The Biggest Alzheimer’s Risk Gene May Shrink Brain Cells Years Before Symptoms

APOE4, the most common genetic risk factor for late-onset Alzheimer’s, may begin harming the brain far earlier than thought — shrinking neurons and making them abnormally excitable years before any memory problems appear.

Researchers at the Gladstone Institutes found that in mice carrying APOE4, the variant drives up levels of a protein called Nell2 specifically inside neurons, not their supporting astrocytes. The findings appeared in Nature Aging.

Why APOE4 has been so frustrating

APOE4 is the strongest common genetic risk factor for Alzheimer’s. Carrying one copy raises risk substantially; carrying two raises it a great deal more and shifts onset earlier.

What has been unclear is why. The APOE protein normally transports lipids in the brain, and the E4 variant differs from the common E3 form by very little. Numerous mechanisms have been proposed — impaired amyloid clearance, altered lipid handling, vascular effects, inflammation — and the difficulty has been an abundance of plausible explanations rather than a shortage.

APOE is also produced mainly by astrocytes, the support cells, rather than by neurons themselves, which has made the causal chain harder to trace: a protein made by one cell type somehow damaging another.

What they found

Elevated Nell2 made neurons smaller and more electrically excitable, and this early hyperactivity in the hippocampus — the brain’s memory hub — predicted later cognitive decline.

The specificity matters. The effect was on Nell2 inside neurons, not astrocytes, which identifies where the damage occurs rather than only where the risk gene is expressed.

Why hyperactivity is a meaningful finding

Neurons becoming more active before disease might seem paradoxical, but it fits a pattern observed in humans.

Hippocampal hyperactivity has been detected in people at elevated Alzheimer’s risk and in early mild cognitive impairment, before atrophy and before symptoms. It is not compensation for weakness — it appears to be dysfunction, and excessive neuronal activity is metabolically costly and potentially damaging over time.

There is a further link worth noting: excessive excitability connects to the elevated seizure risk observed in Alzheimer’s, which has often been treated as a complication rather than as a clue to mechanism.

The reversal experiment

Crucially, reducing Nell2 in adult mice reversed these changes.

That is the result that elevates this from description to something actionable. Demonstrating a correlation between a risk gene and a molecular change establishes association; removing the change and watching the phenotype normalise establishes that the molecule is doing the work.

“We found fundamental changes in brain circuits occurring in young mice that still had normal learning and memory,” said Gladstone’s Misha Zilberter.

Senior author Yadong Huang added that “APOE4 accelerates a process that resembles normal aging,” and stressed that “the damage is not irreversible, and that there may be a window for intervention even after disease processes have been triggered.”

Why reversibility in adults is the key detail

Many developmental findings identify a critical window that closes — interesting biologically, useless therapeutically, because by the time anyone could act the opportunity has passed.

Intervening in adult mice and reversing established changes means the process is maintained rather than merely initiated. Something must be actively sustaining the abnormality, and removing that something restores normal function.

For a genetic risk factor present from conception, that is the difference between a finding that explains risk and one that suggests it might be modifiable.

Accelerated ageing rather than a distinct disease

Huang’s framing — that APOE4 accelerates a process resembling normal ageing — carries an implication worth drawing out.

If the variant speeds a process everyone undergoes rather than causing something categorically different, then understanding it may illuminate ordinary brain ageing, and interventions might apply beyond carriers. It also fits the epidemiology: APOE4 shifts Alzheimer’s earlier rather than causing a distinct disease, and non-carriers develop it too, later.

The caveats

This is preclinical mouse research. Mouse models carrying human APOE variants reproduce some features of the human situation and not the disease itself — these mice do not develop Alzheimer’s.

Whether reducing Nell2 helps people would need testing, and reducing a protein throughout the brain is not straightforward: Nell2 has normal functions in neural development and guidance, so suppressing it could carry consequences the mouse experiments were not designed to detect.

What carriers should take from this

APOE status is increasingly known to people who did not set out to learn it, through consumer genetic testing and through screening required before anti-amyloid treatment, so it is worth being clear about what a finding like this does and does not mean for them.

Carrying APOE4 raises risk and does not determine outcome. Many carriers never develop Alzheimer’s, and many people who do develop it carry no copies. The variant shifts probability and timing rather than deciding anything.

This study also describes changes in mice, using a molecule whose role in humans is uncharacterised, with no intervention available. Nothing in it changes what a carrier can do today — which remains managing cardiovascular risk factors, staying physically and cognitively active, and treating conditions such as hearing loss and sleep disorders that are associated with dementia risk.

What it offers is a reason for cautious optimism about the longer term: a specific mechanism, in a risk factor that had none, that turned out to be reversible in the setting where it was tested.

What the work provides is a specific molecular target, a mechanism connecting a well-established risk gene to observable circuit dysfunction, and evidence that the dysfunction is not permanent. This summarises early research and is not medical advice.