Affinity and avidity

Affinity is the strength of a single binding interaction: one site, one target, described by an on-rate, an off-rate and their ratio. Avidity is what happens when several interactions act together, so that releasing the whole molecule requires all of them to let go at once.

The practical difference is enormous. A single site with modest affinity, presented twice against a target that is present twice, can produce an apparent affinity orders of magnitude tighter. Nothing about the binding site changed.

Where avidity comes from

The molecule. An IgG has two arms. Tandem VHHs, bispecifics and Fc fusions all carry more than one site.

The target. Receptors cluster, antigens repeat on a viral surface, and fibrils present the same epitope many times along their length. A monovalent binder can still gain avidity when its target is a polymer.

The experiment. Capturing an antibody at high density on a sensor chip creates avidity at the surface even when the molecule is monovalent in solution. So does coating an ELISA plate heavily.

That last one is the source of most confusion, because it is a property of the assay rather than of the biology.

How to tell them apart

• Compare formats. Measure the Fab or the monomeric VHH next to the bivalent molecule. A large gap that closes when valency is removed is avidity.
• Vary surface density. If the apparent affinity improves as density rises, avidity is contributing.
• Look at the off-rate. Avidity mostly slows dissociation. An off-rate that looks impossible for the on-rate you measured usually means rebinding or multivalency.
• Check the fit model. Data that only fit with a bivalent analyte model are telling you something about valency.

Which number you want

It depends on the question.

For ranking binding sites, engineering affinity or comparing designs, you want monovalent affinity, measured with one site against one target and with orientation held constant.

For predicting what happens on a cell or on a fibril, you want the avid number, because that is the situation the molecule will meet. Just label it as apparent affinity under those conditions.

Programs get into trouble when the two get mixed: a panel ranked monovalently and then advanced bivalently can reorder, because avidity rescues weak, fast-off binders more than it helps tight ones.

Reporting it so nobody is misled

State the format, the orientation, the surface density if there is one, and the model used to fit. A KD without those is a number without a referent. In a report, showing the monovalent and avid measurements side by side is more useful than choosing one, and it removes the argument later about which number was meant.