An AAV ELISA measures assembled capsid, full or empty, by binding an antibody raised against the intact particle. It is quick, it needs no nucleic acid extraction, and it answers a question no genome assay answers: how much protein shell is in the preparation. It also does not tell you how much of that shell contains a genome, or how much of it will deliver one, and most of the confusion around AAV titres comes from reading one of those numbers as though it were another.
Capsid titre and genome titre are different numbers, and an aav titer quote says which
A capsid ELISA counts particles by their shell. A genome assay, by quantitative PCR or droplet digital PCR, counts copies of the packaged sequence. The ratio between them is a property of the preparation, not a constant, and it is one of the more informative things you can know about a process, because it says how much of what you made is empty. An infectivity or transduction assay measures a third thing again. A preparation described by a single titre is a preparation whose empty fraction somebody has not measured, and the honest specification quotes at least the first two numbers with the method for each.
Serotype specificity decides which kit works
The antibodies in these kits are raised against conformational features of a particular capsid, so a kit for one serotype may read a related one weakly, differently, or not at all, and an engineered or chimeric capsid may fall outside every catalog kit. That is the first question to ask, before sensitivity or price: has this kit been shown to bind this capsid, and with what standard. A kit used outside its validated serotype produces a number, which is worse than producing nothing.
Where the empty capsid problem shows up
Because the ELISA cannot distinguish a full particle from an empty one, a process change that improves packaging can look like it did nothing, and one that produces more empties can look like it improved yield. Analytical ultracentrifugation, charge-based separation or mass photometry are the methods that resolve the population; the ELISA's contribution is the denominator. Use it that way, and pair every capsid reading with the genome reading from the same sample rather than from a neighbouring run.
Validating the kit against your own process
Run the kit against your own material across the range you actually see, including a deliberately diluted and a deliberately concentrated sample, and check linearity and recovery rather than trusting the kit's own curve. Ask what the reference standard is and whether it is the same serotype. Watch for matrix effects from your formulation buffer, surfactant or excipients, which are a common cause of a titre that moves when the process did not. For material heading into a regulated study, the expectations for analytical methods sit with the therapy's own manufacturing and control requirements.
Where a cgmp elisa differs from a research one
The immunochemistry is the same and everything around it is not. A kit used for release has to come with a documented lot history, a stated shelf life, a change control that tells you before the antibody lot moves, and a validation in YOUR matrix rather than the vendor's. That last point is the one that costs time: a kit qualified on a clean buffer says little about a concentrated harvest with surfactant in it, and the bridging exercise between the research kit and the release kit is a project rather than a comparison. Where a program is heading for a submission, ask the supplier at the research stage whether a compliant version of the same kit exists, because switching antibody pairs later means requalifying everything measured with the first one.
aav concentration, and which number it means
Concentration is ambiguous for a viral vector and the ambiguity is expensive. Genome copies per millilitre counts packaged genomes by qPCR or digital PCR. Capsid particles per millilitre counts protein shells by ELISA whether or not they carry a genome. Infectious units per millilitre counts what transduces a permissive line. A preparation can read high on capsids and low on genomes, which is an empty-heavy lot, and the ratio is what a dose calculation and a release specification are actually built on. Ask which assay produced a figure before comparing two preparations.
aav characterization, and the panel it takes
Characterisation of a vector lot is a panel rather than a test: identity of the transgene by sequencing, genome titre, capsid titre and the full to empty ratio, purity by gel or capillary electrophoresis for the three capsid proteins, aggregation by size exclusion or light scattering, residual host cell DNA, plasmid and protein, endotoxin and sterility, and potency in a cell-based assay. Analytical ultracentrifugation or charge detection mass spectrometry is what settles empty and full where the ratio matters. The panel, not the titre, is what a comparability exercise between batches rests on.
aav package, and what a packaging reaction includes
Packaging is the step where a genome flanked by inverted terminal repeats is placed inside a capsid, and what it includes decides the yield: the rep proteins that recognize the repeats and resolve them, the cap proteins that assemble the shell, and the adenoviral helper functions that make the cell permissive. Payload size is the hard limit, since a genome much beyond about 4.7 kilobases packages poorly and produces truncated species that assays read as a genome. Self-complementary designs halve the usable payload and gain expression speed.
aav vs lentivirus, and what decides between them
The two vectors answer different questions. AAV does not integrate, is non-replicating and gives long expression in non-dividing tissue, with a payload limit around 4.7 kilobases and pre-existing immunity in a share of the population. Lentivirus integrates, which means expression survives cell division and passes to daughter cells, carries eight or nine kilobases, and is the vector for ex vivo cell modification. So in vivo delivery to muscle, retina or liver is AAV territory, and stable modification of dividing or cultured cells is lentiviral territory.
A tcid50 assay, and what it measures
The fifty per cent tissue culture infectious dose is a functional titre: serial dilutions are added to replicate wells of permissive cells and the dilution at which half the wells show infection is calculated, usually by the Reed-Muench or Spearman-Karber method. It measures what actually infects, which neither a genome copy number nor a capsid count does, and it is therefore the potency-linked number for a viral preparation. The costs are time, since it waits on a cytopathic effect or a stain, and variability, which is why it is run with a reference preparation on the same plate.
aav9, and the tropism it is chosen for
AAV9 is chosen because it crosses the blood brain barrier to a useful degree and transduces heart and skeletal muscle widely after systemic delivery, which is why it carries the neuromuscular programs. What comes with that reach is dose: a systemic body-weight dose is orders of magnitude above a local injection, which brings manufacturing scale, immune response and liver exposure with it. Pre-existing neutralising antibodies exclude a share of patients, and capsid variants engineered from it aim to keep the tropism at a lower dose.
Questions people ask about aav elisa
Does an AAV ELISA measure infectious virus?
No. It measures assembled capsid whether or not it contains a genome and whether or not it can transduce a cell. Infectivity needs a cell-based assay and correlates only loosely with either titre.
Which AAV ELISA kit fits my serotype?
Only one shown to bind your capsid with a standard of the same serotype. Engineered and chimeric capsids frequently fall outside every catalog kit, and the honest answer there is a custom assay rather than the closest available.
Why does my capsid to genome ratio look wrong?
Most often the two numbers came from different samples, different dilutions or different laboratories. Measure both on the same aliquot before concluding anything about the empty fraction.