When insulin falters, exercise finds another route

Imagine glucose building up in your bloodstream while the muscles that normally help clear and use it for energy are struggling to do so.

This is part of what happens in type 2 diabetes: insulin’s message to the muscles becomes weaker, so glucose is cleared less efficiently from the blood. Yet make the muscles contract – by walking, cycling or climbing a flight of stairs – and they can still pull in more glucose and use it for energy, despite responding poorly to insulin.

“You can think of insulin as a message telling the muscle to take up glucose,” says Kei Sakamoto, Professor at the University of Copenhagen and Vice Executive Director of the Novo Nordisk Foundation Center for Basic Metabolic Research. “That message has to be received, relayed and translated inside the muscle. In type 2 diabetes, somewhere in that communication network the message gets lost in translation and no longer produces the same glucose uptake.”

Exercise gets another message through

“What fascinated me was that even when insulin does not work properly, exercise still works,” Sakamoto says. “It can bypass part of the defect and use a different communication network to help the muscle take up glucose from the blood.”

How can the same muscle struggle to hear insulin – yet respond when you start moving? Kei Sakamoto has spent more than three decades tracing this question from whole-body exercise down to the molecular communication networks inside cells.

“Sometimes we focus too much on what is going wrong,” he says. “There can be a lot of problems, and the problems can also be very diverse. Instead, I ask: when there are defects, what still works?”

This way of thinking has shaped Sakamoto’s research for decades – and helped earn him the 2026 EASD–Novo Nordisk Foundation Diabetes Prize for Excellence.

It began with a question he could not answer in a fitness studio in Japan.

The exercise question he could not answer

Before Sakamoto studied molecular signalling, he was a professional aerobics instructor and competitor in Japan, eventually finishing fifth in the 1992 national championships. He taught full-body aerobic exercise and tailored the type and intensity of exercise according to the guidelines available at the time.

Some people came to get fit or lose weight. Others had type 2 diabetes and had been told by their doctors that exercise could help keep their blood glucose within a healthy range.

The benefits of exercise were clear. But the molecular explanation for those benefits was not.

“There was a huge gap. What does exercise actually do to the body? Could we help those who need these beneficial effects most by understanding what it is doing? To answer this, I had to go deeper into molecular biology and signalling.”

This led Sakamoto to pursue a master’s degree in exercise physiology at the University of Tsukuba in 1993.

Kei Sakamoto with fellow students and colleagues at the University of Tsukuba in Japan in 1993, around the time he began studying exercise physiology and asking what exercise actually does inside the body.

Looking for how exercise sends its message

As a graduate student, Sakamoto attended the American College of Sports Medicine Annual Meeting in Florida in 1998, where he heard Laurie J. Goodyear from the Joslin Diabetes Center explain how insulin and exercise regulate glucose uptake inside muscle cells.

“I had no idea that proteins inside the cell were passing messages to one another – what scientists call signalling – to tell the muscle what to do when you exercise,” he says. “I thought: this is what I want to study.”

He travelled to Boston without arranging an appointment with Goodyear.

“Laurie was probably wondering why some Japanese guy had suddenly appeared in the lab,” Sakamoto says. “I told her I wanted to study exercise signalling under her supervision.”

She gave him the opportunity to carry out his PhD research in her laboratory.

At Joslin, the puzzle became clearer: in people with type 2 diabetes, moderate exercise could still produce a marked decrease in blood glucose even when insulin signalling was not working properly.

“That was pretty remarkable to me. Insulin signalling was impaired, but exercise could still work.”

The observation was clear. Understanding the molecular explanation was much more challenging.

“What does exercise mean to muscle cells?” Sakamoto says. “It means physical contraction. Somehow, that physical event has to be translated into changes in metabolism, protein function and gene expression, allowing muscle cells to adapt and reprogramme.”

How does a muscle translate exercise into a signal?

Contraction changes many things at once: calcium levels rise, energy is consumed, metabolites shift and muscle cells experience mechanical stress. One strong candidate for sensing some of these changes was a protein called AMP-activated protein kinase, or AMPK.

“You can think of ATP as the cell’s energy currency. When energy is being used rapidly, AMP levels can rise, signalling that the cell’s energy balance is under pressure,” Sakamoto says. “AMPK monitors that energy state and helps activate pathways that restore it.”

As contracting muscle uses ATP, the cell’s energy balance changes and AMP levels can rise, helping to activate AMPK. AMPK looked like part of exercise’s alternative route to glucose uptake. But one key piece was still missing: what actually switched it on in a working muscle?

Muscle can increase glucose uptake through partly different signalling routes. During contraction, changes in the cell’s energy state can activate AMPK through LKB1, helping to move the glucose transporter GLUT4 to the cell surface so that more glucose can enter the muscle.

Could a tumour suppressor protein link exercise to AMPK?

In 2003, Sakamoto joined Dario R. Alessi’s laboratory as a postdoctoral researcher at the MRC Protein and Ubiquitylation Unit at the University of Dundee in Scotland, a leading centre for research into protein signalling and its role in human disease.

