March 4, 2022 | News & Notices Verification Testing Started with AI Startup Epistra for Joint Development of Solution for Optimizing Cell Culturing Parameters

Shimadzu Corporation and the AI startup company Epistra Inc. (Address: Minato City, Tokyo; Representative: Representative Director and CEO Yosuke Ozawa; hereinafter “Epistra”) agreed to jointly develop a solution for optimizing culturing parameters used to cultivate cells. The companies intend to commercialize the solution by 2024.

The solution being developed will use a Shimadzu triple quadrupole liquid chromatograph mass spectrometer (LC-MS/MS) system with the LC/MS/MS Method Package for Cell Culture Profiling Ver. 2 software to measure components (125 components) in culture supernatant, use Epistra Accelerate AI-Based Design of Experiment Support software to predict the culture status and then create a design of experiment based on that predicted culture status. The solution will increase the efficiency of determining cultivation parameters for production of antibody drugs, gene therapy, cell therapy, or regenerative medicine.

When certain substances are produced inside cells, activation of gene networks (signal transduction and metabolic systems) and other factors result in the metabolite content and morphology of cells also changing accordingly. Therefore, identifying the characteristics of component patterns in culture supernatants that contain metabolites can result in shortening the time required to evaluate parameters for the purpose of increasing production quantities of target substances. As members of the Manufacturing Technology Association of Biologics (MAB), Shimadzu Corporation and Epistra received cells* from MAB for the purpose of jointly developing a solution for optimizing cell cultivation parameters. They also received help and technical guidance from individuals involved in MAB for starting experiments intended to demonstrate the solution. Shimadzu and Epistra will also develop AI solutions for improving the final yield and quality of antibodies based on analyzing culture supernatant in an LC-MS/MS system.

  • * This research was supported by the Japan Agency for Medical Research and Development (AMED) under grant number JP20ae0101066.

Photo: C2MAP System Configured with a C2MAP-2030 Automatic Pretreatment System and an LC-MS/MS System for Verification Testing

Photo: C2MAP System Configured with a C2MAP-2030 Automatic Pretreatment System and an LC-MS/MS System for Verification Testing

LC/MS/MS Method Package for Cell Culture Profiling Ver. 2

This LC/MS/MS method can be used to analyze 125 components, including culture media components and metabolites secreted from cells, within 20 minutes. Previously, component profiling was a very time-consuming process because each group of compounds, such as amino acids and vitamins, had to be analyzed separately, but now the LC/MS/MS Method Package for Cell Culture Profiling can be used to analyze all the components simultaneously within 20 minutes. Furthermore, the analytical method is optimized for simultaneous analysis of culture media that contains high concentrations of some components, such as glucose and amino acids, while also containing trace concentrations of other components, such as vitamins. The ready-to-use method allows analysis to be started immediately, without the need to perform tedious processes like determining separation parameters or optimizing MS parameters for each compound.

Epistra Accelerate Automatic Experiment Optimization System

Life sciences R&D work involves a variety of challenges due to the presence of manual experiment processes and repetition of trial and error steps. In contrast, the Epistra Accelerate automatic experiment optimization system uses AI technology to identify optimal parameter settings using fewer trial iterations than ever before.
The unique algorithm used in the automatic design of experiment system developed independently by Epistra is based on Bayesian optimization technology enhanced with AI for solving life science problems. The unique Epistra algorithm resolves all three problems associated with using standard Bayesian optimization technology to solve life science problems (high dimension, high noise, and high cost).
In joint research with the Institute of Physical and Chemical Research, Epistra successfully used Epistra Accelerate to automatically search for protocols that differentiate colonies more efficiently than a skilled expert. Epistra Accelerate is now increasingly being adopted by major biomaterial, food, and pharmaceutical companies.