Signal Nonadditivity and Model Mismatch in Microscale Thermophoresis Can Severely Bias Kd Determination

Accurately determined solution-phase equilibrium dissociation constants (Kd) between molecular-recognition elements and analytes are essential for biosensor development. Microscale thermophoresis (MST) is widely used for this purpose because it requires little sample and follows a simple titration workflow. Routine MST analysis, however, assumes that the normalized fluorescence response of an equilibrium mixture is a mole-fraction-weighted combination of the responses of pure free target and pure target-analyte complex. This signal additivity is required for accurate Kd determination, but it is not automatically guaranteed. Starting from thermophoretic redistribution and the fluorescence normalization used in MST, we show that additivity is recovered only under specific conditions, most notably when the free target and target-analyte complex have the same preheating fluorescence brightness per labeled molecule. Binding-induced quenching or enhancement makes the normalized response fluorescence-weighted rather than purely population-weighted, so standard fitting can yield biased Kd values. Simulations show that moderate fluorescence imbalance can cause substantial systematic error, especially when the target concentration is high relative to Kd. Ligand-dependent optical artifacts and sample heterogeneity provide additional model-mismatch routes, whereas early post-heating readout is not intrinsically nonadditive if applied consistently. These results establish signal additivity as a testable requirement for reliable MST-based recognition-element characterization.