Why does a nanobody aggregate, and can it be fixed without losing binding?
The problem this addresses
Single-domain antibodies are attractive because they are small, stable and easy to format. The recurring disappointment is solubility: a clone binds beautifully, then aggregates on concentration, needs arginine in every buffer, or comes off the column polydisperse. At that point most programmes either add an additive and live with it, or go back to the panel.
The conventional explanation is the four hallmark positions in framework region 2. Camelid VHHs carry hydrophilic residues there; human VH domains carry hydrophobic ones, because in a conventional antibody that face is buried against the light chain. A single-domain antibody with VH-like hallmark residues has an exposed hydrophobic patch and tends to self-associate.
But the literature also contains sdAbs with VH-like hallmarks that are perfectly soluble — so the hallmark residues cannot be the whole story. This study takes two real problem clones and works out which part of the explanation applies to each.
What was done
Two single-domain antibodies, sdAb A and sdAb B, were isolated from an alpaca immune library. Both carried VH-like hydrophobic hallmark residues and both had poor solubility. Four VHH-like substitutions were introduced into framework region 2 of each: V37Y, G44E, L45R and W47L.
- sdAb A was solved by the hallmark swap alone. After mutation it could be prepared as a monomer with no additive in the buffer.
- sdAb B was not. It remained polydisperse unless arginine was present — so something outside framework region 2 was driving the aggregation.
- The second cause was found computationally. Spatial aggregation propensity calculations predicted the hydrophobic surface of sdAb B and identified Trp99 as the residue responsible. Introducing W99A gave a monomer.
- Function survived. The engineered sdAbs were characterised structurally, physicochemically and biophysically, and retained their binding activity.
What this means if your nanobody is misbehaving
There are two distinct failure modes and they need different treatment. If the hallmark positions are VH-like, the fix is a known four-residue edit and it is fast. If the hallmarks are already VHH-like — or the swap does not fully solve it — the aggregation is coming from a CDR or framework residue specific to that clone, and it has to be located rather than guessed. Surface-property prediction narrows that search to a small number of candidates before any protein is made.
The practical consequence is that “this nanobody aggregates” is not a verdict on the clone. It is a question with two possible answers, and both are usually addressable while keeping the binding you selected for.
Have a single-domain antibody that will not stay monomeric?
Send the sequence and what you are seeing — SEC profile, the buffer it needs, the concentration where it fails. A PhD scientist reads it and replies within two business days. No NDA needed to start.
Talk to a scientist →Frequently asked questions
Can you fix solubility without a crystal structure?
Yes. This study explicitly demonstrates that, using sequence-based hallmark analysis plus computational surface-property prediction to identify the offending residue. A structure helps when one exists, but it is not a prerequisite for starting.
Will the mutations cost me binding?
In this work the engineered sdAbs retained their functionality, which is the point of choosing framework and surface positions rather than paratope residues. Every variant we return comes with its affinity measured so the trade is visible rather than assumed.
Is this only for camelid VHH, or does it apply to human VH domains too?
The hallmark logic is precisely about the boundary between the two. It applies to any single-domain construct whose framework region 2 face is exposed — camelid-derived sdAbs with VH-like hallmarks, and human VH domains being used as standalone binders.
Can you do this on a clone we discovered elsewhere?
Yes. Engineering campaigns routinely start from a sponsor-supplied parent, whether it came from our discovery work or someone else’s.
Do we keep the sequences?
Yes, with no downstream royalties in standard scope.
Want this run on your target?
Keeping a single domain soluble through engineering is the practical problem behind this paper. These are the campaigns where it matters.
- Llama / alpaca (VHH) humanization — hallmark framework residues assessed position by position, with solubility measured
- VHH / single-domain discovery — de novo camelid single domains, reformattable to VHH-Fc or bivalent
- Developability optimization — aggregation, self-association and viscosity engineered out
Every campaign is scoped individually and quoted. Email info@abwizbio.com with your target or lead sequence, or see all 25 publications and the full service menu.