Menu

Close

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

Preprint posted on March 28, 2020 https://www.biorxiv.org/content/10.1101/2020.02.28.969618v1

How are nucleosomes equally spaced across the genome? This new preprint demonstrates a ruler-like element of chromatin remodelers that may play a key role.

Selected by Katie Weiner

Background: A key feature of genomic organization is the appropriate localization of nucleosomes. Within eukaryotic cells, ATP-dependent nucleosome remodelers of the SWI/SNF, ISWI, CHD, and INO80 families, align nucleosomes to genomic reference points, such as promoters. However, how nucleosomes are equally spaced from a given reference location is poorly understood. In this work, the authors ask if remodelers have intrinsic ruler-like capabilities using an in vitro chromatin reconstitution assay.

Key findings: The authors define two ruler functionalities for ATP-dependent remodelers to set the distance between nucleosomes or between nucleosomes and a barrier: clamping (when a set spacing is established independently from nucleosome density) and remodeler-specificity (when individual remodelers establish different spacing patterns on the same chromatin). For Chd1, ISW2, and ISW1a, the authors demonstrate remodeler-specific clamping functions and define the intrinsic ruler length for the inter-nucleosome linker lengths of each factor (12-13 bp at all densities, 21-26 bp at all densities, and 54-58 bp at low/medium density, respectively). They further find that INO80 does not appear to have ruler-like properties. However, after investigating a set of INO80 mutants, they demonstrate that the HSA helix with minor contribution from the HMG domain has potential inter-nucleosome linker ruler-like activity.

To align nucleosome profiles, the authors induce two distinct barrier conditions: the addition of the DNA binding factor Reb1 or the introduction of a DNA break by the enzyme BamH1. Interestingly, both resulted in well-spaced symmetrical nucleosome arrays. As nucleosome spacing occurred independently of barrier type, the authors next sought to determine if spacing is DNA-sequence independent. To test this, the authors created chromatin from S. pome and E. coli plasmid DNA libraries. They observed no significant difference in nucleosome spacing or positioning, although note that the E. coli plasmids had a decreased nucleosome occupancy in certain conditions. Combined, these data demonstrate that nucleosome remodelers have a DNA-sequence independent remodeler-specific ruler length.

The authors speculate that the combinatorial contribution from each of the remodelers sets the in vivo linker lengths at various different genomic regions. They suggest that the short ruler length of Chd1 and ISW1a sets the ~18 bp linker length found within transcribed gene bodies and dense nucleosome arrays. In contrast, the longer ruler length of INO80 and ISW2 are more relevant in the context of aligning the +1 nucleosome adjacent to the nucleosome depleted region (NDR) barrier established by SWI/SNF.

They propose a model by which remodelers slide nucleosomes in preferred directions based on the distance from a barrier. They note, that the barrier can vary and may include other nucleosomes, histone modifications, DNA sequences, DNA shape, or DNA binding proteins. If the remodeler is within its intrinsic ruler reach from a barrier, the nucleosome will preferentially slide towards the barrier. However, once the remodeler gets too close to the barrier, the bias will switch and the nucleosome will slide away from the barrier. These opposing forces will create a stably positioned nucleosome at the remodelers set length.

What I like about this work: The article uses an in vitro technique to systematically test the intrinsic nucleosome positioning properties of key ATP-dependent remodelers in yeast. Yeast is a powerful model organism to better understand remodeler properties as there is a wealth of in vivo data available in which to compare their in vitro results. The authors establish the intrinsic ruler length for each of the tested remodelers for the first time, values that will be instrumental to determine the combinatorial contribution of these remodelers in vivo. Interestingly, the authors demonstrate that the remodeler’s activity is DNA sequence independent. This suggests that barrier regions may not be DNA sequence itself, but rather DNA binding proteins or other features such as previously positioned nucleosomes, DNA breaks, or DNA structure. Combined, this work moves in the direction of understanding the mechanisms by which cells phase nucleosomes. This is extremely important to understand as nucleosome disruption causes gene expression changes that often result in human disease such as cancer.

