma power had a negative linear relationship (Fig. 3E).

We further investigated how ANT-DBS affected seizure severity and brain oscillations. Following ANT-DBS, the normalized EEG showed a significant decrease of theta power (Fig. 3F) but an increase of gamma power (Fig. 3G). A decrease in normalized theta power due to ANT-DBS was highly correlated with a decrease in the number of seizures (Fig. 3H) and the distance traveled in home-cages (Fig. 3I). An increase in normalized gamma power also showed a strong correlation with the reduced number of seizures (Fig. 3J). However, it did not show a significant correlation with locomotor behavior in home-cages (Fig. 3K).

Additional analyses were performed to determine whether the stimulation itself affected locomotive behavior that could change theta-band EEG signals. When the speed of movement was aligned according to each stimulation session, there was no significant behavior changes following the stimulation onset (Supplementary Fig. S8A–C). The average velocity and theta power of naïve animals were 0.4126 cm/s and 0.0208 mV²/Hz, respectively. When the velocity and theta power of chronic epileptic mice were classified into above and below average groups, there was no direct relationship between heightened locomotion and theta power (Supplementary Fig. S8D-G). Therefore, we could conclude that locomotion did not directly contribute to changes in theta oscillations. ANT-DBS might have indirectly affected the locomotive behavior by altering brain state related to chronic epilepsy rather than directly affecting locomotion.

ANT-DBS affects c-fos expression and neuronal numbers in hippocampal regions

We then investigated whether seizure reduction by ANT-DBS was related to neuronal changes in the hippocampus. We measured c-fos expression levels and counted the number of NeuN⁺ cells in the hippocampus. In chronic epileptic mice, all hippocampal subregions showed an increase in the expression level of c-fos (Fig. 4A-D). However, c-fos expression was reduced in the ANT-DBS group, similar to that of naïve mice (Fig. 4E). The number of NeuN⁺ cells was decreased in the hippocampi of epileptic mice, especially in CA1 and DG (CA1: 3849.47 ± 457.58 cells/mm² in naïve and 2783.73 ± 464.57 cells/mm² in epilepsy; DG: 6113.57 ± 751.88 cells/mm² in naïve and 3998.07 ± 493.76 cells/mm² in epilepsy) (Fig. 5A–E). Surprisingly, the number of NeuN⁺ cells was increased in DG when ANT-DBS was applied (epilepsy + ANT-DBS: 5163.24 ± 451.32 cells/mm²) (Fig. 5D and E). These results indicate that the efficacy of ANT-DBS might result from decreased neuronal excitability and alleviated neuronal loss in the hippocampus, especially in DG.

ANT-DBS increases the number of GABAergic neurons and BrdU ⁺ cells in hippocampal regions

We then investigated whether effects of ANT-DBS in the pilocarpine-induced TLE epileptic model were related to changes in GABAergic inhibitory interneurons of the hippocampus. Specific focus was given to PV⁺ and SOM⁺-expressing neurons as two major subtypes of GABAergic inhibitory interneurons in the hippocampus (PV: ~35%, SOM: ~25%) (Pelkey et al. 2017).

