The binding of salicylate ion to human serum albumin (HSA) was studied in 100 mM potassium phosphate buffer (pH 7.4, 25 degreesC), using equilibrium dialysis and fluorescence titration methods. The protein samples tested were (a) dialyzed human plasma and (b) a commercial preparation of HSA, essentially free of globulin and fatty acids. Independent of the analytical method used, Scatchard and nonlinear regression analyses of the data pointed to a single class of high-affinity salicylate binding sites. On the other hand, the binding parameters were found to be method dependent. K-d ranged between 25 +/- 2.4 and 62 +/- 15 muM in equilibrium dialysis and between 10 +/- 1.3 and 40 +/- 3.0 muM in fluorescence titration. (The higher limits refer to plasma samples at high [HSA]). Following the same pattern, the apparent stoichiometry of binding (though independent of sample identity and concentration) was higher in equilibrium dialysis (n(app) 3.2 +/- 0.10) than in fluorescence titration (n(app) 1.9 +/- 0.30). The difference between the two methods could be reconciled by invoking two distinct classes of binding sites (I and II), which had identical (or marginally different) I, values, while differing in the magnitude of the fluorescence signal (Deltaf) generated upon ligand binding (Deltaf, PLI = Deltaf(I); Deltaf, PLII = 0). Further, it was assumed that the state of occupation of class II sites affected the fluorescence efficiency of class I sites, such that Deltaf, PLI,II = beta Deltaf(I) (beta = interaction factor). A random binding scheme involving P, PLI, PLII, and PLI,II was formulated. The model adequately predicted the behavior of the system when monitored through the change in protein fluorescence: Taking K-d = 25 muM and n(T) = 3, the interaction factor beta was found to be 0.62 +/- 0.10. It was concluded that the correct parame-ters for the binding of salicylate ion to HSA are K-d = 25 +/- 2.4 muM and n(T) = 3.2 +/- 0.10, as indicated by equilibrium dialysis of purified HSA. Besides updating information relating to the salicylate binding potential of HSA, this study serves to illustrate a likely complication in the study of protein-ligand interactions by fluorometric methods. (C) 2001 Academic Press.