- Email: [email protected]
RETRACTED: Structure and Functional Analysis of the MYND Domain
doi:10.1016/j.jmb.2006.01.087
J. Mol. Biol. (2006) 358, 498–508
Structure and Functional Analysis of the MYND Domain
EMBL Heidelberg, Meyerhofstrasse 1, 69117 Heidelberg, Germany 2
Inserm U412, Laboratoire de Virologie Humaine, ENS-Lyon 46, alle´e d’Italie, 69364 Lyon France 3
q 2006 Elsevier Ltd. All rights reserved.
Keywords: MYND; DEAF-1; BS69; zinc binding; transcriptional regulation
ET
*Corresponding author
R
AC
Istituto di Ricerche di Biologia Molecolare, P. Angeletti, Via Pontina, Km 30.600; 00040 Pomezia (RM), Italy
The MYND domain (named after myeloid translocation protein 8, Nervy, and DEAF-1) is a conserved zinc binding domain. It is defined by seven conserved cysteine residues and a single histidine residue that are arranged in a C4–C2HC consensus. MYND domains exist in a large number of proteins that play important roles in development or are associated with cancers and have been shown to mediate protein–protein interactions, mainly in the context of transcriptional regulation. We have determined the three-dimensional structure of the MYND domain from human deformed epidermal autoregulatory factor-1 (DEAF1). The structure reveals a novel zinc binding fold, in which the C4–C2HC motif forms two sequential zinc binding sites. The first and second zinc binding modules comprise a small b hairpin and two short a helices, respectively. The sequential topology of the two zinc binding sites is distinct from the cross-brace PHD and RING finger folds but has some resemblance to LIM domains. The structure reveals that the MYND domain is a novel member of the treble-clef family of zinc binding domains. The MYND domains of BS69 and BOP bind ligands comprising a PXLXP peptide motif. On the basis of the solution structure of the DEAF-1 MYND domain we calculated a homology model of the MYND domain of the BS69 tumor suppressor. A mutational analysis of the BS69 MYND domain indicates that positively charged residues located on one face of its MYND domain are crucial for the molecular interactions of BS69. Different binding specificities of MYND domains may depend on distinct surface charge distributions.
TE
1
D
Roberta Spadaccini1, He´le`ne Perrin2, Matthew J. Bottomley3 Ste´phane Ansieau2 and Michael Sattler1*
Introduction
R
Modification of chromatin structure plays a crucial regulatory role in various processes including transcription, DNA modification and repair. Multiple protein complexes, through different ways, have been implicated in remodeling of chromatin structure. Many components of these complexes share common protein domains that behave as independent functional modules and create or recognize unique histone modifications. For instance, some domains create post-translationally modified lysine residues in histone N termini (HATand SET domains) and others
Abbreviations used: DEAF-1, deformed epidermal autoregulatory factor-1; MTG, myeloid translocation protein; NOE, nuclear Overhauser effect; NOESY, NOE spectroscopy; HSQC, heteronuclear single quantum coherence; GST, glutathione-S-transferase. E-mail address of the corresponding author: [email protected]
recognize these modifications (bromo- and chromodomains, respectively).1,2 The MYND domain (myeloid translocation protein 8, Nervy, DEAF-1) is a novel cysteine-rich structure present in proteins generally implicated in gene regulation and associated with cancers.3,4 MTG8 (also named CBFA2T1, ETO, CDR and Nervy) is a member of the MTG (myeloid translocation protein) family and was first identified as one of the translocation partners of the AML1 protein in acute myeloid leukemia.5 In the AML-MTG8 fusion protein, MTG8 represses AML1 target gene expression and contributes to tumor progression by recruiting the transcriptional corepressor complex Sin3Aa/SMRT(N-CoR)/ HDAC1.6,7 MTG8 was also shown to associate with Bcl-6, a transcription factor involved in normal B cell development, the expression of which is altered in non-Hodgkin lymphoma.8 DEAF-1 is required for embryonic development and is linked to clinical depression and suicidal behaviour in humans.9 DEAF-1 has been shown to
0022-2836/$ - see front matter q 2006 Elsevier Ltd. All rights reserved.
499
Structure and Functional Analysis of the MYND Domain
TE
D
The solution structure of the MYND domain was determined using standard heteronuclear NMR techniques.21,22 Elements of regular secondary structure were identified on the basis of nuclear Overhauser effect (NOE) patterns observed in 3D NOE spectroscopy (NOESY) experiments, and by 13 a C and 13 C b secondary chemical shifts (Figure 1(c)). The presence of dab(i,iC3) NOEs for residues 526–529, 528–531, 530–533 and 535–539, and positive 13Ca secondary chemical shifts of Thr526–Lys531 and Trp533–His538 indicate an a-helical conformation of these residues. 15 N relaxation data were recorded with a 0.5 mM 15 N-labeled MYND domain sample. The rotational correlation time is tcz9.1 ns as estimated from the trimmed average T1/T2 ratio. This value seems unusually high for a protein of 6 kDa molecular mass. Comparison of 1H–15N heteronuclear single quantum coherence (HSQC) spectra recorded at 0.5 mM and 20 mM concentration shows small chemical shift changes for some of the backbone amide signals. The residues affected correspond to conserved hydrophobic residues exposed on the surface of the MYND domain. This may indicate a slight non-specific aggregation at higher concentrations, such as used in the NMR experiments. Furthermore, the MYND domain comprises only a few elements of secondary structure, and a large part of the fold is in an extended conformation. In addition, about 20% of the residues in the protein construct studied are unstructured and highly flexible in solution based on 15 N relaxation measurements (Figure 1(c)). Taken together, the largely extended conformation of the MYND domain and a small tendency to aggregate may explain the unusually high correlation time. Consistent with this, the molecular mass determined by analytical ultracentrifugation experiments indicates that the protein is monomeric in solution (data not shown).
R
ET
R
AC
regulate transcription,3,10 and this function requires the recognition of target DNA sequences by its SAND domain.11 Through LMO4, DEAF-1 also interacts with the tumor suppressor BRCA1, potentially linking DEAF-1 to breast cancer development.12,13 Among the MYND domain-containing proteins, BS69 has been described as a co-repressor in association with various transcription factors such as c-Myb, B-Myb, Ets-2 and the Myc-related MGA protein14 and as a partner of the BRCA2 repressor EMSY.15 Interestingly, the adenoviral oncoprotein E1A and the Epstein-Barr virus-induced nuclear antigen 2 (EBNA2) directly bind to the BS69 MYND domain and disrupt BS69 complexes with cellular factors. Like the cellular partners MGA and EMSY, these viral proteins bind to BS69 through a PXLXP peptide motif (XZany amino acid).14,15 The binding of PXLXP motifs was also observed for the MYND domain of the muscle-specific transcriptional repressor BOP.16,17 However, distinct ligand binding specificities exist since the MYND domain of RACK7 fails to interact with BS69 PXLXP ligands.14 MTG8 and BS69 MYND domains have also been shown to bind directly to N-CoR and SMRT.6,7,18 However, SMRT lacks a PXLXP motif. Similarly, BIP, the binding partner of the Bra-1 MYND domain does not display a PXLXP peptide motif,19 indicating diversification in MYND domain binding properties. Despite the biological importance of MYND domain-containing proteins, the three-dimensional structure and molecular functions of the MYND domain have not yet been described. Here, we present the structure of the human DEAF-1 MYND domain. Our structure reveals a novel zinc binding fold comprising two sequential zinc binding sites, which shares some features with LIM domains.20 The amino acid sequences of DEAF-1 and BS69 MYND domains are closely related. Based on the structural information derived for the DEAF-1 MYND, we performed a structural and functional analysis of BS69. Our data highlight the biological importance of charged residues located on one face of the BS69 MYND domain.
Results
Structure determination On the basis of multiple sequence alignments two MYND domain constructs, comprising residues 496–544 and 496–565 of human DEAF-1, were designed. Both proteins were solubly expressed and 1D NMR spectra were acquired (data not shown). The shorter fragment was chosen for structure determination, since the NMR spectra of both proteins are very similar and additional signals in the 1D spectrum of the longer construct exhibit random coil chemical shifts, indicating that the additional residues are unstructured.
Structure and backbone dynamics of the MYND domain The solution structure of the DEAF-1 MYND domain (Figure 1(a) and (b)) was calculated using CNS/ARIA 1.2.23,24 All distance restraints were derived from manually assigned NOE cross-peaks. Experimental restraints and structural statistics are summarized in Table 1. Residues 503–541 adopt a well-defined tertiary structure. The ten N and three C-terminal residues are disordered as indicated by the paucity of NOEs and negative heteronuclear {1H}–15N NOE values (Figure 1(c)). Some variations are observed for the local residue-specific correlation times as judged from the ratio of 15N T1/T2 relaxation times. These may indicate some flexibility for the backbone of the MYND fold at microto milliseconds time-scales. The first and second zinc binding sites are formed by Cys504, Cys507, Cys515, Cys518 and Cys524, Cys528, His536, Cys540, respectively. The Nd1 position of the His536 side-chain is protonated as
500
R
ET
R
AC
TE
D
Structure and Functional Analysis of the MYND Domain
Figure 1. Structure and dynamics of the DEAF-1 MYND domain. (a) Stereo view of the ensemble of ten lowest energy structures of the DEAF-1 MYND domain (residues 501–541). Secondary structure elements are colored in red and blue for a helices and b strands, respectively. The zinc atoms are shown as green spheres. (b) Ribbon representation of the DEAF-1-MYND domain. Side-chains of the residues coordinating the zinc atoms and conserved hydrophobic residues (W533, Y523, A511 and V505) are shown in green and yellow, respectively. (c) Amino acid sequence, secondary structure and NMR data for the DEAF-1 MYND domain. Residues coordinating the first and second zinc are printed red and blue, respectively. Below: the sequence is shown: on top: difference of 13Ca and 13Cb secondary chemical shifts; middle: heteronuclear NOE; bottom: ratio of 15N T1/T2 relaxation times.
