First Human Tests! Chinese Team Installs “Switches” on Brain Cells, Cutting Seizures by 90%

 

Epilepsy is a chronic neurological disorder caused by abnormal, excessively synchronized discharge of brain neurons. During seizures it may manifest as convulsions, impaired consciousness, sensory abnormalities, or behavioral abnormalities. Based on the distribution of lesions, it can be classified into focal and generalized seizures.

 

Today, there are about 50 million epilepsy patients worldwide, of which roughly 30% have drug‑resistant epilepsy, meaning that seizures cannot be controlled even after the appropriate use of two or more antiepileptic drugs. In the end, many patients can only consider temporal lobe resection surgery, but only about half of drug‑resistant epilepsy patients meet the criteria for surgery, and the cure rate is only between 60% and 80%. When the lesion is close to language, motor or other functional areas, the risk is extremely high.

 

Chemogenetics, which has long been used solely for neuroscience experiments, provides another possibility for this dilemma.

 

On November 5, Professor Zhao Guoguang’s team from Xuanwu Hospital, Capital Medical University, and Professor Luo Minmin’s team from the Institute of Brain Science and Brain‑like Research in Beijing, together with Jianda Kyushu (Beijing) Biotechnology Co., Ltd., published a correspondence in the New England Journal of Medicine (NEJM), proposing an alternative approach: researchers use a one‑time gene delivery to install a drug‑controllable “switch” on neurons in the epileptogenic zone, thereby suppressing the aberrant firing neurons.

 

This is the first time chemical genetics neuro‑modulation has entered the human body,and if future progress proceeds smoothly,the impact of this technology will extend beyond just epilepsy treatment, but also hold promise for Parkinson’s disease, refractory neuropathic pain and other conditions.

 

Manually designed “Silent Switch”

 

The origins of chemical genetics can be traced back to 2007. At that time, Bryan Roth’s team at the University of North Carolina published a study in the Proceedings of the National Academy of Sciences (PNAS), using directed molecular evolution to engineer muscarinic receptors in yeast, creating a family of G‑protein‑coupled receptors that are activated only by inert compounds, including hM4D, which can hyperpolarize hippocampal neurons and achieve neuronal silencing.

 

These artificially designed, receptors that no longer respond to natural neurotransmitters and are activated only by inert small‑molecule ligands are collectively called DREADDs (Designer Receptors Exclusively Activated by Designer Drugs). Over the past 20 years, DREADDs have become one of the most widely used tools in neuroscience labs, allowing researchers to dissect neural circuits and confirm the behavioral regulatory functions of specific neuronal populations.

 

Since its inception, DREADD has never been able to leave the laboratory. Safety concerns about intracranial delivery of viral vectors have persisted for a long time. Brian once said that over the years people have approached him about commercialization, but no one is willing to take on the almost-from-scratch translational risk.

 

The drug used in this study is called GA002, a vector based on adeno‑associated virus serotype 9 (AAV9), carrying a gene that encodes a receptor not found in nature, named hM4D(Gi), which is highly sensitive to a specific small‑molecule drug.

 

This small‑molecule drug is the classic antipsychotic clozapine. After clozapine binds to hM4D(Gi), it suppresses firing via the Gi protein‑coupled signaling pathway. Because the gene expression of GA002 determines that the receptor is primarily expressed in excitatory glutamatergic neurons, which are the main source of seizure‑like discharges in the hippocampus.

 

Compared with traditional gene therapy, the biggest advantage of this approach is controllability: traditional gene replacement therapy delivers a functional gene, after which expression levels are essentially fixed, the dose cannot be adjusted, and the gene cannot be withdrawn. Chemogenetics, on the other hand, intervenes at the gene level once, with oral drugs modulating the treatment intensity.

 

In NEJM simultaneously published commentary, external experts specifically emphasized that one of the study’s highlights lies in adjusting treatment intensity on demand: increase the clozapine dose to strengthen inhibition, reduce or discontinue the dose when problems arise, and the receptors immediately return to a silent state.

 

In three patients, the monthly seizure frequency dropped from 4.33 to 0.33

 

Because GA002 involves both clozapine, which has central activity, and gene therapy, the team must answer a key question before moving into clinical trials: if patients experience fewer seizures, should the improvement be attributed to gene therapy, clozapine, or a combination of both?

