Reply to Campos and Muñoz: Why phosphate is a bad buffer for guanidinium chloride titrations.

Liu et al. (1) alleged that they could not repeat our work on the protein BBL (Protein Data Bank ID code 1W4H) at the most basic level. Campos and Muñoz now admit (2), as we explained (3), that they used a phosphate buffer that greatly changed pH during the guanidinium chloride (GdmCl) titration relative to our Mops buffer. They can now repeat our work. However, the authors now claim (2), misleadingly, that the amine buffer Mops is inferior to phosphate in GdmCl titrations and that we miscalculated the effects of GdmCl on glass-electrode pH readings (4) (Fig. 1). The primary role of a buffer under conditions that can change the pKa of a reagent is to maintain the ionization state of that reagent with changing conditions, which may be very different from the role of maintaining “true” pH. Addition of salt affects amine, carboxylate, and phosphate buffers very differently. At low ionic strengths, as known experimentally and by the Debye–Hückel equation, the pH of amine buffers is unaffected by increasing salt concentrations, the pH of carboxylate buffers drops slightly, and the pH of phosphate buffers drops greatly. At high salt concentrations, specific interactions may alter recorded pH as well as pKa values. Accordingly, when performing titrations with salts such as GdmCl, one should use buffers that are chemically similar to the protein’s ionizing groups that titrate at the buffer pH. Then, pKa values of buffer and protein change in parallel, so minimizing changes in ionization state of the protein during the salt titration. In the pH range of 3.6–5.6, for example, acetate is a good buffer because it is a chemical mimic of the carboxyl groups of the protein that titrate in this region. Around pH 7, imidazole and other amine buffers are best because histidines and N-terminal α-amino groups titrate in this range. Phosphate is particularly bad at pH 7 because it behaves so differently from amines during salt titrations (Fig. 1). We can show formally that changes in true or recorded pH readings are irrelevant and that the buffer ratio and chemical nature are the key factors. Suppose group HB on the protein ionizes with thermodynamic dissociation constant KB and buffer HA ionizes with KA. Then: