Unique protein interaction networks define the chromatin remodelling module of the NuRD complex.

The combination of four proteins and their paralogues including MBD2/3, GATAD2A/B, CDK2AP1 and CHD3/4/5, which we refer to as the MGCC module, form the chromatin remodelling module of the nucleosome remodelling and deacetylase (NuRD) complex. To date, mechanisms by which the MGCC module acquires paralogue-specific function and specificity have not been addressed. Understanding the protein-protein interaction (PPI) network of the MGCC subunits is essential for defining underlying mechanisms of gene regulation. Therefore, using pulldown followed by mass spectrometry analysis (PD-MS), we report a proteome-wide interaction network of the MGCC module in a paralogue-specific manner. Our data also demonstrate that the disordered C-terminal region of CHD3/4/5 is a gateway to incorporate remodelling activity into both ChAHP (CHD4, ADNP, HP1γ) and NuRD complexes in a mutually exclusive manner. We define a short aggregation-prone region (APR) within the C-terminal segment of GATAD2B that is essential for the interaction of CHD4 and CDK2AP1 with the NuRD complex. Finally, we also report an association of CDK2AP1 with the nuclear receptor co-repressor (NCOR) complex. Overall, this study provides insight into the possible mechanisms through which the MGCC module can achieve specificity and diverse biological functions.

The topology of chromatin-binding domains in the NuRD deacetylase complex.

Class I histone deacetylase complexes play essential roles in many nuclear processes. Whilst they contain a common catalytic subunit, they have diverse modes of action determined by associated factors in the distinct complexes. The deacetylase module from the NuRD complex contains three protein domains that control the recruitment of chromatin to the deacetylase enzyme, HDAC1/2. Using biochemical approaches and cryo-electron microscopy, we have determined how three chromatin-binding domains (MTA1-BAH, MBD2/3 and RBBP4/7) are assembled in relation to the core complex so as to facilitate interaction of the complex with the genome. We observe a striking arrangement of the BAH domains suggesting a potential mechanism for binding to di-nucleosomes. We also find that the WD40 domains from RBBP4 are linked to the core with surprising flexibility that is likely important for chromatin engagement. A single MBD2 protein binds asymmetrically to the dimerisation interface of the complex. This symmetry mismatch explains the stoichiometry of the complex. Finally, our structures suggest how the holo-NuRD might assemble on a di-nucleosome substrate.