Home › Services › Llama / Alpaca (VHH) Humanization

The residues that make a VHH soluble are the ones humanization wants to remove. How do you keep both?

Carefully, and by selection rather than by substitution alone. A camelid VHH is soluble as a single domain precisely because four hallmark framework-2 residues replaced the hydrophobic patch that would otherwise sit against a light chain. Reverting them to their human VH equivalents is what humanization nominally asks for — and it is also what re-exposes that patch and invites aggregation. Abwiz Bio substitutes hallmark residues with human equivalents while preserving the solubility and stability that make single domains attractive, using STEM™ to hold function constant through the change, and delivers solubility and stability data on the humanized domain rather than sequence identity alone.

Why VHH humanization is its own problem

Humanizing a conventional antibody is a grafting exercise: move CDRs onto a human framework and repair the loop support. Humanizing a VHH is a trade-off between two properties that are physically linked.

Camelid heavy-chain-only antibodies function without a light chain, and the framework-2 hallmark residues are the adaptation that makes this work — they convert the former VH/VL interface from a hydrophobic surface into a soluble one. Human VH residues at those positions expect a light chain to cover them. Put them into a domain that has none, and the well-behaved single domain becomes a sticky one.

There is a second complication. VHH domains often carry an extra-long HCDR3 that folds back over the former interface, and its conformation depends on framework contacts and on the additional disulfide bonds many VHHs form between HCDR2 or framework 2 and HCDR3. Changing framework residues can disturb that arrangement and, with it, the long loop that reaches into the cleft or catalytic site the VHH was selected for.

How we approach it

• Position-by-position evaluation of the hallmarks. The framework-2 hallmark positions are treated individually rather than as a block. Some tolerate the human residue in a given domain; some do not. Which is which depends on the specific HCDR3 and disulfide arrangement, so it is determined for your VHH rather than assumed from a general rule.
• Universal framework grafting where appropriate. The rest of the framework is brought to human germline in the standard way, so humanness is maximized at the positions that carry no solubility burden.
• STEM™ to retain function. Variant libraries around the humanized domain are selected on your antigen, so the substitutions that would cost binding are removed by the selection instead of discovered afterwards in a single hand-built construct.
• Solubility and stability as measured outputs. Because the entire risk of VHH humanization is biophysical, we deliver solubility and stability data on the humanized domain, not merely a humanness score. Our published work on single-domain solubility is summarized in the single-domain solubility study.
• Thermostability and polyreactivity filters run through every selection stage, so the humanized domain retains the robustness that motivated the format.

Formats and related services

Humanized VHH domains are rarely the final format. They are reformatted to VHH-Fc for half-life and effector function, built into bivalent and biparatopic constructs, or used as arms in multispecifics — and the humanization has to survive all of that, which is why stability is engineered rather than hoped for.

• VHH / single-domain discovery — de novo camelid discovery when you do not yet have a domain, including selection for extra-long HCDR3 and HCDR2/HCDR3 cysteines.
• Thermostability engineering — heat-challenge selection, which single domains take particularly well.
• Developability optimization — aggregation, viscosity, sequence liabilities and polyreactivity.
• Cross-species reactivity — the small VHH paratope reaches conserved epitopes, which often makes ortholog binding easier to engineer.

Our published work on a VHH binding a flexible epitope in complement factor H is described in that study, and the resulting domain bound at 1.71 pM by Biacore.

You receive

• Humanized VHH sequence and protein
• Solubility and stability data on the humanized domain
• Binding data benchmarked against the parent VHH
• Reformatting to VHH-Fc or bivalent constructs on request
• No downstream royalties, stage-gated approval before each next step

Frequently asked questions

Why is humanizing a VHH different from humanizing a conventional antibody?

Because the residues humanization targets are the ones that make the format work. Camelid VHH domains carry four hallmark framework-2 residues that replaced the hydrophobic VH/VL interface with a soluble surface, allowing the domain to function without a light chain. Reverting them to human VH equivalents re-exposes that patch and can cause aggregation, so the substitutions must be evaluated position by position rather than applied as a block.

Will humanization make my nanobody aggregate?

That is the principal risk, and it is why we deliver measured solubility and stability data on the humanized domain rather than a humanness score alone. Hallmark positions are assessed individually for your specific VHH, since tolerance depends on the HCDR3 conformation and the disulfide arrangement of that domain.

Does humanization affect the long HCDR3?

It can. Extra-long HCDR3 loops often fold back over the former interface and are held by framework contacts and, in many VHHs, by disulfide bonds to HCDR2 or framework 2. Changing framework residues can disturb that arrangement, which is why function is held constant by STEM selection on your antigen rather than checked only at the end.

Can you humanize both llama and alpaca VHH domains?

Yes, both are supported, as are single domains from other camelid sources.

Can the humanized domain be reformatted?

Yes. Humanized VHHs can be converted to VHH-Fc, bivalent or biparatopic constructs, or used as arms in multispecific formats. Because stability has to survive reformatting, it is engineered through the campaign rather than assessed only on the isolated domain.

What do you need to start?

The VHH amino-acid sequence and the antigen. Send them to info@abwizbio.com for a scoped quote.