Inefficiencies that limit the global supply of viral vectors are negatively impacting the gene therapy industry, according to researchers, who say there is an urgent need for better downstream purification materials.
The call came from Stefano Menegatti, PhD, professor, Department of Chemical and Biomolecular Engineering at North Carolina State University, who says the limited availability of viral vectors has become a major bottleneck.
“Gene therapy is one of the most transformative frontiers in modern medicine. It holds the promise of curing devastating diseases with a single treatment. Adeno-associated viruses (AAVs) are the leading delivery vehicles for these therapies.
“Yet for all the excitement in the field, manufacturing remains a critical bottleneck: producing AAVs at the purity, potency, and scale needed for clinical use is enormously challenging,” he tells GEN.
A key obstacle is that current purification materials—usually resin-based affinity adsorbents—cannot distinguish between full AAV capsids that carry genetic material and empty capsids that have no payload.
Another issue with current purification technologies is the need to use low flow rates, which increases processing costs and, ultimately, gene therapy prices, Menegatti says.
“Purification technologies operate at slow flow rates, require harsh chemical conditions that can damage the product, and wear out quickly, all of which drive up manufacturing time and cost.
“For patients waiting on life-saving treatments, these are not abstract engineering problems. They translate directly into delayed access and higher prices,” he says.
Purification materials research
In May, Menegatti and colleague Michael Daniele, PhD, were awarded a NIIMBL grant to further develop a purification method with novel materials that they claim can differentiate between full and empty capsids.
“At the heart of it are AvXcel affinity adsorbents—developed by ChromaGenix—which substantially accelerate purification. More importantly, they selectively enrich full, gene-loaded AAV capsids directly at the capture step.
“Our preliminary data show that the fraction of full capsids increases from roughly 20−30% in the raw material to 34−48% in the affinity eluate, far outperforming the industry benchmark,” he says.
In addition, the membranes can also withstand harsh cleaning processes—up to 50 cycles with caustic solutions—which is in line with industry needs, Menegatti adds.
“The combination of speed, selectivity, and durability is genuinely unprecedented. We are not making incremental improvements; we are redesigning the purification step from the ground up, with the goal of transforming a months-long process development campaign into one that takes weeks.”
Machine learning
The NIIMBL grant will also support the ongoing development of a machine learning-based analytical software platform, called Beacon, designed to help manufacturers optimize vaccine purification.
Daniele tells GEN, “Even with next-generation purification materials, optimizing the process for each new AAV target remains a major challenge.
“Today, the standard approach is a trial-and-error methodology that requires dozens to hundreds of experiments to map out the right operating conditions for each new product. This must be repeated essentially from scratch for every new AAV serotype and transgene combination, which is both time-consuming and expensive.”
Beacon, or Bayesian-Enhanced AAV Chromatography Optimization Network, is a machine learning platform designed to help process developers avoid such repetition.
“It uses a Gaussian Process algorithm, a type of Bayesian AI that learns from each experiment and predicts the next most informative one to run. Rather than blindly sweeping through conditions, Beacon intelligently navigates the optimization landscape, reducing the experimental burden by 30−50% compared to conventional DOE while simultaneously optimizing multiple performance criteria: yield, full capsid enrichment, impurity clearance and productivity,” Daniele continues.
Another key Beacon feature is the use of North Carolina’s VVIRAL database, which comprises thousands of AAV purification experiments, to “warm-start” new campaigns.
Danielle tells GEN, “In other words, it doesn’t start from zero for each new product; it leverages everything we’ve already learned. And unlike a black-box AI, Beacon provides interpretable, quantitative outputs through SHAP analysis, so scientists understand why a particular protocol works, enabling smarter decisions and better risk management.
“Beacon will be released as an open-access, cloud-hosted platform to the broader biomanufacturing community, democratizing access to cutting-edge AI tools that were previously available only to a handful of computational specialists,” he adds.
