- Email: [email protected]
Crystal structure of MEF2A core bound to DNA at 1.5 Å resolution1
doi:10.1006/jmbi.2000.3568 available online at http://www.idealibrary.com on
J. Mol. Biol. (2000) 297, 437±449
Crystal Structure of MEF2A Core Bound to DNA at Ê Resolution 1.5 A Eugenio Santelli and Timothy J. Richmond* ETH Zurich, Institut fuÈr Molekularbiologie und Biophysik, ETH-HoÈnggerberg CH 8093, ZuÈrich, Switzerland
Members of the myocyte enhancer factor-2 (MEF2) family of transcription factors bind to and activate transcription through A T-rich DNA sequences found primarily, but not exclusively, in the promoters of muscle-speci®c genes. Their importance has been established for myogenic development and in activation of the immediate-early gene, c-jun, and recently further functional roles in the immune system have emerged. The MEF2 factors belong to the MADS-box superfamily, sharing homology in a 58 amino acid domain that mediates DNA binding and dimerization. The structures of two MADS-box proteins, SRF and MCM1, bound to their cognate DNA have been previously reported and shown to share extensive similarity in their mode of DNA binding. We have solved the structure of MEF2A 2-78 bound to its DNA consensus Ê resolution. It reveals how the absence of amino acids sequence at 1.5 A N-terminal to the MADS-box contributes to the DNA binding properties of MEF2 proteins and shows that the MEF domain C-terminal to the MADS-box adopts a conformation considerably different from the same region in SRF and MCM1. # 2000 Academic Press
*Corresponding author
Keywords: transcription factor; muscle development; protein-DNA recognition; MADS-box; X-ray crystallography
Introduction The myocyte enhancer factor-2 (MEF2) family of transcription factors have an essential role in the myogenesis of cardiac and skeletal muscle cells (Black & Olson, 1998; Kaushal et al., 1994; Lilly et al., 1995; Lin et al., 1997). In vertebrates, it comprises the members MEF2A, B, C and D, each having several splicing variants and together showing more than 85 % identity in an 86-amino acid core domain responsible for DNA binding and dimerization (Breitbart et al., 1993; Gossett et al., 1989; Martin et al., 1993; Yu et al., 1992) (Figure 1(a)). MEF2 proteins bind as homo- and heterodimers to a common DNA consensus sequence, YTA(A/T)4 TAR, found mainly in the control regions of many muscle and nerve speci®c genes but also of other unrelated genes (Andres et al., 1995; Pollock & Treisman, 1991) (Figure 1(b) and (c)). The core domain also forms complexes with proteins that cooperate with or repress MEF2 stimulated tranAbbreviations used: MEF2, myocyte enhancer factor 2; TR, thyroid hormone receptor; SRF, serum response factor. E-mail address of the corresponding author: [email protected] 0022-2836/00/020437±13 $35.00/0
scription activation, most notably myogenic and neurogenic bHLH proteins (e.g. MyoD, myogenin, MASH-1), thyroid hormone receptor (TR), MAP kinase (BMK1/ERK5), and histone deacetylase proteins and associated factors (HDAC4 and MITR) (Black et al., 1996; Kato et al., 1997; Lee et al., 1997; Miska et al., 1999; Molkentin et al., 1995; Sparrow et al., 1999). MEF2 factors display a high level of complexity in their expression pattern and activation potential both temporally and spatially in body development. The best-established role of MEF2 is the regulation of genes essential for differentiation and maintenance of the skeletal and cardiac muscle phenotype, but other important functions crucially dependent on MEF2 factors have come to light since their discovery. Among these are the regulation of smooth muscle (Katoh et al., 1998), vascular (Lin et al., 1998) and nervous system development (Mao et al., 1999; Schulz et al., 1996), multiple roles in the immune system (Han et al., 1997; Rao et al., 1998; Youn et al., 1999), seruminduction of c-jun (Han & Prywes, 1995) and participation in the Epstein-Barr virus lytic switch and mediation of the response to calcium ion signaling (Liu et al., 1997). The ability of MEF2 to activate genes in response to mitogens via the p38 or the # 2000 Academic Press
438
Structure of MEF2A Bound to DNA
Figure 1. (a) Sequence alignment of the DNA-binding/dimerization domains of myocyte-speci®c enhancer factor 2 proteins (four human members MEF2A-D; Drosophila D-MEF2), human serum response factor (SRF) and yeast MCM1. Invariant (*) and highly conserved () amino acid residues for all known MADS-box proteins and secondary structure elements for MEF2A are indicated (Kabsch & Sander, 1983). The MEF2A protein used in this study comprises residues 2-78 (black). The MEF domain homology is not found in SRF or MCM1. (b) DNA site consensus sequences for MEF2A (MEF site) and SRF (CArG-box) determined by in vitro selection (Shore & Sharrocks, 1995). (c) DNA in the crystal structure. The near palindromic duplex with single overhanging A and T bases is required to obtain high quality crystals.
BMK1/ERK5 pathways has also been demonstrated (Han et al., 1997; Kato et al., 1997; Yang et al., 1999; Zetser et al., 1999; Zhao et al., 1999). Therefore, MEF2 must activate, or in certain cases repress, different sets of genes depending on cellular context. MEF2 may be able to modulate its own transactivation properties through various heterodimeric isoforms that it can form (Ornatsky & McDermott, 1996). However, such functional diversity is most likely achieved through interaction with other proteins (Black et al., 1996; Kaushal et al., 1994; Krainc et al., 1998; Lee et al., 1997; Mao & Nadal-Ginard, 1996; Miska et al., 1999; Molkentin et al., 1995; Sparrow et al., 1999; Wilson-Rawls et al., 1999). The MEF2 proteins belong to the MADS-box superfamily of transcription factors characterized by strong sequence homology in a 58-amino acid DNA-binding domain that is N-terminal in MEF2 and in most other family members (Shore & Sharrocks, 1995) (Figure 1(a)). Following immediately C-terminal to the MADS-box is the highly conserved, 28-residue MEF domain which is not found in any other known MADS-box protein and is responsible for the ability of MEF2 factors to form DNA-binding heterodimers with themselves but not with other MADS factors (Molkentin et al., 1996a). MEF2A is 507 amino acid residues in length with at least one activation domain following the core region in the sequence (Black & Olson, 1998). In the case of MEF2C, two activation domains have been identi®ed, but at least while part of the intact protein, they require the MEF domain for full func-
tion (Molkentin et al., 1996a). The MEF domain itself is necessary and suf®cient in the core protein alone to impart transcriptional silencing via interaction with proteins involved with histone deacetylation (Miska et al., 1999; Sparrow et al., 1999). All the splicing isoforms of MEF2 proteins display full DNA binding activity and transactivation potential, and are differentially expressed in vivo (Breitbart et al., 1993; Martin et al., 1993). Different MADS-box proteins bind to DNA sites of related but distinct consensus sequences (Figure 1(b)). The basis of these differences has been the focus of extensive mutational analyses, af®nity/speci®city measurements and proteininduced, DNA distortion experiments (Dodou et al., 1995; Meierhans & Allemann, 1998; Meierhans et al., 1997; Nurrish & Treisman, 1995; Pollock & Treisman, 1991; Sharrocks et al., 1993; West & Sharrocks, 1999; West et al., 1997). The crystal structures of two MADS proteins bound to their cognate DNA, human serum response factor (SRF) and the yeast factor MCM1 in combination with MATa2, have been previously solved by X-ray methods and show that base-speci®c contacts only partially explain the observed sequence speci®city (Pellegrini et al., 1995; Tan & Richmond, 1998). Both SRF and MCM1 are bound to a highly bent DNA duplex, and DNA distortion here as for several other transcription factors plays a major role in sequence speci®c recognition and generation of complex architecture important for promoter function (Perez-Martin & de Lorenzo, 1997). We preÊ X-ray structure of MEF2A core sent here the 1.5 A
439
Structure of MEF2A Bound to DNA
which includes the MADS box and part of the MEF domain bound to its consensus sequence DNA (Figure 1(c)). The crystal structure of MEF2A shows in comparison to SRF and MCM1 similarities and differences in the MADS-box and prominent divergence in the C-terminal domain. The atomic structures of MEF2A, SRF and MCM1 bound to their speci®c DNA sites reveal the variations on DNA binding that occur within this protein family and provide answers as to how they arise.
Results and Discussion MEF2A-DNA complex The MEF2A core protein (amino acid residues 278) is folded in the same three-layer organization observed for SRF and MCM1 (Figure 2). Each monomer of MEF2A contains a long a-helix (aI) and a 12-amino acid N-terminal random coil (Nextension) comprising the DNA contacting layer, and two b-strands (bI, bII) connected by a b-turn forming the middle layer (b-hairpin). Together these two layers de®ne the MADS-box. A second a-helix (aII) makes up the top layer or the major part of the MEF domain and packs against the b-hairpin in an orientation diagonal to the analogous element in SRF (Figure 3(a)). The r.m.s.d. between corresponding Ca atoms of SRF and Ê for MADS box regions MEF2A is 0.50 and 1.77 A 5-38 and 39-58, respectively, calculated after superposing amino acids 145-178 of SRF and 5-38 of MEF2A. The primary differences within the MADS domains occur at amino acids 2-4 of the N-extension bound to DNA and 49-54 of the b-turn. The dimerization surface consists of one face of the monomer analogous with SRF, where aII in MEF2A additionally contacts the b-sheet of the opposite monomer via hydrophobic and hydrogenbonding interactions. Electron density is not visible for the C-terminal amino acid residues 74-78 for either copy of the complex apparently because of disorder. The X-ray structure shows that the DNA is bent overall to only 17 as compared to 72 for SRF (Figure 4). The b-turn in the middle layer of the protein structure is not used for DNA binding in MEF2A whereas it is an important feature of the DNA interaction seen in the SRF-DNA complex. The MEF2A core dimer/DNA assembly crystallized in space group P1 with two copies of the complex per asymmetric unit. The r.m.s.d. after coordinate alignment between all atoms of the two Ê for independent dimers are 0.06, 0.16 and 0.33 A protein Ca atoms only, all protein atoms and all DNA atoms, respectively. The corresponding values comparing subunits in one of the dimers Ê . The solvent accessible are 0.20, 0.58 and 0.74 A Ê probe) buried in the dimerizasurface area (1.4 A Ê 2. All amino acid residues tion interface is 3864 A that could have a major role in the dimerization speci®city, those in aI, bI and aII, are conserved in MEF2A-D so that hetero dimerization can occur
unobstructed. Conservation of amino acid residues is so high throughout the core sequence, it is unlikely that surface side-chains play a selective role in hetereo dimerization either. Although the aI and bI structures of MEF2 and SRF have only minor sequence differences, hetero dimerization of a MEF2 subunit with the SRF subunit would be blocked by incompatibility between the MEF domain and the SRF ``top layer'' structure.
MEF2 domain The MEF2 domain starts as a short coil region (sequence STD) from the C terminus of the b-hairpin, bends back over the b-sheet and continues as the aII helix packed against the underlying b-sheet (Figure 2, Figure 3(a)). The aII helix axis runs perpendicular to the direction of the b-sheet so that it contributes signi®cantly to the dimerization interface of the whole core through predominantly hydrophobic contacts with the twofold related aII helix and b-hairpin of the adjacent subunit. This scheme contrasts with SRF for which the aII helices are mainly associated with the b-strands within the same subunit. The aII helices of MEF2A are separÊ axis to axis as compared to, ated by 13.3(0.3) A for example, the helices in the aI coiled coil spaced Ê . The pair of antiparallel aII helices by 10.1(0.3) A form a substantial cleft in the top layer that likely provides the interaction site in the core required for transactivation, and the binding of the MITR protein and homologous regions in HDAC2-4. Mutation of the VLL sequence in the center of aII to ASR removes the activation potential of MEF2C. Although the V64A alteration in this mutant may allow the aII to fold similarly in the core, since the valine is buried, the L65S and L66R replacements would substantially change the character of the exposed surface of the proposed binding pocket. Most exposed hydrophilic side-chains, those of Asp61, Asp63, Lys64 and Leu67, in the MEF domain are not clearly de®ned by the electron density in any of the four copies. The exception is Lys68 which extends outward from the side of aII and hydrogen bonds in a mixed population with Glu42 at the beginning of bI in the same subunit and Asn52 in the b-turn of the adjacent subunit. Phosphorylation of Ser59 has been shown to increase both the transactivation potential and the DNA binding af®nity of MEF2C, although the biological signi®cance of this modi®cation is not clear (Molkentin et al., 1996b). Ser59 is the ®rst amino acid of the MEF domain and its hydroxyl group is hydrogen bonded to the main-chain amide of Ile43. Phosphorylation at this site could perturb the structure of aII, particularly the relationship of the Lys68 to Glu42 interaction described. It may affect the ability of the MEF2 domain to combine with further factors. Examination of the structure does not suggest any simple explanation for the modest increase in DNA binding observed for phosphorylated species (Molkentin et al., 1996b).
