Extended Data Fig. 5: MNase-seq, ODM-seq and RNA-seq analysis.

a, MNase-seq data for whole yeast genome reconstitution show that INO80 could not generate nucleosomal arrays phased to Abf1 sites without Abf1. Two independent replicates are shown, as in Fig. 6b, but with the indicated conditions. b, The INO80 remodeller concentration was titrated to assess whether Abf1 IDRs contribute to INO80 recruitment. If this were the case, differences in ± IDR at low INO80 concentration would be expected, whereas those differences should vanish in excess INO80. Our results here do not show this behaviour and suggest that the IDRs of Abf1 are not required for remodeller recruitment. This panel is as in Fig. 6b, but with the indicated concentrations of INO80. The fact that wild-type Abf1 vs. ΔIDR1/IDR2 overlapped almost perfectly across different INO80 concentrations indicated that we were not in a regime in the experiment shown in Fig. 6b where INO80 concentration was so high that IDR-dependent recruitment to Abf1 was masked. n = 1 for each condition c, Replicate in vitro reconstitution experiment as in Fig. 6b, but with some altered conditions (for example, Drosophila histones), as detailed in the Methods. d, When Abf1 was depleted from the nucleus, the NFR around Abf1 sites filled up with nucleosomes, whereas the flanking nucleosomes (−1, +1) and up- and downstream nucleosome arrays became disordered. These results are reported as decreased peak-to-trough ratios and shifts of occupancy peaks towards the NFR. This panel is as in Fig. 6e; that is, the data are aligned at responder Abf1 sites (classes I and II; n = 397) as identified by Kubik et al. based on MNase-seq data in an analogous Abf1 anchor-away strain and oriented such that a gene is to the right for unidirectional promoters⁶⁹. e, As in d, but for non-responder sites (classes III to V; n = 478), as defined by Kubik et al.⁶⁹. f, Different Abf1 constructs enhance or suppress RNA levels for genes (n = 279) linked to Abf1 responder sites (d and e; Methods). Shown are the changes in total RNA levels in Abf1 anchor-away strains with the indicated Abf1 constructs left in the nucleus relative to a strain with wild-type Abf1 left in the nucleus. n = 3 for FUS¹⁻¹⁶³12E + Y/M, no Abf1; n = 2 for all other constructs. NCS506 is a random mutagenesis construct (Fig. 5d) with just 13 point mutations. Boxes span the middle two quartiles; whiskers extend to 1.5× the interquartile range; and horizontal bars denote median log₂[fold change] values of the Abf1 variant relative to wild-type Abf1. The naive expectation that the degree of downregulation in RNA-seq data for these genes with Abf1 responder promoter sites would scale with the degree of viability in Abf1 IDR variants was not borne out. Instead, there were reproducible but idiosyncratic transcription responses. In hindsight, such varied outcomes may even be expected given that total RNA-seq after 75 min of rapamycin treatment should include indirect effects. In addition, the range of tested IDRs may mediate more or less specific interactions with various partner proteins, exerting diverse effects depending on the Abf1 sites, that is, where in the genome these IDRs are recruited via the Abf1 DBD. Different IDRs need not necessarily lead to a loss of function in terms of an inability to recruit native Abf1 partners, but also a gain of function in recruiting other nuclear components.