Optimization of Processes and Advanced Platforms for Viral Vector Processing

Viral vectors are synonymous with gene therapies, so their development, production, and processing are of upmost importance to all gene therapy researchers and manufacturers. Every year, I look forward to attending the Cell and Gene Bioprocessing and Commercialization conference in Boston and talking to leaders in industry and academia about their current approaches to advancing gene therapies. Like most other meetings this year, the conference was entirely online and had to provide a shortened agenda. Nonetheless, there was no shortage of interesting presentations, plenaries, speaker panels, and even live virtual laboratory tours. During 19–22 October, two of the five conference tracks focused on the progress and current concerns for ex vivo and in vivo gene therapies. Below, I highlight a few of my favorite presentations and speaker panels (by topic). They focused on current steps toward implementing new technologies and strategies for reducing timelines and costs and improving existing methods. Many presenters talked about high-throughput applications for sequencing and screening, optimizing production platforms for increasing yields, improving transfection methods, or enabling analytics for characterizing and identifying lentiviral and adenoassociated viral (AAV) vectors and capsid particles.

Viral Vector Production
Nicole Nuñez (Eureka Therapeutics) presented “Optimizing Viral Vector Production.” She provided an overview of her company’s transgene delivery system and lentiviral manufacturing strategy. The company optimized its plasmid design, lentiviral manufacturing process, and quality control assays to enable a scalable, high-titer vector-production system. Some companies use a p24 enzyme-linked immunosorbent assay (ELISA) physical titer for measuring protein (a method that does not provide an indication of vector performance). But Eureka Therapeutics developed a lentiviral vector functional-titer assay to measure protein expression and to evaluate viable lentiviral particles, batch-to-batch consistency, and potency. Nuñez highlighted Eureka’s other optimization decisions, including adapting human embryonic kidney (HEK) 293T cells to suspension culture in serum-free media to ease scale-up. A “third-generation lentiviral vector plasmid system” is used to separate genetic elements onto three accessory plasmid constructs, with the transgene located on a fourth separate transfer vector. The system also has a deletion in the long-terminal repeat sequence to ensure self-activation. In her presentation, Nuñez also emphasized the importance of the transfection step. Factors to consider include choice of reagent, incubation time and media, cell density during transfection, and the ratio of DNA to polyethylenimine (PEI). Finally, she highlighted the regulatory landscape of lentiviral vectors and compared her company’s manufacturing process with that of other developers in terms of final product functional titer and number of patients that could be treated.

Improving the production step also was a subject of two speaker panels, both titled, “High-Throughput Applications in Production Platforms and Enabling Manufacturing Technology.” The first panel was led by Rachel Legmann (Pall) and included speakers Sam Wadsworth (Ultragenyx), Thomas Krell (Takeda), and Jayanthi Grebin (Colder Products Company). One critical step is to develop a highly productive cell line. Speakers were asked what they recommend for expediting this step.

Wadsworth suggested beginning early. “You develop a paradigm for cell-line screening typically based on screening for high yield and follow up quickly with other critical quality attributes for the product. The only way to speed up the process is to develop the screening method first [in a way that is] efficient, rapid, and accurate.”

Another question related to ensuring viral control to prevent contamination. Panelists were asked to comment on the main challenges for gene therapy — compared with those for monoclonal antibodies (MAbs) — when validating a viral clearance filter panel for potential sources of adventitious viruses. Krell said, “There is no major challenge in the implementation of virus filters for gene therapy manufacturing if you chose a feasible option for that. For different gene therapy vectors, an upstream viral barrier is entirely doable. It’s actually easier than [it is] for MAbs because the volumes [of viral vectors] typically are much smaller. And for downstream, there are midsize nanofilters that allow for the passage of these vectors and still retain adventitious viruses.”

