Don't have a precise answer, but I would try out different acrylamide percentages and gradients first. 

There are a number of things you can try.  Easiest, as was said already, is to try different types of gels: here's a set of migration charts from BioRad that give you an idea of what sorts of gels will separate well at each mW region: http://www.bio-rad.com/en-us/category/protein-gel-migration-charts .  To my eyes, it looks like a 7.5% or 10% Tris-Glycine gel might be a good place to start.

You can also try running the gels in the cold room, with different currents, etc.  Much less predictable than changing the gel type, but it might help a bit.

You could also just switch system entirely - if the proteins are too similar in mass to separate by standard SDS, you could try a non-denaturing gel, where shape plays more of a role, and quaternary structure becomes critical.  Or, you could try isoelectric focusing, which separates based on isoelectric point, rather than size.  Or go to 2D gels, which separate on both axes.

Finally, since you're doing biotinylation, you could always probe with streptavidin rather than an antibody to the protein itself.  I find streptavidin to be spectacularly dirty as 'antibodies' go, despite how clean you'd expect it to be, but still, it SHOULD be at least somewhat selective for the higher weight, tagged protein.

I don't fully understand your question but you should be able to distinguish a 3kD difference in protein size with an 6% or 8% acrylamide gel when you run the 50kD marker down to the bottom.  Also, you can probe your western blots with streptavidin-HRP to insure that you are getting biotinylation.  

Thank you so much, everyone, especially Willian P Katt  for responding. I will go through the suggestion you have mentioned and then update my result.