Multi-subunit remodellers

CHD family members are all single subunit remodellers and for the multi-subunit ISWI remodeller family, only the ATPase domain has been structurally resolved so far.

Therefore, until now structural information on the organisation of the auxiliary domains of multi-subunit remodellers were only available for the INO80 family. The structure of the full RSC remodeller in complex with the nucleosomes solved here delivers insights into the SWI/SNF family. This enables us to appreciate how energy gained by the same basic remodelling motor is transformed into very different remodelling activities based on the specific subunit combination.

Both families use an extensive multi-subunit body as an assembly scaffold. While in the INO80 family the scaffold is formed by the heterohexameric Rvb1/Rvb2 complex to which the other subunits are attached, the RSC backbone is made mainly by the two Rsc8 proteins intertwined with the other subunits. One example is the anchoring of the ATPase subunits. Whereas the N-terminal part of Sth1 largely contributes to the scaffolding network of RSC, the Ino80 ATPase is anchored to the heterohexamer via the long insertion between the two ATPase lobes. Interestingly, the shorter linker between the ATPase lobes in Sth1 forms the protrusion 1 region that plays a role in the regulation of the enzyme.

All multi-subunit remodellers show several, yet very distinct interactions with their nucleosome substrate. In addition to the ATPase module, INO80 contacts the nucleosome at SHL –2 and –3 with a module formed by the Arp5 and Ies6 subunits (Ayala et al., 2018; Eustermann et al., 2018). These interactions are mirrored by the Arp5-Swc6 proteins in SWR1 (Willhoft et al., 2018). During remodelling, DNA gets translocated from the linker into the nucleosome against those modules in small steps. SWR1 uses a restrained translocation against the block to induce histone distortions that facilitate the exchange of histone variant H2A to H2AZ, and no net translocation is observed (Willhoft et al., 2018). INO80 likely also uses the created internal DNA loop to facilitate the exchange back to H2A histones, but the accumulated DNA gets released in larger ratchet steps of 10 – 20 bp to achieve nucleosome sliding (Eustermann et al., 2018).

Since RSC is not involved in the exchange of histone variants no such roadblock-like interaction with the nucleosome would be expected. Indeed, the RSC structure only shows contacts to the nucleosomal DNA with the ATPase module.

However, RSC extensively interacts with both sides of the octamer face via the finger helix and the SnAC domain. While the engagement of the finger helix with the acidic patch of the inner octamer face could be resolved in the presented RSC-nucleosome structure, how exactly the SnAC domain interacts with the outer octamer face is still elusive. The density suggests a helical structure at the acidic patch as well, but higher resolution structures are still needed to determine that with certainty. The extended octamer interaction interface of the SnAC domain inspires the fascinating model that this interaction might contribute to the differential recognition of H2A.Z and H2A variant containing nucleosomes. In this way, specific targeting of RSC to the promoter nucleosomes (which are enriched in H2AZ), but not to other nucleosomes along the gene body, may be achieved. Similar models have been suggested for the INO80’s sensor foot (Eustermann et al., 2018).

The HSA region of INO80 is involved in remodeller control (Clapier et al., 2016;

Knoll et al., 2018), and is bound by the ARP modules of the corresponding remodeller.

The INO80 ARP^INO80 module is composed of Arp8, N-actin and Arp4 proteins and binds to a segmented HSA helix (Knoll et al., 2018). In contrast, the RSC ARP^RSC module is built by Arp7, Arp9 and the non-actin-related protein Rtt102 that together bind to a long, unsegmented HSA helix (Schubert et al., 2013). Given these similarities, the HSA regions and ARP modules of both families are found at very different positions (Figure 27). The

HSA region of RSC extends from the ATPase module towards the body module causing the ARP module to bind in between them. INO80’s ARP module binds to the HSA helix which binds to the exit DNA and follows its trajectory away from the rest of the remodeller (Knoll et al., 2018).

This explains why the two remodellers are regulated in a different way by their ARP modules. INO80 efficiently translocates DNA only if the linker length of DNA is

>40 bp (Zhou et al., 2018) which matches the footprint of the ARP^INO80 module on the linker DNA (Knoll et al., 2018). Therefore, the ARP^INO80 module was suggested to increase nucleosome sliding by preventing DNA ‘back-slippage’ during remodelling (Knoll et al., 2018; Zhou et al., 2018). However, it is possible that the ARP^INO80 module in addition applies a similar regulation mechanism as proposed here for RSC (section 3.6). The shortening of the DNA linker would lead to less stable interactions with the ARP^INO80 module resulting in increased dynamics of the module and causing a conformational change of the bilobal ATPase transmitted through the HSA and post-HSA region.

The ARP^RSC module seems to regulate the ATPase independent of DNA as suggested by its location in the RSC-nucleosome complex. The dynamics necessary for the proposed mechanism would be in accordance with the observed flexibility of the ARP Figure 27 | Schematic representation of the RSC complex and the INO80 complex showing the location of the HSA region and ARP module of the remodellers. ATPase module and HSA region in orange, nucleosome and extra-nucleosomal DNA in light grey and other remodeller subunits in dark grey. lobe 1, 2, RecA-like ATPase lobes 1, 2.

The schematic representation of the INO80 complex with the ARP module adapted from figure 5 in (Knoll et al., 2018).

module. Understanding the interactions of the ARP^RSC with the HSA and post-HSA regions and possibly the ATPase lobes in context of the RSC-nucleosome complex is key to a more comprehensive model of RSC remodelling. Therefore, further research needs to be done to decipher the exact molecular nature of RSC regulation through ARP^RSC.