Available online at www.sciencedirect.com ScienceDirect 13 C metabolic flux analysis of recombinant expression hosts 1,2 Jamey D Young Identifying host cell metabolic phenotypes that promote high since metabolic enzymes are largely regulated by allo- recombinant protein titer is a major goal of the biotech industry. steric feedback and substrate availability, and mRNA or 13 C metabolic flux analysis (MFA) provides a rigorous approach protein abundances often do not correlate strongly with 13 to quantify these metabolic phenotypes by applying isotope pathway fluxes [3,4]. On the other hand, C MFA relies tracers to map the flow of carbon through intracellular on least-squares regression of direct metabolite measure- metabolic pathways. Recent advances in tracer theory and ments (isotope labeling patterns and extracellular measurements are enabling more information to be extracted exchange rates) to determine the adjustable flux 13 from C labeling experiments. Sustained development of parameters in a mathematical model of cellular metab- publicly available software tools and standardization of olism (Figure 1). The resulting output is a comprehensive experimental workflows is simultaneously encouraging flux map that depicts the flow of carbon throughout 13 increased adoption of C MFA within the biotech research intracellular metabolism under the experimental con- 13 community. A number of recent C MFA studies have ditions of interest. Recent theoretical and experimental 13 identified increased citric acid cycle and pentose phosphate advances are pushing the boundaries of C MFA to new pathway fluxes as consistent markers of high recombinant heights of precision and flexibility, while increased avail- protein expression, both in mammalian and microbial hosts. ability of public software tools is streamlining MFA Further work is needed to determine whether redirecting flux workflows and providing improved access to these tech- into these pathways can effectively enhance protein titers while nologies within the biotech research community. In this maintaining acceptable glycan profiles. article, we provide an overview of these advances and Addresses highlight some important findings that have been 1 13 Department of Chemical and Biomolecular Engineering, Vanderbilt obtained through the application of C MFA to industrial University, PMB 351604, Nashville, TN 37235-1604, USA 2 expression hosts. We also point to emerging areas where Department of Molecular Physiology and Biophysics, Vanderbilt 13 C MFA is likely to provide new insights for improving University, PMB 351604, Nashville, TN 37235-1604, USA the quantity and quality of recombinant proteins pro- Corresponding author: Young, Jamey D ([email protected]) duced by these hosts. 13 Current Opinion in Biotechnology 2014, 30:238–245 Recent advances in C MFA tools and methodologies This review comes from a themed issue on Pharmaceutical biotechnology Optimal experiment design (OED) 13 The earliest C MFA experiments involved feeding Edited by Beth Junker and Jamey D Young 13 13 either [1- C] or [U- C6]glucose as a single tracer. Nowa- days, the increased precision obtainable from combining multiple tracers, whether fed simultaneously or in parallel http://dx.doi.org/10.1016/j.copbio.2014.10.004 experiments, is widely understood and accepted. How- 0958-1669/# 2014 Elsevier Ltd. All right reserved. ever, it is not always appreciated that the optimal tracer combination can depend strongly on the network top- ology and the available measurements, which vary from system to system. This implies that there is a need to tailor the labeling strategy to the system of interest rather than naı¨vely following prior conventions. Furthermore, 2 15 Introduction the use of H or N tracers to probe redox or nitrogen Industrial bioprocesses place high demands on the inter- metabolism, respectively, has proven to be a powerful 13 mediary metabolism of host cells to meet the biosynthetic complement to C MFA. For example, the Metallo and requirements for maximal growth and protein expression. Rabinowitz labs have recently applied an arsenal of 13 2 13 C metabolic flux analysis (MFA) has become the pre- different H and C tracers to assess the contributions mier approach to quantitatively assess the metabolic from multiple pathways to compartment-specific phenotypes of cultured cells and is now playing a growing NADPH production in the cytosol and mitochondria of role in host cell engineering and bioprocess optimization. mammalian cells [5 ,6 ]. Although transcriptomics [1] and proteomics [2] can be used to infer metabolic pathway alterations indirectly from The first systematic treatment of optimal design of iso- changes in enzyme expression, this can be misleading tope labeling experiments was introduced by