High affinity binding of MEF-2C correlates with DNA bending
High affinity binding of MEF-2C correlates with DNA bendingDaniel Meierhans, Martin Sieber and Rudolf K. Allemann*
Department of Chemistry, ETH-Zürich, Universitätstrasse 16, CH-8092 Zurich, Switzerland
Received August 5, 1997;Revised and Accepted September 23, 1997
ABSTRACT
To regulate lineage-specific gene expression in many cell types, members of the myocyte enhancer factor-2 (MEF-2) family of transcription factors cooperate with basic helix-loop-helix (bHLH) proteins, which show only limited intrinsic DNA binding specificity. We investigated the DNA binding properties of MEF-2C in vitro and show that the inherent bendability of the MEF site is one of the principal structural characteristics recognized by MEF-2C. Measurements of the apparent dissociation constants of MEF-2C complexes with several DNA sequences revealed that MEF-2C bound with high affinity to DNA sequences containing a MEF site. Mutations in the MEF site which did not affect the bendability of the DNA changed the free energy of binding only marginally. However, reducing the intrinsic bendability of the DNA binding site through an AA -> GC substitution increased the half-maximal binding concentration of MEF-2C by almost one order of magnitude. Electrophoretic mobility shift assays revealed markedly reduced MEF-2C binding to DNA containing 2,6-diaminopurine. On binding to MEF-2C the maximum ellipticity at 275 nm in the CD spectrum of DNA containing a MEF site was red shifted by 4 nm and its intensity reduced significantly, while a slight blue shift of <1 nm was observed for a mutant DNA sequence with reduced bendability (AA -> GC). Bending analysis by circular permutation assay revealed that the DNA in the cognate complex was bent by 49°, while the DNA in the complex with the mutant oligonucleotide was largely unbent.
INTRODUCTION
Cellular determination and differentiation during embryonal development rely on the establishment of regulatory programmes that control the temporal and spatial specificity of gene expression. Members of the basic helix-loop-helix (bHLH) family of transcription factors are involved in control of lineage-specific gene expression in many cell types, including skeletal muscle and neurons (1 -4 ).
In the skeletal muscle lineage the four bHLH proteins Myf-5, myogenin, MyoD and MRF-4 make up a regulatory system that controls myoblast differentiation (5 ,6 ). Expression of any one of these factors can activate myogenesis in a wide variety of both muscle and non-muscle cell types (1 -3 ). Expression of the bHLH protein MASH-1, on the other hand, promotes differentiation of committed neuronal precursor cells (4 ). In sharp contrast to their biological specificity, these bHLH proteins display only limited intrinsic specificity of DNA binding (7 ,8 ). There is strong biochemical evidence suggesting that they rely on members of the myocyte enhancer factor-2 (MEF-2) family of transcription factors to activate cell type-specific gene expression (9 -12 ).
MEF-2 was originally identified as a muscle-specific DNA binding activity which is induced during differentiation of skeletal myoblasts (13 ). In humans and mice the MEF-2 DNA binding activity is encoded by four genes, mef-2A-mef-2D (14 -17 ). At their N-termini the MEF-2 proteins share a region of high sequence similarity comprising a MADS box and a MEF-2 domain (Fig. 1 A). The MADS domain contains 56 amino acids and is named after the first four proteins identified as containing this domain, namely MCM1 in yeast, the plant homeotic proteins Agamous and Deficiens and serum response factor (SRF) in vertebrates (18 ). Directly adjacent to the MADS box is the 29 amino acid MEF-2 domain, which is unique to the MEF-2 family of proteins. Outside these two domains the MEF-2 sequences are highly diverged.
MATERIALS AND METHODS
Expression of MEF-2C(1-117) and protein purification
An N-terminal fragment of murine MEF-2C protein was produced in BL21(DE3)pLys cells (22 ) from the T7 promoter in plasmid pJGetita essentially as previously described (7 ,8 ,23 ). SDS-PAGE of the purified protein showed a single band. MALDI-TOF mass spectroscopy revealed a molecular mass of 13 386 mu, which corresponded well with the calculated mass of 13 377 mu for MEF-2C(1-117) lacking the N-terminal methionine. Edman sequencing confirmed lack of the N-terminal methionine as well as the identity of the first eight amino acid residues of the recombinant protein. The protein concentration was determined by measuring the UV absorption at 215 and 220 nm (24 ). The protein yield was ~5 mg purified protein/l culture.
