Reverse the electrode. If size of your protein is above 50 kDa run a 10% Native PAGE.

If your protein binds infrequently to DNA (and which DNA will you use for this???) it might not change the neg charge on the DNA sufficiently to affect its migration, so you might not need to switch electrodes.

If your protein is binding to DNA in the cell itself or a large fragment of DNA as in that case it will be difficult for the protein to migrate in PAGE and Navite PAGE might not help. You have to perform AGROSE DNA gel. In case you want to see shift in the migration pattern of protein upon binding of DNA you can  do gelshift assay or EMSA. 

Thanks Mrityunjay and Jeoff, I did try agarose gel electrophoresis and it didn't work. I can use either 45 bp / 18 bp DNA sequences. The mode of binding is non-specific- protein binds to the P-sugar backbone, may be in multiple nos. with a single DNA molecule. Protein size is small, less than 20 KD.

This new information would have been helpful upfront!!

We run 12.5 or 15% acrylamide gels using DS oligos from 13 to 23 (and higher) bp. These are HEX tagged and are easily detected on a phosphorimager. Our proteins are 13 or 22 kDa and the shift is easy to detect. To reduce the slight chance of dissociation as the gel heats up, we run with ice cold buffers (TBE or TAE). But, unlike yours, these proteins usually bind only once per oligo (unless large amounts of protein are used).

If you stay with normal polarity, you might see a decrease in oligo band intensity as you increase your protein concentration - that would at least hint at some binding. Then if you reverse polarity, you might see some new bands in lanes that contain oligo and protein. I think the problem may be that a smear would be expected until all the sites on the oligo are occupied, when a distinct band would be expected. with an 18-mer binding your 20kDa protein on each phosphate you would have (maximally) a 30kDa complex, so Mrityunjay is right to suggest a 10% gel. However, I would overlayer this with a 5% gel, just in case migration into the 10%  is very slow.