Sulfolobus solfatarius Sso7d - dsDNA binding protein could be an option.

As you exclude (for unknown reason) histones, one of the more or less valid candidates, in my opinion, are restriction endonucleases that recognize non-complex sequences (4-mers, for example CGCG (BstUI), GGCC (HaeIII) etc) that, depending on the length of the DNA might be more or less frequently presented within the DNA (less complexity correlates with higher frequency). In the case, when the enzymatic activity of the respective endonuclease is inhibited or kept at minimum (by either temperature or divalent metal cation constrains or mutations that inhibit catalytic activity but have no effect on binding properties), such proteins might be of potential use to you. Of course, they are NOT 100% non-specific DNA-binding proteins, but given a rather limited complexity of their sites, one may consider them as quasi-non-specific. The only potential problem may arise from structural view on these candidates, but I do not know how important is that for you. Another potential candidate might be a synthetic peptide that carries numeral spaced Lys residues that will provide (cumulatively) reasonable binding characteristics, purely non-specific, as the binding will depend on their interactions with phosphate backbone. What do you think?

RPA, helicases.

Thanks everyone! I truly appreciate the inputs. Oleg, thanks for your thoughts. I cannot use histones as they will form inaccessible aggregates/clumps with DNA. Since my downstream assay is to assess binding of a certain protein to DNA, I cannot afford to distort the substrate DNA--whether topological, aggregative or due to cleavage. Therefore, an inactivated/mutated restriction enzyme might be thought of--thanks for the suggestion. I fear the multi-Lys peptide might form aggregate with the DNA, but it is worth trying. Shivasankari, thanks. Sorry, I did not mention that my substrate DNA is ds, and RPA binds ssDNA. For reasons I said above, I cannot use a helicase.

Rad51

RecA