Origins and consequences of ligand-induced multiphasic thermal protein denaturation.

The presence of subsaturating levels of a high-affinity ligand has been demonstrated both by experiment and calculation to have far-reaching consequences on thermally induced protein denaturation due to the coupling between the protein denaturation and ligand-binding equilibria. Under such circumstances, a protein may undergo biphasic denaturation even though in the absence of ligand it exhibits a thermogram comprised of a single essentially symmetric endotherm. Up to now, the presence of just 2 maxima in the thermogram has been presented merely as an experimental observation or as the result of equilibrium computations. Here we develop a thermodynamic description of the linkage between these equilibria in which the number of cusps present in the thermogram correlates with the number of resolved steps in the plot of saturation level of remaining native protein vs temperature (i.e., the thermal binding curve). During thermally induced denaturation, the concentration of native protein decreases; thus, the native protein in effect is titrated with ligand at constant total ligand concentration. The free ligand concentration The free ligand concentration increases substantially through the release of bound ligand by unfolding protein thereby increasing the saturation level of the remaining native protein. The form of this thermal binding curve is a function of the number of ligand-binding sites on the protein, the magnitudes of the association constants, and the total ligand and total protein concentrations. As a result, the model predicts multiphasic denaturation of a single cooperative unit when the thermal binding curve consists of discrete multiple steps. The presence of only 2 maxima (i.e., a single cusp) in a thermogram for a protein with multiple sites on the native species derives from the form of the thermal binding curve, which in this case is a single-step sigmoidal plot, and not from the predominant denaturation of unliganded and fully liganded native species. In addition, it is shown that, in general, the contributions from the denaturation of individual native protein species are decidedly non-two-state in character; thus, simple deconvolution should not be carried out. The effects of nonzero values of delta Cp and d delta Cp/dT for denaturation and of changes in enthalpy and in heat capacity for ligand binding, as well as the interaction of ligand with the denatured protein, are explored also.