Building Successful Viral Vector Production Processes

Viral vectors are key delivery tools for advanced therapies, enabling gene transfer to treat inherited disorders, cancer and other serious diseases. Despite rapid clinical progress, manufacturing remains a bottleneck due to cost, scale limitations and process variability. As the viral vector field matures, focus is shifting from early clinical supply toward scalable, robust and cost-effective commercial production. Achieving this requires coordinated optimization of upstream and downstream processes, supported by robust analytics and scalable process design.

Cell Line Considerations

The choice of a producer cell line (mammalian vs. insect) and culture format (adherent vs. suspension) underlies any viral vector production system. Insect cells were initially favored for Adeno-Associated Virus (AAV) production due to scalability, cost-effectiveness and perceived advantages in product quality. More recently, advances in process optimization and successful scale up have addressed concerns around mammalian HEK293 systems, shifting focus back to this cell line for its now-demonstrated flexibility, scalability, robustness and product quality.

For HEK293 systems, several factors drive selection. First, suspension cultures are generally preferred for scalability and compatibility with chemically defined, serum-free media. Second, monoclonal lines with documented single-cell origins are increasingly favored over polyclonal populations for consistent, high-titer production. Third, while HEK293T cells have been used clinically, concerns about the T antigen are shifting preference toward lines without the T antigen.

Cell Banking   

Large-scale cGMP production of HEK293 cell banks is critical but often overlooked. Reliable post-thaw recovery is essential to maintain production timelines, yet HEK293 cells are more sensitive to banking conditions than CHO cells, requiring tailored best practices to reduce risk, especially when making cell banks at larger scales.  Aside from the preparation of the cells themselves, large-scale freezing conditions, like controlled-rate freezing, must be optimized specifically for HEK293 cells, as standard protocols for other cell lines often do not translate effectively to HEK293 cells.  

PRO TIP:  Successful cell banking starts with healthy, log-phase cells grown under well-characterized culture conditions, such as optimal growth vessels, culture volumes, shaking speeds and shaking orbits.  In all instances, cells scaled up for cell bank production should grow identically in the larger-scale vessels as in the smaller shake flasks used for routine cell culture maintenance.  Any increase in doubling times or decrease in viability during the scale up process may lead to inferior cell performance.  Materials compatibility is also key: DMSO (5–10%) is commonly used for cryoprotection, but prolonged exposure at larger scales increases the risk of incompatibility. Glass pipettes should be used for transferring concentrated DMSO and DMSO-compatible containers like polypropylene for freeze media preparation. (can cite if needed: https://tools.thermofisher.com/content/sfs/brochures/D20480.pdf)  

Considerations for Obtaining Cell Density and Viability Assessments Post-Thaw

Accurate viability assessment immediately post-thaw can be challenging at low cell densities. At this stage, certain components found broadly in culture media may interact with trypan blue, forming trypan blue aggregates that may be miscounted as dead cells on some cell counters.  In instances where production batch records require cell viability post-thaw, this simple issue could put an entire production run at risk.   

PRO TIP:  Dilute a small sample of cells directly from the thawed freeze vial into phosphate buffered saline (PBS) solution before counting (e.g., 50 µL of cells from the freeze vial into 450 µL PBS) to prevent trypan blue aggregation and improve accuracy. Once cells recover and reach standard densities, this step is no longer necessary.

Scaling from Shake Flasks to Bioreactors

Overall, production cell lines should grow as well, or better, in large-scale bioreactors as in shake flask scale during the seed train and production phases. 

PRO TIP:  If cell growth is reduced or inconsistent compared to small-scale shake flask cultures, prioritize optimizing core bioreactor conditions like mixing and gassing first before attempting to address the issues with supplements that could affect the upstream or downstream processes in unknown ways, such as anti-clumping agents and surfactants. Optimizing cell health is essential for consistent transfection and high viral vector yield across scales.  

Optimization of Critical Quality Attributes

Using AAV as an example, once cell culture conditions are established, process optimization is key to achieving high yields and product quality. Development starts with DNA design—selecting optimal capsids and genomes, refining promoters, codon usage and minimizing sequence liabilities to ensure strong, safe expression.  Capsid choice is critical, as it determines tissue targeting and efficiency. Beyond natural serotypes, engineered capsids—via rational design, directed evolution and computational methods—expand performance options. After selecting DNA and capsid, production is optimized using design-of-experiments (DOE) to tune the amounts of DNA, transfection conditions and plasmid ratios. 

