A 2025 multi-omics study found that 46% of HEK293T cells in a standard triple transfection AAV production run lacked transcripts from at least one of the three plasmids — meaning nearly half of cells were structurally incapable of producing full, genome-containing AAV particles. This single finding explains most of what we observe as “high empty capsid ratio.” This post breaks down the causes and what you can realistically do about them.
What are empty capsids and why do they matter?
AAV capsids are proteinaceous shells assembled from 60 copies of VP1, VP2, and VP3 in a 1:1:10 ratio. Full capsids contain the packaged single-stranded DNA genome (the therapeutic transgene). Empty capsids are structurally identical shells that assembled and completed the packaging process without encapsidating a genome. They are:
- Biologically inactive: empty capsids do not transduce cells — they bind and internalise but have no DNA to express
- Immunogenic: empty capsids present the same capsid epitopes as full capsids — they consume patient immune tolerance to the capsid without delivering therapeutic benefit, potentially reducing the effectiveness of re-dosing
- A regulatory concern: FDA and EMA both require characterisation of the empty/full ratio in the drug substance specification, and high empty ratios require additional purification steps that add cost and product loss
Typical empty/full ratios from triple transient transfection in HEK293T: 50–90% empty capsids before purification. After iodixanol gradient or AEC purification: 10–30% empty. The goal is to push the pre-purification ratio as low as possible to simplify downstream processing.
The three root causes of empty capsid generation
1. Unequal plasmid delivery to individual cells
Triple transfection requires all three plasmids — transfer vector (transgene + ITRs), packaging plasmid (rep + cap), and helper plasmid (adenoviral helper functions) — to be present in the same cell simultaneously. The 2025 multi-omics data showing 46% of cells lacking transcripts from at least one plasmid reveals that transfection efficiency is highly heterogeneous at the single-cell level.
Cells that receive cap plasmid but not the transfer vector will assemble capsids — but have no genome to package. These produce empty capsids. Cells that receive all three plasmids produce primarily full capsids. The empty/full ratio is therefore directly determined by the proportion of cells that receive all three plasmids simultaneously — a function of transfection efficiency and plasmid stoichiometry.
2. Genome packaging is kinetically slower than capsid assembly
Capsid assembly and genome packaging are sequential but not tightly coupled processes in transient transfection. Capsid proteins are expressed from the cap plasmid, fold, and assemble into procapsids — these procapsids then package AAV genomes that have been replicated from the transfer vector. If cap protein expression outpaces genome replication (which occurs when the packaging plasmid is over-represented relative to the transfer vector), excess procapsids will complete assembly without packaging a genome.
This is why plasmid ratio optimisation is so impactful: a transfer vector:packaging plasmid:helper plasmid molar ratio that is biased toward more transfer vector reduces the relative excess of cap protein and increases the proportion of procapsids that successfully package a genome.
3. Cell stress drives premature capsid release
AAV production is cytotoxic — the Rep proteins induce cell cycle arrest and eventual cell death as part of their normal replication function. Cells under stress upregulate autophagy and exosome secretion, which can release incompletely assembled or genome-lacking capsids into the medium before packaging is complete. Cell stress is exacerbated by endotoxin in FBS (TLR4-driven antiviral signalling), high passage number, and over-confluency at transfection — all of which accelerate the apoptotic timeline and reduce the window for genome packaging.
Process parameters that reduce empty/full ratio
| Parameter | Optimised approach | Expected impact on empty/full |
|---|---|---|
| Plasmid ratio (transfer:packaging:helper) | Increase transfer vector ratio — typically 1:1:1 molar is suboptimal; 2:1:1 or higher transfer vector reduces empty capsids by 20–40% | High — 20–40% reduction in empty fraction |
| FBS endotoxin content | Switch to FBS Low Endotoxin (<1 EU/mL) — reduces TLR4-driven antiviral signalling and cell stress during production | Moderate — 15–30% improvement in full capsid fraction |
| Transfection reagent and ratio | Optimise PEI:DNA ratio (typically 3:1 w/w for linear PEI 25kDa) — over-complexed PEI is cytotoxic; under-complexed gives poor delivery | High — directly affects per-cell plasmid delivery uniformity |
| Cell confluency at transfection | 70–80% — S-phase cells have higher nuclear import efficiency for all three plasmids | Moderate — 10–20% improvement through better per-cell delivery |
| Harvest timing | Serotype-dependent: AAV2 at 48–72h, AAV5/8/9 at 72–96h — harvesting too late increases empty fraction as stressed cells release incomplete capsids | Moderate — 10–25% depending on serotype |
| Cell passage number | Use low-passage HEK293T (<30 passages) — high-passage cells have reduced transfection efficiency and higher baseline stress | Moderate — particularly important for consistency |
| BAX/BAK1 knockout | Published data: BBKO HEK293T increases full capsid AAV2 titre 55% and AAV9 titre 66% without increasing empty capsid fraction | High — but requires cell line engineering investment |
The plasmid ratio optimisation — what to test
If you have not optimised your plasmid ratio, this is the highest-impact experiment you can run to reduce empty capsids. The standard 1:1:1 molar ratio (transfer:packaging:helper) is a reasonable starting point but rarely optimal for any specific transgene/serotype combination. Recommended optimisation matrix:
- Keep helper plasmid constant at 1 molar equivalent (helper functions are needed in excess)
- Test transfer:packaging ratios of 1:1, 2:1, 3:1, and 4:1 (transfer:packaging)
- Measure both physical titre (qPCR for genome copies) and full capsid fraction (AUC-based empty/full separation or TEM)
- Optimal ratio gives maximum physical titre × maximum full fraction — not maximum physical titre alone
Why FBS grade matters for empty/full ratio
The connection between FBS endotoxin and empty/full ratio is indirect but mechanistically clear. Endotoxin activates TLR4 → NF-κB → type I interferon response in HEK293T cells. The interferon response upregulates PKR (double-stranded RNA-activated protein kinase), which phosphorylates eIF2α — globally suppressing translation. In a state of suppressed translation, cap protein expression is reduced, but AAV genome replication (which uses cellular DNA replication machinery) may be less affected. This shifts the balance toward more genomes relative to cap protein — paradoxically potentially reducing empty capsids. However, the dominant effect of endotoxin-driven interferon activation is overall titre reduction, including full capsid titre. The net effect of high-endotoxin FBS is always worse, not better.
Using FBS Low Endotoxin (<1 EU/mL) removes this variable entirely — cells produce cap protein and replicate AAV genomes without the confounding effect of innate immune activation.
Purification strategies for removing empty capsids
Even with an optimised production process, some empty capsid removal by downstream purification is typically required:
- Iodixanol gradient ultracentrifugation: separates full (higher density ~1.41 g/mL) from empty (<1.32 g/mL) capsids by density — the oldest and most reliable method; suitable for research scale
- Anion exchange chromatography (AEC): full and empty capsids have different surface charge profiles at specific pH — AEC can separate them at scale; required for GMP production
- Ultracentrifugation in CsCl: highest resolution for empty/full separation but not scalable to manufacturing; primarily used for analytical characterisation
FBS Low Endotoxin for AAV Production — Reduce Process Variability
Endotoxin in standard FBS activates antiviral signalling in HEK293T — reducing both total titre and full capsid fraction. SeamlessBio supplies FBS Low Endotoxin (<1 EU/mL, LAL-tested per lot) with batch reservation up to 24 months.
English 
Deutsch