Unlocking Scalable rAAV Production Through Molecular Vector Engineering

Recombinant adeno-associated virus (rAAV) has become one of the most clinically validated platforms for gene therapy, supported by its favorable safety profile, broad tissue tropism, and capacity for long-term transgene expression. However, as AAV-based programs expand beyond rare-disease proof-of-concept studies into systemic indications and larger patient populations, manufacturing has become a major bottleneck. Many systemic AAV therapies require extremely high vector doses, placing increasing pressure on the field to improve upstream productivity, product quality, scalability, and cost efficiency. Consequently, the strategic engineering of vector plasmids and producer cell lines has emerged as a vital necessity for optimizing the molecular foundations of upstream production.

Minji Lee and Jung Eun Park recently published a review paper “The evolving landscape of recombinant adeno-associated virus (rAAV) manufacturing: Engineering vector plasmids for upstream production” in Biotechnology Advances. The article provides a comprehensive review of decades of progress in recombinant AAV production, with a particular focus on the HEK293-based transient transfection system, which remains one of the most widely used platforms for clinical and commercial AAV vector manufacturing. Importantly, the review frames rAAV manufacturing not only as a bioreactor scale-up or downstream purification challenge, but also as a molecular engineering problem. As the authors highlight, a new wave of innovation is emerging around the rational redesign of the genetic components that control vector production, including rep, cap, ITRs, helper genes, plasmid architecture, and stable cell-line systems.

Rep Engineering: Tuning Rep Function for Higher rAAV Productivity

The rep gene is one of the most critical genetic components in the rAAV production system. It encodes four non-structural proteins—Rep78, Rep68, Rep52, and Rep40—that play indispensable roles in the AAV life cycle. These proteins are produced through alternative transcription initiation and splicing from a single rep open reading frame. The large Rep proteins (Rep78 and Rep68) are responsible for recognizing and binding to the inverted terminal repeats (ITRs) at both ends of the viral genome. They perform site-specific endonuclease and ATP-dependent helicase activities required for initiating and resolving replication intermediates. The small Rep proteins (Rep52 and Rep40), by contrast, are primarily involved in genome packaging. They promote the accumulation and translocation of single-stranded progeny genomes into preformed capsids, ensuring the formation of fully packaged virions.

Three Engineering Strategies for Rep Optimization

Rep engineering aims to improve rAAV production by balancing Rep activity with host-cell tolerance. Because Rep proteins are essential for AAV genome replication and packaging but can also be cytotoxic when overexpressed, recent strategies focus on precise tuning rather than simple overexpression.

Three major Rep engineering strategies are emerging (Fig.1):

• Rep minimization: This strategy reduces Rep-associated cytotoxicity by selectively retaining essential Rep isoforms while removing or replacing more toxic elements. For example, Rep78 may be replaced with Rep68 to reduce growth-arrest effects, while Rep52 and Rep40 are retained to support genome packaging. However, results vary by system. Some studies show reduced cellular stress, while others suggest that Rep78 is still required for efficient production in certain rAAV contexts. This indicates that Rep minimization is feasible but highly platform- and serotype-dependent.
• Hybrid Rep engineering: AAV2-derived Rep is widely used because of its compatibility with AAV2 ITRs, but it may not perform optimally with capsids from other serotypes. Hybrid Rep designs combine functional domains from different AAV serotypes to improve Rep–ITR compatibility and genome packaging efficiency. Chimeric Rep variants involving AAV1, AAV2, AAV6, or AAV8 sequences have been shown to improve vector yield, increase genome-containing capsids, and reduce empty capsid formation in specific production systems.
• Directed evolution and codon-level Rep optimization: More advanced approaches use DNA family shuffling, iterative selection, and single-codon mutagenesis to identify Rep variants with improved production performance. For example, highly chimeric variants such as Rep1.03 have been reported to improve packaging efficiency while reducing empty capsid production. At the single-amino-acid level, the S110R substitution in the Rep origin-binding domain nearly doubled viral genome titers and transducing particle output across multiple serotypes.

