Why Cells Cannot Simply Store ATP
Every living cell runs on adenosine triphosphate, or ATP. Theo Wallimann, who spent his career at ETH Zurich studying the enzyme that manages this currency, describes ATP as the dollar of the cell: a universal unit of exchange that every energy consuming reaction accepts. The energy itself is released when the terminal phosphate group, known as the gamma phosphate, is split off. What remains after that split is ADP, a free phosphate ion and a proton.
This is where the difficulty begins. According to Wallimann, ATP hydrolysis is heavily inhibited by its own products. As ADP, phosphate and protons pile up, the reactions that depend on ATP slow down. A cell that needs a sudden burst of energy, such as a muscle fibre in the first seconds of a sprint or a neuron firing rapidly, cannot solve the problem by hoarding ATP, because burning a large stock quickly would flood the cell with the very products that jam the machinery.
Wallimann explains that nature answered this problem with a separate, metabolically inert store of energy. Rather than keeping large amounts of ATP, cells keep phosphocreatine, a compound that holds a phosphate group at an even higher energy than ATP does. Standard biochemistry references put the free energy released by phosphocreatine hydrolysis at about 43 kilojoules per mole, compared with roughly 30.5 kilojoules per mole for ATP. In his lecture Wallimann rounds these to about 12 and 7 kilocalories per mole; the textbook figures are closer to 10 and 7, but the point stands. Phosphocreatine sits high enough on the energy ladder to hand its phosphate back to ADP and rebuild ATP.
The enzyme that performs this handover is creatine kinase. Wallimann points out that creatine kinase is not alone; adenylate kinase, nucleoside diphosphate kinase and the energy sensing enzyme AMPK all contribute to energy balance. His lecture concentrates on creatine kinase, which he and colleagues have studied for five decades.
Phosphocreatine as a Metabolic Buffer
The textbook picture, as Wallimann describes it, is that the creatine kinase system is a temporal energy buffer. It is expressed most strongly in cells with high and fluctuating energy demand, and in those cells the concentration of phosphocreatine can be several times that of ATP. Wallimann quotes a figure of up to ten times higher; published measurements in skeletal muscle usually place phosphocreatine at around three to four times the ATP concentration, so the exact ratio depends on the tissue and the method, but phosphocreatine is consistently the larger pool.
The practical result is that when a muscle starts to contract, ATP levels stay remarkably stable while phosphocreatine is drawn down. Wallimann notes that ATP concentration is so well defended that it only begins to fall when phosphocreatine is nearly exhausted, at which point the cell is in serious trouble. The rise in AMP that follows activates AMPK, the cell’s emergency energy sensor.
Phosphocreatine has another advantage that Wallimann emphasises: unlike ATP and ADP, it does not react with any other enzyme in the cell. That makes it a safe place to park energy. It also diffuses freely through the cytoplasm, which matters for reasons that become clear later.
Wallimann adds that the system does more than buffer. By removing ADP and protons at the site of ATP use, creatine kinase relieves product inhibition. It also releases inorganic phosphate and free creatine, and he argues that free creatine acts as a signal that stimulates mitochondrial respiration, so the creatine kinase system connects the moment of energy use to the machinery of energy production. A 2011 review in the journal Amino Acids by Wallimann and colleagues summarises this as three roles: a temporal buffer, a spatial energy transport system and a metabolic regulator.
Where Creatine Kinase Lives in the Body
Creatine kinase is found in almost every cell type, and Wallimann links this to the wide range of tissues where creatine supplementation has been reported to have effects. He lists skeletal muscle, heart, smooth muscle, brain, the retina, sperm, skin cells and, notably, immune cells such as macrophages and lymphocytes.
One organ stands out for having very little creatine kinase: the liver. Wallimann explains the apparent paradox. The liver, together with the kidney, is where the body makes creatine. In the two step synthesis described in the major 2000 review in Physiological Reviews, the kidney produces the intermediate guanidinoacetate and the liver methylates it into creatine. Roughly half the body’s creatine comes from this internal synthesis and half from food, mainly meat and fish. The liver exports what it makes. Because neither creatine nor phosphocreatine can cross cell membranes on its own, tissues that take creatine up through a dedicated transporter effectively trap it. That is why muscle and brain are rich in creatine while the liver, which is a factory rather than a consumer, has hardly any creatine kinase to speak of.
Wallimann recalls that creatine kinase was long considered a bulk enzyme, present at high concentration in muscle and evenly spread through the cell. The work he and others did overturned this. Muscle cells contain a soluble pool of the enzyme that performs the general buffering job, but a substantial fraction is bound to specific structures where ATP is consumed, and each of those bound pools forms a functional microcompartment with its own ATP consumer.
