Structural and functional analysis of the DEAF-1 and BS69 MYND domains

Abstract

DEAF-1 is an important transcriptional regulator that is required for embryonic development and is linked to clinical depression and suicidal behavior in humans. It comprises various structural domains, including a SAND domain that mediates DNA binding and a MYND domain, a cysteine-rich module organized in a Cys(4)-Cys(2)-His-Cys (C4-C2HC) tandem zinc binding motif. DEAF-1 transcription regulation activity is mediated through interactions with cofactors such as NCoR and SMRT. Despite the important biological role of the DEAF-1 protein, little is known regarding the structure and binding properties of its MYND domain.Here, we report the solution structure, dynamics and ligand binding of the human DEAF-1 MYND domain encompassing residues 501-544 determined by NMR spectroscopy. The structure adopts a ββα fold that exhibits tandem zinc-binding sites with a cross-brace topology, similar to the MYND domains in AML1/ETO and other proteins. We show that the DEAF-1 MYND domain binds to peptides derived from SMRT and NCoR corepressors. The binding surface mapped by NMR titrations is similar to the one previously reported for AML1/ETO. The ligand binding and molecular functions of the related BS69 MYND domain were studied based on a homology model and mutational analysis. Interestingly, the interaction between BS69 and its binding partners (viral and cellular proteins) seems to require distinct charged residues flanking the predicted MYND domain fold, suggesting a different binding mode. Our findings demonstrate that the MYND domain is a conserved zinc binding fold that plays important roles in transcriptional regulation by mediating distinct molecular interactions with viral and cellular proteins.

Figure 1. Primary sequence and NMR analysis of the DEAF-1 MYND domain.

(a) Sequence alignment of different MYND domains. Residues coordinating the first and second zinc ions are highlighted with red and blue background, respectively. Residues involved in binding to corepressor peptides are indicated with a green star at the bottom, and those interacting through their side chains are highlighted in green. The positions of mutations performed on BS69 are indicated at the bottom. (b) Long range ¹H, ¹⁵N HSQC spectrum correlating H^ε1 and H^δ2 to N^δ1 and N^ε2 through ² J _HN and ³ J _HN couplings (Pelton et al 1993). The spectrum reveals a different protonation pattern for each histidine sidechain corresponding to the three possible tautomeric states. (c) ¹³C secondary chemical shifts (top), {¹H}-¹⁵N heteronuclear NOE (middle), and ¹⁵N R₂/R₁ relaxation rates ratio (bottom) are plotted versus DEAF-1 MYND residue numbers. The secondary structure elements and the amino acid sequence of the protein are indicated at the top of the figure. Residues coordinating the first and second zinc are colored red and blue respectively.

Figure 2. Three-dimensional structure of the DEAF-1 MYND domain.

(a) Stereo view of the ensemble of the twenty lowest energy structures of the DEAF-1 MYND domain. α helices and β strands are colored in green and purple respectively, whereas zinc atoms are depicted as red spheres. (b) Ribbon representation of the DEAF-1 MYND domain. Side-chains of residues coordinating the zinc atoms are shown as sticks. The zinc coordination geometry is indicated by red dotted lines. (c) Superposition of DEAF-1 (green), ETO (red), ZNF10 (Blue), SMYD1 (yellow), SMYD2 (orange) and SMYD3 (gray) MYND structures shown in ribbon representation. The two zinc ions are depicted as red spheres. (d) Schematic representation of the zinc-binding pattern and secondary structure elements in MYND, RING, PHD and LIM domains. (e) Cartoon representation of DEAF1-MYND domain. Side chains of residues for which medium and long-range NOEs are observed that unambiguously define the cross-brace zinc binding topology are shown in magenta. Green lines indicate NOEs between C524 H^N/C504 H^β*, C524 H^N/C528 H^β* for the first binding site; and H536 H^ε1/C540 H^N, H536 H^ε1/C518 H^β1, H536 H^ε1/K520 H^N, and H536 H^ε1/C515 H^β2 for the second binding site.

Figure 3. Binding of DEAF-1 MYND to SMRT and NCoR peptides.

(a) Chemical Shift Perturbations (CSP, see methods) observed for the interaction between DEAF-1 MYND domain and SMRT (top) and NCoR (bottom) corepressor peptides. The sequence of SMRT and NCoR peptides used for the titrations are indicated in each graph. Secondary structure elements and amino acid sequence of the DEAF-1 MYND domain are shown at the top of the panel. Residues expected to interact with the corepressors are colored magenta. CSPs of residues that are most strongly affected are shown (cross symbols) as a function of ligand concentration for both titration with SMRT (top right) and NCoR (bottom right). The observed CSPs were fitted to a binding isotherm yielding dissociation constanst of 5.30±0.54 mM and 3.08±0.12 mM for the SMRT and NCoR peptides, respectively. The binding curves are shown as dashed lines. (b) Superposition of ¹H, ¹⁵N HSQC spectra of a 1 mM sample of the free DEAF-1 MYND domain (red) and upon addition of unlabeled corepressor peptides (cyan) up to a final concentration of 8 mM and 5 mM of SMRT and NCoR peptide (1∶8 and 1∶5 molar ratio), respectively. The intermediate steps of each titration are zoomed for a sub-region of the corresponding spectrum. In either case binding takes place on the fast exchange regime with respect to the chemical shift time scale. (c) Ribbon representation of DEAF-1 (left) and ETO (right) MYND domains. Residues experiencing the largest chemical shift perturbation upon addition of the corepressor peptides are shown as magenta sticks. In the ETO-SMRT complex structure the corresponding residues are shown as sticks as well, and the SMRT ligand peptide is colored orange and shown in cartoon representation.

Figure 4. Binding studies and mutational analysis of BS69 MYND domain.

(a) Two views of an electrostatic surface representation of the DEAF-1 MYND domain. Positive (blue) and negative (red) electrostatic surface potential is shown at ±3 k_B T/e⁻ and was determined using the program APBS in Pymol (www.pymol.org). (b) Corresponding surface views for the BS69 MYND structure obtained from homology modelling. Positive (blue) and negative (red) electrostatic surface potential is displayed at ±3 k_B T/e⁻. The location of the residues that were mutated for binding studies are labelled on the surface representation of the BS69 homology model. (c),(d) Analysis of binding of E1A, EBNA2 and MGA to wild-type or mutant BS69 proteins expressed as a GST-fusion protein or to the GST (G−) moiety as control; Inp: Input 10%; G-RRKR-559-562G4: mutation of residues 559–562 into four glycines. (e) A single point mutation in BS69 abrogates the binding to E1A. QT6 fibroblasts were transfected with 12SE1A and/or FLAG-tagged BS69 expression vectors as indicated. Cellular complexes were immunoprecipitated with an M2 anti-FLAG antibody. Co-immunoprecipitated and ectopic expressions of E1A were revealed by immunoblotting using an M73-E1A antibody.