Abstract
Wang J, Yan J, Li D, He S, Li X, Xing Y, Lai H, Gui Y, Zhang N, Huang W, Yang X. Brain. 2025:awaf024. PMID: 39847527 Epilepsy is a network disorder, involving neural circuits at both the micro- and macroscale. While local excitatory–inhibitory imbalances are recognized as a hallmark at the microscale, the dynamic role of distinct neuron types during seizures remain poorly understood. At the macroscale, interactions between key nodes within the epileptic network, such as the central median thalamic nucleus (CMT), are critical to the hippocampal epileptic process. However, precise mechanisms underlying these interactions remain unclear. In this study, we investigated the microcircuit dynamics within the seizure onset zone and secondary spreading regions, as well as the network connectivity between the hippocampus and the CMT, using a 4-aminopyridine (4-AP) induced hippocampal seizure model. Rats were allocated into three experimental groups. The first group used a 3D tetrode array to monitor hippocampal seizure activity and microcircuit dynamics, including seizure propagation across the macroscale network. In the second group, a chemical lesion was induced in the CMT to assess its impact on hippocampal seizures. In the third group, chemogenetic techniques were used to selectively suppress pyramidal neurons in the CMT and observe changes in neural network connectivity between the CMT and hippocampus during seizures. Offline single-unit sorting was performed using KlustaKwik and further analysis was conducted with CellExplorer. At seizure onset, the narrow interneurons exhibited increased firing rates, initiating recruitment of other neurons, followed by increased activity in pyramidal neuron. Wide interneurons also showed heightened activity subsequent to pyramidal neurons. Interneurons played a more prominent role in the microcircuit during seizures compared to baseline. The CMT exhibited characteristic seizure activity and a decrease in narrow interneuron activity, whereas the cortex did not display seizure activity during hippocampal seizures. Lesioning the CMT resulted in the loss of the tonic component of hippocampal seizures and reduced overall neuronal activity in the hippocampal. Selective suppression of CMT pyramidal neurons resulted in shortened hippocampal seizures while preserving the tonic component. Narrow interneuron activity remained unchanged, while pyramidal neuron and wide interneuron activity significantly decreased. Our findings underscore the critical role of interneurons in the micronetwork of the seizure onset zone and secondary spreading region. Narrow interneurons were particularly vital in seizure initiation, whereas wide interneurons may contribute to seizure termination within the onset zone but not in the secondary spreading region. Pyramidal neurons in the CMT influence hippocampal seizures by modulating of both hippocampal pyramidal neurons and wide interneurons.
Commentary
We now accept that epilepsy is a network disorder, 1 where both micro- and macro-scale networks interact across space and time to drive seizures. While recent work has started to investigate how these complex interactions evolve, 2 many questions remain about how specific cell types and micro-circuits influence seizure initiation and dynamics. Furthermore, at a larger scale, much remains unknown about how local networks within the seizure initiation zone interact with those outside of the area. However, understanding these cross-scale interactions will be essential to develop novel treatment and control strategies. 3
In their recent study, Wang et al 4 made initial steps to address some of these important questions. Using a 3D tetrode array, they were able to simultaneously record both local field potential and single neuron activity across three distinct regions: the motor cortex, the CA1 region of the hippocampus, and the central median thalamic nucleus (CMT). Additionally, the use of tetrode arrays allowed them to further sort the spike data into three distinct types: putative pyramidal neurons, narrow interneurons—putative PV+ cells, and wide interneurons—putative SOM+ cells. In order to reliably induce seizures with a known onset zone, the authors injected 4-aminopyridine (4-AP) into the CA3 region of the hippocampus. This induced a characteristic low-voltage fast (LVF) onset seizure that was first observed in the hippocampus before spreading to the CMT. Interestingly, the cortex did not display any seizure activity in this model, so the authors were confined to studying the interaction between circuitry in the hippocampus and CMT.
The authors first characterize the properties of seizure dynamics in the hippocampus—the seizure onset zone. While the duration and inter-seizure interval varied over time, the seizures were generally characterized by an initial LFV onset during the tonic phase. Just before the sentinel spike of the seizure, the authors saw an initial activation of PV + interneurons which was followed by a subsequent activation of pyramidal neurons, and finally recruitment of SOM + interneurons. The firing rates of the neurons remained high at the beginning of the seizures but then fell as the seizures progressed.
The authors next examined the neuron dynamics as the seizure spread to the CMT. In this case, they did not see an initial activation of any neurons before the sentinel spike and surmised that in this spreading region, the seizure was initiated directly from the hippocampal activity as opposed to being initiated through local circuitry. During the seizure in the CMT, the firing rates of PV+ cells were reduced from baseline, but the firing rates of the pyramidal and SOM+ cells remained at their baseline values, in contrast to what was observed in the hippocampus.
The most exciting part of this study was that the authors next explored how cell and seizure dynamics changed when the interaction between the hippocampus and CMT was disrupted. In the first set of experiments, the authors removed the CMT from the network by using tetrodotoxin to induce a chemical lesion across the entire complex. Despite the fact that no lesions were introduced in the hippocampus, this resulted in a loss of the characteristic LVF onset that was previously observed in the hippocampus. This was accompanied by a significant reduction in the firing rates of all neurons in the hippocampus, indicating that the CMT significantly affects seizure initiation and dynamics in the hippocampus through larger network interactions.
In their second set of experiments, instead of disrupting the entire CMT, the authors chemogenetically inhibited only pyramidal cells in the CMT. In this case, the characteristic LVF onset of seizures remained intact. The authors also observed that the firing rates of PV+ cells in the hippocampus were unaffected, but the firing rates of both pyramidal and SOM+ cells in the hippocampus were decreased when the pyramidal cells in the CMT were inhibited. The duration of seizures was also reduced.
This study demonstrates that what might be considered to be secondary spreading locations of seizures can still have substantial impacts on seizure initiation. Furthermore, specific cell types can differentially drive seizure dynamics through both their local and network connections. Future work will be needed to investigate the roles of other cell types and regional interactions, but this study represents an encouraging start to this type of work.
Although the presented findings are impactful, the study wasn’t without its limits. Here the authors are using the 4-AP model to chemically induce seizures. While this approach does allow them to specifically and consistently define the seizure onset zone, this is not a chronic model of epilepsy that displays known neuronal loss and rewiring. One therefore must question if similar results would be seen in chronic models of epilepsy.
The authors also appear to be limited in the amount of analysis they can perform due to the relatively low number of single neurons that could be recorded during an experiment. While they could record from multiple pyramidal cells, they were able to only record a handful of interneurons in a single recording. The authors tried to use this data to look at causal relationships between the firing of neuron pairs and present many figures showing network diagrams of these interactions. While this type of analysis is of interest, unfortunately, the methodology is not well described in the manuscript, nor are the findings given much discussion in the text. While the evolution of network diagrams showing changes as the seizures progress seems highly relevant, the findings appear anecdotal and cannot be systematically quantified across experiments. Hopefully, future work in this area will focus more on this type of analysis, and methods that can quantify neuron interactions over time and across experiments will be enhanced.
Regardless of these limitations, this work offers insight into the importance of simultaneous investigation of brain dynamics across multiple spatial and temporal scales during epilepsy. As our experimental techniques continue to advance, more multi-modal and multi-scale data will be increasingly available, both in human 5 and animal models. 6 It will be of the utmost importance to develop novel analytical methods to properly interpret this data. The development of new multi-scale computational models of brain dynamics will also be essential in this endeavor. 7 Ultimately, through a combination of computational, analytical, and experimental efforts, it will become possible to tease apart how network structure drives dynamics across all of these scales.