440
Structure of MEF2A Bound to DNA
Figure 2. The MEF2A structure in two orthogonal views. The 2-fold axes of the protein dimer and DNA (silver) are aligned in the crystals. The three-layer organization of a protein monomer extends from the DNA as an N-extension (brown, N-ext) and an a-helix (red, aI) of the antiparallel coiled coil in the ``bottom layer'', followed by two strands (blue, bI-bII) of the four-strand b-sheet making up the ``middle layer''. The top and bottom layers comprise the MADS-box. The a-helix (yellow, aII) of the MEF domain forms the ``top layer'' with its counterpart in the 2-fold related subunit (green).
MEF2A-DNA interactions The MEF2A core protein bends the DNA double helix containing the MEF site consensus sequence by 17 , in good agreement with the value of 19 measured previously in solution (West et al., 1997) (Figure 2). The DNA is essentially straight in comparison to the 72 bend seen in the SRF/DNA
structure (Pellegrini et al., 1995) (Figure 4). However, the twist of the double helix within the binding site is over wound to 9.9 bp/turn while the minor groove at the four central A T base-pairs Ê , about half that for the canoninarrows to 2.7-3.1 A cal B-form conformation (Table 1). The two most important protein units for DNA binding are the aI helix and the N-extension of the
Structure of MEF2A Bound to DNA
441
Figure 3. Main-chain paths for MADS-box proteins (stereographic). (a) Monomers of MEF2A (brown) and SRF (green) are shown in superposition. The b-hairpin of MEF2A is noticeably twisted clockwise compared to SRF, and the b-turn is displaced toward the dimer interface (left). Amino acid residues 5-39 in the N-extension and aI helix were used for the alignment. (b) Comparison of the DNA-binding segments of the N-extensions of MCM1 (17-21), SRF (142-146) and MEF2A (2-6). The structures were aligned by superposing MEF2A amino acid residues 5-39 with the corresponding residues in MCM1 and SRF. The arginine and glycine amino acid residues labeled penetrate into the minor groove making key contacts with the DNA.
MADS-box (Figure 5(a)). As for SRF, the central two turns of the aI helix (amino acid residues 2331) provide the primary DNA-binding contacts over the consensus-binding site (Figure 6). Lys30 hydrogen bonds to a phosphate oxygen of base A 2 from the major groove side, and Lys31 similarly bonds to the adjacent T-3 in the complementary strand from the minor groove side. Arg24 hydrogen bonds to the phosphate group of A 4 from the minor groove side while Lys23 reaches into the major groove and forms hydrogen bonds to the N7 atom of A 4 and simultaneously with the O6 atom of G 5, specifying a purine at the former position and a guanine at the latter. Lys23 makes the only direct contact with bases in the major groove. Arg24 and Lys30 from opposite subunits are linked together through hydrogen bonds with Glu34 forming a seemingly rigid constellation of side-chains important for specifying the local DNA conformation (Table 1). At the extremities of the DNA site, Arg15 bridges over the major groove in two different conformers to make either a
hydrogen bond or salt link to the phosphate group of A-8. The guanidinium group of Arg15 interacts with the phosphate and N7 atom of A-8, in accord with previous results indicating that purine is preferred at this position (Dodou et al., 1995; Pollock & Treisman, 1991). All of these amino acid residues, and including Gly27 which allows the close approach of aI to the DNA backbone, are invariant in all MADS proteins with the exception of Arg15. The side-chains of Ile8, Val19, Thr20, Phe26, Tyr33 and Leu38 appear to further restrict the conformational ¯exibility of the long side-chains involved in these interactions. Tightly bound water molecules mediate further contacts between the DNA phosphate groups and the side-chains of Glu34 and Thr20 and the main-chain carbonyl groups of Lys23, Arg24 and Gly27. The minor groove formed by the eight central base-pairs of the MEF site is in close contact with Gly2 and Arg3. In each half site, the arginine side-chain lies in the minor groove with the opposite orientation as compared to the homologous
442
Structure of MEF2A Bound to DNA
Figure 4. The MEF2A and SRF core/DNA complexes. The aI helices (red) were used for the alignment of the complexes. The packing of the b-sheet (green) on the coiled coil is similar, although not identical, in the two MADS domains, whereas the interaction of the N-extension (yellow) with DNA (blue-cyan) and the orientation of the SRF C-terminal and MEF domains (magenta) on the b-sheet are substantially different for the two complexes. The DNA is bent over an angle four times greater for SRF than MEF2A (72 versus 17 ).
arginine for SRF (Figure 5(a)). Nevertheless, two of three side-chain methylene groups contact the C2 atom of A 4 and the guanidinium group hydrogen bonds to the O2 and O40 atoms of T-4 so that an A T base-pair is clearly speci®ed at this position (Figure 5(b)). The Lys23 further speci®es the orientation of this base-pair via the major groove. The N-terminal amino group of Gly2 is pointed toward the center of the DNA binding site and hydrogen bonds to the N3 atom of A-2, and its amide atom hydrogen bonds to the N3 atom of A 4. The main-chain conformation of the N-extension between the aI helix at Asp13 and Lys4 is virtually identical with that for SRF (Figure 3(a)). Lys4 and Ile6 interact with the phosphodiester backbone to ``push'' the Arg3 side-chain into the DNA groove and the intervening Lys5 makes a salt link with the phosphate group of A 1. Comparison of sequences for different MADSbox factors reveals that the amino acid residues to either side of the highly conserved arginine residue at MADS-box position three are variable (Shore & Sharrocks, 1995). The conformation of the polypeptide chain in this segment of the N-extension is different in the structures of the SRF, MCM1 and MEF2A (Figure 3(b)). DNA complexes due to the size, charge or hydrophobicity of the amino acid
residues immediately ¯anking position 3. For MEF2C and SRF, direct contact by the entire arginine side-chain with the surface of the minor groove affects the sequence of the central A T base-pairs beyond the requirement for a narrow minor groove dictated by the aI coiled coil (Pellegrini et al., 1995). For MCM1, the arginine at position 3 does not invade the minor groove apparently because it is preceded by glutamic acid (Tan & Richmond, 1998). Instead, only the guanidinium group of a second arginine at position 4 is inserted into the minor groove and this arrangement permits greater variability in the binding site sequence. DNA bending by MEF2A and SRF MEF2A and SRF differ greatly in their apparent ability to bend DNA into the major groove. SRF bends DNA around the N-terminal third of the aI helix; MEF2A does not. In the SRF/DNA complex, Lys154 and Lys165 facilitate DNA bending through their interaction with DNA phosphate groups, allowing ¯anking DNA outside the CArGbox to bind to Thr191 and His193 in the b-turn of the middle layer (Pellegrini et al., 1995). The path of the b-turn in MEF2A (amino acid residues 49-53,
443
Structure of MEF2A Bound to DNA Table 1. DNA parameters Minor groove Base-pair A ÿ 8/T 8 A ÿ 7/T 7 G ÿ 6/C 6 C ÿ 5/G 5 T ÿ 4/A 4 A ÿ 3/T 3 TA ÿ 2/AT 2 T ÿ 1/A 1 A 1/T ÿ 1 TA 2/AT ÿ 2 T 3/A ÿ 3 A 4/T ÿ 4 G 5/C ÿ 5 C 6/G ÿ 6 T 7/A ÿ 7 T 8/A ÿ 8
Propeller twist (deg.) ÿ8.9 ÿ15.7 ÿ8.2 ÿ13.0 ÿ11.7 ÿ17.4 ÿ14.6 ÿ17.9 ÿ17.0 ÿ14.2 ÿ17.0 ÿ11.9 ÿ13.2 ÿ8.3 ÿ14.9 ÿ7.1
Roll (deg.)
Tilt (deg.)
Twist (deg.)
6.8 8.6 ÿ4.8 9.3 ÿ7.5 7.3 ÿ1.6 7.9 ÿ2.3 7.5 ÿ7.5 8.8 ÿ4.8 8.1 6.3
ÿ7.5 ÿ0.4 ÿ0.4 5.6 ÿ3.2 ÿ3.7 ÿ0.3 ÿ0.3 0.9 4.4 2.9 ÿ5.1 0.6 0.7 6.8
34.5 31.9 39.2 31.3 49.0 28.8 38.9 37.4 39.4 28.8 48.1 31.6 39.4 32.0 35.0
Major groove
Ê) Width (A
Ê) Depth (A
Ê) Width (A
Ê) Depth (A
4.4 5.8 5.5 3.1 2.7 2.7 3.3 5.5 5.8 4.3
5.5 5.9 6.1 5.4 5.3 5.3 5.4 6.1 5.8 5.6
12.2 11.0 12.1 12.3 13.0 12.9 12.3 12.0 11.0 12.2
3.4 2.9 3.2 3.6 3.9 4.1 3.6 4.2 2.7 3.5
Parameters are averages over two duplexes and two conformers.
sequence NSSNKL) deviates appreciably from that seen in SRF by being displaced toward the dimer interface (Figure 3(a)). Nevertheless, superposition of the SRF and MEF2A structures shows that Gln49, Ser51 and Lys53 in the MEF2A b-turn could in principle provide binding interactions for bent DNA (Figure 7). Lys154 found in SRF is replaced by Glu14 in MEF2A (a glutamine residue in MEF2B). This exchange of oppositely charged sidechains contributes strongly to the lack of signi®cant DNA bending for MEF2A as indicated by mutagenesis experiments (Nurrish & Treisman, 1995; West et al., 1997). This is perhaps surprising since the glutamic acid side-chain could in principle bind to a DNA phosphate group via a water molecule in conjunction with a neighboring arginine side-chain as seen in the nucleosome core structure (Karolin Luger & T.J.R., unpublished results). The structural context of Glu14 within the framework of the MADS-box must therefore be important; the reciprocal exchange experiment of introducing K154E into SRF is suf®cient to signi®cantly reduce the DNA bend (Nurrish & Treisman, 1995; Sharrocks et al., 1993). By comparison, MEF2C which also contains Glu14 does shows a degree of bending intermediate between SRF and MEF2A and suggests that analysis conditions may affect the values obtained (Meierhans et al., 1997). Furthermore, the bend angle measured may not represent a static bend, but the average for at least two states of DNA bending. Conversion of SRF to MEF site specificity SRF N-terminally truncated to Gly142, corresponding to Gly2 in MEF2, binds with high af®nity to MEF2 sites while still displaying residual binding to the CArG-box sequence (Nurrish &
Treisman, 1995; Sharrocks et al., 1993). This truncated version of SRF would allow the N terminus to lie under the aI helices and Arg143 to now specify the base-pair at 4/ÿ 4 as seen from the MEF2A complex (Figure 5). With the native sequence, the SRF N terminus must lie to the outside of the complex, requiring the arginine sidechain to be tucked between the protein and DNA. Truncated SRF bends DNA only slightly less well than the wild-type protein. Conversion to MEF2 site speci®city is completed with the additional mutation K154E (Nurrish & Treisman, 1995; Sharrocks et al., 1993), which would prevent the DNA from contacting the b-loop.
Conclusions MEF2A recognizes its speci®c DNA binding site by selecting a narrow minor groove and by making speci®c contacts with two base-pairs per halfsite. The MADS-box motif has been adapted in different proteins to bind equally well to DNA in a highly bent form as for SRF and MCM1 or to relatively unbent DNA as for MEF2A. The core region immediately C-terminal to the MADS-box, the MEF domain in MEF2 proteins, has evolved to interact with several factors that relay or deliver a transcription activating or silencing signal. The entire core region of MEF2A, as well as SRF and MCM1, is important for interactions with other DNA-binding transactivator proteins that account for promoter site speci®city. Further structural studies pertinent to these multifactor complexes are essential to understand the atomic mechanisms of transcription activation and silencing.
