Charge-based protein separations

I recently performed an experimental separation with 4 proteins in 100 and 10 mM KCl.  All proteins are negatively charged at the pH of the experiments, and we would expect that less would pass through the membrane in the 10 mM KCl solution.  Here’s the gel:

You can see that the beta-gal in particular does not make it through the 10 mM KCl experiment.  For reference, here is a histogram of the pore sizes:

I’ve recently been working on including an electrostatic contribution into my diffusion model, and I’ve finally had some success.  Briefly I’ve included an electrostatic component in the steric term of the hindrance (see this post).  I’ve used neutral axisymmetric theory for the frictional component, as literature suggests that like charges should push the molecule toward the center anyhow.  I’ve also ignored the effects of electrokinetic relaxation during the movement of the particle, because I’ve read that it does not have a large effect in the size range/debye length that we’re operating at.  However these are important assumptions to keep in mind.

The above figure shows the prediction for 100 and 10 mM KCl solutions.  Zeta potentials of -20 mV were used for both membrane and molecules in the calculation.  Once membrane diffusion coefficients were calculated, the simulations were performed in the 3D COMSOL model, and the figure shows the sieving coefficients obtained from the model and the densitometry of the above gels.

Results:

1. The experimental results are similar to the model prediction, though the 10 mM results fit better than the 100 mM.  The 100 mM results diffuse quicker than even free diffusion expects.

2. There’s not a huge difference between the two experimental separations, nor does the model predict there to be.  In the past we thought this difference would be more pronounced.