A muscle signal restores memory in Alzheimer’s disease – without touching plaques

More than 55 million people worldwide have Alzheimer’s disease, yet effective treatments remain elusive. Despite decades of effort, most experimental therapies still target amyloid plaques and chronic neuroinflammation – with only modest effects on memory and daily function.

A growing alternative view is that memory also depends on signals from the rest of the body – chemical messages released by organs such as muscle – and that these signals may still matter even when the brain is already diseased.

In a new collaborative study from the University of Copenhagen and Florida Atlantic University, researchers put that idea to a direct – and risky – test by turning to the body’s largest organ: skeletal muscle. By inducing muscle tissue to produce the exercise-linked protease cathepsin B – a protein released into the bloodstream when muscles are active, acting as a molecular messenger between muscle and brain – they improved movement, memory and hippocampal neurogenesis in a mouse model of Alzheimer’s.

“Muscle is the largest tissue in the body, and yet we rarely think about how signals from muscle influence the brain,” says Atul S. Deshmukh, co-corresponding author of the study and Associate Professor at the Novo Nordisk Foundation Center for Basic Metabolic Research at the University of Copenhagen in Denmark. That blind spot, he adds, may have limited how Alzheimer’s treatments have been conceived.

The work builds on decades of evidence linking exercise to better cognition.

“Exercise releases many signals at the same time,” says Henriette van Praag, Associate Professor at Florida Atlantic University and co-corresponding author. “The challenge has been to identify which of these signals actually matter for the brain.” One candidate is cathepsin B, which Henriette van Praag’s group has previously shown rises after exercise and correlates with memory performance for both animals and humans.

What remained unknown, Hazal Haytural, a Postdoctoral Fellow at the Novo Nordisk Foundation Center for Basic Metabolic Research explains, was whether any signal from outside the brain can still improve memory once neurodegeneration is already underway – or whether the window has already closed.

The body–brain connection Alzheimer’s research has overlooked

For decades, Alzheimer’s research has been dominated by a single framework: amyloid plaques, tau protein pathology and chronic neuroinflammation as the primary drivers of disease. This view has shaped generations of therapies aimed at removing or neutralising these features – often with only modest effects on patients’ memory and daily functioning.

“At the same time, an alternative perspective has emerged,” says van Praag. “Brain function is continuously influenced by signals from the rest of the body, and those signals may become especially important when the brain is under disease-related stress.”

“Alzheimer’s is a very heterogeneous disease,” says Haytural. “Amyloid is important, but it is clearly not the only thing that determines whether cells are still functioning properly.” This heterogeneity helps to explain why targeting a single pathological hallmark has repeatedly failed to produce meaningful cognitive improvement.

Population and animal studies point toward a broader biological picture. Physical inactivity and loss of muscle mass are associated with higher dementia risk, whereas exercise improves memory, mood, brain structure and hippocampal neurogenesis – even later in life.

“From population studies, we know that physical activity is one of the most robust non-pharmacological predictors of brain health,” says Atul S. Deshmukh.

Identifying the mechanism has been difficult, because exercise affects almost every organ at once.

“Exercise biology is incredibly complex,” Deshmukh says – a complexity that has kept the field stuck at correlation rather than mechanism. “A signal that is beneficial in one physiological state may be neutral – or even harmful – in another.”

This context-dependence is one reason that simple “exercise pills” have proved so hard to make.

Cathepsin B: a muscle signal with effects in the brain

Blood-sharing and plasma-transfer experiments reinforce this view, showing that circulating factors can directly modulate brain function.

The protease cathepsin B emerged as one such candidate. Once studied mainly in pathological contexts such as cancer or tissue injury, it has taken on a new role. “We and others have shown that cathepsin B levels increase in the blood after exercise in both humans and animals,” van Praag explains. “Importantly, those levels correlate with memory performance.”

Genetic evidence further strengthened the case. Mice lacking cathepsin B fail to gain the cognitive and neurogenic benefits of running, whereas interventions that boost muscle metabolic activity increase cathepsin B levels and improve cognition – suggesting that this single molecule is required for exercise to influence hippocampal function and related mechanisms, van Praag says.

“However, no one had shown whether a muscle-derived signal like cathepsin B could counteract cognitive decline once neurodegeneration was already underway,” Haytural says. “That was the gap this study set out to address.”

Isolating one signal from the noise of exercise

To test whether a signal originating in muscle alone could influence brain function, the researchers deliberately avoided any direct intervention in the central nervous system. Rather than delivering cathepsin B to the brain or bloodstream, they engineered skeletal muscle to produce the protease itself – ensuring that any effects on the brain would have to occur indirectly, via circulation.