Overall, rAAV productivity is not determined simply by the presence of Rep proteins, but by the balance among Rep isoforms, domains, expression levels, and sequence variants. Future manufacturing platforms will likely rely on multi-scale Rep tuning to reduce toxicity, improve genome replication, enhance packaging efficiency, and support more predictable high-yield rAAV production.

Cap Engineering: Designing Capsids for Function and Manufacturability

The cap gene is one of the most actively engineered components in rAAV development because the AAV capsid controls multiple critical functions, including tissue tropism, receptor binding, cellular uptake, immune recognition, genome packaging, and overall vector productivity. The capsid is formed by VP1, VP2, and VP3, which assemble into a 60-subunit icosahedral particle, typically at an approximate 1:1:10 ratio. Maintaining the correct VP1/VP2/VP3 balance is essential, because changes in capsid protein stoichiometry can directly affect capsid assembly, vector yield, and infectivity.

The cap gene also encodes accessory proteins such as AAP, MAAP, and X protein, which can influence capsid assembly, nuclear localization, DNA replication, and production efficiency. As a result, capsid engineering is no longer focused only on improving tissue targeting or transduction. It is increasingly being used to balance biological performance with manufacturability, including packaging efficiency, full capsid formation, and scalable production.

Three major capsid engineering strategies are emerging (Fig 2):

• Rational capsid design: This strategy uses existing structural and functional knowledge to introduce targeted capsid modifications, such as point mutations, peptide insertions, or domain swaps between serotypes. Historically, rational design has been used to alter tissue tropism, receptor binding, and transduction efficiency. Although it may be less powerful for discovering unexpected capsid phenotypes, it remains useful when the structure–function relationship is well understood and predictable engineering is desired.
• Directed evolution: This approach generates large capsid libraries through methods such as error-prone PCR, DNA shuffling, peptide display, or saturation mutagenesis, followed by selection under defined biological pressures. High-throughput sequencing is then used to identify enriched variants with desired properties, such as improved target-cell transduction, immune evasion, or packaging fitness. Directed evolution has been especially valuable for discovering functional capsids that would be difficult to design rationally. Comprehensive AAV2 capsid mutagenesis studies have also generated large sequence–function datasets and helped uncover important production-related features, including the role of MAAP in AAV biology.
• Machine learning-guided capsid design: ML-guided engineering uses large sequence–function datasets from mutagenesis, NGS-based screening, and capsid libraries to predict capsid performance and design improved variants. Supervised learning models can predict packaging fitness or tissue tropism from experimentally labeled datasets, while generative and unsupervised models can explore broader capsid sequence space when labeled data are limited. Advanced frameworks such as Pareto optimization, protein language models, and multi-objective design can help identify capsids that balance manufacturability, tissue specificity, immune evasion, and vector potency.

Cap engineering is moving from empirical screening toward data-driven, multi-objective vector design. The goal is no longer simply to create capsids with better transduction, but to develop AAV variants that combine strong biological function with high production fitness, improved full capsid formation, reduced off-target activity, and scalable manufacturability. As sequence datasets, computational tools, and screening platforms continue to improve, machine learning-guided capsid design is likely to become an increasingly important driver of next-generation rAAV vector development.

ITR Engineering: Optimizing the Cis-Acting Elements of rAAV Production

Inverted terminal repeats, or ITRs, are the only cis-acting elements required for rAAV genome replication and packaging. Each ITR is approximately 145 nucleotides long and forms a compact hairpin structure containing several essential functional regions, including the Rep-binding element, the secondary Rep-binding element, and the terminal resolution site. These elements enable Rep protein binding, genome rescue, replication, and encapsidation.

Although ITRs are small, they have a major impact on vector yield, genome integrity, transgene expression, and safety. As a result, ITR engineering has become an important strategy for improving both rAAV productivity and vector performance.

