My guess- since digitonin is frequently used to solubilize membrane protein complexes, it will preserve protein protein interactions during CO-IP as well. It might be more suitable for CO -IP of membrane proteins. However, I am not sure why one would use digitonin in washing buffer. The lysis buffer should contain digitonin from 1% to 5% (concentration should be optimized according to the protein of interest), NaCl not more than 50 mM (for membrane proteins) and pH should be around 7 (50 mM Bis-Tris neutralized with HCl).

Thank you Tushar! My bad that I didn't specify the issue. I'm now co-IPing a cytosolic protein (RhoGDI) and that with a membrane protein (TrkB-t1) which has a very short intracellular domain, the latter could also exists in the intracellular organelles. I tried 1% digitonin to lyse and wash cells as recommended by published papers. The combination of the buffer is kind of weird or rare, it goes with 10mM triethanolamine, 10mM Iodoacetoamide, which are not usual, the rest are the regular NaCl, EDTA and proteinase inhibitors. It didn't work out, neither did it work with 1% NP-40, 0.3% Chaps, or 1% Triton. The antibodies were not a problem, cuz it could IP the target protein very well. So I might need to try multiple methods before I go to the cross linking stuffs. Thanks! 

Thanks Srgjan! My antibodies work fine for IP and WB, I can clearly see the strong bands both on input and pulldown respectively, and not in the IgG lanes. Just that they can't pull down each other. I always put phosphatase inhibitors in the lysis and washing buffer, and I don't feel like the presence of GDP/GTP were an important issue in this experiment. I will definitely optimize the lysis buffer conditions like you say and try more times. Thanks!