“This discovery has implications for understanding muscle–brain communication,” says Henriette van Praag. “We used gene therapy strictly as a research tool – a way to switch on a single muscle signal for months at a time – not as a treatment. But identifying peripheral signals like this is a necessary step toward evaluating whether gene-therapy approaches could ever be relevant in neurodegenerative disease.”

Protease injections would have produced only transient exposure; using muscle instead enabled a single exercise-linked signal to be isolated over time.

Early action, late test: does the effect last?

The experiments were conducted in a well-established transgenic mouse model of Alzheimer’s disease (amyloid precursor protein and presenilin delta 9 mutations (APP/PS1)), which develops amyloid plaques, neuroinflammation, and progressive cognitive impairment. Four-month-old male mice – before overt symptoms emerge – were randomly assigned to treatment or control groups. Treatment was a single adeno-associated virus injection designed to drive cathepsin B production specifically in skeletal muscle.

The timing was deliberate. “Most Alzheimer’s therapies are now trying to find the right window – before too much synaptic content is lost,” says Atul S. Deshmukh. “That is why the intervention was introduced early, even though the full pathology develops much later.”

Rather than focusing on short-term effects, the researchers intervened early and then waited several months while the disease progressed.

“The idea was to intervene early and then wait,” van Praag explains, “to see whether any benefit would still be visible once the disease phenotype was clearly established.”

Six months later – once clear Alzheimer-like deficits had emerged – the mice underwent a behavioural test battery covering movement and hippocampus-dependent learning and memory.

“No single behavioural test is definitive,” van Praag notes. “You need several that converge on the same brain functions.”

Measuring memory, movement – and brain plasticity

In parallel, newly generated cells in the hippocampus were labelled to quantify adult neurogenesis, a form of brain plasticity known to decline with ageing and Alzheimer’s pathology. Following behavioural testing, brain tissue, skeletal muscle and blood were collected for large-scale proteomic analysis.

“We did not just want to know whether the animals performed better,” Haytural says. “We wanted to understand which biological systems were shifting – in the brain, in the muscle and in the circulation.”

Using proteomics, the researchers quantified thousands of proteins across tissues to see how Alzheimer’s reshapes systemic biology – and how muscle-derived cathepsin B shifts these patterns.

“If muscles really talk to the brain,” Deshmukh says, “you should be able to see the fingerprints across the organism.”

The experiments were conducted in male mice only – a common but important limitation. “Alzheimer’s disease affects women disproportionately,” Haytural notes. “Sex differences in metabolism and brain biology are still underexplored, and they may strongly influence how muscle-derived signals act.” Deshmukh agrees. “With the tools we have now, there is no excuse not to include sex as a biological variable going forward,” he says.

Memory improves – even as plaques stay the same

Six months after treatment, behavioural testing revealed a clear – and in several respects unexpected – pattern. Alzheimer’s mice whose skeletal muscles produced cathepsin B performed markedly better than untreated disease controls across both motor and cognitive tasks – showing improvements in balance, learning and memory.

“What really stood out was how consistent the effect was,” says van Praag. “We saw improvements in movement, spatial memory and fear learning – functions that normally deteriorate in this model.”

In balance and spatial-memory tests, untreated Alzheimer’s mice performed poorly, whereas animals producing muscle-derived cathepsin B performed at levels comparable to healthy controls.

“This is not just about learning the task faster,” Haytural says. “It is about memory retention – whether the animals actually remember what they learned.”

Brain plasticity returns: neurogenesis restored in the hippocampus

The behavioural improvements were mirrored by changes in brain plasticity. In the hippocampus, adult neurogenesis – the birth of new neurons in memory-relevant brain regions – was substantially reduced in untreated Alzheimer’s mice but restored to near-normal levels in animals receiving muscle-derived cathepsin B.

“Loss of neurogenesis is one of the earliest and most robust changes we see in Alzheimer’s disease models,” says Haytural. “Seeing that process preserved, despite ongoing pathology, was striking.”

Equally striking was what did not change. The overall burden of amyloid plaques and markers of neuroinflammation in the brain were largely unaffected by treatment — although the study did not assess whether plaque composition or toxicity may have shifted.

“We did not see fewer plaques, and inflammation was essentially the same,” Deshmukh says. “Yet brain function improved. This challenges the idea that these markers are always tightly coupled to cognition.”

