An in vivo data-driven framework for classification and quantification of enzyme kinetics and determination of apparent thermodynamic data Andre ´ B. Canelas n , Cor Ras, Angela ten Pierick, Walter M. van Gulik, Joseph J. Heijnen Department of Biotechnology, Kluyver Centre for Genomics of Industrial Fermentation, Delft University of Technology, Julianalaan 67, 2628 BC, The Netherlands article info Article history: Received 27 August 2010 Received in revised form 10 January 2011 Accepted 15 February 2011 Available online 24 February 2011 Keywords: In vivo enzyme kinetics Biochemical thermodynamics Genome-scale kinetic modeling Baker’s yeast Targeted quantitative metabolomics Aerobic glucose-limited conditions abstract Kinetic modeling of metabolism holds great potential for metabolic engineering but is hindered by the gap between model complexity and availability of in vivo data. There is also growing interest in network-wide thermodynamic analyses, which are currently limited by the scarcity and unreliability of thermodynamic reference data. Here we propose an in vivo data-driven approach to simultaneously address both problems. We then demonstrate the procedure in Saccharomyces cerevisiae, using chemostats to generate a large flux/metabolite dataset, under 32 conditions spanning a large range of fluxes. Reactions were classified as pseudo-, near- or far-from-equilibrium, allowing the complexity of mathematical description to be tailored to the kinetic behavior displayed in vivo. For 3/4 of the reactions we derived fully in vivo-parameterized kinetic descriptions which can be readily incorporated into models. For near-equilibrium reactions this involved a new simplified format, dubbed ‘‘Q-linear kinetics’’. We also demonstrate, for the first time, systematic estimation of apparent in vivo K eq values. Remarkably, comparison with E. coli data suggests they constitute a suitable in vivo interspecies thermodynamic reference. & 2011 Elsevier Inc. All rights reserved. 1. Introduction The quantitative study of the relations between fluxes and metabolites, as well as how they are affected by the influence of gene expression on enzyme levels, is pivotal in our efforts to understand the regulation of metabolic reaction networks and predict their dynamic behavior, in terms of the direction and value of in vivo fluxes (Dauner, 2010). The incorporation of this knowledge into mathematical models will also expand the array of tools at our disposal for targeted engineering of cellular phenotypes (Nielsen, 1998). The technological advances we are witnessing in metabolo- mics (Wu et al., 2005; Bajad et al., 2006; Cipollina et al., 2009; Chalcraft et al., 2009; Buescher et al., 2010), fluxomics (Sauer, 2006; Wiechert et al., 2007; Niklas et al., 2010) and proteomics (Gerber et al., 2003; Beynon et al., 2005; Picotti et al., 2009) continue to extend our ability to monitor quantitatively the relevant variables and, thus, to characterize in detail the states of metabolic systems. Yet, great challenges remain largely unresolved in the integration and interpretation of these datasets. Here, we propose a data-driven approach to simultaneously address two key problems: the scarcity and unreliability of thermodynamic reference data (Kummel et al., 2006; Fleming et al., 2009) (needed to predict flux directions) and the difficulty in adjusting the high degree of complexity of enzyme rate equations (used to predict flux values) to the limited availability of in vivo data (Nikerel et al., 2009b; Gunawardena, 2010). We then demonstrate the approach by applying it to a large number of reactions in the central metabolism of S. cerevisiae, under aerobic glucose-limited conditions. The flux directions in biochemical reactions are determined by the metabolite activities, which for simplicity we shall assume equal to their concentrations. According to the second law of Contents lists available at ScienceDirect journal homepage: www.elsevier.com/locate/ymben Metabolic Engineering 1096-7176/$ - see front matter & 2011 Elsevier Inc. All rights reserved. doi:10.1016/j.ymben.2011.02.005 Abbreviations: Metabolites: 2PG, 2-phosphoglycerate; 3PG, 3-phosphoglycerate; 6PG, 6-phospho gluconate; AcAld, acetaldehyde; CIT, citrate; DHAP, dihydroxya- cetone phosphate; E4P, erythrose-4-phosphate; F6P, fructose-6-phosphate; FBP, fructose-1,6-bis-phosphate; FUM, fumarate; G1P, glucose-1-phosphate; G3P, gly- cerol-3-phosphate; G6P, glucose-6-phosphate; GAP, glyceraldehyde-3-phosphate; Isocit, isocitrate; M1P, mannose-1-phosphate; M6P, mannose-6-phosphate; Mtl1P, mannitol-1-phosphate; MAL, malate; OGL, oxoglutarate; PEP, phosphoe- nolpyruvate; Pi, phosphate; PYR, pyruvate; R5P, ribose-5-phosphate; Rbu5P, ribulose-5-phosphate; S7P, sedoheptulose-7-phosphate; SUC, succinate; T6P, trehalose-6-phosphate; UDP-G, UDP-glucose; X5P, xylulose-5-phosphate; Enzymes and/or the reactions they catalyze: ACO, aconitate hydratase; ADH, alcohol dehydrogenase; ADK, adenylate kinase; APT, alanine transaminase; ENO, phosphopyruvate hydratase; FBA, fructose-bisphosphate aldolase; FMH, fumarate hydratase; G3PDH, glycerol-3-phosphate dehydrogenase; GLT, glycerol transport; GPP, glycerol-1-phosphatase; GAPDH, glyceraldehyde-3-phosphate dehydrogen- ase; GPM, phosphoglycerate mutase; HXK, hexokinase; HX, hexose transporte; PDC, pyruvate decarboxylase; PFK, 6-phosphofructokinase; PGI, glucose-6-phos- phate isomerase; PGK, phosphoglycerate kinase; PGM, phosphoglucomutase; PMI, mannose-6-phosphate isomerase; PMM, phosphomannomutase; PYK, pyruvate kinase; RPE, ribulose-phosphate 3-epimerase; RPI, ribose-5-phosphate isomerase; TAL, transaldolase; TK1, transketolase (S7P-producing reaction); TK2, transketo- lase (E4P-consuming reaction); TPI, triose-phosphate isomerase; TPP, trehalose- phosphatase; TPS, alpha,alpha-trehalose-phosphate synthase. n Corresponding author. Fax: + 31 15 278 2355. E-mail addresses: a.canelas@tudelft.nl, canelasab@gmail.com (A.B. Canelas). Metabolic Engineering 13 (2011) 294–306