Binding Assay Formats: What Each One Measures
Binding establishes that a molecule arrived. It establishes nothing about what happened once it got there. Agonists, antagonists and inverse agonists are all perfectly capable of binding with comparable affinity, and separating them requires a functional readout. Binding and function are therefore complementary rather than interchangeable, and any compound described solely by its affinity has been characterized halfway.
Even that half depends on how it was obtained. Three formats share the name “binding assay” while measuring three different quantities: how much compound is bound once equilibrium is reached, how rapidly it associates and dissociates, and whether it can displace something else. Quote an affinity without naming its format and the information required to interpret it has already been discarded.
Two ways to run the experiment
Saturation is the direct route. Rising concentrations of labeled ligand are applied to a fixed quantity of receptor until binding levels off, and the resulting curve delivers both a dissociation constant and the total count of binding sites. The catch is that it demands a labeled version of the compound under study.
Competition sidesteps that. A labeled reference ligand is held at fixed concentration while increasing amounts of unlabeled test compound displace it, and the displacement curve’s midpoint yields an IC50. Converting that IC50 into an inhibition constant requires the reference ligand’s own affinity and concentration. Since no labeled test compound is needed, this is the format that actually gets run most of the time, and its output rests on a conversion that presumes straightforward competition at a single site. Allosteric binding breaks the presumption, and so does the existence of more than one site. Which constant results, and what each means, is the subject of Ki, IC50 and EC50.
The subtraction that defines the number
Labeled ligand adheres to the receptor, to everything else in the preparation, and to the tube itself. Specific binding is then defined by subtraction: total binding minus whatever persists when a large excess of unlabeled ligand is present.
Defining it that way makes non-specific binding operational rather than physical, which also makes it adjustable. Blocking agent, filter type and wash conditions each move it. When non-specific binding accounts for a large share of the total, the specific signal becomes a small difference between two large numbers, and its uncertainty grows to match.
Peptides are especially exposed here on account of how readily they adsorb, for the reasons given in adsorptive loss to surfaces. Any binding curve worth reading is accompanied by the proportion of total binding that was specific. Somewhere under half, and the numbers merit little confidence.
Three platforms
| Radioligand | Still the reference approach. Its chief virtue is that the label is small enough not to perturb binding behavior. Against that, it requires handling radioactive material and a separation step to strip away unbound ligand. |
| Fluorescence polarization | Detects the slowed tumbling of a small labeled ligand once it binds a large receptor. No separation is needed, and performance improves as the size difference grows, which makes it a poor fit for a small peptide binding a small target. The fluorophore is also bulky next to a short peptide and may itself change how that peptide binds. |
| Surface plasmon resonance | One partner is immobilized and mass accumulating at the surface is tracked in real time, producing association and dissociation rates rather than an equilibrium constant alone. Immobilization is the trade: anchoring receptor or peptide at a given point risks obstructing the very surface that does the binding. |
A ratio conceals the two rates that make it
Any equilibrium constant is a quotient of two rate constants. Two compounds can share an affinity precisely and still differ enormously in how long each stays bound, and a method that reports only the quotient has no way of distinguishing them.
This becomes important when analogs are compared. A modification made in pursuit of tighter binding might have altered the off-rate, the on-rate, or both, and the affinity figure hides which. It is also the reason a lone constant from a competition assay describes a compound more thinly than its precision suggests.
Has it reached equilibrium yet?
Equilibrium constants presuppose that equilibrium was actually attained before measurement. Reaching it takes time, and the time required is dictated by the off-rate, so a compound that dissociates slowly may need hours.
Stop early and the apparent affinity comes out weaker than the real one, with the error growing largest for exactly those compounds that bind most tightly. Verification means demonstrating that binding no longer changes with additional incubation. A methods section that names an incubation time without showing it was long enough has left the assumption standing untested.
Receptor source and buffer
Membranes, whole cells and purified receptor all support binding measurements and all give different answers. Stripping receptors into membrane preparations discards the cellular context that shapes receptor conformation. Whole cells keep that context while introducing uptake and degradation. Purified receptor eliminates both complications and may fail to fold the way it would inside a membrane.
Buffer carries equal weight. Ionic strength, pH, and whether divalent cations and guanine nucleotides are present all shift apparent affinity at a G-protein-coupled receptor. Give two laboratories the same compound and the same receptor in different buffers and they will report different constants, which is one specific case of the general difficulty set out in why two laboratories get different results.
The limits of a selectivity claim
Selectivity statements inherit the same constraint as affinity statements. Calling a compound selective for one subtype nearly always means it was tested against some portion of the family rather than all of it. Stated honestly, the claim names which subtypes were examined and which were not, as discussed for the melanocortin family in MC1R and MC4R selectivity.
