Not sure what you exactly mean, but if it is distinguishing antigens,, antibodies and complexes thereof on a non-reducing gel, it should work (presuming the antigen is large enough to make for a difference by apparent MW).
If you can do crosslinks with a bidentate crosslinker, it also should work on reducing gels)
Yes, you can do that.. but you won't see the bands of the same molecular weight of the protein. As primary antibodies are made against certain epitope of protein. So generally, you will see the bands of molecular weight little higher (due to epitope) than 25 KD (light chain of antibody) and little higher (due to epitope) than 50 KD ( heavy chain of antibody)