When function and pathology come apart

This dissociation echoes a growing tension in Alzheimer’s research: several recent drugs reduce amyloid burden but produce only modest cognitive gains.

“Our results suggest that, at least over the time frame studied, functional improvements can be achieved without detectable changes in amyloid burden or inflammatory markers,” Deshmukh says.

The researchers are careful not to dismiss amyloid pathology altogether. Plaques may still drive long-term neurodegeneration, even if short- to medium-term cognitive performance can be partly uncoupled from plaque burden. “This is not an either–or situation,” Haytural emphasises and adds that the key is understanding which biological levers affect which outcomes.

Clues to these levers emerged from large-scale proteomic analysis – not through a single dominant target but through coordinated shifts across biological systems.

“Proteomics lets us see whether the system as a whole is moving toward a healthier state – not just in the brain but across the body,” Deshmukh explains.

The study did not test whether blocking these downstream pathways would eliminate the behavioural benefits, leaving open – as Deshmukh emphasises – which molecular changes are necessary and which may be secondary.

In the hippocampus of treated Alzheimer’s mice, proteins involved in messenger RNA metabolism, RNA processing and protein synthesis were upregulated – processes known to be disrupted early in the disease.

“What we see looks like a partial normalisation of the brain’s protein-production machinery,” Deshmukh says. “These processes are tightly linked to synaptic maintenance and plasticity, making them plausible contributors to the functional improvements we observe – even if they are not yet proven to be the direct cause.”

A crucial warning: the same signal harms healthy brains

Although the extent of adult neurogenesis in the human brain remains debated, both researchers stress that the underlying molecular processes – including protein synthesis and synaptic plasticity – are conserved across species and central to memory formation.

“Even if humans generate fewer new neurons than rodents, protein synthesis and plasticity are fundamental to memory across species,” Haytural says.

Across the body, protein profiles in blood and muscle shifted toward those seen in healthy control mice, reinforcing that the intervention acted system-wide. But this broader improvement came with a crucial caveat: in healthy mice, the same intervention impaired memory and triggered disease-like molecular changes – showing that the effect is strongly state-dependent.

“This result was central to how we interpreted the study,” Deshmukh says. “A diseased brain is not just a weaker version of a healthy one – it is biologically different, and it responds differently to the same signals.”

What muscle–brain communication changes about Alzheimer’s

The findings push attention beyond visible lesions toward signalling pathways that link the brain to the rest of the body. If memory and neurogenesis can be improved without reducing plaques or inflammation, a fundamental question follows: should treatment aim primarily to remove pathology – or to preserve function?

“We show that functional outcomes can be changed without altering the classical disease markers,” says van Praag. “This does not mean that plaques are irrelevant, but it does mean that the relationship between pathology and cognition is more complex than we often assume.”

For Haytural, that complexity reflects the biological diversity of the disease itself. “Alzheimer’s is a very heterogeneous condition,” she says. “Amyloid is an important biomarker, but it is clearly not the only process that determines whether neurons maintain their functional capacity.”

The muscle–brain axis highlighted in the study opens a potential therapeutic space, particularly for patients who are unable to achieve the full cognitive benefits of physical exercise. At the same time, the results carry a clear warning: the same intervention that improved cognition in Alzheimer’s mice impaired memory in healthy animals.

“This is not a supplement you would give to everyone,” Deshmukh says. “Biology is deeply context-dependent. A diseased brain responds very differently from a healthy one.”

Timing matters – and so does who benefits

Importantly, the same intervention impaired memory in healthy mice – underscoring that the effect is strongly context-dependent.

“A signal that is beneficial in disease may be disruptive in a healthy system,” Deshmukh says.

“This is not a universal treatment,” adds van Praag. “But it highlights a new biological principle: that muscle–brain communication can meaningfully shape cognitive outcomes in neurodegenerative disease.”

Timing emerges as another critical constraint. In the study, treatment was initiated early – around the onset of disease-related changes but long before severe cognitive impairment. “There is a growing consensus that interventions are most likely to work while the brain still has a degree of plasticity,” Haytural says.

Whether similar body-mediated strategies could help later in the disease course remains an open question.

Taken together, the findings indicate a broader shift in how neurodegenerative diseases may need to be understood mechanistically.

“Rather than thinking of Alzheimer’s as a disease confined to the brain, we may need to think of it as a breakdown in communication across the whole body,” says Deshmukh. “Signals that normally help organs coordinate with the brain may arrive too weakly, too late or not at all – and this shift in perspective opens up new ways of thinking about mechanism, timing and treatment.”