Cartagena, Colombia, Octubre 6 a 8 de 2014 Determination of the optimal operation conditions to maximize the biomass production in plant cell cultures of thevetia peruviana using multi-objective optimization Adriana Villegas a,c *, Juan Pablo Arias b, Daira Aragón d, Mario Arias b, Silvia Ochoa,a a Research Group in Simulation, Design, Control and Optimization of Chemical processes (SIDCOP), Facultad de ingeniería, Universidad de Antioquia. Calle 67 No. 53 108. Medellín Colombia b Research Group in Industrial Biotechnology, Facultad de Ciencias, Universidad Nacional de Colombia Sede Medellín. Calle 59A No 63-20 Medellín, Colombia b Termomec Research Group, Facultad de Ingeniería, Universidad Cooperativa de Colombia. Calle 48 No 28 00 Medellín Colombia d Audubon Sugar Institute, LSU AgCenter. 3845 Highway 75, St. Gabriel, LA 70776, United States *E-mail:Adrianavillegas@udea.edu.co Abstract In this work, multi-objective optimization is used to determine the initial sucrose concentration and initial inoculum concentration in order to increase the biomass production for plant cell cultures of Thevetia peruviana, while minimizing the operating costs. For this approach, multi-objective genetic algorithm (MOGA) provided by matlab toolbox was used. The best solutions were chosen from the pareto front in agreement with expert criterion. Results obtained indicate that the best interval for initial conditions is between 30g/L and 20g/L of initial sucrose concentration and between 3.6g/L and 4.2g/L for initial inoculum concentration. Multi-objective optimization gives a new vision of the optimization problem. Keywords: Multi-objective optimization, Biomass production, Plant Cell Cultures, Mathematical Modelling, Parameter Determination, Pareto Front. 1. Introduction In optimization, the main objective is to find one or more solutions to minimize or maximize a given objective function. Multi-objective optimization attempts to find the solutions that are optimal for several conflicting objectives simultaneously. (Vera et al., 2003) These solutions are called Pareto Optimal solution. Despite the existence of multiple Pareto Solutions, in practice, only one of these solutions will be chosen taking into account previous knowledge. (Branke et al., 2008) Multi-objective optimization has been reported by: (Vera et al., 2003) to maximize ethanol concentration and minimize each internal metabolite concentration for ethanol production from Saccharomyces cerevisiae. (Wang and Sheu, 2000) used multi-objective optimization to estimate the kinetic model parameters of batch and fed batch fermentation processes for ethanol production using Saccharomyces diastaticul, and (Brunet et al., 2011) introduce a novel framework for the optimal development of biotechnological processes using optimization tools such as multi-objective mixed-integer nonlinear programming.
1192 A. Villegas, J.P. Arias, D. Aragón, M. Arias, S. Ochoa. In the case of plant cell cultures, it is necessary to produce the biomass in order to stimulate the secondary metabolite production. One of the problems associated with biomass and metabolite production are the operating costs. Therefore, in this paper the multi-objective optimization is used to increase biomass production at lower operating costs. In-vitro plant cell cultures are a promising biotechnological alternative to the cultivation of whole plants since these cultures can be conducted on large scale fermenters, eliminating the need of large areas of land and the diseases that affect the crops. When the biomass that grow in the bioreactor has been optimized, new conditions are implemented in order to stimulate the secondary metabolites production and other substances of biological interest (Weathers et al., 2010). In this context, Thevetia peruviana appears as a promising plant with medical applications and availability for the production of important secondary metabolites such as cardiac glicosydes, thevetin A and B, thevetoxin, peruvoside, iridoid glycosides cerberin, among others. (Arias et al., 2010) Given its applications, in this theoretical and experimental work, modelling studies of plant cell cultures of Thevetia peruviana were developed under batch operation in shake flask cultivation in order to determine the best initial conditions of inoculum and sucrose concentration that maximize the biomass production using multi-objective optimization. Finding the optimal initial conditions for maximizing biomass production is the first step towards assuring metabolite production from plant cell cultures of Thevetia peruviana. The results obtained in this phase of the project will be used to optimize this process, which will contribute to technological developments for future industrial implementation. Positive impacts from these developments are expected not only economically but also environmentally by reducing the pressure on native forests. The paper is organized as follows: in section 2, the materials and methods for callus induction, maintenance of cell suspensions and analytical methods for determining biomass concentration and sugars concentrations are presented. In the first part of section 3 the proposed model and methodology for parameter determination is explained. Furthermore, a multi-objective optimization problem is formulated and solved to determine the best initial conditions of substrate and inoculum concentration. 