Structure of the MeCP2-TBLR1 complex reveals a molecular basis for Rett syndrome and related disorders

Abstract

Rett syndrome (RTT) is an X-linked neurological disorder caused by mutations in the methyl-CpG-binding protein 2 (MeCP2) gene. The majority of RTT missense mutations disrupt the interaction of the MeCP2 with DNA or the nuclear receptor corepressor (NCoR)/silencing mediator of retinoic acid and thyroid receptors (SMRT) corepressor complex. Here, we show that the "NCoR/SMRT interaction domain" (NID) of MeCP2 directly contacts transducin beta-like 1 (TBL1) and TBL1 related (TBLR1), two paralogs that are core components of NCoR/SMRT. We determine the cocrystal structure of the MeCP2 NID in complex with the WD40 domain of TBLR1 and confirm by in vitro and ex vivo assays that mutation of interacting residues of TBLR1 and TBL1 disrupts binding to MeCP2. Strikingly, the four MeCP2-NID residues mutated in RTT are those residues that make the most extensive contacts with TBLR1. Moreover, missense mutations in the gene for TBLR1 that are associated with intellectual disability also prevent MeCP2 binding. Our study therefore reveals the molecular basis of an interaction that is crucial for optimal brain function.

Keywords: MeCP2; NCoR/SMRT; Rett syndrome; TBL proteins; intellectual disability.

Fig. 1.

TBL1/TBLR1 and MeCP2 interact with a common fragment of NCoR1. (A) Domain overview of MeCP2 (orange) and the NCoR/SMRT complex components TBLR1 (blue) and NCoR1 (brown). A sequence of MeCP2 NID (orange) is shown, with residues mutated in RTT highlighted in yellow. Above the sequence are RTT mutations used in this study. The T308 phosphorylation site is marked with a P in a red circle. A previously characterized interaction between TBL1 and NCoR1 is indicated with gray dotted lines. Domains are annotated as follows: DAD, deacetylase-activating domain; MBD, methylated DNA-binding domain; NID, NCoR interaction domain; WD40, domain containing eight WD40 motif repeats. Brown lines indicate NCoR1 deletion mutants. Results of binding interactions between NCoR1 fragments with TBL1 and MeCP2 are summarized. n.d., not determined. (B) Pull-down of FLAG-NCoR1 fragments from HeLa cell extracts by MeCP2-NID peptide. in, input; IP, immunoprecipitate; WT, wild type. (C) Co-IP of endogenous TBLR1 with FLAG-NCoR1 fragments from transfected HeLa cells.

Fig. S1.

(A) Sequence alignment for MeCP2-NID domain. Residues that interact with TBLR1 are marked with blue dots. T308, which can be phosphorylated, is marked with a yellow dot. Known RTT missense mutations used in this study are marked below. (B) Sequence alignment of TBL1 and TBLR1 from vertebrates and sponge. Secondary structure elements are shown above the sequence. Residues interacting with MeCP2 are marked with orange dots. Residues mutated in this study are marked with red triangles.

Fig. 2.

MeCP2 binds directly to the TBL1 C-terminal domain. (A) Overview of TBL1/TBLR1 domain structure (blue). Blue lines show deletion mutants of TBL1. Results from pull-down assays are summarized beside the overview of deletion mutants. (B) Pull-down assay of full-length and deletion fragments of TBL1 from HeLa cell extracts with MeCP2-NID peptide. (C) Pull-down assay of purified TBLR1-CTD with wild-type MeCP2-NID peptide and peptides encoding RTT mutations or phosphoThr308 modification. Samples were bound (B) to beads containing biotinylated peptide. Peptides were run as a control in adjacent lanes (P). Mw, molecular weight. Asterisks denote nonspecific bands. (D) Binding interaction between MeCP2 peptides and TBLR1-CTD by SPR. Steady-state binding curves were derived from binding reactions with MeCP2-NID (blue), MeCP2-NID_S (gray), and MeCP2-NID R306C mutant (red) peptides.

Fig. S2.

(A) Coomassie-stained gel of purified TBLR1-CTD. (B) Raw SPR traces for wild-type TBLR1-CTD with MeCP2 mutant and wild-type peptides. The key indicates the concentrations of TBLR1 used in a dilution series. Binding curves in Fig. 2 were calculated from equilibrium measurements from these data. (C) Fluorescence anisotropy binding with a minimal 11-mer peptide of MeCP2 (residues 298–309). Anisotropy is measured in arbitrary units (A.U.). The binding constant was derived from fitting three independent experiments and is expressed as the mean K_D ± SE.

Fig. 3.

TBL1 chimeras allow identification of MeCP2-binding residues. (A) Overview of mouse-sponge chimeras. Wild-type and mutant MeCP2-NID peptides were used to pull down FLAG-tagged TBL1 chimeras. Binding interactions with MeCP2 are summarized next to the schema. (B) Pull-down assays of MeCP2 NID with FLAG-tagged mouse TBL1 mutants with individual amino acids converted to residues found in sponge.

Fig. S3.

(A) Pull-down assay of biotinylated MeCP2-NID wild-type or mutant peptide with TBL1 from sponge and Drosophila. (B) Sequence alignment of TBL1 from mouse and sponge, as well as TBL1-chimera2, indicating segments of the sequence that were used for mutagenesis analysis. (C) Pull-down assays between MeCP2-NID and local TBL1 mouse→sponge mutants from HeLa cell extracts. B, bound; in, input. (D) Co-IP of HDAC3 with wild-type and mutant TBL1. IP, immunoprecipitate. (E) Location of residue equivalent to E364 on the human TBLR1 structure (PDB ID code 4LG9).