There are also several different creatine kinase proteins. Wallimann describes cytosolic isoforms of muscle type and brain type, separable on a native gel, and two mitochondrial isoforms, one specific to striated muscle (the sarcomeric form) and one found in brain, smooth muscle and many other tissues (the ubiquitous form). All are encoded by separate genes and expressed in a tissue specific and developmentally regulated way.
Creatine Kinase Inside the Muscle Cell
The first bound compartment Wallimann describes is the M band, the thin line at the exact centre of each sarcomere. Using fluorescent antibodies on frozen sections of intact muscle, his group showed muscle type creatine kinase decorating the M band precisely, with none in the neighbouring actin only zone. Electron microscopy resolved the enzyme as two fine lines corresponding to the bridging structures that hold the myosin filaments in register.
Wallimann explains that this localisation is specific to the muscle isoform. By genetically engineering fluorescent versions of the enzyme and diffusing them into skinned muscle fibres, his group showed that the muscle type enzyme finds the M band, that mutated versions bind much less well, and that the brain type enzyme does not bind at all. The binding is mediated by a cluster of positively charged lysine residues near the start of the protein, which his laboratory called a charge clamp, and it engages a specific region of the M band proteins myomesin and M protein.
According to Wallimann, the point of placing creatine kinase here is that the M band sits right where the myosin heads pull on actin. ATP is hydrolysed at those heads during each stroke of contraction. An enzyme anchored a few nanometres away, fed by freely diffusing phosphocreatine, can rebuild ATP on the spot and clear ADP before it accumulates.
The proof that this matters came from genetically modified mice, work led by Bé Wieringa’s group in Nijmegen. The first animals lacking the soluble muscle creatine kinase were published in the journal Cell in 1993. Wallimann describes the result of stimulating their muscles repeatedly: force falls away rapidly compared with normal mice. The published study is careful about the details. Single contraction force in these mice was normal, but the muscles lacked the ability to perform burst activity, and the rate of phosphate exchange between phosphocreatine and ATP was reduced at least twentyfold. In Wallimann’s words the phenotype is very significant, and it shows the enzyme is needed directly for force development under repeated demand. Mice lacking both the cytosolic and the mitochondrial isoforms fare worse still.
Wallimann notes that some researchers dismissed the phenotype because the animals live and move around. His answer is that the knockout has profound consequences for adaptation. Fast twitch skeletal muscle in a normal mouse contains few mitochondria between its myofibrils, relying on the phosphocreatine system to ferry energy across long diffusion distances. In the knockout, mitochondria proliferate and pack themselves close to the contractile filaments, shortening the diffusion path. The 1993 paper confirms increased mitochondrial volume between the myofibrils of fast fibres. Wallimann calls this a lesson in how organisms adapt to a missing enzyme.
The Calcium Pump and Muscle Relaxation
The second compartment Wallimann describes is the sarcoplasmic reticulum, the membrane network that releases calcium to trigger contraction and pumps it back to allow relaxation. Pumping calcium is expensive, and the pump consumes ATP.
Wallimann’s group isolated sarcoplasmic reticulum vesicles by gentle sonication and labelled them with gold tagged antibodies, finding creatine kinase bound tightly to the membrane. In a functional test, the vesicles were given phosphocreatine and a small amount of ADP, and they pumped calcium. When the creatine kinase was inactivated with a specific reagent, pumping stopped. When ATP was added directly instead, the pumping rate was similar. The 1990 paper in the Journal of Biological Chemistry reports that membrane bound creatine kinase could support a significant fraction of the maximal calcium uptake rate and that its capacity to regenerate ATP matched the pump’s rate of ATP consumption. Wallimann concludes that this is a functional microcompartment in which the pump’s apparent affinity for ATP is improved by being kinetically coupled to the enzyme.
The knockout mice back this up. Wallimann reports that muscles lacking creatine kinase show a modest slowing of relaxation, which becomes larger with repeated stimulation. The animals partly compensate by expanding their sarcoplasmic reticulum.
The most interesting step is the move to humans. Wallimann cites the work of Peter Hespel’s group in Leuven, who measured relaxation time in volunteers before and after creatine loading. In the 1999 study, sixteen men took either 20 grams of creatine monohydrate a day or a placebo for five days, then performed twelve maximal elbow flexions. Relaxation time fell by roughly 20 percent in the creatine group, consistently from the first contraction to the last, while peak torque and contraction time did not change. Wallimann’s reading is that creatine has a measurable effect on muscle physiology beyond raw strength: in repeated sprint efforts, a muscle has to relax quickly as well as contract quickly. The study is small and it measured one muscle group under laboratory conditions, so it is supportive rather than conclusive, but the direction of the effect matches the mechanism.