But viral clearance for lentiviral vectors could be more difficult than for other viral vectors. Grebin pointed out that lentiviruses (associated with ex vivo therapies) are more challenging to purify because they are bigger than in vivo viruses such as AAV. So a 0.2-µm filter might not be usable with lentivectors. “I think our hands are tied in the manner of what viruses we use. There are challenges in the viruses you use based on the type of treatment that you want to do for your process [in vivo or ex vivo].” Krell added, “There is a complication with every manipulation that needs to occur before a treatment is brought back to a patient. If you do the production right, you can use closed systems that [do not require] interference with operators. But with an ex vivo system, there is more exposure [and] more manipulation that also is going to provide more opportunities for contamination events.”

On a follow-up question regarding viral contamination precautions for cell and gene therapies, Krell said, “We need to strike a careful balance in making those treatments available, and at the same time making sure they come within an adequate safety margin. Essentially, the technology is there; we have a toolbox to prevent [contamination].” Wadsworth brought up that one common source of contamination is serum, which can be removed from a process as early as the cell-banking step. Single-use components also offer a major benefit. Grebin suggested that manufacturers evaluate their whole processes when it comes to preventing contamination, not just downstream steps. “What you’re doing upstream [to prevent contamination] is going to improve your process downstream. In some cases, you won’t have a good method for filtering impurities and contamination, so the upstream platform you use and how well you are getting your validation done will affect contamination. ”

The second speaker panel of the same title was led by Tom Broughton (Informa Connect) and included speakers Nicole Nuñez, Sha Sha (Center for Biomedical Innovation, Massachusetts Institute of Technology), and Aleš Štrancar (BIA Separations). One question posed to the panelists concerned the most critical step for producing viral vectors with high titers. Nuñez said that transfection is most important, both for lentiviral and AAV processes. She said that biomanufacturers “have to think about the plasmid ratios [they are] using, the cell densities, different types of media, and the transfection reagents. We spend a considerable amount of time optimizing that step. Once you nail down an efficient and robust transfection step, that can help you with purification and ensuring you have high starting yields.”

She added that one challenge arises with the amount of time and effort required to optimize a process. “Many parameters must be investigated, and that investigation must be repeated for each product type. In academia, we try to consider whether any approach can be useful to tackle those challenges other than spending time with empirical optimization. One approach is to determine how those parameters interact mechanistically and affect how viral vectors are produced in cells. There are variables in cells that we cannot capture completely, but we hope to get the core of those variables validated.” Other questions pertained to regulatory acceptance, the bottlenecks of AAV manufacturing (e.g., empty capsids, low yield, high costs, and the need for flexible platforms for scalability), and technologies that can improve processing (two in particular: miniaturized bioreactors based on microfluidics and on-line sensors).

Viral Vector Development and Manufacturing
Sam Wadsworth presented “Next-Generation Cell Line Development for AAV Gene Therapy Products.” He highlighted the features of his company’s gene therapy manufacturing processes based on HEK293 suspension/plasmid transfection and HeLa producer cell line (PCL) suspension/adenovirus helper platforms. In collaboration with Bayer, the company has conducted engineering improvements to its HeLa PCL platform. Specifically, the company made molecular changes to the plasmid components and multiple process and screening changes to the original “HeLa 1.0” platform. The improved “HeLa 2.0” stage has a 10× increase in yield and a streamlined clonal selection. The technology now is used for the company’s work in finding a gene therapy for Wilson’s disease. Wadsworth showed how further improvements to the platform (“HeLa 3.0”) could be used to increase production of AAV vector yield and how the use of a siRNA knockdown of specific genes could improve AAV production titer. He highlighted work on Wilson’s disease potency assays and the use of clustered regularly interspaced short palindromic repeats (CRISPR) and CRISPR-associated protein 9 (Cas 9) to delete ATP7B (the loss of which causes copper toxicity in Wilson’s disease) in the HepG2 cell line.