At his interview, Alessi placed a stack of manuscripts on the table.

“There were maybe five or ten,” Sakamoto recalls. “He said: ‘Kei, you can select any research topic you want.’ I eventually picked one about the protein LKB1.”

It seemed an unlikely candidate for regulating exercise metabolism in muscle. At the time, LKB1 was mainly known for its role in cancer. When LKB1 is missing or does not work properly, it can cause Peutz–Jeghers syndrome, a rare inherited condition that increases the risk of several cancers.

Researchers in Dundee and elsewhere had recently shown that LKB1 could activate AMPK in cultured cancer and other cell lines.

“It was clear that LKB1 was important, but it was not clear whether it was the physiological regulator in muscle,” Sakamoto says. “That was the question I wanted to test.”

Sakamoto established methods to make isolated muscles contract using electrical stimulation and to measure what happened to AMPK activity and glucose uptake. But to find out whether LKB1 was really responsible for switching on AMPK during exercise, he needed muscle without LKB1 – and had to see what happened when those muscles were made to contract in the laboratory.

Kei Sakamoto with Dario Alessi during his years in Dundee. In Alessi’s laboratory, Sakamoto investigated whether LKB1 was the physiological regulator that switches on AMPK in working muscle – work that became a major part of his contribution to metabolic signalling research.

Take away LKB1 – and the AMPK signal almost disappears

Sakamoto and his colleagues created mice lacking LKB1 only in muscle. At first, almost nothing seemed wrong with the mice.

“The animals developed normally, and the muscle looked quite normal at rest. Perhaps there is no phenotype,” Sakamoto says.

Then he made the muscle contract.

“In the first experiment, there was almost no AMPK activation,” he recalls. “I was astonished – almost speechless.”

The result was striking. Sakamoto had expected the muscle to have a backup way of switching on AMPK, given how important this energy-sensing system is. Instead, without LKB1, the AMPK signal almost disappeared when the muscle contracted.

He then measured how much glucose the muscle took up immediately after contraction or after treatment with an AMPK-activating drug. Glucose uptake was also reduced.

“This showed that LKB1 was needed to activate AMPK in contracting muscle, and that this pathway contributed to the increase in glucose uptake.”

Insulin, however, still worked.

“That was an important control,” Sakamoto says. “The muscle had not simply lost its ability to take up glucose. Insulin could still work, while an important part of the response to contraction was gone.”

Muscle contraction normally increases glucose uptake markedly. In muscle lacking LKB1, however, this response was strongly reduced both in living mice and in isolated muscle. The experiment provided genetic evidence that LKB1 is an important part of the pathway through which contracting muscle increases glucose uptake (EMBO Journal, 2005) 

A major route – but not the only one

Published in 2005, the study established LKB1 as a major physiological activator of AMPK in skeletal muscle and showed that the pathway contributes importantly to glucose uptake during muscle contraction.

“The data were very clear, and then other groups independently found the same thing.”

A protein best known for its connection to cancer had become part of an insulin-independent route used by working muscle.

But this did not make LKB1–AMPK “the exercise pathway”. Some contraction-stimulated glucose uptake remained, and LKB1 regulates other proteins too.

“We had identified a major route, but not the only route,” Sakamoto says. “Exercise is too complex to be explained by one protein.”

When fuel itself becomes a message

Sakamoto next asked whether cells also had more than one way of sending metabolic information.

“I studied language and communication before I became a scientist, and I think cells are similar,” Sakamoto says. “They have different languages and a kind of grammar. To understand what a cell is doing, you need to understand how those messages are translated.”

To explore this, he studied another important physiological question: how glucose is stored in muscle so that it is ready to be used when the body needs energy during exercise. This stored form of glucose is called glycogen.

Insulin helps trigger glycogen storage, but Sakamoto suspected that the hormonal message was not the whole story.

The clue was glucose-6-phosphate, a molecule formed almost immediately after glucose enters the cell. It can be used to produce energy or stored as glycogen. But Sakamoto found that it can also act as a message of its own, activating glycogen synthase – the protein that builds glycogen.

His group then made a precise change to glycogen synthase so that glucose-6-phosphate could no longer activate it, while leaving the enzyme’s basic function intact. This allowed them to separate the effect of the glucose-derived signal from the usual hormonal signal from insulin.

The experiment revealed a broader principle of cellular metabolism: the molecules produced as cells process nutrients can themselves carry information, helping the cell decide what to do with those nutrients.

Metformin works when insulin does not – but how?

Metformin is one of the most widely used treatments for type 2 diabetes and has been prescribed for decades. One of its important effects is to lower blood glucose by limiting how much new glucose the liver produces.

“For about 10 years after 2001, people thought this was essentially the answer,” Sakamoto says. “The idea was that metformin activates AMPK in the liver, and this would explain how the drug reduced glucose production and lipid accumulation in the liver.”

