Scaling Up Viral Vector Manufacturing – Moving From the Lab to the Clinic

Dr Kathrin Schneider, Senior Vice President, Global Business Segments, Life Sciences, at Revvity. Dr Irene Ferreira, Director, Head of CGT Platform Innovation & Management, Life Sciences, at Revvity. Nikki Withers, Scientific Content Writer, Life Sciences, at Revvity. 

The adeno-associated virus (AAV) is a key viral vector vehicle extensively exploited in clinical gene therapy studies. Scaling up manufacturing of this therapeutic vector to amounts needed to treat patients is a prerequisite for a successful path to new gene therapies. Manufacturing starting materials, like producer cell lines and packaging plasmids, have been identified as influencers of Critical Quality Attributes (CQAs) for purity and potency. In addition, the starting materials can positively affect manufacturability parameters like productivity, which in turn decrease cost of goods for clinical development and the commercialised gene therapy.

This article explores how starting materials, specifically producer cell lines and the design of transfer plasmids, affect the quality attributes of AAV vectors (Figure 1).

Figure 1. The manufacturing process is the rAAV-based product. Starting materials (e.g., the producer cell line, the transfer plasmid) and the manufacturing process determine the rAAV quality attributes related, for example with potency and purity.

Influence of the Producer Cell Line on the rAAV Quality Attributes

Recombinant AAV vectors (rAAVs) are highly complex molecules produced by living cells through a multi-step manufacturing process. The two leading upstream approaches for producing rAAV vectors are transiently transfected human HEK293 cells and live baculovirus infection of Spodoptera frugiperda (Sf9) insect cells. Although both methods have been widely adopted, questions have been raised about the comparative quality and performance of the vectors produced using these different systems.

The most extensively used method for large-scale production of rAAV vectors involves the co-transfection of HEK293 cells with three plasmids. These plasmids contain either the AAV rep and cap genes, the cargo gene flanked by inverted terminal repeat (ITR) sequences, or adenoviral helper genes.

An alternative avenue involves infecting Sf9 insect cells using the baculovirus expression vector system (BEVS). This method utilises baculoviruses containing the AAV rep and cap genes and the transgene. Baculovirus-Sf9 cells can be grown in suspension at high densities, presenting an attractive solution for improving scalability. Although the BEVS-based production platform has yield advantages, there are still major knowledge gaps in terms of its performance. Concerns include potential immunotoxicity against insect impurities, especially when the therapies are administered at high doses. Additionally, the influence of post-translational modifications (PTMs) such as glycosylation, acetylation, phosphorylation, and methylation on the functional activity and immunogenicity of rAAV vectors remains a subject of investigation.

Differences Between Sf9- and HEK293-Derived AAV Vectors: Unveiling Variations in PTMs, Impurities, and Potency

In a recent study, Rumachik and colleagues employed an array of techniques to gain insights into the differences between rAAV vectors produced by both manufacturing platforms.

Deep proteomic profiling and liquid chromatography-tandem mass spectrometry revealed PTMs on rAAV8 capsids produced by both platforms, with a greater number of PTMs detected on baculovirus-derived capsids. The presence of host-cell protein (HCP) impurities was observed in all vectors, but the researchers detected HCP impurities with N-linked glycans from baculovirus-derived vectors, posing safety concerns due to their allergenic potential in humans.

To determine any differences in potency between the two production methods, in vitro functional transduction assays were performed across various cell types. This included immortalised human HEK293T and Huh7 cells, primary human fibroblasts, primary human iPSCs, and immortalised mouse C2C12 myoblasts, all of which were transduced with ssAAV1-EF1a-FLuc. Analysis revealed that human-produced rAAV1 exhibited higher potency than the baculovirus-derived vector in all cases. Further experimentation showed that the presence of insect/baculoviral impurities, as well as empty capsids in baculovirus-Sf9 preparations, could be influencing the potency of these vector preparations. Findings were corroborated by in vivo experiments, where human-produced rAAVs demonstrated superior potency compared to baculovirus-produced vectors, both in mouse skeletal muscle and humanised liver models.

Differences Between Sf9- and HEK293-Derived AAV Vectors: Unveiling Variations in Empty/Full Capsids and VP Ratios

In a separate investigation, Giles et al. conducted a comprehensive comparison of AAV serotypes 1, 8, and 9 produced by either HEK293 or baculovirus-Sf9 methodologies.2 The study also assessed the influence of purification processes on any observed differences in vector characteristics.

 Analytical ultracentrifugation was used to determine the relative quantities of empty, full, and partially full capsids. On average, HEK293-produced preparations had the largest percentages of full capsids and fewer partially packaged genomes when compared to baculovirus-produced vectors. Furthermore, HEK293-derived vectors exhibited superior purity, as well as in vitro and in vivo potency.

