Ruler elements in chromatin remodelers set nucleosome array spacing and phasing

Elisa Oberbeckmann; Vanessa Niebauer; Shinya Watanabe; Lucas Farnung; Manuela Moldt; Andrea Schmid; Patrick Cramer; Craig L. Peterson; Sebastian Eustermann; Karl-Peter Hopfner; Philipp Korber

To better assess distances to barrier (phasing) and linker lengths (spacing), we aligned the MNase-seq data for each remodeler/barrier/density combination to either in vivo Reb1 sites or BamHI sites ( Figure 2A ). For each replicate ( Figure S2A -C), we called nucleosome peaks and determined the distances to barrier and linker lengths as defined in Figure 2B . All remodelers symmetrically aligned regular arrays to BamHI sites, which are palindromic and therefore inherently symmetrical, and most of them also to Reb1 sites (Figures 2A, S2A ,B) regardless of site orientation and position relative to genes (groups 1 to 3; Figure S3A ,B). However, if INO80 aligned arrays at promoter Reb1 sites (groups 1 to 3, Figure S3A , accompanying paper by Oberbeckmann & Krietenstein et al.) , nucleosome occupancy (peak height) was higher over genic versus non-genic regions at low and medium nucleosome density leading to asymmetric patterns with regard to peak heights in groups 1 and 2. Reb1 site orientation had no effect (group 1 vs. 2). This asymmetry in nucleosome occupancies reflected that positioning of +1 nucleosomes, per definition the first nucleosomes downstream of transcription start sites, i.e. at gene starts, was not only guided by Reb1 bound to promoter sites but also synergistically by underlying DNA shape features (accompanying paper Oberbeckmann & Krietenstein et al.) . We recapitulated here that INO80 was able to position in vivo-like +1 nucleosomes in the absence of a barrier at low and medium densities ( Figure S2C ,D). This synergism between Reb1-and DNA shape-guided +1 positioning at low and medium density resulted in higher occupancy at the +1 nucleosomes, which are alignment points for +2 nucleosomes and so on. Therefore, all array peaks over genes were higher than their counterparts over non-genic regions. However, such synergism was not seen at high density where in vivo-like +1 nucleosomes positioning by INO80 alone was much less pronounced ( Figure S3C ,D). This inability was not due to a general inability of INO80 to slide densely packed nucleosomes as INO80 could generate Reb1aligned arrays at these high nucleosome densities, too (Figures 1C, D, 2A, S2A, B) . Nonetheless, this activity was apparently incompatible with or dominant over DNA shape-guided nucleosome positioning (see Discussion). This showed again that our here generated high nucleosome density was higher than the nucleosome density used previously , otherwise in vivo-like +1 nucleosome positioning by INO80 would not have been clearly observed in our earlier study. Heat maps of MNase-seq data for SGD chromatin of the indicated nucleosome density and after incubation with indicated remodelers and Reb1. Chd1 refers to the Chd1/FACT complex. Heat maps are aligned at in vivo +1 nucleosome positions and sorted according to decreasing (top to bottom) anti-Reb1 SLIM-ChIP score shown in leftmost heat map. Horizontal red shading highlights genes with strong in vivo Reb1-binding in their promoters. Merged data of replicates are shown, individual replicates in Figure S1B ,C. (D) Composite plots for MNase-seq data averaged over the indicated number of replicates (n) as in panel B but only for genes highlighted in red in panel B.