Hippocampal Inhibitory Neuron Loss Is Epileptogenic But Not Likely a Sole Cause of Mesial Temporal Lobe Epilepsy (MTLE)

Abstract

Transient Seizure Clusters and Epileptiform Activity Following Widespread Bilateral Hippocampal Interneuron Ablation.

Dusing MR, LaSarge CL, Drake AW, Westerkamp GC, McCoy C, Hetzer SM, Kraus KL, Pedapati EV, Danzer SC. Transient Seizure Clusters and Epileptiform Activity Following Widespread Bilateral Hippocampal Interneuron Ablation. eNeuro. 2024 Apr 22;11(4):ENEURO.0317-23.2024. doi: 10.1523/ENEURO.0317-23.2024. PMID: 38575351; PMCID: PMC11036118. eNeuro

Interneuron loss is a prominent feature of temporal lobe epilepsy in both animals and humans and is hypothesized to be critical for epileptogenesis. As loss occurs concurrently with numerous other potentially proepileptogenic changes, however, the impact of interneuron loss in isolation remains unclear. For the present study, we developed an intersectional genetic approach to induce bilateral diphtheria toxin-mediated deletion of Vgat-expressing interneurons from the dorsal and ventral hippocampus. In a separate group of mice, the same population was targeted for transient neuronal silencing with DREADDs. Interneuron ablation produced dramatic seizure clusters and persistent epileptiform activity. Surprisingly, after 1 week seizure activity declined precipitously and persistent epileptiform activity disappeared. Occasional seizures (≈ 1/day) persisted to the end of the experiment at 4 weeks. In contrast to the dramatic impact of interneuron ablation, transient silencing produced a large number of interictal spikes, a significant but modest increase in seizure occurrence, and changes in EEG frequency band power. Taken together, findings suggest that the hippocampus regains relative homeostasis—with occasional breakthrough seizures—in the face of an extensive and abrupt loss of interneurons.

Commentary

The contribution of altered hippocampal inhibitory neuronal circuitry to epileptogenesis and mesial temporal lobe epilepsy (MTLE) has been debated for over 40 years. The early clinical observation of endfolium sclerosis, a lesion characterized by neuronal loss limited to the dentate hilus, suggested that this limited lesion was all that was necessary to cause seizures in MTLE patients. ¹ Of course, it was unclear if the hilar neuron loss was a cause or consequence of seizure activity. The dentate hilus is populated by GABAergic inhibitory neurons expressing NPY, somatostatin, and parvalbumin (PV), and excitatory mossy cells. In the normal hippocampus, the hilus regulates granule cell excitability, and the loss of inhibition after hilar cell loss is hypothesized to contribute to epileptogenesis and temporal lobe epilepsy. ² Inhibitory neuron loss in the dentate hilus and other subregions of the hippocampus is a common feature of several experimental models of epilepsy including prolonged perforant path stimulation, ³ traumatic brain injury (TBI) ⁴ and status epilepticus (SE), with seizure frequency often correlated with the degree of hilar cell loss. ⁵ The observation that not all GABAergic neurons are lost in human and experimental sclerotic hippocampal tissue led to the somewhat controversial “dormant basket cell” hypothesis that suggested the loss of inhibition was partially due to the loss of the excitatory mossy cells. ⁶ However, GABAergic neuron and mossy cell loss are variable and often accompanied by synaptic reorganization and alterations in neurogenesis, making it difficult to determine the exact contribution of inhibitory neuron loss to epileptogenesis and spontaneous recurrent seizure (SRS) activity.

The highlighted study by Dusing et al ⁷ was designed to determine the contribution of hippocampal inhibitory neuron loss to epileptogenesis. Two elegant approaches were used to selectively remove or silence hippocampal GABAergic neurons. The first used a genetic approach to bilaterally target dorsal and ventral hippocampal GABAergic neurons expressing the vesicular inhibitory amino acid transporter (Vgat) using diphtheria toxin. The second was to transiently silence the same population of neurons using DREADDs. For both approaches, Slc32a1^{Tm1.1(FlpO)Hze}/J (Vgat-FlpO) mice were treated with viral vectors (AAV9) constructed such that Vgat-expressing neurons expressed either the fluorescent protein mCherry, a diphtheria toxin receptor for ablation, or the silencing DREADD hM4D_i.

