Silencing of glycolysis in muscle: experimental observation and numerical analysis

The paradox of fatty-acid β-oxidation in muscle insulin resistance: Metabolic control and muscle heterogeneity

The skeletal muscle is a metabolically heterogeneous tissue that plays a key role in maintaining whole-body glucose homeostasis. It is well known that muscle insulin resistance (IR) precedes the development of type 2 diabetes. There is a consensus that the accumulation of specific lipid species in the tissue can drive IR. However, the role of the mitochondrial fatty-acid β-oxidation in IR and, consequently, in the control of glucose uptake remains paradoxical: interventions that either inhibit or activate fatty-acid β-oxidation have been shown to prevent IR. We here discuss the current theories and evidence for the interplay between β-oxidation and glucose uptake in IR. To address the underlying intricacies, we (1) dive into the control of glucose uptake fluxes into muscle tissues using the framework of Metabolic Control Analysis, and (2) disentangle concepts of flux and catalytic capacities taking into account skeletal muscle heterogeneity. Finally, we speculate about hitherto unexplored mechanisms that could bring contrasting evidence together. Elucidating how β-oxidation is connected to muscle IR and the underlying role of muscle heterogeneity enhances disease understanding and paves the way for new treatments for type 2 diabetes.

Measurements of in vivo skeletal muscle oxidative capacity are lower following sustained isometric compared with dynamic contractions

Human skeletal muscle oxidative capacity can be quantified non-invasively using 31-phosphorus magnetic resonance spectroscopy (31P-MRS) to measure the rate constant of phosphocreatine (PCr) recovery (kPCr) following contractions. In the quadricep muscles, several studies have quantified kPCr following 24-30 s of sustained maximal voluntary isometric contraction (MVIC). This approach has the advantage of simplicity but is potentially problematic because sustained MVICs inhibit perfusion, which may limit muscle oxygen availability or increase the intracellular metabolic perturbation, and thus affect kPCr. Alternatively, dynamic contractions allow reperfusion between contractions, which may avoid limitations in oxygen delivery. To determine whether dynamic contraction protocols elicit greater kPCr than sustained MVIC protocols, we used a cross-sectional design to compare quadriceps kPCr in 22 young and 11 older healthy adults following 24 s of maximal voluntary: (1) sustained MVIC and (2) dynamic (MVDC; 120°·s-1, 1 every 2 s) contractions. Muscle kPCr was ∼20% lower following the MVIC protocol compared with the MVDC protocol (p ≤ 0.001), though this was less evident in older adults (p = 0.073). Changes in skeletal muscle pH (p ≤ 0.001) and PME accumulation (p ≤ 0.001) were greater following the sustained MVIC protocol, and pH (p ≤ 0.001) and PME (p ≤ 0.001) recovery were slower. These results demonstrate that (i) a brief, sustained MVIC yields a lower value for skeletal muscle oxidative capacity than an MVDC protocol of similar duration and (ii) this difference may not be consistent across populations (e.g., young vs. old). Thus, the potential effect of contraction protocol on comparisons of kPCr in different study groups requires careful consideration in the future.

Systematic Bayesian posterior analysis guided by Kullback-Leibler divergence facilitates hypothesis formation

Bayesian inference produces a posterior distribution for the parameters of a mathematical model that can be used to guide the formation of hypotheses; specifically, the posterior may be searched for evidence of alternative model hypotheses, which serves as a starting point for hypothesis formation and model refinement. Previous approaches to search for this evidence are largely qualitative and unsystematic; further, demonstrations of these approaches typically stop at hypothesis formation, leaving the questions they raise unanswered. Here, we introduce a Kullback-Leibler (KL) divergence-based ranking to expedite Bayesian hypothesis formation and investigate the hypotheses it generates, ultimately generating novel, biologically significant insights. Our approach uses KL divergence to rank parameters by how much information they gain from experimental data. Subsequently, rather than searching all model parameters at random, we use this ranking to prioritize examining the posteriors of the parameters that gained the most information from the data for evidence of alternative model hypotheses. We test our approach with two examples, which showcase the ability of our approach to systematically uncover different types of alternative hypothesis evidence. First, we test our KL divergence ranking on an established example of Bayesian hypothesis formation. Our top-ranked parameter matches the one previously identified to produce alternative hypotheses. In the second example, we apply our ranking in a novel study of a computational model of prolactin-induced JAK2-STAT5 signaling, a pathway that mediates beta cell proliferation. Within the top 3 ranked parameters (out of 33), we find a bimodal posterior revealing two possible ranges for the prolactin receptor degradation rate. We go on to refine the model, incorporating new data and determining which degradation rate is most plausible. Overall, while the effectiveness of our approach depends on having a properly formulated prior and on the form of the posterior distribution, we demonstrate that our approach offers a novel and generalizable quantitative framework for Bayesian hypothesis formation and use it to produce a novel, biologically-significant insight into beta cell signaling.

Butyrate oxidation attenuates the butyrate-induced improvement of insulin sensitivity in myotubes

Skeletal muscle insulin resistance is a key pathophysiological process that precedes the development of type 2 diabetes. Whereas an overload of long-chain fatty acids can induce muscle insulin resistance, butyrate, a short-chain fatty acid (SCFA) produced from dietary fibre fermentation, prevents it. This preventive role of butyrate has been attributed to histone deacetylase (HDAC)-mediated transcription regulation and activation of mitochondrial fatty-acid oxidation. Here we address the interplay between butyrate and the long-chain fatty acid palmitate and investigate how transcription, signalling and metabolism are integrated to result in the butyrate-induced skeletal muscle metabolism remodelling. Butyrate enhanced insulin sensitivity in palmitate-treated, insulin-resistant C2C12 cells, as shown by elevated insulin receptor 1 (IRS1) and pAKT protein levels and Slc2a4 (GLUT4) mRNA, which led to a higher glycolytic capacity. Long-chain fatty-acid oxidation capacity and other functional respiration parameters were not affected. Butyrate did upregulate mitochondrial proteins involved in its own oxidation, as well as concentrations of butyrylcarnitine and hydroyxybutyrylcarnitine. By knocking down the gene encoding medium-chain 3-ketoacyl-CoA thiolase (MCKAT, Acaa2), butyrate oxidation was inhibited, which amplified the effects of the SCFA on insulin sensitivity and glycolysis. This response was associated with enhanced HDAC inhibition, based on histone 3 acetylation levels. Butyrate enhances insulin sensitivity and induces glycolysis, without the requirement of upregulated long-chain fatty acid oxidation. Butyrate catabolism functions as an escape valve that attenuates HDAC inhibition. Thus, inhibition of butyrate oxidation indirectly prevents insulin resistance and stimulates glycolytic flux in myotubes treated with butyrate, most likely via an HDAC-dependent mechanism.

Full humanization of the glycolytic pathway in Saccharomyces cerevisiae

Although transplantation of single genes in yeast plays a key role in elucidating gene functionality in metazoans, technical challenges hamper humanization of full pathways and processes. Empowered by advances in synthetic biology, this study demonstrates the feasibility and implementation of full humanization of glycolysis in yeast. Single gene and full pathway transplantation revealed the remarkable conservation of glycolytic and moonlighting functions and, combined with evolutionary strategies, brought to light context-dependent responses. Human hexokinase 1 and 2, but not 4, required mutations in their catalytic or allosteric sites for functionality in yeast, whereas hexokinase 3 was unable to complement its yeast ortholog. Comparison with human tissues cultures showed preservation of turnover numbers of human glycolytic enzymes in yeast and human cell cultures. This demonstration of transplantation of an entire essential pathway paves the way for establishment of species-, tissue-, and disease-specific metazoan models.