Major ITR engineering strategies include:

• ITR minimization: Early ITR engineering focused on deleting regions thought to be non-essential, with the goal of reducing genome size and identifying which ITR subdomains are required for replication and packaging. Some asymmetric or limited deletions can be tolerated with minimal effect on yield. However, larger deletions, especially those affecting key palindromic regions or Rep-binding motifs, can significantly reduce rAAV production. These findings show that ITR minimization is possible, but the structural constraints of the ITR leave little room for extensive modification.
• Functional engineering through scAAV design: Self-complementary AAV, or scAAV, is a major example of functional ITR engineering. By deleting the terminal resolution site and adjacent D element in one ITR, the vector genome forms a double-stranded, transcription-ready structure after delivery. This can accelerate and enhance transgene expression compared with conventional single-stranded AAV. However, scAAV has only about half the packaging capacity of standard AAV, approximately 2.4 kb, making it suitable mainly for smaller expression cassettes.
• Safety-focused ITR modification: CpG-rich regions in AAV vector genomes may activate innate immune sensors such as TLR9, potentially contributing to immune activation and reduced transgene persistence. CpG-free ITRs have been designed to reduce immune recognition, but this can also reduce vector yield, likely by impairing Rep-mediated genome replication. This highlights the need to balance immune safety with production efficiency.
• Synthetic ITR design: More recent approaches use rationally designed synthetic ITRs to improve productivity while reducing cellular stress responses. For example, synITRs have been reported to enhance genome titers while attenuating p53-mediated stress signaling. This suggests that ITR engineering can potentially improve both safety and yield when designed with Rep interactions and host-cell responses in mind.

In summary, ITR is not simply a passive packaging signal but actively coordinates with Rep proteins and host-cell pathways to influence genome replication, vector yield, cellular stress, and transgene expression. Future rAAV manufacturing strategies will likely rely on more precise ITR designs that preserve essential replication functions while improving safety, expression kinetics, and production efficiency.

Helper Gene Engineering: Moving Beyond the Minimal Set

AAV requires helper functions from adenovirus or herpesvirus for productive replication. Early rAAV production used helper-virus co-infection, which created safety and purification challenges. A breakthrough came with helper virus-free HEK293-based production, where essential adenoviral helpers are supplied by plasmids. Since HEK293 cells already express E1A/E1B, modern systems typically add E2A, E4, and VA RNA to support genome replication, Rep/Cap expression, and protein synthesis.

Helper engineering has evolved from identifying a “minimal” set toward tuning helper combinations for specific production contexts.

Key strategies:

• Minimal helper design– E2A supports DNA replication, E4orf6 aids RNA transport and host-cell modulation, and VA RNA prevents PKR-mediated translational shutdown. This approach became the foundation of HEK293-based rAAV manufacturing.
• E4 region optimization– E4orf6 is a key component, but the role of E4orf3 is context-dependent: some studies find it dispensable, others show improved productivity when co-expressed with E4orf6. Helper requirements vary by serotype, platform, and expression level.
• L4-22K/33K integration– These late adenoviral proteins (not in the original minimal set) regulate late transcripts and improve rep/cap expression. Their inclusion enhances genome packaging, full capsid ratios, and vector release.

In practice, helper engineering is shifting from a universal minimal-gene model to combinatorial optimization. The goal is to tune helper combinations to maximize productivity, quality, and safety for each production system. Future work must validate these designs in stable cell lines and large-scale manufacturing.

Genetic Consolidation and Vector Architecture Optimization

Reducing the number of plasmids required for rAAV production simplifies the system and improves scalability. The conventional three-plasmid system is flexible but requires precise stoichiometry, large amounts of GMP-grade plasmid DNA, and careful control of transfection variability.

