An AAV capsid that failed to package its genome ("empty") and one that succeeded ("full") are nearly identical in size and surface chemistry, so chromatography struggles to tell them apart. What differs is buoyant density: the packaged ~4.7 kb single-stranded DNA genome adds mass without adding volume, so full capsids are measurably denser.
Isopycnic banding: a particle stops migrating when
ρ(medium at radius r) = ρ(particle)
Sedimentation drift (this model):
dv/dt = ω²·k·(ρ_particle − ρ_local(y)) − γ·v + η(t)
ω = rotor angular velocity (∝ rpm²)
ρ_local(y) = ρ_top + Δρ · (y / tube height) [linear gradient]
γ = viscous drag, η(t) = thermal/Brownian jitter (band broadening)
In practice a CsCl or iodixanol step/continuous gradient is spun at ~60,000–65,000 rpm for hours in an ultracentrifuge. Empty particles equilibrate near ρ ≈ 1.32 g/mL; full particles equilibrate higher, around ρ ≈ 1.40–1.45 g/mL depending on capsid serotype and genome size — the two populations settle into separate, visible bands that can be fractionated and pulled off separately.
- Rotor speed — higher ω sharpens each band (stronger restoring force relative to diffusion) and shortens time-to-equilibrium.
- Gradient steepness — a steeper Δρ compresses the whole tube's density range, pulling the two bands closer together and lowering resolution; a shallow gradient spreads them out but takes longer to run.
- Empty : Full ratio — mimics a real upstream production batch, where the empty fraction can be 50–90% before purification.
- Resolution Rs — computed the same way as chromatographic resolution: Rs = 2·(yfull − yempty) / (wfull + wempty), using each band's mean position and standard deviation. Rs ≥ 1.0 is generally considered baseline-separated.
- This is exactly why AUC (analytical ultracentrifugation) and density-gradient ultracentrifugation remain gold-standard empty/full quality-control assays for clinical AAV lots, alongside orthogonal methods like charged-detector mass spectrometry.