An AAV plasmid is a circular piece of bacterial DNA engineered to carry the genetic instructions needed to manufacture adeno-associated virus vectors, the most widely used delivery vehicles in gene therapy. No single plasmid does the job alone. The standard production method relies on three separate plasmids, each carrying a distinct set of instructions, introduced simultaneously into human cells. How those plasmids are designed, built, and combined determines nearly everything about the final therapeutic product, from how much virus you get to how pure it is and which tissues it reaches.
The Three-Plasmid System
AAV does not replicate on its own. In nature, it needs a helper virus (usually adenovirus) to complete its life cycle. Early lab production of AAV vectors mimicked this by infecting cells with actual adenovirus, which was messy and created contamination problems. Researchers solved this by identifying the specific adenovirus genes that AAV needs and assembling them onto a single “helper plasmid.” The critical adenovirus genes turned out to be the E4, E2A, and VA RNA regions. When this helper plasmid was delivered into human embryonic kidney 293 cells (which already supply the adenovirus E1 gene), AAV vectors could be produced just as efficiently as with live adenovirus infection, but without the contamination headache.1Gene Therapy. Adeno-associated virus vectors can be efficiently produced without helper virus
The second plasmid is the rep/cap plasmid, sometimes called the packaging plasmid. It encodes two families of proteins: the Rep proteins, which recognize and replicate the viral genome, and the Cap proteins (VP1, VP2, and VP3), which assemble into the icosahedral shell that protects the DNA and delivers it to target cells. The ratio of these capsid proteins in the final particle is not tightly controlled by any organizing principle. Instead, capsids assemble stochastically from whatever pool of VP1, VP2, and VP3 is available, meaning the relative expression levels of these proteins directly shape what ends up in the final product.2Nature Communications. Adeno-associated virus capsid assembly is divergent and stochastic A persistent challenge is that the rep/cap plasmid is present at low copy number inside the cell, limiting how much Rep and Cap protein gets made and ultimately capping vector yields.3PubMed. Replication of rep-cap genes is essential for the high-efficiency production of recombinant AAV
The third plasmid is the vector plasmid, also called the cis plasmid or transfer plasmid. This one carries the actual therapeutic gene you want delivered to a patient, flanked on both ends by the inverted terminal repeats (ITRs) of AAV. Everything between those ITRs gets packaged into the viral capsid. Everything outside them does not, at least in theory. The ITRs are the only viral sequences the vector plasmid retains, and they are the only piece of AAV DNA that the final product is supposed to contain.4PubMed. A User’s Guide to the Inverted Terminal Repeats of Adeno-Associated Virus
Why the ITRs Matter So Much
The inverted terminal repeats are short DNA sequences, about 145 bases each, that sit at both ends of the AAV genome. They fold into distinctive hairpin structures through internal self-pairing, forming T-shaped configurations that serve as the starting points for DNA replication.5Wiley Online Library. AAV Genome Topology Decides ITR Secondary Structure Rep proteins recognize these hairpins and use them to initiate copying the vector genome. Without intact ITRs, the therapeutic DNA cannot be replicated inside the production cell or packaged into capsids.
The ITRs also direct recombination events after the vector enters a patient’s cells. They promote the formation of circular DNA molecules and head-to-tail concatemers, which are important for long-term gene expression in non-dividing tissues like the liver or retina.6PubMed Central. Inverted terminal repeat sequences are important for intermolecular recombination and circularization of adeno-associated virus genomes In short, the ITRs are doing double duty: they are essential for manufacturing the vector and for making it work once it reaches the target tissue.
The ITR Instability Problem
Here is one of the most frustrating practical challenges in AAV vector production: the very hairpin structures that make ITRs functional also make them unstable when you try to grow the vector plasmid in bacteria. Bacteria are the workhorse for producing plasmid DNA at scale, but bacterial enzymes tend to chew up those palindromic hairpin sequences over the course of normal growth. The damage is not random. Detailed sequencing has shown that the most common outcome is internal deletion of one hairpin arm with inversion of the neighboring arm, accounting for the majority of degraded sequences. Some plasmids lose the inverted repeat structure almost entirely.7Nucleic Acids Research. Degradation and stable maintenance of adeno-associated virus inverted terminal repeats in E. coli
This degradation happens over a timescale of days, which is exactly the window needed for large-scale plasmid production. The ITR closer to the bacterial origin of replication tends to degrade faster than the one farther away. The practical consequence is that every batch of vector plasmid needs careful quality-control testing to confirm the ITRs are still intact before you use it to make virus. A batch with heavily degraded ITRs will produce less vector and potentially vector with altered properties.