Fig. S4.

(A) Divergent stereo view of the MeCP2-NID peptide (yellow) bound to TBLR1 (blue) with a Fo-Fc omit map calculated from coordinates where the MeCP2 peptide was omitted. This map is contoured at 2.5σ. (B) Comparison of peptide recognition in two histone H3-binding WD40 domains, WDR5 (PDB ID code 4A7J) and EED (PDB ID code 3K26), that feature recognition of a basic motif. The orientation of the viewpoint is rotated 90° around the y axis with respect to A. Each WD40 domain was superposed using the SSM function in Coot, and they are aligned primarily on the first four repeats. Root mean square deviations of fit are 2.6 Å and 2.7 Å, respectively. WD40 repeats 4 and 5 have been removed for clarity. Methylated arginine and lysine residues from histone H3 are recognized by aromatic cages, in contrast to the ionic interactions observed with basic residues in MeCP2.

Fig. 4.

Cocrystal structure of the TBLR1–MeCP2 complex. (A) Cartoon representation of TBLR1-CTD (blue) with MeCP2 NID (orange) shown as a ribbon. WD40 blades are colored shades of blue, with a gradient of dark to light from the N terminus to the C terminus. (B) Surface representation of TBLR1-CTD with MeCP2 NID. The view is rotated 90° around the horizontal axis compared with A, and WD40 repeats 1–3 are removed for clarity. (C) Surface representation of TBLR1 with the view as in A. The area buried by MeCP2 is colored yellow. (D) Electrostatic surface representation of TBLR1-CTD (APBS) colored red to blue from −5.5 to +5.5 kT/e. (E) Surface conservation analysis of TBLR1 (CONSURF), colored from blue (variable) to purple (conserved).

Fig. 5.

RTT mutations disrupt key interactions between MeCP2 and TBLR1. (A) Close-up of MeCP2 NID (orange) bound to TBLR1 (blue). Only residue side chains that make contacts with MeCP2 are shown. Hydrogen bonds are gray dotted lines. (B) Same view, rotated 180° around the vertical axis. MeCP2 is shown with RTT mutations modeled, and wild-type side chains are shown as transparent ghosts: R306C (C), K304E (D), P302R (E), and K305R (F). Orange arrows indicate the mutated residue.

Fig. 6.

TBL1/TBLR1 surface mutations do not bind MeCP2. (A) Map of mutations in TBLR1 and TBL1. (B) Steady-state SPR binding curves of wild-type (blue) and mutant (gray and black) TBLR1 proteins interacting with immobilized MeCP2-NID peptide. Wild-type TBLR1 with R306C NID peptide (red) is a negative control. (C) Recruitment of TBL1-mCherry to heterochromatin foci in NIH 3T3 cells by EGFP-MeCP2. DAPI staining (blue), green and red fluorescence, and a merged image are shown for wild-type and mutant proteins. (Scale bars, 5 μm.) Wild-type cells show colocalization in 63% of cells (n = 27). Mutants E184A, E364A, E382A, and D326N showed colocalization in 0 to 1 cells out of 14 cells. Y459F shows colocalization in 55% of cells (n = 27).

Fig. S5.

(A) Raw SPR traces for TBLR1-CTD mutants with MeCP2-NID peptide. (B) Alterations in melting temperature (Tm) of TBLR1 mutants using a thermal denaturation assay. Change in Tm is compared with wild type. (C) Co-IP of TBLR1-mCherry or importin α4 (KPNA4)-mCherry with NCoR/SMRT components. Samples tested were as follows: KPNA4 (negative control, has a similar molecular mass as TBLR1), wild-type TBLR1 (positive control), and TBLR1 point mutants. Copurification of NCoR/SMRT was determined by probing with α-HDAC3.

Fig. 7.

Missense mutations associated with developmental delay block MeCP2 binding to TBL1/TBLR1. (A) Map of mutations in TBLR1 and TBL1. (B) Pull-down assays of equivalent mutations in TBL1 with MeCP2-NID peptide or a mutant peptide containing the K305R RTT mutation. (C) Recruitment of TBL1-mCherry (red) and developmental delay mutants by EGFP-MeCP2 (green) to heterochromatin foci (blue) in NIH 3T3 cells. (Scale bars, 5 μm.)

Fig. S6.

(A) Overview of TBLR1 mutations associated with developmental delay without (dark green) or with (light green) autism. (Insets) Models of P444R and D369E mutations (pink). (B) Co-IP of TBLR1-mCherry or importin α4 (KPNA4)-mCherry with NCoR/SMRT components. Samples tested were as follows: KPNA4 (negative control), wild-type TBL1 (positive control), and TBL1 point mutants. Residues in TBLR1 that are equivalent to these mutations are D369E, P444R, and E171A. Copurification of NCoR/SMRT was determined by probing with α-HDAC3. (C) Thermal denaturation assays for developmental delay mutants showing change in Tm with respect to wild type. (D) SPR binding curves for NID and NID_S peptides with developmental delay mutants. Curves for wild-type TBLR1 protein are shown with wild-type (blue) and RTT (red) mutant peptides. Mutant TBLR1 proteins with wild-type peptide are shown in gray and black.