Roughly 50 million people worldwide have epilepsy, about one in every 130. Scientists have identified more than 1,000 genes that, when disrupted, can individually cause it. And yet, more than half of patients whose epilepsy is suspected to be genetic never receive a genetic diagnosis.
Researchers at the Duncan Neurological Research Institute at Texas Children's Hospital and Baylor College of Medicine have proposed an explanation for part of that gap. In work announced on July 16, 2026, and published in the Journal of Clinical Investigation, they report that epilepsy can be caused not only by a defect in a single gene but also by specific combinations of two or more defective genes acting together.
They also mapped the biological chain that those genes disrupt and found two compounds that quieted seizures in fruit flies carrying the defects.
Starting with Cell Scaffolding, Not Electrical Signaling
Most epilepsy genes studied in detail affect synaptic activity, the communication between brain cells. The Baylor team went somewhere else.
"But what about other seizure-associated genes that do not clearly fall into this category?" said first author Shenzhao Lu, an instructor of molecular and human genetics working in the laboratory of Hugo Bellen. The group focused on seizure-associated genes involved in actin biology.
Actin is a structural protein present in every cell, forming filaments that build and reshape the cell's internal scaffolding. In 2022, the same laboratory reported the first association between defective variants of the human gene TIAM1, which helps regulate actin filament formation in neurons, and a neurological disorder involving developmental delay, intellectual disability, and seizures.
For the new work, they turned to the fruit fly version of that gene, called sif. The mutant flies had seizures. Their actin filaments were shorter than normal and clustered within neurons, a pattern the team observed in both fly and human cells. Notably, the damage was selective: the most affected neurons were excitatory neurons that release glutamate.
A Chain That Runs Through the Mitochondria
The researchers expected damaged scaffolding to manifest as damaged neuronal structures or faulty connections. It did not. "We found that neurons in flies lacking the sif gene were not structurally different from normal neurons, but they were overly active," Lu said.
The answer came from a different property of actin: its role in mitochondrial division. Mitochondria are the structures that supply cellular energy. Neurons with defective actin filaments turned out to have more mitochondria than normal, each one smaller, each one more active, and collectively producing higher levels of reactive oxygen species.
Reactive oxygen species are normally present in small quantities and become damaging in excess. The team's chain runs from reduced actin polymerization, to mitochondrial fission driven by the protein DRP1, to excess reactive oxygen species, to increased glutamate transmission, to seizures. They named it the actin-mitochondria-glutamate pathway, or AMG. The published paper also reports that the glial innate immune pathway, a recently recognized contributor to epilepsy, is activated when the AMG pathway is disrupted.
Two drug experiments in flies supported the model. Mdivi-1, a DRP1 inhibitor that reduces mitochondrial fission, significantly suppressed seizures in the mutants. NACA, an antioxidant compound, reduced seizures and suppressed the excess glutamate signaling.
Why Two Genes Can Do What One Cannot
The diagnostic implication comes from the last part of the study. The team reports that digenic heterozygous loss-of-function variants in AMG pathway genes combine to cause seizures, meaning two half-strength defects that would individually be filed as unremarkable.
The team then looked at human data. According to Bellen, people with epilepsy of unknown origin carry more defective AMG genes than people without the condition, and modeling the specific gene combinations found in those patients in the fruit fly confirmed that many of them increase seizure susceptibility.
That reframes how a genetic test might be read. Current interpretation looks for a single variant severe enough to explain the disease. If the AMG findings hold up, some pairs that are currently dismissed will be diagnosed.
What This Does Not Yet Establish
The caveats are substantial and worth stating plainly.
The mechanistic work was done in fruit flies and cultured cells. Flies are a well-established model for seizure genetics, but a fly is not a person, and a seizure in a fly is not epilepsy in a child.
The human component is a genetic association: patients with unexplained epilepsy carry an increased burden of defective AMG genes compared with controls. Association is not proof of causation, and the confirmation that specific combinations raise seizure risk came from modeling those combinations back in flies.
Mdivi-1 and NACA are research compounds. Neither is an approved epilepsy treatment; neither was tested in humans here, and nothing in this work supports anyone changing their medication. The authors describe the pathway as suggesting potential therapeutic targets, which is a starting point for drug development rather than a result of it.
The realistic near-term payoff is diagnostic rather than therapeutic. If laboratories begin evaluating variant pairs within a defined pathway instead of hunting single culprits, some families who have been told their genetic testing was uninformative may eventually get an answer. Anyone with questions about their own epilepsy or genetic testing should raise them with their neurologist.
Key Questions Answered
What did the researchers find? That epilepsy can result from specific combinations of two or more defective genes rather than only from a single faulty gene, and that these genes converge on a newly described actin-mitochondria-glutamate pathway.
Why does this matter for diagnosis? A genetic cause cannot be found in about half of people with suspected genetic epilepsy. Looking for damaging combinations of variants, rather than one severe variant, may explain some of those unsolved cases.
What is the AMG pathway? A chain in which reduced actin polymerization promotes mitochondrial fission, which raises reactive oxygen species, which increases glutamate signaling and raises seizure susceptibility.
Were any treatments tested? Two compounds reduced seizures in fruit flies: Mdivi-1, which blocks mitochondrial fission, and NACA, an antioxidant. Neither has been tested in people for this purpose, and neither is an approved epilepsy treatment.
How strong is the human evidence? It is an association. People with epilepsy of unknown origin carried an increased burden of defective pathway variants, and those combinations were then modeled in flies. This does not establish causation in humans.
Should anyone change their epilepsy care because of this? No. This is early-stage research. Questions about diagnosis, genetic testing, or treatment should go to a treating neurologist.
