Pichia pastoris as an efficient host for the production of recombinant proteins is mostly cultivated in fed-batch mode in which the cell’s environment is continuously changing. Therefore, to fine-tune bioreactor performance in respect to the associated metabolic changes of the microorganism, it is crucial to understand the influence of feeding strategy parameters on the intracellular reaction network. In this study, dynamic flux balance analysis (DFBA) integrated with transcriptomics data was used to simulate the recombinant P.pastoris (Muts) growth during induction phase for 3 fed-batch µ-stat strategies. The induction phase was divided into equal time intervals and the correlated reactions with protein yield were identified in the 3 fed-batch strategies using the Pearson correlation coefficient. Subsequently, Principal Component Analysis was applied to cluster induction phase time intervals and identify the role of correlated reactions on metabolic differentiation of time intervals. It was found that increasing fluxes through the methanol dissimilation pathway increased protein yield. By adding a methanol assimilation pathway inhibitor (HgCl2) to the shake flask medium containing 10% (v/v) glycerol, the protein titer increased by 60%. Using the DFBA revealed that the higher the dimensionless flux of methanol, the higher amounts of protein yield. Finally, a novel feeding strategy was developed so that the dimensionless methanol flux increased compared to the performed cultivations. Protein titer increased by 16% compared to the optimally performed cultivation, while production yield increased by 85%.

Ammonia is a toxic byproduct of CHO cell metabolism, which inhibits cell growth, reduces cell viability, alters glycosylation, and decreases recombinant protein productivity. In an attempt to minimize the ammonium accumulation in cell culture media, different amino acids were added individually to the culture medium before the production phase to alleviate the negative effects of ammonium on cell culture performance. Among all the amino acids examined in this study, valine showed the most positive impact on CHO cell culture performance. When the cultured CHO cells were fed with 5 mM valine, EPO titer was increased by 25% compared to the control medium, and ammonium and lactate production were decreased by 23 and 26%, respectively, relative to the control culture. Moreover, the sialic acid content of the EPO protein in valine-fed culture was higher than in the control culture, most likely because of the lower ammonium concentration. Flux balance analysis (FBA) results demonstrated that the citric acid cycle was enriched by valine feeding. The analysis revealed that there might be a link between promoting tricarboxylic acid (TCA) cycle metabolism in valine-fed culture and reduction in lactate and ammonia accumulation. Furthermore, in valine-fed culture, FBA outcomes showed that alanine was excreted into the medium as the primary mechanism for reducing ammonium concentration. It was predicted that the elevated TCA cycle metabolism was concurrent with an increment in recombinant protein production. Taken together, our data demonstrate that valine addition could be an effective strategy for mitigating the negative impacts of ammonium and enhancing glycoprotein production in both quality and quantity.

Recent noteworthy advances in the development of high-performing microbial and mammalian strains have enabled the sustainable production of bio-economically valuable substances such as bio-compounds, biofuels, and biopharmaceuticals. However, to obtain an industrially viable mass-production scheme, much time and effort are required. The robust and rational design of fermentation processes requires analysis and optimization of different extracellular conditions and medium components, which have a massive effect on growth and productivity. In this regard, knowledge- and data-driven modeling methods have received much attention. Constraint-based modeling (CBM) is a knowledge-driven mathematical approach that has been widely used in fermentation analysis and optimization due to its capabilities of predicting the cellular phenotype from genotype through high-throughput means. On the other hand, machine learning (ML) is a data-driven statistical method that identifies the data patterns within sophisticated biological systems and processes, where there is inadequate knowledge to represent underlying mechanisms. Furthermore, ML models are becoming a viable complement to constraint-based models in a reciprocal manner when one is used as a pre-step of another. As a result, more predictable models are produced. This review highlights the applications of CBM and ML independently and the combination of these two approaches for analyzing and optimizing fermentation parameters.