It is long known that transcription factors (TFs) which bind to common DNA motifs in vitro occupy distinct sets of promoters in vivo. The reason for that is not clear. A new study may provide some hints. Spoiler warning: it involves intrisincally disordered regions.
Brodsky et al. reports in Molecular Cell their study of Msn2 and Yap1, two TFs of different families that contain long intrinsically disordered regions (IDRs) outside their DNA-binding domains. They found that the IDRs are both necessary and sufficient for localizing TFs to the majority of target promoters. They found that IDR-mediated binding does not dependent on any single localized domain in the IDR, but rather results from a multitude of weak determinants distributed throughout the entire IDR sequence.
Therefore, the authors proposes a two-step search model of TF-promoter binding: first IDR recognizes promoters via multiple weak interactions, and then DNA-binding domain binds to its preferential motif to regulate gene transcription.
The results are on one side not really surprising, but nevertheless reassuring and interesting, in the sense that non-DNA binding domains must somehow contribute to the promoter-specificity of TFs. On the other side, they are indeed intriguing that the intrinsic disordered regions contribute to promoter selective consistently and substantially in two transcription factors.
I think the results are even more intriguing because de Boer et al. reported their findings from 100 million random promoters that expression level is influence by weak regulatory interactions between TFs and promoter sequences. Are we looking at two levels of weak interactions - between IDR and promoter sequences, and between different TFs and promoter sequences - or just one level? I have no clear answer yet.
In any case, these studies seem to suggest that weak interactions between TF and promoter sequences matter.