Philip Lee (Senti Biosciences) presented “Gene-Modified Allogeneic Mesenchymal Stromal Cells (MSCs) and Natural Killer (NK) Cells.” He introduced his company’s “gene circuit” platform, which is a genetic “logic system” created by a method of combining DNA. Lee also discussed a case study focused on the use of engineered MSCs to treat solid tumors through lentiviral vector transduction. The cells were derived from bone marrow and engineered to express two immune-stimulated payloads (interleukins 12 and 21). Data from preclinical trials showed that the combination of those payloads is highly potent and that it yields a robust antitumor response across every tumor model evaluated. Lee presented his company’s strategy for transferring its genetic modification process to good manufacturing practice (GMP) scales. He also showed how the MSC platform can be expanded to other gene-modified cell therapies. The second study focused on the early stage development of an allogeneic natural killer (NK) cell treatment for cancer. Lee provided an overview of the Senti Biosciences method for robust cell isolation and cryopreservation and emphasized the company’s continued work in optimizing its viral vector process and NK cell transduction and expansion strategies.

Sha Sha presented “Advancing Recombinant Adenoassociated Virus (rAAV) Manufacturing By Mechanistic Insights and Modeling.” Currently, the viral vector processing industry is having difficulty in trying to meet AAV production-scale demands, given that approved in vivo gene therapies so far require from 1011 to 1015 vector genomes per dose (larger amounts for whole-body treatments). Production demand is even greater if such products are meant for large patient populations. Most rAAV production methods generate defective capsids. An MIT team and its collaborators are developing a continuous viral vector manufacturing process based on mechanistic modeling and novel process analytics. The goal is to develop a first-principles mathematical model for rAAV design.

“We expect that analyses using mechanistic models can help us understand the trajectory and bottlenecks in rAAV production, which can help us to get information that we cannot get from empirical experiments for improving our capability of process designs.” Sha discussed the triple-plasmid transfection method, which is the most used and most flexible transfection approach. She described the steps used to build the mechanistic model and demonstrated how the model captured the dynamics of the plasmid content in cells. Sha also showed how the model revealed the activities along the AAV production pathways and how such a mechanistic model can be used to configure a continuous “assembly line” of AAV production in which plasmids and media are fed continuously into a process.

Analytics
This year’s conference included several presentation on characterizing viral vectors and on identifying and quantifying empty and full capsid particles. Franz Schnetzinger (Gyroscope Therapeutics) presented “The Case for rAAV Titration By HPLC.” His company develops treatments for age-related macular degeneration (AMD) by induced expression of complement factor 1. He discussed an AAV vector study on enumerating genomic capsid titers. Traditional and well-established methods for achieving that include the use of quantitative polymerase chain reaction (qPCR), digital PCR (dPCR), and ELISA. But as Schnetzinger explained, such methods “suffer from inherent limitations, including extensive sample preparation.” His study was performed to determine whether high-performance liquid chromatography (HPLC) methods could serve as an alternative.

Results showed HPLC with ion-exchange (IEX) chromatography and with size-exclusion chromatography (SEC) assays showed good correlation with PCR- and ELISA-based methods. HPLC-IEX also enabled genomic titer determination and estimations of the ratios of full to empty particles. In his study, IEX was used as a screening method for spotting outliers in such ratios. Schnetziner showed results of HPLC-SEC capsid titer determination studies for sample stability and linearity. He demonstrated that up to 70 samples were quantified accurately and precisely and that the method was robust. And because HPLC has fewer sample-preparation steps than traditional methods require, operator variability is minimized. However, Schnetziner pointed out that HPLC-IEX methods are best used for pure samples because results can be confounded by impurities. He concluded that the method is limited as an in-process screening tool rather than for quality control: “The relative quantification requires accurate reference and is the most limiting factor of all.”

Michael J. DiBiasio-White (Ring Therapeutics) presented “Emerging Upstream Bioprocessing and Analytical Tools for Vector Manufacturing.” He pointed out some key bottlenecks in process development: scale-up (or scale-out), transfection efficiency, plasmid requirements, process development studies for transfer to bioreactors, and determining which analytics are needed for those studies. Adherent culture systems for vector production typically have limited scalability, poor yield, and high variability between batches. The ideal production system is a stable producer cell line. “A lot of the field is working on it, and it’s something that I think will be the next technological advancement in viral vector production and manufacturing.” Stable producer cell line systems are a “gold standard of where we want to be.”