The Roughly Five-Kilobase Ceiling
AAV capsids can hold only about five kilobases of single-stranded DNA. That sounds like plenty until you account for the ITRs themselves (about 290 bases combined) plus whatever regulatory elements you need: a promoter to drive gene expression, a polyadenylation signal to terminate it, and sometimes enhancers or introns to boost output. After those elements claim their share, the space left for the actual protein-coding sequence is often closer to 3.5 to 4 kilobases.8PubMed Central. Adeno-associated Virus (AAV) Dual Vector Strategies for Gene Therapy Encoding Large Transgenes
Many disease-causing genes exceed this limit. The dystrophin gene behind Duchenne muscular dystrophy, for instance, has a coding sequence far larger than any single AAV vector can accommodate. Researchers have developed dual-vector strategies that split a large gene across two separate AAV vectors. When both vectors infect the same cell, the two halves recombine to produce a full-length messenger RNA. The trans-splicing approach, where each half carries a splice signal that the cell’s own machinery joins together, has proven considerably more effective than the overlapping approach, where the two vectors share a region of identical sequence that drives homologous recombination. In muscle tissue, trans-splicing was roughly twelve times more effective at reconstituting the full gene product.9Molecular Therapy. Expanding AAV Packaging Capacity with Trans-splicing or Overlapping Vectors: A Quantitative Comparison
Self-Complementary Vectors
Standard AAV vectors deliver single-stranded DNA. Before the therapeutic gene can be expressed, the cell has to convert that single strand into a double-stranded form, a step that can be slow and inefficient depending on the tissue. Self-complementary AAV vectors sidestep this bottleneck entirely. The vector plasmid is designed so that the packaged genome folds back on itself, forming a double-stranded molecule immediately upon release from the capsid. The tradeoff is that the effective payload capacity is cut roughly in half, to about 2.2 kilobases of coding sequence, because the genome is essentially carrying two complementary copies of the same information.10PubMed. Self-complementary AAV vectors; advances and applications
For genes small enough to fit, the gains can be dramatic. In a large animal study comparing self-complementary and standard single-stranded AAV vectors in the retina, the self-complementary version produced detectable gene expression much earlier and at higher levels, with expression remaining stable for the seven-month study duration.11PubMed Central. AAV retinal transduction in a large animal model species: Comparison of a self-complementary AAV2/5 with a single-stranded AAV2/5 vector This makes self-complementary designs especially attractive for diseases where you need fast onset and the therapeutic gene is compact enough to fit.
Controlling Where the Gene Turns On
The promoter sequence on the vector plasmid determines which cell types will actually express the therapeutic protein. A strong ubiquitous promoter will drive expression in virtually any cell the vector reaches, which is useful in some contexts but dangerous in others. If a protein meant only for liver cells gets expressed in immune cells, for example, it can trigger an unwanted immune response against the therapeutic product itself.
Tissue-specific promoters address this by restricting expression to the desired cell type. In neonatal mice given AAV9 vectors intravenously, a cardiac-specific promoter confined transgene expression to the heart, while a desmin promoter drove expression in both cardiac and skeletal muscle.12PubMed Central. Tissue specific promoters improve specificity of AAV9 mediated transgene expression following intra-vascular gene delivery in neonatal mice These promoters act as an additional layer of safety on top of the capsid’s own tissue preferences.
More recently, deep-learning methods have been used to design entirely synthetic promoters from scratch, guided by single-cell chromatin accessibility data. When tested for targeting specific retinal cell types in mice, these AI-designed promoters outperformed rationally designed ones, achieving stronger and more specific expression in the intended cells.13PubMed Central. Deep learning-guided design of cell type-specific AAV promoters This is a shift from choosing among known natural promoters to engineering custom regulatory sequences tailored to the clinical need.
Steering Tissue Tropism Through the Rep/Cap Plasmid
While the promoter controls gene expression after the vector enters a cell, the capsid determines which cells the vector enters in the first place. Different AAV serotypes have different surface properties that make them bind to different receptors on different cell types. AAV9, for instance, crosses the blood-brain barrier efficiently; AAV8 homes to the liver; AAV1 and AAV6 are strong in muscle. All of these serotype choices are encoded on the rep/cap plasmid.