Then researchers, including Sakamoto as a contributing author, removed AMPK from the liver and treated the mice with metformin. Unexpectedly, the drug still lowered blood glucose.

The results challenged the prevailing explanation. If AMPK was not required, how could metformin lower blood glucose?

Sakamoto began to ask whether another energy-sensing mechanism could help explain the effect. Metformin was known to modestly lower cellular levels of ATP – the cell’s main energy currency – while increasing AMP. That rise in AMP does more than activate AMPK: it can also directly inhibit fructose-1,6-bisphosphatase, or FBP1, a key enzyme the liver needs to make new glucose.

The idea offered a simple biochemical explanation. When cellular energy is low, rising AMP could act as a brake on an energy-demanding process such as glucose production – independently of AMPK.

But testing that idea required more than simply removing FBP1. The researchers needed to make FBP1 resistant to AMP while leaving the enzyme itself working normally. Finding a change that could do one without disrupting the other was the key experimental challenge.

“We changed the AMP-binding regulation while preserving the basic function of the enzyme,” Sakamoto says.

AMP binds directly to the liver enzyme FBP1 and can inhibit its activity when cellular energy is low. Sakamoto and colleagues altered the AMP-binding site so that FBP1 became much less sensitive to AMP while retaining its basic function, allowing them to test whether this interaction contributes to metformin’s glucose-lowering effect (Nature Medicine, 2018).

Another route – but not the whole story

His team then created mice carrying this AMP-resistant version of FBP1. When the mice received metformin, its acute glucose-lowering effect was substantially weakened.

The results supported another route: metformin lowers the energy state of the liver cell; AMP rises; AMP inhibits FBP1; and glucose production falls – without requiring AMPK to carry the signal.

Metformin also activates AMPK, but the findings suggest that AMPK does not directly account for this acute suppression of glucose production in the liver. Instead, AMPK may contribute to other effects of the drug, including effects on lipid metabolism, helping to limit the production of new lipids and promote lipid breakdown.

“I would not say this is the entire story,” he says. “Metformin appears to act through multiple mechanisms and context-dependent effects, including in the intestine and other parts of the body.”

The experiment identified an important contribution to metformin’s acute effect on glucose production by the liver. It did not show that FBP1 explains all of metformin’s glucose-lowering effects in people.

Finding different ways to switch on AMPK

By this point, AMPK was attracting growing interest as a potential target for treating metabolic disease.

In 2012, Sakamoto moved to the Nestlé Institute of Health Sciences on the EPFL campus in Lausanne, Switzerland, where he began thinking more seriously about how basic discoveries could be translated into health benefits.

“We knew that activating AMPK could have beneficial effects,” he says. “But simply switching it on was not enough. We needed to understand how to control it – how strongly, where and for how long – to get the desired effect without causing unwanted effects.”

This became a central question in Sakamoto’s research: how can the same energy-sensing system be controlled differently in different tissues to produce a specific metabolic benefit?

Can AMPK be switched on where it is needed?

When Sakamoto returned to academia at the Novo Nordisk Foundation Center for Basic Metabolic Research at the University of Copenhagen in 2019, he continued his basic mechanistic research on AMPK while turning his attention to a broader challenge: how do you target a useful metabolic pathway without affecting the wrong tissues?

“There were still missing components of AMPK biology,” he says. “We needed to understand the different switches – how we could selectively regulate AMPK to deliver a specific health benefit.”

One organ that matters is the liver, where fat can accumulate in obesity and type 2 diabetes.

“Our current focus is to develop a way to deliver an AMPK activator selectively into the liver,” he says. “The aim is to reduce fat accumulation and avoid its toxic effects specifically in the liver.”

This work has led to a spinout company, but it remains a research programme, not a treatment.

“It is very important to control AMPK activation – to limit it to a particular cell or tissue, and also control the timing and duration,” Sakamoto says.

Whether liver-selective AMPK activation can produce sustained metabolic benefits safely in people remains unknown.

Kei Sakamoto and his research group at the Novo Nordisk Foundation Center for Basic Metabolic Research, University of Copenhagen. The group studies how cells sense energy and nutrients and how these signalling systems might be targeted in metabolic disease.

What still works?

In 2026, Sakamoto receives the EASD–Novo Nordisk Foundation Diabetes Prize for Excellence for his contributions to metabolic signalling and diabetes research.

“When you look at the previous recipients, they represent major discoveries, important applications and contributions to patient care,” he says. “To be included among them is overwhelming.”

But the scientific question that began in a fitness studio remains open.

Sakamoto compares metabolism with travelling through a city when a metro line closes.

“If one route is blocked, there are alternative ways to get to the destination,” he says. “You can take another train, a bus or another route.”

“When something does not function, you try to find a way to bypass it,” Sakamoto says. “If the cell cannot solve the problem itself, perhaps we can use a drug to provide an alternative pathway.”

More than three decades later, he is still asking the question that began in the fitness studio:

What still works?