RESUMO
Soy hydrolysates are widely used as a nutrient supplement in mammalian cell culture for the production of recombinant proteins. The batch-to-batch variability of a soy hydrolysate often leads to productivity differences. This report describes our metabolomics platform, which includes a battery of LC-MS/MS modes of operation, and advanced data analysis software for automated data processing. The platform was successfully used for screening productivity markers in soy hydrolysates during the production of two therapeutic antibodies in two Chinese hamster ovary cell lines. A total of 123 soy hydrolysate batches were analyzed, from which 62 batches were used in the production runs of cell line #1 and 12 batches were used in the production runs of cell line #2. For cell line #1, out of 19 amino acids, 106 other metabolites and 4,131 peptides identified in the soy hydrolysate batches being used, several nucleosides and short hydrophobic peptides showed negative correlation with antibody titer, while ornithine, citrulline and several amino acids and organic acids correlated positively with titer. For cell line #2, only ornithine and citrulline showed strong positive correlation. When ornithine was spiked into the culture media, both cell lines demonstrated accelerated cell growth, indicating ornithine as a root cause of the performance difference. It is proposed that better soy hydrolysate performance resulted from better bacterial fermentation during the hydrolysate production. A few selected markers were used to predict the performance of other soy hydrolysate batches for cell line #1. The predicted titers agreed with the experimental values with good accuracy.
