Extracorporeal bioartificial liver (BAL) devices, which incorporate living hepatocytes in a bioreactor, represent a promising technology to support and extend the life of critically ill patients until a liver transplantation becomes available. The major limitation of BAL devices is that hepatocytes are anchorage-dependent cells which rapidly lose their liver-specific functions when they are removed from their host and cultured in an artificial environment [1]. In order to maintain liver-specific cellular functions, a variety of approaches have been proposed including sandwich collagen culture, hormone preconditioning and amino acid supplementation [2-4]. Although these experimental designs improve the differentiated cell functions to some extent, the levels of liver-specific functions are still lower in culture than in vivo. The main goal of this work is to achieve the best performance of liver-specific functions of hepatocytes in culture by adjusting amino acid supplementation in the medium.

We utilize flux balance analysis (FBA) as a computational tool to design amino acid supplementation protocols and to interpret the experimental results. Since the objective in terms of hepatic function is not restricted to maximizing one production rate but a series of different liver-specific functions (detoxification, protein synthesis, urea secretion), some of them were found to compete against each other [5]. In this work, the optimal flux distribution for the hepatic network was identified by considering different objectives simultaneously in terms of urea secretion and albumin synthesis. A number of different solution approaches have been developed to solve the multiobjective optimization in order to determine the pareto optimal set [6-7]. The -constraint method [8] was utilized in this work to identify the optimal flux distributions, which maximizes a primary objective while constraining the remaining objective(s). The constraints of linear programming are important to reduce the solution space and to find the solutions close to real conditions of cells. The flux bounds are determined using the full range of a set of previously reported experimental settings [9]: (a) high/ low insulin preconditioned unsupplemented plasma cultures (HIP/ LIP), and (b) high/ low insulin preconditioned, amino acid-supplemented plasma cultures (HIP_AA/LIP_AA) in order to explore the hepatocyte capabilities based on the assumption that this represents a sampling of accessible physiological states. The classification of reversible versus irreversible reaction was made based on the information given in the metabolic map of KEGG as thermodynamic constraints [10]. The irreversible reactions are considered to have positive fluxes whereas the reversible reactions can have either positive or negative flux values. This mathematical model helps to elucidate the flux distribution associated with the maximum value of cell functions (urea secretion and albumin synthesis). By comparing the results with those in the literature [9], the computationally predicted optimal level of urea production was found to be increased more than two fold than has been observed experimentally. Higher urea production is achieved by an increase of fluxes of gluconeogenesis, TCA cycle, urea cycle fluxes, pentose phosphate pathway and uptake of some amino acids such as arginine, serine, glycine etc.

Based on the optimal amino acid uptake rates obtained from the solutions of linear programming, we can determine the optimal concentration of amino acids in the media for the desired output of maximum urea and albumin production. However, the ability of the liver to metabolize large amounts of amino acids depends on its capability to transport amino acids through the cell membrane. In order to investigate whether transport limitation exist when higher amino acids supplementations are available, we are performing experiments, in which each amino acid is examined separately by changing its concentration in the plasma. Their amino acid concentrations are measured using HPLC and their fluxes calculated from the changes of concentration for a specific time period. By analyzing these experimental data, we can determine whether the fluxes of amino acids vary linearly with the concentration and whether there exists a limitation of transport for particular amino acid: (1) when the relationship is found to be linear, there is no limitation in amino acids transport, (2) for a particular amino acid, when a nonlinear function is determined to characterize the relation between its flux and concentration, it suggests that a transporter limitation exists for this particular amino acid. These transport constraints are incorporated back into the FBA model and the optimal flux distribution recalculated.

In this work, ammonia detoxification and albumin synthesis are considered as the study end-points to investigate our hypothesis that liver specific functions will be improved by modulation of hepatic network using optimal amino acids supplementation. Having quantified and incorporated amino acid transport constraints into the model, we expect closer approach between model and experiment. Remaining discrepancies, which may reflect gene regulatory constraints, will also be discussed.

Reference: [1] Allen, J. W., Hassanein, T. and Bhatia, S. N., 2001. Advances in bioartificial liver devices. Hepatology 34(3), 447-455. [2] Washizu, J., Berthiaume, F., Chan, C., Tompkins, R. G., Toner, M. and Yarmush, M. L., 2000. Optimization of rat hepatocyte culture in citrated human plasma. Journal of Surgical Research 93(2), 237-246. [3] Bhatia, S. N., Yarmush, M. L. and Toner, M., 1997. Controlling cell interactions by micropatterning in co-cultures: Hepatocytes and 3T3 fibroblasts. Journal of Biomedical Materials Research 34(2), 189-199. [4] Chan, C., Berthiaume, F., Lee, K. and Yarmush, M. L., 2003. Metabolic flux analysis of hepatocyte function in hormone- and amino acid-supplemented plasma. Metab Eng 5(1), 1-15. [5] Sharma, N. S., Ierapetritou, M. G. and Yarmush, M. L., 2005. Novel quantitative tools for engineering analysis of hepatocyte cultures in bioartificial liver systems. Biotechnol Bioeng 92(3), 321-335. [6] Das I, Dennis JE. 1998. Normal-boundary intersection: A new method for generating the Pareto surface in nonlineae multicreteria optimization problems. Siam J Optimization 8: 631-657. [7] Hillermeier C. 2001. Generalized homotopy approach to multiobjective optimization. J Optimization Theory Appl 110: 557-583. [8] Haimes YY. 1973. Integrated system identification and optimization. Cont DynSys: Adv Theory Appl 10: 435. [9] Chan, C., Hwang, D., Stephanopoulos, G. N., Yarmush, M. L. and Stephanopoulos, G., 2003c. Application of multivariate analysis to optimize function of cultured hepatocytes. Biotechnol Prog 19(2), 580-598. [10] Kanehisa, M. and Goto, S., 2000. KEGG: kyoto encyclopedia of genes and genomes. Nucleic Acids Res 28(1), 27-30.