A step toward systems metabolomics

Renée Ventura-Clapier, Hilary A. Coller

while levels of genes and proteins provide information on a cell's possibilities, the levels of metabolic intermediates provide a very direct indication of a cell's current activities. Methods to monitor the levels of metabolites as well as the dynamic flux through metabolic pathways can be integrated to generate a systems-level understanding of the metabolic activity in a cell at a given time. Nemutlu et al. (5a) describe a new methodology for metabolomic analysis that provides dynamic and systems-level information on phosphate transfer rates that can be used to understand the integration of multiple cellular signaling pathways in normal, genetically altered and diseased tissue.

Energy Generation in Cardiomyocytes

The heart is the organ that consumes the most energy in the entire body. In the heart, energy is stored in the form of ATP and phosphocreatine (PCr), which is formed when creatine kinases add a phosphate derived from ATP to creatine. While ATP and phosphocreatine allow for limited storage of energy, >90% of the energy required for the heart muscle to perform its energy-intensive function of pumping blood is derived from mitochondrial respiration. To maintain constant, global cellular ATP and PCr, there is a close correlation between cardiac oxygen consumption and cardiac work, showing that the bioenergetics of the heart functions in tight flux mode (10). Thus, the heart is somewhat unusual in that it maintains metabolic homeostasis, expressed as constancy in concentrations of ATP, PCr, and creatine, despite large variations in the energetic demands placed on it (6, 9).

Cardiac metabolic homeostasis requires the existence of a mechanism whereby energy signaling pathways can ensure a close match between oxygen consumption and energy utilization. The cardiomyocyte is not a well-mixed bag (13), and the reactions involved in ATP generation (glycolysis and oxidative phosphorylation) and utilization [ATPases of the plasma membrane, sarcoplasmic reticulum (SR), and myofilaments] are localized within structural and functional entities and are spatially and temporarily coordinated. The cardiomyocyte has a complex cytoarchitecture formed by rows of mitochondria and myofilaments surrounded by an SR network in which the cytosolic compartment represents not more than a few percent of the cell volume. This complex cytoarchitecture creates energetic microdomains close to the sites of energy production and consumption, where the concentration of ATP and ADP are limited by diffusion and can be different from the rest of the cell. In particular, energy must be quickly and efficiently transferred from the mitochondria to the ATPases according to the immediate demand to ensure an optimal energetic microenvironment in the vicinity of these energy consumers.

There are multiple different energy transfer systems in cardiomyocytes, the most important being creatine kinase (CK). These systems rely on the presence of mitochondrial isoforms bound to mitochondrial membranes and cytosolic isoforms bound to myofilaments, SR, and plasma membrane that reside close to the ATPases (for reviews see Refs. 3, 8, 11, 12). For example, mitochondrial CK transfers the phosphate group from ATP generated in the mitochondria to Cr, thereby producing PCr, which diffuses to the energy-consuming sites through near-equilibrium reactions of the cytosolic CK. CK bound to the myofilaments, SR, and the plasma membrane rephosphorylates locally produced ADP by using PCr. Other energy transfer systems are adenylate kinase (AK), nucleotide diphosphokinase, glycolytic complexes, and direct energy channeling between mitochondria and ATP-consuming organelles (5). Muscle cells use these phospho-transfer networks, consisting of multiple near-equilibrium enzyme reactions, to convey energy-rich phosphoryl groups between cellular compartments in a kinetically and thermodynamically efficient manner (2, 7, 12). These complex interconnected networks of energy transfer may have partially redundant or additive effects depending on the energy state, and CK deficiency is a hallmark of cardiovascular diseases. Mechanisms that compensate for reduced energy transfer functions have been described using genetic models of AK or CK deficiency showing the plasticity in cardiac energy supply and substrate utilization networks (1, 5).

Local compartmentation of adenine nucleotides at the interface between ATPases on one side and mitochondria, glycolytic complexes, or phosphotransfer enzymes on the other side ensures a high ATP/ADP ratio. This subcellular organization of energetic components has been elucidated through work on isolated organelles or permeabilized preparations using biochemical (competitive enzyme systems) and functional (contractile force, calcium movements, ion fluxes …) assays.

Technical Approaches for Monitoring Metabolism and the Advances Described in Nemutlu et al.

Methologies for understanding energy generation in cardiomyocytes have included 31P NMR spectroscopy, which allows researchers to noninvasively follow the ATP and PCr concentrations in vivo or in isolated organs and, by using magnetization transfer of the phosphate moiety as a tracer of energetic fluxes, to monitor the enzymology of CK. However, global approaches to defining cellular fluxes with NMR do not allow the discrimination of fluxes among the different CK isoforms, nor do they take into account the compartmentation of energy fluxes. A combination of 31P NMR together with mathematical modeling could provide evidence for the unidirectional fluxes of localized CKs in a living organ (4). Nevertheless, even with the sophistication of saturation transfer, 31P NMR has the limitation that long-standing acquisitions are necessary to estimate kinetic constants of seconds to minutes.

The dynamics of high-energy phosphoryl exchanges using advanced phosphoryl 18O-labeling analysis was introduced in the 1980s (14). The 18O-labeling procedure is based on incorporation of one 18O atom, provided from [18O]H2O into inorganic phosphate for each step of ATP hydrolysis, and the subsequent distribution of 18O-labeled phosphoryls among other high-energy phosphoryl-carrying molecules.