501
Structure and Functional Analysis of the MYND Domain
Table 1. Structural statistics for the DEAF-1 MYND domain
hSAi 0.023G0.002 0.90G0.12
hSAiwater refined 0.035G0.004 1.64G0.23
0.094G0.003
0.119G0.010
0.40G0.21 0.90G0.20
0.47G0.11 1.01G0.12
0.6G0.8
0G0
66.9G3.6 28.0G5.5
69.7G5.6 26.0G5.4
D
˚ )b All distance restraints (A Dihedral anglesc C. Q-factor for experimental residual dipolar coupling restraintsd 1 DHN ˚ )e D. Coordinate precision (A N, Ca, C 0 All heavy atoms E. Structural qualityf Bad contacts Ramachandran plot Residues in most favored region (%) Residues in additionally allowed region (%)
627 (624/3) 174 80 61 311 27 f, 27 j 13
TE
A. Number of structural restraints All NOE (unambiguous/ambiguous) Sequential (jiKjjZ1) Medium-range (1!jiKjj% 4) Long-range (jiKjjO4) Intraresidual Dihedral angles 1 H–15N residual dipolar couplings ˚ ) from experimental restraintsa B. r.m.s. deviation (A
R
AC
a hSAi is an ensemble of ten lowest-energy solution structures (out of 100 calculated) of the MYND domain before water-refinement, hSAiwater-refined is the hSAi ensemble after refinement in a shell of water.39 The CNS Erepel function was used to simulate van der Waals ˚ K4 using PROLSQ van der Waals radii.39 r.m.s. deviations for bond lengths, interactions with an energy constant of 25.0 kcal molK1 A ˚ , 0.426 (G0.012)8 and 0.365 (G0.015)8 before and 0.0055 (G0.0002) A ˚, bond angles and improper dihedral angles are 0.0023 (G0.0001) A 0.782 (G0.043)8 and 2.15 (G0.21) after water-refinement. 1 kcalZ4.18 kJ. b ˚ 2. For restraining Distance restraints were employed with a soft square-well potential using an energy constant of 50 kcal molK1 A the Zn coordination geometry 20 distance restraints were applied as described in Materials and Methods. No distance restraint in the ˚. hSAi structures was violated by more than 0.5 A c Dihedral angle restraints derived from TALOS36 were applied to f, j backbone angles using energy constants of 200 kcal molK1 radK2. No dihedral angle restraint in the hSAi ensemble was violated by more than 5.58. d Quality factor for the RDC refinement.43 Residual dipolar couplings were applied with a final energy constant of 1.0 kcal molK1 HzK2 for an alignment tensor with an axial component of 9 Hz and a rhombicity of 0.4. e Coordinate precision is given as the Cartesian coordinate r.m.s. deviation of the ten lowest-energy structures in the NMR ensemble with respect to their mean structure for residues 504–540. f Structural quality was analyzed using PROCHECK.41
R
ET
determined from the chemical shifts of the 15Nd nitrogen atoms in a 2D 1H–15N long-range HMQC spectrum.25 Thus, His536 coordinates the second zinc atom via the N32 nitrogen of the imidazole ring. Only a few secondary structure elements are present in the MYND domain fold. The first zinc domain comprises a small b hairpin, while the second zinc binding domain contains two short a helices (Thr526–Lys531 and Trp533–His538), connected by a turn around Asp532. The MYND domain has a zinc-dependent fold. Introduction of an excess (20 mM) of the metal-chelator EDTA leads to complete unfolding of the protein as indicated by the loss of chemical shift dispersion in the HSQC spectra (data not shown). Removal of EDTA and the addition of two equivalents of Zn result in complete refolding of the domain. This demonstrates that the uptake of zinc is completely reversible and crucial for proper folding. ˚ and are The two zinc ions are separated by z10 A coordinated by conserved Cys and His residues in the C4 and C2HC clusters, respectively. Aside from the C4–C2HC motif, hydrophobic residues, corresponding to Val505, Ala511, Tyr523 and Trp533 in DEAF-1, are very well conserved in other MYND domains. With the exception of Val505, which
contributes to a small hydrophobic core, these residues are (partially) accessible on the surface of the domain (Figure 1(b)). In addition, the surface of the DEAF-1 MYND domain is highly charged, with a cluster of negatively charged side-chains being exposed from the first zinc binding module (Figure 3(a), below). The combined hydrophobic and negatively charged surface suggests that the DEAF-1 MYND domain could mediate protein– protein interactions and is not involved in nucleic acid binding. Consistent with this, point mutations of zinc-binding Cys residues do not abolish the interaction of full-length (Drosophila) DEAF-1 with DNA.3 Since these mutations are expected to disrupt the three-dimensional fold of the protein it can be concluded that the MYND domain does not contribute to DNA binding, a function that rather involves the DEAF-1 SAND domain.11 Comparison with other zinc binding domains The MYND domain adopts a unique zincdependent protein fold. Searches with DALI26 did not reveal any significant structural homologues. However, the sequential topology of the two zinc coordination sites in MYND resembles that of LIM
502
TE
D
Structure and Functional Analysis of the MYND Domain
AC
Figure 2. Comparison of tandem zinc binding folds. (a) Alignment of consensus sequences for MYND, LIM, RING and PHD domains. Residues coordinating the first and second zinc are colored red and blue, respectively. (b) Schematic representation of the zinc binding pattern and secondary structure elements in MYND, LIM, RING and PHD domains. (c) Ribbon representation of MYND (PDB accession 2FV6), LIM (1A7I), RING (1CHC) and PHD (1XWH) domains. Sidechains of zinc coordinating or conserved aromatic residues are shown in green and yellow, respectively. Zinc atoms are shown as green spheres.
R
ET
R
domain27 and is distinct to the cross-brace coordination employed by RING and PHD fingers20 (Figure 2). Similar to the LIM domain the residue spacing between the zinc coordinating sites is highly conserved. However, the order of the two zinc binding modules is reversed between the MYND (C4–C2HC) and LIM domains (C2HC–C4) and different secondary structure elements are found. Distinct from the other zinc binding folds shown in Figure 2, the first and second zinc binding modules exhibit b strands and a helices, respectively. The first two zinc coordinating Cys residues in the C4 module are located in a small b-sheet, while the remaining two Cys residues are in an extended conformation. The four zinc coordinating ligands in the second zinc binding site (C2HC) are located within and flanking the two a helices. Notably, in contrast to LIM, RING and PHD, a ˚ ) is found between the relatively short distance (10 A two zinc atoms in the MYND domain (Figure 2(c)). When comparing the three-dimensional structures of MYND, LIM, RING and PHD domains (Figure 2(c)) with a multiple sequence alignment of these zinc binding folds different structural roles of conserved hydrophobic residues can be discerned. For example, a tryptophan residue is conserved in the primary sequences of MYND and PHD domains. However, while this tryptophan contributes to the hydrophobic core in PHD domains28 the corresponding residue (Trp533 in DEAF-1) is solvent-exposed in the structure of the MYND domain (Figures 1 and 2). Similarly, an aromatic residue (Tyr/Phe) precedes the fifth metal coordinating ligand in MYND, PHD and RING domains. However, while in PHD and RING domains this residue is part of the hydrophobic core the
corresponding Tyr523 is surface-exposed in the MYND domain. We thus conclude that the MYND domain is a new member of the treble clef domain family29 that shares some structural features with LIM domains. Significant differences in structural details suggest that it is a novel fold with distinct molecular functions. Molecular interactions of MYND domains In glutathione-S-transferase (GST)-pulldown and immunoprecipitation assays, BS69 and BOP MYND domains have been shown to bind to viral and/or cellular partners via a PXLXP motif 14,15. We could not detect any interaction of the known BS69 binding partners with the DEAF-1 MYND domain precluding further analysis of the binding properties of the DEAF-1 MYND domain with such ligands. Consequently, we decided to examine the binding properties of the BS69 MYND domain, which displays a high level of sequence similarity with the DEAF1 MYND domain (41/47% identity/ similarity over 41 residues) For BS69, various PXLXP partners have been identified, including the Myc-related protein MGA and the viral proteins E1A and EBNA2. Interestingly, despite its homology with BS69 (Figure 4(a), below) indicating a conserved fold, the MYND domain of the RACK7 protein fails to interact with BS69 PXLXP ligands, offering therefore an additional possibility to identify important functional residues.14 Despite considerable efforts, we could not obtain recombinant BS69 MYND domain in amounts suitable for NMR analysis. Therefore, we calculated a homology model based on our structure of the DEAF-1 MYND domain using the program
503
D
Structure and Functional Analysis of the MYND Domain
ligands, but absent from RACK7, ETO/MTG8 and DEAF-1. Mutation of the positively charged residues (RRKR559–562GGGG) disrupts the binding to PXLXP ligands (Figure 4(d)). Individual mutations of these residues further revealed that among these four residues only Arg560 is essential for binding of BS69 to its ligands (Figure 4(e)). As a control, all GST-fusions bind similarly to a novel identified BS69 co-factor, the binding site of which is located upstream from the MYND domain (data not shown). Together the mutational analysis suggests that a positively charged face of the BS69 MYND domain, which comprises positively charged residues at the N and C-terminal ends of the zinc binding fold, is crucial for the interaction with PXLXP ligands (Figure 4(f)).
R
ET
R
AC
MODELLER.30 A surface representation of the BS69 MYND model colored by electrostatic potential indicates a striking separation of charges (Figure 3(b)), which is quite distinct compared to the DEAF-1 MYND domain (Figure 3(a)). An extended positively charged patch is found on one face comprising KKK518-520 and RRKR559-562 at the N and C terminus of the BS69 MYND domain, respectively, while hydrophobic and negative regions are located on the opposite side. A BS69 C523S mutation completely disrupts the binding of all partners without any distinction.14 This strongly suggests that the MYND domain fold is necessary for the interaction with the ligands, since the Cys mutant most likely destroys the first zinc binding site and thus the fold of the MYND domain structure. To further characterize the molecular interactions of the BS69 MYND domain, we generated BS69 proteins harboring point mutations affecting either the positively or negatively charged patches and examined their binding properties in GST pull-down assays (Figure 4). Simultaneous mutation of the three positive charges (KKK518–520DDD) preceding the first zinc binding site abolishes the binding of BS69 to its ligands (Figure 4(b)). Similarly, the charge reversal of Glu527-Glu528 (EE527-528KK), which are located in a neighboring region in the first zinc binding module, strongly reduce the binding of all three PXLXP ligands (Figure 4(c)). Introduction of a negative charge (S515D; Figure 4(d)) also strongly impairs the binding to all three proteins, while the W522Y mutation mainly affects the interaction with the cellular ligand (MGA) but not the viral binding partners (Figure 4(d)). The sequence alignment of MYND domains (Figure 4(a)) highlights a set of positively charged residues located at the extreme C terminus of BS69 (RRKR559–562). Some of these residues are conserved in BS69 and BOP orthologues, which are known to bind PXLXP
TE
Figure 3. Homology modeling and electrostatic surface of MYND domains. (a) Ribbon (left) and surface representation (middle and right) of the DEAF-1 MYND domain. The first and second zinc binding modules are colored orange and purple, respectively. Side-chains of the conserved aromatic residues are shown on the left. Surfaces are colored blue and red for positive and negative electrostatic surface potential, respectively. (b) Ribbon (left) and surface representation (middle and right) of the BS69-MYND domain model. Colors are as for (a). Side-chains of aromatic residues present only in the BS69 are shown in cyan.