Wallimann also mentions, without going into detail, that the same principle has been shown for other membrane ATP consumers. He credits work by Petrus Andrée on ATP gated potassium channels and the sodium potassium pump, where local nucleotide concentrations near the membrane differ from the bulk cytoplasm and creatine kinase is closely associated with the transporters. He describes a further compartment at the I band of the sarcomere, where a more weakly bound pool of creatine kinase associates with glycolytic enzymes such as pyruvate kinase, one of the ATP producers of glycolysis, and can be reconstituted by infusing the enzymes into skinned fibres.
Mitochondrial Creatine Kinase and Energy Transport
A small fraction of the cell’s creatine kinase, first detected in 1964, sits inside mitochondria. Wallimann recounts pioneering experiments by Bessman, Kammermeier, Jacobus and Saks showing that when respiring mitochondria are given creatine, oxygen consumption rises. In his own laboratory’s later work with visiting scientist Laurence Kay, this stimulation was present in normal muscle and in muscle lacking the cytosolic enzyme, but absent in muscle lacking the mitochondrial isoform and in double knockouts. The 2000 paper in the Journal of Biological Chemistry showed that creatine plus ATP increased respiration by about two and a half times compared with ATP alone, even when all ADP outside the mitochondria was mopped up by a trapping system, and that the effect disappeared only when the mitochondrial isoform was missing. The conclusion is that ADP produced inside the mitochondrial intermembrane space by creatine kinase is an important regulator of oxidative phosphorylation.
The story of the enzyme’s structure is one Wallimann tells with evident pleasure. A postdoctoral colleague managed to purify the mitochondrial enzyme, and electron microscopy revealed a molecule with a channel through its middle. Wallimann describes taking the picture to his mentor Hans Eppenberger and announcing that he would work on it for the next ten years. The enzyme turned out to have a strong affinity for membranes and formed two dimensional crystals on lipid films. His doctoral student Thomas Schnyder grew three dimensional protein crystals, and in collaboration with the crystallography group of Wolfgang Kabsch at the Max Planck Institute in Heidelberg, the X ray structure was solved and published in Nature in 1996. It shows an octamer: four dimers arranged around a central cavity, forming a cube about 93 angstroms on a side with a channel roughly 20 angstroms wide running through it. The human ubiquitous mitochondrial enzyme was solved later by Michael Eder and Uwe Schlattner at 2.7 angstrom resolution and proved very similar.
Wallimann describes the catalytic mechanism as direct, in line transfer of the phosphate group from ATP to creatine, with a flexible loop closing over the active site once the nucleotide binds. He states that the enzyme turns over roughly 200 times per second; that figure could not be independently confirmed for this article and should be treated as his estimate.
The octamer’s shape explains its job. Wallimann points out that the flat faces of the cube carry clusters of positive charge and a hydrophobic patch that bind to negatively charged mitochondrial membranes. Because the molecule is symmetrical, it can bind the inner and outer membranes at the same time, physically linking them. In that position it interacts with the adenine nucleotide translocator in the inner membrane, which exports ATP, and with the voltage dependent anion channel in the outer membrane. The respiratory chain and ATP synthase make ATP, the translocator hands it to mitochondrial creatine kinase, the enzyme converts it to phosphocreatine, and phosphocreatine leaves through the outer membrane channel into the cytoplasm.
Why Cells Need an Energy Shuttle
Why go to the trouble of converting ATP to phosphocreatine and back? Wallimann’s favourite illustration is the sperm cell. Its mitochondria sit in the midpiece, just behind the head, while the flagellum that has to beat stretches far away. Sea urchin sperm run entirely on mitochondrial energy, and experiments from the 1970s and 1980s, notably by Tombes and Shapiro, showed that when creatine kinase is inhibited, the flagellar wave weakens progressively from the tip. Computer analysis of stroboscopic images in a 1987 Biophysical Journal paper confirmed that inhibiting the enzyme causes bending waves to damp out as they travel along the tail. The interpretation, which Wallimann shares, is that adenine nucleotides cannot diffuse fast enough over that distance, but phosphocreatine and creatine, being smaller and less charged, diffuse considerably better.
Wallimann describes calculations of diffusion flux from the sperm mitochondrion to the axoneme, carried out with colleagues in Nijmegen, showing that ADP is the limiting species. Glycolysis and mitochondria both produce ATP; mitochondrial creatine kinase converts it to phosphocreatine; phosphocreatine travels; and the cytosolic isoforms at the myosin heads or the dynein arms convert it back to ATP where it is needed.
