Your Liver, Not Your Legs, May Shield Your Brain From Alzheimer's

Two-year-old mice — the rodent equivalent of seventy-year-old humans — had forgotten how to navigate a water maze. Then researchers at the University of California, San Francisco raised the level of a single liver enzyme. The mice remembered. In an Alzheimer’s model, amyloid plaques shrank and cognitive deficits reversed. The enzyme is called GPLD1. It never enters the brain. It works entirely on the blood vessels that guard it.

Published in Cell on March 5, 2026, the study closes a gap that had been open since 2020. That year, the same lab led by Saul Villeda showed that blood plasma from exercising mice could rejuvenate the brains of sedentary old ones. They identified GPLD1 — a liver-secreted enzyme — as the key factor. The puzzle was how a protein that cannot cross the blood-brain barrier could improve cognition. The answer turned out to be deceptively simple: it repairs the barrier itself.

A Liver Enzyme Patching Holes in the Brain’s Armor

Blood-brain barrier (BBB) — a layer of specialized endothelial cells lining the brain’s blood vessels. It allows oxygen and glucose through while blocking toxins, pathogens, and immune cells. With age, it develops leaks, triggering inflammation and neurodegeneration.

As we age, a protein called TNAP — tissue-nonspecific alkaline phosphatase — accumulates on BBB endothelial cells. Picture it as rust building up on a water pipe: more TNAP, more leaks. Inflammatory molecules seep through, and the brain pays the price.

Exercise activates the liver. The liver secretes GPLD1 into the bloodstream — an enzyme that specializes in cutting proteins tethered to cell membranes via a molecular anchor called GPI. TNAP happens to be one of those tethered proteins. GPLD1 travels through the blood, reaches the brain’s vasculature, and shears TNAP off the endothelial surface.

GPI anchor (glycosylphosphatidylinositol) — a molecular «rope» that attaches certain proteins to the outer face of a cell membrane. GPLD1 cuts that rope, releasing the protein from the surface.

The barrier seals up. Hippocampal inflammation drops. Neurons get a reprieve. And GPLD1 accomplishes all of this without ever setting foot inside the brain — its entire action is external, on the vascular wall.

Six Experiments, One Chain of Evidence

Villeda’s team ran a cascade of experiments, each confirming a different link.

They genetically boosted TNAP on brain vasculature in young mice. Animals that should have been at their cognitive peak failed memory tests — as though they had aged decades in weeks.

Then the reverse. In two-year-old mice, they genetically suppressed TNAP. The barrier tightened. Brain inflammation fell. The mice performed better on the Morris water maze and object recognition tasks.

Next, they increased GPLD1 expression in the liver of elderly mice — no treadmill involved. Transcriptomic analysis revealed that brain endothelial cell profiles had shifted toward a youthful signature.

A pharmacological TNAP inhibitor reproduced the same cognitive benefits as GPLD1 overexpression. This matters enormously for drug development: if delivering GPLD1 as a therapeutic proves difficult, blocking its target could be a viable shortcut.

Transcriptomics — simultaneous measurement of all active genes in a cell. It reveals whether a cell operates in «young mode» or «old mode.»

In an Alzheimer’s transgenic model, either raising GPLD1 or suppressing TNAP reduced amyloid plaque burden and rescued cognitive function.

The most encouraging detail: the intervention worked late in the animals’ lives. «We were able to tap into this mechanism late in life for the mice and it still worked, ” says first author Gregor Bieri.

An Orchestra of Exercise Signals

The liver is not the only organ sending dispatches to the brain during a workout. Muscles release irisin, cathepsin B, and interleukin-6. Platelets release PF4, linked in 2023 to hippocampal neurogenesis rejuvenation. The brain itself ramps up BDNF — a growth factor that strengthens synapses.

Exerkines — biologically active molecules secreted by various organs in response to physical exercise and carried by the bloodstream to distant tissues.