444
Structure of MEF2A Bound to DNA
Figure 5 (legend opposite)
Methods Production and crystallization of the MEF2A/ DNA complex The gene coding for MEF2A 1-78 was created by PCR ampli®cation from synthetic oligonucleotides and cloned into the pET3a expression vector. The bacterially expressed polypeptide (Escherichia coli strain BL21(DE3), 25 C) was puri®ed from the cytosolic soluble fraction by successive chromatographic steps on heparine-Sepharose (Pharmacia), SP-5PW and Phenyl-5PW (TSK) columns run at 4 C and concentrated to approximately 1 mg mlÿ1 in 300 mM NaCl, 5 mM Hepes-Na (pH 7.0), 0.5 mM EDTA and 1 mM DTT to give a ®nal yield of 0.3 mg lÿ1. The polypeptide was also puri®ed from inclusion bodies: the cells grown at 37 C were lysed by freeze-thawing, homogenized, sonicated and centrifuged. The pellet was dissolved in 6 M guanidinium chloride and centrifuged again. The supernatant was ®ltered through a XM50 membrane (Amicon) and the ®ltrate eluted on a Superdex S-200 column with a buffer containing 8 M urea, 100 mM NaCl and 10 mM b-mercaptoethanol. The relevant fractions were precipitated by extensive dialysis against 10 mM b-mercaptoethanol and lyophlized. The polypeptide was folded in two successive dialysis steps: (1) 6 M urea, 0.5 M NaCl, 10 mM Hepes-Na buffer (pH 7.0), 0.5 mM EDTA, 5 mM b-mercaptoethanol to a buffer containing 0.5 M arginine, 0.75 M guanidinium chloride, 0.05 % Triton -100, 100 mM NaCl, 25 mM bis-Tris base (pH 6.3), (2) 10 mM Hepes-Na buffer (pH 7.0), 0.5 M NaCl, 10 mM Hepes-Na (pH 7.0), 0.5 mM EDTA and 1 mM DTT. A ®nal chromatographic step on a Phenyl5PW (TSK) column followed. N-terminal sequencing indi-
cated that the polypeptide began at Gly2, and MALDITOF mass spectrometry analysis showed that the mass was that expected from the expression construct lacking Met1. It is not known if MEF2A Met1 is present in eukaryotic cells. However, since glycine is the second amino acid in the MEF2A sequence, the expectation is that Met1 is removed in vivo (Flinta et al., 1986). All oligonucleotides were synthesized with a Applied Biosystems 380B synthesizer using standard phosphoramidite chemistry and puri®ed by reverse phase HPLC. After addition of 10-30 % excess MEF2 site containing DNA, the complex was puri®ed by gel electrophoresis (BioRad) at 4 C on a 9 % (w/v) acrylamide (acrylamide to bis-acrylamide, 40:1, w/w) gel containing 35.6 mM Tris base/boric acid and 1.0 mM EDTA and eluted into 5 mM Hepes-Na buffer (pH 7.0), 10 mM NaCl, 1 mM DTT. Several DNA duplexes of different length and sequence were tried, and crystals suitable for X-ray diffraction analysis were obtained with a 16-bp oligonucleotide having A-T overhanging bases (Figure 1(c)). Cystallization was by hanging drop vapor diffusion conditions using 15-18 % PEG6000, 50 mM NaNO3, 5 mM Ba(NO3)2, 50 mM bis-Tris buffer (pH 5.8-6.3), 25 mM NaOAc, 2 mM DTT, 22 C in the well. Crystallographic structure solution The needle-shaped crystals (20-60 mm 100-250 mm 400-1200 mm) used for data collection diffracted to Ê using a Rigaku RU200 X-ray generator equipped 1.8 A Ê at with a MAR180 image plate detector and to 1.3 A beam line ID14, ESRF Grenoble using a MAR345. Crystals were ¯ash-cooled in propane at ÿ120 C using 20 %
Structure of MEF2A Bound to DNA
445
Figure 5. The MADS-box aI coiled coil and N-extension bound to DNA. (a) The two aI helices in the protein dimer Ê (green and orange) bind primarily to the narrow minor groove of the MEF site DNA (accessible surface to a 1.4 A probe in red-white-blue, electrostatic negative-neutral-positive). The charged side-chains of Lys23, Arg24, Lys30, Lys31, Glu34, and Arg15 in one DNA half site are shown (CPK colors). Lys30 is shown in two alternative conformations; both bind a DNA phosphate oxygen, but only one binds Glu34. Lys23 is the only residue to make a speci®c contact in the major groove. The N-terminal Gly2 and side-chain of Arg3 (CPK colors) lie buried in the minor groove with Arg3 bound to base-pair A 4/T ÿ 4. (b) Electron density from a simulated annealing omit map (10,000 degrees, amino acid residues 2-17 omitted, contoured at 0.6s) shows the proximity of two Arg3 methylene groups to the C2 atom of base A 4.
(w/v) glycerol as a cryoprotectant. A complete dataset Ê was collected also at beam line ID14, ESRF Greto 1.5 A noble (Table 2). Data reduction was performed using programs DENZO and SCALEPACK (Otwinowski, 1993). The MEF2A structure was solved by a combination of molecular replacement using the SRF structure (pdb accession code 1SRS) and SIR phasing using data collected in the laboratory from a derivative with 5-iodouracil replacing T-4. A poly-alanine model comprising SRF residues 142-196 and base-pairs ÿ5 to 5 was used for the initial search using AMoRe (Navaza, 1994). The heavy-atom positions for the iodo-derivative were predicted from the molecular replacement solution and precisely determined by difference Patterson and Fo ÿ Fc maps. Phases were calculated with MLPHARE Ê (Otwinowski, 1991) and electron density maps to 2.2 A were calculated using combined phases with CNS
(BruÈnger et al., 1998). Density modi®cation by solvent ¯attening and NCS averaging was performed with DM (Collaborative Computational Project, 1994; Cowtan, 1994) and the atomic model was built using the graphics program O (Jones et al., 1991). Re®nement was carried out using data from 20.0 to Ê resolution (Table 3). NCS restraints were applied 1.5 A between the two copies of the complex in the crystal asymmetric unit during the initial simulated annealing re®nement using CNS. The model was improved through cycling energy minimization without NCS using alternately CNS and REFMAC (Murshudov et al., 1999) followed by manual rebuilding with O. Water molecules were added using CNS and WARP (Perrakis et al., 1997). The high resolution of the diffraction data permitted observation of two separate conformers for the DNA in the electron density map. They exist most probably
446
Structure of MEF2A Bound to DNA Table 2. Crystallographic data Dataset Ê) Resolution (A Completeness Overall Last shell (range) Unique reflections Average redundancy Rmerge (%) Overall Ê) Last shell (range A Phasing power Cullis R Unit cell a b c a b g
Native
Iodo derivative
1.5
2.15
96.0 92.0 (1.55-1.50) 87,264 2.6
97.4 89.4 (2.23-2.15) 29,659 4.2
7.5 10.4 (1.55-1.50)
7.9 23.7 (2.23-2.15) 0.70 0.84
41.37 60.70 63.99 115.18 89.99 90.00
41.36 60.73 64.18 115.33 90.05 89.95
obtained by including the DNA model twice, each at half occupancy, and restraining the base-pairs 4/ÿ 4, 3/ÿ 3, 1/ÿ 1, ÿ1/ 1, ÿ3/ 3 and ÿ4/ 4, and bases 5 and ÿ5 between the two atomic models. The maximum rms deviation values for all atoms occuring for unrestrained (8/ ÿ 8) and restrained (ÿ4/ 4) Ê and 0.64 A Ê (0.13 A Ê without phosbase-pairs are 1.73 A phate), respectively. The conformer with the greatest number of hydrogen bonds between protein and DNA was used for display. Several amino acid side-chains were also included in two conformations.
Table 3. Structure re®nement
Figure 6. Schematic of MEF2A/DNA interaction. The direct hydrogen bonds (arrows) and hydrophobic contacts (continuous lines) from a single protein monomer (amino acid labels) to DNA base-pairs (MEF2 site shaded) are shown. Highly localized water molecules mediate some interactions (dotted lines).
because an approximate 2-fold rotational degeneracy superimposes both complexes in the crystal asymmetric unit imperfectly around the 2-fold axis of the complex. The iodo-derivative used for phasing con®rms this degeneracy. Completely symmetric DNA sequences and overhangs did not yield well diffracting crystals. Of the several options tested, the best re®nement results were
Ê) Resolution (A R-factor 20-1.5/1.57-1.50 (%) Free-R-factor 20-1.5/1.57-1.50 (%) Reflections work/free Atoms (incl. multiple conformers) Total Protein DNA Solvent r.m.s.d. from ideality Ê) Bond lengths (A Bond angles (deg.) Dihedral angles (deg.) r.m.s.d. B-factors for bonded atoms Main-chain/target s Side-chains/target s Ramachandran plot Allowed (%) Additionally allowed (%) Average G-factor (PROCHECK)a Ê 2) Average B-factors (A Main-chain Side-chain Solvent DNA backbone DNA bases
20.0-1.5 20.6/25.7 22.9/28.6 77,972/8679 5940 2520 2764 656 0.010 1.44 19.6 1.2/2.0 1.7/3.0 93.8 6.2 0.33 21.1 23.4 38.9 26.1 21.6
No sigma cutoff was applied. The Free R set was created in spacegroup P21, expanded to P1 and combined with the diffraction data, due to the high correlation between re¯ections related by the pseudo-symmetry axis. a All mean values are above 0.
Structure of MEF2A Bound to DNA
447
Figure 7. Superposition of MEF2A aI helix and b-turn and SRF DNA structures. The Ca atoms of the aI helices of both the MEF2A/DNA and SRF/DNA (Pellegrini et al., 1995) complexes were aligned. In this hypothetical complex, the SRF DNA is in near proximity to the b-turn of MEF2A such that interaction might occur similar to SRF. Glu14 apparently inhibits this interaction and prevents signi®cant DNA bending as revealed by its replacement with a lysine residue in a MEF2A mutant protein (West et al., 1997).
Programs CNS, PROCHECK (Laskowski et al., 1993) and CURVES (Lavery & Sklenar, 1988) were used for model analysis. Molecular images were generated with programs MIDASPlus (Ferrin et al., 1988), O (Jones et al., 1991) and WebLab ViewerPro 3.5 (Molecular Simulations Inc). Protein Data Bank accession code The atomic coordinates and structure factors of the MEF2A core/DNA structure were deposited in the Protein Data Bank, accession code 1egw.
Acknowledgments We thank Dr David Sargent for expert assistance in data collection and the staff at ESRF Grenoble, beamline ID14, for excellent technical support. This project was supported in part by the TMR Program of the European Union.
References Andres, V., Cervera, M. & Mahdavi, V. (1995). Determination of the consensus binding site for MEF2 expressed in muscle and brain reveals tissue-speci®c sequence constraints. J. Biol. Chem. 270, 23246-23349.