PRO TIP:  Because maximizing for titer alone can reduce product quality, optimization must balance multiple attributes, including titer, full capsid percentage, collateral DNA packaging and potency. To accelerate AAV development and identify optimal vector design and production conditions, use automated liquid handlers and statistical modeling in scale-down models to run large DOE studies, enabling evaluation of multiple variables simultaneously to define processes that maximize both titer and vector quality.

Plasmid DNA Technologies

High-quality plasmid DNA is foundational to robust viral vector production. Impurities, inaccurate concentration, poor topology or lot-to-lot variability can negatively affect transfection efficiency, yield and final product quality. Alternative DNA technologies such as nanoplasmids, doggybone DNA and minicircles are attracting interest because they may reduce bacterial backbone content, improve expression and lower the risk of unwanted sequence carryover. However, adoption depends on whether these benefits outweigh the added costs and operational and regulatory complexities compared to traditional plasmids. 

Indicators of a Successful Transfection

Successful transfection can be first detected by characteristic changes in cell growth and viability. Within the first 24 hours post-transfection, cells in suspension culture typically undergo one division, facilitating nuclear entry of plasmid DNA. This is often followed by a reduction in growth rate and a decline in viability, driven by the metabolic burden and cellular stress associated with transfection and viral replication processes. 

PRO TIP:  Depending on the system and production time, cell viabilities in the range of ~60–80% are commonly observed at harvest. Conversely, continued rapid cell division with minimal or no loss of cell viability is a strong indicator of poor or failed transfection.

PRO TIP:  A more direct assessment of transfection efficiency can be obtained by monitoring expression of the transfer gene in producer cells, when applicable. This is typically evaluated at early time points like 18 hours post-transfection to distinguish true transgene expression from later auto-transduction effects. Depending on the system, 60–80% of cells may be transgene-positive. However, the percentage of positive cells alone is not sufficient as the overall expression level across the population—such as mean fluorescence intensity (MFI) in fluorescent-based methods—is often a more representative indicator of transfection efficiency.

Considerations for Optimizing Plasmid DNA Complexation at Scale

Successful large-scale transfection requires understanding and careful control of key complexation parameters, including reagent stability, order of addition, mixing behavior and complexation kinetics. While these variables can be tightly managed at a small scale, larger volumes inherently introduce longer addition, incubation and mixing times. Because failures at scale are disproportionately costly, it is essential to assess process robustness by systematically evaluating extended holding times during each step of complex formation.  Small-scale studies should be performed to define the design space of critical steps—such as reagent dilution and DNA complexation.  

Mixing is a key consideration and becomes increasingly complex at larger volumes. Insufficient mixing can result in incomplete or heterogeneous complex formation, leading to variability in particle size and charge. Conversely, excessive mixing can either inhibit or accelerate complexation kinetics (depending on the reagent used), producing particles that are less efficiently taken up by cells. 

PRO TIP: Mock large-scale complexations can be performed using dyes to visualize mixing efficiency and identify any unexpected flow patterns that lead to incomplete plasmid DNA complexation. 

Scalable Purification: Large Scale AAV Purification

AAV purification involves seven key steps: harvest, clarification, concentration, affinity purification, full capsid enrichment, buffer exchange and fill–finish. During harvest, cells are lysed and treated with nucleases to release AAV particles and reduce residual DNA. The lysate is clarified to remove debris, typically via centrifugation or depth filtration, both of which are scalable through continuous or staged systems.

A concentration step is often performed before affinity capture to improve efficiency, given the high binding capacity of AAV resins. Tangential flow filtration (TFF) is widely used due to its scalability and recovery. Affinity chromatography serves as a robust primary purification step, with columns scalable across a range of sizes, while membrane-based methods offer faster processing but limited scalability.

