Researchers from MIT have developed a bioreactor simulator to predict adeno-associated virus (AAV) yields for insect cell/baculovirus manufacturing. Francesco Destro, PhD, a postdoctoral associate from the department of chemical engineering, created the digital twin for modeling the pathways inside cells as they manufacture AAV, hoping to reduce the future costs of gene therapy.
“These types of therapeutics address severe diseases, but they are complex biological products and can be expensive,” Destro explains. “Digital tools allow scientists to test processes on computer, reducing the experiments that need to be done physically, and better addressing bottlenecks.”
The model simulates the processes inside insect cells during intracellular AAV manufacturing by modeling pathways involved in baculovirus infection of the cells, through protein synthesis, through to AAV production. Manufacturers can input information on the baculovirus construct and promoter cassette that they plan to use for manufacturing. The model then predicts AAV yields, along with the number of empty AAV capsids, he says, adding that empty capsids are known to negatively affect the effectiveness of gene therapies.
Two projects
Details of the team’s first project, using the digital twin to predict AAV titers when manufacturing AAV in batches, has recently been published in Molecular Therapy Methods. The team has also worked on a second project, using the digital twin to model continuous processing.
Unlike with AAV manufacturing using plasmids and mammalian cells, insect cells infected with baculoviruses manufacture those viruses as well as the AAV product, he explains. This is a benefit in continuous processing, as the new baculoviruses can infect new cells. However, the baculovirus genomes slowly mutate as the process continues.
“The challenge [for continuous AAV manufacturing in insect cells] is that, as continuous manufacturing continues, there are so many defective viruses that the process eventually stops,” he points out.
The new digital twin models the genetic instability in the baculoviruses, and manufacturers will be able to use it to optimize their process towards higher AAV yield, according to Destro.
Going forward, the team plans to adapt the digital twin to model a range of small- and large-scale processes.
“We hope it will be a game changer because the small molecule and food industries have used digital twins to reduce costs,” he says.
