Improved Prediction of Phosphorus Dynamics in Biotechnological Processes by Considering Precipitation and Polyphosphate Formation: A Case Study on Antibiotic Production with Streptomyces coelicolor

The multiplicity of physico-chemical and biological processes, where phosphorus is involved, makes their accurate prediction using current mathematical models in biotechnology quite a challenge. In this work, an antibiotic production model of Streptomyces coelicolor is chosen as a representative case study in which major difficulties arise in explaining the measured phosphate dynamics among some minor additional issues. Thus, the utilization of an advanced speciation model and a multiple mineral precipitation framework is proposed to improve phosphorus predictions. Furthermore, a kinetic approach describing intracellular polyphosphate accumulation and consumption has been developed and implemented. A heuristic re-estimation of selected parameters is carried out to improve overall model performance. The improved process model predicts phosphate dynamics (Root Mean Squared Error ≤52h: -90 %, Relative Average Deviation ≤52h: -96 %) very accurately in comparison to the original implementation, where biomass growth/decay was the only phosphorus source-sink. In addition, parameter re-estimation achieved an improved description of the available measurements for biomass, total ammonia, dissolved oxygen and actinorhodin concentrations. This work contributes to the existing process knowledge of biotechnological systems in general and especially to antibiotic production with S. coelicolor, while emphasizing the (unavoidable) need of considering both physico-chemical and biological processes to accurately describe phosphorus dynamics.