The Barrier
A rare fibronectin mutation may explain why some people who carry APOE4 stay sharp into their 80s.
A rare mutation explains some of the people who carry the worst Alzheimer's risk gene and stay sharp.
About one in five people carries at least one copy of APOE ε4, the strongest common genetic risk factor for Alzheimer's disease. Some of them never develop it. They stay sharp into their 80s, and for years nobody could say why.
Two years ago a Columbia team noticed that a subset of those resilient carriers also carry a rare mutation in a different gene, FN1, which makes fibronectin, a protein that helps hold tissue together. The new paper, in Nature Aging, with Caghan Kizil as corresponding author, shows what that mutation is protecting them from.
APOE ε4 drives fibronectin to accumulate around the brain's blood vessels. In mice carrying the human APOE ε4 gene, fibronectin in the brain ran at nearly twice the level seen in mice carrying APOE ε3. The lab then pushed the protein both ways. Raise fibronectin in the astroglia and the blood-brain barrier starts to leak. Lower it and the barrier holds. Prabesh Bhattarai, co-first author, said the animal work is what let them move past association.
The damage has a route. Excess fibronectin signals through integrins — the proteins cells use to feel their surroundings — and an enzyme called focal adhesion kinase. That signal throws off three growth factors, VEGF, HBEGF and IGF1, which astrocytes and blood-vessel cells use to maintain the barrier together. Human tissue agreed with the mice. Higher fibronectin in brain tissue and in cerebrospinal fluid tracked with inflamed astrocytes. Badri Vardarajan, who worked on the original FN1 finding, said the independent human datasets converge on the same protein.
Richard Mayeux, chair of Columbia's neurology department, drew the clinical line. A major risk gene is now tied to an early process, and early processes are the ones you might still modify. There is no drug. The mutation is rare, so a therapy would have to recreate what the variant does, not wait for people to inherit it.
This is not the APOE4 paper that sat in yesterday's file. That was Leyla Akay and Li-Huei Tsai at the Picower Institute, on oligodendrocytes and myelin, and it scored 6.65 and never aired. Different gene, different mechanism, different lab.
Key Facts
- APOE ε4: about 1 in 5 people carry at least one copy. Strongest common genetic risk factor for Alzheimer's.
- Protective variant: rare mutation in FN1, the fibronectin gene. Identified in resilient carriers two years ago.
- Mouse result: brain fibronectin nearly twice as high in human-APOE-ε4 mice as in APOE-ε3 mice.
- Causation test: increasing fibronectin made the barrier permeable; reducing it improved barrier function.
- Pathway: integrins, focal adhesion kinase, then disruption of VEGF, HBEGF and IGF1.
- Human data: higher fibronectin in brain tissue and cerebrospinal fluid associated with inflamed astrocytes.
- Corresponding author: Caghan Kizil, Columbia / Taub Institute. Co-first author: Prabesh Bhattarai. Collaborators: Badri Vardarajan, Richard Mayeux.
- Journal: Nature Aging. Columbia release reported 19 October 2026.
Why This Matters
Most Alzheimer's drugs are aimed at amyloid, late, and they have been humbling. This paper points at the blood-brain barrier, and at a mutation that already exists in people who beat the odds. If a treatment can do what that mutation does — limit fibronectin before the barrier fails — it is a different target, and it would act earlier than the drugs that chase plaques. Mayeux is careful, and he is right to be. A mechanism is not a medicine.
What We Don't Know Yet
No patient has been dosed with anything that lowers fibronectin. The mouse work shows the barrier can be pushed in both directions. It does not show that lowering fibronectin prevents dementia in a person. The protective variant is rare, so inheritance is not a strategy. And "sharp into their 80s" describes the carriers the original genetic study found. It is not a promise, and it is not a result of this experiment.