(175am) Macroscopic Modeling of Bioreactors for Recombinant Protein Producing Pichia Pastoris in Defined Medium

The methylotrophic yeast Pichia pastoris is widely used as the microbial host for recombinant protein production. This system has the advantages of (i) tight gene regulation and high product titers with methanol-induced alcohol oxidase (AOX) promoter, (ii) growth on simple media at high cell density with preference for respiratory growth, (iii) eukaryotic post-translational modification such as glycosylation and disulfide bond formation, and (iv) simplified downstream purification with foreign protein secretion and low amount of naturally secreted proteins [1].

Bioreactor modeling can inform understanding of cellular metabolism—including aspects of growth kinetics, productivity, media consumption, and metabolite secretion—and can be used to optimize bioreactor operations. Models for bioreactors are developed with different level of details and complexity depending on their purpose. For P. pastoris, complex cellular models such as genomic-scale metabolic models have been developed [2] to describe detailed cellular metabolism using data on gene expression. These models, however, have too many parameters to be accurately estimated from experimental data, due to large number of states and parameters. As such, most of the parameters in these models are highly uncertain, even when additional assumptions and/or approximations are made such as the metabolic fluxes being at steady state. In contrast, macroscopic models can yield a reasonable description of the bioreactor while using simpler mathematical formulations with low computational costs. These types of models can also facilitate real-time applications such as adaptive parameter estimation and model predictive control.

Although P. pastoris has been used to produce numerous recombinant proteins, limited macroscopic models have been published, with their main focus being on the substrate consumption, growth, and production [3]–[6]. In this work, an extensive macroscopic bioreactor model was constructed for substrate (glycerol and methanol), biomass, and recombinant protein, and for other medium components and off-gas components (oxygen and carbon dioxide). The modeling of the medium components was facilitated by using a chemically defined medium [7], which also reduces run-to-run variability and simplifies purification in the downstream processes. Species and elemental balances were introduced to describe uptake and evolution rates for medium components and off-gas components. Additionally, a pH model was constructed using an overall charge balance, acid/base equilibria, and activity coefficients to describe the dependence of recombinant protein production on pH, precipitation of medium components, and carbon dioxide to carbonate species reactions.

For a set of bioreactors operated in fed-batch with P. pastoris, variations among individual examples—especially with respect to volumetric productivity for an exemplar recombinant subunit vaccine component for rotavirus—were observed even with highly similar target operating conditions with controlled temperature, pH, and dissolved oxygen. These observations suggested that the experimental data for the bioreactors are not describable by a mathematical model with a single set of parameters. An additional consideration is that any macroscopic model does not describe all of the complex biology that occurs in the bioreactor; such models effectively lump multiple biological pathways into some of the model parameters. The extent of run-to-run variability was modeled by using distributions for a subset of the model parameters, in which each run was associated with a single set of model parameters. The model parameters with low uncertainty were collected from the literature, and the distributions of the uncertain parameters in the macroscopic bioreactor model were estimated using the maximum likelihood method. The probability distribution of the model predictions quantified from the parameter distribution was quantifiably consistent with the run-to-run variability observed in the experimental data.

The uncertainty description in this macroscopic bioreactor model identifies the model parameters that have large variability and provides guidance as to which aspects of cellular metabolism should be the focus of additional experimental studies. The model for medium components, pH, and precipitation can be used for improving the chemically defined medium by minimizing the amount of components needed while meeting cellular requirements.

References:

[1] J. L. Cereghino and J. M. Cregg, Heterologous protein expression in the methylotrophic yeast Pichia pastoris, FEMS Microbiol. Rev., vol. 24, no. 1, pp. 45–66, 2000.

[2] M. Tomàs-Gamisans, P. Ferrer, and J. Albiol, Integration and validation of the genome-scale metabolic models of Pichia pastoris: a comprehensive update of protein glycosylation pathways, lipid and energy metabolism, PLOS One, vol. 11, no. 1, art. no. e0148031, 2016.

[3] M. Jahic, J. Rotticci-Mulder, M. Martinelle, K. Hult, and S.-O. Enfors, Modeling of growth and energy metabolism of Pichia pastoris producing a fusion protein, Bioprocess Biosyst. Eng., vol. 24, no. 6, pp. 385–393, 2002.

[4] H. T. Ren, J. Q. Yuan, and K.-H. Bellgardt, Macrokinetic model for methylotrophic Pichia pastoris based on stoichiometric balance, J. Biotechnol., vol. 106, no. 1, pp. 53–68, 2003.

[5] E. Çelik, P. Çalık, and S. G. Oliver, A structured kinetic model for recombinant protein production by Mut+ strain of Pichia pastoris, Chem. Eng. Sci., vol. 64, no. 23, pp. 5028–5035, 2009.

[6] J. M. Barrigon, F. Valero, and J. L. Montesinos, A macrokinetic model‐based comparative meta‐analysis of recombinant protein production by Pichia pastoris under AOX1 promoter, Biotech. Bioeng., vol. 112, no. 6, pp. 1132–1145, 2015.

[7] C. B. Matthews, A. Kuo, K. R. Love, and J. C. Love, Development of a general defined medium for Pichia pastoris, Biotech. Bioeng., vol. 115, no. 1, pp. 103–113, 2018.