2. Materials and Methods 2.1 Callus induction Collected fruits of Thevetia peruviana were subjected to disinfection protocol. Small pieces of pulp obtained by dissection were used as explants for callus induction in SH medium supplemented with 2,4-D, kinetin, agar and sucrose. ph of the medium was adjusted to 5.8 before autoclaving at 120 C and 15 psi for 15 minutes. The cultures were maintained at ambient temperature under natural photoperiod and were sub-cultured every 15 days (Arias et al., 2010). 2.2 Cell suspensions Cell suspensions were obtained from friable callus which were transferred into a shake flask with 80ml of liquid SH medium. The cultures were maintained in an orbital shaker (IK KS501 digital) at 110rpm and ambient temperature under natural photoperiod. These cultures were sub-cultured every 12-15 days (Arias et al., 2010). The experimental data were performed in liquid medium supplemented with sucrose in concentrations of 30g/L and 20g/L with and inoculum concentrations of 4.6g/L and 4.8/L (DW basis) respectively.
Multiobjective Optimization in biomass production 1193 2.3 Analytical methods Cell growth was determined by dry weight (DW). Samples were collected every 3 days of culture for 15 days, and then filtered on a vacuum filter unit. Retained cells were washed 3 times with distilled water and dried at 57ºC in a convection oven for 48 hours. The filtered medium was stored at -4 C for further analysis of extracellular sugars. Quantification of reducing sugars was determined by HPLC coupled to a refractive index detector using an amino column, and acetonitrile/water as mobile phase. (Arias et al., 2010) 3. Results and Discussion 3.1 Unstructured model The optimization approach is based on a mathematical model of the biochemical system. For this work, an unstructured model was built and is presented in equations (1) - (5). This model describes the dynamics of cell growth [ ] (equation 2), the hydrolysis of sucrose [ ] into glucose [ ] and fructose [ ] (equation 3), and the dynamic behaviour of glucose and fructose (equations 4 and 5). All concentrations were calculated in [ ]. ( + ) ( ) (1) (2) ( ) ( ) (3) ( ) [( ) ] (4) ( ) [( ) ] (5) The model parameters were estimated using as objective function a mean squared error between the model predictions and the experimental data. The Simulated annealing algorithm was used and a mean squared error of 7.7413g 2 /L 2 was obtained. The parameters of the model are presented in Table 1. 3.2 Determination of operating conditions that maximize the biomass productivity of biomass. A multi-objective optimization is used to determine the initial sucrose concentration and initial inoculum concentration that maximize biomass production while the operating costs are minimized using the model proposed by equations (1)-(5) and the operating costs. The costs function is presented by equations (6) and (7) ( + ) + (6)
1194 A. Villegas, J.P. Arias, D. Aragón, M. Arias, S. Ochoa. Table 1: Model parameters for plant cell cultures of Thevetia peruviana Parameter Symbol Value Unit Maximum specific growth rate 5.8 Saturation constant based on glucose 35.7 Saturation constant based on fructose 24.1 Michaelis Menten constant 7.9 Reaction rate of sucrose into glucose 2.03 Reaction rate of sucrose into fructose 2.4 Yield of biomass from glucose 0.98 Yield of biomass from fuctose 0.83 Where: : is the cost of energy consumption per day, : is the cost associated to the personal per day, : is the total volume used for plant cell production in shake flashes, is the initial sucrose concentration and :is the cost of raw materials. Therefore, the final cost function is presented in equation (7) + (7) For this approach, the multi-objective genetic algorithm (MOGA) provided by matlab was used. The formulation of the multi-objective problem is the following: ( ) (8) ( ) (9) Were, ( ) correspond to biomass production in which is the period corresponding to each measurement and is the initial inoculum concentration. Subject to: (10) (11) (12) The summary of several pareto solutions is presented in Figure 1. In Table 2 the most relevant results in the case of plant cell cultures of Thevetia peruviana are selected.
