An automated data-driven DSP development approach for glycoproteins from yeast

Area of Science:

• Biotechnology
• Process Chemistry
• Data Science

Background:

• Downstream process development for yeast-derived recombinant glycoproteins is challenging due to hyperglycosylation.
• Previous work utilized a two-step flowthrough approach with monolithic columns for glycoprotein purification.

Purpose of the Study:

• To investigate a novel automated data science approach for identifying purification conditions for glycoproteins using monolithic columns.
• To assess the efficiency of monolithic columns in separating recombinant glycoproteins.

Main Methods:

• Performed automated design of experiments at analytical scale for three recombinant horseradish peroxidase (HRP) isoenzymes.
• Introduced a relative impurity removal (IR) term for quantifying separation efficiency.
• Automated experimental procedures and data analysis, completing each run in under 40 minutes.

Main Results:

• Identified optimal purification conditions for HRP isoenzymes using the automated approach.
• Validated analytical scale findings at laboratory scale, confirming results with offline analyses.
• Established a strong correlation between online-estimated IR and offline-determined IR.

Conclusions:

• Presents a novel, automated, data-driven methodology for rapid downstream process development.
• Leverages analytical scale advantages for efficient purification of recombinant glycoproteins from yeast.
• Eliminates the need for offline analyses, significantly accelerating process optimization.