Hi Caroline.

I have never use a gradient gel but whenever im running a SDS-PAGE i set my electrophoresis chamber and I put inside a Polystyrene fridge (thats the way it's called in my country) and then I add enough ice to make sure it's about the same level as my running buffer. Also I try not to use my running buffer more than 3 times.

Hope it helps.

It looks like a sample issue to me (not a heating issue). You may have more protein (or other components) in those samples that are running slower. The excess protein or other components may be enough to affect the pH of the gel in that region, and cause migration differences. This can be tested by running a well-behaved sample (a known protein, or marker) in all the lanes. If it is heat-induced, you should still see the problem. If it is due to your specific samples, then the problem should go away. Also, if it was heat-induced, you would not have such a big migration difference between lanes 4 and 5 (This would mean there is no heat generated in lane 4, but lots in lane 5, which is unlikely.)

If the migration problem is specific to your samples, then you may be able to precipitate your samples and re-dissolve them in running buffer. That will remove other components (salt, buffers, etc.) that may affect sample migration. If the problem is caused by the amount of protein in those lanes, you may want to try to dilute the problematic samples before loading them on the gel.

Good luck!

I imagine you could get this effect if the gel in the middle  shrank within the plates due to heating or buffer depletion effects then gas bubbles from the electrophoresis lodged in the gel plates so slowing down electrophoresis because the gel is no longer in contact with the buffer. Hopefully your apparatus will allow you to see this if it is the cause