Oligonucleotides
Oligonucleotides were purchased from Microsynth or the University of Zurich (Institute for Zoology), desalted on Sephadex and precipitated with ethanol. Phosphoramidites of 6-deaza-2'-desoxyadenosine and 2-amino-2'-desoxyadenosine (DAP) were purchased from Glen Research and incorporated into synthetic oligonucleotides using standard chemistry on an Applied Biosystems DNA synthesizer. Single-stranded oligonucleotides were labelled with [[gamma]-32P]ATP (Amersham) in the presence of T4 polynucleotide kinase (New England Biolabs). Complementary strands were annealed by heat denaturation followed by slow cooling to room temperature.
Electrophoretic mobility shift assays
Electrophoretic mobility shift assays were performed as previously described (7 ,8 ). Bacterially expressed MEF-2C was serially diluted into EMSA buffer [50 mM Tris, pH 7.9, 6 mM MgCl2, 40 mM ammonium sulphate, 0.2 mM EDTA, 1 mM DTT, 100 mM KCl or potassium glutamate (as indicated) and 5% glycerol]. This solution was incubated in the presence of 10 nM labelled oligonucleotide for 10 min at room temperature. Samples were applied to 4% polyacrylamide gels in 0.9× TAE, pH 7.9. After electrophoresis the gels were dried and exposed to Kodak X-OMAT-S film at -70oC. Quantitative data were obtained with a Packard Instantimager using system software. The fraction [Phi] of DNA bound was determined as the activity of the retarded band (corresponding to the protein-DNA complex) divided by the sum of the activities of the retarded and unretarded (corresponding to the free DNA) bands.
CD spectroscopy
Circular permutation assays
The plasmids pbMEF and pbMEF-GC were prepared by inserting the synthetic oligonucleotides 5'-CTAGATGCTGCTATAAATAGAGTG and 5'-CTAGATGCTGCTATGCATAGAGTG respectively between the XbaI and SalI restriction sites in plasmid pTK401 (25 ). DNA fragments were labelled by PCR amplification in the presence of [[alpha]-32P]dATP (Amersham) using primers complementary to sequences flanking the inserts. To generate circularly permuted DNA fragments the PCR products were digested with the restriction enzymes MluI, BglII, XhoI, PvuII, BglI, StuI, RsaI and BamHI as indicated in Figure 5 A. The probes were purified by agarose gel electrophoresis and recovered by electroelution and ethanol precipitation.
EMSAs were performed essentially as described above, except that the samples were applied to 8% polyacrylamide gels. Electrophoresis was performed at 4oC and 17 V/cm for 3 h. Gels were dried and exposed to Kodak X-OMAT-S film at -70oC. The mobility of the MEF-2C-DNA complexes, normalized to the mobilities of the free probes, was plotted against the flexure displacement of the probes, which was defined as the distance of the centre of the MEF-2 site from the 5'-end of the probe devided by the total length of the probe (26 ).
RESULTS
Expression, purification and characterization of MEF-2C(1-117)
A gene encoding the first 117 amino acids of MEF-2C was constructed, expressed in E.coli under control of the T7 promoter and the gene product purified to apparent homogeniety (Fig. 2 A).
Circular dichroism spectroscopy of the recombinant MEF-2C, which lacked the N-terminal methionine, revealed that, unlike many transcription factors (7 ,8 ,27 -30 ), the protein adopted a folded structure in solution even in the absence of DNA (Fig. 2 B). Approximately 23% of MEF-2C was in an [alpha]-helical conformation, corresponding to 27 of the 117 amino acids (31 ). The high sequence similarity between MEF-2C and SRF in the N-terminal region of the MADS box (Fig. 1 ) and the crystal structure of SRF (19 ) suggest that the amino acids from Asp13 to Cys39 adopt an [alpha]-helical conformation in MEF-2C. This region was found to interact with DNA in SRF (19 ). The structure of MEF-2C did not change for concentrations of MEF-2C between 1 µM and 100 nM (data not shown), the concentration at which DNA binding occurs (vide infra).
The secondary structure of MEF-2C was not altered by addition of an equimolar amount of a double-stranded oligonucleotide containing either a MEF site (Fig. 2 B) or MEF-4G,5C (data not shown).
