Accelerating the scale up of adeno-associated virus (AAV) manufacture is highly desirable to meet the increased demand for gene therapies. However, the development of bioprocesses for AAV gene therapies remains time-consuming and challenging. The quality by design (QbD) approach ensures bioprocess designs that meet the desired product quality and safety profile. Rapid stress tests, developability screens, and scale-down technologies have the potential to streamline AAV product and manufacturing bioprocess development within the QbD framework.

The Rapid Emergence of AAV-Based Gene Therapies

The therapeutic effects of gene therapy are achieved via the manipulation of a genetic sequence or the delivery of genetic materials. The delivery of genetic materials often requires their incorporation into viral or nonviral gene transfer vectors such as lipid nanoparticles, adenovirus, or AAV. AAV, a member of the parvovirus family, is a popular viral vector used extensively in gene therapies and vaccines due to its wide cell tropism and low immunogenicity. The genome of wild-type AAV comprises three genes, rep, cap, and aap, encoding, respectively, key viral replication proteins, capsid proteins (VP1, VP2, and VP3), and the assembly-activating protein.

Fig. 1 Structure and genome of adeno-associated virus (AAV) serotype 2.

To date, over 150 AAV genotypes and 13 serotypes have been discovered in humans (AAV1, AAV2, AAV3, AAV5, AAV6, and AAV9) and nonhuman primates, while more have been identified in other species. They vary in cell tropism and transduction efficiency, leading to different functionalities.

Tools to Accelerate Bioprocess Development

The demand for AAV is growing rapidly as clinical pipelines expand, making the need for a platform large-scale AAV production process more urgent than ever. Their manufacture is complex, requiring cell expansion, transfection, cell culture, virus harvest, and multiple purification and buffer exchange steps. The development and scale up of manufacturing processes for AAV is highly time-consuming and risky due to their inherent instability and complex biology and the relatively few long-term clinical data that can inform the ideal product profile to achieve. Their development also builds on relatively few precedents and lacks the benefits of technical maturity in AAV-specific bioprocesses, cell lines, and analytical instrumentation that are found with more established platforms such as antibodies. These challenges will continue to evolve for many years as future developments increase the process intensity of individual AAV manufacture steps.

Several experimental tools are anticipated to be able to accelerate AAV product formulation and early-stage process development. These build on their well-established success in the development of monoclonal antibodies (mAbs) and include, in approximate order of implementation:

• Developability screening of product candidates;
• Scale-down and USD tools for rapid bioprocess development
• Accelerated (forced) degradation screening for rapid formulation.

Each of these tools is designed to be used at their respective development stages to accelerate the selection of manufacturable product candidates, to optimise the manufacturing process, or for the rapid development of stable dosage formulations, although learning in one can also be used to inform other stages of development.

Fig. 2 Workflow of the quality by design (QbD) framework.

Scale-Down and USD Tools for AAV Manufacturing Bioprocess Development

Scaled-down platforms use the same process unit operations as pilot- to large-scale manufacture and with sufficiently tight control to achieve high reproducibility, but with much smaller operating volumes that can often be run in parallel to rapidly explore and optimise process variables. Early pilot-scale optimisation of manufacturing processes strongly informs the acceptable design space for CMAs and CPPs as well as potential control strategies to maintain them in the design space. Therefore, scaled-down platforms can extend this to a broader exploration of bioprocess operating parameters, giving a deeper understanding.

This approach is now well established for mAb bioprocess development, and some aspects can be readily transferred to AAV despite the significant difference between these products. For example, a scaled-down approach using 24 deep-square well plates has already been used to optimise the production of lentiviral vectors in suspension culture and so has good potential to succeed similarly for AAV production. Other powerful scaled-down systems, such as microscale chromatography columns and microlitre batch incubation, have also been widely used as high-throughput screening methods for chromatographic conditions and resin selection.

In contrast to scale down, USD tools do not replicate a process operating at a small scale but are instead designed to precisely mimic the critical conditions and exposures encountered in a large-scale process unit of operation using only millilitres of samples. They can be used to optimise products, solution conditions, or bioprocess parameters and define the CPP design space for each unit of operation. Fill/finish and ultrafiltration/diafiltration (UF/DF) steps are often found to be major causes of functional AAV titre loss, and so USD could potentially be used to identify the causes of these losses. Such losses are proposed to derive from shear-induced virus aggregation during the concentration steps. To date, USD models have successfully simulated large-scale bioprocess conditions for numerous biologics, including plasmid DNA, mAbs, adenovirus, and cells.

Comparison Between Stress Tests for mAbs and AAV

Stress tests play an important role in determining product stability under process, storage, shipping, and in-use conditions. Stress tests apply moderate (for manufacture developability) or extreme (for formulation development) conditions to biologics that can be grouped into physical stress (e.g., temperature, light, mechanical shear, surface adsorption) and chemical stress (e.g., pH, denaturants, oxidation, reduction). Through years of experimentation, stress tests used for the development of mAbs have become relatively standardised through industry consensus around established product and manufacturing platforms. However, mAb stress conditions do not translate simply to AAV products and so these still need to be investigated and refined. A standardised set of stress tests, perhaps tuned to certain serotypes, would facilitate platform-based process and formulation development for AAV products.

Fig. 3 A comparison between the conventional monoclonal antibody (mAb) and adeno-associated virus (AAV) production processes.

Current Challenges in Analytical Method Development for Large-Scale AAV Manufacture

During AAV production, their CQAs, such as the potency, identity, impurity, and stability of the viruses, are monitored to ensure product safety and efficacy in clinical applications. Each of these qualities is attributed to a different property, such as infectious titre, viral genome titre, genome identity, capsid stoichiometry, or aggregation. Any changes in these properties could undermine the efficacy of the product or lead to serious safety concerns.

Current analytics for AAV characterisation and monitoring were largely designed and optimised for smaller biologics, leading to various insufficiencies with AAV. The adaptation of existing analytics to AAV products is particularly challenging due to their large molecular weight, their oligomeric complexity, the mix of protein and ssDNA, and the low concentrations of viral particles compared with typical mAb concentrations during bioprocessing. Although the current techniques are sufficient to analyse AAV CQAs during production, they are not always highly accurate, are often time-consuming, and are not available for real-time monitoring. Therefore, more robust analytical methods are needed for almost all AAV CQAs, with higher throughput, greater sensitivity and dynamic range for typical AAV titres, and ideally the ability to monitor product CQAs in real time.

Concluding Remarks

Process development challenges prevent the establishment of a robust large-scale AAV production platform. To meet these challenges, it will be necessary to establish validated standards, analytics, product CQAs, and suitable developability screens. To achieve this, attempts should be made to better understand what are the essential product CQAs and to anticipate which of the other potential CQAs are likely to arise as we learn more about products in the clinic. Analytical methods should also be further improved to enable faster analysis on smaller samples, as the gradual shift from "potential" to "actual" CQAs will bring a greater analytical burden.