This article was originally published by Phacilitate on August 24, 2022.
When scaling up production of any material, important questions come under consideration. For this blog, based on a presentation delivered at Advanced Therapies Week 2022, the Phacilitate team explores a possible solution for a technology capable of assisting the scale up of AAV production.
An efficient AAV production scale-up enables rapid generation of high throughput AI-designed AAV libraries to accelerate gene therapy development.
In this blog, we discuss how Dyno Therapeutics is leveraging Corning’s Ascent™ Fixed Bed Bioreactor system to efficiently scale-up the AAV production that is integral to their novel AI-driven development of improved capsid libraries.
Corning’s Bioreactor in Practice: Dyno Therapeutics
Dyno Therapeutics is pioneering the use of AI to optimize the engineering of novel AAV capsid have improved targeting ability, payload size, immune evasion, and manufacturability.
To discover improved capsids, Dyno’s internal research loop starts with the Machine Learning-driven design libraries consisting of hundreds of thousands of novel capsid sequences to test. After producing these variants, DNA barcode tracking enables for the high-throughput measurement of capsid properties in vitro and in vivo. These measurements help generate an increasingly large dataset that Dyno uses to train its machine learning models to predict even better AAV capsid variants over time to enable improved gene therapies.
At these high-throughput scales, Dyno’s method of AAV production requires scalability, consistency, control and speed. They need to make AAV in greater quantity and quality, while shortening cycle time, to test the capsids quicker, capture more data and learn more about the landscape.
Dyno decided to investigate the benefits of using Corning’s Ascent FBR PD system with a 5m2 bioreactor in their AI AAV capsid engineering platform.
Investigating the Bioreactor
Could this new technology produce Dyno’s AAV libraries with quality while improving speed and efficiency sufficiently to warrant replacing their current platform, Corning HYPERFlask® vessels?
Dyno found that in its initial test the HYPERFlask vessel performed slightly more efficiently than the Ascent FBR system on a production per cm2 basis, but as they became more familiar with the new technology over subsequent runs, performance equalized.
Having worked with HYPERFlasks for years, scientists at Dyno hope to achieve faster process optimization with the Ascent FBR system given these early results.
What Does the Bioreactor Offer?
Conclusions as to the benefit of using the fixed bed bioreactor technology drawn from the study were as follows:
- More work could be completed with fewer staff. It took 1 person half a week to run the bioreactor compared to 3 people over 1 week to run the same amount of HYPERFlasks. This allows Dyno’s team to be more time efficient.
- The bioreactor provided rich data. After ‘plugging in’, the data can be pulled and monitored continuously in the lab or remotely. Several monitors are attached to the bioreactors creating continuous metabolite readings, as well as controlling the onboard pH and d0. This also means staff handovers can be automated, allowing for greater flexibility. Such close process control and feedback are not available when using HYPERFlasks.
- The Ascent FBR produced Dyno’s libraries with enough yield. This is very important for pipelines.
- Using bioreactors will likely shorten overall production timelines. Running a large unit can really accelerate the product pipeline, providing more and more robust data to push the AI engine, and design and discover variants faster, to deliver new capsids for a profile faster, to allow for a gene therapy to actually be delivered to the target cells.
If we consider making a volume of output using the most basic method of AAV production – in a petri dish – it could be estimated to take 10–50 weeks from the start of production for data to be produced, and this will be via manual collection. Also, every single step multiplied by every single plate creates a chance for human error.
The Corning HYPERFlask vessel is a significant upgrade to a standard petri dish or flask for smaller scale work or seed train, offering a more than 10-fold increase in area per unit. However, data is still recorded manually, so there is room for human error and production is time consuming.
When using the Ascent FBR, the number of units collapses, and the time scale is reduced. Fewer units mean fewer opportunities for mistakes. The automated data captured is robust and incredibly valuable
The Future
It is easy to justify that automated AAV production has a viable future.
“The faster we can run our cycle time, the faster we can design new capsids, and the less someone who is making the gene has to worry,” comments Chris Reardon, Associate Scientist at Dyno Therapeutics.
He adds that the efficiency and accuracy of Corning’s bioreactor could translate to “the less someone who manufactures has to worry about off-target effects, and they may be able to manufacture a thousand-fold less because it targets much better.”
Dyno’s Reardon is looking forward to evaluating the Ascent FBR Pilot system, which scales up to a 100 m2 bioreactor. The larger platform may enable Dyno to consolidate production into one run, reducing production time and opportunities for mistakes, while producing a higher yield and more robust data.
As the industry demands more from scaling AAV production, it is certainly a very exciting time for AAV and gene therapy design.