Cytotoxicity and cell-killing assays
A killing assay asks the end-to-end question: does the molecule reduce the number of living target cells? It is the readout that matters for antibody drug conjugates, for effector-function antibodies and for cell-engaging formats, and it folds binding, internalization, payload release and payload potency into one number.
Choosing a readout
ATP luminescence measures metabolically active cells and is the workhorse: fast, sensitive, and read on a plate reader.
Resazurin reduction is cheaper and slower, with a wider dynamic range at the cost of sensitivity.
Membrane integrity readouts, such as release of lactate dehydrogenase, measure lysis specifically and are the usual choice for effector-mediated killing, where the mechanism is lysis rather than growth arrest.
Imaging or live cell counting distinguishes killing from growth inhibition, which the bulk metabolic readouts cannot. If a molecule stops cells dividing without killing them, an ATP readout shows a partial effect that looks like weak killing.
Controls that carry the interpretation
- A target-negative line. Killing that happens on cells without the target is not target-mediated.
- An isotype or irrelevant conjugate, matched for format and for payload load, which separates payload toxicity from targeted delivery.
- Free payload alone, titrated, which sets the ceiling for what the payload can do and shows whether the conjugate is delivering.
- Unconjugated antibody, which reveals any effect of the antibody by itself.
- Untreated and vehicle wells, and a positive control that kills completely, to set both ends of the window.
For effector-function assays, the effector to target ratio, the effector cell source and the donor all move the result, so they belong in the method and in the comparison.
Time is a variable, not a detail
Conjugates usually need days, because internalization, trafficking and release take time and the payload then has to act, often on dividing cells. Reading too early underestimates potency and can reorder a panel. Reading too late lets untreated wells overgrow and compresses the window. Pick the time from a pilot time course rather than from habit, and use the same time across the panel.
Seeding density matters for the same reason: a plate seeded too densely runs out of room before the payload acts.
Reporting potency
Fit the full curve, report the midpoint with its confidence interval, and report the maximum killing as well. Two conjugates with identical midpoints and different plateaus are not equivalent, and the plateau is often the more informative number, because incomplete killing at saturating concentrations points at a resistant subpopulation or at a delivery limit.
Normalize to the controls on the same plate, not to a historical value, and run independent experiments before believing a difference.