BIOLOGICAL RECEPTORS: IMPLICATIONS OF RADIOLIGAND QUALITY IN BINDING ASSAYS

Abstract

We tested the ability of different fitting procedures to discern differences in the binding affinity between the labeled and unlabeled ligand in a receptor assay, in order to obtain the correct value of the radiotracer specific activity. This point is critical for the correct conversion of the estimated binding constants to meaningful units. We simulated sets of receptor assays which considered distinct experimental situations, and fitted the data with two models considering either a unique or two different affinities for the two classes of ligand. We compared direct non-linear regression, linear robust regression on transformed data, and an experimental approach called pseudo-saturation (labeled plus unlabeled ligand in constant ratio) assay. It emerges from the results that none of these methods would be sensitive enough, even in commonly acceptable error levels in the measured variables, to indicate the appropriate model with a reasonable certainty. The a priori consideration of a different binding capacity for both categories of the ligand gives an acceptable estimation of the parameters, regardless of the fact that the fitting criteria reflect a better fitting of the other model. We conclude that a two-affinity binding model should be routinely fitted to experimental data and the binding constants compared to the simple binding model when the specific activity of the radiotracer needs to be assessed.