Probing the metabolism of an inducible mammalian expression system using extracellular isotopomer analysis

Abstract

In an effort to quantitatively assess the impact of recombinant protein expression on the primary metabolism of mammalian cells in culture, we have employed an efficient inducible expression system and conducted a comparative study of the intracellular flux map distribution with and without the induction of recombinant protein synthesis. Cells were grown in parallel semi-continuous cultures with various singly and uniformly labeled substrates and the resulting mass isotopomer distributions of lactate and extracellular amino acids were measured by mass spectrometry. These distributions were used to quantify the main intracellular fluxes. The analysis revealed that, under mild hypothermic conditions, the onset of protein expression is correlated with small but significant changes in several key pathways related to ATP and NADPH formation. More specifically, we observed that induced cells exhibit a more efficient utilization of glucose, characterized by an increased flux of pyruvate into the TCA cycle. In contrast, the catabolic rates of most amino acids remained relatively unaffected. Such analysis is instrumental to guide the identification of robust biomarkers of productivity, as well as the development of medium formulations optimized for recombinant protein production.

Introduction

Mammalian cells continue to be the preferred expression system for the production of valuable therapeutic proteins and vaccines, which require specific post-translational modifications that cannot be performed in other microorganisms. Monoclonal antibodies (MAb) produced by mammalian cell culture form the most important group of new biopharmaceuticals approved or under development and their market is expected to grow significantly in the next few years (Walsh, 2010). In order to meet the growing demand for MAb, there is a continuing need to develop robust, cost-effective and efficient cell culture processes. Conventional process development for mammalian cell culture MAb platforms has been conducted mainly by experimental and especially trial-and-error methods (Baughman et al., 2010). Despite significant advances in the field of cell culture technology, the establishment of an optimal set of process conditions remains mostly cell line-specific. Typically, improvements in product titers have been essentially achieved through an increase of the cumulative integrated viable cell density, but factors limiting the cell specific productivity of mammalian cells are still poorly understood (Meleady et al., 2011). Efforts to further enhance processes have thus rapidly underscored the need for better characterization and control of the complex mammalian cell metabolism in culture. In particular, compared to parental cells, recombinant cells typically exhibit reduced specific growth rates and increased nutrient utilization (Yallop et al., 2003). Assessing and analysing the impact of recombinant protein expression on the cells’ central metabolism is thus of utmost importance for the establishment and optimization of a productive mammalian cell expression platform.

To this end, quantification of metabolic fluxes can provide critical insights into the fundamental processes of biological systems and allow the identification of possible metabolic bottlenecks (Boghigian et al., 2010, Metallo et al., 2009, Park et al., 2008). In turn, such information has proven to be invaluable to enable the targeted optimization of cell culture processes. Many comprehensive metabolic models have now been developed for the most relevant industrial cell lines such as CHO, BHK, HEK-293 and hybridoma cells (Niklas and Heinzle, 2011). However, a complete and accurate determination of the flux distribution in mammalian cell networks is generally impossible to obtain from routine measurements and simple metabolite balancing, due to the typically large number of biochemical reactions involved on the one hand, and the existence of reversible, cyclic and parallel pathways on the other hand. To cope with this problem and alleviate the need to resort to questionable assumptions or constraints, additional independent information can be derived with the use of isotopic tracers. ¹³C-metabolic flux analysis (¹³C-MFA) has recently been increasingly applied to mammalian systems in order to obtain a more detailed and accurate description of cellular physiology. The technique is based on culturing cells with specific ¹³C-labeled substrates and subsequently measuring the label distribution in the network's metabolites using either mass spectroscopy (Ahn and Antoniewicz, 2011, Bonarius et al., 2001, Hofmann et al., 2008, Metallo et al., 2009, Niklas et al., 2011, Sengupta et al., 2011) or NMR analysis (Bonarius et al., 2001, Goudar et al., 2010, Mancuso et al., 1998). However, these approaches are experimentally challenging; they typically require the reliable and accurate determination of the mass distributions of several free intracellular metabolites, which necessitates careful and labor-intensive extraction/analysis procedures. Moreover, to ensure the achievement of proper metabolic and isotopic steady-states, such experiments are preferably carried out in a chemostat, which requires the setup of a complex culture system. Recent studies emphasized that adequate (pseudo-)steady conditions may be difficult to achieve in typical batch/fed-batch cultures (Deshpande et al., 2009). While transient isotopic studies can be performed (Ahn and Antoniewicz, 2011, Maier et al., 2008, Young et al., 2007), they are even more experimentally/computationally demanding since, in addition to mass isotopomer distributions, the intracellular pool concentrations must either be accurately measured or estimated along with the unknown fluxes.

In the current contribution, we demonstrate that reliable estimates can be obtained for several key intracellular fluxes using an approach that alleviates the need for a complex chemostat setup and the difficulties inherent in the extraction/analysis of intracellular metabolites. Cells were grown in parallel semi-continuous cultures containing various labeled glucose and glutamine tracers and only the resulting mass isotopomer distributions of extracellular metabolites (three excreted amino acids and lactate) were measured by LC-QTOF MS. The semi-continuous (or repeated batch) mode provides a simple and flexible operation with good approximation of (pseudo)steady-state conditions (Henry et al., 2008). This method was used to analyze the central carbon metabolism of a mammalian cell line in relation with cellular productivity. To this end, we have employed a CHO cell line that has been engineered with an inducible expression system called the “cumate gene-switch” (Gaillet et al., 2007, Mullick et al., 2006). These cells are capable of robust growth in serum-free and protein-free suspension cultures and, upon addition of a non-toxic small molecule (cumate) in the culture medium, they start to express a recombinant antibody. Comparing the metabolism of the induced and non-induced cells using ¹³C-MFA allowed investigating the effect of the onset of protein expression on the primary metabolism of the cells. Confidence intervals for the evaluated fluxes were calculated to properly judge the significance of the observed similarities/differences. Such analysis is invaluable for the rational development of improved cell culture processes on both the metabolic and process scales (Boghigian et al., 2010).