besides common gel-problems, there are only a few pit-falls with EMSA.
Anyway, check whether your (purified?renatured?) protein is detectable by e.g. Western.
Pitfalls:
1) Your protein does not bind. Maybe due to extraction, renaturing, lacking modifications, etc. ...
2) Your nucleotide probe is not properly labelled and you detect only residual labelled nucleotides (mainly with radioactive labelling)
3) Your nucleotide probe is degraded.
4) Your protein binds, but your nucleotide probe is too big (>50 bp). Then you will not see a nicely separated band.
5) Miration of your complex is competed by other proteins - especially, when crude extracts are used
6) Try supershifts by adding an antibody that specifically detects your protein. This needs to be done anyway to show that the shift is really due to your prtein of interest.
Such supershifts sometimes lead to bands, that were cryptic before - but, in most of all possible cases, you don't see bands in EMSA because of oints 1) - 5).
Thanks to answer. I analysed my nuclear extracts by western and they were ok. I was thinking that maybe I have problems with salts, glycerol? I changed it, and I will see tomorrow.
You can use ficoll (4% final concentration) instead of glycerol in loading buffer to avoid smiles or/and change conditions of electrophoresis (different buffer: 0,5x TB, TBM, TB with Triton). Maby you should use also some additives to prevent nonspecific binding or aggregation of complexes (tRNA, BSA).