Dose-response titration and EC50
A dose-response curve is the backbone of assay work. You vary concentration over a wide range, measure a response, and fit a curve. The fitted midpoint, EC50 in a functional assay or IC50 in an inhibition assay, is what usually gets reported and compared.
The fit
The standard model is the four-parameter logistic: a bottom, a top, a midpoint and a slope. All four matter.
Bottom and top define the assay window. If the curve does not reach a plateau at both ends, the midpoint is an extrapolation and should be treated as one. Constraining the top or bottom to a value the data did not reach is sometimes reasonable and always worth stating.
The Hill slope describes steepness. A slope far from one is a finding, not a nuisance parameter to be constrained away. It can indicate cooperativity, multiple sites, aggregation, or a problem in the assay.
The midpoint is a property of the assay, not only of the molecule.
EC50 is not KD
This is the most common confusion in antibody work. EC50 is the concentration that produces half the maximal response under these conditions. KD is a thermodynamic constant for the interaction. They coincide only under restrictive conditions.
Two effects break the equivalence routinely. Ligand depletion happens when the target concentration is comparable to the KD, so a substantial fraction of the antibody is consumed by binding; the measured midpoint then reflects the target concentration rather than the affinity. Avidity and signal amplification shift the midpoint when the readout saturates before binding does, which is common in cell assays and in ELISA with an enzyme-linked secondary.
Neither effect makes the measurement wrong. They make it conditional.
Designing a titration
- Span enough range. Aim for points above and below the expected midpoint, with the plateaus captured. A titration that spans three orders of magnitude and lands with the midpoint in the middle is worth more than twice as many points bunched together.
- Use a consistent dilution factor, and prepare the series in one operation to avoid compounding pipetting error.
- Replicate across the plate, not down a column. Edge effects and gradients are real.
- Include controls on every plate: a reference molecule, a no-antibody well, and an isotype or irrelevant binder control matched for format and concentration.
- Randomize or at least rotate plate positions across a panel so that layout does not correlate with candidate.
Comparing across candidates and across days
A single midpoint carries little information on its own. What makes comparisons trustworthy is running candidates on the same plate with the same reagent lots, keeping a reference molecule in every run, and reporting the spread across independent experiments rather than the tightest replicate set.
Between-day variation is usually larger than within-plate variation. If two candidates differ by less than the between-day spread of the reference, they have not been distinguished.