Discussion We describe the solution structure of the DEAF-1 MYND domain, in which two zinc ions are coordinated in a sequential manner. Based on the conservation of structurally important residues, the fold is representative for other MYND domains. Although the sequential arrangement of the two zinc binding modules is similar to LIM domains, the MYND domain adopts a unique zinc dependent fold, comprised of a short b hairpin and two short a helices in the first and second zinc binding modules, respectively. Thus, the MYND domain represents a novel member of the treble-clef family of zinc binding folds.29,31 The BS69 and BOP MYND domains have previously been shown to depend on a PXLXP peptide motif present in their binding partners.14,16,17 However, as the BS69-related RACK7 MYND domain fails to interact with BS69 ligands it was unclear whether only a subset of MYND domains bind to PXLXP peptide motifs or whether
504
ET
R
AC
TE
D
Structure and Functional Analysis of the MYND Domain
R
Figure 4. Binding studies and mutational analysis of the BS69 MYND domain. (a) Sequence alignment of MYND domain sequences of human BS69 (accession number X86098, residues 515–562), RACK7 (AB032951, residues 1020–1066), ETO/ MTG8 (X70990, residues 477–525), BOP (NP938015, residues 44–95) and DEAF-1 (NP066288, residues 496–544) proteins. The positions of the mutations performed in the BS69 protein are indicated at the top. (b), (c) and (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 glycine residues. (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. Coimmunoprecipitated and ectopic expressions of E1A were revealed by immunoblotting using an M73-E1A antibody. (f) Ribbon representation of the BS69-MYND domain homology model. Side-chains of residues probed by mutations are colored in red and blue, for negatively and positively charged residues, respectively, and in green for other residues. The orientation of the BS69 MYND domain corresponds to Figure 3(b) after rotation by 908 along the x-axis.
all of them bind to PXLXP motifs with different sequence specificities. We found that the MYND domain of the human DEAF-1 protein does neither bind to short PXLXP peptides nor to established full-length PXLXP containing BS69 ligands. This reenforces that binding to PXLXP ligands is restricted to a subset of MYND domains. To further characterize the molecular interactions and identify the residues that confer the ability to bind to PXLXP ligands we studied the BS69 MYND
domain. Notably, compared to DEAF-1 and most other MYND-containing proteins (based on sequence alignments; Figure 4(a)) the BS69 MYND domain exhibits a unique distribution of charged residues at its surface (Figure 3(b)). An extended positively charged surface is present on one side of the MYND domain, while hydrophobic and negative regions are located on the opposite side. Based on these peculiar surface features, we performed a mutational analysis of BS69
505
Structure and Functional Analysis of the MYND Domain
Materials and Methods Protein expression and purification
TE
D
Expression constructs of human DEAF-1 MYND were prepared in modified pET-24d vectors (Novagen) that express proteins with N-terminal His6 tags, removable by cleavage with TEV protease. The DEAF-1 MYND construct used for structure determination spans wildtype residues 496 to 544 plus an N-terminal insertion of three amino acid residues GAM. The two plasmid constructs were verified by DNA sequencing and the recombinant proteins by mass spectrometry (MALDI). Chemically transformed Escherichia coli (strain BL21-DE3) cells were grown in LB rich medium supplemented with 50 mM ZnCl2 at 37 8C until the absorbance at 595 nm reached 0.4. The cells were further incubated at 23 8C until the absorbance reached 0.6, upon which gene expression was induced by the addition of IPTG to a final concentration of 0.4 mM. After 18 h incubation at 23 8C with shaking, cells were harvested by centrifugation. The cells were resuspended in de-gassed lysis buffer (40 mM Tris–HCl (pH 8.0), 0.2 M NaCl, 5 mM imidazole, 15 mM 2-mercaptoethanol, 50 mM ZnCl2, 0.05% (v/v) NP-40 detergent) plus EDTA-free protease inhibitors (Roche) and were lysed using a microfluidizer (Microfluidics Corporation, MA). Cell debris was removed by centrifugation at 12,000 r.p.m. for 30 min at 4 8C. The soluble His6tagged MYND protein was purified from the supernatant using Ni2C-NTA Superflow affinity resin (QIAGEN). After several washing steps, protein was eluted in 40 mM Tris (pH 8.0), 0.2 M NaCl, 0.3 M imidazole, 15 mM 2-mercaptoethanol, 50 mM ZnCl2. The His6 tag was removed by overnight incubation at 23 8C with TEV protease. During incubation, the sample was dialyzed against 40 mM Tris (pH 8.0), 0.2 M NaCl, 5 mM imidazole, 15 mM 2-mercaptoethanol, 50 mM ZnCl2. The noncleavable His6-tagged TEV protease was then removed using Ni2C-NTA resin. The protein sample was concentrated to a volume of 1 ml and further purified on a Superdex-75 (16/60) gel-filtration column, from which it eluted with a single symmetrical peak profile. The purified sample was exchanged into appropriate buffers for NMR (50 mM sodium phosphate (pH 6.5), 0.15 M NaCl, 5 mM DTT, 50 mM ZnCl2, 0.02% (w/v) NaN3, 10% (v/v) 2H2O, stored under argon gas). For NMR studies, uniformly 15N and 13C/15N-labeled proteins were prepared using M9 minimal growth medium appropriately supplemented with 15N-labeled ammonium chloride and 13 C-labeled glucose. NMR experiments were performed at a protein concentration of 0.5–1.0 mM.
R
ET
R
AC
(Figure 4). Mutations affecting the positive patches strongly reduce the affinity for PXLXP partners, while charge reversal of two glutamate residues has a smaller effect on the interaction. The mutational analysis suggests that a positively charged face of the BS69 MYND domain, comprising arginine and lysine residues at the N and C-terminal ends of the zinc binding fold, is crucial for the interaction with PXLXP ligands (Figure 4(f)). Positively charged residues are found at the C-terminal end of the MYND domains of BS69 and BOP homologues, but not of other proteins, including ETO, RACK7 or DEAF1 (Figure 4(a)). The importance of these residues in BS69 for binding suggests that they may contribute to binding selectivity of certain MYND domains with PXLXP containing proteins. For example, the long hydrophobic lysine and arginine side-chains might interact with proline residues and thus contribute to the molecular recognition of the PXLXP motif by the MYND domain. The positive charges of these side-chains could mediate long-range electrostatic interactions with negative charges, which may be located in regions flanking the PXLXP motif. The MYND interaction might thus involve a larger binding epitope of which PXLXP merely represents a conserved core motif. Interestingly, the number and location of aromatic and specifically tryptophan residues is quite variable amongst MYND domains (Figure 4(a)). Two conserved aromatic residues, which are shared amongst all MYND domain family members, contribute to a hydrophobic patch at this binding surface (Figure 3). In addition, an aromatic residue is found between the first two zinc-coordinating Cys residues in both BS69 (Tyr524) and BOP (His58) MYND domains, but not in other proteins, suggesting that hydrophobic interactions of this aromatic residue could further contribute to the ligand binding of BS69 and BOP. In MTG family proteins, a Trp exists at this position, which is more bulky, and thus might sterically interfere with ligand binding. Two additional Trp residues are found in BS69 and RACK7 (Trp522, Trp536 in BS69) compared to other MYND domains. It is likely that these residues also contribute to distinct molecular interactions of MYND domains. Our studies suggest that the zinc-dependent fold of MYND domains presents a binding surface that involves hydrophobic (aromatic) residues and exhibits a unique distribution of charged surfaceexposed residues that are important for binding. These results are somewhat reminiscent of the binding of viral proteins to the retinoblastoma protein (RB) that not only depends on hydrophobic residues in the pocket binding domain but also involves a patch of lysine residues located at the rim of the binding site.32 Structural analysis of MYND– ligand complexes will be required to clarify details of the molecular recognition.
NMR spectroscopy and resonance assignments The 1H, 13C and 15N resonances of the DEAF-1 MYND domain were assigned using standard triple resonance experiments (HNCACB, HNCA, CBCA(CO)NH), HN detected side-chain TOCSY (H(CCO)NH, (H)C(CO)NH), and HCCH-TOCSY acquired at 22 8C. Experiments were recorded on in-house Bruker DRX600 and DRX500 spectrometers equipped with cryo-probes. For structure determination, 2D-NOESY, 15N-edited HSQC-NOESY and 13C-edited HMQC-NOESY spectra were recorded at 800 MHz or 900 MHz with mixing times of 60 ms and 90 ms. Data were processed with NMRPIPE33 and analyzed using NMRVIEW.34 The tautomeric state of the histidine ring was determined with a 2D 1H–15N HMQC experiment optimized for histidine side-chains.25
506
Structure and Functional Analysis of the MYND Domain
Structure calculations
The coordinates and NMR restraints for the DEAF-1 MYND domain have been deposited at the Protein Data Bank under accession number 2FV6.
Acknowledgements We thank A. Gasch, B. Simon and G. Stier (EMBL Heidelberg) for help; and M. Beyermann (FMP, Berlin) for peptide synthesis, M. Collard and J. Huggenvik (Southern Illinois University, USA) for useful discussions. R.S. and M.J.B. are grateful for long-term EMBO and Marie-Curie fellowships, respectively. This work was supported by the Deutsche Forschungsgemeinschaft and EMBL.
Supplementary Data
Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j. jmb.2006.01.087
ET
R
AC
Structures were calculated using the experimentally derived restraints with ARIA1.223/CNS.24 A total of nine iterations (20 structures in the first seven iterations) were performed, in the last two iterations 100 structures were calculated. Initial rounds of refinement using only NOE data defined the general fold of the domain and revealed the zinc coordination. In the final refinement calculation, ˚) distance restraints were added for the Zn–Sg (1.8–2.8 A ˚ ) bonds, and between the four zinc and Zn–N32 (1.5–2.5 A ˚ ) to ensure coordinating atoms at each site (3.6–4.0 A tetrahedral zinc coordination geometry. The structures were refined in a shell of water molecules as described.39 For the water refinement the Zn coordination geometry ˚ ; Zn–N32 2.0 A ˚ ) and was defined by bonds (Zn–Sg 2.3 A angles (109.58 for S–Zn–S, N32–Zn–S, Zn–S–C and 1208 for Zn–N–C angles, respectively). Residual dipolar couplings were employed as orientational restraints as described.40 The quality of the structure was assessed using PROCHECK-NMR.41 Molecular images were generated using MOLMOL.42 Structure similarity searches were performed using DALI†. The structure of BS69-MYND (residues E423 to E485) was generated by comparative modeling with the program MODELLER30 based on a sequence alignment (Figure 4(a)) using the coordinates of DEAF-1 MYND as template.