For the in vivo ablation experiment, mice with inhibitory neurons expressing the diphtheria toxin receptor received bilateral injections of the toxin into the dorsal and ventral hippocampus, resulting in significant, but not complete, bilateral loss of Vgat-expressing neurons in the dorsal and ventral hippocampus. This was confirmed with immunohistochemical staining for GAD67, PV, somatostatin, and NPY. Control animals whose Vgat cells did not express the diphtheria toxin receptor did not exhibit cell loss. Consistent with the loss of dentate hilar PV neurons, there was a significant loss of granule cell perisomatic parvalbumin inputs, however, there was a less dramatic decrease in granule cell somatic staining for gephyrin, a postsynaptic marker of inhibitory synapses. Video/EEG (24/7) continuously recorded for 2–4 weeks after the injections found that during the first week after the toxin injections the animals exhibited transient clusters of seizures followed by a prolonged period of persistent epileptiform activity. Behavioral seizures were scored using a modified Racine scale. ⁸ However, in the second week, there was a significant drop off in seizure clusters with the animals exhibiting isolated seizure activity of approximately one/day for up to 4 weeks of EEG recording.

For the silencing experiments, mice with hippocampal GABAergic neurons expressing the DREADD were implanted with EEG electrodes and then treated with two injections of the hM4D_i ligand clozapine N-oxide (CNO). Since CNO was administered systemically, the silencing approach had a reduced chance of inducing an injury response that may have been present in the ablation experiment. The electrographic and behavioral response to silencing Vgat-expressing neurons was significantly different from what was observed after ablation. CNO-treated mice exhibited a 10-fold increase in interictal spiking (IS) accompanied by repeated, milder electrographic seizures, with only 20% of the animals exhibiting convulsive seizures. CNO-treated animals did not develop persistent epileptiform activity. An analysis of EEG power revealed a significant increase in gamma power, which was surprising given the role of PV neurons in the generation of gamma activity. Neither the selective removal nor silencing of hippocampal GABAergic neurons resulted in the development of SE.

The results from the current study ⁷ support the hypothesis that loss of hippocampal inhibitory neurons is epileptogenic. However, the difference in seizure response between the two approaches used to selectively remove these neurons and experimental confounds, including the occurrence of spontaneous seizures in control animals and the potential of an injury response associated with ablation, make it difficult to conclude that inhibitory neuron loss alone can be the sole cause of the long-term SRS seen in MTLE. In the ablation experiment, there was an abrupt drop off in seizure activity one week after interneuron ablation. This was also observed in an earlier study that used diphtheria toxin to unilaterally remove GABAergic neurons in CA1. ⁹ However, a somewhat different response was observed in a study that used bilateral injections of substance P-saporin to ablate hilar inhibitory neurons. ¹⁰ This study observed limited behavioral seizure activity 4–8 days after the injection. However, 3–4 months after ablation, spontaneous electrographic and behavioral seizures were observed. This seizure activity was accompanied by a sclerotic hippocampus with extensive neuronal loss in the CA1 and CA3 pyramidal cell layers.

The significant differences in the types of seizure activity generated by the two approaches used to selectively remove hippocampal inhibitory neurons are surprising since both methods targeted the same cells. One explanation for the sudden drop off in seizure activity was that the combination of the toxin injection and the death of the inhibitory neurons generated an injury response that contributed to the seizure activity observed the first week after ablation. Another possible explanation was that surviving inhibitory neurons sprouted fibers that re-established inhibitory input to the granule cells. However, this is unlikely to occur within one week. The observation that perisomatic gephyrin staining of the granule cells was not reduced to the same degree as perisomatic parvalbumin inputs suggests that some inhibitory inputs to the granule cells remained in place. The functionality of the sprouted inputs or gephyrin-labeled inhibitory synapses could be determined by whole cell recording of miniature IPSCs in the granule cell during the first and second weeks after ablation. It is possible that if the EEG sampling period for the ablation study were extended several months or if the treatment with CNO was prolonged, more robust SRS activity would have developed.

In conclusion, the study by Dusing et al ⁷ provides additional evidence that the selective, bilateral loss of hippocampal inhibitory neurons is epileptogenic but by itself may not be sufficient to cause the development of spontaneous recurrent seizures common in MTLE. It is likely other changes such as synaptic reorganization and alterations in neurogenesis ¹¹ contribute to the eventual development of MTLE with the hippocampal inhibitory neuronal loss being the trigger or combining with other proepileptogenic changes. ¹²