What if amyloid plaques are not the whole story of Alzheimer's disease? New research suggests that a broad range of Alzheimer's pathologies may be orchestrated by a molecular switch called ERBB4 when it turns on in the wrong neurons. Once activated, this misplaced signal can set off a chain reaction that disrupts brain circuits, eliminates synapses, activates glial cells, increases amyloid pathology and impairs memory.
A research team led by Associate Director CHUNG Won-Suk at the Center for Vascular Research within the Institute for Basic Science (IBS) has identified aberrant ERBB4 expression in excitatory neurons as an early driver of multiple Alzheimer's disease pathologies. The researchers found that when ERBB4, a receptor normally associated mainly with inhibitory neurons, appears in excitatory neurons, it can trigger neuronal hyperactivity, abnormal synapse loss, reactive gliosis, amyloid accumulation, and cognitive impairment.
Alzheimer's disease is not one malfunction but many. Synapses—the tiny contact points through which neurons communicate—disappear. Neural circuits become unstable. Astrocytes and microglia become reactive. Amyloid plaques accumulate, and memory and thinking abilities gradually decline. Scientists have long struggled to explain why these seemingly different abnormalities emerge and worsen together.
The researchers initially focused on astrocytes and microglia, two types of glial cells that normally support and protect the brain. These cells can remove unnecessary synapses, raising the possibility that they become overactive and destroy too many connections in Alzheimer's disease.
But the results pointed to a more complicated picture.
Using two mouse models of Alzheimer's disease, the team found that astrocytes and microglia increasingly engulfed excitatory synapses while removing fewer inhibitory synapses. This was not simply a global increase in glial "appetite." Instead, the cells were selectively remodeling different parts of the neural circuit.
The researchers then manipulated neuronal activity directly. When neural activity increased, glial synapse engulfment also increased; when activity was suppressed, engulfment declined. The glial cells were not acting alone. They appeared to be responding to abnormal signals coming from neurons.
That observation led the team to ask a critical question: what had gone wrong inside the neurons themselves?
Using single-nucleus RNA sequencing, which profiles gene activity in individual cell nuclei, the researchers identified a distinct population of excitatory neurons that emerged early in disease. These neurons had unexpectedly switched on ERBB4, a receptor that helps cells receive and transmit signals.
In the healthy brain, ERBB4 is expressed mainly by inhibitory neurons—the cells that help prevent neural circuits from becoming overexcited. In the Alzheimer's disease models, however, ERBB4 appeared in a subset of excitatory neurons, which normally drive circuit activity. It was, in effect, the wrong molecular switch turned on in the wrong cells.
The researchers named these cells "Early Responsive Excitatory Neurons", or ERENs.
To determine whether ERBB4 was merely a sign of disease or an active driver of pathology, the team used targeted gene editing to selectively remove Erbb4 from hippocampal excitatory neurons of Alzheimer's disease model mice.
The effects extended far beyond the targeted neurons. Reducing ERBB4 dampened neuronal hyperactivity, helped rebalance inhibitory circuit activity and corrected abnormal synaptic changes. It also reduced reactive changes in astrocytes and microglia, lowered amyloid plaque burden and improved performance in several tests of memory and spatial cognition. The beneficial effects also persisted over time, and reducing ERBB4 after substantial disease progression still reduced several pathological features.
The reverse experiment produced an equally striking result. When ERBB4 was switched on in a small subset of excitatory neurons in otherwise healthy mice, the animals developed circuit hyperactivity, synaptic imbalance, reactive gliosis and cognitive impairment—even without amyloid plaques.
Together, these experiments suggest that abnormal ERBB4 expression is not simply a consequence of Alzheimer's pathology. In mouse models, it can actively drive several major features of the disease at once.
Further experiments identified mTOR signaling as a major pathway through which ERBB4 produces its effects. This places the ERBB4–mTOR axis at a potentially important control point connecting neuronal hyperactivity with synaptic, glial and cognitive abnormalities.
The team also examined whether the findings might be relevant to people. Analyses of postmortem brain samples and transcriptomic data from 446 individuals showed elevated ERBB4 expression in excitatory neurons in Alzheimer's disease. Higher levels of ERBB4-expressing excitatory neurons were associated with greater amyloid plaque burden and poorer cognitive performance, while further statistical modeling linked ERBB4 to amyloid pathology, subsequent tau pathology, and cognitive decline.
These human data do not prove that ERBB4 directly causes Alzheimer's disease in people. They do, however, support the possibility that the abnormal neuronal state identified in mice also occurs in the human brain.
ERBB4 is not the single cause of Alzheimer's disease, and these findings do not mean that an ERBB4-targeted treatment is ready for patients. Rather, the study identifies aberrant ERBB4 expression in excitatory neurons as a potential early driver of a broader pathological cascade, linking several interconnected features of the disease.
That distinction could matter for treatment. Therapies aimed at only one feature of Alzheimer's disease may struggle to control the self-reinforcing damage already spreading through neural circuits. Targeting an early neuronal abnormality such as ERBB4 could instead offer a way to influence several pathological processes at once.
The findings therefore shift the focus from asking which individual hallmark causes Alzheimer's disease to asking what connects those hallmarks. The answer, the researchers suggest, may begin with a small population of neurons that have switched on the wrong signal.