Protein Data Bank accession codes
D
J (HN,Ha) coupling constants were measured to derive restraints for f dihedral angles.35 Additional f/j restraints were obtained from backbone chemical shifts using TALOS.36 1H–15N residual dipolar couplings were measured in isotropic and anisotropic phases, the latter being created by a media containing 5% (w/v) hexaethylene glycol monododecyl ether (C12E6)/n-hexanol (molar ratio 0.64).37 Heteronuclear {1H}–15N NOEs as well as longitudinal and transversal 15N autorelaxation rates were measured at 500 MHz 1H frequency using standard methods.38 The relaxation decays were sampled in an interleaved fashion at 10/14 different time-points, between 20 and 1600 ms (T1) and 14.4 and 244.8 ms (T2), respectively. The relaxation data were analyzed with NMRView version 4.34
TE
3
Binding experiments
R
Mutations were performed using the QuickChange mutagenesis kit (Stratagene). Wild-type and mutants BS69 GST fusion proteins encompass the residues 411 to 562 of the human BS69 protein (accession number X86098). GST-protein production, GST pull-down assay, cell transfections and immunoprecipitation experiments have been described.14 For immunoprecipitation experiments, briefly, the BS69 fragment (amino acid residues 451–562 R560G) was sub-cloned as the FLAG-pcDNA3 vector. Combination of BS69 and 12SE1A expression vectors were transiently transfected in fibroblasts (QT6). Cells were harvested 24 h post-transfection, lysed and successively incubated with an anti-FLAG M2 antibody (Integra Biosciences) and protein A Sepharose (Amersham Biosciences Inc.). After separation of proteins on SDS-PAGE, BS69 and E1A were visualized with an M2 anti-FLAG antibody and an M73 anti-E1A antibody (Calbiochem), respectively. † http://www.ebi.ac.uk/dali/index.html
References
1. Jenuwein, T. & Allis, C. D. (2001). Translating the histone code. Science, 293, 1074–1080. 2. Bottomley, M. J. (2004). Structures of protein domains that create or recognize histone modifications. EMBO Rep. 5, 464–469. 3. Gross, C. T. & McGinnis, W. (1996). DEAF-1, a novel protein that binds an essential region in a Deformed response element. EMBO J. 15, 1961–1970. 4. Huggenvik, J. I., Michelson, R. J., Collard, M. W., Ziemba, A. J., Gurley, P. & Mowen, K. A. (1998). Characterization of a nuclear deformed epidermal autoregulatory factor-1 (DEAF-1)-related (NUDR) transcriptional regulator protein. Mol. Endocrinol. 12, 1619–1639. 5. Rossetti, S., Hoogeveen, A. T. & Sacchi, N. (2004). The MTG proteins: chromatin repression players with a passion for networking. Genomics, 84, 1–9. 6. Wang, J., Hoshino, T., Redner, R. L., Kajigaya, S. & Liu, J. M. (1998). ETO, fusion partner in t(8;21) acute myeloid leukemia, represses transcription by interaction with the human N-CoR/mSin3/HDAC1 complex. Proc. Natl Acad. Sci. USA, 95, 10860–10865. 7. Lutterbach, B., Sun, D., Schuetz, J. & Hiebert, S. W. (1998). The MYND motif is required for repression of basal transcription from the multidrug resistance 1 promoter by the t(8;21) fusion protein. Mol. Cell Biol. 18, 3604–3611. 8. Chevallier, N., Corcoran, C. M., Lennon, C., Hyjek, E., Chadburn, A., Bardwell, V. J. et al. (2004). ETO protein of t(8;21) AML is a corepressor for Bcl-6 B-cell lymphoma oncoprotein. Blood, 103, 1454–1463. 9. Lemonde, S., Turecki, G., Bakish, D., Du, L., Hrdina, P. D., Bown, C. D. et al. (2003). Impaired repression at a 5-hydroxytryptamine 1A receptor gene polymorphism associated with major depression and suicide. J. Neurosci. 23, 8788–8799.
507
Structure and Functional Analysis of the MYND Domain
26. 27.
TE
28.
R
ET R
D
25.
Crystallography & NMR system: a new software suite for macromolecular structure determination. Acta Crystallog. sect. D, 54, 905–921. Pelton, J. G., Torchia, D. A., Meadow, N. D. & Roseman, S. (1993). Tautomeric states of the active-site histidines of phosphorylated and unphosphorylated IIIGlc, a signal-transducing protein from Escherichia coli, using two-dimensional heteronuclear NMR techniques. Protein Sci. 2, 543–558. Holm, L. & Sander, C. (1995). DALI: a network tool for protein structure comparison. Trends Biochem. Sci. 20, 478–480. Kadrmas, J. L., Beckerle, M. C. & The, L. I. M. (2004). domain: from the cytoskeleton to the nucleus. Nature Rev. Mol. Cell. Biol. 5, 920–931. Bottomley, M. J., Stier, G., Pennacchini, D., Legube, G., Simon, B., Akhtar, A. et al. (2005). NMR structure of the first PHD finger of autoimmune regulator protein (AIRE1). Insights into autoimmune polyendocrinopathy-candidiasis-ectodermal dystrophy (APECED) disease. J. Biol. Chem. 280, 11505–11512. Grishin, N. V. (2001). Treble clef finger–a functionally diverse zinc-binding structural motif. Nucl. Acids Res. 29, 1703–1714. Marti-Renom, M. A., Stuart, A., Fiser, A. S. A., Sa´nchez, R., Melo, F. & Sali, A. (2000). Comparative protein structure modeling of genes and genomes. Annu. Rev. Biophys. Biomol. Struct. 29, 291–325. Krishna, S. S., Majumdar, I. & Grishin, N. V. (2003). Structural classification of zinc fingers: survey and summary. Nucl. Acids Res. 31, 532–550. Lee, J. O., Russo, A. A. & Pavletich, N. P. (1998). Structure of the retinoblastoma tumour-suppressor pocket domain bound to a peptide from HPV E7. Nature, 391, 859–865. Delaglio, F., Grzesiek, S., Vuister, G. W., Zhu, G., Pfeifer, J. & Bax, A. (1995). NMRPipe: a multidimensional spectral processing system based on UNIX pipes. J. Biomol. NMR, 6, 277–293. Johnson, B. A. (2004). Using NMRView to visualize and analyze the NMR spectra of macromolecules. Methods Mol. Biol. 278, 313–352. Kuboniwa, H., Grzesiek, S., Delaglio, F. & Bax, A. (1994). Measurement of HN-H alpha J couplings in calcium-free calmodulin using new 2D and 3D waterflip-back methods. J. Biomol. NMR, 4, 871–878. Cornilescu, G., Delaglio, F. & Bax, A. (1999). Protein backbone angle restraints from searching a database for chemical shift and sequence homology. J. Biomol. NMR, 13, 289–302. Ru¨ckert, M. a. O. G. (2000). Alignment of biological macromolecules in novel nonionic crystalline media for NMR experiments. J. Am. Chem. Soc. 122, 7793–7797. Farrow, N. A., Muhandiram, R., Singer, A. U., Pascal, S. M., Kay, C. M., Gish, G. et al. (1994). Backbone dynamics of a free and phosphopeptide-complexed Src homology 2 domain studied by 15N NMR relaxation. Biochemistry, 33, 5984–6003. Linge, J. P., Williams, M. A., Spronk, C. A., Bonvin, A. M. & Nilges, M. (2003). Refinement of protein structures in explicit solvent. Proteins: Struct. Funct. Genet. 50, 496–506. Tjandra, N., Omichinski, J. G., Gronenborn, A. M., Clore, G. M. & Bax, A. (1997). Use of dipolar 1H–15N
29.
30.
AC
10. Michelson, R. J., Collard, M. W., Ziemba, A. J., Persinger, J., Bartholomew, B. & Huggenvik, J. I. (1999). Nuclear DEAF-1-related (NUDR) protein contains a novel DNA binding domain and represses transcription of the heterogeneous nuclear ribonucleoprotein A2/B1 promoter. J. Biol. Chem. 274, 30510–30519. 11. Bottomley, M. J., Collard, M. W., Huggenvik, J. I., Liu, Z., Gibson, T. J., Sattler, M. & The, S. A. N. D. (2001). domain structure defines a novel DNAbinding fold in transcriptional regulation. Nature Struct. Biol. 8, 626–633. 12. Sum, E. Y., Peng, B., Yu, X., Chen, J., Byrne, J., Lindeman, G. J. & Visvader, J. E. (2002). The LIM domain protein LMO4 interacts with the cofactor CtIP and the tumor suppressor BRCA1 and inhibits BRCA1 activity. J. Biol. Chem. 277, 7849–7856. 13. Hahm, K., Sum, E. Y., Fujiwara, Y., Lindeman, G. J., Visvader, J. E. & Orkin, S. H. (2004). Defective neural tube closure and anteroposterior patterning in mice lacking the LIM protein LMO4 or its interacting partner Deaf-1. Mol. Cell Biol. 24, 2074–2082. 14. Ansieau, S. & Leutz, A. (2002). The conserved Mynd domain of BS69 binds cellular and oncoviral proteins through a common PXLXP motif. J. Biol. Chem. 277, 4906–4910. 15. Hughes-Davies, L., Huntsman, D., Ruas, M., Fuks, F., Bye, J., Chin, S. F. et al. (2003). EMSY links the BRCA2 pathway to sporadic breast and ovarian cancer. Cell, 115, 523–535. 16. Gottlieb, P. D., Pierce, S. A., Sims, R. J., Yamagishi, H., Weihe, E. K., Harriss, J. V. et al. (2002). Bop encodes a muscle-restricted protein containing MYND and SET domains and is essential for cardiac differentiation and morphogenesis. Nature Genet. 31, 25–32. 17. Sims, R. J., III, Weihe, E. K., Zhu, L., O’Malley, S., Harriss, J. V. & Gottlieb, P. D. (2002). m-Bop, a repressor protein essential for cardiogenesis, interacts with skNAC, a heart- and muscle-specific transcription factor. J. Biol.Chem. 277, 26524–26529. 18. Masselink, H. & Bernards, R. (2000). The adenovirus E1A binding protein BS69 is a corepressor of transcription through recruitment of N-CoR. Oncogene, 19, 1538–1546. 19. Sugawara, K., Morita, K., Ueno, N. & Shibuya, H. (2001). BIP, a BRAM-interacting protein involved in TGF-beta signalling, regulates body length in Caenorhabditis elegans. Genes Cells, 6, 599–606. 20. Capili, A. D., Schultz, D. C., Rauscher, I. F. & Borden, K. L. (2001). Solution structure of the PHD domain from the KAP-1 corepressor: structural determinants for PHD, RING and LIM zinc-binding domains. EMBO J. 20, 165–177. 21. Clore, G. M. & Gronenborn, A. M. (1998). New methods of structure refinement for macromolecular structure determination by NMR. Proc. Natl Acad. Sci. USA, 95, 5891–5898. 22. Sattler, M., Schleucher, J. & Griesinger, C. (1999). Heteronuclear multidimensional NMR experiments for the structure determination of proteins in solution employing pulsed field gradients. Prog. NMR Spectrosc. 34, 93–158. 23. Linge, J. P., O’Donoghue, S. I. & Nilges, M. (2001). Automated assignment of ambiguous nuclear overhauser effects with ARIA. Methods Enzymol. 339, 71–90. 24. Bru¨nger, A. T., Adams, P. D., Clore, G. M., DeLano, W. L., Gros, P., Grosse-Kunstleve, R. W. et al. (1998).
31.
32.
33.
34. 35.
36.
37.
38.
39.
40.
508
Structure and Functional Analysis of the MYND Domain
and 1H–13C couplings in the structure determination of magnetically oriented macromolecules in solution. Nature Struct. Biol. 4, 732–739. 41. Laskowski, R. A., Rullmannn, J. A., MacArthur, M. W., Kaptein, R. & Thornton, J. M. (1996). AQUA and PROCHECK-NMR: programs for checking the quality of protein structures solved by NMR. J. Biomol. NMR, 8, 477–486.
42. Koradi, R., Billeter, M. & Wuthrich, K. (1996). MOLMOL: a program for display and analysis of macromolecular structures. J. Mol. Graph. 14. 51-55. 43. Cornilescu, G., Marquardt, J. L., Ottiger, M. & Bax, A. (1998). Validation of protein structure from anisotropic carbonyl chemical shifts in a dilute liquid crystalline phase. J. Am. Chem. Soc. 120, 6836–6837.
