Abstract
Teoh J, Simko J, Camp CR, Liu CJ, Wang W, Williams D, Ma L, Soundararajan D, Martin C, Taylor NK, François E, Petri S, Kanber A, Ravichandra A, Pero ME, Bartolini F, Swayne TC, Lutz CM, Zuberi A, Rubinstein M, Hausman Kedem M, Yuan H, Gelinas JN, Sands TT, Harper SQ, Traynelis SF, Makinson CD, Frankel WN. Brain. 2025 Apr 9:awaf125. doi:10.1093/brain/awaf125. Epub ahead of print. PMID: 40200555. Gain-of-function (GoF) variants in the GRIN2D gene, encoding the GluN2D subunit of the N-methyl-
Commentary
The number of known monogenic “epilepsy genes” has increased in recent years to over 1000. 1 About 90% of these genes have been associated with developmental and epileptic encephalopathies (DEEs), in which their mutations can be highly destructive. 2 Unfortunately, the rapid advancements in gene sequencing have not yet been translated into accelerated development of precision therapies, at least in part due to the genetic heterogeneity and rarity of DEEs. The good news is that the Clustered Regularly Interspaced Short Palindromic Repeats-associated protein 9 (CRISPR-Cas9) revolution in the generation of transgenic preclinical models now facilitates establishing disease causality, developing gene therapies, and testing novel treatments.
When it comes to DEEs, transgenic animals may be perceived as straightforward models compared to those of common epilepsies. One identifies a genetic variant, generates mice that carry the edited, knocked-out, or knocked-in gene, and obtains a model that phenocopies key aspects of the human disease. Nevertheless, even models of monogenic epilepsies have limitations, including species differences, genetic background variability, phenotypic heterogeneity, and off-target effects of genetic editing.3–5 Specifically for DEEs, many mouse models do not fully capture the seizure component of the respective human disease: seizures are subtle, nonconvulsive, or rare.
In this context, a hallmark of the new murine DEE model, developed by Wayne Frankel's group,
6
is a severe epilepsy phenotype with several developmental and behavioral aspects of the human disease. The mice carry the gain-of-function V664I variant in Grin2d, encoding a subunit of the N-methyl-
The development of the new model was challenging. The first attempt involved the use of CRISPR mutagenesis; however, the pups died prematurely. Consequently, the team employed a different genetic approach (“knockout-first”) for maintaining the mutant mouse line. The heterogeneous offspring of these mice, which expressed the mutation throughout the brain, featured better survival than those generated in the first attempts. The team also developed mouse strains that expressed the variant gene in a neuron- and brain region-specific manner.
The Grin2d mice featured abnormal locomotor behavior and complex spike and wave discharge activity (representative of absence seizures), but only when the variant gene was expressed across the brain. Notably, spontaneous seizures were apparent even when the mutation was restricted to inhibitory neurons. The mouse line that exhibited the highest mortality rate (100% by postnatal day 31) was Gad2-Cre, in which the mutation was expressed in all gamma-aminobutyric acid (GABA)ergic neurons. A lower mortality rate, 67%, among mice carrying Gad2;Emx1double-Cre (the mutation expressed in both GABA and forebrain glutamatergic neurons) suggested a partial protective effect for the V664I mutation in excitatory neurons. This finding aligns with the results previously obtained in a murine model of Dravet syndrome. 8 In addition, mortality occurred during the second postnatal week, shortly after the switch from GABA-mediated cortical depolarization to hyperpolarization in mice, potentially implicating GABA in the lethal seizures. Hippocampal neurons exhibited ex vivo elevated responses to electrical stimulation, and hyperexcitability was not normalized by the GABA agonist clonazepam. The latter finding underscores the limited ability of traditional antiseizure medications treatment to re-establish an excitatory–inhibitory balance that is lost.
Following model establishment, an inhibitory RNA sequence was packed into a viral vector and injected into the brain ventricles to partially knock down the overactive gene. The treatment improved survival, seizure burden, and body weight. The best outcomes were achieved with an intermediate dose, which was likely less toxic than the high dose.
Despite this impressive work, some inherent limitations of animal DEE models have not been resolved. Key issues include the extent to which the Grin2D gain-of-function mutation leads to hyperactivity in mice versus humans and the contribution of other genetic, epigenetic, and environmental factors. The differences between the human GRIN2D-DEE and its respective animal model are best demonstrated in an accompanying paper, 7 which describes a proconvulsive effect of ketamine in Grin2d-mutant mice, whereas in patients with GRIN2D-DEEs, ketamine suppressed seizures. Having said that, the accompanying paper also represents an important strength of the study: an external validation of the model at a remote lab by the team of Moran Rubinstein. 7
Interestingly, neither lab reported the testing of the NMDA modulator radiprodil. Radiprodil has been evaluated in 15 patients with GRIN1, GRIN2A, GRIN2B, and GRIN2D mutations in the Honeycomb open-label trial (NCT05818943) and has led to a median reduction of 86% in seizure frequency. The honeycomb trial is still active, and its future findings could have been complemented by data obtained from the Grin2d-mutated mice. Specifically, it would be interesting to know if the pharmacological allosteric modulation of N-methyl-D-aspartate (NMDA) receptors by radiprodil could be as effective as the silencing of Grin2d gene, at least in mice.
The featured study demonstrates the advances made since Merritt and Putnam's work on cats with the resultant discovery of phenytoin. 9 Yet even with the best animal model, the road to therapies is long. A key bottleneck in the development of precision treatments for DEE is their testing in clinical trials. As of August 2025, none of the 46 FDA-licensed gene or cell-based treatments target variant genes associated with epilepsies 10 (cerliponase alpha is a precision but not genetic therapy, and its primary indication is not treating seizures). Several therapeutic candidates are under evaluation in relevant populations, for example, children with SCN1A-positive Dravet syndrome, 5 but the road to precision treatments in other DEEs is still long.
The efforts made by Dr Frankel's team bring to mind the story by Loren Eiseley about a child who tosses starfish one by one back into the ocean: They probably can’t help all the children with DEEs, but they might be able to make a difference for those who carry GRIN2D variants.
