An examination of experimental design in relation to receptor binding assays.

1. Real and computer-simulated data were used to examine the efficiency of designs for receptor binding assays. 2. Initially, several different concentrations of [3H]-ketanserin were used in receptor binding studies, using membrane preparations from rat cerebral cortex to establish the form of the binding curves and to investigate the relationship between the variance of the binding measurements and their means. 3. The data showed that the specific binding could be modelled by a simple rectangular hyperbola, and were consistent with the assumption that the binding measurements have a constant coefficient of variation (0.1). 4. Computer-simulated data, with a coefficient of variation of 0.1, were then used to look at the precision of estimates of KD and Bmax obtained through the use of assay designs based on replicate incubations at two radioligand concentrations, or through the use of a geometric sequence of 5-7 radioligand concentrations. In each case, the influence of varying amounts (3.8-50%) of non-specific binding at KD on the precision of these estimates was monitored. 5. The results (a) illustrate the problems which arise in the analysis of receptor binding data when there are relatively high amounts of non-specific binding in combination with a constant coefficient of variation, (b) calculate the errors involved and (c) quantitate the relative merits of 2, 5, 6 and 7 point saturation curves in the estimation of Bmax and KD.