Future directions and questions for the authors:

  • The authors mention that the histone chaperone FACT was used in the purification of Chd1 and the figure legends indicate that FACT is still present in the nucleosome positioning assay. As FACT is a histone chaperone that also affects nucleosome positioning, it is possible that the observed Chd1 effects are partially caused by FACT. Have the authors added FACT to other chromatin assays or tried the purification with other histone chaperones/binding partners of Chd1 to try to separate these effects?
  • MNase-seq is a great technique to determine nucleosome spacing, but is less accurate at interpreting absolute occupancy. Have the authors thought of combining their MNase-seq approach with other techniques, such as histone ChIP, to better understand how remodelers affect occupancy?
  • One potentially exciting future direction would be to add modified histones or histone variants into the in vitro This could help determine if and how histone modifications/variants affect nucleosome positioning and how nucleosome remodelers interpret histone modifications/variants.
  • To test if their combinatorial model is correct, have the authors tried adding multiple remodelers onto the same chromatin. It would be interesting to see how the remodelers interact: do all remodelers affect each other or do only a subset interact? If they do interact, is it cooperative, additive, competitive, antagonistic, etc.?

Tags: chd1, chromatin remodeler, ino80, isw1a, isw2, yeast

Posted on: 28th March 2020

doi: https://doi.org/10.1242/prelights.17991

Read preprint (No Ratings Yet)




Author's response

Elisa Oberbeckmann and Philipp Korber shared

  1. The authors mention that the histone chaperone FACT was used in the purification of Chd1 and the figure legends indicate that FACT is still present in the nucleosome positioning assay. As FACT is a histone chaperone that also affects nucleosome positioning, it is possible that the observed Chd1 effects are partially caused by FACT. Have the authors added FACT to other chromatin assays or tried the purification with other histone chaperones/binding partners of Chd1 to try to separate these effects?

We are working on another paper that focuses on Chd1. There we tested whether nucleosome positioning by Chd1 is influenced by FACT. It seems that this is not altered by FACT, but Chd1 activity in general can be reconstituted more reliably in the presence of FACT. Hence, we decided to add FACT throughout our remodeler ruler study to obtain more reliable results.

We have not tested other histone chaperones and have not combined FACT with other remodelers so far. It is indeed possible that chaperones affect nucleosome positioning. For now, we wished to probe and show the nucleosome positioning information input of remodelers just on their own.

  1. MNase-seq is a great technique to determine nucleosome spacing, but is less accurate at interpreting absolute occupancy. Have the authors thought of combining their MNase-seq approach with other techniques, such as histone ChIP, to better understand how remodelers affect occupancy?

Indeed, the Korber lab published just recently an entire study dedicated to the question and measurement of absolute nucleosome occupancy (Oberbeckmann et al., 2019, Genome Research, PMID: 31694866). We established genome-wide determination of absolute occupancy using DNA methyltransferases (ODM-seq). (Using histone-ChIP for measuring absolute occupancy is tricky due to difficulties in calibrating IP efficiencies.) However, experiments for our remodeler ruler story were conducted before we established ODM-seq. Thus, we did not measure absolute occupancy in our remodeler assays here, but we plan to do this in a related context. Nonetheless, ODM-seq cannot replace MNase-seq in terms of assessing nucleosome positioning, as the ODM-seq resolution is limited to 9 bp on average. Thus, we still rely on MNase-seq to precisely determine nucleosome positions. Also note that remodelers with spacing activity do not have nucleosome eviction activity. Therefore, we do not expect changes in average occupancies, just changes in positioning.

  1. One potentially exciting future direction would be to add modified histones or histone variants into the in vitro system. This could help determine if and how histone modifications/variants affect nucleosome positioning and how nucleosome remodelers interpret histone modifications/variants.

Definitely! This is a direction we are exploring.

  1. To test if their combinatorial model is correct, have the authors tried adding multiple remodelers onto the same chromatin. It would be interesting to see how the remodelers interact: do all remodelers affect each other or do only a subset interact? If they do interact, is it cooperative, additive, competitive, antagonistic, etc.?

In our earlier study (Krietenstein et al., 2016, Cell, PMID: 27768892), we combined different remodelers on the same SGD chromatin and showed, for example, that ISW1a was dominant over ISW2 in terms of setting tight spacing, and that titration of increasing amounts of RSC against a constant amount of ISW2 could push the +1 nucleosome position further downstream. However, we did not try all remodeler combinations yet, and we did not take different nucleosome densities into account as Krietenstein et al., 2016 used only one nucleosome density, which was, in retrospect, lower than our high density in the remodeler ruler study. Thus, it might be worthwhile repeating these experiments with our new knowledge about the effects of nucleosome densities. Furthermore, we may include Chd1 now into these combinatorics, as we have now a version of Chd1 that is highly active in our genome-wide assays, in contrast to our study Krietenstein et al., 2016.

Have your say

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Sign up to customise the site to your preferences and to receive alerts

Register here
Close