Multiobjective Optimization in biomass production 1195 Figure 1: Pareto solutions for plant cell cultures of Thevetia peruviana. Table 2: Efficient optimum profile solutions in plant cell cultures of Thevetia Peruviana. Test Initial inoculum concentration [g/l] Initial sucrose concentration [g/l] Costs of production*10e5 [$] Biomass production [g/l] 1 3.598 35.492 12.8812849 15.130 2 3.610 30.758 12.0311843 13.163 3 4.236 32.626 12.3913149 13.956 4 3.536 35.867 12.941284 15.291 From table 2, it can be seen that the best conditions for biomass production correspond to the tests 1 and 4, giving a biomass production of 15.13g/L and 15.291g/L respectively. However, the points 2 and 3 present acceptable values of biomass with respect to points 1 and 4 with the lower production costs. Therefore, these points are selected in this work for obtaining biomass production fixing the intervals of 3.6g/L and 4.2g/L of initial inoculum concentration and 30g/L and 32g/L of initial sucrose concentration. These initial conditions will be used for obtaining high biomass concentrations with low operating costs 4. Conclusions In this work, a multi-objective optimization was carried out for determining the initial inoculum concentration and initial sucrose concentration that maximize the biomass production with lower production costs in plant cell cultures of Thevetia Peruviana. Results indicate that inoculum concentration between 3.6g/L and 4.2g/L and concentration of sucrose between 30g/L and 32g/L are recommended for obtaining acceptable values in biomass production with lower operating costs. Currently, in order to validate these results, experiments implementing the initial conditions found in this study are being carried out.
1196 A. Villegas, J.P. Arias, D. Aragón, M. Arias, S. Ochoa. Acknowledgement Financial support from the CODI committee at the Universidad de Antioquia Project EO- 1615 and from the Universidad Nacional de Colombia. Adriana Villegas thanks the financial support from the Universidad Cooperativa de Colombia. Referencias ARIAS, M., ANGARITA, M., RESTREPO, J., CAICEDO, L. & PEREA, M. 2010. Elicitation with methyl-jasmonate stimulates peruvoside production in cell suspension cultures of Thevetia peruviana. In Vitro Cellular & Developmental Biology - Plant, 46, 233-238. BRANKE, J., DEB, K., MIETTINEN, K. & SLOWINSKI, R. 2008. Multiobjective optimization: Interactive and evolutionary approaches, Springer. BRUNET, R., GUILLÉN-GOSÁLBEZ, G. & JIMÉNEZ, L. 2011. Cleaner Design of Single-Product Biotechnological Facilities through the Integration of Process Simulation, Multiobjective Optimization, Life Cycle Assessment, and Principal Component Analysis. Industrial & Engineering Chemistry Research, 51, 410-424. VERA, J., DE ATAURI, P., CASCANTE, M. & TORRES, N. V. 2003. Multicriteria optimization of biochemical systems by linear programming: Application to production of ethanol by Saccharomyces cerevisiae. Biotechnology and Bioengineering, 83, 335-343. WANG, F.-S. & SHEU, J.-W. 2000. Multiobjective parameter estimation problems of fermentation processes using a high ethanol tolerance yeast. Chemical Engineering Science, 55, 3685-3695. WEATHERS, P. J., TOWLER, M. J. & XU, J. 2010. Bench to batch: advances in plant cell culture for producing useful products. Appl Microbiol Biotechnol, 85, 1339-1351.