Canada-based researchers have investigated the impact of transferring a COVID-19 vaccine manufacturing process between two similar-sized bioreactors. Juan Sebastian Reyes Davila, a PhD student from the Polytechnique Montreal, was tasked with transferring a fed-batch process for manufacturing of SARS-CoV-2 spike proteins between bioreactors.
The aim was to support Canada’s future pandemic response by ensuring universities, as well as national centers, could contribute to the scale up of vaccine production.
“[The idea was] to exchange information, so we’re all on the same page and could work on cells right away,” explains Davila, who is due to present at BioProcess International in Boston.
Although the bioreactors were all approximately one liter and benchtop scale, he explains that the process transfer was “not trivial.” The National Research Council of Canada used a Multifors bioreactor whereas Polytechnique Montreal had a Dasgip multisystem bioreactor.
According to Davila, each manufacturer’s bioreactor had a different impeller for stirring the cells, which meant the strategy for aerating the cells needed to be different for each bioreactor. To understand what changes needed to be made, Davila used machine learning to study how different variables, such as pH and lactate, affected the yield of spike proteins.
He discovered that high levels of lactate led to lower yields and that slower feeding strategies helped to combat it.
“What we learned from our production process was that, to control lactate metabolism, if you were going to feed 100 milliliters [of enriched nutrient supplement to the cells] between Monday and Wednesday, you should feed it over 48 hours at a very slow rate,” he explains.
He believes this finding may be useful to industry. Other findings of his work include a model to predict how the cells would perform over time. This, he says, could be used in industry to automatically monitor oxygen uptake and adjust feeding of the cells to keep them healthy.
