The use of gel chromatography for the determination of sizes and relative molecular masses of proteins. Interpretation of calibration curves in terms of gel-pore-size distribution.

The separation of proteins by gel-exclusion chromatography has been explained in terms of partitioning of the macromolecules within the gel by a distribution of pores of various radii. The assumption that the distribution of pore sizes is Gaussian has led to the prediction of a linear relationship between the molecular Stokes radius (RS) of the protein and the function erf-1 (1-KD), where KD is the partition coefficient [Ackers (1967) J. Biol. Chem. 242, 3237-3238]. Since careful calibrations of classical (agarose and dextran) gels and h.p.l.c. gels have shown that such a linear relationship is not verified experimentally over a wide range of native protein sizes, we have reinvestigated the model of Ackers (above reference). We show that Ackers' (above reference) derivation is not valid except for a particular Gaussian distribution of pore sizes centred at the origin. Relaxation of this restriction to allow for other types of Gaussian distributions cannot account for the non-linear calibration curves that we have obtained. Instead we show that the pore-size distribution can be calculated from the experimentally determined function KD = f(RS) and that this distribution is bimodal (non-Gaussian). One distribution is centred below 2 nm, whereas the mean value of the second one is around 6-8 nm. The minimum in this bimodal distribution corresponds, for some gels, to a region of poor resolution, which needs to be appreciated for the proper use of gel chromatography in the determination of molecular size.