DNA binding specificity of MEF-2C
In EMSA titrations the concentration of MEF-2C at which half maximal binding to a double-stranded oligonucleotide containing a consensus MEF site occurs was measured. In these experiments the concentration of the oligonucleotide was held constant while the concentration of MEF-2C was successively increased (Fig. 3 A). The best fit for the binding isotherm (equation 1) was found under the assumption of one protein monomer binding to one oligonucleotide (Fig. 3 B):
[Phi] = K[P]/(1 + K[P])
1
Characterization of the DNA binding reaction of MEF-2C
CD spectroscopy was used to obtain structural information about the DNA binding reaction of MEF-2C. As mentioned above, CD did not reveal any significant changes in the conformation of MEF-2C on binding to DNA (Fig. 2 B). However, informative changes in the CD spectrum of the DNA were observed upon binding to MEF-2C. The spectra of double-stranded oligonucleotides containing either a MEF (Fig. 4 A) or a MEF-4G,5C site (Fig. 4 B) exhibited the positive maximum of ellipticity at 275 nm characteristic of B-DNA (36 ,37 ). The contribution of MEF-2C to the ellipticity in this region was negligible (Fig. 4 A and B). Adding one equivalent of MEF-2C dimer to the double-stranded oligonucleotide containing a MEF site led to a red shift of 4 nm of the maximum ellipticity at 275 nm and a reduction of the intensity by >30%, indicative of a change in DNA conformation (Fig. 4 A). The change in the spectrum of MEF-4G,5C-DNA upon binding to MEF-2C was much smaller. The maximum ellipticity at 275 nm was blue shifted by <1 nm to ~274 nm and the reduction in the intensity of the CD signal was smaller (Fig. 4 B).
Characterization of DNA complexes of MEF-2C by circular permutation assays
DISCUSSION
The problem of sequence-specific DNA recognition by proteins is central to understanding cellular events such as transcription initiation. The precise recognition of a defined DNA sequence by a given protein necessitates an optimal complementarity between the binding sites of the interacting species. Both the DNA and the protein can thereby adapt their conformation in order to ensure an optimal fit between the respective recognition elements. While many proteins bind to relatively unperturbed B-DNA in a sequence-specific manner via a pattern of hydrogen bonds and van der Waals interactions from amino acid sidechains to atoms along the floor of the major groove (41 -45 ), it has long been realised that DNA sequences can also be recognized through their capacity to be deformed (46 ,47 ). It was commonly assumed, however, that deformations such as DNA bending are induced by the protein.
Crystal structure analysis of numerous double-stranded oligonucleotides has revealed that certain nucleotide sequences have a high intrinsic tendency to adopt a bent conformation (20 ,21 ). For instance, DNA bending by ~10-20o has been observed at junctions between G/C- and A/T-rich regions (48 ,49 ) as a consequence of the inherent deformability of groove width in A/T-rich regions (48 ,50 ) and due to the lack of steric hindrance involving the 2-amino group in G/C-rich regions (21 ,32 -34 ). The helical bend is thereby generated by large positive roll angles at the G/C to A/T steps. The A/T-rich core displays a small but significant average roll which is made possible through increased propeller twists in this region. The increased propeller twist in the A/T-rich region is facilitated because A-T pairs are held together by only two hydrogen bonds rather than the three formed between guanine and cytosine (33 ). However, it is important to stress that while such sequences can be bent, they do not necessarily have to adopt a bent conformation. The X-ray analysis of a double-stranded oligonucleotide with the palindromic sequence CGCGAATTCGCG showed that the GC/AT junction at one end of the DNA was bent by 18o, while the junction at the other end was unbent (20 ,49 ). Therefore, the GC/AT junction is a hinge that makes that region of the double helix more flexible and capable of adopting a bent conformation. Proteins that bind short A/T runs might simply take advantage of and potentiate this intrinsically high tendency to be bent.
The core sequence of the MEF site is TATAAATA and we tested the hypothesis that MEF-2C recognizes its DNA target site partly by its potential to adopt a bent conformation. To this end we measured the affinities of MEF-2C for a number of oligonucleotides with mutated MEF sites (Fig. 1 B). Neither replacing adenine 4 or 5 with thymine nor changing thymine 3 to adenine significantly changed the concentration of MEF-2C needed to bind 50% of the respective oligonucleotides (Table 1 ). However, replacing adenines 4 and 5 respectively with guanine and cytosine increased the half maximal binding concentration by approximately one order of magnitude.
