Effect of binding of ions and other small molecules on protein structure. VII. The role of aliphatic acid binding and of convection in the electrophoresis of serum albumin at low pH.

Bovine serum albumin undergoes two electrophoretically distinguishable types of interaction in acidic media (1). The first (l-6), which is also shown by ovalbumin, bovine y-pseudoglobulin and oxidized ribonuclease, is observed only in media containing acetate buffer (NaAc-HAc) or other carboxylic acid buffers. The various peaks shown by the electrophoretic patterns obtained in these media correspond neither to single stable protein components nor to single components involved in a slowly adjusted equilibrium as evidenced by the results of fractionation experiments (3, 6). Resolution of the peaks is intimately related to changes in conductance and pH produced in the Tiselius cell by the electrophoretic process. This should not be interpreted to mean that bovine serum albumin and other proteins do not undergo some interaction with these media; rather, it would appear that resolution of their electrophoretic patterns into multiple peaks results from coupling of such interactions with electrophoretic transport of the small ions of the solvent medium for the protein. The electrophoretic patterns have been interpreted qualitatively in terms of reversible binding of undissociated buffer acid by the protein a-ith the assumption that the resulting protein-acid complexes have more positive electrophoretic mobilities than the uncomplexed protein molecules (5, 6). This interpretation predicts that during electrophoresis the pH at certain levels in the Tiselius cell will change in such a manner as to induce a type of nonenantiography in the two patterns typical of that actually observed under a variety of experimental conditions. Furthermore, observed changes in pH are in the direction predicted by the model. Since it is assumed that protein-acid complexes exist in instantaneous equilibrium with uncomplexed protein molecules and undissociated buffer acid, the various peaks in the patterns should constitute a single reaction boundary in which a homogeneous phase cannot be generated between any two peaks. Consequently, the protein gradient should not become zero between the peaks. Although this last prediction has been realized under a variety of conditions of buffer composition (see, for example, Fig. 2 (2); Figs. 1, 2, and 4 (3) ; and Fig. 2B (I)), the gradient does appear to become