In the article in focus, the group of Petras Dzeja has combined the previous Goldberg study of energy fluxes using [18O]H2O, substrate uptake, and metabolism with 13C and 31P NMR techniques and high-performance liquid chromatography (HPLC) followed by gas chromatography-mass spectrometry (GC/MS) for metabolomic analysis. One of the challenges in the development of 18O-labeling technology has been the detection of the metabolic products after 18O is incorporated into phosphates. By combining different analytical methods for detecting the downstream metabolites, the Dzeja group has developed a methodology that allows them to measure metabolite levels and metabolic fluxes through phosphotransfer systems and thereby characterize the activity of different energetic pathways. In particular, the methodology that they developed allows for simultaneous recordings of ATP synthesis and utilization, phosphotransfer fluxes through AK, CK, and glycolytic pathways, as well as mitochondrial Krebs cycle activity and glycogen turnover. The rate of each reaction is determined based on the extent of 18O incorporation, evaluated in individual metabolites after HPLC followed by GC/MS. For instance, G-6-P[18O] labeling indicates glycolysis, G-3-P[18O] labeling provides information on the rate of substrate shuttling, and G-1-P[18O] is a marker for glycogenolysis. AK and CK flux rates can be determined from the rate of appearance of 18O-containing β-phosphoryls in ADP and ATP or PCr. One drawback of the approach is that only phosphates are labeled, and while phosphates are central to metabolic regulation, some types of information are less direct than with other methods. For instance, the flow of carbons through central carbon metabolism can be more directly ascertained with 13C labeling.

The ability to monitor all of these phosphate-related activities simultaneously is nevertheless very powerful, especially given that the methodology can be applied directly to tissue, and the response of the entire network to perturbation of a single node can be determined. The hearts of AK knockout mice, for instance, differed from wild-type controls in their response to high calcium stress not only for enzymatic activities directly downstream of AK, but also with regard to glycolytic and glycogenolytic pathways, Krebs and urea cycles, and nucleotide metabolism. A similar result was observed for hearts deficient in CK. Hearts from these mice exhibited higher G-6-[18O]P turnover and higher γ/β-phosphoryl GTP[18O] turnover. β-[18O]ATP and β-[18O]ADP labeling were also altered, reflecting changes in AK metabolic flux. These changes likely indicate rearrangements in phosphotransfer to compensate for genetic CK deficiency. The results are consistent with previous reports that CK-deficient cardiomyocytes have greater reliance on glycolytic metabolism and that when stressed, they cannot sustain their ATP/ADP ratios, indicating a deficiency in the communication between ATP-consuming and ATP-generating activities. Thus, the methodology developed by the authors allows for monitoring systems-wide metabolic adjustments to the absence of a specific node, in this case, CK.

Looking Ahead

Understanding the rate of all metabolic pathways in the cell in different conditions would provide enormous insight into the structure of cellular pathways. However, because of the diversity in the physicochemical properties of the relevant cellular metabolites, it is necessary to utilize a wide range of analytical techniques, each capable of monitoring a small subset of all cellular metabolites, to generate a comprehensive view. The methodology presented by Nemutlu et al. (5a) represents a significant step forward in that it allowed the authors to detect the substrates for multiple important biochemical reactions and to assess their rates. While the applications of the methodology in this publication involve genetic alteration of proteins central to energy metabolism and thus are essentially proof of principle demonstrations, the availability of a methodology for monitoring metabolic flux more broadly opens up a wide array of possibilities for applying it to cellular processes that likely involve shifts in phosphate metabolism.

One important application of the methodology is in defining the metabolic networks present in normal cells. Maintaining optimal cardiac function, for instance, requires coordination between ATP consuming processes and ATP generating processes, which are interconnected by phosphotransfer networks created by AK, CK, and glycolytic/glycogenolytic network nodes. This network contains redundancies that add to its robustness and allow the cell to develop compensatory mechanisms to adapt to genetic or pathological defects but may leave it vulnerable to specific types of stress. Intracellular phosphate transfer is likely disrupted in clinically relevant situations such as ischemia-reperfusion, heart failure, and heart genetic deficiencies. The methodology presented has proven valuable in elucidating this network in cardiomyocytes and would likely be valuable in defining other networks, for instance, the systems-level metabolic response to insulin and other hormones.

The methodology described would also be very valuable for defining the important contributors to disease states. For instance, mutations that alter the glycerophosphate substrate shuttle are linked to Brugada syndrome, a hereditary syndrome in which apparently healthy individuals experience unexpected cardiac death. Metabolic changes may prove to be valuable biomarkers for specific disease states and may therefore aid in disease classification and diagnosis. The methodology will also be valuable for classifying and quantifying the changes in metabolic profile that result from drugs or toxins. As one example, creatine or creatine analogs are used to treat patients with hypertension, congestive heart failure, and myocardial infarction. The changes in creatine phosphate metabolism that occur with these different types of heart disease and the effects of creatine-based treatments would be another promising application. The methodology also makes it possible to characterize the effects of differences in lifestyle including nutritional choices and exposure to environmental agents on metabolism. The genetic underpinnings of metabolic rates could also be defined. In sum, the proposed methodology offers an opportunity to advance our understanding of metabolic networks in a wide variety of physiological and pathophysiological states.


H. A. Coller is supported by National Institute of General Medical Sciences (NIGMS) Center of Excellence Grant P50 GM-071508, a Focused Funding Grant from the Johnson & Johnson Foundation, and NIGMS Grants 1R01 GM-081686 and 1R01 GM-086465. R. Ventura-Clapier is a senior scientist of the Centre National de la Recherche Scientifique.


No conflicts of interest, financial or otherwise, are declared by the author(s).


Author contributions: R.V.-C. and H.A.C. drafted manuscript; R.V.-C. and H.A.C. edited and revised manuscript; R.V.-C. and H.A.C. approved final version of manuscript.


The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 5a.
  7. 6.
  8. 7.
  9. 8.
  10. 9.
  11. 10.
  12. 11.
  13. 12.
  14. 13.
  15. 14.