Major architecture strategies:

• Dual-plasmid systems– pDG and pQT combine rep/cap and helper genes on one plasmid, with the vector genome on a second. pQT added kanamycin resistance and cloning flexibility for multiple capsids. Yields and quality remain comparable to three-plasmid systems.
• Alternative dual design (pOXB)– Places GOI and rep/cap on one plasmid and helper genes on another. This improves stoichiometric balance, increasing productivity (2.5-2.9×), full capsid ratios, and purity.
• Single-plasmid systems– Integrate GOI, rep/cap, and helper genes into one molecule. Reduces batch variability and backbone impurities, but larger plasmids can be unstable and prone to recombination.
• Alternative DNA templates– Minicircle, Nanoplasmid, doggybone DNA, and in situ recombination (AAVPureMfg) reduce bacterial backbone contamination, improve packaging efficiency, and increase full capsid ratios.

A central theme is stoichiometric balance – the right levels and timing of genome replication, Rep activity, Cap expression, and helper function. As the field moves toward dual-, single-, and alternative-template platforms, vector architecture becomes a key driver of yield, quality, and GMP compatibility.

Optimization of Cis-Acting Regulatory Elements

Cis-acting elements in the transfer cassette influence both therapeutic potency and manufacturability. Promoters, enhancers, WPRE variants, polyadenylation signals, and CpG content affect expression, genome size, immune activation, and vector stability. Cassette design must balance expression strength, safety, size, and production performance.

Key optimization strategies:

• Promoter optimization– Strong constitutive promoters (CMV, CAG) drive high expression but are large, reducing packaging capacity. Compact, tissue-specific, or synthetic promoters improve cell-type specificity and preserve space, especially for scAAV vectors.
• Post-transcriptional element engineering– WPRE enhances transcript stability and nuclear export. Safety-truncated variants (e.g., WPRE3) retain activity while removing oncogenic X-protein sequences.
• CpG motif management– CpG-rich sequences activate TLR9, causing inflammation and transgene silencing. CpG optimization reduces immunogenicity, but excessive depletion may destabilize plasmids. A balanced approach is recommended.
• Integrated cassette design– Promoter choice, WPRE variant, CpG content, transgene sequence, and genome size should be optimized together, not independently. The best designs balance potency, durability, safety, and manufacturability.

Cis-regulatory engineering connects therapeutic function with manufacturing performance. Future AAV vector design will increasingly use integrated, computationally assisted optimization to generate expression cassettes that are potent, compact, safer, and compatible with scalable rAAV production.

Stable Packaging and Producer Cell Line Engineering

Stable packaging and producer cell lines are being developed to overcome key limitations of transient transfection, including high plasmid demand, transfection variability, scalability constraints, and batch-to-batch inconsistency. The basic concept is to integrate essential rAAV production components into the host cell genome, allowing cells to provide part or all of the machinery needed for vector production.

A packaging cell line provides key viral production functions, such as rep, cap, and/or helper genes, while the gene of interest is delivered separately. A producer cell line is more complete: it contains all essential components, including the vector genome carrying the gene of interest, and can generate rAAV particles after induction without additional plasmid or viral delivery steps.

Major stable cell line engineering strategies include:

• Inducible control of toxic viral genes: AAV Rep proteins and certain helper genes can impair cell growth if expressed continuously. Early packaging and producer cell line systems showed that stable rAAV production is possible, but also revealed the importance of tightly controlling Rep and helper gene expression. Inducible systems, such as chemically regulated promoters or Cre/Lox-based switches, help suppress toxic genes during cell expansion and activate them only during production.
• Helper virus-free stable systems: To reduce biosafety and purification challenges associated with helper-virus infection, researchers developed systems in which rep/cap and adenoviral helper genes are stably integrated into HEK293 or 293T-derived cells. These virus-free approaches reduce contamination risks and simplify downstream processing, but they still require careful control of Rep expression, helper gene activity, genome packaging efficiency, and long-term genetic stability.
• Suspension-adapted industrial platforms: Scalable manufacturing requires producer cells that grow efficiently in suspension culture and perform consistently in bioreactors. Platforms such as ELEVECTA integrate rep and essential helper functions into HEK293 or CAP suspension cells under Tet-inducible control. These systems can serve as foundational packaging cell lines that are further customized with capsid genes and vector genomes for specific AAV products. Their ability to scale into large bioreactors makes them attractive for industrial rAAV production.
• Modular and tunable expression systems: Newer designs separate the vector genome, Rep machinery, Cap expression, and helper functions into independently regulated modules. This allows better control over expression timing and stoichiometry. By optimizing induction timing, limiting excessive transgene expression, and reducing capsid protein degradation, these systems can improve productivity and bring yields closer to transient transfection levels.
• CRISPR-mediated site-specific integration: The latest generation of stable cell line engineering uses CRISPR-Cas9 to insert rep, cap, and helper genes into defined safe-harbor loci such as AAVS1, ROSA26, or CCR5. Compared with random integration, site-specific integration improves genetic consistency, reduces clonal variability, and provides a more predictable expression architecture. Although productivity optimization is still ongoing, this approach offers strong potential for regulatory compatibility and clinical manufacturing.

