Estimation of ligand binding parameters by simultaneous fitting of association and dissociation data: a Monte Carlo simulation study.

A new procedure for analysis of ligand binding kinetics was evaluated by Monte Carlo simulations. In this, all association and dissociation data were fitted simultaneously to a set of nonlinear equations. This should have several advantages over more conventional methods; data are better used in a single fitting procedure in which the degrees of freedom are maximized and the error term is spread over more observations; all relevant parameters (Bmax, k1, and k-1) are obtained directly; values obtained from measurements are not treated as errorless; and it yields a single residual term that can be used for statistical comparison among binding models and/or experiments. We have compared this approach with the common practice of analyzing the association and dissociation phases separately, either by nonlinear regression or by linear regression after suitable transformations. With respect to both the precision and accuracy of parameter estimates, the simultaneous procedure was superior to the other two methods. The properties of the simultaneous procedure were further investigated, concerning both parameter estimation and the probability of reliably detecting a second binding site. For the latter, the relative density of receptor subtypes and the dissociation rate constants were found to be of major importance, whereas association rate constants and ligand concentration were of minor importance in this respect. The probability of resolving two sites by kinetic or equilibrium data under similar conditions with the aid of a single labeled ligand was examined. When the selectivity of the ligand was low, the resolution was found to be more probable when based on kinetic, rather than equilibrium, data. This was true at higher selectivities as well, provided kinetic data were obtained at two different ligand concentrations.