Structure and function insights into the NuRD chromatin remodeling complex

Abstract

Transcription regulation through chromatin compaction and decompaction is regulated through various chromatin-remodeling complexes such as nucleosome remodeling and histone deacetylation (NuRD) complex. NuRD is a 1 MDa multi-subunit protein complex which comprises many different subunits, among which histone deacetylases HDAC1/2, ATP-dependent remodeling enzymes CHD3/4, histone chaperones RbAp46/48, CpG-binding proteins MBD2/3, the GATAD2a (p66α) and/or GATAD2b (p66β) and specific DNA-binding proteins MTA1/2/3. Here, we review the currently known crystal and NMR structures of these subunits, the functional data and their relevance for biomedical research considering the implication of NuRD subunits in cancer and various other diseases. The complexity of this macromolecular assembly, and its poorly understood mode of interaction with the nucleosome, the repeating unit of chromatin, illustrate that this complex is a major challenge for structure-function relationship studies which will be tackled best by an integrated biology approach.

Fig. 1

Schematic description of the NuRD complex. The three-dimensional structure of the overall complex has not been determined to date and precise subunit interactions are unknown. All seven proteins of the complex are represented and annotated, and the stoichiometry obtained from mass-spectrometry analysis was taken into account, suggesting that the NuRD complex contains one CHD3 or CHD4 protein, one HDAC1 or HDAC2, three MTA1/2/3, one MBD2 or MBD3, six RbAp46/48, two p66α/β and two DOC-1 [12] subunits. Out of these proteins, the three paralogs of MTA (MTA1, 2 and 3) are found to be mutually exclusive, as well as the two paralogs of MBD (MBD2 and MBD3). Whether the other proteins are also part of distinct NuRD complexes is currently unknown

Fig. 2

a Schematic description of the CHD3 and CHD4 domains. CD chromo domain, DEAH-box Asp-Glu-Ala-His-box. b–d The published NMR structures of the two PHD domains and the second chromodomain of CHD4 are represented with their pdb accession number. Zinc ions are represented by gray spheres. In particular, one can notice the π-cation stacking interaction (represented with black dashes) allowing discrimination between the methylated and non-methylated state of H3K9 by residue F451 of the second PHD domain of CHD4 (2c). e The SAXS envelope of CHD4 suggests an elongated topology comprising ATPase (in red), PHD (in blue) and chromo domains (in green), as proposed by Morra et al. [36]. The structure can be divided into a head (ATPase domain) and a stalk (chromodomains), with the tandem PHD domain linking both. The tight association of these three domains explains their functional interdependence

Fig. 3

a Schematic description of the HDAC1 and HDAC2 domains. b The global X-ray structure of HDAC2 highlights the lipophilic tube as well as the foot pocket forming the active site of this enzyme. On the right, a closer view of the active site of HDAC8 (not found in NuRD but structurally highly similar, and the only one for which the structure in complex with acetylated lysine, K^ac, is available) shows crucial residues for the zinc ion coordination and for natural substrate interaction. Zinc ions involved in coordinated bonds with the ligand are represented by gray spheres. c, d The two structures of HDAC2 in complex with antagonist inhibitors of the hydroxymate class (SAHA) and benzamide class (20Y: 4-acetylamino-N-2-amino-5-thiophen-2-ylphenylbenzamide) show the M31 and L140 residues, forming the gate of the foot pocket which opens up to accommodate the thiophene group of benzamide inhibitors. This results in slower kinetics but also higher specificity towards this class of inhibitors, buried deeper in the active site than hydroxymates

Fig. 4

a Schematic description of the MTA1, MTA2 and MTA3 domains. NLS nuclear localization sequence. b The structure shows how an inositol phosphate molecule (Ins[1, 4, 5, 6]P4) can accommodate in the basic pocket (in blue) formed at the interface between HDAC1 and MTA1. In yellow, the limitation of the HDAC1–MTA1 interface. Structure superimposed and docked on 4A69 (HDAC3-SMRT). Negative, neutral and positive surface electrostatic potentials are displayed in red, white, and blue, respectively. c The global X-ray structure highlights HDAC1, represented in gray (see also Fig. 3), and shows how MTA1 peptide (represented in colour) is wrapped around the deacetylase (in orange, the SANT domain; in green/blue, the ELM2 domain). Crucial residues involved in the HDAC1–MTA1 interaction are annotated

Fig. 5

a Schematic description of the MBD2 and MBD3 domains. GR glycine-arginine-rich region, MBD methyl-CpG binding domain, TRD transcription repression domain, Poly-E poly-glutamate. b The NMR structure of MBD2^MBD shows the MBD domain of MBD2 interacting with a symmetrically methylated CpG island within an 11-bp DNA. Residues R22 and R44 are shown pointing at guanines inside the major groove of the DNA. c Close-up of the interaction interface highlights the crucial residues, and shows Van der Waals forces (represented with black dashes) between the methylated cytosines and the guanidinium group of arginines. d, e Complementary CG-base pairs and their specific hydrogen bonds with MBD2 are shown. Water molecules engaged in water-mediated hydrogen bonds are represented by black dots. The displayed arginine residues interact with guanines within the CpG island and are stabilized by residues D32 and Y34. The latter are also involved in cytosine recognition within the CpG

Fig. 6

a Schematic description of the RbAp46 and RbAp48 domains. WD tryptophan-aspartate domain. b RbAp46/H4 complex shows a binding interface located on the side of the barrel, in a pocket formed by the PP loop (in orange) and the long N-terminal helix (in dark blue). Crucial hydrophobic residues involved in the RbAp46–H4 interaction are annotated in the close-up view. c The structure of the RbAp48/MTA1 complex shows a noticeably similar interaction interface, on the side of the barrel, suggesting RbAp proteins cannot interact with histones and MTA proteins at the same time. d The RbAp48/FOG1 complex shows a different binding interface than the previous two, on the top of the barrel and extending towards the central channel