Edited by M. F. Summers
R
ET
R
AC
TE
D
(Received 16 November 2005; received in revised form 24 January 2006; accepted 24 January 2006) Available online 8 February 2006
J. Mol. Biol. (2006) 358, 498–508
Structure and Functional Analysis of the MYND Domain
EMBL Heidelberg, Meyerhofstrasse 1, 69117 Heidelberg, Germany 2
Inserm U412, Laboratoire de Virologie Humaine, ENS-Lyon 46, alle´e d’Italie, 69364 Lyon France 3
q 2006 Elsevier Ltd. All rights reserved.
Keywords: MYND; DEAF-1; BS69; zinc binding; transcriptional regulation
ET
*Corresponding author
R
AC
Istituto di Ricerche di Biologia Molecolare, P. Angeletti, Via Pontina, Km 30.600; 00040 Pomezia (RM), Italy
The MYND domain (named after myeloid translocation protein 8, Nervy, and DEAF-1) is a conserved zinc binding domain. It is defined by seven conserved cysteine residues and a single histidine residue that are arranged in a C4–C2HC consensus. MYND domains exist in a large number of proteins that play important roles in development or are associated with cancers and have been shown to mediate protein–protein interactions, mainly in the context of transcriptional regulation. We have determined the three-dimensional structure of the MYND domain from human deformed epidermal autoregulatory factor-1 (DEAF1). The structure reveals a novel zinc binding fold, in which the C4–C2HC motif forms two sequential zinc binding sites. The first and second zinc binding modules comprise a small b hairpin and two short a helices, respectively. The sequential topology of the two zinc binding sites is distinct from the cross-brace PHD and RING finger folds but has some resemblance to LIM domains. The structure reveals that the MYND domain is a novel member of the treble-clef family of zinc binding domains. The MYND domains of BS69 and BOP bind ligands comprising a PXLXP peptide motif. On the basis of the solution structure of the DEAF-1 MYND domain we calculated a homology model of the MYND domain of the BS69 tumor suppressor. A mutational analysis of the BS69 MYND domain indicates that positively charged residues located on one face of its MYND domain are crucial for the molecular interactions of BS69. Different binding specificities of MYND domains may depend on distinct surface charge distributions.
TE
1
D
Roberta Spadaccini1, He´le`ne Perrin2, Matthew J. Bottomley3 Ste´phane Ansieau2 and Michael Sattler1*
Introduction
R
Modification of chromatin structure plays a crucial regulatory role in various processes including transcription, DNA modification and repair. Multiple protein complexes, through different ways, have been implicated in remodeling of chromatin structure. Many components of these complexes share common protein domains that behave as independent functional modules and create or recognize unique histone modifications. For instance, some domains create post-translationally modified lysine residues in histone N termini (HATand SET domains) and others
Abbreviations used: DEAF-1, deformed epidermal autoregulatory factor-1; MTG, myeloid translocation protein; NOE, nuclear Overhauser effect; NOESY, NOE spectroscopy; HSQC, heteronuclear single quantum coherence; GST, glutathione-S-transferase. E-mail address of the corresponding author: [email protected]
recognize these modifications (bromo- and chromodomains, respectively).1,2 The MYND domain (myeloid translocation protein 8, Nervy, DEAF-1) is a novel cysteine-rich structure present in proteins generally implicated in gene regulation and associated with cancers.3,4 MTG8 (also named CBFA2T1, ETO, CDR and Nervy) is a member of the MTG (myeloid translocation protein) family and was first identified as one of the translocation partners of the AML1 protein in acute myeloid leukemia.5 In the AML-MTG8 fusion protein, MTG8 represses AML1 target gene expression and contributes to tumor progression by recruiting the transcriptional corepressor complex Sin3Aa/SMRT(N-CoR)/ HDAC1.6,7 MTG8 was also shown to associate with Bcl-6, a transcription factor involved in normal B cell development, the expression of which is altered in non-Hodgkin lymphoma.8 DEAF-1 is required for embryonic development and is linked to clinical depression and suicidal behaviour in humans.9 DEAF-1 has been shown to
0022-2836/$ - see front matter q 2006 Elsevier Ltd. All rights reserved.
499
Structure and Functional Analysis of the MYND Domain
TE
D
The solution structure of the MYND domain was determined using standard heteronuclear NMR techniques.21,22 Elements of regular secondary structure were identified on the basis of nuclear Overhauser effect (NOE) patterns observed in 3D NOE spectroscopy (NOESY) experiments, and by 13 a C and 13 C b secondary chemical shifts (Figure 1(c)). The presence of dab(i,iC3) NOEs for residues 526–529, 528–531, 530–533 and 535–539, and positive 13Ca secondary chemical shifts of Thr526–Lys531 and Trp533–His538 indicate an a-helical conformation of these residues. 15 N relaxation data were recorded with a 0.5 mM 15 N-labeled MYND domain sample. The rotational correlation time is tcz9.1 ns as estimated from the trimmed average T1/T2 ratio. This value seems unusually high for a protein of 6 kDa molecular mass. Comparison of 1H–15N heteronuclear single quantum coherence (HSQC) spectra recorded at 0.5 mM and 20 mM concentration shows small chemical shift changes for some of the backbone amide signals. The residues affected correspond to conserved hydrophobic residues exposed on the surface of the MYND domain. This may indicate a slight non-specific aggregation at higher concentrations, such as used in the NMR experiments. Furthermore, the MYND domain comprises only a few elements of secondary structure, and a large part of the fold is in an extended conformation. In addition, about 20% of the residues in the protein construct studied are unstructured and highly flexible in solution based on 15 N relaxation measurements (Figure 1(c)). Taken together, the largely extended conformation of the MYND domain and a small tendency to aggregate may explain the unusually high correlation time. Consistent with this, the molecular mass determined by analytical ultracentrifugation experiments indicates that the protein is monomeric in solution (data not shown).
R
ET
R
AC
regulate transcription,3,10 and this function requires the recognition of target DNA sequences by its SAND domain.11 Through LMO4, DEAF-1 also interacts with the tumor suppressor BRCA1, potentially linking DEAF-1 to breast cancer development.12,13 Among the MYND domain-containing proteins, BS69 has been described as a co-repressor in association with various transcription factors such as c-Myb, B-Myb, Ets-2 and the Myc-related MGA protein14 and as a partner of the BRCA2 repressor EMSY.15 Interestingly, the adenoviral oncoprotein E1A and the Epstein-Barr virus-induced nuclear antigen 2 (EBNA2) directly bind to the BS69 MYND domain and disrupt BS69 complexes with cellular factors. Like the cellular partners MGA and EMSY, these viral proteins bind to BS69 through a PXLXP peptide motif (XZany amino acid).14,15 The binding of PXLXP motifs was also observed for the MYND domain of the muscle-specific transcriptional repressor BOP.16,17 However, distinct ligand binding specificities exist since the MYND domain of RACK7 fails to interact with BS69 PXLXP ligands.14 MTG8 and BS69 MYND domains have also been shown to bind directly to N-CoR and SMRT.6,7,18 However, SMRT lacks a PXLXP motif. Similarly, BIP, the binding partner of the Bra-1 MYND domain does not display a PXLXP peptide motif,19 indicating diversification in MYND domain binding properties. Despite the biological importance of MYND domain-containing proteins, the three-dimensional structure and molecular functions of the MYND domain have not yet been described. Here, we present the structure of the human DEAF-1 MYND domain. Our structure reveals a novel zinc binding fold comprising two sequential zinc binding sites, which shares some features with LIM domains.20 The amino acid sequences of DEAF-1 and BS69 MYND domains are closely related. Based on the structural information derived for the DEAF-1 MYND, we performed a structural and functional analysis of BS69. Our data highlight the biological importance of charged residues located on one face of the BS69 MYND domain.
Results
Structure determination On the basis of multiple sequence alignments two MYND domain constructs, comprising residues 496–544 and 496–565 of human DEAF-1, were designed. Both proteins were solubly expressed and 1D NMR spectra were acquired (data not shown). The shorter fragment was chosen for structure determination, since the NMR spectra of both proteins are very similar and additional signals in the 1D spectrum of the longer construct exhibit random coil chemical shifts, indicating that the additional residues are unstructured.
Structure and backbone dynamics of the MYND domain The solution structure of the DEAF-1 MYND domain (Figure 1(a) and (b)) was calculated using CNS/ARIA 1.2.23,24 All distance restraints were derived from manually assigned NOE cross-peaks. Experimental restraints and structural statistics are summarized in Table 1. Residues 503–541 adopt a well-defined tertiary structure. The ten N and three C-terminal residues are disordered as indicated by the paucity of NOEs and negative heteronuclear {1H}–15N NOE values (Figure 1(c)). Some variations are observed for the local residue-specific correlation times as judged from the ratio of 15N T1/T2 relaxation times. These may indicate some flexibility for the backbone of the MYND fold at microto milliseconds time-scales. The first and second zinc binding sites are formed by Cys504, Cys507, Cys515, Cys518 and Cys524, Cys528, His536, Cys540, respectively. The Nd1 position of the His536 side-chain is protonated as
500
R
ET
R
AC
TE
D
Structure and Functional Analysis of the MYND Domain
Figure 1. Structure and dynamics of the DEAF-1 MYND domain. (a) Stereo view of the ensemble of ten lowest energy structures of the DEAF-1 MYND domain (residues 501–541). Secondary structure elements are colored in red and blue for a helices and b strands, respectively. The zinc atoms are shown as green spheres. (b) Ribbon representation of the DEAF-1-MYND domain. Side-chains of the residues coordinating the zinc atoms and conserved hydrophobic residues (W533, Y523, A511 and V505) are shown in green and yellow, respectively. (c) Amino acid sequence, secondary structure and NMR data for the DEAF-1 MYND domain. Residues coordinating the first and second zinc are printed red and blue, respectively. Below: the sequence is shown: on top: difference of 13Ca and 13Cb secondary chemical shifts; middle: heteronuclear NOE; bottom: ratio of 15N T1/T2 relaxation times.