Mo¨llney Current Opinion in Biotechnology 2014, 30:238–245 www.sciencedirect.com 13 C MFA of recombinant expression hosts Young 239 Figure 1 (1) Develop metabolic 100 network model (1) A 20 80 C G B 40 140 F D E (2) (2) Optimize tracers 70 30 and measurements H (3) Perform isotope (3) labeling experiment (4) (4) Measure extracellular exchange rates and 13C isotopomer abundances 13C MFA INST-MFA OR M0 M1 M2 M3 MID (5) Apply 13C MFA (5) Simulated to estimate fluxes Measured Time Current Opinion in Biotechnology 13 Typical workflow of C MFA studies. (Step 1) Development of a mathematical model comprising a system of mass balances and isotopomer balances on all metabolites in the biochemical reaction network is the initial step of MFA. (Step 2) This model can be applied to perform simulation studies and optimal experiment design (OED) computations to select tracer/measurement combinations that maximize flux resolution. (Step 3) A series of isotope labeling experiments is then performed by introducing the selected tracer(s) into culture and collecting cell and media samples at various time points during the growth phase of interest. (Step 4) MS and/or NMR is used to analyze the isotope labeling of target metabolites in these samples, while a parallel set of biochemical analyses is applied to profile the exchange rates of extracellular metabolites, as 13 well as cell growth and product formation rates. (Step 5) These data sets are regressed using either stationary C MFA or INST-MFA, depending on whether the labeling reaches equilibrium, which enables the adjustable flux parameters in the model to be uniquely determined. A series of statistical tests can be performed to assess the goodness-of-fit of the model and to calculate uncertainties for each of the estimated parameters. Recent theoretical advances and improvements in the availability of MFA software tools have enhanced the efficiency of this workflow, especially steps 1,2, and 5. At the same time, better analytical methods for obtaining isotopomer measurements have provided increased flexibility and precision in steps 3 and 4. et al. [7]. Their approach was based on the classical OED [9] to compute accurate nonlinear confidence intervals on 13 formulation of minimizing a scalar objective function all fluxes. They identified [1,2- C2]glucose and 13 computed from the flux covariance matrix of a reference [U- C5]glutamine as the most useful single tracers for flux map. Recent studies have applied more sophisticated flux determination in glycolysis and citric acid cycle 13 OED algorithms to tailor C MFA experiments to mam- (CAC) pathways, respectively. Walther et al. [10] later malian cell cultures, which pose a greater challenge due to extended this approach to examine mixtures of tracers their use of complex media containing multiple carbon identified using a genetic algorithm. sources. For example, Metallo et al. [8] examined flux identifiability of a carcinoma cell line using a variety of Another exciting development was the introduction of single tracers. Rather than using the parameter covariance the EMU basis vector (EMU-BV) approach, which can be matrix as a local estimate of flux uncertainty, they applied used to express the labeling of any metabolite in the the parameter continuation method of Antoniewicz et al. network as a linear combination of labeled substrate www.sciencedirect.com Current Opinion in Biotechnology 2014, 30:238–245 240 Pharmaceutical biotechnology atoms [11 ]. This allows the influence of fluxes on the expected to become an indispensible tool for extending 13 isotopomer measurements to be decoupled from sub- C MFA approaches to studies of mammalian systems strate labeling, and thereby enables a fully a priori and industrial bioprocesses. Finally, INST-MFA pro- approach to tracer selection that does not depend on vides increased measurement sensitivity to estimate the choice of a reference flux map. Crown et al. [12] reversible exchange fluxes and metabolite pool sizes subsequently applied the EMU-BV method to identify [32,33], which represents a potential framework for inte- 13 two novel tracers for quantifying oxidative pentose phos- grating metabolomic analysis with C MFA. phate pathway (OPPP) flux and anaplerotic pyruvate carboxylase flux using a metabolic model of HEK-293 New measurement strategies 13 cell metabolism. The EMU-BV approach also suggests In addition to theoretical advances, the field of C MFA that maximal flux information can be obtained by inte- is also rapidly expanding due to new isotopomer measure- grating data sets from multiple parallel tracer exper- ment capabilities. Traditionally, these measurements iments, each of which is designed to elucidate specific have been obtained from mass spectrometry (MS) [34] reactions in the