AAV capsids consist of a total of 60 viral proteins (VPs), constituting a mixture of three overlapping gene products: VP1, VP2, and VP3, encoded by the cap open reading frame. These molecules appear in an approximate ratio of 1:1:10, respectively. The researchers observed that the average VP2:VP1 ratio was similar to the expected 1:1 for both production methods. However, baculovirus-derived vectors exhibited a two-fold higher (1:22.5) VP1:VP3 ratio than the expected 1:10 ratio, indicating an underrepresentation of VP1 and VP2. This deficiency has implications for infectivity and efficacy, as VP1 has a phospholipase A2 domain which is critical for virus infection.

Another study by Ebberink et al. evaluated AAVs from two different serotypes (AAV8 and AAV2) produced either using insect cell- or mammalian cell-based platforms, the latter of which included AAVs manufactured by Revvity with HEK293.3 When the team investigated capsid VP ratios, they found that the mammalian-produced vectors had a higher number of VP1 and VP2 than baculovirus-derived vectors. Differences in VP PTMs were also observed between the two production methods, including abundant phosphorylation of VP1 and VP2 in baculovirus-derived vectors.

The three studies highlight the need for careful consideration when selecting the most appropriate method for rAAV vector production. The Sf9-baculovirus production system is often favoured for its scalability; however, the presented studies highlight the overall advantage of HEK293-derived rAAV vectors. These findings underscore the importance of considering factors such as capsid protein content, VP ratios, PTMs, impurities, and potency when making this decision. Researchers and manufacturers need to evaluate these variations carefully, considering the specific therapeutic application, regulatory requirements, scalability needs, and available resources to enable successful downstream gene therapy applications.

Impact of the Therapeutic Payload Transfer Plasmid Design on the Quality Attributes of rAAV Vector Manufactured with HEK293 Producer Cells.

HEK293 producer cell lines receive the “instructions' ' for biosynthesizing the components of the rAAV vector, i.e., capsid and genome, through plasmids (Figure 2). Unmethylated CpG motifs in the genome of AAV vectors, particularly those intended for in vivo administration, is likely a CQA. For example, clinical studies in haemophilia B have provided evidence highlighting the involvement of CpG motifs in triggering the innate immune response. This response ultimately led to the removal of transduced hepatocytes through cytotoxic T lymphocytes (CTL), resulting in the cessation of therapeutic factor IX (FIX) expression4-6. CpG motifs can become part of the AAV vector genome through the design process. Designing an AAV vector genome involves making choices about multiple elements, such as ITRs, promoters, and therapeutic gene sequence. For instance, opting for viral promoters like CAG, CMV, or modifying the therapeutic transgene native codon by applying algorithms originally developed to enhance recombinant protein production, can introduce CpGs into the AAV genome 4,5. CpG motifs are likely to remain unmethylated during the production process, even when using producer cells derived from HEK293 cells, due to the rapid kinetics of replication and packaging.

​​Figure 2. The most employed approach for rAAV production utilises HEK293 producer cell lines in combination with three plasmids. These plasmids contain either the AAV rep and cap genes (i.e., packaging plasmid), the therapeutic payload gene flanked by ITR sequences (i.e., transfer plasmid), or adenoviral helper genes.

DNA encapsidated from plasmids employed in the manufacturing process, like the transfer plasmid, can also be a source of unmethylated CpG motifs5. Analysis of the amount of encapsidated transfer plasmid in a total of 71 rAAV vectors with genome sizes ranging from 1 to 4.9 kb has shown vectors in the 1-1.5 kb size category show the highest amount of encapsidated plasmid (mean =9.2%, n=11), while the 4.5 - 4.9 kb category, which includes the wild-type genome size, had the lowest amount (mean =2.6%, n=12)8 (Figure 3). These results show that encapsidated plasmid DNA, a process-related impurity, is influenced by the size of the rAAV genome, thereby this parameter should be optimised during the development process.

Figure 3. Encapsidated plasmid DNA, a process-related impurity and potential source of unmethylated CpG motifs, is influenced by the size of the rAAV genome. Process development researchers at Revvity analysed the amount of encapsidated transfer plasmid in a total of 71 rAAV vectors with genome sizes ranging from 1 to 4.9 kb. They have found the highest amount of encapsidated plasmid DNA in the 1-1.5 kb size category (mean =9.2%, n=11), while the 4.5 - 4.9 kb category, which includes the wild-type genome size, had the lowest amount (mean =2.6%, n=12).

Outlook

Large-scale manufacturing of rAAV vectors remains a major bottleneck in the gene therapy field. Thoroughly considering the design of the manufacturing process as early as possible will support a smooth transition from preclinical to clinical and the scale up to commercial phase, all whilst reducing time to market.