Bridging Offline Functional Model Carrying Aging-Specific Growth Rate Information and Recombinant Protein Expression:...

Area of Science:

• Biotechnology
• Bioprocess Engineering
• Mathematical Modeling

Background:

• Recombinant protein expression is crucial in biopharmaceutical manufacturing.
• Accurate mathematical models are needed to optimize bioprocesses.
• Existing model selection criteria like AIC can lead to overfitting.

Purpose of the Study:

• To develop and validate a robust mathematical model for recombinant protein expression.
• To address the overfitting issue in model selection.
• To identify key biological parameters influencing protein production.

Main Methods:

• Development of a mathematical model for fed-batch cultivation.
• Application of an entropic extension to the Akaike information criterion (AIC) to prevent overfitting.
• Pseudo-global optimization for estimating recombinant protein concentration.
• Model fitting using data from 70 fed-batch experiments across two biopharmaceutical sites.

Main Results:

• The developed model accurately describes recombinant protein expression.
• Functional models incorporating average cell age and specific growth at induction are optimal descriptors.
• The proposed entropic extension and tuning coefficient effectively mitigate overfitting.
• Model fitting was successfully performed on 327 protein concentration samples.

Conclusions:

• The novel mathematical model enhances the prediction of recombinant protein concentration.
• Cellular factors like age and growth rate are critical for optimizing protein biosynthesis.
• The modified AIC approach facilitates the development of more parsimonious and efficient models.
• This work supports efficient maximization of microbial products in upstream biomanufacturing.