Black, B. L. & Olson, E. N. (1998). Transcriptional control of muscle development by myocyte enhancer factor-2 (MEF2) proteins. Annu. Rev. Cell Dev. Biol. 14, 167-196. Black, B. L., Ligon, K. L., Zhang, Y. & Olson, E. N. (1996). Cooperative transcriptional activation by the neurogenic basic helix-loop-helix protein MASH1 and members of the myocyte enhancer factor-2 (MEF2) family. J. Biol. Chem. 271, 26659-26663. Breitbart, R. E., Liang, C. S., Smoot, L. B., Laheru, D. A., Mahdavi, V. & Nadal-Ginard, B. (1993). A fourth human MEF2 transcription factor, hMEF2D, is an early marker of the myogenic lineage. Development, 118, 1095-1106. BruÈnger, A. T., Adams, P. D., Clore, G. M., DeLano, W. L., Gros, P., Grosse-Kunstleve, R. W., Jiang, J. S., Kuszewski, J., Nilges, M., Pannu, N. S., Read, R. J., Rice, L. M., Simonson, T. & Warren, G. L. (1998). Crystallography and NMR system: a new software suite for macromolecular structure determination. Acta Crystallog. sect. D, 54, 905-921. Collaborative Computational Project Number 4 (1994). The CCP4 suite: programs for protein crystallography. Acta Crystallog. sect. D, 50, 760-763. Cowtan, K. (1994). ``dm'': an automated procedure for phase improvement by density modi®cation. CCP4/ ESF-EACBM Newsletter on Protein Crystallog. 31, 3438. Dodou, E., Sparrow, D. B., Mohun, T. & Treisman, R. (1995). MEF2 proteins, including MEF2A, are
448 expressed in both muscle and non-muscle cells. Nucl. Acids Res. 23, 4267-4274. Ferrin, T. E., Huang, C. C., L.E., J. & Langridge, R. (1988). The MIDAS display system. J. Mol. Graph. 6, 13-27. Flinta, C., Persson, B., Jornvall, H. & von Heijne, G. (1986). Sequence determinants of cytosolic N-terminal protein processing. Eur. J. Biochem. 154, 193196. Gossett, L. A., Kelvin, D. J., Sternberg, E. A. & Olson, E. N. (1989). A new myocyte-speci®c enhancerbinding factor that recognizes a conserved element associated with multiple muscle-speci®c genes. Mol. Cell Biol. 9, 5022-5033. Han, T. H. & Prywes, R. (1995). Regulatory role of MEF2D in serum induction of the c-jun promoter. Mol. Cell Biol. 15, 2907-2915. Han, J., Jiang, Y., Li, Z., Kravchenko, V. V. & Ulevitch, R. J. (1997). Activation of the transcription factor MEF2C by the MAP kinase p38 in in¯ammation. Nature, 386, 296-299. Jones, T. A., Zou, J. Y., Cowan, S. W. & Kjeldgaard, M. (1991). Improved methods for building protein models in electron density maps and the location of errors in these models. Acta Crystallog. sect. A, 47, 110-119. Kabsch, W. & Sander, C. (1983). Dictionary of protein secondary structure: pattern recognition of hydrogen-bonded and geometrical features. Biopolymers, 22, 2577-2637. Kato, Y., Kravchenko, V. V., Tapping, R. I., Han, J., Ulevitch, R. J. & Lee, J. D. (1997). BMK1/ERK5 regulates serum-induced early gene expression through transcription factor MEF2C. EMBO J. 16, 7054-7066. Katoh, Y., Molkentin, J. D., Dave, V., Olson, E. N. & Periasamy, M. (1998). MEF2B is a component of a smooth muscle-speci®c complex that binds an A T-rich element important for smooth muscle myosin heavy chain gene expression. J. Biol. Chem. 273, 1511-1518. Kaushal, S., Schneider, J. W., Nadal-Ginard, B. & Mahdavi, V. (1994). Activation of the myogenic lineage by MEF2A, a factor that induces and cooperates with MyoD. Science, 266, 1236-1240. Krainc, D., Bai, G., Okamoto, S., Carles, M., Kusiak, J. W., Brent, R. N. & Lipton, S. A. (1998). Synergistic activation of the N-methyl-D-aspartate receptor subunit 1 promoter by myocyte enhancer factor 2C and Sp1. J. Biol. Chem. 273, 26218-26224. Laskowski, R. A., MacArthur, M. W., Moss, D. S. & Thornton, J. M. (1993). PROCHECK: a program to check the stereochemical quality of protein structures. J. Appl. Crystallog. 26, 283-291. Lavery, R. & Sklenar, H. (1988). The de®nition of generalised helicoidal parameters and of axis curvature for irregular nucleic acids. J. Biomol. Struct. Dynam. 6, 63-91. Lee, Y., Nadal-Ginard, B., Mahdavi, V. & Izumo, S. (1997). Myocyte-speci®c enhancer factor 2 and thyroid hormone receptor associate and synergistically activate the alpha-cardiac myosin heavy-chain gene. Mol. Cell. Biol. 17, 2745-2755. Lilly, B., Zhao, B., Ranganayakulu, G., Paterson, B. M., Schulz, R. A. & Olson, E. N. (1995). Requirement of MADS domain transcription factor D-MEF2 for muscle formation in Drosophila. Science, 267, 688693.
Structure of MEF2A Bound to DNA Lin, Q., Schwarz, J., Bucana, C. & Olson, E. N. (1997). Control of mouse cardiac morphogenesis and myogenesis by transcription factor MEF2C. Science, 276, 1404-1407. Lin, Q., Lu, J., Yanagisawa, H., Webb, R., Lyons, G. E., Richardson, J. A. & Olson, E. N. (1998). Requirement of the MADS-box transcription factor MEF2C for vascular development. Development, 125, 45654574. Liu, S., Liu, P., Borras, A., Chatila, T. & Speck, S. H. (1997). Cyclosporin A-sensitive induction of the Epstein-Barr virus lytic switch is mediated via a novel pathway involving a MEF2 family member. EMBO J. 16, 143-153. Mao, Z. & Nadal-Ginard, B. (1996). Functional and physical interactions between mammalian achaetescute homolog 1 and myocyte enhancer factor 2A. J. Biol. Chem. 271, 14371-14375. Mao, Z., Bonni, A., Xia, F., Nadal-Vicens, M. & Greenberg, M. E. (1999). Neuronal activity-dependent cell survival mediated by transcription factor MEF2. Science, 286, 785-790. Martin, J. F., Schwarz, J. J. & Olson, E. N. (1993). Myocyte enhancer factor (MEF) 2C: a tissue-restricted member of the MEF-2 family of transcription factors. Proc. Natl Acad. Sci. USA, 90, 5282-5286. Meierhans, D. & Allemann, R. K. (1998). The N-terminal methionine is a major determinant of the DNA binding speci®city of MEF-2C. J. Biol. Chem. 273, 26052-26060. Meierhans, D., Sieber, M. & Allemann, R. K. (1997). High af®nity binding of MEF-2C correlates with DNA bending. Nucl. Acids Res. 25, 4537-4544. Miska, E. A., Karlsson, C., Langley, E., Nielsen, S. J., Pines, J. & Kouzarides, T. (1999). HDAC4 deacetylase associates with and represses the MEF2 transcription factor. EMBO J. 18, 5099-5107. Molkentin, J. D., Black, B. L., Martin, J. F. & Olson, E. N. (1995). Cooperative activation of muscle gene expression by MEF2 and myogenic bHLH proteins. Cell, 83, 1125-1136. Molkentin, J. D., Black, B. L., Martin, J. F. & Olson, E. N. (1996a). Mutational analysis of the DNA binding, dimerization, and transcriptional activation domains of MEF2C. Mol. Cell. Biol. 16, 2627-2636. Molkentin, J. D., Li, L. & Olson, E. N. (1996b). Phosphorylation of the MADS-Box transcription factor MEF2C enhances its DNA binding activity. J. Biol. Chem. 271, 17199-171204. Murshudov, G. N., Lebedev, A., Vagin, A. A., Wilson, K. S. & Dodson, E. J. (1999). Ef®cient anisotropic re®nement of macromolecular structures using FFT. Acta Crystallog. sect. D, 55, 247-255. Navaza, J. (1994). AMoRe: an automated package for molecular replacement. Acta Crystallog. sect. A, 50, 157-163. Nurrish, S. J. & Treisman, R. (1995). DNA binding speci®city determinants in MADS-box transcription factors. Mol. Cell. Biol. 15, 4076-4085. Ornatsky, O. I. & McDermott, J. C. (1996). MEF2 protein expression, DNA binding speci®city and complex composition, and transcriptional activity in muscle and non-muscle cells. J. Biol. Chem. 271, 2492724933. Otwinowski, Z. (1991). Isomorphous replacement and anomalous scattering. In Proceedings of the CCP4 Study Weekend 25-26 January 1991 (Wolf, W. E. P. R. & Leslie, A. G. W., eds), pp. 80-88, SERC Daresbury Laboratory, Warrington, UK.
Structure of MEF2A Bound to DNA Otwinowski, Z. (1993). Oscillation data reduction program. In Proceedings of the CCP4 Study Weekend: Data Collection and Processing, 29-30 January (Sawyer, L. I. N. & Bailey, S., eds), pp. 55-62, SERC Daresbury Laboratory, Daresbury, UK. Pellegrini, L., Tan, S. & Richmond, T. J. (1995). Structure of serum response factor core bound to DNA. Nature, 376, 490-498. Perez-Martin, J. & de Lorenzo, V. (1997). Clues and consequences of DNA bending in transcription. Annu. Rev. Microbiol. 51, 593-628. Perrakis, A., Sixma, T. K., Wilson, K. S. & Lamzin, V. S. (1997). wARP: improvement and extension of crystallographic phases by weighted averaging of multiple-re®ned dummy atomic models. Acta Crystallog. sect. D, 53, 448-455. Pollock, R. & Treisman, R. (1991). Human SRF-related proteins: DNA-binding properties and potential regulatory targets. Genes Dev. 5, 2327-2341. Rao, S., Karray, S., Gackstetter, E. R. & Koshland, M. E. (1998). Myocyte enhancer factor-related B-MEF2 is developmentally expressed in B cells and regulates the immunoglobulin J chain promoter. J. Biol. Chem. 273, 26123-26129. Schulz, R. A., Chromey, C., Lu, M. F., Zhao, B. & Olson, E. N. (1996). Expression of the D-MEF2 transcription in the Drosophila brain suggests a role in neuronal cell differentiation. Oncogene, 12, 1827-1831. Sharrocks, A. D., von Hesler, F. & Shaw, P. E. (1993). The identi®cation of elements determining the different DNA binding speci®cities of the MADS box proteins p67SRF and RSRFC4. Nucl. Acids Res. 21, 215-221. Shore, P. & Sharrocks, A. D. (1995). The MADS-box family of transcription factors. Eur. J. Biochem. 229, 1-13. Sparrow, D. B., Miska, E. A., Langley, E., ReynaudDeonauth, S., Kotecha, S., Towers, N., Spohr, G., Kouzarides, T. & Mohun, T. J. (1999). MEF-2 func-
449 tion is modi®ed by a novel co-repressor, MITR. EMBO J. 18, 5085-5098. Tan, S. & Richmond, T. J. (1998). Crystal structure of the yeast MATalpha2/MCM1/DNA ternary complex. Nature, 391, 660-666. West, A. G. & Sharrocks, A. D. (1999). MADS-box transcription factors adopt alternative mechanisms for bending DNA. J. Mol. Biol. 286, 1311-1323. West, A. G., Shore, P. & Sharrocks, A. D. (1997). DNA binding by MADS-box transcription factors: a molecular mechanism for differential DNA bending. Mol. Cell. Biol. 17, 2876-2887. Wilson-Rawls, J., Molkentin, J. D., Black, B. L. & Olson, E. N. (1999). Activated notch inhibits myogenic activity of the MADS-Box transcription factor myocyte enhancer factor 2C. Mol. Cell. Biol. 19, 28532862. Yang, S. H., Galanis, A. & Sharrocks, A. D. (1999). Targeting of p38 mitogen-activated protein kinases to MEF2 transcription factors. Mol. Cell. Biol. 19, 40284038. Youn, H. D., Sun, L., Prywes, R. & Liu, J. O. (1999). Apoptosis of T cells mediated by Ca2-induced release of the transcription factor MEF2. Science, 286, 790-793. Yu, Y. T., Breitbart, R. E., Smoot, L. B., Lee, Y., Mahdavi, V. & Nadal-Ginard, B. (1992). Human myocyte-speci®c enhancer factor 2 comprises a group of tissue- restricted MADS box transcription factors. Genes Dev. 6, 1783-1798. Zetser, A., Gredinger, E. & Bengal, E. (1999). p38 mitogen-activated protein kinase pathway promotes skeletal muscle differentiation. Participation of the Mef2c transcription factor. J. Biol. Chem. 274, 51935200. Zhao, M., New, L., Kravchenko, V. V., Kato, Y., Gram, H., di Padova, F., Olson, E. N., Ulevitch, R. J. & Han, J. (1999). Regulation of the MEF2 family of transcription factors by p38. Mol. Cell. Biol. 19, 21-30.