Magnetic resonance reveals mitochondrial dysfunction and muscle remodelling in spinal muscular atrophy

Genetic therapy has changed the prognosis of hereditary proximal spinal muscular atrophy, although treatment efficacy has been variable. There is a clear need for deeper understanding of underlying causes of muscle weakness and exercise intolerance in patients with this disease to further optimize treatment strategies. Animal models suggest that in addition to motor neuron and associated musculature degeneration, intrinsic abnormalities of muscle itself including mitochondrial dysfunction contribute to the disease aetiology.

To test this hypothesis in patients, we conducted the first in vivo clinical investigation of muscle bioenergetics. We recruited 15 patients and 15 healthy age and gender-matched control subjects in this cross-sectional clinico-radiological study. MRI and ³¹P magnetic resonance spectroscopy, the modality of choice to interrogate muscle energetics and phenotypic fibre-type makeup, was performed of the proximal arm musculature in combination with fatiguing arm-cycling exercise and blood lactate testing. We derived bioenergetic parameter estimates including: blood lactate, intramuscular pH and inorganic phosphate accumulation during exercise, and muscle dynamic recovery constants. A linear correlation was used to test for associations between muscle morphological and bioenergetic parameters and clinico-functional measures of muscle weakness.

MRI showed significant atrophy of triceps but not biceps muscles in patients. Maximal voluntary contraction force normalized to muscle cross-sectional area for both arm muscles was 1.4-fold lower in patients than in controls, indicating altered intrinsic muscle properties other than atrophy contributed to muscle weakness in this cohort. In vivo³¹P magnetic resonance spectroscopy identified white-to-red remodelling of residual proximal arm musculature in patients on the basis of altered intramuscular inorganic phosphate accumulation during arm-cycling in red versus white and intermediate myofibres. Blood lactate rise during arm-cycling was blunted in patients and correlated with muscle weakness and phenotypic muscle makeup. Post-exercise metabolic recovery was slower in residual intramuscular white myofibres in patients demonstrating mitochondrial ATP synthetic dysfunction in this particular fibre type.

This study provides the first in vivo evidence in patients that degeneration of motor neurons and associated musculature causing atrophy and muscle weakness in 5q spinal muscular atrophy type 3 and 4 is aggravated by disproportionate depletion of myofibres that contract fastest and strongest. Our finding of decreased mitochondrial ATP synthetic function selectively in residual white myofibres provides both a possible clue to understanding the apparent vulnerability of this particular fibre type in 5q spinal muscular atrophy types 3 and 4 as well as a new biomarker and target for therapy.

Magnetic Resonance-Compatible Arm-Crank Ergometry: A New Platform Linking Whole-Body Calorimetry to Upper-Extremity Biomechanics and Arm Muscle Metabolism

INTRODUCTION

Evaluation of the effect of human upper-body training regimens may benefit from knowledge of local energy expenditure in arm muscles. To that end, we developed a novel arm-crank ergometry platform for use in a clinical magnetic resonance (MR) scanner with 31P spectroscopy capability to study arm muscle energetics. Complementary datasets on heart-rate, whole-body oxygen consumption, proximal arm-muscle electrical activity and power output, were obtained in a mock-up scanner. The utility of the platform was tested by a preliminary study over 4 weeks of skill practice on the efficiency of execution of a dynamic arm-cranking task in healthy subjects.

RESULTS

The new platform successfully recorded the first ever in vivo 31P MR spectra from the human biceps brachii (BB) muscle during dynamic exercise in five healthy subjects. Changes in BB energy- and pH balance varied considerably between individuals. Surface electromyography and mechanical force recordings revealed that individuals employed different arm muscle recruitment strategies, using either predominantly elbow flexor muscles (pull strategy; two subjects), elbow extensor muscles (push strategy; one subject) or a combination of both (two subjects). The magnitude of observed changes in BB energy- and pH balance during ACT execution correlated closely with each strategy. Skill practice improved muscle coordination but did not alter individual strategies. Mechanical efficiency on group level seemed to increase as a result of practice, but the outcomes generated by the new platform showed the additional caution necessary for the interpretation that total energy cost was actually reduced at the same workload.