GPLD1 occupies a distinct niche in this orchestra. Other exerkines act on neurons directly — promoting growth, enhancing plasticity. GPLD1 repairs infrastructure: the vascular shield that protects the brain from the outside world. A leaky barrier undermines every other protective mechanism. In that sense, Villeda’s discovery is the missing piece. It explains why exercise can help even after neurodegeneration has begun: the barrier can still be patched, giving the brain a window for recovery.

Mice Are Not Humans (But the Signals Are Encouraging)

The study was published in Cell and has undergone peer review, though all experiments were performed in mouse models.

The mechanistic completeness here is unusual. Six experiments close a full causal loop: TNAP up → barrier leaks → cognition drops; GPLD1 up or TNAP down → barrier seals → cognition recovers. An independent group confirmed in 2024 that TNAP inhibition reduces amyloid burden in mice — the finding does not rest on a single lab’s data alone.

Species differences remain real, though. The human BBB differs from the mouse version in thickness, pericyte density, and transporter profiles. The experiments lasted weeks to months; human Alzheimer’s develops over decades. GPLD1 can potentially cleave over 100 GPI-anchored proteins — the side effects of wholesale clipping remain unknown. The authors hold a patent on the technology and co-founded Ceiba Bio, a conflict worth weighing when evaluating the enthusiasm.

An observational study from 2025 (Marlatt et al.) found that physically active adults aged 65–85 had higher circulating GPLD1 and better cognitive scores. Correlation, not causation — but the direction aligns. Clinical trials in humans are the logical next step, and funding from NIH and Cure Alzheimer’s Fund suggests the scientific community takes these results seriously.

What You Can Do Right Now

While GPLD1 drugs or TNAP inhibitors are years from the clinic, the mechanism they serve is available for free. Every workout triggers your liver to release a dose of GPLD1. The optimal type, intensity, and minimum effective dose for this specific pathway are unknown, but converging evidence from BDNF, irisin, and other exerkine research consistently points to moderate-intensity aerobic exercise — walking, running, swimming, cycling.

Villeda’s work reframes the search. For decades, Alzheimer’s research focused almost exclusively on what happens inside the brain — neurons, synapses, amyloid plaques. «We’re uncovering biology that Alzheimer’s research has largely overlooked, ” he says. „It may open new therapeutic possibilities beyond the traditional strategies that focus almost exclusively on the brain.“ The liver, the vasculature, the barrier — the key to a disease we have spent fifty years failing to defeat at the center may lie at the periphery.

Frequently Asked Questions

What type of exercise raises GPLD1?

The study used mice running in wheels, which most closely resembles aerobic exercise. Data on other exerkines (BDNF, irisin) consistently show benefits from moderate aerobic activity: walking, jogging, swimming, cycling. Whether resistance training raises GPLD1 specifically has not yet been studied.

Can I take GPLD1 as a supplement instead of exercising?

No. GPLD1 is a large enzyme that would be destroyed in the stomach. Oral delivery is not possible. Injectable delivery is theoretically feasible but has not been tested in humans. A more realistic pharmaceutical approach would target TNAP (the protein GPLD1 removes), which may be easier to inhibit with a small-molecule drug.

Does this work in humans, or only in mice?

Direct human evidence is still missing. However, a 2025 observational study found that physically active older adults (ages 65–85) had higher blood levels of GPLD1, which correlated with better cognitive performance. The mechanism is plausible, but a clinical trial is still needed.

Could this help someone already diagnosed with Alzheimer’s?

In a mouse Alzheimer’s model, boosting GPLD1 reduced amyloid plaques and improved memory even at late stages. The authors emphasize the intervention «still worked» late in life. For humans, this is an encouraging signal, but the gap between a mouse model and human disease remains substantial.

How does GPLD1 relate to other known benefits of exercise for the brain?

GPLD1 complements, rather than replaces, other mechanisms. BDNF strengthens synapses, irisin promotes neurogenesis, PF4 rejuvenates the hippocampus. GPLD1 operates at a different level — it repairs the brain’s vascular infrastructure. The blood-brain barrier is the first line of defense; if it leaks, other protective mechanisms become less effective.