Compared to naïve mice, chronic epileptic mice showed a decreased number of PV⁺ neurons in the hippocampi (Fig. 6A,B,G). Surprisingly, chronic epileptic mice that were given ANT-DBS showed an increased number of PV⁺ neurons compared to epileptic mice (Fig. 4C,G). Interestingly, their dramatic changes were observed in DG and CA1 (Fig. 6A–C,G). ANT-DBS restored the number of PV⁺ neurons in chronic epileptic mice to that observed in DG and CA1 of naïve mice (CA1: 27.82 ± 6.61 cells/mm in naïve, 12.76 ± 2.56 cells/mm² in epilepsy, and 34.22 ± 4.77 cells/mm² in epilepsy + ANT-DBS; DG: 14.69 ± 3.21 cells/mm in naïve, 7.34 ± 3.04 cells/mm² in epilepsy, and 15.93 ± 3.75 cells/mm² in epilepsy + ANT-DBS) (Fig. 6G). Likewise, numbers of SOM⁺ neurons were reduced in DG and CA1 of epileptic mice. However, delivering ANT-DBS restored these numbers similar to those of naïve mice (CA1: 20.13 ± 2.41 cells/mm² in naïve, 11.30 ± 1.25 cells/mm² in epilepsy, and 16.44 ± 2.54 cells/mm² in epilepsy + ANT-DBS; DG: 13.54 ± 1.19 cells/mm² in naïve, 6.72 ± 1.97 cells/mm² in epilepsy, and 13.92 ± 2.18 cells/mm² in epilepsy + ANT-DBS) (Fig. 6D–F,H). On the other hand, in the contralateral side of ANT-DBS, the number of PV⁺ and SOM⁺ neurons was less than that of the ipsilateral side and was similar to that of epileptic mice in DG and CA1 (Supplementary Fig. S9). To confirm whether the recovery of PV⁺ and SOM⁺ neurons in number was related to neurogenesis, we performed BrdU assay. Notably, BrdU⁺ cells were increased in the hippocampus of the ANT-DBS group, compared to the naïve and the epilepsy group. Both PV⁺ and SOM⁺ neurons colocalized with BrdU⁺ were increased in DG, compared to the naïve and the epilepsy group (Fig. 6I and J). These results suggest that ANT-DBS can induce neurogenesis of GABAergic interneurons in the hippocampus, which may lead to augmented GABAergic inhibition and reduced chronic seizures.

Discussion

We investigated the underlying cellular mechanism involved in the effect of ANT-DBS on brain activities and seizure severity. Results are summarized in Fig. 7. Delivering ANT-DBS to chronic epileptic mice resulted in a reduction in seizure frequency, a decrease in theta power, and an increase in gamma power in interictal brain activity. ANT-DBS also reduced abnormal locomotor behaviors. Furthermore, the number and c-fos activity of hippocampal neurons were restored following ANT-DBS. In particular, numbers of PV⁺ and SOM⁺ inhibitory neurons were restored in DG and CA1 regions. BrdU-positive inhibitory neurons were apparently increased in DG of the ANT-DBS group, compared to the naïve and the epilepsy group. These results indicate that such neuronal changes might be a driving force for basal brain activity changes and seizure reduction following ANT-DBS.

Theta brainwaves in mice with epilepsy and after ANT-DBS

EEG signals measured in our study might involve neural signals not only from the motor cortex, but also from other adjacent brain regions, including the prefrontal cortex, hippocampus, striatum, and thalamus (Herweg et al. 2016; Seeber et al. 2019). Therefore, the elevated theta-band neural activity might indicate altered neural state in and between multiple brain regions, resulting in higher seizure severity and behavioral changes. Epilepsy patients show an increase in cortical theta-band activity compared to healthy controls (Quraan et al. 2013). Many studies have reported that alterations in specific EEG waves are closely related to epilepsy-related behavioral changes (Holmes and Lenck-Santini 2006).

ANT-DBS is known to affect various brain regions, including the hippocampus, anterior cingulate cortex, and prefrontal cortex (Stypulkowski et al. 2013). These brain regions are known to play a key role in the control of brain network excitability in TLE (Blumenfeld et al. 2004). The reduced theta power of EEG signals by ANT-DBS may indicate that it can potentially modulate brain activity at circuit levels and affect epileptogenic brain networks. We surmise that ANT-DBS may affect the epileptic neural state via both direct and indirect neural connections between different brain regions. The reduced theta power by ANT-DBS has also been reported in human epilepsy (Scherer et al. 2020). Further study is needed to investigate the origin of theta-band EEG signal changes by exploring altered neural activities in various regions.