CONCLUSION

The presented platform integrates dynamic in vivo 31P MRS recordings from proximal arm muscles with whole-body calorimetry, surface electromyography and biomechanical measurements. This new methodology enables evaluation of cyclic motor performance and outcomes of upper-body training regimens in healthy novices. It may be equally useful for investigations of exercise physiology in lower-limb impaired athletes and wheelchair users as well as frail patients including patients with debilitating muscle disease and the elderly.

Quantification of skeletal muscle mitochondrial function by 31P magnetic resonance spectroscopy techniques: a quantitative review

Magnetic resonance spectroscopy (MRS) can give information about cellular metabolism in vivo which is difficult to obtain in other ways. In skeletal muscle, non-invasive (31) P MRS measurements of the post-exercise recovery kinetics of pH, [PCr], [Pi] and [ADP] contain valuable information about muscle mitochondrial function and cellular pH homeostasis in vivo, but quantitative interpretation depends on understanding the underlying physiology. Here, by giving examples of the analysis of (31) P MRS recovery data, by some simple computational simulation, and by extensively comparing data from published studies using both (31) P MRS and invasive direct measurements of muscle O2 consumption in a common analytical framework, we consider what can be learnt quantitatively about mitochondrial metabolism in skeletal muscle using MRS-based methodology. We explore some technical and conceptual limitations of current methods, and point out some aspects of the physiology which are still incompletely understood.

Multi-tasking of biosynthetic and energetic functions of glycolysis explained by supply and demand logic

After more than a century of research on glycolysis, we have detailed descriptions of its molecular organization, but despite this wealth of knowledge, linking the enzyme properties to metabolic pathway behavior remains challenging. These challenges arise from multi-layered regulation and the context and time dependence of component functions. However, when viewed as a system that functions according to the principles of supply and demand, a simplifying theoretical framework can be applied to study its regulation logic and to assess the coherence of experimental interpretations. These principles are universally applicable, as they emphasize the common metabolic tasks of glycolysis: the provision of free-energy carriers, and precursors for biosynthesis and stress-related compounds. Here we will review the regulation of multi-tasking by glycolysis and consider how an understanding of this central metabolic pathway can be pursued using general principles, rather than focusing on the biochemical details of constituent components.

Optimal experiment design for model selection in biochemical networks

Background

Mathematical modeling is often used to formalize hypotheses on how a biochemical network operates by discriminating between competing models. Bayesian model selection offers a way to determine the amount of evidence that data provides to support one model over the other while favoring simple models. In practice, the amount of experimental data is often insufficient to make a clear distinction between competing models. Often one would like to perform a new experiment which would discriminate between competing hypotheses.

Results

We developed a novel method to perform Optimal Experiment Design to predict which experiments would most effectively allow model selection. A Bayesian approach is applied to infer model parameter distributions. These distributions are sampled and used to simulate from multivariate predictive densities. The method is based on a k-Nearest Neighbor estimate of the Jensen Shannon divergence between the multivariate predictive densities of competing models.

Conclusions

We show that the method successfully uses predictive differences to enable model selection by applying it to several test cases. Because the design criterion is based on predictive distributions, which can be computed for a wide range of model quantities, the approach is very flexible. The method reveals specific combinations of experiments which improve discriminability even in cases where data is scarce. The proposed approach can be used in conjunction with existing Bayesian methodologies where (approximate) posteriors have been determined, making use of relations that exist within the inferred posteriors.