Removal of the amino group in the 6 position of adenine leads to a purine-thymine base pair with only a single hydrogen bond, thereby facilitating formation of propeller twist angles. Replacing adenine 5 with purine (Fig. 1 B) did not significantly change the stability of the complex with MEF-2C. Purines in position 3' or 4, on the other hand, reduced the complex stability slightly.
There is experimental evidence suggesting that bending A runs into the minor groove at position 2 of the base reduces the unfavourable interaction between the stacked exocyclic amino groups of adenine (33 ) and can lead to attractive bifurcated interstrand hydrogen bonds between the exocyclic amino group of adenine and O(4) of the adjacent thymine (51 ). Our data suggest that the reduction in exergonic character of the binding reaction due to removal of the amino group can be compensated for by increased flexibility of the purine-thymine base pair.
Interpretation of the results obtained with the unnatural base 2,6-diaminopurine was more straightforward. The 2-amino group of DAP allows formation of a third hydrogen bond to thymine, introduces steric hinderance, when the DNA is bent into the minor groove, and reduces the intrinsic curvature and bendability of the DNA (21 ,32 -35 ). Indeed, when DAP replaced adenine in the 4 or 5 position of the MEF site the affinity for MEF-2C was substantially reduced (Table 2 and Fig. 3 C).
Taken together, the results of the binding studies strongly suggest that base changes within the core region of the MEF site significantly reduce the affinity for MEF-2C mainly if these mutations diminish the intrinsic bendability of the DNA. However, it must be mentioned in this context that the introduction of an amino group on C(2) of the purine bases might also introduce steric hindrance for interaction with the protein. Indeed, crystal structure analysis of the complex of SRF with its binding site CCT(1)A(2)A(3)TTAGG revealed that one of the sidechain amino groups of Arg143 of SRF is simultaneously hydrogen bonded to N(3) of adenine 3 and to O(2) of the thymine, which forms a base pair with adenine 2. It is likely that Arg3 of MEF-2C and Arg143 of SRF (Fig. 1 A) form similar contacts with their respective DNA binding sites, in that Arg3 of MEF-2C could contact N(3) of adenine 4 (Fig. 1 B) and the base in position 3. Inspection of the structure of the DNA complex of SRF (19 ) revealed that introduction of an amino group in the 2 position of the central adenine (for instance in MEF-4G,5C or by replacing adenine with DAP) should not greatly disturb the hydrogen bond between Arg3 and N(3) of the base because N[eta] is rather remote from C(2) of adenine 4 and O(2) of thymine 4'. However, the introduction of an additional hydrogen bond might reduce the propeller twist between the bases in position 4, thereby leading to a reduction in the favourable interaction with N[eta] of R(3).
The dissociation constant of the complex of MEF-2C with an oligonucleotide containing a heterologous sequence was 10 times greater than for the complex with the MEF-4G,5C site (Table 1 ), indicating that the bendability of the MEF site accounts for approximately half of the binding specificity of MEF-2C, while the other half is achieved through sequence-specific contacts between the DNA bases flanking the core region and the protein.
While the results of the binding studies suggested that the intrinsic bendability of the DNA was important for high affinity MEF-2C binding, only the CD experiments and the data obtained from circular permutation assays revealed a direct correlation between the intrinsic bendability of the binding site and the extent of the actual distortion of the DNA in complexes with MEF-2C. When MEF-2C bound to an oligonucleotide containing a MEF site the maximum positive ellipticity at 275 nm, which is characteristic of unperturbed B-DNA, was shifted by 4 nm to 279 nm and its intensity reduced significantly (Fig. 4 A), suggesting a major structural reorganization of the DNA. This red shift is not readily interpretable in structural terms. cAMP receptor, a protein known to cause DNA bending, induces a red shift and a slight reduction in the CD signal around 275 nm upon binding to its cognate DNA target site (52 ). On the other hand, while the HMG box protein SOX-5 bends DNA by ~74o, the positive ellipticity in the CD spectrum of its target DNA is blue shifted on binding (53 ).
In sharp contrast, binding of MEF-2C to the MEF-4G,5C site led to a slight blue shift (Fig. 4 B). The small reduction in ellipticity around 275 nm suggested that the heterologous DNA also underwent a conformational change upon binding. However, the structural reorganization of the mutant DNA appears to be much less dramatic and of a different nature than that of the cognate MEF site. These observations suggested that the DNA in the complex between MEF-2C and MEF site DNA adopted a bent conformation, while the structure of the DNA in the complex with the mutant MEF-4G,5C site was closer to the conformation of B-DNA.
