What exactly are kon, koff and Kd?
In a simple reversible 1:1 interaction, the association rate constant kon describes complex formation and the dissociation rate constant koff describes complex separation. The equilibrium dissociation constant is KD = koff / kon. Lower equilibrium Kd corresponds to tighter binding under the assumptions of that model.
Real antibodies can deviate from a 1:1 system. Bivalent IgG binding, heterogeneous antigens or heterogeneous ligand surfaces may require different interpretation. Always state whether a result is a model-derived apparent affinity or a defensible intrinsic interaction estimate.
SPR and BLI: practical differences
| Consideration | SPR | BLI |
|---|---|---|
| Detection | Surface refractive-index response | Interference shift at biosensor tip |
| Fluidics | Typically flow-based | Dip-and-read plate format |
| Study design | Well suited to detailed kinetic methods | Convenient for multi-sample screening |
| Common risks | Mass transport, bulk effects, surface artifacts | Sensor loading, shaking/mixing, nonspecific signal |
| Interpretation | Both require controls, quality review and an appropriate binding model | |
Artifacts to check before believing a strong binder
- Immobilization and orientation: a captured antigen may not present the relevant epitope as it appears on cells.
- Surface density and avidity: multivalent formats can show slower apparent dissociation than a monovalent interaction.
- Mass transport: diffusion or mixing can limit observed rates rather than true molecular kinetics.
- Nonspecific interactions: sticky molecules can generate misleading responses even against reference surfaces.
- Antigen quality: aggregation, misfolding, tags or inactive fractions compromise the experiment.
- Model fit: apparent numerical precision should not exceed the experimental evidence.
Useful mitigations may include alternative orientations, lower surface densities, reference/control proteins, matched buffers, replicate dilution series and orthogonal cell or binding measurements when biology demands them.
What to request in a CRO data package
- Raw or suitably exported sensorgrams, not just a Kd summary.
- Antigen construct, source, quality, concentration series and sample handling conditions.
- Capture/immobilization design, reference subtraction and regeneration approach, if used.
- Model assumptions, fitted values with limits and a clear statement of data-quality flags.
- A conclusion that considers specificity and developability alongside affinity.
When comparing candidates, use consistent target lots, assay formats and controls where feasible. Never rank a panel solely by apparent Kd when expression, aggregation or target presentation differs materially.
Related reading: validating AI-generated antibodies and preclinical developability decision design.
Scientific context
- Sartorius best practices: minimizing kinetic assay artifacts
- Blueprint for antibody biologics developability (2023)
These links provide background context and do not imply partner endorsement.