Edited by J. Karn (Received 5 January 2000; received in revised form 31 January 2000; accepted 31 January 2000)
J. Mol. Biol. (2000) 297, 437±449
Crystal Structure of MEF2A Core Bound to DNA at Ê Resolution 1.5 A Eugenio Santelli and Timothy J. Richmond* ETH Zurich, Institut fuÈr Molekularbiologie und Biophysik, ETH-HoÈnggerberg CH 8093, ZuÈrich, Switzerland
Members of the myocyte enhancer factor-2 (MEF2) family of transcription factors bind to and activate transcription through A T-rich DNA sequences found primarily, but not exclusively, in the promoters of muscle-speci®c genes. Their importance has been established for myogenic development and in activation of the immediate-early gene, c-jun, and recently further functional roles in the immune system have emerged. The MEF2 factors belong to the MADS-box superfamily, sharing homology in a 58 amino acid domain that mediates DNA binding and dimerization. The structures of two MADS-box proteins, SRF and MCM1, bound to their cognate DNA have been previously reported and shown to share extensive similarity in their mode of DNA binding. We have solved the structure of MEF2A 2-78 bound to its DNA consensus Ê resolution. It reveals how the absence of amino acids sequence at 1.5 A N-terminal to the MADS-box contributes to the DNA binding properties of MEF2 proteins and shows that the MEF domain C-terminal to the MADS-box adopts a conformation considerably different from the same region in SRF and MCM1. # 2000 Academic Press
*Corresponding author
Keywords: transcription factor; muscle development; protein-DNA recognition; MADS-box; X-ray crystallography
Introduction The myocyte enhancer factor-2 (MEF2) family of transcription factors have an essential role in the myogenesis of cardiac and skeletal muscle cells (Black & Olson, 1998; Kaushal et al., 1994; Lilly et al., 1995; Lin et al., 1997). In vertebrates, it comprises the members MEF2A, B, C and D, each having several splicing variants and together showing more than 85 % identity in an 86-amino acid core domain responsible for DNA binding and dimerization (Breitbart et al., 1993; Gossett et al., 1989; Martin et al., 1993; Yu et al., 1992) (Figure 1(a)). MEF2 proteins bind as homo- and heterodimers to a common DNA consensus sequence, YTA(A/T)4 TAR, found mainly in the control regions of many muscle and nerve speci®c genes but also of other unrelated genes (Andres et al., 1995; Pollock & Treisman, 1991) (Figure 1(b) and (c)). The core domain also forms complexes with proteins that cooperate with or repress MEF2 stimulated tranAbbreviations used: MEF2, myocyte enhancer factor 2; TR, thyroid hormone receptor; SRF, serum response factor. E-mail address of the corresponding author: [email protected] 0022-2836/00/020437±13 $35.00/0
scription activation, most notably myogenic and neurogenic bHLH proteins (e.g. MyoD, myogenin, MASH-1), thyroid hormone receptor (TR), MAP kinase (BMK1/ERK5), and histone deacetylase proteins and associated factors (HDAC4 and MITR) (Black et al., 1996; Kato et al., 1997; Lee et al., 1997; Miska et al., 1999; Molkentin et al., 1995; Sparrow et al., 1999). MEF2 factors display a high level of complexity in their expression pattern and activation potential both temporally and spatially in body development. The best-established role of MEF2 is the regulation of genes essential for differentiation and maintenance of the skeletal and cardiac muscle phenotype, but other important functions crucially dependent on MEF2 factors have come to light since their discovery. Among these are the regulation of smooth muscle (Katoh et al., 1998), vascular (Lin et al., 1998) and nervous system development (Mao et al., 1999; Schulz et al., 1996), multiple roles in the immune system (Han et al., 1997; Rao et al., 1998; Youn et al., 1999), seruminduction of c-jun (Han & Prywes, 1995) and participation in the Epstein-Barr virus lytic switch and mediation of the response to calcium ion signaling (Liu et al., 1997). The ability of MEF2 to activate genes in response to mitogens via the p38 or the # 2000 Academic Press
438
Structure of MEF2A Bound to DNA
Figure 1. (a) Sequence alignment of the DNA-binding/dimerization domains of myocyte-speci®c enhancer factor 2 proteins (four human members MEF2A-D; Drosophila D-MEF2), human serum response factor (SRF) and yeast MCM1. Invariant (*) and highly conserved () amino acid residues for all known MADS-box proteins and secondary structure elements for MEF2A are indicated (Kabsch & Sander, 1983). The MEF2A protein used in this study comprises residues 2-78 (black). The MEF domain homology is not found in SRF or MCM1. (b) DNA site consensus sequences for MEF2A (MEF site) and SRF (CArG-box) determined by in vitro selection (Shore & Sharrocks, 1995). (c) DNA in the crystal structure. The near palindromic duplex with single overhanging A and T bases is required to obtain high quality crystals.
BMK1/ERK5 pathways has also been demonstrated (Han et al., 1997; Kato et al., 1997; Yang et al., 1999; Zetser et al., 1999; Zhao et al., 1999). Therefore, MEF2 must activate, or in certain cases repress, different sets of genes depending on cellular context. MEF2 may be able to modulate its own transactivation properties through various heterodimeric isoforms that it can form (Ornatsky & McDermott, 1996). However, such functional diversity is most likely achieved through interaction with other proteins (Black et al., 1996; Kaushal et al., 1994; Krainc et al., 1998; Lee et al., 1997; Mao & Nadal-Ginard, 1996; Miska et al., 1999; Molkentin et al., 1995; Sparrow et al., 1999; Wilson-Rawls et al., 1999). The MEF2 proteins belong to the MADS-box superfamily of transcription factors characterized by strong sequence homology in a 58-amino acid DNA-binding domain that is N-terminal in MEF2 and in most other family members (Shore & Sharrocks, 1995) (Figure 1(a)). Following immediately C-terminal to the MADS-box is the highly conserved, 28-residue MEF domain which is not found in any other known MADS-box protein and is responsible for the ability of MEF2 factors to form DNA-binding heterodimers with themselves but not with other MADS factors (Molkentin et al., 1996a). MEF2A is 507 amino acid residues in length with at least one activation domain following the core region in the sequence (Black & Olson, 1998). In the case of MEF2C, two activation domains have been identi®ed, but at least while part of the intact protein, they require the MEF domain for full func-
tion (Molkentin et al., 1996a). The MEF domain itself is necessary and suf®cient in the core protein alone to impart transcriptional silencing via interaction with proteins involved with histone deacetylation (Miska et al., 1999; Sparrow et al., 1999). All the splicing isoforms of MEF2 proteins display full DNA binding activity and transactivation potential, and are differentially expressed in vivo (Breitbart et al., 1993; Martin et al., 1993). Different MADS-box proteins bind to DNA sites of related but distinct consensus sequences (Figure 1(b)). The basis of these differences has been the focus of extensive mutational analyses, af®nity/speci®city measurements and proteininduced, DNA distortion experiments (Dodou et al., 1995; Meierhans & Allemann, 1998; Meierhans et al., 1997; Nurrish & Treisman, 1995; Pollock & Treisman, 1991; Sharrocks et al., 1993; West & Sharrocks, 1999; West et al., 1997). The crystal structures of two MADS proteins bound to their cognate DNA, human serum response factor (SRF) and the yeast factor MCM1 in combination with MATa2, have been previously solved by X-ray methods and show that base-speci®c contacts only partially explain the observed sequence speci®city (Pellegrini et al., 1995; Tan & Richmond, 1998). Both SRF and MCM1 are bound to a highly bent DNA duplex, and DNA distortion here as for several other transcription factors plays a major role in sequence speci®c recognition and generation of complex architecture important for promoter function (Perez-Martin & de Lorenzo, 1997). We preÊ X-ray structure of MEF2A core sent here the 1.5 A
439
Structure of MEF2A Bound to DNA
which includes the MADS box and part of the MEF domain bound to its consensus sequence DNA (Figure 1(c)). The crystal structure of MEF2A shows in comparison to SRF and MCM1 similarities and differences in the MADS-box and prominent divergence in the C-terminal domain. The atomic structures of MEF2A, SRF and MCM1 bound to their speci®c DNA sites reveal the variations on DNA binding that occur within this protein family and provide answers as to how they arise.
Results and Discussion MEF2A-DNA complex The MEF2A core protein (amino acid residues 278) is folded in the same three-layer organization observed for SRF and MCM1 (Figure 2). Each monomer of MEF2A contains a long a-helix (aI) and a 12-amino acid N-terminal random coil (Nextension) comprising the DNA contacting layer, and two b-strands (bI, bII) connected by a b-turn forming the middle layer (b-hairpin). Together these two layers de®ne the MADS-box. A second a-helix (aII) makes up the top layer or the major part of the MEF domain and packs against the b-hairpin in an orientation diagonal to the analogous element in SRF (Figure 3(a)). The r.m.s.d. between corresponding Ca atoms of SRF and Ê for MADS box regions MEF2A is 0.50 and 1.77 A 5-38 and 39-58, respectively, calculated after superposing amino acids 145-178 of SRF and 5-38 of MEF2A. The primary differences within the MADS domains occur at amino acids 2-4 of the N-extension bound to DNA and 49-54 of the b-turn. The dimerization surface consists of one face of the monomer analogous with SRF, where aII in MEF2A additionally contacts the b-sheet of the opposite monomer via hydrophobic and hydrogenbonding interactions. Electron density is not visible for the C-terminal amino acid residues 74-78 for either copy of the complex apparently because of disorder. The X-ray structure shows that the DNA is bent overall to only 17 as compared to 72 for SRF (Figure 4). The b-turn in the middle layer of the protein structure is not used for DNA binding in MEF2A whereas it is an important feature of the DNA interaction seen in the SRF-DNA complex. The MEF2A core dimer/DNA assembly crystallized in space group P1 with two copies of the complex per asymmetric unit. The r.m.s.d. after coordinate alignment between all atoms of the two Ê for independent dimers are 0.06, 0.16 and 0.33 A protein Ca atoms only, all protein atoms and all DNA atoms, respectively. The corresponding values comparing subunits in one of the dimers Ê . The solvent accessible are 0.20, 0.58 and 0.74 A Ê probe) buried in the dimerizasurface area (1.4 A Ê 2. All amino acid residues tion interface is 3864 A that could have a major role in the dimerization speci®city, those in aI, bI and aII, are conserved in MEF2A-D so that hetero dimerization can occur
unobstructed. Conservation of amino acid residues is so high throughout the core sequence, it is unlikely that surface side-chains play a selective role in hetereo dimerization either. Although the aI and bI structures of MEF2 and SRF have only minor sequence differences, hetero dimerization of a MEF2 subunit with the SRF subunit would be blocked by incompatibility between the MEF domain and the SRF ``top layer'' structure.
MEF2 domain The MEF2 domain starts as a short coil region (sequence STD) from the C terminus of the b-hairpin, bends back over the b-sheet and continues as the aII helix packed against the underlying b-sheet (Figure 2, Figure 3(a)). The aII helix axis runs perpendicular to the direction of the b-sheet so that it contributes signi®cantly to the dimerization interface of the whole core through predominantly hydrophobic contacts with the twofold related aII helix and b-hairpin of the adjacent subunit. This scheme contrasts with SRF for which the aII helices are mainly associated with the b-strands within the same subunit. The aII helices of MEF2A are separÊ axis to axis as compared to, ated by 13.3(0.3) A for example, the helices in the aI coiled coil spaced Ê . The pair of antiparallel aII helices by 10.1(0.3) A form a substantial cleft in the top layer that likely provides the interaction site in the core required for transactivation, and the binding of the MITR protein and homologous regions in HDAC2-4. Mutation of the VLL sequence in the center of aII to ASR removes the activation potential of MEF2C. Although the V64A alteration in this mutant may allow the aII to fold similarly in the core, since the valine is buried, the L65S and L66R replacements would substantially change the character of the exposed surface of the proposed binding pocket. Most exposed hydrophilic side-chains, those of Asp61, Asp63, Lys64 and Leu67, in the MEF domain are not clearly de®ned by the electron density in any of the four copies. The exception is Lys68 which extends outward from the side of aII and hydrogen bonds in a mixed population with Glu42 at the beginning of bI in the same subunit and Asn52 in the b-turn of the adjacent subunit. Phosphorylation of Ser59 has been shown to increase both the transactivation potential and the DNA binding af®nity of MEF2C, although the biological signi®cance of this modi®cation is not clear (Molkentin et al., 1996b). Ser59 is the ®rst amino acid of the MEF domain and its hydroxyl group is hydrogen bonded to the main-chain amide of Ile43. Phosphorylation at this site could perturb the structure of aII, particularly the relationship of the Lys68 to Glu42 interaction described. It may affect the ability of the MEF2 domain to combine with further factors. Examination of the structure does not suggest any simple explanation for the modest increase in DNA binding observed for phosphorylated species (Molkentin et al., 1996b).