In short, stable packaging and producer cell line engineering is moving from simple gene complementation toward precise, tunable control of viral gene expression. The goal is to balance cellular viability with high vector productivity by controlling where production genes are integrated, when they are expressed, and how strongly each component is produced (Fig 4). As demand for scalable and cost-efficient AAV manufacturing grows, stable and inducible producer cell lines are likely to become increasingly important industrial platforms.

Complementary Platforms and Process Engineering

While the review centers on HEK293-based systems, parallel advances in the Sf9/baculovirus platform—including OneBac cell lines and optimized Rep/Cap expression—continue to support commercial production. Broader process engineering improvements, such as suspension-adapted cultures, chemically defined media, fed-batch and perfusion strategies, advanced transfection reagents, and multi-omics-guided optimization, enhance productivity across platforms.

Key Takeaways and Future Directions

Modern rAAV manufacturing has evolved from empirical, multi-plasmid transient transfection toward integrated, data-driven, and highly engineered platforms. Coordinated optimization of Rep, Cap, ITR, helper genes, plasmid architecture, and host cells is delivering substantial gains in yield, quality, and scalability. Machine learning, synthetic biology, and precise inducible systems are accelerating this progress.

Remaining challenges include achieving consistently high full-to-empty ratios, minimizing impurities, and ensuring platform flexibility across diverse serotypes and payloads. Future success will depend on continued mechanistic understanding, convergence of computational and biological tools, and the development of unified, modular manufacturing systems. These advances will be instrumental in realizing the full therapeutic potential of AAV gene therapies for a broad range of patients.

How PackGene Supports Next-Generation AAV Manufacturing

As rAAV manufacturing continues to move toward higher productivity, improved vector quality, and better scalability, integrated technical support across vector design, plasmid engineering, production, purification, and analytical testing becomes increasingly important. PackGene supports AAV developers by providing end-to-end services that help translate these upstream engineering principles into practical vector production workflows.

PackGene’s AAV platform can support projects from early discovery to preclinical and GMP-oriented development, including customized vector design, plasmid construction, serotype and capsid selection, scalable AAV production, downstream purification, and quality-focused analytical characterization. For programs requiring improved yield, reduced empty capsids, better genome integrity, or optimized expression cassette design, PackGene can help evaluate key design elements such as promoter selection, regulatory elements, vector genome size, serotype choice, and manufacturability considerations.

By combining AAV process development experience with flexible production platforms and analytical testing capabilities, PackGene helps researchers and gene therapy developers address key challenges in rAAV manufacturing, including productivity, purity, full capsid ratio, residual impurities, vector genome integrity, potency, and batch-to-batch consistency. These capabilities make PackGene a strong partner for advancing AAV programs from molecular design to high-quality vector material for downstream research, translational development, and clinical manufacturing preparation.

Reference:

Lee M, Park JE. The evolving landscape of recombinant adeno-associated virus (rAAV) manufacturing: Engineering vector plasmids for upstream production. Biotechnol Adv. 2026 Apr 27