Swapping the cap gene on the packaging plasmid is the simplest way to change tissue tropism while keeping the same therapeutic payload. But researchers have gone further, creating hybrid capsids by grafting specific amino acid regions from one serotype onto another. Studies mapping the tropism determinants of AAV1 found that replacing amino acids 350 to 736 of AAV2’s capsid protein with the corresponding AAV1 sequence produced a hybrid vector that behaved like AAV1 in muscle, both in cell culture and in living animals. Narrowing the critical region further suggested that amino acids 350 to 430 function as a major tissue tropism determinant.14PubMed Central. Characterization of tissue tropism determinants of adeno-associated virus type 1 These findings enable a toolbox approach where capsid modules can be mixed and matched to create vectors with customized targeting profiles.15PubMed. Custom adeno-associated virus capsids: the next generation of recombinant vectors with novel tropism
Reducing Immune Triggers Built Into the DNA
An underappreciated aspect of vector plasmid design involves the CpG dinucleotide content of the packaged genome. CpG motifs are short DNA sequences that mammalian immune systems recognize as foreign, primarily through a receptor called TLR9. A vector genome loaded with CpG sequences can provoke an innate immune response that destroys transduced cells and limits the duration of gene expression. In one striking demonstration, vectors with CpG-depleted genomes were able to establish persistent transgene expression and evade immune detection in a model system where unmodified vectors were cleared.16PubMed Central. CpG-depleted adeno-associated virus vectors evade immune detection
Reducing CpG content has become something of a best practice in vector plasmid design. Strategies include codon-modifying the therapeutic gene’s coding sequence (choosing synonymous codons that avoid CpG) and altering non-coding elements like promoters and polyadenylation signals to minimize CpG density throughout the packaged genome.17Molecular Therapy. Quantification of CpG Motifs in rAAV Genomes: Avoiding the Toll This adds another constraint to an already tight design space: you need to fit your gene, your regulatory elements, and your ITRs within roughly five kilobases while also keeping the CpG count low.
Optimizing Plasmid Ratios for Manufacturing
Getting the ratio of the three plasmids right turns out to be surprisingly important and surprisingly non-obvious. For years, many labs defaulted to a 1:1:1 weight ratio of helper, rep/cap, and vector plasmid. Systematic optimization studies using design-of-experiment approaches have consistently found that this default is suboptimal. For AAV production in suspension HEK293 cells, one group found that using lower amounts of total plasmid DNA and vector plasmid combined with higher levels of the rep/cap plasmid yielded about 4.6-fold more vector than their previous one-factor-at-a-time method, reaching over 1014 viral genomes per liter after purification.18Molecular Therapy – Methods & Clinical Development. Creation of a High-Yield AAV Vector Production Platform in Suspension Cells Using a Design-of-Experiment Approach
The optimal ratio also varies by serotype. An optimization study for AAV2 and AAV9 found that the ideal helper-to-rep/cap-to-vector weight ratios were quite different between the two: roughly 1:3.5:0.5 for AAV2 versus 1:1.4:0.3 for AAV9. Compared to the standard 1:1:1 ratio, these optimized mixtures roughly doubled genome titers for both serotypes and reduced the proportion of empty capsids by about a quarter.19PubMed. Enhancing the production of adeno-associated virus (AAV)2 and AAV9 with high full capsid ratio in HEK293 cells through design-of-experiment optimization of triple plasmid ratio The lesson is that plasmid ratio optimization should be treated as serotype-specific rather than one-size-fits-all.
What Else Gets Packaged
In an ideal world, only the DNA between the ITRs would end up inside capsids. Reality is messier. During production, fragments of all three plasmids and even host cell DNA get encapsidated at low but measurable levels. Long-read sequencing of clinical-grade AAV batches has shown that over 95% of packaged DNA is the intended vector genome, with the vector plasmid backbone being the largest impurity source at roughly 2 to 4%, helper plasmid sequences at around 1% or less, and host cell DNA at less than 0.1%.20Molecular Therapy. Comprehensive long-read sequencing analysis of recombinant adeno-associated virus vector batches
A reasonable concern is whether these stray DNA fragments could produce unwanted proteins once delivered to a patient. Detailed analysis of encapsidated plasmid DNA found that most reads were partial and lacked complete open reading frames. For different impurity genes, the fraction of reads containing a complete coding sequence ranged from as low as 2% for E4 to 63% for the kanamycin resistance gene, suggesting that most encapsidated plasmid fragments are too broken up to be transcribed into functional proteins.21PubMed Central. Residual DNA impurities in AAV vectors—nature and transcription Reassuring, but not zero risk, which is why regulatory agencies pay close attention to residual DNA levels in approved products.
One reason empty capsids dominate many preparations is a timing mismatch during production. Modeling work has shown that capsid proteins start being made earlier than viral DNA replication ramps up. By the time plenty of replicated vector DNA is available for packaging, capsid production has already slowed down, and many empty shells have already been exported from the nucleus. The result is a harvest where the majority of capsids often contain no DNA at all.22Molecular Therapy Methods & Clinical Development. Mechanistic modeling of adeno-associated viral vector production by triple plasmid transfection
Alternatives to Traditional Plasmids
The baculovirus/insect cell system offers a fundamentally different production route. Instead of transfecting plasmids into human cells, the AAV genes and the therapeutic cassette are encoded in recombinant baculoviruses that infect insect cells (typically Sf9 cells derived from fall armyworm). A large-scale comparison found that insect cell-derived AAV had substantially higher yields than HEK293-derived AAV, with roughly 40-fold more viral genomes at comparable scales. The full capsid ratio was also higher: about 93% for insect cell product versus about 71% for HEK293 product.23PubMed Central. Systematic comparison of rAAV vectors manufactured using large-scale suspension cultures of Sf9 and HEK293 cells Insect cell vectors also showed a six-fold reduction in contaminating DNA packaging for certain serotypes and higher biological potency in some comparisons.24Molecular Therapy. A Scalable Platform for Production of High-Titer Recombinant AAV Vectors Reproducing Serotype-Specific Variants Despite these manufacturing advantages, the two systems produced comparable results in animal efficacy studies, meaning the choice between them is driven more by manufacturing economics than biology.