440
Structure of MEF2A Bound to DNA
Figure 2. The MEF2A structure in two orthogonal views. The 2-fold axes of the protein dimer and DNA (silver) are aligned in the crystals. The three-layer organization of a protein monomer extends from the DNA as an N-extension (brown, N-ext) and an a-helix (red, aI) of the antiparallel coiled coil in the ``bottom layer'', followed by two strands (blue, bI-bII) of the four-strand b-sheet making up the ``middle layer''. The top and bottom layers comprise the MADS-box. The a-helix (yellow, aII) of the MEF domain forms the ``top layer'' with its counterpart in the 2-fold related subunit (green).
MEF2A-DNA interactions The MEF2A core protein bends the DNA double helix containing the MEF site consensus sequence by 17 , in good agreement with the value of 19 measured previously in solution (West et al., 1997) (Figure 2). The DNA is essentially straight in comparison to the 72 bend seen in the SRF/DNA
structure (Pellegrini et al., 1995) (Figure 4). However, the twist of the double helix within the binding site is over wound to 9.9 bp/turn while the minor groove at the four central A T base-pairs Ê , about half that for the canoninarrows to 2.7-3.1 A cal B-form conformation (Table 1). The two most important protein units for DNA binding are the aI helix and the N-extension of the
Structure of MEF2A Bound to DNA
441
Figure 3. Main-chain paths for MADS-box proteins (stereographic). (a) Monomers of MEF2A (brown) and SRF (green) are shown in superposition. The b-hairpin of MEF2A is noticeably twisted clockwise compared to SRF, and the b-turn is displaced toward the dimer interface (left). Amino acid residues 5-39 in the N-extension and aI helix were used for the alignment. (b) Comparison of the DNA-binding segments of the N-extensions of MCM1 (17-21), SRF (142-146) and MEF2A (2-6). The structures were aligned by superposing MEF2A amino acid residues 5-39 with the corresponding residues in MCM1 and SRF. The arginine and glycine amino acid residues labeled penetrate into the minor groove making key contacts with the DNA.
MADS-box (Figure 5(a)). As for SRF, the central two turns of the aI helix (amino acid residues 2331) provide the primary DNA-binding contacts over the consensus-binding site (Figure 6). Lys30 hydrogen bonds to a phosphate oxygen of base A 2 from the major groove side, and Lys31 similarly bonds to the adjacent T-3 in the complementary strand from the minor groove side. Arg24 hydrogen bonds to the phosphate group of A 4 from the minor groove side while Lys23 reaches into the major groove and forms hydrogen bonds to the N7 atom of A 4 and simultaneously with the O6 atom of G 5, specifying a purine at the former position and a guanine at the latter. Lys23 makes the only direct contact with bases in the major groove. Arg24 and Lys30 from opposite subunits are linked together through hydrogen bonds with Glu34 forming a seemingly rigid constellation of side-chains important for specifying the local DNA conformation (Table 1). At the extremities of the DNA site, Arg15 bridges over the major groove in two different conformers to make either a
hydrogen bond or salt link to the phosphate group of A-8. The guanidinium group of Arg15 interacts with the phosphate and N7 atom of A-8, in accord with previous results indicating that purine is preferred at this position (Dodou et al., 1995; Pollock & Treisman, 1991). All of these amino acid residues, and including Gly27 which allows the close approach of aI to the DNA backbone, are invariant in all MADS proteins with the exception of Arg15. The side-chains of Ile8, Val19, Thr20, Phe26, Tyr33 and Leu38 appear to further restrict the conformational ¯exibility of the long side-chains involved in these interactions. Tightly bound water molecules mediate further contacts between the DNA phosphate groups and the side-chains of Glu34 and Thr20 and the main-chain carbonyl groups of Lys23, Arg24 and Gly27. The minor groove formed by the eight central base-pairs of the MEF site is in close contact with Gly2 and Arg3. In each half site, the arginine side-chain lies in the minor groove with the opposite orientation as compared to the homologous
442
Structure of MEF2A Bound to DNA
Figure 4. The MEF2A and SRF core/DNA complexes. The aI helices (red) were used for the alignment of the complexes. The packing of the b-sheet (green) on the coiled coil is similar, although not identical, in the two MADS domains, whereas the interaction of the N-extension (yellow) with DNA (blue-cyan) and the orientation of the SRF C-terminal and MEF domains (magenta) on the b-sheet are substantially different for the two complexes. The DNA is bent over an angle four times greater for SRF than MEF2A (72 versus 17 ).
arginine for SRF (Figure 5(a)). Nevertheless, two of three side-chain methylene groups contact the C2 atom of A 4 and the guanidinium group hydrogen bonds to the O2 and O40 atoms of T-4 so that an A T base-pair is clearly speci®ed at this position (Figure 5(b)). The Lys23 further speci®es the orientation of this base-pair via the major groove. The N-terminal amino group of Gly2 is pointed toward the center of the DNA binding site and hydrogen bonds to the N3 atom of A-2, and its amide atom hydrogen bonds to the N3 atom of A 4. The main-chain conformation of the N-extension between the aI helix at Asp13 and Lys4 is virtually identical with that for SRF (Figure 3(a)). Lys4 and Ile6 interact with the phosphodiester backbone to ``push'' the Arg3 side-chain into the DNA groove and the intervening Lys5 makes a salt link with the phosphate group of A 1. Comparison of sequences for different MADSbox factors reveals that the amino acid residues to either side of the highly conserved arginine residue at MADS-box position three are variable (Shore & Sharrocks, 1995). The conformation of the polypeptide chain in this segment of the N-extension is different in the structures of the SRF, MCM1 and MEF2A (Figure 3(b)). DNA complexes due to the size, charge or hydrophobicity of the amino acid
residues immediately ¯anking position 3. For MEF2C and SRF, direct contact by the entire arginine side-chain with the surface of the minor groove affects the sequence of the central A T base-pairs beyond the requirement for a narrow minor groove dictated by the aI coiled coil (Pellegrini et al., 1995). For MCM1, the arginine at position 3 does not invade the minor groove apparently because it is preceded by glutamic acid (Tan & Richmond, 1998). Instead, only the guanidinium group of a second arginine at position 4 is inserted into the minor groove and this arrangement permits greater variability in the binding site sequence. DNA bending by MEF2A and SRF MEF2A and SRF differ greatly in their apparent ability to bend DNA into the major groove. SRF bends DNA around the N-terminal third of the aI helix; MEF2A does not. In the SRF/DNA complex, Lys154 and Lys165 facilitate DNA bending through their interaction with DNA phosphate groups, allowing ¯anking DNA outside the CArGbox to bind to Thr191 and His193 in the b-turn of the middle layer (Pellegrini et al., 1995). The path of the b-turn in MEF2A (amino acid residues 49-53,
443
Structure of MEF2A Bound to DNA Table 1. DNA parameters Minor groove Base-pair A ÿ 8/T 8 A ÿ 7/T 7 G ÿ 6/C 6 C ÿ 5/G 5 T ÿ 4/A 4 A ÿ 3/T 3 TA ÿ 2/AT 2 T ÿ 1/A 1 A 1/T ÿ 1 TA 2/AT ÿ 2 T 3/A ÿ 3 A 4/T ÿ 4 G 5/C ÿ 5 C 6/G ÿ 6 T 7/A ÿ 7 T 8/A ÿ 8
Propeller twist (deg.) ÿ8.9 ÿ15.7 ÿ8.2 ÿ13.0 ÿ11.7 ÿ17.4 ÿ14.6 ÿ17.9 ÿ17.0 ÿ14.2 ÿ17.0 ÿ11.9 ÿ13.2 ÿ8.3 ÿ14.9 ÿ7.1
Roll (deg.)
Tilt (deg.)
Twist (deg.)
6.8 8.6 ÿ4.8 9.3 ÿ7.5 7.3 ÿ1.6 7.9 ÿ2.3 7.5 ÿ7.5 8.8 ÿ4.8 8.1 6.3
ÿ7.5 ÿ0.4 ÿ0.4 5.6 ÿ3.2 ÿ3.7 ÿ0.3 ÿ0.3 0.9 4.4 2.9 ÿ5.1 0.6 0.7 6.8
34.5 31.9 39.2 31.3 49.0 28.8 38.9 37.4 39.4 28.8 48.1 31.6 39.4 32.0 35.0
Major groove
Ê) Width (A
Ê) Depth (A
Ê) Width (A
Ê) Depth (A
4.4 5.8 5.5 3.1 2.7 2.7 3.3 5.5 5.8 4.3
5.5 5.9 6.1 5.4 5.3 5.3 5.4 6.1 5.8 5.6
12.2 11.0 12.1 12.3 13.0 12.9 12.3 12.0 11.0 12.2
3.4 2.9 3.2 3.6 3.9 4.1 3.6 4.2 2.7 3.5
Parameters are averages over two duplexes and two conformers.
sequence NSSNKL) deviates appreciably from that seen in SRF by being displaced toward the dimer interface (Figure 3(a)). Nevertheless, superposition of the SRF and MEF2A structures shows that Gln49, Ser51 and Lys53 in the MEF2A b-turn could in principle provide binding interactions for bent DNA (Figure 7). Lys154 found in SRF is replaced by Glu14 in MEF2A (a glutamine residue in MEF2B). This exchange of oppositely charged sidechains contributes strongly to the lack of signi®cant DNA bending for MEF2A as indicated by mutagenesis experiments (Nurrish & Treisman, 1995; West et al., 1997). This is perhaps surprising since the glutamic acid side-chain could in principle bind to a DNA phosphate group via a water molecule in conjunction with a neighboring arginine side-chain as seen in the nucleosome core structure (Karolin Luger & T.J.R., unpublished results). The structural context of Glu14 within the framework of the MADS-box must therefore be important; the reciprocal exchange experiment of introducing K154E into SRF is suf®cient to signi®cantly reduce the DNA bend (Nurrish & Treisman, 1995; Sharrocks et al., 1993). By comparison, MEF2C which also contains Glu14 does shows a degree of bending intermediate between SRF and MEF2A and suggests that analysis conditions may affect the values obtained (Meierhans et al., 1997). Furthermore, the bend angle measured may not represent a static bend, but the average for at least two states of DNA bending. Conversion of SRF to MEF site specificity SRF N-terminally truncated to Gly142, corresponding to Gly2 in MEF2, binds with high af®nity to MEF2 sites while still displaying residual binding to the CArG-box sequence (Nurrish &
Treisman, 1995; Sharrocks et al., 1993). This truncated version of SRF would allow the N terminus to lie under the aI helices and Arg143 to now specify the base-pair at 4/ÿ 4 as seen from the MEF2A complex (Figure 5). With the native sequence, the SRF N terminus must lie to the outside of the complex, requiring the arginine sidechain to be tucked between the protein and DNA. Truncated SRF bends DNA only slightly less well than the wild-type protein. Conversion to MEF2 site speci®city is completed with the additional mutation K154E (Nurrish & Treisman, 1995; Sharrocks et al., 1993), which would prevent the DNA from contacting the b-loop.
Conclusions MEF2A recognizes its speci®c DNA binding site by selecting a narrow minor groove and by making speci®c contacts with two base-pairs per halfsite. The MADS-box motif has been adapted in different proteins to bind equally well to DNA in a highly bent form as for SRF and MCM1 or to relatively unbent DNA as for MEF2A. The core region immediately C-terminal to the MADS-box, the MEF domain in MEF2 proteins, has evolved to interact with several factors that relay or deliver a transcription activating or silencing signal. The entire core region of MEF2A, as well as SRF and MCM1, is important for interactions with other DNA-binding transactivator proteins that account for promoter site speci®city. Further structural studies pertinent to these multifactor complexes are essential to understand the atomic mechanisms of transcription activation and silencing.
