Usually tubulin, and the solution is to titre your primary and/or secondary until it disappears, since it is, as you suspected, an abundance problem.

Thanks for the suggestions!  I did try filtering my BME to remove keratin contamination a while back, but that did not help.  Since I've replaced both RIPA and Laemmli and always use gloves, I'm not sure what else could be introducing contamination between collecting the lysates and running the gels.  I have not done a Coomassie stain, but I expect that the band would be there too; this protein must be a component of the sample when it gets loaded into the gel since it appears as such a sharp band.  Unfortunately my targets (including phospho-Akt) are apparently quite a bit lower in abundance, so reducing the antibody concentrations could be tricky.

Albumin runs around there, actually 66kD. It will be in anything but especially cell culture samples and any tissue that is well vascularised. It will always be present in ponceau or other protein stains sadly. For antibody staining Id recommend blocking in 5% bsa for as long as you can, then using the most dilute anybody concentration you can get away with. Using bsa to mix your antibodies will help also. If you are looking at proteins in the high 40s range, (isnt akt in the 40s?) Id cut the blot at the 60 kd line before staining. Let me know if that works! Our paper in MBoC used that method.