BACKGROUND The effects of alternative co‐substrate feeding strategies on recombinant human erythropoietin (rHuEPO) production by Pichia pastoris‐Mut+ strain were investigated. RESULTS Five different production strategies were designed and performed in the production phase of the bioprocesses with the use of either mannitol or sorbitol as the co‐substrate. The highest rHuEPO production was achieved as CrHuEPO= 0.65 g L−1 at t=9 h, with the cell concentration Cx=55 g L−1 in the production phase; wherein methanol was fed to the bioreactor with a predetermined dynamic feeding rate calculated for constant μM0=0.03 h−1; and three consecutive pulses of the co‐substrate mannitol were introduced at t=0, 6, and 12 h, so that CMan=50 g L−1. The overall cell and product yields on the substrates methanol and mannitol were found, respectively, as YX/St=0.22 g g−1 and YrHuEPO/St=3.74 mg g−1. CONCLUSIONS The use of mannitol as the co‐substrate enhanced rHuEPO production and furthermore shortened the bioprocess time. The design of semi‐batch feeding strategy, based on the selection of the substrates and the co‐substrates, and their dynamic feeding into the bioreactor, is important to increase the production and productivity in r‐protein production by Mut+ strains of P. pastoris, and the developed strategy would be especially important for therapeutic glycoprotein productions by P. pastoris. © 2013 Society of Chemical Industry

BACKGROUND Biochemical reaction network of Pichia pastoris was improved by including N‐ glycosylation pathway reactions to determine the intracellular reaction rates for glycosylated protein production. RESULTS When co‐substrate sorbitol was replaced with mannitol, 12‐fold increase in erythropoietin flux was obtained as a result of 1.2‐ and 2.4‐fold increase in glucose‐6‐phosphate (G6P) formation from fructose‐6‐phosphate (F6P) and 3‐phospho‐D‐glycerate (3PG) formation from glyceraldehyde‐3‐phosphate (G3P), respectively. Thus, to analyse the fluxes around F6P and the fluxes of the glycosylation pathway reactions in depth, the biochemical reaction network of P. pastoris containing 204 reactions and 129 metabolites was developed and used. The flux analysis for glycosylated erythropoietin production with three glycan residues having 12 mannoses (EPO‐Man12) reveals that the fluxes from G3P and guanosine 5′‐diphosphate (GDP) to F6P synthesis were 1.6‐ and 3‐fold higher than that of the fluxes determined for erythropoietin production with no glycans (EPO), respectively. Furthermore, the results of theoretical data‐based overproduction capacity show that the fluxes of EPO‐Man12 and partially humanized EPO with three glycan residues having five mannoses (EPO‐Man5) were 3.6 and 3.8 µmol gDCW‐1 h‐1, respectively, when the methanol and mannitol uptake rates were 2.10 mmol gDCW‐1 h‐1 and 1.49 mmol gDCW‐1 h‐1. For glycoprotein production, fluxes around F6P were found to be important as UDP‐N‐acetyl‐α‐D‐glucosamine (UDP‐GlcNAc) and guanosine 5′‐diphosphate‐D‐Mannose (GDP‐Man) were produced from F6P. Thus the co‐substrate mannitol does not repress AOX promoter but is easily converted to F6P, enhancing glycoprotein production. CONCLUSION A metabolic reaction network for P. pastoris with glycosylation reactions is proposed for the determination of fluxes for glycoprotein production by P. pastoris. © 2013 Society of Chemical Industry