Glia Preferentially Gobble on GABAergic Neurons

Abstract

GABA-Dependent Microglial Elimination of Inhibitory Synapses Underlies Neuronal Hyperexcitability in Epilepsy

Chen ZP, Zhao X, Wang S, et al. Nat Neurosci. 2025;28:1404-1417.

Neuronal hyperexcitability is a common pathophysiological feature of many neurological diseases. Neuron–glia interactions underlie this process but the detailed mechanisms remain unclear. Here, we reveal a critical role of microglia-mediated selective elimination of inhibitory synapses in driving neuronal hyperexcitability. In epileptic mice of both sexes, hyperactive inhibitory neurons directly activate surveilling microglia via GABAergic signaling. In response, these activated microglia preferentially phagocytose inhibitory synapses, disrupting the balance between excitatory and inhibitory synaptic transmission and amplifying network excitability. This feedback mechanism depends on both GABA–GABA_B receptor-mediated microglial activation and complement C3–C3aR-mediated microglial engulfment of inhibitory synapses, as pharmacological or genetic blockage of both pathways effectively prevents inhibitory synapse loss and ameliorates seizure symptoms in mice. Additionally, putative cell–cell interaction analyses of brain tissues from males and females with temporal lobe epilepsy reveal that inhibitory neurons induce microglial phagocytic states and inhibitory synapse loss. Our findings demonstrate that inhibitory neurons can directly instruct microglial states to control inhibitory synaptic transmission through a feedback mechanism, leading to the development of neuronal hyperexcitability in temporal lobe epilepsy.

Commentary

Neuronal hyperexcitability is a hallmark of many neurologic disorders, including epilepsy. Imbalances in excitatory (mostly glutamatergic) and inhibitory (primarily GABAergic) synaptic transmission underlie abnormal neuronal hyperexcitability. Aberrant synaptic remodeling is a proposed mechanism that creates excitatory/inhibitory (E/I) imbalances, which may contribute to neuronal hyperexcitability and seizures.

Over the past 20 years, it has become increasingly appreciated that microglia, the brain's resident macrophages, actively survey the surrounding neuropil and other microglia. Moreover, microglia-mediated synaptic stripping occurs in a homeostatic fashion during normal development^1–3 and in a pathological fashion in neurodegenerative diseases. ⁴ Despite exhaustive research on the role of microglia in organizing neuronal circuitry in epilepsy and other neurologic conditions, it remains unresolved whether microglia selectively target neuronal populations (excitatory vs inhibitory) and synapses, thereby contributing to seizures.

Using the systemic kainic acid (KA) model of temporal lobe epilepsy, Chen et al. ⁵ exhaustively and eloquently delineate a selective synaptic remodeling mechanism by which neuron-derived GABA and astrocyte-derived complement 3 (C3) align to drive microglia-dependent removal of inhibitory synapses. Within hours of KA-induced neuronal hyperactively, microglial processes contacted VGAT + inhibitory presynaptic boutons, and this interaction persisted up to 21 days post-KA injection. Inhibitory synapses were reduced by ∼25% in the stratum pyramidale within 3 hours of KA injection. By 21 days post-KA, the inhibitory synapses further decreased by ∼65% from baseline. Going further still, Chen and colleagues pharmacologically blocked microglial activity (with minocycline) or depleted microglia (with the anti-CSF1R blocker PLX3397), and these strategies prevented the removal of inhibitory synapses. The histological observations were accompanied by whole-cell patch-clamp recordings that revealed electrophysiological characteristics consistent with an E/I imbalance 5 days after KA administration, further supporting an early microglia-mediated remodeling of neuronal circuitry. These findings align with previous studies revealing an increased association between microglial processes and neurons and reduced GABAergic postsynaptic currents following febrile seizures in young mice. ⁶

To glean mechanistic insight and identify the relevant GABA receptors, the authors used 2 complementary in vivo approaches. In the first experiment, the GABA_B receptor (CGP52432) antagonist was used to block the GABA_B receptor systemically. In the second, mice with conditional genetic ablation of the GABA_BR2 subunit in microglia were generated. In both conditions, microglial engulfment of inhibitory synapses in the stratum pyramidale was partially rescued, and interictal-like discharges were rescued 5 days post-KA. Adeno-associated virus-mediated knockdown of neuronal GABA_BR2 did not rescue inhibitory synapse loss. Finally, conditional ablation of C3 from astrocytes also prevented inhibitory synapse loss in the stratum pyramidale. The authors conclude that inhibitory neuron-derived GABA and astrocyte-derived C3 coordinate to drive microglial elimination of synapses in the latent phase after KA.

Next, the authors asked if targeting C3 attenuates seizure outcomes in the chronic phase. Genetic ablation of C3 from astrocytes attenuated spontaneous seizure frequency and duration by ∼50% at the 21-day time point. Moreover, pharmacological inhibition of C3 also reduced seizure frequency and duration at 21 days. The authors conclude that blocking C3 signaling after the onset of acute seizures exerts antiepileptic effects during the chronic phase. However, this interpretation is based on EEG recordings obtained at a single time point (day 21 post-KA), which may not be sufficient to fully assess the effects of the genetic and pharmacologic manipulations on spontaneous recurrent seizures.

Finally, the authors’ mouse data were supplemented by single-nucleus RNA sequencing (snRNA-seq) on freshly collected cortical tissue from patients with TLE and non-TLE controls (individuals undergoing benign tumor or trauma surgery). Subclustering of the total microglial pool to define microglial phenotypes revealed several distinct cell populations. Not surprisingly, one microglial population expressed high levels of homeostatic markers (TMEM119, CX3CR1, CSF1R). Another pool was enriched in markers previously identified in damage-associated microglia (DAM), including SPP1 and CLEC7A. ⁷ Surprisingly, the homeostatic microglia (HM), not the DAM, expressed high levels of GABA_BR1 and GABA_BR2. Additionally, Gene Ontology (GO) enrichment analysis revealed overactivation of synaptic engulfment and GABAergic pathways in the HM cluster, but not the DAM cluster. These results suggest that the HM pool is activated and involved in inhibitory synapse elimination in TLE patients.

The authors must be commended for their rigorous histological and electrophysiological analyses. The use of male and female mice and 2 timepoints after KA administration adds value and enhances interpretability. However, it is worth noting that the pharmacological agents used to perturb microglia activity are not specific for microglia; both minocycline and PLX3397 have wide-ranging pleiotropic effects. Information regarding the efficiency of the conditional knockout approaches, that is, what percentage of cells were subject to gene ablation, would have enhanced the interpretation of the studies. Despite these reservations, the complementary use of pharmacological agents and conditional genetic approaches provides a more complete picture of the multicellular control of neuronal circuitry that might apply to other neurologic conditions beyond epilepsy.

The findings presented by Chen et al. add to and further reinforce the concept of the neuroimmune connectome, recently reviewed by Wheeler et al. ⁸ Briefly, the neuroimmune connectome comprises the molecular and functional interactions that enable communication between immune cells and neurons. More specifically, the neuroimmune synapse represents a fundamental and specialized multicellular unit of this communication. These concepts can be extended to include system-level networks integrating synaptic units across cells, tissues, and organs. The recognition that neurons are immune effector cells fundamentally alters our understanding of neuroimmunology, as neurons are not limited to their electrophysiological properties. Understanding this might provide a conceptual framework that could serve as a novel platform for therapeutic development for epilepsy.