104
incorporated into the above regression schemes based on equation (4-6)
by simply setting C-| = 0 and concomitantly replacing By with B$ No
attempts were made to fit models containing two or more high-affinity,
saturable binding sites, since conventional Scatchard plots offered no
evidence of binding site heterogeneity.
In order to further examine binding isotherms for apparent "fine
structure" that might be indicative of heterogeneity or apparent
cooperativity, but that is not obvious from Scatchard or other simple
methods of analysis that assume a single class of noninteracting binding
sites, some isotherm data sets were transformed into continous affinity
distributions or spectra of equilibrium constants by the approximate
method of finite differences described by Hunston (1975) and by Thakur,
Munson, Hunston and Rodbard (1980). Because the approximate method of
finite differences generates a rather broad spectral line (instead of a
narrow spike or 6-function) even from "perfect" data derived from models
containing only a single class of noncooperative receptors, these
transformations of the experimental values of total binding (By) were
then compared with spectral transformations ("comparison spectra"
representating a single class of saturable, noninteracting binding
sites) of binding data (By) "predicted" by the 3-parameter nonlinear
regression model (described above) derived from the same set of
experimental data used to generate the "unconstrained" spectrum to be
analyzed. Thus, the comparison spectrum consists of a single broad
"line" having the minimum line width that can be "resolved" by the
finite differences method. Spectra A and B of figure 4-22 are
representative, respectively, of the comparison and unconstrained
spectra generated by transformation of the data from a single binding