An Enzyme That Rises With Age May Be Part of Why Some Brains Resist Decline

Some people reach old age with brains full of the amyloid deposits that define Alzheimer's disease, and remain mentally sharp. Pathologists have recorded this mismatch for decades, and the explanation usually offered is cognitive reserve: the idea that the ageing brain runs compensatory processes that hold decline at bay.

A study published in Nature Neuroscience on October 15, 2026 identifies one of those processes in molecular form. A team led by researchers at Massachusetts General Hospital reports that an enzyme called PLA2G2F rises in the ageing brain, protects memory in mice, and is most active in the brains of people whose thinking stayed intact despite amyloid pathology.

What cognitive reserve means

The gap between pathology and symptoms is one of the more awkward facts in dementia research. Amyloid plaques and tau tangles are the defining features of Alzheimer's disease, yet how much of either a person has predicts their cognition only loosely. Some people carry a heavy burden of plaques and function normally until they die; others decline steeply with comparatively little.

Cognitive reserve is the framework built to explain this. In its clinical form it is largely a statistical observation, associated with education, occupation and social engagement, which tells you that something protective exists without saying what it is. Showing that a specific molecule is part of that protection, and that changing its level changes outcomes, is a different kind of claim.

Looking for what the brain does about damage

The approach was indirect, and that is the interesting part. Rather than asking what goes wrong in ageing, the group asked what the brain switches on when something goes wrong.

They ran a screen in mice for factors induced by synapse loss in dentate granule cells, the principal neurons of the dentate gyrus, a region of the hippocampus central to forming new memories. One gene, Pla2g2f, responded sharply. It encodes a secreted phospholipase, an enzyme that breaks down fats. Its expression turned out to climb in these neurons during normal ageing.

That pattern, a gene that becomes more active as the brain ages and as synapses are lost, is what a compensatory response should look like. The question was whether it was doing anything useful or simply marking the damage.

Removing it, and adding it back

The mouse experiments answer that directly, in both directions.

When the researchers deleted Pla2g2f from dentate granule cells, ageing went worse. The animals showed accelerated synapse loss, inflammatory microglia, reactive astrogliosis, impaired formation of new neurons, disordered lipid handling, accumulation of lipid droplets, and memory impairment on hippocampal-dependent tasks. Middle-aged mice lacking the gene scored lower on spatial memory tests than controls of the same age.

When they boosted the enzyme in those same neurons during ageing, the pattern reversed. The paper reports that this was sufficient to prevent those changes across hippocampal-cortical circuits, protect memory, and, in the authors' phrase, prophylactically modify the trajectory of cognitive decline. Older animals with extra Pla2g2f outperformed older controls.

The two-way result is what makes the evidence stronger than a correlation. Deleting the gene made ageing worse and adding it made ageing better, in the same cell population, which is the standard test for whether a molecule is doing work rather than passively tracking a process.

"It seems to act as a compensatory cognitive-resilience mechanism," said study co-author Amar Sahay of Massachusetts General Hospital, who added that the enzyme can be used to shift the trajectory of pathological ageing in mice.

Lipids, microglia and a signal between cells

The mechanism the paper describes runs through fat handling rather than through amyloid.

Brain cells accumulate lipid droplets with age, and droplet-laden microglia, the brain's resident immune cells, are associated with inflammation and poor neuronal support. The researchers found that neuronal PLA2G2F acts on this problem from outside the cell that makes it. It is secreted, and it signals to neighbouring cells: in human stem cell-derived microglia, neuronal PLA2G2F reduced the load of lipid droplets and cholesterol esters.

They also tested whether the enzyme's catalytic activity was necessary, by mutating its active site. The mutated version failed to mobilize lipid droplets and failed to reduce reactive gliosis and inflammatory microglia, which indicates the effect depends on the enzyme actually doing chemistry rather than simply being present. The authors describe PLA2G2F as a "broadcast signal" that maintains lipid homeostasis across cell types.

That framing matters for how the finding might be used. A molecule that acts only inside the neuron producing it would have to be delivered to every cell at risk. A secreted signal released by neurons and acting on surrounding glia is, at least in principle, easier to exploit, because raising its level in one population could influence the tissue around it.

The human evidence

Two strands connect this to people. First, analysing existing large datasets, the team found that PLA2G2F activity in the human brain rises with age, mirroring the mouse pattern.

Second, and more pointedly, they worked with researchers holding gene-expression data from brains of deceased donors who had undergone cognitive testing before death. PLA2G2F activity was highest in the resilient group: people whose cognition had remained intact even though their brains carried amyloid-β deposits.

Timothy Hohman of the Vanderbilt Memory & Alzheimer's Center, who was not involved in the work, called it "a fantastic, very thoughtful approach to get at what could be protecting neurons," and "a great example of how you can use a model system to get at some of these resilience mechanisms that are super challenging to study in humans."

Limitations

• The human data are correlational and small. The analysis draws on 13 human brains. An association between enzyme activity and preserved cognition does not establish that the enzyme caused the preservation; both could follow from something else.
• Causal evidence is entirely in mice. Deletion and overexpression experiments are powerful, but mouse ageing is not human ageing, and the history of neurodegeneration research is full of mouse findings that did not transfer.
• Boosting was done genetically. Increasing expression in a defined neuronal population in a laboratory animal is a long way from a drug that raises enzyme activity safely in an older person's brain.
• A phospholipase is an awkward target. These enzymes sit in branched lipid-signalling networks, and related family members have been pursued in the opposite direction. A paper published a day earlier reported that suppressing a different phospholipase, PLA2G7, in neural stem cells improved cognition in ageing mice. Direction of benefit appears to be enzyme-specific.
• Resilience is not treatment. Even if this pathway is real in humans, strengthening a compensatory mechanism may delay decline without addressing the underlying pathology.

What to watch

• Replication of the human association in larger brain-bank cohorts with cognitive testing before death.
• Whether PLA2G2F activity can be measured in living people, through imaging or fluid biomarkers, which would make it testable rather than purely post-mortem.
• Pharmacological rather than genetic ways of raising enzyme activity, and whether they are tolerable.
• Whether the lipid-droplet mechanism holds in human microglia beyond the stem cell-derived systems used here.
• How this fits the broader turn toward lipid metabolism in dementia research, which has become notably active.

This article is for information only and is not medical advice. Treatment decisions should be made with a qualified clinician.

Sources: Nature Neuroscience; Nature news