On the potentially spurious contributions of cosolutes in protein 15N relaxation dispersion measurements

A rigorous quantification of the thermodynamics, kinetic and structural changes can be approached by relaxation dispersion methods, when such motions are in the microsecond to millisecond timescale. Albeit some past efforts, it is still unclear how cosolutes modulate the fitted parameters extracted from relaxation dispersion analyses. Here, we have studied how the systematic measurement of ¹⁵N relaxation dispersion on a well-studied enzyme, undergoing two-state chemical exchange, are affected in terms of the populations of the minor state (thermodynamics) and the exchange rates (kinetics) in the presence of different cosolutes, from ions to denaturants. The cosolute-induced changes are modest but can significantly affect protein function. Specifically, exchange rates can be associated to a subtle leverage in the exchange rates. For most canonical residues, i.e. without spurious effects caused by the cosolute, the chemical shift difference ($\Delta \omega$) between both states is essentially unaffected with respect to the expected shift of water resonance. Yet, solvent exposed residues do not always follow this canonical trend. Instead, residue-cosolute interactions are large enough to induce a residue specific shift that ultimately modulates the fitted parameter $\Delta \omega$. Unfortunately, solvent accessibility analysis is not accurate enough to a priori discriminate between canonical and non-canonical residues and a full experimental analysis at varying concentrations of the cosolute is required. This observation aims to act as a word of caution for measurements of relaxation dispersion at a single-solvent condition that contains large amounts of cosolutes.