Introduction
Quantification of metabolite flows through metabolic networks by 13C-based flux analysis is a key technology for systems-oriented research in microbes1. Thus far, however, the intrinsic complexity of mammalian cells has hindered application of this method to human disease. In this issue, Munger et al.2 succeed in quantifying flux in cultured human fibroblasts by combining multiple 13C-tracer experiments with rigorous mathematical analysis of labeling kinetics3, 4, 5, 6. Notably, they show that cytomegalovirus infection increases flux through the fatty acid biosynthesis pathway and that inhibitors of this pathway suppress viral replication. This study represents a conceptual shift from the popular use of metabolomics to hunt for biomarkers in body fluids to a more elegant exploitation of cellular models to probe disease-associated metabolic changes at the level of integrated network responses.
The rapid pace of analytical developments in the canonical 'omics technologies—transcriptomics, proteomics and metabolomics—continues to drive advances in pharmacology, functional genomics and systems biology. Nonetheless, it should not be forgotten that 'omics techniques assess only the concentrations of components in biological systems and not their functions. Methods to quantify the integrated, functional output from the interactions between the components are strikingly underrepresented in today's arsenal of technologies. One example of such a method is quantitative imaging, which enables observation of developmental processes in multicellular organisms7 and of spatial and temporal movement of and within cells8, 9.
For metabolism, the integrated, functional outputs are the time-dependent metabolite fluxes through networks of enzymatic reactions1, 10. The relationship between the concentrations of transcripts, proteins and metabolites provided by 'omics techniques and the functional flux output might be compared to that between parking lots (mRNAs), roads (proteins), cars (metabolites) and traffic (flux). The molecular traffic is the integrated response that emerges from the nonlinear interactions between genes, proteins and metabolites across multiple catalytic and regulatory layers. It cannot be obtained by looking at any of the components in isolation (Fig. 1).
Figure 1: Interactions between molecular components give rise to metabolic network responses.
Complexity arises through multiple levels of regulatory feedback (black arrows) between the components that mediate the primary flow of genetic information (red arrows). Viral infection alters metabolic flux through certain parts of the network.
Full size image (77 KB)One way of determining the traffic pattern is to predict it with mechanistic models of component interactions using previously determined parameters of, for example, enzyme kinetics. Alternatively, as in the paper by Munger et al.2, the pattern can be inferred from the measured 13C distributions to which the nonmeasurable fluxes are fitted with mass- and isotope-balancing models, typically only for a given steady state1. The advance here and in some earlier reports3, 4, 5 is the explicit consideration of labeling dynamics and their mathematical interpretation using ordinary differential equation models. Munger et al.2 increase the accuracy of flux estimates not only by using state-of-the-art mass spectrometry to detect the 13C labeling kinetics of intermediates in central carbon metabolism, but also by merging the results of five independent experiments involving different isotopic tracers.
The authors demonstrate the relevance of quantifying the integrated network response by identifying and rationalizing antiviral drug targets after comparing flux data obtained from healthy and infected cells. In previous work they determined the steady-state metabolite concentrations in similarly infected human fibroblasts and observed severely altered metabolism associated with viral replication11. Their current flux data defines these alterations in terms of functional pathways: the most striking changes in virally infected cells are substantially increased fluxes through the energy-generating pathways of glycolysis and the respiratory tricarboxylic acid cycle as well as a strong efflux into fatty acid biosynthesis.
Compared with data on altered metabolite levels, understanding of upregulated fluxes provides more direct leads to antiviral targets. Munger et al.2 searched for virally upregulated pathways that might be inhibited without undue toxicity to host cells. Although it is unclear how an increased rate of fatty acid biosynthesis is related to viral replication, the value of the authors' practical approach was demonstrated by their finding that two pharmacological inhibitors of fatty acid biosynthesis diminish replication of both cytomegalovirus and influenza A virus without apparent toxicity or induction of apoptosis in uninfected cells.
Flux data are extremely useful for monitoring the integrated metabolic system response but they do not reveal its underlying mechanistic basis. Does the control program of enveloped viruses actively redirect cellular metabolism to meet their requirement for the lipoprotein coat as they bud from infected cells? Or is the observed metabolic dysregulation merely a consequence of infection? Evidence that enveloped viruses bind to or modify key enzymes or regulators of fatty acid biosynthesis might resolve this issue.
ADVERTISEMENT
Beyond identifying effective antiviral drug targets, flux quantification in mammalian cells has broader implications for understanding disease-associated perturbations of metabolic networks. A few obvious examples of profound economic and public-health import are diabetes, obesity, cancer, drug metabolism and the impact of nutrition on health. All of these phenomena involve biochemical alterations that cause substantial metabolic responses (Fig. 1).
The current molecular target–based discovery paradigm in biology and pharmacology struggles to cope with complex metabolic responses for two primary reasons12. First, the integrated network response to interventions remains largely unpredictable. Second, metabolic networks are robust, allowing them to maintain stable function in the face of perturbations. Quantification of the in vivo flux response in cell types and tissues relevant to disease should allow unprecedented insights into the emergent and causal properties of metabolic networks. This knowledge may enable the development of kinetic biomarkers with improved predictive power compared with current, static biomarkers12. Although methods like that of Munger et al.2 provide a computational framework to deal with 13C-tracer data from cultured mammalian cells, more sophisticated analytics will be needed to harness the full power of flux analysis for higher cells and organisms. Challenges for future research include understanding single-cell dynamics, subcellular compartmentation and system-wide flux resolution, and addressing the need to average measurements over time spans of hours to days.