501
Structure and Functional Analysis of the MYND Domain
Table 1. Structural statistics for the DEAF-1 MYND domain
hSAi 0.023G0.002 0.90G0.12
hSAiwater refined 0.035G0.004 1.64G0.23
0.094G0.003
0.119G0.010
0.40G0.21 0.90G0.20
0.47G0.11 1.01G0.12
0.6G0.8
0G0
66.9G3.6 28.0G5.5
69.7G5.6 26.0G5.4
D
˚ )b All distance restraints (A Dihedral anglesc C. Q-factor for experimental residual dipolar coupling restraintsd 1 DHN ˚ )e D. Coordinate precision (A N, Ca, C 0 All heavy atoms E. Structural qualityf Bad contacts Ramachandran plot Residues in most favored region (%) Residues in additionally allowed region (%)
627 (624/3) 174 80 61 311 27 f, 27 j 13
TE
A. Number of structural restraints All NOE (unambiguous/ambiguous) Sequential (jiKjjZ1) Medium-range (1!jiKjj% 4) Long-range (jiKjjO4) Intraresidual Dihedral angles 1 H–15N residual dipolar couplings ˚ ) from experimental restraintsa B. r.m.s. deviation (A
R
AC
a hSAi is an ensemble of ten lowest-energy solution structures (out of 100 calculated) of the MYND domain before water-refinement, hSAiwater-refined is the hSAi ensemble after refinement in a shell of water.39 The CNS Erepel function was used to simulate van der Waals ˚ K4 using PROLSQ van der Waals radii.39 r.m.s. deviations for bond lengths, interactions with an energy constant of 25.0 kcal molK1 A ˚ , 0.426 (G0.012)8 and 0.365 (G0.015)8 before and 0.0055 (G0.0002) A ˚, bond angles and improper dihedral angles are 0.0023 (G0.0001) A 0.782 (G0.043)8 and 2.15 (G0.21) after water-refinement. 1 kcalZ4.18 kJ. b ˚ 2. For restraining Distance restraints were employed with a soft square-well potential using an energy constant of 50 kcal molK1 A the Zn coordination geometry 20 distance restraints were applied as described in Materials and Methods. No distance restraint in the ˚. hSAi structures was violated by more than 0.5 A c Dihedral angle restraints derived from TALOS36 were applied to f, j backbone angles using energy constants of 200 kcal molK1 radK2. No dihedral angle restraint in the hSAi ensemble was violated by more than 5.58. d Quality factor for the RDC refinement.43 Residual dipolar couplings were applied with a final energy constant of 1.0 kcal molK1 HzK2 for an alignment tensor with an axial component of 9 Hz and a rhombicity of 0.4. e Coordinate precision is given as the Cartesian coordinate r.m.s. deviation of the ten lowest-energy structures in the NMR ensemble with respect to their mean structure for residues 504–540. f Structural quality was analyzed using PROCHECK.41
R
ET
determined from the chemical shifts of the 15Nd nitrogen atoms in a 2D 1H–15N long-range HMQC spectrum.25 Thus, His536 coordinates the second zinc atom via the N32 nitrogen of the imidazole ring. Only a few secondary structure elements are present in the MYND domain fold. The first zinc domain comprises a small b hairpin, while the second zinc binding domain contains two short a helices (Thr526–Lys531 and Trp533–His538), connected by a turn around Asp532. The MYND domain has a zinc-dependent fold. Introduction of an excess (20 mM) of the metal-chelator EDTA leads to complete unfolding of the protein as indicated by the loss of chemical shift dispersion in the HSQC spectra (data not shown). Removal of EDTA and the addition of two equivalents of Zn result in complete refolding of the domain. This demonstrates that the uptake of zinc is completely reversible and crucial for proper folding. ˚ and are The two zinc ions are separated by z10 A coordinated by conserved Cys and His residues in the C4 and C2HC clusters, respectively. Aside from the C4–C2HC motif, hydrophobic residues, corresponding to Val505, Ala511, Tyr523 and Trp533 in DEAF-1, are very well conserved in other MYND domains. With the exception of Val505, which
contributes to a small hydrophobic core, these residues are (partially) accessible on the surface of the domain (Figure 1(b)). In addition, the surface of the DEAF-1 MYND domain is highly charged, with a cluster of negatively charged side-chains being exposed from the first zinc binding module (Figure 3(a), below). The combined hydrophobic and negatively charged surface suggests that the DEAF-1 MYND domain could mediate protein– protein interactions and is not involved in nucleic acid binding. Consistent with this, point mutations of zinc-binding Cys residues do not abolish the interaction of full-length (Drosophila) DEAF-1 with DNA.3 Since these mutations are expected to disrupt the three-dimensional fold of the protein it can be concluded that the MYND domain does not contribute to DNA binding, a function that rather involves the DEAF-1 SAND domain.11 Comparison with other zinc binding domains The MYND domain adopts a unique zincdependent protein fold. Searches with DALI26 did not reveal any significant structural homologues. However, the sequential topology of the two zinc coordination sites in MYND resembles that of LIM
502
TE
D
Structure and Functional Analysis of the MYND Domain
AC
Figure 2. Comparison of tandem zinc binding folds. (a) Alignment of consensus sequences for MYND, LIM, RING and PHD domains. Residues coordinating the first and second zinc are colored red and blue, respectively. (b) Schematic representation of the zinc binding pattern and secondary structure elements in MYND, LIM, RING and PHD domains. (c) Ribbon representation of MYND (PDB accession 2FV6), LIM (1A7I), RING (1CHC) and PHD (1XWH) domains. Sidechains of zinc coordinating or conserved aromatic residues are shown in green and yellow, respectively. Zinc atoms are shown as green spheres.
R
ET
R
domain27 and is distinct to the cross-brace coordination employed by RING and PHD fingers20 (Figure 2). Similar to the LIM domain the residue spacing between the zinc coordinating sites is highly conserved. However, the order of the two zinc binding modules is reversed between the MYND (C4–C2HC) and LIM domains (C2HC–C4) and different secondary structure elements are found. Distinct from the other zinc binding folds shown in Figure 2, the first and second zinc binding modules exhibit b strands and a helices, respectively. The first two zinc coordinating Cys residues in the C4 module are located in a small b-sheet, while the remaining two Cys residues are in an extended conformation. The four zinc coordinating ligands in the second zinc binding site (C2HC) are located within and flanking the two a helices. Notably, in contrast to LIM, RING and PHD, a ˚ ) is found between the relatively short distance (10 A two zinc atoms in the MYND domain (Figure 2(c)). When comparing the three-dimensional structures of MYND, LIM, RING and PHD domains (Figure 2(c)) with a multiple sequence alignment of these zinc binding folds different structural roles of conserved hydrophobic residues can be discerned. For example, a tryptophan residue is conserved in the primary sequences of MYND and PHD domains. However, while this tryptophan contributes to the hydrophobic core in PHD domains28 the corresponding residue (Trp533 in DEAF-1) is solvent-exposed in the structure of the MYND domain (Figures 1 and 2). Similarly, an aromatic residue (Tyr/Phe) precedes the fifth metal coordinating ligand in MYND, PHD and RING domains. However, while in PHD and RING domains this residue is part of the hydrophobic core the
corresponding Tyr523 is surface-exposed in the MYND domain. We thus conclude that the MYND domain is a new member of the treble clef domain family29 that shares some structural features with LIM domains. Significant differences in structural details suggest that it is a novel fold with distinct molecular functions. Molecular interactions of MYND domains In glutathione-S-transferase (GST)-pulldown and immunoprecipitation assays, BS69 and BOP MYND domains have been shown to bind to viral and/or cellular partners via a PXLXP motif 14,15. We could not detect any interaction of the known BS69 binding partners with the DEAF-1 MYND domain precluding further analysis of the binding properties of the DEAF-1 MYND domain with such ligands. Consequently, we decided to examine the binding properties of the BS69 MYND domain, which displays a high level of sequence similarity with the DEAF1 MYND domain (41/47% identity/ similarity over 41 residues) For BS69, various PXLXP partners have been identified, including the Myc-related protein MGA and the viral proteins E1A and EBNA2. Interestingly, despite its homology with BS69 (Figure 4(a), below) indicating a conserved fold, the MYND domain of the RACK7 protein fails to interact with BS69 PXLXP ligands, offering therefore an additional possibility to identify important functional residues.14 Despite considerable efforts, we could not obtain recombinant BS69 MYND domain in amounts suitable for NMR analysis. Therefore, we calculated a homology model based on our structure of the DEAF-1 MYND domain using the program
503
D
Structure and Functional Analysis of the MYND Domain
ligands, but absent from RACK7, ETO/MTG8 and DEAF-1. Mutation of the positively charged residues (RRKR559–562GGGG) disrupts the binding to PXLXP ligands (Figure 4(d)). Individual mutations of these residues further revealed that among these four residues only Arg560 is essential for binding of BS69 to its ligands (Figure 4(e)). As a control, all GST-fusions bind similarly to a novel identified BS69 co-factor, the binding site of which is located upstream from the MYND domain (data not shown). Together the mutational analysis suggests that a positively charged face of the BS69 MYND domain, which comprises positively charged residues at the N and C-terminal ends of the zinc binding fold, is crucial for the interaction with PXLXP ligands (Figure 4(f)).
R
ET
R
AC
MODELLER.30 A surface representation of the BS69 MYND model colored by electrostatic potential indicates a striking separation of charges (Figure 3(b)), which is quite distinct compared to the DEAF-1 MYND domain (Figure 3(a)). An extended positively charged patch is found on one face comprising KKK518-520 and RRKR559-562 at the N and C terminus of the BS69 MYND domain, respectively, while hydrophobic and negative regions are located on the opposite side. A BS69 C523S mutation completely disrupts the binding of all partners without any distinction.14 This strongly suggests that the MYND domain fold is necessary for the interaction with the ligands, since the Cys mutant most likely destroys the first zinc binding site and thus the fold of the MYND domain structure. To further characterize the molecular interactions of the BS69 MYND domain, we generated BS69 proteins harboring point mutations affecting either the positively or negatively charged patches and examined their binding properties in GST pull-down assays (Figure 4). Simultaneous mutation of the three positive charges (KKK518–520DDD) preceding the first zinc binding site abolishes the binding of BS69 to its ligands (Figure 4(b)). Similarly, the charge reversal of Glu527-Glu528 (EE527-528KK), which are located in a neighboring region in the first zinc binding module, strongly reduce the binding of all three PXLXP ligands (Figure 4(c)). Introduction of a negative charge (S515D; Figure 4(d)) also strongly impairs the binding to all three proteins, while the W522Y mutation mainly affects the interaction with the cellular ligand (MGA) but not the viral binding partners (Figure 4(d)). The sequence alignment of MYND domains (Figure 4(a)) highlights a set of positively charged residues located at the extreme C terminus of BS69 (RRKR559–562). Some of these residues are conserved in BS69 and BOP orthologues, which are known to bind PXLXP
TE
Figure 3. Homology modeling and electrostatic surface of MYND domains. (a) Ribbon (left) and surface representation (middle and right) of the DEAF-1 MYND domain. The first and second zinc binding modules are colored orange and purple, respectively. Side-chains of the conserved aromatic residues are shown on the left. Surfaces are colored blue and red for positive and negative electrostatic surface potential, respectively. (b) Ribbon (left) and surface representation (middle and right) of the BS69-MYND domain model. Colors are as for (a). Side-chains of aromatic residues present only in the BS69 are shown in cyan.
