Response noise affects the graphical evaluation of response versus concentration curves

Several theories of chemoreceptor stimulation predict the same simple relationship between response R and stimulus concentration C:

R/Rmax=1/(1+K/C)

(1 + K/C),(in which R max is maximum response and K an equilibrium constant), also known as Beidler's taste equation. To test whether data points fit this equation and estimate R max and K, several transformations are in use which convert the theoretical curve into a straight line (Lineweaver-Burk, Scatchard, Eadie-Hofstee and Beidler plots). However, even modest amounts of response variability may interfere badly with the evaluation of these plots. This is not always appreciated; therefore this paper presents an illustration of the extent of this effect, using a realistic example. In addition, this effect may also obscure the presence of theoretically relevant deviations from the above equation, caused by multiple receptor sites, or convergence of receptors, which are expressed in a Hill coefficient nH ≠ 1. These effects are also exemplified. The illustrations are shown graphically and by a Montre-Carlo computer simulation. The conclusion is drawn that the linearizing plots should not be used at all for the quantitative evaluation of data. Direct, numerical iterative curve-fitting methods seem to give more reliable results.