444
Structure of MEF2A Bound to DNA
Figure 5 (legend opposite)
Methods Production and crystallization of the MEF2A/ DNA complex The gene coding for MEF2A 1-78 was created by PCR ampli®cation from synthetic oligonucleotides and cloned into the pET3a expression vector. The bacterially expressed polypeptide (Escherichia coli strain BL21(DE3), 25 C) was puri®ed from the cytosolic soluble fraction by successive chromatographic steps on heparine-Sepharose (Pharmacia), SP-5PW and Phenyl-5PW (TSK) columns run at 4 C and concentrated to approximately 1 mg mlÿ1 in 300 mM NaCl, 5 mM Hepes-Na (pH 7.0), 0.5 mM EDTA and 1 mM DTT to give a ®nal yield of 0.3 mg lÿ1. The polypeptide was also puri®ed from inclusion bodies: the cells grown at 37 C were lysed by freeze-thawing, homogenized, sonicated and centrifuged. The pellet was dissolved in 6 M guanidinium chloride and centrifuged again. The supernatant was ®ltered through a XM50 membrane (Amicon) and the ®ltrate eluted on a Superdex S-200 column with a buffer containing 8 M urea, 100 mM NaCl and 10 mM b-mercaptoethanol. The relevant fractions were precipitated by extensive dialysis against 10 mM b-mercaptoethanol and lyophlized. The polypeptide was folded in two successive dialysis steps: (1) 6 M urea, 0.5 M NaCl, 10 mM Hepes-Na buffer (pH 7.0), 0.5 mM EDTA, 5 mM b-mercaptoethanol to a buffer containing 0.5 M arginine, 0.75 M guanidinium chloride, 0.05 % Triton -100, 100 mM NaCl, 25 mM bis-Tris base (pH 6.3), (2) 10 mM Hepes-Na buffer (pH 7.0), 0.5 M NaCl, 10 mM Hepes-Na (pH 7.0), 0.5 mM EDTA and 1 mM DTT. A ®nal chromatographic step on a Phenyl5PW (TSK) column followed. N-terminal sequencing indi-
cated that the polypeptide began at Gly2, and MALDITOF mass spectrometry analysis showed that the mass was that expected from the expression construct lacking Met1. It is not known if MEF2A Met1 is present in eukaryotic cells. However, since glycine is the second amino acid in the MEF2A sequence, the expectation is that Met1 is removed in vivo (Flinta et al., 1986). All oligonucleotides were synthesized with a Applied Biosystems 380B synthesizer using standard phosphoramidite chemistry and puri®ed by reverse phase HPLC. After addition of 10-30 % excess MEF2 site containing DNA, the complex was puri®ed by gel electrophoresis (BioRad) at 4 C on a 9 % (w/v) acrylamide (acrylamide to bis-acrylamide, 40:1, w/w) gel containing 35.6 mM Tris base/boric acid and 1.0 mM EDTA and eluted into 5 mM Hepes-Na buffer (pH 7.0), 10 mM NaCl, 1 mM DTT. Several DNA duplexes of different length and sequence were tried, and crystals suitable for X-ray diffraction analysis were obtained with a 16-bp oligonucleotide having A-T overhanging bases (Figure 1(c)). Cystallization was by hanging drop vapor diffusion conditions using 15-18 % PEG6000, 50 mM NaNO3, 5 mM Ba(NO3)2, 50 mM bis-Tris buffer (pH 5.8-6.3), 25 mM NaOAc, 2 mM DTT, 22 C in the well. Crystallographic structure solution The needle-shaped crystals (20-60 mm 100-250 mm 400-1200 mm) used for data collection diffracted to Ê using a Rigaku RU200 X-ray generator equipped 1.8 A Ê at with a MAR180 image plate detector and to 1.3 A beam line ID14, ESRF Grenoble using a MAR345. Crystals were ¯ash-cooled in propane at ÿ120 C using 20 %
Structure of MEF2A Bound to DNA
445
Figure 5. The MADS-box aI coiled coil and N-extension bound to DNA. (a) The two aI helices in the protein dimer Ê (green and orange) bind primarily to the narrow minor groove of the MEF site DNA (accessible surface to a 1.4 A probe in red-white-blue, electrostatic negative-neutral-positive). The charged side-chains of Lys23, Arg24, Lys30, Lys31, Glu34, and Arg15 in one DNA half site are shown (CPK colors). Lys30 is shown in two alternative conformations; both bind a DNA phosphate oxygen, but only one binds Glu34. Lys23 is the only residue to make a speci®c contact in the major groove. The N-terminal Gly2 and side-chain of Arg3 (CPK colors) lie buried in the minor groove with Arg3 bound to base-pair A 4/T ÿ 4. (b) Electron density from a simulated annealing omit map (10,000 degrees, amino acid residues 2-17 omitted, contoured at 0.6s) shows the proximity of two Arg3 methylene groups to the C2 atom of base A 4.
(w/v) glycerol as a cryoprotectant. A complete dataset Ê was collected also at beam line ID14, ESRF Greto 1.5 A noble (Table 2). Data reduction was performed using programs DENZO and SCALEPACK (Otwinowski, 1993). The MEF2A structure was solved by a combination of molecular replacement using the SRF structure (pdb accession code 1SRS) and SIR phasing using data collected in the laboratory from a derivative with 5-iodouracil replacing T-4. A poly-alanine model comprising SRF residues 142-196 and base-pairs ÿ5 to 5 was used for the initial search using AMoRe (Navaza, 1994). The heavy-atom positions for the iodo-derivative were predicted from the molecular replacement solution and precisely determined by difference Patterson and Fo ÿ Fc maps. Phases were calculated with MLPHARE Ê (Otwinowski, 1991) and electron density maps to 2.2 A were calculated using combined phases with CNS
(BruÈnger et al., 1998). Density modi®cation by solvent ¯attening and NCS averaging was performed with DM (Collaborative Computational Project, 1994; Cowtan, 1994) and the atomic model was built using the graphics program O (Jones et al., 1991). Re®nement was carried out using data from 20.0 to Ê resolution (Table 3). NCS restraints were applied 1.5 A between the two copies of the complex in the crystal asymmetric unit during the initial simulated annealing re®nement using CNS. The model was improved through cycling energy minimization without NCS using alternately CNS and REFMAC (Murshudov et al., 1999) followed by manual rebuilding with O. Water molecules were added using CNS and WARP (Perrakis et al., 1997). The high resolution of the diffraction data permitted observation of two separate conformers for the DNA in the electron density map. They exist most probably
446
Structure of MEF2A Bound to DNA Table 2. Crystallographic data Dataset Ê) Resolution (A Completeness Overall Last shell (range) Unique reflections Average redundancy Rmerge (%) Overall Ê) Last shell (range A Phasing power Cullis R Unit cell a b c a b g
Native
Iodo derivative
1.5
2.15
96.0 92.0 (1.55-1.50) 87,264 2.6
97.4 89.4 (2.23-2.15) 29,659 4.2
7.5 10.4 (1.55-1.50)
7.9 23.7 (2.23-2.15) 0.70 0.84
41.37 60.70 63.99 115.18 89.99 90.00
41.36 60.73 64.18 115.33 90.05 89.95
obtained by including the DNA model twice, each at half occupancy, and restraining the base-pairs 4/ÿ 4, 3/ÿ 3, 1/ÿ 1, ÿ1/ 1, ÿ3/ 3 and ÿ4/ 4, and bases 5 and ÿ5 between the two atomic models. The maximum rms deviation values for all atoms occuring for unrestrained (8/ ÿ 8) and restrained (ÿ4/ 4) Ê and 0.64 A Ê (0.13 A Ê without phosbase-pairs are 1.73 A phate), respectively. The conformer with the greatest number of hydrogen bonds between protein and DNA was used for display. Several amino acid side-chains were also included in two conformations.
Table 3. Structure re®nement
Figure 6. Schematic of MEF2A/DNA interaction. The direct hydrogen bonds (arrows) and hydrophobic contacts (continuous lines) from a single protein monomer (amino acid labels) to DNA base-pairs (MEF2 site shaded) are shown. Highly localized water molecules mediate some interactions (dotted lines).
because an approximate 2-fold rotational degeneracy superimposes both complexes in the crystal asymmetric unit imperfectly around the 2-fold axis of the complex. The iodo-derivative used for phasing con®rms this degeneracy. Completely symmetric DNA sequences and overhangs did not yield well diffracting crystals. Of the several options tested, the best re®nement results were
Ê) Resolution (A R-factor 20-1.5/1.57-1.50 (%) Free-R-factor 20-1.5/1.57-1.50 (%) Reflections work/free Atoms (incl. multiple conformers) Total Protein DNA Solvent r.m.s.d. from ideality Ê) Bond lengths (A Bond angles (deg.) Dihedral angles (deg.) r.m.s.d. B-factors for bonded atoms Main-chain/target s Side-chains/target s Ramachandran plot Allowed (%) Additionally allowed (%) Average G-factor (PROCHECK)a Ê 2) Average B-factors (A Main-chain Side-chain Solvent DNA backbone DNA bases
20.0-1.5 20.6/25.7 22.9/28.6 77,972/8679 5940 2520 2764 656 0.010 1.44 19.6 1.2/2.0 1.7/3.0 93.8 6.2 0.33 21.1 23.4 38.9 26.1 21.6
No sigma cutoff was applied. The Free R set was created in spacegroup P21, expanded to P1 and combined with the diffraction data, due to the high correlation between re¯ections related by the pseudo-symmetry axis. a All mean values are above 0.
Structure of MEF2A Bound to DNA
447
Figure 7. Superposition of MEF2A aI helix and b-turn and SRF DNA structures. The Ca atoms of the aI helices of both the MEF2A/DNA and SRF/DNA (Pellegrini et al., 1995) complexes were aligned. In this hypothetical complex, the SRF DNA is in near proximity to the b-turn of MEF2A such that interaction might occur similar to SRF. Glu14 apparently inhibits this interaction and prevents signi®cant DNA bending as revealed by its replacement with a lysine residue in a MEF2A mutant protein (West et al., 1997).
Programs CNS, PROCHECK (Laskowski et al., 1993) and CURVES (Lavery & Sklenar, 1988) were used for model analysis. Molecular images were generated with programs MIDASPlus (Ferrin et al., 1988), O (Jones et al., 1991) and WebLab ViewerPro 3.5 (Molecular Simulations Inc). Protein Data Bank accession code The atomic coordinates and structure factors of the MEF2A core/DNA structure were deposited in the Protein Data Bank, accession code 1egw.
Acknowledgments We thank Dr David Sargent for expert assistance in data collection and the staff at ESRF Grenoble, beamline ID14, for excellent technical support. This project was supported in part by the TMR Program of the European Union.
References Andres, V., Cervera, M. & Mahdavi, V. (1995). Determination of the consensus binding site for MEF2 expressed in muscle and brain reveals tissue-speci®c sequence constraints. J. Biol. Chem. 270, 23246-23349.
