Adeno-associated virus is one of the most demanding feeds in modern downstream processing. The product is a 25-nm protein capsid; the impurities are host-cell proteins, residual DNA, and — most awkwardly — capsids of the same size and shape as the product, but without the genome inside. Separating these full and empty capsids is the polishing step that defines how a clinical-grade AAV process performs. It is also the step where most chromatography formats stop being good enough.
This note is aimed at process-development teams choosing media for AAV polishing. It focuses on anion-exchange chromatography — the technique that has emerged as the standard for full/empty separation — and on why doing it well requires both resolution and throughput in the same column.
The full/empty problem
An empty AAV capsid and a full one are structurally almost identical: same proteins (VP1, VP2, VP3), same overall geometry, similar surface charge profile. The difference is the encapsidated genome — the negatively charged DNA inside. That genome shifts the apparent surface charge of the capsid just enough that an anion-exchange resin sees the full and empty populations as slightly different species and elutes them at slightly different conductivities.
The word that matters in that sentence is slightly. The charge difference between full and empty is small. A polishing step that wants to deliver a high-percentage full population needs a chromatographic system that can resolve closely-running peaks — the kind of resolution traditionally associated with packed-bed resins, not with the convective formats AAV processes prefer for throughput.
On top of full/empty, the same anion-exchange step is usually asked to clear residual host-cell protein and residual host-cell DNA from the upstream affinity capture. These impurities elute earlier in the gradient and are typically the easy part of the separation — but the column has to handle them without compromising the harder full/empty window.
Why neither incumbent format is quite right
Membrane adsorbers
Anion-exchange membrane adsorbers are widely used for AAV polishing because they tolerate the flow rates needed to process clinical-scale harvests and because they are gentle on capsids. Their limitation is peak resolution. Pleated membrane stacks have heterogeneous flow paths — the centre of a pleat sees a different velocity than its edges — and that heterogeneity broadens elution peaks. For impurities with a wide elution window this is harmless. For full/empty separation, where the windows overlap, it makes the difference between a process that hits a 90 %-full target on the first column and one that needs a second polishing pass.
Membrane formats also have relatively high void volume, which dilutes the eluate. Diluted product fractions usually mean an extra ultrafiltration step downstream — another unit operation, another yield hit.
Packed-bed resins
Packed-bed anion-exchange resins give the resolution membranes lack — but at a cost that AAV processes can rarely afford. AAV capsids are large, and mass transport into porous beads is slow. To preserve dynamic binding capacity, packed-bed columns have to run at residence times of several minutes; the cycle times that follow do not scale to commercial AAV harvest volumes. The same diffusion limitation also means that as flow rate goes up, peak shape degrades — exactly the parameter that has to stay sharp for full/empty separation.
There is a secondary issue. Packed-bed resins for AAV are typically smaller-bead, higher-density formats, and the resulting pressure regime is harder on capsid integrity at high flow than the open architecture of a convective column. For some sensitive AAV serotypes, this matters.
Where a cellulose monolith Q sits
A monolith is a single continuous porous body with interconnected channels rather than packed particles. Mass transport is convective — like a membrane — so the format runs at high flow without giving up dynamic binding capacity. Unlike a pleated membrane, the channel network is homogeneous: the residence-time distribution stays narrow, and peaks stay sharp at process flow rates.
MonoCore™ Q is Lumatix's anion-exchange monolith built on a cross-linked cellulose backbone in a radial-flow geometry. The feed enters along the outer wall and exits at the core. Path length stays short, cross-section stays large, and the pressure drop is in the membrane regime even at flow rates that would shut a packed-bed column down. The result is the design point AAV polishing actually needs: convective throughput with peak resolution close to a packed-bed resin.
On the chemistry side, MonoCore Q is available with both strong and weak anion-exchange ligands. Strong Q is the conventional choice for AAV full/empty work; weak Q opens an alternative gradient regime that some development teams use to optimise resolution at lower conductivity.
Host-cell protein and residual DNA
The same anion-exchange column that resolves full and empty also catches the residual HCP and DNA that survive upstream capture. Both impurity classes are more strongly retained than AAV capsids on Q chemistry, so a well-designed gradient elutes the AAV window first and leaves the high-charge contaminants on the column until a strip step at the end of the cycle. The convective transport in a monolith helps here too: sharper peaks at the AAV window mean less co-elution of trailing HCP into the product fraction, which improves measured HCP clearance even when the gradient itself is unchanged.
Scale-up behaviour
AAV polishing methods developed at bench scale frequently break when transferred to larger devices, because the flow-distribution behaviour of pleated stacks and packed beds both change with scale. A monolith with a single continuous channel network does not have that problem to the same degree. The pressure profile and residence-time distribution stay consistent from a 1.8 mL bench device to larger formats; method-transfer effort is materially lower, even if it is never zero.
That said, monolith scale-up is not magic. Conductivity step heights, hold times, and gradient slope still need to be re-checked at the new scale. What changes is that the hydraulic envelope you tuned in at bench does not move under you on the way to pilot.
Practical method-development pointers
Load conditions matter more than the resin choice for first-cut full/empty separation. AAV typically loads in low-conductivity buffer at near-neutral pH; the actual load conductivity is the lever that decides where the AAV window sits relative to HCP and DNA. Spend the development time on conductivity scouting before tuning gradient slope.
Shallow linear gradients in the AAV elution window give the cleanest full/empty resolution. Step gradients are tempting at process scale because they shorten cycle time, but they typically cost a percentage point or two of full purity. On a monolith, the cycle-time penalty of a shallow gradient is mild — the convective transport keeps total run time short even at low slope — so the trade-off is usually worth it.
Plan a high-salt strip step at the end of every cycle to clear retained HCP and DNA, and a low-salt re-equilibration before the next load. Both steps are standard for AAV anion-exchange polishing regardless of media format.
When this is the right tool
Monolith Q is positioned for AAV polishing where the full/empty separation is the limiting step — which is most clinical-grade AAV processes. For very early discovery-scale work where load volumes are small and full/empty content is not yet a release specification, a simple membrane polishing step may still be the right place to start. The crossover happens when the process moves into preclinical development and full/empty percentage becomes a real number on a real CoA.
Further reading
For the underlying transport physics that lets a monolith sustain peak resolution at convective flow rates, see our article on convective vs diffusive mass transport in chromatography. For a broader side-by-side of monoliths and membrane adsorbers across applications, the monolith vs membrane comparison covers hydraulics, dynamic binding capacity, and scale-up in more detail.