Cell Culture and Media
A total of 25 14-day fed-batch production cell culture conditions were used to build a training dataset for the model. For all 25 batches, CHO cells were cultured in AMBR® 15 (Sartorius Stedim, GE) mammalian bioreactors seeded at 2E6 cells/mL with an initial working volume of 13 mL. Bioreactors were operated at dissolved oxygen of 30%, pH setpoint of 6.95±0.15 controlled by either CO2 or base addition, agitation at 900 RPM, and incubated at 36.5°C. Control cultures were fed with a high nutrient feed based on a previously defined process. The remaining cultures were fed varying amounts of the same high nutrient feed following the same feeding schedule to obtain a varying distribution of growth and productivity profiles for the training dataset. Glucose was supplemented back to 6 g/L if extracellular glucose levels fell below 3 g/L. Similarly, glutamic acid was supplemented as 3 mM bolus additions if the extracellular glutamic acid levels fell below 4 mM.
To validate the MVDA models for cell growth and mAb productivity, an experiment containing 46-batches was performed. Three baseline controls were designed in which cells were fed with three CD feed media corresponding to high, moderate, and low nutrient levels. In addition to the ranging nutrient feeds (high, moderate, and low), key amino acids identified from either the growth or production model were formulated into amino acid cocktail feeds and supplemented to the cultures at model-informed time points.
Each cocktail feed contained a 100X concentration of the specific amino acids with pH adjusted via 10N NaOH to maintain solubility. Previous studies indicated that AMBR® 15 bioreactors incur a 1 – 1.5% of evaporation rate (v/v per day), which can lead to inaccurate quantification of cell growth and mAb productivity due to the small working volume (Hsu, Aulakh, Traul, & Yuk, 2012). Therefore, 150 µL of sterile water was supplemented daily to compensate for evaporation. To compensate for osmotic and pH differences of amino acid cocktails, control solutions with similar osmolality and pH to the most concentrated amino acid cocktail feed were added at the same volume. In addition, a control was included to confirm the impact of basal media dilution.
Cell culture samples were taken on alternate days up to and including day 14 for offline analysis of VCD, cell viability, cell diameter, pH, pCO2, extracellular metabolites (lactate, ammonia, glutamate, glutamine, and glucose), mAb titer, and amino acids. Viable cell density, viability, and cell diameter were measured using the trypan blue exclusion method on a CEDEX HiRes cell counter (Roche Diagnostics GmbH, Mannheim, Germany). The ABL80 blood gas analyzer (Radiometer, Denmark) was used to capture offline pH and pCO2 values. The extracellular metabolites were measured using the RX IMOLA Clinical Chemistry Tester (Randox, UK). Antibody titer was quantified using a previously described protein-A reverse-phase high-performance liquid chromatography (HPLC) method (Hoshan et al., 2019). Amino acids were measured using Waters Acquity ultra‐performance liquid chromatography (UPLC) H-Class Bio System with an AccQ‐Tag Ultra Derivatization Kit (Waters, MA). The amino acid standard kit (A9781) was obtained from Millipore Sigma (St. Louis, MO).