Pressure denaturation of proteins: Evaluation of compressibility effects

One of the key pieces of information from pressure denaturation experiments is the standard volume change for unfolding (Delta V-o). The pressure dependence of the volume change, the standard compressibility change (Delta K-T(o)), is typically assumed to be zero in the analysis of these experiments. We show here that this assumption can be incorrect and that the neglect of compressibility differences can skew the interpretation of experimental results. Analysis of experimental, variable-pressure NMR data for bovine pancreatic ribonuclease A in (H2O)-H-2 at pH* 2.0 and 295 K yielded the following statistically significant, non-zero values: Delta K-T(o) = 0.015 +/- 0.002 mL mol(-1) bar(-1), Delta V-o = -21 +/- 2 mL mol(-1), and Delta G(o) = 2.8 +/- 0.3 kcal mol(-1). The experimental protein stability is in good agreement with one (Delta G(o) = 2.5 kcal mol(-1)) determined independently for the same protein by calorimetry at atmospheric pressure under equivalent conditions [Makhatadze, G. L., Clore, G. M., and Gronenborn, A. M. (1995) Nat. Struct. Biol. 2, 852-855]. The positive value for Delta K-T(o) indicates that the denatured form of ribonuclease A is more compressible than the native form; this is explained in terms of an interplay between the intrinsic compressibility of the protein and solvation effects. When the same data were fitted to a model that assumes a zero compressibility change, the Delta G(o) value of 4.0 +/- 0.1 kcal mol(-1) returned by the model no longer agreed with the independent measurement, and the Delta V-o returned by the model was a Very different -59 +/- 1 mL mol(-1). By contrast, it was not possible to carry out a similar thermodynamic analysis of fluorescence spectroscopic data for the denaturation of staphylococcal nuclease to yield well-defined values of Delta G(o), Delta V-o, and Delta K-T(o). This limitation was shown by evaluation of synthetic data to be intrinsic to spectroscopic data whose analysis requires fitting of the plateaus at either side of the transition. Because NMR data do not have this requirement, they can be analysed more rigorously.