Lost in transition: start-up of glycolysis yields subpopulations of nongrowing cells

Cells need to adapt to dynamic environments. Yeast that fail to cope with dynamic changes in the abundance of glucose can undergo growth arrest. We show that this failure is caused by imbalanced reactions in glycolysis, the essential pathway in energy metabolism in most organisms. The imbalance arises largely from the fundamental design of glycolysis, making this state of glycolysis a generic risk. Cells with unbalanced glycolysis coexisted with vital cells. Spontaneous, nongenetic metabolic variability among individual cells determines which state is reached and, consequently, which cells survive. Transient ATP (adenosine 5'-triphosphate) hydrolysis through futile cycling reduces the probability of reaching the imbalanced state. Our results reveal dynamic behavior of glycolysis and indicate that cell fate can be determined by heterogeneity purely at the metabolic level.

Biochemical competition makes fatty-acid β-oxidation vulnerable to substrate overload

Fatty-acid metabolism plays a key role in acquired and inborn metabolic diseases. To obtain insight into the network dynamics of fatty-acid β-oxidation, we constructed a detailed computational model of the pathway and subjected it to a fat overload condition. The model contains reversible and saturable enzyme-kinetic equations and experimentally determined parameters for rat-liver enzymes. It was validated by adding palmitoyl CoA or palmitoyl carnitine to isolated rat-liver mitochondria: without refitting of measured parameters, the model correctly predicted the β-oxidation flux as well as the time profiles of most acyl-carnitine concentrations. Subsequently, we simulated the condition of obesity by increasing the palmitoyl-CoA concentration. At a high concentration of palmitoyl CoA the β-oxidation became overloaded: the flux dropped and metabolites accumulated. This behavior originated from the competition between acyl CoAs of different chain lengths for a set of acyl-CoA dehydrogenases with overlapping substrate specificity. This effectively induced competitive feedforward inhibition and thereby led to accumulation of CoA-ester intermediates and depletion of free CoA (CoASH). The mitochondrial [NAD⁺]/[NADH] ratio modulated the sensitivity to substrate overload, revealing a tight interplay between regulation of β-oxidation and mitochondrial respiration.

Lipid metabolism plays an important role in the development of metabolic syndrome, a major risk factor for cardiovascular disease and diabetes. Furthermore, inborn errors in lipid oxidation cause rare, but severe diseases in children. To obtain more insight into the response of lipid oxidation to dietary and medical interventions, we constructed a computational model. The model correctly simulated the rate of lipid oxidation and the time courses of most acyl carnitines. The latter are used as diagnostic markers in blood. Subsequently, we subjected the model to an increased supply of lipids, as often happens in obese people. We discovered that the lipid-oxidation machinery easily becomes overloaded, very much like a highway during rush hours: the more cars enter the road, the slower they proceed and the more they clog the road. Analogously, an overload of lipids slowed down the lipid oxidation and led to an accumulation of intermediate metabolites in the pathway. Potential protection mechanisms of cells consist of restricted entry of lipids into the oxidation pathway or efficient downstream processing of reaction products. In future research we will use the model to test dietary or medical interventions in silico and thereby guide the development of new treatment and prevention strategies.

Parameter trajectory analysis to identify treatment effects of pharmacological interventions

The field of medical systems biology aims to advance understanding of molecular mechanisms that drive disease progression and to translate this knowledge into therapies to effectively treat diseases. A challenging task is the investigation of long-term effects of a (pharmacological) treatment, to establish its applicability and to identify potential side effects. We present a new modeling approach, called Analysis of Dynamic Adaptations in Parameter Trajectories (ADAPT), to analyze the long-term effects of a pharmacological intervention. A concept of time-dependent evolution of model parameters is introduced to study the dynamics of molecular adaptations. The progression of these adaptations is predicted by identifying necessary dynamic changes in the model parameters to describe the transition between experimental data obtained during different stages of the treatment. The trajectories provide insight in the affected underlying biological systems and identify the molecular events that should be studied in more detail to unravel the mechanistic basis of treatment outcome. Modulating effects caused by interactions with the proteome and transcriptome levels, which are often less well understood, can be captured by the time-dependent descriptions of the parameters. ADAPT was employed to identify metabolic adaptations induced upon pharmacological activation of the liver X receptor (LXR), a potential drug target to treat or prevent atherosclerosis. The trajectories were investigated to study the cascade of adaptations. This provided a counter-intuitive insight concerning the function of scavenger receptor class B1 (SR-B1), a receptor that facilitates the hepatic uptake of cholesterol. Although activation of LXR promotes cholesterol efflux and -excretion, our computational analysis showed that the hepatic capacity to clear cholesterol was reduced upon prolonged treatment. This prediction was confirmed experimentally by immunoblotting measurements of SR-B1 in hepatic membranes. Next to the identification of potential unwanted side effects, we demonstrate how ADAPT can be used to design new target interventions to prevent these.