Black, B. L. & Olson, E. N. (1998). Transcriptional control of muscle development by myocyte enhancer factor-2 (MEF2) proteins. Annu. Rev. Cell Dev. Biol. 14, 167-196. Black, B. L., Ligon, K. L., Zhang, Y. & Olson, E. N. (1996). Cooperative transcriptional activation by the neurogenic basic helix-loop-helix protein MASH1 and members of the myocyte enhancer factor-2 (MEF2) family. J. Biol. Chem. 271, 26659-26663. Breitbart, R. E., Liang, C. S., Smoot, L. B., Laheru, D. A., Mahdavi, V. & Nadal-Ginard, B. (1993). A fourth human MEF2 transcription factor, hMEF2D, is an early marker of the myogenic lineage. Development, 118, 1095-1106. BruÈnger, A. T., Adams, P. D., Clore, G. M., DeLano, W. L., Gros, P., Grosse-Kunstleve, R. W., Jiang, J. S., Kuszewski, J., Nilges, M., Pannu, N. S., Read, R. J., Rice, L. M., Simonson, T. & Warren, G. L. (1998). Crystallography and NMR system: a new software suite for macromolecular structure determination. Acta Crystallog. sect. D, 54, 905-921. Collaborative Computational Project Number 4 (1994). The CCP4 suite: programs for protein crystallography. Acta Crystallog. sect. D, 50, 760-763. Cowtan, K. (1994). ``dm'': an automated procedure for phase improvement by density modi®cation. CCP4/ ESF-EACBM Newsletter on Protein Crystallog. 31, 3438. Dodou, E., Sparrow, D. B., Mohun, T. & Treisman, R. (1995). MEF2 proteins, including MEF2A, are
448 expressed in both muscle and non-muscle cells. Nucl. Acids Res. 23, 4267-4274. Ferrin, T. E., Huang, C. C., L.E., J. & Langridge, R. (1988). The MIDAS display system. J. Mol. Graph. 6, 13-27. Flinta, C., Persson, B., Jornvall, H. & von Heijne, G. (1986). Sequence determinants of cytosolic N-terminal protein processing. Eur. J. Biochem. 154, 193196. Gossett, L. A., Kelvin, D. J., Sternberg, E. A. & Olson, E. N. (1989). A new myocyte-speci®c enhancerbinding factor that recognizes a conserved element associated with multiple muscle-speci®c genes. Mol. Cell Biol. 9, 5022-5033. Han, T. H. & Prywes, R. (1995). Regulatory role of MEF2D in serum induction of the c-jun promoter. Mol. Cell Biol. 15, 2907-2915. Han, J., Jiang, Y., Li, Z., Kravchenko, V. V. & Ulevitch, R. J. (1997). Activation of the transcription factor MEF2C by the MAP kinase p38 in in¯ammation. Nature, 386, 296-299. Jones, T. A., Zou, J. Y., Cowan, S. W. & Kjeldgaard, M. (1991). Improved methods for building protein models in electron density maps and the location of errors in these models. Acta Crystallog. sect. A, 47, 110-119. Kabsch, W. & Sander, C. (1983). Dictionary of protein secondary structure: pattern recognition of hydrogen-bonded and geometrical features. Biopolymers, 22, 2577-2637. Kato, Y., Kravchenko, V. V., Tapping, R. I., Han, J., Ulevitch, R. J. & Lee, J. D. (1997). BMK1/ERK5 regulates serum-induced early gene expression through transcription factor MEF2C. EMBO J. 16, 7054-7066. Katoh, Y., Molkentin, J. D., Dave, V., Olson, E. N. & Periasamy, M. (1998). MEF2B is a component of a smooth muscle-speci®c complex that binds an A T-rich element important for smooth muscle myosin heavy chain gene expression. J. Biol. Chem. 273, 1511-1518. Kaushal, S., Schneider, J. W., Nadal-Ginard, B. & Mahdavi, V. (1994). Activation of the myogenic lineage by MEF2A, a factor that induces and cooperates with MyoD. Science, 266, 1236-1240. Krainc, D., Bai, G., Okamoto, S., Carles, M., Kusiak, J. W., Brent, R. N. & Lipton, S. A. (1998). Synergistic activation of the N-methyl-D-aspartate receptor subunit 1 promoter by myocyte enhancer factor 2C and Sp1. J. Biol. Chem. 273, 26218-26224. Laskowski, R. A., MacArthur, M. W., Moss, D. S. & Thornton, J. M. (1993). PROCHECK: a program to check the stereochemical quality of protein structures. J. Appl. Crystallog. 26, 283-291. Lavery, R. & Sklenar, H. (1988). The de®nition of generalised helicoidal parameters and of axis curvature for irregular nucleic acids. J. Biomol. Struct. Dynam. 6, 63-91. Lee, Y., Nadal-Ginard, B., Mahdavi, V. & Izumo, S. (1997). Myocyte-speci®c enhancer factor 2 and thyroid hormone receptor associate and synergistically activate the alpha-cardiac myosin heavy-chain gene. Mol. Cell. Biol. 17, 2745-2755. Lilly, B., Zhao, B., Ranganayakulu, G., Paterson, B. M., Schulz, R. A. & Olson, E. N. (1995). Requirement of MADS domain transcription factor D-MEF2 for muscle formation in Drosophila. Science, 267, 688693.
Structure of MEF2A Bound to DNA Lin, Q., Schwarz, J., Bucana, C. & Olson, E. N. (1997). Control of mouse cardiac morphogenesis and myogenesis by transcription factor MEF2C. Science, 276, 1404-1407. Lin, Q., Lu, J., Yanagisawa, H., Webb, R., Lyons, G. E., Richardson, J. A. & Olson, E. N. (1998). Requirement of the MADS-box transcription factor MEF2C for vascular development. Development, 125, 45654574. Liu, S., Liu, P., Borras, A., Chatila, T. & Speck, S. H. (1997). Cyclosporin A-sensitive induction of the Epstein-Barr virus lytic switch is mediated via a novel pathway involving a MEF2 family member. EMBO J. 16, 143-153. Mao, Z. & Nadal-Ginard, B. (1996). Functional and physical interactions between mammalian achaetescute homolog 1 and myocyte enhancer factor 2A. J. Biol. Chem. 271, 14371-14375. Mao, Z., Bonni, A., Xia, F., Nadal-Vicens, M. & Greenberg, M. E. (1999). Neuronal activity-dependent cell survival mediated by transcription factor MEF2. Science, 286, 785-790. Martin, J. F., Schwarz, J. J. & Olson, E. N. (1993). Myocyte enhancer factor (MEF) 2C: a tissue-restricted member of the MEF-2 family of transcription factors. Proc. Natl Acad. Sci. USA, 90, 5282-5286. Meierhans, D. & Allemann, R. K. (1998). The N-terminal methionine is a major determinant of the DNA binding speci®city of MEF-2C. J. Biol. Chem. 273, 26052-26060. Meierhans, D., Sieber, M. & Allemann, R. K. (1997). High af®nity binding of MEF-2C correlates with DNA bending. Nucl. Acids Res. 25, 4537-4544. Miska, E. A., Karlsson, C., Langley, E., Nielsen, S. J., Pines, J. & Kouzarides, T. (1999). HDAC4 deacetylase associates with and represses the MEF2 transcription factor. EMBO J. 18, 5099-5107. Molkentin, J. D., Black, B. L., Martin, J. F. & Olson, E. N. (1995). Cooperative activation of muscle gene expression by MEF2 and myogenic bHLH proteins. Cell, 83, 1125-1136. Molkentin, J. D., Black, B. L., Martin, J. F. & Olson, E. N. (1996a). Mutational analysis of the DNA binding, dimerization, and transcriptional activation domains of MEF2C. Mol. Cell. Biol. 16, 2627-2636. Molkentin, J. D., Li, L. & Olson, E. N. (1996b). Phosphorylation of the MADS-Box transcription factor MEF2C enhances its DNA binding activity. J. Biol. Chem. 271, 17199-171204. Murshudov, G. N., Lebedev, A., Vagin, A. A., Wilson, K. S. & Dodson, E. J. (1999). Ef®cient anisotropic re®nement of macromolecular structures using FFT. Acta Crystallog. sect. D, 55, 247-255. Navaza, J. (1994). AMoRe: an automated package for molecular replacement. Acta Crystallog. sect. A, 50, 157-163. Nurrish, S. J. & Treisman, R. (1995). DNA binding speci®city determinants in MADS-box transcription factors. Mol. Cell. Biol. 15, 4076-4085. Ornatsky, O. I. & McDermott, J. C. (1996). MEF2 protein expression, DNA binding speci®city and complex composition, and transcriptional activity in muscle and non-muscle cells. J. Biol. Chem. 271, 2492724933. Otwinowski, Z. (1991). Isomorphous replacement and anomalous scattering. In Proceedings of the CCP4 Study Weekend 25-26 January 1991 (Wolf, W. E. P. R. & Leslie, A. G. W., eds), pp. 80-88, SERC Daresbury Laboratory, Warrington, UK.
Structure of MEF2A Bound to DNA Otwinowski, Z. (1993). Oscillation data reduction program. In Proceedings of the CCP4 Study Weekend: Data Collection and Processing, 29-30 January (Sawyer, L. I. N. & Bailey, S., eds), pp. 55-62, SERC Daresbury Laboratory, Daresbury, UK. Pellegrini, L., Tan, S. & Richmond, T. J. (1995). Structure of serum response factor core bound to DNA. Nature, 376, 490-498. Perez-Martin, J. & de Lorenzo, V. (1997). Clues and consequences of DNA bending in transcription. Annu. Rev. Microbiol. 51, 593-628. Perrakis, A., Sixma, T. K., Wilson, K. S. & Lamzin, V. S. (1997). wARP: improvement and extension of crystallographic phases by weighted averaging of multiple-re®ned dummy atomic models. Acta Crystallog. sect. D, 53, 448-455. Pollock, R. & Treisman, R. (1991). Human SRF-related proteins: DNA-binding properties and potential regulatory targets. Genes Dev. 5, 2327-2341. Rao, S., Karray, S., Gackstetter, E. R. & Koshland, M. E. (1998). Myocyte enhancer factor-related B-MEF2 is developmentally expressed in B cells and regulates the immunoglobulin J chain promoter. J. Biol. Chem. 273, 26123-26129. Schulz, R. A., Chromey, C., Lu, M. F., Zhao, B. & Olson, E. N. (1996). Expression of the D-MEF2 transcription in the Drosophila brain suggests a role in neuronal cell differentiation. Oncogene, 12, 1827-1831. Sharrocks, A. D., von Hesler, F. & Shaw, P. E. (1993). The identi®cation of elements determining the different DNA binding speci®cities of the MADS box proteins p67SRF and RSRFC4. Nucl. Acids Res. 21, 215-221. Shore, P. & Sharrocks, A. D. (1995). The MADS-box family of transcription factors. Eur. J. Biochem. 229, 1-13. Sparrow, D. B., Miska, E. A., Langley, E., ReynaudDeonauth, S., Kotecha, S., Towers, N., Spohr, G., Kouzarides, T. & Mohun, T. J. (1999). MEF-2 func-
449 tion is modi®ed by a novel co-repressor, MITR. EMBO J. 18, 5085-5098. Tan, S. & Richmond, T. J. (1998). Crystal structure of the yeast MATalpha2/MCM1/DNA ternary complex. Nature, 391, 660-666. West, A. G. & Sharrocks, A. D. (1999). MADS-box transcription factors adopt alternative mechanisms for bending DNA. J. Mol. Biol. 286, 1311-1323. West, A. G., Shore, P. & Sharrocks, A. D. (1997). DNA binding by MADS-box transcription factors: a molecular mechanism for differential DNA bending. Mol. Cell. Biol. 17, 2876-2887. Wilson-Rawls, J., Molkentin, J. D., Black, B. L. & Olson, E. N. (1999). Activated notch inhibits myogenic activity of the MADS-Box transcription factor myocyte enhancer factor 2C. Mol. Cell. Biol. 19, 28532862. Yang, S. H., Galanis, A. & Sharrocks, A. D. (1999). Targeting of p38 mitogen-activated protein kinases to MEF2 transcription factors. Mol. Cell. Biol. 19, 40284038. Youn, H. D., Sun, L., Prywes, R. & Liu, J. O. (1999). Apoptosis of T cells mediated by Ca2-induced release of the transcription factor MEF2. Science, 286, 790-793. Yu, Y. T., Breitbart, R. E., Smoot, L. B., Lee, Y., Mahdavi, V. & Nadal-Ginard, B. (1992). Human myocyte-speci®c enhancer factor 2 comprises a group of tissue- restricted MADS box transcription factors. Genes Dev. 6, 1783-1798. Zetser, A., Gredinger, E. & Bengal, E. (1999). p38 mitogen-activated protein kinase pathway promotes skeletal muscle differentiation. Participation of the Mef2c transcription factor. J. Biol. Chem. 274, 51935200. Zhao, M., New, L., Kravchenko, V. V., Kato, Y., Gram, H., di Padova, F., Olson, E. N., Ulevitch, R. J. & Han, J. (1999). Regulation of the MEF2 family of transcription factors by p38. Mol. Cell. Biol. 19, 21-30.
Edited by J. Karn (Received 5 January 2000; received in revised form 31 January 2000; accepted 31 January 2000)