Implementation of Design of Experiments (DOE) for Optimization of Feeding Strategy and Glyco-Engineering of Trastuzumab Biosimilar

Fed-batch cell culture is the most commonly used process for antibody production in biopharmaceutical industries. Basal media, feed, feeding strategy and glycan structures are always among the most important concerns during process development and optimization. In this study, first, a traditional screening study was performed to identify the top media/feed combinations by evaluating the cell culture performance including cell growth and protein titre. Optimization of the process was also performed using response surface methodology in order to find the most optimum feeding strategy and glucose set point regarding final titre of the recombinant monoclonal antibody being produced in Chinese hamster ovary cell line. The focus of this study is not only on titre, but also on product quality and comparability especially protein glycosylation. The prediction model of product titre as a function of feeding percentage and glucose set point was successfully applied for the second set of experiments that was performed for glycan improvement. Statistical design of experiments was applied to determine the most important factors and their effects on galactosylated and afucosylated glycans. Uridine, manganese, galactose and fucosyltransferase inhibitor were chosen to evaluate if their presence can affect glycans and to obtain their best combination for fed-batch culture supplementation. We determined that 2.5 % daily feeding combined with maintaining the glucose set point on 2.5±0.2 g/L could achieve final titre of 2.5± 0.1 g/L. Galactosylation of antibody was increased about 25% using MnCl2 and galactose while afucosylation was increased about 8% in presence of fucosyltransferase inhibitor. Galactose and Mn2+ led to a shift from G0F to G1F and presence of Fucosyltransferase inhibitor caused to an increase in G0 compared to its absence. These results demonstrated that supplementation of culture with all these components can provide exact control of antibody galactosylation and fucosylation with minimal impact on culture characteristics and product quality attributes. Subsequently, validation experiments were also carried out in 5L STR bioreactors which showed that similar results could be achieved in bioreactors compared to shake flasks regarding both titre and quality.

115 cells/ml for at least 6 passages. Then fed-batch cultures were started with initial working volume 116 of 100 ml in 500 ml shake flasks with the same initial cell density. Cells were incubated in an 117 incubator with 5% CO 2 at 37 °C on orbital shaker with 100 rpm shaking speed and 30 mm orbital 118 shaking diameter. In all culture conditions feeding was started on 2 nd day of culture with a daily 119 bolus addition according to Table 1. Batches were harvested on either day 15 or when viability 120 dropped below 70%, which one occurred earlier.  125 In a typical fed-batch culture, the basal medium usually supports cell growth, while the 126 feed medium contributes in increasing the productivity. After determining the top basal media/feed 127 combination, the next development phase was started to improve the feeding strategy in order to 128 fine tune the batch duration combined with final expression. To optimize the feeding strategy, two     253 One of the most efficient feeding strategies is glucose-based feeding. This strategy 254 although can help to maintain glucose levels during the batch but can be problematic due to 255 accumulation or depletion of certain amino acids (12, 15, 16). Meanwhile once-daily feeding 256 strategy cannot ensure that glucose level is being held near the set points pre-defined (13). Based 257 on these results, the decision was made to use fixed feeding strategy while increasing the total 258 amount of daily feed added to the culture. Simultaneously, glucose stock solution of 400 g/L was 259 also used to bring glucose concentration to a level that prevents depletion before the next bolus  (Table 2). In addition, glucose was fed daily according to the glucose concentration at 268 the moment of feeding and the anticipated glucose uptake rate for the next 24 h. By this method 269 the glucose was held within a pre-determined set point irrespective of the amount of feed added 270 daily. Table 4 shows the final titre of the 13 generated experimental runs after execution.

271
The results showed that the max VCC (viable cell count) of 16.5 ± 0.5 ×10 6 cells/ As can be seen in Table 4, increasing the feed percentage from 2.5% to 3% when glucose 309 set point is on 2 g/L had no significant effect on titre. However in the same scenario when glucose 310 set point was adjusted on 3 g/L, final titre was decreased. Lower titre may be due to the 311 accumulation of specific amino acids which showed inhibitory effect on final productivity.

312
The results from this experiment shed light in the feeding strategy based on glucose set      Table 8. The verification process generated experimental results close to the predicted response 473 (p-value < 0.05) as represented in Table 9. Our optimized feeding strategy combined with glucose, 474 galactose, MnCl 2 and FTI supplementation resulted in mean expression of 2510 ± 65 mg/L.
475 Simultaneously the level of galactosylation and afucosylation were also increased to 32.5 ± 0.6 % 476 and 7.3 ± 0.5 %, respectively which were completely comparable to Herceptin originator drug.
477 The good agreement between the experimental and predicted results verifies the validity of the 478 models. 479