Effects of DBS on c-fos activity and neuronal loss in hippocampus

In our results, along with theta-band EEG activity, chronic epileptic mice showed a decreased number and an increased c-fos activity in pyramidal layers of hippocampus. When ANT-DBS was applied to chronic epileptic mice, it not only decreased c-fos activity, but also increased number of neurons in the DG. Neuronal loss and hyperactivity have also been reported in other epilepsy studies (Cossart et al. 2001; Encinas and Sierra 2012). Moreover, hippocampal neurons are the most vulnerable to hyperexcitable conditions (Murray and Holmes 2011). Our data showed that ANT-DBS significantly reduced the increase of c-fos activity in chronic epileptic mice. Considering that the expression of c-fos in neurons is an indirect marker of neuronal activity (Hunt et al. 1987; Sagar et al. 1988), ANT-DBS can highly suppress neuronal hyperactivity in chronic epileptic condition. Furthermore, other than neuronal activity, increased c-fos activity might be also related to other downstream processes such as cellular transformation, proliferation, differentiation, and apoptosis (Chinenov and Kerppola 2001; Hu et al. 2002; Güller et al. 2008; Chen et al. 2015). Especially, the expression of c-fos protein is known to play a key role in regulating neuronal cell survival versus death (Preston et al. 1996; Zhang et al. 2002; Chen et al. 2015). Apoptotic neuronal death can be induced by a prolonged increase of c-fos proteins (Ellwardt et al. 2018). Therefore, the elevated c-fos activity in neurons may not only indicate hyperactivity, but also indicate a neuronal state that is prone to cell death. Excessive neuronal activity in chronic epilepsy may result in neuronal loss in hippocampus, especially in DG and CA1 regions (André et al. 2001; Borges et al. 2003; Schartz et al. 2016). Our results suggest that ANT-DBS may alleviate pathological state related to neuronal death by downregulating c-fos activity in neurons.

ANT-DBS changes numbers of PV+ and SOM+ inhibitory interneurons

Our data showed decreases of PV⁺ and SOM⁺ inhibitory interneurons in DG and CA1 of epileptic mice, consistent with a previous study showing loss of PV⁺ and SOM⁺ neurons in the DG of a chronic epilepsy model (Buckmaster and Dudek 1997). Another study has also reported that PV⁺ and SOM⁺ neurons in CA1 are the most vulnerable interneurons in chronic epilepsy (Cossart et al. 2001). Importantly, excessive neuronal activity can initiate negative feedback in multiple intracellular pathways, leading to neuronal degeneration (Motte et al. 1998), which can be more severe and long-lasting in GABAergic inhibitory neurons than in excitatory neurons (Peng and Houser 2005). Given this evidence, hyperexcitable brain activity in chronic epileptic conditions might be primarily responsible for the loss of GABAergic neurons. Loss of inhibitory interneurons in the hippocampus can worsen hyperexcitable conditions in chronic epilepsy since GABAergic interneuron activity is important for suppressing seizure activity and affecting seizure thresholds (Krook-Magnuson et al. 2013). Inhibitory neuronal loss in DG can substantially deteriorate chronic epileptogenesis (Mello et al. 1993; Hattiangady et al. 2004; Dudek and Sutula 2007). A recent study has also shown that inhibiting PV⁺ activity can lead to a reduction in seizure threshold (Drexel et al. 2017). Elimination of GABAergic interneurons in the hippocampus not only can elevate network hyperexcitability, but also can impair cognitive behaviors (Antonucci et al. 2012).

Interestingly, ANT-DBS restored numbers of PV⁺ and SOM⁺ neurons in the hippocampi of chronic epileptic mice to be comparable to those of control naïve mice. Notably, proportions of PV⁺ and SOM⁺ neurons colocalized with a cell proliferation marker (BrdU) were increased in DG compared to those of the naïve and the epileptic mice. These results suggest that ANT-DBS might have neuroprotective and neurogenic effects in the hippocampus as shown in other reports (Encinas,and Sierra 2012; Yang et al. 2015). Recovery of GABAergic interneurons might lead to restoration of brain activity in limbic circuits. Therefore, cellular recovery might have contributed to changes in EEG signals, especially significant reductions in theta-band activity.