Empty and full capsids are typically separated using chromatography, especially anion exchange resins, membranes or monoliths. Resins provide the greatest scalability, whereas membranes and monoliths enable higher flow rates but have format constraints. Finally, TFF enables buffer exchange and concentration, followed by sterile filtration and aseptic filling.

PRO TIP: The pre-affinity concentration step is mainly used to reduce loading time rather than being strictly required. With high-flow affinity resins and membrane-based capture technologies, clarified harvest can often be loaded directly onto the affinity step. However, feasibility should be evaluated based on feed volume, impurity load and capacity to ensure consistent performance at scale.

Large Scale Lentiviral Vector (LV) Purification

LV purification is more challenging than AAV, with approaches including ion exchange, size exclusion and affinity methods. Due to the vector’s fragility, some processes rely on minimal purification with clarification and concentration. Unlike AAV, lysis is not required since LV particles are secreted into the supernatant, though nuclease treatment is used to reduce residual DNA.

Clarification removes cells and debris, typically via depth filtration, sometimes preceded by low-speed centrifugation. Depth filters are favored for scalability and gentle handling. Because LV titers are relatively low, a concentration step is often needed, with TFF commonly used; however, shear and membrane conditions must be optimized to preserve infectivity.

When included, chromatography typically uses membrane- or monolith-based ion exchange methods, as high flow rates reduce the risk of irreversible binding and may eliminate the need for prior concentration. Size-exclusion chromatography can also be used, while affinity methods are emerging but not yet widely adopted. Buffer exchange and formulation are usually performed with TFF, followed by sterile filtration and aseptic filling with attention to filter compatibility.

PRO TIP: Due to LV’s large size and envelope sensitivity, measurable losses can occur during sterile filtration. In addition, LV is prone to degradation and loss of infectivity during extended hold times. To mitigate these losses, processes should be designed for continuous operation where possible, minimizing pauses or overnight holds. Sterile filtration is best limited to a single final step, with the addition of a large pore size prefilter to preserve vector yield and potency.

Analytics

Viral vector process development requires robust analytical assays to measure key quality attributes. For AAV, genome titer is assessed by quantitative PCR (qPCR), digital/droplet digital PCR (d/ddPCR), with d/ddPCR offering higher precision and qPCR enabling higher throughput. Empty-to-full capsid ratios are increasingly measured by mass photometry, while impurities like residual DNA are analyzed using qPCR or d/ddPCR or next-generation sequencing. Additional methods, like CE-SDS, LC-MS and SEC, assess capsid identity, purity and aggregation, alongside cell-based assays for functional infectivity, though these do not fully predict in vivo potency.

PRO TIP:  AAV qPCR or d/ddPCR titer assays typically include a nuclease step to remove non-encapsidated DNA. When working with crude lysates, media components can inhibit nuclease activity, leading to incomplete digestion and artificially inflated titers. To improve accuracy, dilute lysates (e.g., 1:10) in PBS + 0.001% Pluronic F-68 before the nuclease step to ensure more complete removal of plasmid DNA.

For LV, process development focuses on maximizing yield while preserving functional titer and minimizing impurities. As with AAV, consistent product quality depends on monitoring multiple attributes using robust analytical and cell-based assays. Historically, the predominance of ex vivo applications reduced pressure to develop highly discriminating assays. However, as in vivo use expands, improved analytics are needed to distinguish functional lentiviral particles from extracellular vesicles and enable more precise product characterization.

Viral vector manufacturing continues to evolve as the industry moves toward commercial scale. Consistent, high-quality production requires integrated process design, robust scale-up strategies and strong analytical control. Advances in automation, high-throughput tools and data-driven strategies are improving efficiency and reproducibility. Continued innovation will be essential to reduce costs and expand patient access, enabling scalable production of next-generation therapies.

Jonathan Zmuda, Ph.D. is a Senior Director of Cell Biology R&D at Thermo Fisher Scientific located in Frederick, MD (USA).  Jon leads a team of scientists dedicated to developing new products for protein expression, viral vector production and nucleic acid delivery.  Dr. Zmuda received his Ph.D. in Cell Biology from the University of Maryland, College Park and his BSc degree from Dickinson College in Carlisle, PA.

With contributions from Arjen Van den Berg, Emily Jackson-Holmes, Kenneth Thompson, and Nils Williston.