He also addresses a long standing objection. Measurements of creatine kinase flux by phosphorus magnetic resonance appeared to show no net flow, which some took as evidence against any shuttle. Wallimann explains that at the mitochondrial end the reaction runs from ATP to phosphocreatine, while at the consumer end it runs from phosphocreatine to ATP. The two fluxes are equal, opposite and beautifully timed, so they cancel in a whole tissue measurement even though energy is being carried from one place to the other. The apparent enigma dissolves once compartments are taken into account.
Wallimann draws a thermodynamic conclusion from all this. By keeping the ATP to ADP ratio very high right where ATP is consumed, the creatine kinase system maximises the free energy released by each ATP hydrolysis. By keeping that ratio lower at the mitochondria, where ADP is delivered efficiently, it minimises the energy the cell must spend to make ATP in the first place. In his words, the cell gets more out of every ATP it uses and pays less for every ATP it makes. This is a model rather than a measured quantity, but it is consistent with the enzyme’s placement and with the respiration data.
Creatine and Mitochondrial Protection
The second half of Wallimann’s lecture turns from physiology to disease. He observes that a long list of degenerative conditions, from muscle wasting to neurodegeneration, share a common denominator: low ATP, calcium overload and excess production of reactive oxygen species. If the creatine kinase system stabilises energy and calcium handling, creatine might help cells resist these insults.
His group tested this with a transgenic mouse that expresses mitochondrial creatine kinase in the liver, an organ that normally has almost none. Liver mitochondria from these animals were challenged with calcium overload, which opens the mitochondrial permeability transition pore and causes the organelle to swell, a gold standard early event in programmed cell death. Measured by light scattering, mitochondria carrying the enzyme were protected by creatine, whereas ordinary liver mitochondria were not. The 2003 paper by Dolder and colleagues shows that creatine and its analogue cyclocreatine delayed pore opening, that a different analogue which is not a good substrate did not, and that protection vanished when the enzyme’s activity was reduced by leaving magnesium out of the medium. Wallimann’s conclusion is that the enzyme, sitting in the intermembrane space and coupled to the adenine nucleotide translocator, can hold the pore closed as long as creatine is available.
Wallimann then cites what he calls a very important experiment by Meyer and Galina in Brazil, published in the Journal of Biological Chemistry in 2006, showing that mitochondrial creatine kinase, by recycling ADP inside the mitochondrion, minimises the generation of hydrogen peroxide and other oxygen radicals in a creatine dependent way. Efficient ADP recycling keeps the respiratory chain busy and reduces electron leakage. This provides a mechanistic basis for describing creatine as indirectly antioxidant, though Wallimann is careful to root the effect in the enzyme rather than in any radical scavenging by creatine itself.
At the level of intact cells, his group used single myotubes loaded with calcium indicators and exposed them to osmotic shock and calcium overload. Cells pretreated with creatine kept their internal calcium close to normal; untreated cells did not. The same idea was tested in the mdx mouse, an animal model of Duchenne muscular dystrophy where the membrane is fragile and calcium floods in. Wallimann reports that creatine fed mdx mice show much less inflammation and necrosis than untreated animals, and the 2002 paper by Passaquin and colleagues in Neuromuscular Disorders describes reduced muscle degeneration and improved mitochondrial function. Earlier cell culture work by Pulido and colleagues found better calcium handling and survival in creatine treated dystrophic muscle cells.
Wallimann adds a telling negative control: feeding animals a creatine analogue that depletes the cell’s creatine pool causes mitochondria to develop inclusions and abnormal enlargement, a pathology resembling what is seen in some human mitochondrial diseases. The system, in other words, needs its substrate.
Creatine, Brain Cells and Stroke Models
Neurons are among the most energy hungry cells in the body, and Wallimann describes experiments with Gregory Brewer on cultured rat hippocampal neurons exposed to glutamate, which kills cells through overexcitation, and to amyloid beta, the peptide associated with Alzheimer’s disease. Creatine protected against both in a dose dependent manner. Wallimann quotes an improvement in viability from around 20 percent to 90 percent in the medium; the published 2000 paper in the Journal of Neurochemistry describes glutamate toxicity as greatly reduced by creatine above 0.1 millimolar and amyloid toxicity as partially prevented. It also reports that creatine could be added as late as two hours after the glutamate insult and still protect, and, importantly, that protection was weaker in neurons taken from old animals. The exact percentages are the speaker’s; the direction and the dose dependence are supported.