We observed changes in PV and SOM neurons at one week after ANT-DBS, indicating that effects of ANT-DBS lasted for 1 week at least. It has been reported that a single 1 h ANT-DBS can induce an increased expression of BDNF for a few hours (Selvakumar et al. 2015). Chen et al. (2017) have also reported that gene and protein expression of NeuN, BDNF, Ki67, and DCX, which are neuroprotective and neurogenic markers, are increased until after 1 week of ANT-DBS. Effects of DBS involve not only neurogenesis, but also anti-inflammatory effects and anti-apoptotic effects through phosphoinositide 3-kinase pathways (Yasuhara et al. 2017). In addition, DBS can induce dramatic cellular changes in stroke animal models. Such changes can last for 30 days (Baba et al. 2009; Morimoto et al. 2011). These studies indicate that DBS can result in long-term cellular changes by increasing neuroprotective and neurogenic substances. Therefore, cellular changes observed in PV and SOM neurons might maintain for a few weeks. Moreover, repeated ANT-DBS at regular intervals may extend its neuroprotective and neurogenic effects. Future study is necessary to confirm how long such cellular changes are maintained after discontinuing ANT-DBS and how long its effects can last if ANT-DBS is repeatedly applied. Overall, our results suggest that specific changes in brainwaves following ANT-DBS might be a manifestation of cellular recovery of GABAergic inhibitory neurons in the hippocampus and restored brain activity due to ANT-DBS.

Clinical relevance

The ANT is considered a useful target for DBS to treat medically refractory epilepsy owing to its distinct connections with the limbic circuit (Child and Benarroch 2013). ANT-DBS has a beneficial role in restoring brain network activity of the limbic circuit (Stypulkowski et al. 2013). It can increase seizure threshold in patients with TLE (Middlebrooks et al. 2018). A recent clinical study has demonstrated that ANT-DBS can desynchronize local field potentials over a broad frequency range within the ipsilateral hippocampus, attenuating the frequency of interictal spikes and high-frequency oscillations within the hippocampus (Yu et al. 2018).

It is currently unclear whether ANT-DBS works better for patients with a particular clinical syndrome. However, patients with TLE show somewhat more favorable outcomes in both seizure reduction and cognitive improvement (Fisher et al. 2010; Oh et al. 2012; Kim et al. 2017), which might be attributable to participation of efferent thalamo-hippocampal projections from the hippocampus along with ANT in the limbic circuit of Papez. Additional clinical experience with well-designed and larger sample studies may help establish the best candidate for ANT-DBS treatment in drug-refractory epilepsy.

Limitations

Although unilateral stimulation is effective enough to be used for clinical treatment (Liu et al. 2012), bilateral stimulation should also be tested since it can be more effective in reducing seizures (Hamani et al. 2004). More importantly, further study is needed to address how long effects of ANT-DBS observed in this study could persist even after DBS is discontinued. Measurements of neural signals in various brain regions in the limbic circuit should be carried out to evaluate effects of ANT-DBS more precisely. It also remains unclear how specific subtypes of inhibitory neurons are differentially affected and restored in specific hippocampal subregions after applying ANT-DBS. According to our observations, the number of BrdU⁺ cells did not apparently increase in the CA1 region, although the number of PV⁺ and SOM⁺ cells was increased. The CA1 region is known to differently express precursor cell markers involved in neurogenesis (Becq et al. 2005). Our results may indicate that ANT-DBS can induce other indirect pathways that lead to its neuroprotective effects or protein profiling changes in inhibitory neurons in CA1 regions. More detailed studies are needed to determine neuronal changes in the CA1 region. We will further explore the underlying cellular mechanisms of ANT-DBS at the limbic circuit level.

Data availability

Requests for further information, resources and reagents should be directed to and will be fulfilled by the corresponding authors, Dr. Young-Min Shon and Dr. Minah Suh. All relevant data are available from the corresponding authors upon reasonable request. The custom-written MATLAB codes used for the analysis are also available from the corresponding authors upon reasonable request for the purposes of academic research.

Funding

This work was supported by the Institute for Basic Science (grant: IBS-R015-D1); Samsung Medical Center (grant: #SMO1161881 and #SMX1200251); the National Research Foundation of Korea (NRF) funded by the Korea government (MSIT) (grant: 2018R1D1A1B07048147 and 2020R1A2C1012017); Basic Science Research Program through the NRF funded by the Ministry of Education (grant: 2021R1I1A1A01059610); and Institute of Information & Communications Technology Planning & Evaluation (IITP) funded by the Korea government (MSIT) (grant: 20200002610011001). This work was also supported by IMNEWRUN Inc.

Notes

We thank Hyesook Lee for her help with immunostaining.

Conflict of interest statement: None of the authors have any conflicts of interest relevant to this study to disclose.

References

Author notes