Species cross-reactivity

A therapeutic antibody has to be tested for safety in an animal before it reaches people, and that testing only means something if the antibody binds the animal's version of the target. Species cross-reactivity is therefore a design constraint, not a late characterization step.

Why it bites late

The default discovery path optimizes for the human target. If the epitope happens to sit on a conserved surface, cross-reactivity comes for free. If it sits on a divergent loop, the panel can be excellent and still unusable for toxicology, and that is usually discovered after the panel has narrowed.

The fix is cheap at the start and expensive later: include the ortholog in the screen, or at least in the triage, from the first round.

Testing it properly

• Use the same assay format for every species. A human ELISA and a cynomolgus cell assay are not comparable.
• Use material of matched quality. The ortholog often comes from a different expression batch, sometimes with a different tag and a different level of aggregation, and each of those changes apparent binding.
• Titrate. A single-concentration binding result cannot distinguish a tenfold difference from a hundredfold one, and that distinction decides whether a tox study is interpretable.
• Test the format you will use. Avidity can hide a real affinity loss on the ortholog.
• Confirm on cells where the target is a membrane protein, because sequence differences outside the epitope can change presentation.

When there is no cross-reactivity

This happens, and there are established routes forward.

Choose a different species. If the target is conserved in another laboratory species, the tox package can move there.

Make a surrogate. A separate antibody against the animal ortholog, used for the animal studies, with the human antibody carried in parallel. It is more work and it introduces a comparability argument, but it is a standard approach.

Engineer cross-reactivity in. Where the difference is one or two residues in the epitope, affinity maturation against both orthologs, or selection alternating between them, can produce a binder that sees both.

Use a transgenic model. Where one exists for the target, it can replace the ortholog requirement.

Each of these changes cost and timeline, which is exactly why the question belongs in the first design conversation.

Record it clearly

Report cross-reactivity as measured affinity or potency per species, in the same format, with the material described. A phrase like "cross-reactive with cyno" without a number is the kind of statement that gets copied into a slide and then into a plan, and it can mean anything from equivalent binding to a hundredfold loss.