 

In response, the team set up four control groups in a mouse epilepsy model: water only, clozapine only, GA002 injection only, and GA002 combined with clozapine. The results showed that only the combination treatment group significantly reduced seizure frequency. Cell experiments also confirmed that neither clozapine alone nor receptor expression could activate downstream signaling pathways; only when both are present can activation occur, providing mechanistic support for advancing clinical trials.

 

The clinical study led by Xuanwu Hospital enrolled three male participants aged 28–37, who were diagnosed with unilateral hippocampal sclerosis causing medial temporal lobe epilepsy, with disease durations of 10–25 years, and all classified as drug‑resistant epilepsy, none of whom had previously undergone epilepsy surgery. Prior to enrollment, the three individuals experienced an average of 4–5 seizures per month.

 

The researchers performed a full‑anesthesia surgery on them, using convection‑enhanced delivery (CED) to continuously deliver GA002 to the head, body, and tail of the patients’ hippocampus via a sustained pressure gradient, thereby covering the entire lesion tissue.

 

During the 8‑week efficacy assessment period, participants’ self‑recorded seizure diaries showed that the monthly average seizure frequency of the three patients dropped from a baseline of 4.33 episodes to 0.33 episodes, an average reduction of 92.6%. Individually, two participants decreased from 4.5 seizures per month to 0.5, while the third went from 4 seizures per month to zero seizures.

 

In addition to subjective records, the study also collected two types of objective indicators of disease status. Electroencephalography showed that the average number of epileptiform spikes per hour during the interictal period decreased from 166 to 67 across the three participants. Imaging revealed that the average hippocampal metabolic asymmetry index (quantifying the metabolic difference between the affected and healthy hippocampi, with higher values indicating more severe disease) dropped from 22.07% to 7.26%.

 

During long‑term follow‑up, one participant unintentionally discontinued the medication for a week about 5 months after dosing, during which two episodes occurred; after resuming the drug the episodes stopped immediately and no further episodes were observed, which is consistent with the expected mechanism of the “drug‑controlled switch”, but the investigators also emphasized that a specifically designed prospective discontinuation and re‑administration protocol is needed to rigorously verify ligand dependence.

 

In terms of safety, no treatment‑related serious adverse events occurred during the trial. As for the issue of vector shedding that requires close monitoring in gene therapy (referring to the viral vector leaving the injection site and entering systemic circulation), laboratory tests showed that vector DNA could be detected in the patient’s blood 2 days post‑operation, and thereafter the vector levels in fluid samples at all time points were below the limit of quantification.

 

It is worth noting that when the team calculated the drug dosage, the starting dose of clozapine was only about 4% of the maximum recommended antipsychotic dose of the drug. The protocol also allowed for stepwise dose escalation, but the three participants reached the predefined response criteria at the lowest dose.

 

A positive outcome does not mean that this therapy will quickly move toward clinical use. In the discussion section of the paper, the team wrote that the study provides evidence that chemogenetic neuromodulation is feasible and preliminarily safe in humans, but it does not constitute proof of efficacy and should not be seen as a replacement for existing temporal lobe resection surgery.

 

The design involved three participants, no control group, and an open‑label format, which cannot rule out placebo effects or the well‑documented “implant effect” in neuromodulation—short‑term improvements caused by the surgical procedure itself rather than the drug.

 

More crucial evidence gaps still remain; at current technical levels, researchers find it difficult to confirm in living human brains whether the receptor is truly expressed in neuronal cells and binds clozapine. The team says they are developing PET tracers for DREADD, planning to deploy them in subsequent clinical trials.

 

(Source: NEJM)

 

A conference that silenced American peers

 

In August 2026, at the annual meeting of the U.S. National Institutes of Health BRAIN Initiative, Brian said that he and a postdoctoral fellow searched the clinical trial registries of both China and the United States and found that at least seven DREADD human trials are underway, all of them in China. According to Chemical & Engineering News, the room fell into a “shocked silence.”According to records from researchers present, Brian also played a video in which Luo Minmin introduced the epilepsy trial.

