Metabolic flux analysis of HEK-293 cells in perfusion cultures for the production of adenoviral vectors

Abstract

To meet increasing needs of adenovirus vectors for gene therapy programs, development of efficient and reproducible production processes is required. Perfusion cultures were employed to allow infection at greater cell concentrations. In an effort to define culture conditions resulting in enhanced productivities, experiments performed at different feed rates and infected at various cell densities were compared using metabolic flux analysis. The highest specific product yields were achieved in experiments performed at high perfusion rates and/or low cell concentrations. The intracellular flux analysis revealed that these experiments exhibited greater glycolytic fluxes, slightly higher TCA fluxes, and greater ATP production rates at the time of infection. In contrast, cultures infected at high cell density and/or low medium renewal rates were characterized by a more efficient utilization of glucose at the time of infection, but the specific product yields achieved were lower. The intracellular flux analysis provided a rational basis for the implementation of a feeding strategy that allowed successful infection at a density of 5×10⁶ cells/ml.

Introduction

Adenoviruses constitute important gene delivery vehicles in the field of gene therapy and are currently involved in one-third of the protocols, mainly for the treatment of cancer. To meet the increasing demand of clinical trials, the development of high-yield processes is necessary. One of the major challenge of adenoviral vector production lies in the fact that high cell-specific productivities are only maintained for infection at low cell densities (typically under 1×10⁶ cells/ml) (Nadeau and Kamen, 2003). Efforts to increase volumetric productivities have aimed at designing new feeding strategy to allow productive infection at higher cell concentrations. Substantial improvements were achieved with sequential batch (Frazzati-Gallina et al., 2001; Garnier et al., 1994; Iyer et al., 1999; Merten et al., 1999), and fed-batch strategies (Lee et al., 2003; Nadeau et al., 1996; Wong et al., 1999). This suggests that the decreased specific productivity at high cell concentrations is due to nutrient depletion or waste inhibition, though the exact nature of these limitations remains unknown. The cell density effect was also reported in perfusion cultures (Henry et al., 2004). For a set perfusion rate, infecting at higher cell concentrations led to a dramatic decrease in productivity. Henry et al. (2004) also showed that for a given cell density at infection, increasing the perfusion rate yields greater productivity.

All these results suggest that the physiological state of the cells at the time of infection is a determinant factor for culture productivity. Consequently, the design and operation of an efficient process must aim at maintaining or driving the cells towards a favourable physiological state before proceeding with the infection. The difficulty arises in establishing a quantitative description of this physiological status by a set of several process variables providing relevant information. In the context of viral vector production, various on-line and off-line methods have been investigated to assess the status of an infection. Increased oxygen uptake rate during virus production have been reported (Garnier et al., 1994; Kussow et al., 1995). A fluorescence probe allowed Gilbert et al. (2000) to determine the harvest time during an infection by an adenovirus containing the gene for Green Fluorescence Protein (GFP). Also, increased glucose uptake and lactate production rates were noted following infection (Garnier et al., 1994; Iyer et al., 1999; Nadeau et al., 2002; Xie et al., 2002). All the aforementioned methods can give valuable information regarding the kinetics of infection but none of them can provide a priori indication on the productivity of the culture. To take a step further, one needs to include information at the intracellular level that may unravel how metabolic flux distribution influences production.

In this work, under the hypothesis that the cell metabolic activity may be governing the productivity, metabolic flux analysis was applied to characterize the metabolism of 293 cells grown and infected in six perfusion cultures to evaluate the effects of varying the perfusion rate and the cell density at infection. The intracellular fluxes prior and after infection were estimated using a 40-flux metabolic network and 25 material balance equations. Observed differences in maximum viral vector concentrations are analysed in relation with the metabolic flux profiles of each culture. Results demonstrate that the productivity of cells is intimately linked with their physiological state, which is largely dictated by the operating conditions. With the goal of maximizing productivity while retaining process simplicity and lowering medium costs, the analysis led to the design of a feeding strategy allowing successful infection at cell density up to 5×10⁶ cells/ml.