A simulation approach to integrated metabolic regulation in working skeletal muscle

For over four decades, phosphorus magnetic resonance spectroscopy (31P MRS) has proved a useful tool to study working muscle metabolism in vivo, its pathophysiology (e.g. in neuromuscular, cardiac, vascular, respiratory, endocrine and metabolic disease) and its responses to treatment and training interventions. Key to quantitative interpretation are the relationships between dynamic 31P MRS measurements and the underlying metabolism: in what one might call the traditional approach, the measurements are analysed to make inferences about the processes (Kemp, 2015); by contrast, in this issue of The Journal of Physiology, Disch et al. (2026) model the processes to simulate the measurements. 31P MRS has limitations: it can only detect mobile phosphates at tissue concentrations above ∼0.1 mm, such that its inherently low sensitivity limits both spatial and temporal resolution (although less so at the higher fields increasingly used in research), and absolute quantification of metabolite measurements is complicated. Nevertheless, what 31P MRS can measure, non-invasively and if necessary repeatedly in vivo, includes important components of muscle energetics, some directly detectable (e.g. inorganic phosphate, phosphocreatine (PCr) and ATP), others indirectly calculable (e.g. cytosolic pH, free ADP and AMP and ΔGATP) (Meyerspeer et al., 2020). Many ways have been proposed to use dynamic in vivo 31P MRS measurements in working muscle to make quantitative inferences about metabolic/physiological processes such as ATP usage for force generation, ATP production by oxidative phosphorylation and glycogenolysis to lactate, the ATP-buffering action of creatine kinase, and cellular acid–base physiology (Kemp, 2015). Within the limits of what 31P MRS can measure, these inferences typically depend, explicitly or implicitly, on simplified models of metabolic regulation and more or less well-supported assumptions about unmeasured quantities and processes. Although there is good consensus on the methodology of 31P MRS data acquisition and processing, there is less agreement about such methods of inference (Meyerspeer et al., 2020). One straightforward application of dynamic 31P MRS is the use of post-exercise PCr recovery kinetics as a surrogate for muscle oxidative ATP synthesis capacity. Various analysis approaches have been advocated, but these are largely equivalent in practical terms (Kemp et al., 2015), and PCr recovery measures have been useful in many studies of disease (Meyerspeer et al., 2020); for example, as a trial end point (Charles-Edwards et al., 2019). Similarly, 31P MRS in low-intensity exercise is straightforward to analyse in terms of ATP usage and mitochondrial function (Kemp, 2015). However, 31P MRS in moderate-to-high intensity exercise, which has often proved useful in research, is complicated by the interactions between glycolytic and oxidative ATP synthesis and proton buffering and efflux, and interpretative analysis is therefore more dependent on assumptions (Kemp, 2015). Disch et al. (2026) take a different approach, modelling the underlying physiology and metabolism, taking parameter values as far as possible from independent sources, to fit an experimental 31P MRS dataset, then using this model to probe more deeply. The main strengths of the paper by Disch et al. (2026) are its computational, kinetic and thermodynamic rigour, as well as its clarity about the underpinning assumptions. It is a highly successful proof of principle, a major step forward in computational research in skeletal muscle. There are also important scientific implications. It is important that known mass-action and feedback effects account well for the data, in view of previous proposals that feed-forward mechanisms are necessary to explain key features of dynamic exercise responses in muscle. Another important point is that lactate production is not a consequence of inadequate O2 supply, arising instead from system properties, as what Disch et al. (2026) summarise as ‘a by-product of metabolic flexibility’. The finding of second-order kinetics at higher intensities is also novel in this context (although with some precedent in other experimental models) but definitive comparison is limited by the time-resolution of 31P MRS data; further work will be needed on this, as noted by Disch et al. (2026). Lastly, it is reassuring to know that creatine kinase is indeed near-equilibrium throughout because this assumption underpins the conventional 31P MRS calculation of ADP, AMP and ΔGATP (Meyerspeer et al., 2020). The simulation approach allows Disch et al. (2026) to test the system using conceptual experiments, and to conduct detailed analysis of parameter sensitivities and correlations. It can also go beyond the data: the analysis of proton fluxes and pH throws useful light on a topic which has become somewhat confused in the literature; the calculations of glycolytic intermediates and cytoplasmic redox potential, which are not currently measurable in vivo, are also conceptually valuable. The limitations of the model are fairly stated, and offer clear objectives for future development: additionally modelling glycogen phosphorylase, fatty acid oxidation and the processes of O2 supply and diffusion; including a fuller treatment of proton efflux and a more detailed model of oxidative phosphorylation; and sharpening the focus on post-stimulation recovery kinetics. In short, a rigorous computational approach has enabled Disch et al. (2026) to explore in detail the relationships between dynamic 31P MRS measurements and the interacting processes of energetics, proton handling and redox balance in a working oxidative-glycolytic muscle, going deeper than currently accessible by any non-invasive method. This not only goes well beyond what is possible by conventional interpretative approaches to muscle 31P MRS, but also will help to improve them. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article. No competing interests declared. G.K. was responsible for the conception or design of the work; drafting the work or revising it critically for important intellectual content; and approved the final version of the manuscript submitted for publication. G.K. agrees to be accountable for all aspects of the work. No funding was received.