13 C-based metabolic flux analysis Nicola Zamboni 1 , Sarah-Maria Fendt 1,2 , Martin Ru ¨hl 1,3 & Uwe Sauer 1 1 Institute of Molecular Systems Biology, ETH Zurich, Zurich 8093, Switzerland. 2 PhD Program Systems Biology of Complex Diseases, ETH Zurich, Zurich, Switzerland. 3 PhD Program Molecular Life Sciences, ETH Zurich, Zurich, Switzerland. Correspondence should be addressed to U.S. (sauer@ethz.ch). Published online 21 May 2009; doi:10.1038/nprot.2009.58 Stable isotope, and in particular 13 C-based flux analysis, is the exclusive approach to experimentally quantify the integrated responses of metabolic networks. Here we describe a protocol that is based on growing microbes on 13 C-labeled glucose and subsequent gas chromatography mass spectrometric detection of 13 C-patterns in protein-bound amino acids. Relying on publicly available software packages, we then describe two complementary mathematical approaches to estimate either local ratios of converging fluxes or absolute fluxes through different pathways. As amino acids in cell protein are abundant and stable, this protocol requires a minimum of equipment and analytical expertise. Most other flux methods are variants of the principles presented here. A true alternative is the analytically more demanding dynamic flux analysis that relies on 13 C-pattern in free intracellular metabolites. The presented protocols take 5–10 d, have been used extensively in the past decade and are exemplified here for the central metabolism of Escherichia coli. INTRODUCTION Cellular metabolism consists of hundreds to thousands of genes, enzymes and metabolites that convert nutrients into biosynthetic building blocks and energy. Thereby, it fuels growth and essentially all other cellular activities. Despite extremely different lifestyles and physiology, the underlying biochemistry of metabolism is remark- ably similar in all kingdoms of life. From Escherichia coli to humans, intermediary metabolism, many anabolic reactions and respiratory energy generation are almost identical. Assembled from decades of in vitro enzyme data on individual reactions, mutant analyses and reaction-focused isotope tracing, the unraveled network structure of metabolism is currently the best understood of any biological network. Beyond the mere network structure, however, one needs to understand how the pieces work together to produce appropriate network responses that enable cellular functioning under ever- changing environmental conditions. This functional output response of metabolism is the traffic of metabolites (i.e., the intracellular metabolic fluxes), which emerges from the intertwined non-linear and dynamic interactions occurring between large numbers of network components 1–3 . In contrast to the concentra- tions of enzymes and metabolites that define the network structure of cellular metabolism, time-dependent fluxes cannot be measured directly but must be inferred from measurable quantities 2,4 . The above historical methods, including the use of isotopes to unravel reaction mechanisms, but also the more recent proteomics or transcriptomics methods cannot assess network fluxes. Earlier, the only possibility was to balance physiological fluxes in and out of cells within assumed reaction networks 5 . Since few such fluxes enter the network but many degrees of freedom exist within the network, only very few intracellular fluxes could be resolved, severely limiting the capacity of this method to obtain new insights. The pivotal advances were network-wide stable isotope balancing methods (mostly using 13 C) and appropriate computational methods that went well beyond the traditional isotope labeling studies 2,6,7 . The protocol described here for metabolic flux analysis is based on mass spectrometry (MS)—tracing patterns of stable isotopes in protein-bound amino acids from growing micro-organisms for extended periods on 13 C-labeled substrates 6,8–10 . This 13 C-based flux analysis takes advantage of the fact that alternative pathways scramble and cleave the substrates’ carbon backbone differently before they converge on the same intermediate. When a partly labeled 13 C-substrate is fed to cells, alternative pathways produce characteristic 13 C pattern in the common products of a pathway, e.g., phosphoenolpyruvate molecules derived through glycolysis exhibit a distinct pattern from those derived through the pentose phosphate (PP) pathway. To quantify fluxes from 13 C data, we use two distinct methods. The first method calculates ratios of fluxes in converging pathways directly from these mass isotope patterns using the program FiatFlux 11,12 . The second method requires additional physiological fluxes that are determined from time courses of extracellular-metabolite concentrations during cultiva- tion 6,13 . Network-wide intracellular fluxes are then estimated indirectly from these physiological data and the above mass isotope pattern in a computational procedure that uses stoichiometric models of atom transitions between intermediates of metabolism using the program 13CFLUX 10 . The two methods are complemen- tary and may serve to cross-validate each other 14,15 . Alternative protocols for metabolic flux analysis are variations on the theme with two key differences. The first difference is in the detection of isotope patterns. Besides the MS method described here, nucleic magnetic resonance (NMR) spectroscopy is some- times used and provides somewhat complementary informa- tion 6,7,16 . The pros and cons of NMR versus MS are discussed later (see Factors affecting flux calculability). The second difference is in the analytes that are used to detect the 13 C patterns. MS or NMR analysis of free metabolic intermediates provides the richest source of information, but high turnover and very low concentra- tions of intermediates poses serious technical challenges to sample preparation, separation and analytical sensitivity. Such methods based on free intracellular intermediates are currently under development for dynamic flux analysis, but so far remain expert methods 17–20 . For flux estimation under steady state conditions, the protocol presented here based on protein-bound amino acids is technically much simpler, robust and widely used 1,2,6,21–23 . Since the carbon backbone of amino acids is synthesized from various p u o r G g n i h s i l b u P e r u t a N 9 0 0 2 © natureprotocols / m o c . e r u t a n . w w w / / : p t t h 878 | VOL.4 NO.6 | 2009 | NATURE PROTOCOLS PROTOCOL