A driving ambition of medical systems biology is to advance our understanding of molecular processes that drive the progression of complex diseases such as Type 2 Diabetes and cardiovascular disease. This insight is essential to enable the development of therapies to effectively treat diseases. A challenging task is to investigate the long-term effects of a treatment, in order to establish its applicability and to identify potential side effects. As such, there is a growing need for novel approaches to support this research. Here, we present a new computational approach to identify treatment effects. We make use of a computational model of the biological system. The model is used to describe the experimental data obtained during different stages of the treatment. To incorporate the long-term/progressive adaptations in the system, induced by changes in gene and protein expression, the model is iteratively updated. The approach was employed to identify metabolic adaptations induced by a potential anti-atherosclerotic and anti-diabetic drug target. Our approach identifies the molecular events that should be studied in more detail to establish the mechanistic basis of treatment outcome. New biological insight was obtained concerning the metabolism of cholesterol, which was in turn experimentally validated.

Parameter uncertainty in biochemical models described by ordinary differential equations

Improved mechanistic understanding of biochemical networks is one of the driving ambitions of Systems Biology. Computational modeling allows the integration of various sources of experimental data in order to put this conceptual understanding to the test in a quantitative manner. The aim of computational modeling is to obtain both predictive as well as explanatory models for complex phenomena, hereby providing useful approximations of reality with varying levels of detail. As the complexity required to describe different system increases, so does the need for determining how well such predictions can be made. Despite efforts to make tools for uncertainty analysis available to the field, these methods have not yet found widespread use in the field of Systems Biology. Additionally, the suitability of the different methods strongly depends on the problem and system under investigation. This review provides an introduction to some of the techniques available as well as gives an overview of the state-of-the-art methods for parameter uncertainty analysis.

Combined in vivo and in silico investigations of activation of glycolysis in contracting skeletal muscle

The hypothesis was tested that the variation of in vivo glycolytic flux with contraction frequency in skeletal muscle can be qualitatively and quantitatively explained by calcium-calmodulin activation of phosphofructokinase (PFK-1). Ischemic rat tibialis anterior muscle was electrically stimulated at frequencies between 0 and 80 Hz to covary the ATP turnover rate and calcium concentration in the tissue. Estimates of in vivo glycolytic rates and cellular free energetic states were derived from dynamic changes in intramuscular pH and phosphocreatine content, respectively, determined by phosphorus magnetic resonance spectroscopy ((31)P-MRS). Computational modeling was applied to relate these empirical observations to understanding of the biochemistry of muscle glycolysis. Hereto, the kinetic model of PFK activity in a previously reported mathematical model of the glycolytic pathway (Vinnakota KC, Rusk J, Palmer L, Shankland E, Kushmerick MJ. J Physiol 588: 1961-1983, 2010) was adapted to contain a calcium-calmodulin binding sensitivity. The two main results were introduction of regulation of PFK-1 activity by binding of a calcium-calmodulin complex in combination with activation by increased concentrations of AMP and ADP was essential to qualitatively and quantitatively explain the experimental observations. Secondly, the model predicted that shutdown of glycolytic ATP production flux in muscle postexercise may lag behind deactivation of PFK-1 (timescales: 5-10 s vs. 100-200 ms, respectively) as a result of accumulation of glycolytic intermediates downstream of PFK during contractions.