An entirely different approach eliminates plasmids altogether. Enzymatic DNA amplification technologies can produce linear DNA constructs in vitro, bypassing bacteria entirely. One such platform, called doggybone DNA, uses an enzymatic process that can generate gram-scale quantities of manufacturing-grade DNA without the ITR instability problems inherent to bacterial propagation.25PubMed Central. Synthetic Biology Design as a Paradigm Shift toward Manufacturing Affordable Adeno-Associated Virus Gene Therapies Another enzymatically produced format has demonstrated equivalent viral genome titers, full-to-empty capsid ratios, and infectivity compared to conventional plasmid controls across multiple AAV serotypes.26Cytotherapy. Process Development and Manufacturing SYNTHETIC, ENZYMATICALLY PRODUCED DNA FOR GENE THERAPY AND VACCINE APPLICATIONS These cell-free DNA technologies are still maturing but could eventually solve the ITR degradation problem and reduce dependence on bacterial fermentation.
Next-Generation Plasmid Architectures
Even within the traditional triple-transfection framework, the plasmids themselves are being redesigned. Minimized-backbone vectors, sometimes marketed under names like Nanoplasmid, strip out unnecessary bacterial sequences to reduce the overall plasmid size. A smaller backbone means less non-therapeutic DNA in the production cell, which can translate to lower impurity levels in the final product and improved safety profiles.27PubMed Central. Improving cell and gene therapy safety and performance using next-generation Nanoplasmid vectors
A more ambitious redesign uses site-specific recombination to convert a standard plasmid into a minicircle inside the production cell. In one system, an enzyme called Bxb1 excises the therapeutic cassette from the bacterial backbone after the plasmid enters the cell, generating a small circular DNA molecule that contains only the ITR-flanked transgene and no prokaryotic sequences. This approach reduced bacterial DNA contaminants in the final AAV product by 10- to 50-fold compared to standard triple transfection while also increasing the ratio of DNA-containing capsids up to threefold.28PubMed Central. High-purity AAV vector production utilizing recombination-dependent minicircle formation and genetic coupling By ensuring that the bacterial backbone is physically separated from the transgene cassette before packaging begins, the system attacks the impurity problem at its root rather than trying to remove contaminants after the fact.
Improved helper plasmids are evolving in parallel. A recent redesign incorporating deletions within the E4 and E2A regions of the helper plasmid led to increased AAV productivity, demonstrating that even these well-established components still have room for optimization decades after their initial development.29PubMed. An Improved Helper Plasmid Containing Deletions Within the E4 and E2a Genes Results in Increased Adeno-Associated Virus Productivity Purification methods for the plasmids themselves are also advancing. A chromatography technique using a kosmotropic salt has been shown to selectively enrich for the supercoiled form of plasmid DNA, which is the most biologically active configuration, while removing relaxed and damaged forms. The method works across plasmids ranging from 3 to 12 kilobases, covering the size range typical of AAV production plasmids.30Wiley Online Library / Biotechnology and Bioengineering. Selective hydrophobic interaction chromatography for high purity of supercoiled DNA plasmids
How AAV Plasmids Were First Built
The molecular cloning of AAV into plasmids happened in the early 1980s through two independent efforts. One team, working in Gainesville, Florida, constructed molecular clones by adding GC tails to AAV2 DNA duplexes and inserting them into the plasmid pBR322. A second group at the National Institute of Allergy and Infectious Diseases took a different approach, adding synthetic DNA linkers to enable clean excision of the intact AAV genome from the plasmid later. Both strategies worked: when the resulting AAV plasmids were transfected into mammalian cells co-infected with adenovirus, infectious AAV particles were produced from the plasmid-borne genome, free of plasmid sequence.31Molecular Therapy. AAV Vectorology, Much More than a Gut Feeling Those early molecular clones laid the foundation for everything that followed, establishing the basic principle that AAV genomes could live as passengers inside bacterial plasmids and be rescued on demand in mammalian cells. Forty years later, the plasmid-based production workflow those experiments enabled remains the dominant manufacturing method for clinical AAV vectors, even as the plasmids themselves have grown far more sophisticated.