# Redefining media blending mathematically: a systematic approach for screening of medium components

## TL;DR

This paper introduces a mathematical workflow for optimizing cell culture media components through systematic media blending, improving efficiency in biopharmaceutical production.

## Contribution

The paper proposes a systematic workflow for media blending using D-optimal design and principal components to screen and optimize culture media components.

## Key findings

- A workflow for media blending was developed using D-optimal design and principal components to handle multicollinearity.
- A case study demonstrated the workflow's effectiveness in identifying key components affecting viable cell concentrations.
- The workflow enabled the design of a dedicated media set for media blending.

## Abstract

Biopharmaceuticals, such as antibody therapeutics, are produced by culturing mammalian cells with chemically defined media that consist of more than 50 synthesized components. The screening of medium components related to culture performance and the subsequent optimization of the composition are required in the development of new modalities, host cells, and culture methods. Screening the components to be optimized is typically labor-intensive. The easiest approach is media blending, which creates variations in the concentrations of the components with only liquid mixing. However, a workflow for systematically determining experimental conditions (i.e., how to blend media) has not been established. Therefore, we reassessed media blending from a mathematical perspective and proposed a workflow for the first time. In the workflow, we evaluated the use of a commercially available chemically defined media to maximize simplicity and applicability. From a mathematical perspective, we clarified that multicollinearity is an inevitable challenge in both experimental design and its analysis. Under the constraint, we showed that one of the most appropriate experimental conditions could be systematically calculated and selected by applying D-optimal design focusing on the principal components. We performed a case study of cell culture to screen medium components under 120 experimental conditions using 11 chemically defined media designed for Chinese Hamster ovary cells. The case study provided a reasonable set of components that explained the variance in viable cell concentrations, which range from 5.8 to 19.4 (× 106) cells/mL. Finally, our mathematical redefinition also enabled the design of a dedicated media set for media blending.

• The constraints in media blending were clearly explained.

• A systematic workflow from blending design to analysis was proposed.

• The workflow also enabled the design of a dedicated media set for media blending.

The online version contains supplementary material available at 10.1007/s00253-025-13594-z.