Wallimann then moves to living brains. With Ernst Martin and colleagues at the Children’s Hospital in Zurich, his group used arterial occlusion to produce hypoxia and ischemia in newborn rats and measured brain oedema by imaging. Creatine pretreatment reduced the oedema volume significantly. The 2002 study by Adcock and colleagues in Developmental Neuroscience is the source, and it has since been cited as part of the rationale for exploring creatine in pregnancy and neonatal medicine, an area still at the research stage.
The adult stroke work came from Konstantin Prass and Josef Priller in Berlin, with Wallimann as a collaborator. Mice were fed creatine at one or two percent of their diet and then subjected to a 45 minute occlusion of the middle cerebral artery. Infarct volume fell significantly. The 2007 paper in the Journal of Cerebral Blood Flow and Metabolism reports a 40 percent reduction in infarct size, and, using magnetic resonance techniques, an increase in cerebral blood flow through the infarcted region during reperfusion, along with better vasodilation in isolated cerebral arteries. Wallimann highlights this as a possible mechanism: creatine fed animals survive stroke better partly because microvascular blood flow into the damaged area is improved. Interestingly, the study found only minor, statistically nonsignificant changes in brain creatine and ATP levels after three weeks of feeding, which suggests the protection does not depend solely on filling the brain’s energy stores. An independent group in Boston had reported in 2004 that oral creatine reduced infarct size and caspase activation in a similar mouse model, so the finding has been reproduced, though only in rodents.
What the Evidence Supports and What It Does Not
Wallimann closes by summarising what he believes fifty years of work have established. The creatine kinase system improves the efficiency and economy of cellular energetics. It enhances the performance of cells. It helps cells deal with metabolic stress, whether from calcium overload, low oxygen, interrupted blood supply or toxins; he mentions that the transgenic liver expressing mitochondrial creatine kinase is strikingly resistant to potent liver poisons. Creatine improves the viability and resilience of cells and counters programmed cell death. In his view, these effects across so many aspects of cell function justify the phrase “creatine for life” and support a role for creatine in healthy aging.
It is worth being precise about how much of this rests on strong evidence. The biochemistry and cell biology described in this article are well established. The isoform structure, the binding of the enzyme to the M band and the sarcoplasmic reticulum, the octameric mitochondrial enzyme, the control of respiration and the phenotype of the knockout mice have all been published in leading journals and reproduced across laboratories. That part of the story is settled science.
The human evidence for creatine’s effects on muscle performance is also strong. The 2017 position stand of the International Society of Sports Nutrition, drawing on hundreds of trials, describes creatine monohydrate as the most effective ergogenic supplement available for high intensity exercise, and finds that supplementation at typical doses, and even up to 30 grams a day for five years in some studies, has been well tolerated in healthy people and in several patient groups. For inherited muscle diseases specifically, a Cochrane review of fourteen randomised trials found that creatine increased muscle strength and improved daily function in muscular dystrophies, but showed no benefit in metabolic myopathies and caused muscle pain and worse function at high doses in McArdle disease. So even within muscle disease, the effect is condition specific.
The protective effects that occupy the second half of Wallimann’s lecture, against mitochondrial permeability transition, oxidative stress, glutamate toxicity, amyloid beta and stroke, are supported by consistent and mechanistically coherent laboratory work. However, nearly all of it comes from cultured cells and rodents. Clinical trials of creatine in human neurodegenerative disease have so far been disappointing; large trials in Parkinson’s and Huntington’s disease did not show slowing of disease. There are no completed trials showing that creatine reduces stroke damage in people. The neonatal and pregnancy applications remain experimental. Claims about healthy aging in humans rest mostly on the muscle and possibly cognitive literature rather than on the cell protection data.
One final line of work that Wallimann mentions shows how the basic science is now being translated. Dialysis patients lose creatine into the dialysis fluid, cannot synthesise enough because their kidneys have failed, and often eat less meat on medical advice, so they become creatine depleted. Wallimann has collaborated with nephrologists in Groningen on adding creatine directly to the dialysate. A double blind, placebo controlled pilot study in sixteen patients, published in 2026, found the approach feasible and well tolerated and found that it raised circulating creatine, which supports a larger trial to test whether it improves quality of life and outcomes. It is early, but it is exactly the kind of study that will tell us whether the promise of the cell biology holds up in patients.
Taken together, the honest summary is this. The mechanism by which creatine works is among the best understood in nutritional science, and its benefits for muscle performance and for some inherited muscle diseases are backed by rigorous human trials. The broader idea that topping up the creatine pool protects cells against energetic stress and supports healthy aging is biologically plausible and well supported in the laboratory, but it remains a promising hypothesis in humans rather than an established fact.