 

The trials conducted in China currently cover three categories of indications: for epilepsy, in addition to two studies at Xuanwu Hospital, Sanbo Neurology Hospital is also conducting a trial using GA002 to treat focal epilepsy; there are three Parkinson’s disease trials, using vectors named STP-001 and ZR001 that target the ventral thalamic nucleus or substantia nigra, aiming to replace hardware‑requiring deep brain stimulation with chemogenetics; another category focuses on refractory neuropathic pain, conducted by Beijing Tiantan Hospital, injecting the vector into the anterior cingulate cortex, which is responsible for the emotional dimension of pain.

 

Brian believes that the most promising of the seven trials is pain. Taking trigeminal neuropathic pain as an example, this condition is so painful that it becomes a suicide risk factor, and current treatments are very limited. If the trial succeeds, it would essentially open the door to circuit therapies for almost all neuropsychiatric disorders.

 

Look at the field globally; the foundational research in chemical genetics is mainly accumulated in Europe and the United States:DREADD originated in the United States, at University College London (UCL)Matthew Walker’s team has done the most preclinical work focusing on gene therapy for epilepsy. They reported in 2014 in Nature Communications a study on chemically genetic inhibition of focal neocortical seizure activity, and have since continued to explore safer activating ligands to replace clozapine.

 

In terms of clinical translation, the UCL team focuses more on gene therapy based on engineered potassium channels (EKC), i.e., a one‑time delivery, non‑adjustable traditional approach. Focalis Bio, co‑founded by Matthew, is developing an epilepsy gene‑therapy pipeline, with the first human trial planned to launch “within the next five years.”

 

In China, the pace of translation progresses a bit faster.GA002, developed by Jianda Kyushu founded by Luo Minmin, launched a Phase I/II registration clinical study at Xuanwu Hospital in July this year, and simultaneously obtained clinical trial approval from the U.S. FDA, making GA002 the world’s first gene therapy drug based on a chemical genetics technology pathway to enter the registration clinical stage, intended for treating focal refractory epilepsy with clearly defined lesions.

 

When interviewed by Nature, Matthew gave a positive assessment of the Chinese team’s latest research, stating that the efficacy of GA002 is “encouraging.” However, he also expressed his concerns: high‑dose, long‑term use of clozapine is associated with cardiovascular adverse effects, and agranulocytosis can occur as a serious complication even at low doses. Chemogenetic therapy means patients may need lifelong medication, and he looks forward to the development of more suitable and safer ligand drugs in the future.

 

(Source: Science)

 

The significance of this study is not limited to a treatment plan for a single disease. To date, gene therapy development has mainly focused on clear monogenic causes, supplementing or correcting defective genes.

 

However, the intervention logic of chemogenetics is entirely different. Luo Minmin said in an interview that, in principle, any neurological or psychiatric disorder that can be attributed to abnormal neuronal activity in a specific brain region—regardless of whether the cause is genetic—can have a genetic “switch” installed in the cells, which can then be remotely controlled with drugs. In this way, chronic pain, Parkinson’s disease, major depression, and drug addiction could all become potential indications for such therapies.

 

In an interview with Science, Luo Minmin disclosed that in the epilepsy and pain trials, the team has so far treated more than ten patients. Only when these data are made public can we better assess the true potential of this therapy.

 

Reference:

https://www.nejm.org/doi/full/10.1056/NEJMc2606815

https://www.nature.com/articles/d41586-026-03566-2

https://cen.acs.org/pharmaceuticals/drug-development/designer-gene-therapy-chemogenetics-dreadd-epilepsy-clinical-trial/104/web/2026/11

https://cen.acs.org/biological-chemistry/biotechnology/human-trial-chemogenetic-brain-therapy/104/web/2026/08

https://www.pnas.org/doi/10.1073/pnas.0700293104

https://www.stdaily.com/web/gdxw/2026-11/05/content_608943.html

https://www.stdaily.com/web/gdxw/2026-07/06/content_543043.html

https://www.science.org/content/article/novel-gene-therapy-lets-drug-dampen-brain-cell-activity-reducing-seizures

Note: Cover/featured image generated with AI assistance