DiBiasio-White said that the industry is working on technology improvements and described what an AAV producer cell line could look like. In particular, the system would consist of an AAV2rep and AAV2cap stably encoded within a baby hamster kidney (BHK) or HeLa cell with a transgene, and the introduction of a helper virus would activate vector production. Alternatively, a baculovirus-infection–based system can be used. Both platforms overcome the limitations of an adherent culture system because they are easily scalable, require few plasmids or none, and could generate vector yield to meet commercial demands.

DiBiaso-White highlighted advancements in analytics for process development and said that the industry is moving into the era of high-throughput process development with optimized media and feed, use of perfusion-based systems, process control, and on-line or in-line analytics. “Understanding how upstream process changes can influence vector properties is critical,” he said. And he predicted that “future viral vector manufacturing will move toward automation and process control strategies. Stricter process control over process changes and the ability to correct in real time will add value.”

David Dobnik (National Institute of Biology, Slovenia) presented “Exposing the Content of Different AAV Fractions After Ultracentrifugation.” His case study used different fractions of AAV with green fluorescent protein (GFP) reporter transgene after ultracentrifugation: two types of viral genome (self-complementary and single-stranded AAV) and four fractions from CsCl ultracentrifugation (empty, intermediate, full, and heavy particles). PCR and droplet-digital PCR (ddPCR) analysis showed that all fractions contained encapsidated vector genomes, but there were significantly lower quantities in the “heavy” and “empty” fractions. Transmission electron microscopy (TEM) and capsid ELISAs were used to determine the percentage of particles in each fraction. Nontargeted high-throughput sequencing (HTS) was used to characterize vector samples: one method for viral genome sequencing and another for sequencing whole-vector constructs. An HTS approach also was performed to evaluate genomes and identify nucleic acid impurities. Finally, Dobnik discussed optimized single-molecule sequencing technologies for additional characterization information.

Qin Zou (Pfizer) presented “Holistic Characterization of AAV Capsid Particle Using Advanced Analytical Tools.” He pointed out that AAV2 capsids are much more complex than IgG antibodies because the capsids are much heavier and larger. Particle content (empty, partial, or full) is an important physicochemical property of viral capsid particles. Zou covered analytical methods for examining AAV capsids and how they behave in solution. One exciting tool is analytical ultracentrifugation (AUC), which can be implemented in different ways. Zou provided details about two applications: boundary sedimentation velocity and isopycnic density-gradient equilibrium. He showed results of multiwavelength UV detection, which can be used to identify empty, partial, and full capsids and to provide “direct and absolute quantification based on extinction coefficients.” Another study involved the use of an isopycnic centrifugation equilibrium method, which separated particles according to their density, and multiwavelength analysis provided identification of different capsids. Results showed higher resolution of full and empty capsids than can be discerned with traditional methods. Using sedimentation velocity (SV) AUC and sedimentation equilibrium (SE) AUC results, researchers determined particle mass using the Svedberg equation. Zou also showed studies of DNA release of phosphorus-31 using nuclear magnetic resonance (NMR) analysis. Results showed that NMR provided direct detection of DNA, but the method was slow and not highly sensitive (encapsulated DNA is invisible to NMR). Finally, Zou showed results of using extrinsic fluorescence analysis of genome DNA release. They demonstrated that the method is “fast, easy, and sensitive but potentially depends on dye binding.”

Conference Postevent Report
For additional highlights, be sure to check out the Cell and Gene Bioprocessing and Commercialization postevent report, which will be posted on the BPI website at
https://bioprocessintl.com in December 2020.

Maribel Rios is managing editor at BioProcess International, a part of Informa Connect; [email protected]