Discussion We describe the solution structure of the DEAF-1 MYND domain, in which two zinc ions are coordinated in a sequential manner. Based on the conservation of structurally important residues, the fold is representative for other MYND domains. Although the sequential arrangement of the two zinc binding modules is similar to LIM domains, the MYND domain adopts a unique zinc dependent fold, comprised of a short b hairpin and two short a helices in the first and second zinc binding modules, respectively. Thus, the MYND domain represents a novel member of the treble-clef family of zinc binding folds.29,31 The BS69 and BOP MYND domains have previously been shown to depend on a PXLXP peptide motif present in their binding partners.14,16,17 However, as the BS69-related RACK7 MYND domain fails to interact with BS69 ligands it was unclear whether only a subset of MYND domains bind to PXLXP peptide motifs or whether
504
ET
R
AC
TE
D
Structure and Functional Analysis of the MYND Domain
R
Figure 4. Binding studies and mutational analysis of the BS69 MYND domain. (a) Sequence alignment of MYND domain sequences of human BS69 (accession number X86098, residues 515–562), RACK7 (AB032951, residues 1020–1066), ETO/ MTG8 (X70990, residues 477–525), BOP (NP938015, residues 44–95) and DEAF-1 (NP066288, residues 496–544) proteins. The positions of the mutations performed in the BS69 protein are indicated at the top. (b), (c) and (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 glycine residues. (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. Coimmunoprecipitated and ectopic expressions of E1A were revealed by immunoblotting using an M73-E1A antibody. (f) Ribbon representation of the BS69-MYND domain homology model. Side-chains of residues probed by mutations are colored in red and blue, for negatively and positively charged residues, respectively, and in green for other residues. The orientation of the BS69 MYND domain corresponds to Figure 3(b) after rotation by 908 along the x-axis.
all of them bind to PXLXP motifs with different sequence specificities. We found that the MYND domain of the human DEAF-1 protein does neither bind to short PXLXP peptides nor to established full-length PXLXP containing BS69 ligands. This reenforces that binding to PXLXP ligands is restricted to a subset of MYND domains. To further characterize the molecular interactions and identify the residues that confer the ability to bind to PXLXP ligands we studied the BS69 MYND
domain. Notably, compared to DEAF-1 and most other MYND-containing proteins (based on sequence alignments; Figure 4(a)) the BS69 MYND domain exhibits a unique distribution of charged residues at its surface (Figure 3(b)). An extended positively charged surface is present on one side of the MYND domain, while hydrophobic and negative regions are located on the opposite side. Based on these peculiar surface features, we performed a mutational analysis of BS69
505
Structure and Functional Analysis of the MYND Domain
Materials and Methods Protein expression and purification
TE
D
Expression constructs of human DEAF-1 MYND were prepared in modified pET-24d vectors (Novagen) that express proteins with N-terminal His6 tags, removable by cleavage with TEV protease. The DEAF-1 MYND construct used for structure determination spans wildtype residues 496 to 544 plus an N-terminal insertion of three amino acid residues GAM. The two plasmid constructs were verified by DNA sequencing and the recombinant proteins by mass spectrometry (MALDI). Chemically transformed Escherichia coli (strain BL21-DE3) cells were grown in LB rich medium supplemented with 50 mM ZnCl2 at 37 8C until the absorbance at 595 nm reached 0.4. The cells were further incubated at 23 8C until the absorbance reached 0.6, upon which gene expression was induced by the addition of IPTG to a final concentration of 0.4 mM. After 18 h incubation at 23 8C with shaking, cells were harvested by centrifugation. The cells were resuspended in de-gassed lysis buffer (40 mM Tris–HCl (pH 8.0), 0.2 M NaCl, 5 mM imidazole, 15 mM 2-mercaptoethanol, 50 mM ZnCl2, 0.05% (v/v) NP-40 detergent) plus EDTA-free protease inhibitors (Roche) and were lysed using a microfluidizer (Microfluidics Corporation, MA). Cell debris was removed by centrifugation at 12,000 r.p.m. for 30 min at 4 8C. The soluble His6tagged MYND protein was purified from the supernatant using Ni2C-NTA Superflow affinity resin (QIAGEN). After several washing steps, protein was eluted in 40 mM Tris (pH 8.0), 0.2 M NaCl, 0.3 M imidazole, 15 mM 2-mercaptoethanol, 50 mM ZnCl2. The His6 tag was removed by overnight incubation at 23 8C with TEV protease. During incubation, the sample was dialyzed against 40 mM Tris (pH 8.0), 0.2 M NaCl, 5 mM imidazole, 15 mM 2-mercaptoethanol, 50 mM ZnCl2. The noncleavable His6-tagged TEV protease was then removed using Ni2C-NTA resin. The protein sample was concentrated to a volume of 1 ml and further purified on a Superdex-75 (16/60) gel-filtration column, from which it eluted with a single symmetrical peak profile. The purified sample was exchanged into appropriate buffers for NMR (50 mM sodium phosphate (pH 6.5), 0.15 M NaCl, 5 mM DTT, 50 mM ZnCl2, 0.02% (w/v) NaN3, 10% (v/v) 2H2O, stored under argon gas). For NMR studies, uniformly 15N and 13C/15N-labeled proteins were prepared using M9 minimal growth medium appropriately supplemented with 15N-labeled ammonium chloride and 13 C-labeled glucose. NMR experiments were performed at a protein concentration of 0.5–1.0 mM.
R
ET
R
AC
(Figure 4). Mutations affecting the positive patches strongly reduce the affinity for PXLXP partners, while charge reversal of two glutamate residues has a smaller effect on the interaction. The mutational analysis suggests that a positively charged face of the BS69 MYND domain, comprising arginine and lysine residues at the N and C-terminal ends of the zinc binding fold, is crucial for the interaction with PXLXP ligands (Figure 4(f)). Positively charged residues are found at the C-terminal end of the MYND domains of BS69 and BOP homologues, but not of other proteins, including ETO, RACK7 or DEAF1 (Figure 4(a)). The importance of these residues in BS69 for binding suggests that they may contribute to binding selectivity of certain MYND domains with PXLXP containing proteins. For example, the long hydrophobic lysine and arginine side-chains might interact with proline residues and thus contribute to the molecular recognition of the PXLXP motif by the MYND domain. The positive charges of these side-chains could mediate long-range electrostatic interactions with negative charges, which may be located in regions flanking the PXLXP motif. The MYND interaction might thus involve a larger binding epitope of which PXLXP merely represents a conserved core motif. Interestingly, the number and location of aromatic and specifically tryptophan residues is quite variable amongst MYND domains (Figure 4(a)). Two conserved aromatic residues, which are shared amongst all MYND domain family members, contribute to a hydrophobic patch at this binding surface (Figure 3). In addition, an aromatic residue is found between the first two zinc-coordinating Cys residues in both BS69 (Tyr524) and BOP (His58) MYND domains, but not in other proteins, suggesting that hydrophobic interactions of this aromatic residue could further contribute to the ligand binding of BS69 and BOP. In MTG family proteins, a Trp exists at this position, which is more bulky, and thus might sterically interfere with ligand binding. Two additional Trp residues are found in BS69 and RACK7 (Trp522, Trp536 in BS69) compared to other MYND domains. It is likely that these residues also contribute to distinct molecular interactions of MYND domains. Our studies suggest that the zinc-dependent fold of MYND domains presents a binding surface that involves hydrophobic (aromatic) residues and exhibits a unique distribution of charged surfaceexposed residues that are important for binding. These results are somewhat reminiscent of the binding of viral proteins to the retinoblastoma protein (RB) that not only depends on hydrophobic residues in the pocket binding domain but also involves a patch of lysine residues located at the rim of the binding site.32 Structural analysis of MYND– ligand complexes will be required to clarify details of the molecular recognition.
NMR spectroscopy and resonance assignments The 1H, 13C and 15N resonances of the DEAF-1 MYND domain were assigned using standard triple resonance experiments (HNCACB, HNCA, CBCA(CO)NH), HN detected side-chain TOCSY (H(CCO)NH, (H)C(CO)NH), and HCCH-TOCSY acquired at 22 8C. Experiments were recorded on in-house Bruker DRX600 and DRX500 spectrometers equipped with cryo-probes. For structure determination, 2D-NOESY, 15N-edited HSQC-NOESY and 13C-edited HMQC-NOESY spectra were recorded at 800 MHz or 900 MHz with mixing times of 60 ms and 90 ms. Data were processed with NMRPIPE33 and analyzed using NMRVIEW.34 The tautomeric state of the histidine ring was determined with a 2D 1H–15N HMQC experiment optimized for histidine side-chains.25
506
Structure and Functional Analysis of the MYND Domain
Structure calculations
The coordinates and NMR restraints for the DEAF-1 MYND domain have been deposited at the Protein Data Bank under accession number 2FV6.
Acknowledgements We thank A. Gasch, B. Simon and G. Stier (EMBL Heidelberg) for help; and M. Beyermann (FMP, Berlin) for peptide synthesis, M. Collard and J. Huggenvik (Southern Illinois University, USA) for useful discussions. R.S. and M.J.B. are grateful for long-term EMBO and Marie-Curie fellowships, respectively. This work was supported by the Deutsche Forschungsgemeinschaft and EMBL.
Supplementary Data
Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j. jmb.2006.01.087
ET
R
AC
Structures were calculated using the experimentally derived restraints with ARIA1.223/CNS.24 A total of nine iterations (20 structures in the first seven iterations) were performed, in the last two iterations 100 structures were calculated. Initial rounds of refinement using only NOE data defined the general fold of the domain and revealed the zinc coordination. In the final refinement calculation, ˚) distance restraints were added for the Zn–Sg (1.8–2.8 A ˚ ) bonds, and between the four zinc and Zn–N32 (1.5–2.5 A ˚ ) to ensure coordinating atoms at each site (3.6–4.0 A tetrahedral zinc coordination geometry. The structures were refined in a shell of water molecules as described.39 For the water refinement the Zn coordination geometry ˚ ; Zn–N32 2.0 A ˚ ) and was defined by bonds (Zn–Sg 2.3 A angles (109.58 for S–Zn–S, N32–Zn–S, Zn–S–C and 1208 for Zn–N–C angles, respectively). Residual dipolar couplings were employed as orientational restraints as described.40 The quality of the structure was assessed using PROCHECK-NMR.41 Molecular images were generated using MOLMOL.42 Structure similarity searches were performed using DALI†. The structure of BS69-MYND (residues E423 to E485) was generated by comparative modeling with the program MODELLER30 based on a sequence alignment (Figure 4(a)) using the coordinates of DEAF-1 MYND as template.
