Shifting nucleosomes along DNA is an essential part of gene expression in eukaryotes — it allows regulatory proteins to gain access to previously inaccessible sequences. Although this nucleosome relocation is known to be the job of ATP-dependent chromatin remodellers, uncertainty has surrounded the question of whether these proteins are specific in terms of the sites to which they move nucleosomes. An investigation of mammalian remodellers now provides evidence that these proteins do show such specificity, and that this is directed by the sequence of the DNA substrate.

If repositioning by remodellers is determined simply by which DNA sequences have the highest affinity for nucleosomes, then, for a particular substrate, different remodellers should relocate nucleosomes to the same high-affinity positions. Rippe, Schrader and colleagues tested this possibility in vitro for seven mammalian remodelling complexes on two DNA substrates: the heat shock protein 70 (hsp70) fragment from Drosophila melanogaster and the mouse ribosomal DNA (rDNA) promoter. Following repositioning by the remodellers, relocation was indeed mainly to sites with high affinities for nucleosomes. However, the seven complexes gave distinct patterns in terms of which combination of these sites were occupied — evidence that the remodellers themselves have a role in determining the new location of nucleosomes.

One potential explanation for this remodeller specificity is that the enzymes are directed by DNA sequence information. On the rDNA substrate, one site of nucleosome positioning by the ACF chromatin-remodelling complex is strongly correlated with a DNA region that is intrinsically curved, with the repositioned nucleosome centred close to the peak of the curvature. The authors took a 40-bp fragment that spanned this peak and moved it into a new sequence environment. As in the rDNA context, ACF positioned a nucleosome close to the curvature peak, and the same result was found when the 40-bp sequence was placed into two other sequence environments. So, it seems that nucleosome repositioning is indeed directed by DNA sequence elements.

Finally, the authors tested two models for how such elements might direct remodellers. One possibility is that the enzyme is released at a particular position because of a low binding affinity for the sequence, thus determining the relocation end point (the 'release model'). Alternatively, the end point could be specified by the remodeller moving into a region that provides a poor substrate for the translocation of the enzyme, so that the remodeller comes to a standstill (the 'arrest model'). In the case of the remodellers chromodomain helicase DNA-binding protein 1 (CHD1) and ACF, among the potential sites of occupancy, nucleosomes were repositioned to the sites with the lowest binding affinity for the enzyme — consistent with the release model.

So, it seems that ATP-dependent chromatin remodellers do more than just providing the brawn when it comes to nucleosome positioning. The diversity of these enzymes and the complexes that they participate in suggest that their repositioning specificity provides an important level of gene regulation.