¹Department of Chemical Engineering and Materials Science, ²Department of Computer Science and Engineering, University of Minnesota

The productivity of recombinant mammalian cells has reached levels that rival professional protein secretors in vivo. Such hyperproductivity is a composite of many superior characteristics. To elucidate the high productivity trait, comparative transcriptome analysis of high producing cell lines and high productivity culture conditions have proved to be illuminating.

Supervised machine learning tools were employed to identify expression patterns that correlate with productivity.  Gene testing (GST) analysis was used to identify physiological functions that are enriched in high producers and high productivity culture conditions.  Major functional classes identified include those involved in protein processing and transport, such as protein modification, vesicle trafficking and protein turnover. A significant proportion of genes involved in mitochondrial ribosomal function, cell cycle regulation, cytoskeleton-related elements are also differentially up-regulated in high producers.

Data-driven, pattern recognition approaches are high effective in revealing the physiological basis of high productivity.  Such a systematic elucidation of the complex gene-trait relationship will help establish the platform for a scientific, cell engineering –based approach for development of high producing cell lines.