Protein Data Bank accession codes
D
J (HN,Ha) coupling constants were measured to derive restraints for f dihedral angles.35 Additional f/j restraints were obtained from backbone chemical shifts using TALOS.36 1H–15N residual dipolar couplings were measured in isotropic and anisotropic phases, the latter being created by a media containing 5% (w/v) hexaethylene glycol monododecyl ether (C12E6)/n-hexanol (molar ratio 0.64).37 Heteronuclear {1H}–15N NOEs as well as longitudinal and transversal 15N autorelaxation rates were measured at 500 MHz 1H frequency using standard methods.38 The relaxation decays were sampled in an interleaved fashion at 10/14 different time-points, between 20 and 1600 ms (T1) and 14.4 and 244.8 ms (T2), respectively. The relaxation data were analyzed with NMRView version 4.34
TE
3
Binding experiments
R
Mutations were performed using the QuickChange mutagenesis kit (Stratagene). Wild-type and mutants BS69 GST fusion proteins encompass the residues 411 to 562 of the human BS69 protein (accession number X86098). GST-protein production, GST pull-down assay, cell transfections and immunoprecipitation experiments have been described.14 For immunoprecipitation experiments, briefly, the BS69 fragment (amino acid residues 451–562 R560G) was sub-cloned as the FLAG-pcDNA3 vector. Combination of BS69 and 12SE1A expression vectors were transiently transfected in fibroblasts (QT6). Cells were harvested 24 h post-transfection, lysed and successively incubated with an anti-FLAG M2 antibody (Integra Biosciences) and protein A Sepharose (Amersham Biosciences Inc.). After separation of proteins on SDS-PAGE, BS69 and E1A were visualized with an M2 anti-FLAG antibody and an M73 anti-E1A antibody (Calbiochem), respectively. † http://www.ebi.ac.uk/dali/index.html
References
1. Jenuwein, T. & Allis, C. D. (2001). Translating the histone code. Science, 293, 1074–1080. 2. Bottomley, M. J. (2004). Structures of protein domains that create or recognize histone modifications. EMBO Rep. 5, 464–469. 3. Gross, C. T. & McGinnis, W. (1996). DEAF-1, a novel protein that binds an essential region in a Deformed response element. EMBO J. 15, 1961–1970. 4. Huggenvik, J. I., Michelson, R. J., Collard, M. W., Ziemba, A. J., Gurley, P. & Mowen, K. A. (1998). Characterization of a nuclear deformed epidermal autoregulatory factor-1 (DEAF-1)-related (NUDR) transcriptional regulator protein. Mol. Endocrinol. 12, 1619–1639. 5. Rossetti, S., Hoogeveen, A. T. & Sacchi, N. (2004). The MTG proteins: chromatin repression players with a passion for networking. Genomics, 84, 1–9. 6. Wang, J., Hoshino, T., Redner, R. L., Kajigaya, S. & Liu, J. M. (1998). ETO, fusion partner in t(8;21) acute myeloid leukemia, represses transcription by interaction with the human N-CoR/mSin3/HDAC1 complex. Proc. Natl Acad. Sci. USA, 95, 10860–10865. 7. Lutterbach, B., Sun, D., Schuetz, J. & Hiebert, S. W. (1998). The MYND motif is required for repression of basal transcription from the multidrug resistance 1 promoter by the t(8;21) fusion protein. Mol. Cell Biol. 18, 3604–3611. 8. Chevallier, N., Corcoran, C. M., Lennon, C., Hyjek, E., Chadburn, A., Bardwell, V. J. et al. (2004). ETO protein of t(8;21) AML is a corepressor for Bcl-6 B-cell lymphoma oncoprotein. Blood, 103, 1454–1463. 9. Lemonde, S., Turecki, G., Bakish, D., Du, L., Hrdina, P. D., Bown, C. D. et al. (2003). Impaired repression at a 5-hydroxytryptamine 1A receptor gene polymorphism associated with major depression and suicide. J. Neurosci. 23, 8788–8799.
507
Structure and Functional Analysis of the MYND Domain
26. 27.
TE
28.
R
ET R
D
25.
Crystallography & NMR system: a new software suite for macromolecular structure determination. Acta Crystallog. sect. D, 54, 905–921. Pelton, J. G., Torchia, D. A., Meadow, N. D. & Roseman, S. (1993). Tautomeric states of the active-site histidines of phosphorylated and unphosphorylated IIIGlc, a signal-transducing protein from Escherichia coli, using two-dimensional heteronuclear NMR techniques. Protein Sci. 2, 543–558. Holm, L. & Sander, C. (1995). DALI: a network tool for protein structure comparison. Trends Biochem. Sci. 20, 478–480. Kadrmas, J. L., Beckerle, M. C. & The, L. I. M. (2004). domain: from the cytoskeleton to the nucleus. Nature Rev. Mol. Cell. Biol. 5, 920–931. Bottomley, M. J., Stier, G., Pennacchini, D., Legube, G., Simon, B., Akhtar, A. et al. (2005). NMR structure of the first PHD finger of autoimmune regulator protein (AIRE1). Insights into autoimmune polyendocrinopathy-candidiasis-ectodermal dystrophy (APECED) disease. J. Biol. Chem. 280, 11505–11512. Grishin, N. V. (2001). Treble clef finger–a functionally diverse zinc-binding structural motif. Nucl. Acids Res. 29, 1703–1714. Marti-Renom, M. A., Stuart, A., Fiser, A. S. A., Sa´nchez, R., Melo, F. & Sali, A. (2000). Comparative protein structure modeling of genes and genomes. Annu. Rev. Biophys. Biomol. Struct. 29, 291–325. Krishna, S. S., Majumdar, I. & Grishin, N. V. (2003). Structural classification of zinc fingers: survey and summary. Nucl. Acids Res. 31, 532–550. Lee, J. O., Russo, A. A. & Pavletich, N. P. (1998). Structure of the retinoblastoma tumour-suppressor pocket domain bound to a peptide from HPV E7. Nature, 391, 859–865. Delaglio, F., Grzesiek, S., Vuister, G. W., Zhu, G., Pfeifer, J. & Bax, A. (1995). NMRPipe: a multidimensional spectral processing system based on UNIX pipes. J. Biomol. NMR, 6, 277–293. Johnson, B. A. (2004). Using NMRView to visualize and analyze the NMR spectra of macromolecules. Methods Mol. Biol. 278, 313–352. Kuboniwa, H., Grzesiek, S., Delaglio, F. & Bax, A. (1994). Measurement of HN-H alpha J couplings in calcium-free calmodulin using new 2D and 3D waterflip-back methods. J. Biomol. NMR, 4, 871–878. Cornilescu, G., Delaglio, F. & Bax, A. (1999). Protein backbone angle restraints from searching a database for chemical shift and sequence homology. J. Biomol. NMR, 13, 289–302. Ru¨ckert, M. a. O. G. (2000). Alignment of biological macromolecules in novel nonionic crystalline media for NMR experiments. J. Am. Chem. Soc. 122, 7793–7797. Farrow, N. A., Muhandiram, R., Singer, A. U., Pascal, S. M., Kay, C. M., Gish, G. et al. (1994). Backbone dynamics of a free and phosphopeptide-complexed Src homology 2 domain studied by 15N NMR relaxation. Biochemistry, 33, 5984–6003. Linge, J. P., Williams, M. A., Spronk, C. A., Bonvin, A. M. & Nilges, M. (2003). Refinement of protein structures in explicit solvent. Proteins: Struct. Funct. Genet. 50, 496–506. Tjandra, N., Omichinski, J. G., Gronenborn, A. M., Clore, G. M. & Bax, A. (1997). Use of dipolar 1H–15N
29.
30.
AC
10. Michelson, R. J., Collard, M. W., Ziemba, A. J., Persinger, J., Bartholomew, B. & Huggenvik, J. I. (1999). Nuclear DEAF-1-related (NUDR) protein contains a novel DNA binding domain and represses transcription of the heterogeneous nuclear ribonucleoprotein A2/B1 promoter. J. Biol. Chem. 274, 30510–30519. 11. Bottomley, M. J., Collard, M. W., Huggenvik, J. I., Liu, Z., Gibson, T. J., Sattler, M. & The, S. A. N. D. (2001). domain structure defines a novel DNAbinding fold in transcriptional regulation. Nature Struct. Biol. 8, 626–633. 12. Sum, E. Y., Peng, B., Yu, X., Chen, J., Byrne, J., Lindeman, G. J. & Visvader, J. E. (2002). The LIM domain protein LMO4 interacts with the cofactor CtIP and the tumor suppressor BRCA1 and inhibits BRCA1 activity. J. Biol. Chem. 277, 7849–7856. 13. Hahm, K., Sum, E. Y., Fujiwara, Y., Lindeman, G. J., Visvader, J. E. & Orkin, S. H. (2004). Defective neural tube closure and anteroposterior patterning in mice lacking the LIM protein LMO4 or its interacting partner Deaf-1. Mol. Cell Biol. 24, 2074–2082. 14. Ansieau, S. & Leutz, A. (2002). The conserved Mynd domain of BS69 binds cellular and oncoviral proteins through a common PXLXP motif. J. Biol. Chem. 277, 4906–4910. 15. Hughes-Davies, L., Huntsman, D., Ruas, M., Fuks, F., Bye, J., Chin, S. F. et al. (2003). EMSY links the BRCA2 pathway to sporadic breast and ovarian cancer. Cell, 115, 523–535. 16. Gottlieb, P. D., Pierce, S. A., Sims, R. J., Yamagishi, H., Weihe, E. K., Harriss, J. V. et al. (2002). Bop encodes a muscle-restricted protein containing MYND and SET domains and is essential for cardiac differentiation and morphogenesis. Nature Genet. 31, 25–32. 17. Sims, R. J., III, Weihe, E. K., Zhu, L., O’Malley, S., Harriss, J. V. & Gottlieb, P. D. (2002). m-Bop, a repressor protein essential for cardiogenesis, interacts with skNAC, a heart- and muscle-specific transcription factor. J. Biol.Chem. 277, 26524–26529. 18. Masselink, H. & Bernards, R. (2000). The adenovirus E1A binding protein BS69 is a corepressor of transcription through recruitment of N-CoR. Oncogene, 19, 1538–1546. 19. Sugawara, K., Morita, K., Ueno, N. & Shibuya, H. (2001). BIP, a BRAM-interacting protein involved in TGF-beta signalling, regulates body length in Caenorhabditis elegans. Genes Cells, 6, 599–606. 20. Capili, A. D., Schultz, D. C., Rauscher, I. F. & Borden, K. L. (2001). Solution structure of the PHD domain from the KAP-1 corepressor: structural determinants for PHD, RING and LIM zinc-binding domains. EMBO J. 20, 165–177. 21. Clore, G. M. & Gronenborn, A. M. (1998). New methods of structure refinement for macromolecular structure determination by NMR. Proc. Natl Acad. Sci. USA, 95, 5891–5898. 22. Sattler, M., Schleucher, J. & Griesinger, C. (1999). Heteronuclear multidimensional NMR experiments for the structure determination of proteins in solution employing pulsed field gradients. Prog. NMR Spectrosc. 34, 93–158. 23. Linge, J. P., O’Donoghue, S. I. & Nilges, M. (2001). Automated assignment of ambiguous nuclear overhauser effects with ARIA. Methods Enzymol. 339, 71–90. 24. Bru¨nger, A. T., Adams, P. D., Clore, G. M., DeLano, W. L., Gros, P., Grosse-Kunstleve, R. W. et al. (1998).
31.
32.
33.
34. 35.
36.
37.
38.
39.
40.
508
Structure and Functional Analysis of the MYND Domain
and 1H–13C couplings in the structure determination of magnetically oriented macromolecules in solution. Nature Struct. Biol. 4, 732–739. 41. Laskowski, R. A., Rullmannn, J. A., MacArthur, M. W., Kaptein, R. & Thornton, J. M. (1996). AQUA and PROCHECK-NMR: programs for checking the quality of protein structures solved by NMR. J. Biomol. NMR, 8, 477–486.
42. Koradi, R., Billeter, M. & Wuthrich, K. (1996). MOLMOL: a program for display and analysis of macromolecular structures. J. Mol. Graph. 14. 51-55. 43. Cornilescu, G., Marquardt, J. L., Ottiger, M. & Bax, A. (1998). Validation of protein structure from anisotropic carbonyl chemical shifts in a dilute liquid crystalline phase. J. Am. Chem. Soc. 120, 6836–6837.
Edited by M. F. Summers
R
ET
R
AC
TE
D
(Received 16 November 2005; received in revised form 24 January 2006; accepted 24 January 2006) Available online 8 February 2006