The results obtained in circular permutation experiments confirmed this view. The relative mobilities of the complex of MEF-2C with DNA containing a MEF site were strongly dependent on the position of the MEF site along the length of the probe (Fig. 5 B and D) and indicated that DNA was bent by ~49o. Circular permutation assays often underestimate bend angles and this value might therefore represent a lower limit for the actual bend angle (40 ). A DNA bend angle of 72o was observed in the co-crystal structure of SRF and its AT-rich target site (19 ). The observation that both SRF and MEF-2C bend their DNA target sequences to a similar extent suggests that the mechanisms of DNA recognition might be similar in both cases.
On the other hand, the mobilities of the complexes with DNA containing the mutant MEF-4G,5C site were only slightly affected by the position of the binding site (Fig. 5 C and D), suggesting that in this case the DNA remained largely unbent. Different mobilities of probes with circularly permuted binding sites cannot only result from bends but also from other distortions of the DNA caused by the extended shape of certain proteins (54 ). However, the crystal structure analysis of SRF bound to DNA showed that SRF adopted an almost globular tertiary structure (19 ) and the sequence similarity between MEF-2C and SRF suggests that their tertiary structures are similar (Fig. 1 A). More importantly, our results revealed a significant difference between the behaviour of the complexes of MEF-2C with MEF site and with mutant DNA. The only difference in the 141 bp probes used for bending analysis of the MEF site and MEF-4G,5C complexes was a mutation, AA -> GC, in the centre of the MEF site core region. Nevertheless, our results depended dramatically on the nature of these two base pairs, while the conformations adopted by MEF-2C in complexes with the MEF site and with mutant DNA were indistinguishable by CD. In addition, quantitative prediction of local DNA bending and curvature propensities showed that replacement of the central AA dinucleotide in the MEF site with GC decreased the ability of the DNA to adopt a bent conformation (55 -57 ). Therefore, the different behaviour of the complexes of MEF-2C with circularly permuted MEF and MEF-4G,5C sites was most likely due to different DNA conformations and the mobility changes in the complexes with the circularly permuted MEF site probes were caused by DNA bending.
Many other proteins take advantage of the inherent deformability of their target sites (58 ). The conformations of the DNA in complexes with TBP (59 -61 ), SRF (19 ), EcoRI (62 ), EcoRV (63 ) and mEcoRI methyltransferase (64 ), all of which recognize short sequences rich in A-T base pairs, are significantly bent in order to optimize complementarity between the binding sites of the interacting species. For DNA complexes with EcoRI and EcoRV the observed DNA bending could be correlated with the bendability of the DNA (57 ).
Proteins that bind to sequences with low bendability often do not bend the DNA and rely for recognition soley on sequence-specific contacts with the DNA bases (41 -45 ,54 ,58 ). No bending was observed in the complex of EcoRV with the non-cognate DNA binding site GAGCTG. The X-ray structure of the same protein with its natural DNA target site GATATC revealed that in this case the DNA was bent by 50o (63 ). However, it is important to stress that even DNA sequences of low deformability can be bent upon binding to a protein, but such bending is energetically costly (58 ,65 ).
In summary, we have shown that the sequence preference of MEF-2C is governed not only by sequence-specific contacts between amino acid residues and bases of the DNA, but also to a significant extent by the intrinsic bendability of the DNA target site. MEF-2C appears to take advantage of the inherent deformability of A/T-rich regions and to bend the DNA further through specific interactions with the flanking base pairs (58 ). The ability of MEF-2C to recognize DNA of increased bendability might allow it to function as an adaptor for transcription factors, such as the members of the bHLH family, which do not have sufficient intrinsic DNA binding specificity to recognize their target sites.
ACKNOWLEDGEMENTS
We thank Eric N.Olson for providing the murine MEF-2C cDNA clone, Tom Kerppola for plasmid pTK401, Peter James for MALDI-TOF mass spectroscopy, Lesley A.Tannahill for critical reading of the manuscript and the members of our laboratory for helpful discussions. This work was supported by a TH grant (R.K.A. and D.M.).
37 Johnson,W.C.,Jr (1996) In Fasman,G.D. (ed.), Circular Dichroism and the Conformational Analysis of Biomolecules. Plenum Press, New York, NY, pp. 433-468.