Computational modeling of mitochondrial energy transduction

Mitochondria are the power plant of the heart, burning fat and sugars to supply the muscle with the adenosine triphosphate (ATP) free energy that drives contraction and relaxation during each heart beat. This function was first captured in a mathematical model in 1967. Today, interest in such a model has been rekindled by ongoing in silico integrative physiology efforts such as the Cardiac Physiome project. Here, the status of the field of computational modeling of mitochondrial ATP synthetic function is reviewed.

Unchanged muscle fiber conduction velocity relates to mild acidosis during exhaustive bicycling

Muscle fiber conduction velocity (MFCV) has often been shown to decrease during standardized fatiguing isometric contractions. However, several studies have indicated that the MFCV may remain constant during fatiguing dynamic exercise. It was investigated if these observations can be related to the absence of a large decrease in pH and if MFCV can be considered as a good indicator of acidosis, also during dynamic bicycle exercise. High-density surface electromyography (HDsEMG) was combined with read-outs of muscle energetics recorded by in vivo ³¹P magnetic resonance spectroscopy (MRS). Measurements were performed during serial exhausting bouts of bicycle exercise at three different workloads. The HDsEMG recordings revealed a small and incoherent variation of MFCV during all high-intensity exercise bouts. ³¹P MRS spectra revealed a moderate decrease in pH at the end of exercise (~0.3 units down to 6.8) and a rapid ancillary drop to pH 6.5 during recovery 30 s post-exercise. This additional degree of acidification caused a significant decrease in MFCV during cycling immediately after the rest period. From the data a significant correlation between MFCV and [H⁺] ([H⁺] = 10^−pH) was calculated (p < 0.001, Pearson’s R = −0.87). Our results confirmed the previous observations of MFCV remaining constant during fatiguing dynamic exercise. A constant MFCV is in line with a low degree of acidification, considering the presence of a correlation between pH and MFCV after further increasing acidification.

Systems biology from micro-organisms to human metabolic diseases: the role of detailed kinetic models

Human metabolic diseases are typically network diseases. This holds not only for multifactorial diseases, such as metabolic syndrome or Type 2 diabetes, but even when a single gene defect is the primary cause, where the adaptive response of the entire network determines the severity of disease. The latter may differ between individuals carrying the same mutation. Understanding the adaptive responses of human metabolism naturally requires a systems biology approach. Modelling of metabolic pathways in micro-organisms and some mammalian tissues has yielded many insights, qualitative as well as quantitative, into their control and regulation. Yet, even for a well-known pathway such as glycolysis, precise predictions of metabolite dynamics from experimentally determined enzyme kinetics have been only moderately successful. In the present review, we compare kinetic models of glycolysis in three cell types (African trypanosomes, yeast and skeletal muscle), evaluate their predictive power and identify limitations in our understanding. Although each of these models has its own merits and shortcomings, they also share common features. For example, in each case independently measured enzyme kinetic parameters were used as input. Based on these 'lessons from glycolysis', we will discuss how to make best use of kinetic computer models to advance our understanding of human metabolic diseases.

The silicon trypanosome

African trypanosomes have emerged as promising unicellular model organisms for the next generation of systems biology. They offer unique advantages, due to their relative simplicity, the availability of all standard genomics techniques and a long history of quantitative research. Reproducible cultivation methods exist for morphologically and physiologically distinct life-cycle stages. The genome has been sequenced, and microarrays, RNA-interference and high-accuracy metabolomics are available. Furthermore, the availability of extensive kinetic data on all glycolytic enzymes has led to the early development of a complete, experiment-based dynamic model of an important biochemical pathway. Here we describe the achievements of trypanosome systems biology so far and outline the necessary steps towards the ambitious aim of creating a 'Silicon Trypanosome', a comprehensive, experiment-based, multi-scale mathematical model of trypanosome physiology. We expect that, in the long run, the quantitative modelling enabled by the Silicon Trypanosome will play a key role in selecting the most suitable targets for developing new anti-parasite drugs.