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

al., 1997; Tsukiyama et al., 1999; Udugama et al., 2011; Varga-Weisz et al., 1997) , but not of the SWI/SNF family, show spacing activity. This activity was suggested to rely on a length-sensor mechanism (Yang et al., 2006; Zhou et al., 2018) where nucleosome sliding rate is regulated by linker DNA length. Sliding one nucleosome back and forth between two other nucleosomes, with a linker length-dependent velocity, would center a nucleosome at steady state when both flanking linkers have the same length. While the length-sensor mechanism may equalize linker lengths and thereby generate spacing distance regularity, it does not by itself determine spacing distance length in absolute terms. This would reciprocally depend on nucleosome density. However, spacing in vivo (Gossett and Lieb, 2012; Hennig et al., 2012; van Bakel et al., 2013) , as well as generated in vitro (Lieleg et al., 2015; Zhang et al., 2011) , remained constant despite changes in nucleosome density. This was called "active packing" (Zhang et al., 2011) or "clamping" (Lieleg et al., 2015) , but it remained unclear if remodeler or nucleosome features led to such density-independent spacing. Structural studies suggested that the yeast ISW1a remodeler contacts a neighboring nucleosome and may set the linker length by a "protein ruler" (Yamada et al., 2011) . Two ISWI family remodelers, yeast ISW1a and ISW2, each generated regular arrays aligned at DNA-bound Reb1 or Abf1 in vitro, but with different spacing at the same nucleosome density . This points towards a remodeler-specific linker length determining ruler mechanism. Also suggestive of a built-in ruler, INO80 required a minimum linker length for nucleosome sliding (Zhou et al., 2018) and recognized linker DNA via a structural module that was important for sliding . The ruler metaphor may indeed describe a remodeler mechanism that measures and sets the phasing and spacing distances of arrays. However, so far it is mainly suggestive and has to be substantiated in molecular terms. This would be exceedingly convoluted in vivo but requires a defined system that allows to assay the generation of phased regular arrays by remodelers and to dissect if and how a ruler mechanism is at work. Are there rulers within some or all remodelers with spacing activity? Are linker length vs. distance to barrier determined in the same or different way? Are rulers autonomous or does the outcome depend on nucleosome density or underlying DNA sequence? Ultimately, is it possible to tune a ruler, i.e., can a remodeler be mutated to generate arrays with altered spacing and/or phasing distances? Here, we used genome-wide in vitro chromatin reconstitution with purified remodelers ( , accompanying paper Oberbeckmann & Krietenstein et al.) to answer these questions. All yeast remodelers with spacing activity, ISW1a, ISW2, Chd1, and INO80 have rulers that are largely autonomous regarding underlying DNA sequence but some may respond to nucleosome density. Remodeler-specific rulers mechanistically explain earlier in vivo observations. Structure-guided mutations in recombinant INO80 complexes led to shorter or longer spacing and phasing distances and showed that these quantities may be uncoupled. Finally, we propose a model how remodeler rulers position nucleosomes by regulating sliding direction bias according to (epi)genetic information in the nucleosome environment. purified yeast remodelers (Figure S1A), and the barrier Reb1 or the restriction enzyme Bam