Structure of the Trichomonas vaginalis Myb3 DNA-binding domain bound to a promoter sequence reveals a unique C-terminal β-hairpin conformation.

Trichomonas vaginalis Myb3 transcription factor (tvMyb3) recognizes the MRE-1 promoter sequence and regulates ap65-1 gene, which encodes a hydrogenosomal malic enzyme that may play a role in the cytoadherence of the parasite. Here, we identified tvMyb3(53-180) as the essential fragment for DNA recognition and report the crystal structure of tvMyb3(53-180) bound to MRE-1 DNA. The N-terminal fragment adopts the classical conformation of an Myb DNA-binding domain, with the third helices of R2 and R3 motifs intercalating in the major groove of DNA. The C-terminal extension forms a β-hairpin followed by a flexible tail, which is stabilized by several interactions with the R3 motif and is not observed in other Myb proteins. Interestingly, this unique C-terminal fragment does not stably connect with DNA in the complex structure but is involved in DNA binding, as demonstrated by NMR chemical shift perturbation, (1)H-(15)N heteronuclear-nuclear Overhauser effect and intermolecular paramagnetic relaxation enhancement. Site-directed mutagenesis also revealed that this C-terminal fragment is crucial for DNA binding, especially the residue Arg(153) and the fragment K(170)KRK(173). We provide a structural basis for MRE-1 DNA recognition and suggest a possible post-translational regulation of tvMyb3 protein.

INTRODUCTION

Trichomonas vaginalis is one of the most widespread sexually transmitted human pathogens. Infection with T. vaginalis causes the disease trichomoniasis and is also associated with several adverse health consequences, including increased HIV transmission and infertility (1,2). With the increasing number of drug-resistant clinical T. vaginalis strains (3,4), the infection caused by T. vaginalis could become a major threat to public health. In T. vaginalis, the iron-responsive promoter region of the gene ap65-1 has two discrete Myb protein-recognition sites, MRE-1/MRE-2r (TATA, MRE-1 in italic and MRE-2r underlined) and MRE-2f (TATCGT), interspersed among several closely spaced DNA elements (5,6). The ap65-1 gene encodes a 65-kDa hydrogenosomal malic enzyme that is reputed to be a prominent trichomonad adhesin and plays a role in the cytoadherence of T. vaginalis (7,8). Three T. vaginalis Myb transcription factors, tvMyb1, tvMyb2 and tvMyb3 (Figure 1), were identified to regulate ap65-1 (9–11). The tvMyb1 and tvMyb2 proteins, identified by Southwestern library screening with a probe of concatenated MRE-2f sequences, conferred antagonistic actions on basal and iron-inducible transcription of ap65-1 through dual recognition and differential promoter selection toward both the MRE-1/MRE-2r and MRE-2f sites (10,11). Recently, tvMyb3 protein was identified with use of a probe with concatenated MRE-1/MRE-2r sequences. The tvMyb3 protein was found to recognize only the MRE-1 element to activate basal and iron-inducible ap65-1 transcription. These studies demonstrate that tvMyb2 and tvMyb3 may coactivate basal and iron-inducible ap65-1 transcription against tvMyb1 through conditional and competitive promoter entries (9).

Figure 1.

Sequence alignment of DBD of tvMyb3, tvMyb1, tvMyb2 and c-Myb. Secondary structure of tvMyb353–180 is shown on the top of the amino acid sequence. Amino acid numbering for each protein is indicated in the termini of sequence.

Myb proteins are characterized by a highly conserved DNA-binding domain (DBD), the Myb domain (12). Myb DBD generally consists of 1–4 imperfect repeats (R), with each repeat typically containing approximately 50 amino acids divided by three regularly spaced Trp (or hydrophobic) residues that form a hydrophobic core in a 3D structure. The repeat forms a three-helix bundle structure, and the second and the third helices adopt a helix-turn-helix conformation, with the third helix (the recognition helix) directly contacting DNA in the major groove (12). In vertebrates, c-Myb protein was first identified to contain a three-repeat R1R2R3 DBD (13). Structural study showed that c-Myb R2 and R3 motifs are packed closely with each other so that the two recognition helices can cooperatively recognize specific DNA in the major groove. In Arabidopsis thaliana, the Myb protein family is large and contains more than 190 distinct members, most of which contain an R2R3 DBD (126 genes) with at least 40% sequence identity (12,14). Recently, Myb DBD from plant telomere-binding protein was found to contain a C-terminal extension required for binding plant telomeric DNA (15,16). Structural studies further revealed that this Myb DBD adopts a novel four-helix tetrahedron structure and the C-terminal extension forms the fourth helix, which, although not interacting with DNA directly, can stabilize the overall structure of DBD and is crucial for telomeric DNA recognition (17–19).

The tvMyb3 protein contains R1, R2 and R3 motifs and shares high sequence homology with tvMyb1, tvMyb2 and human c-Myb in the R2 and R3 motifs (Figure 1). However, tvMyb3 protein contains a C-terminal extension rich in basic amino acid residues (R168SKKRK173), which resembles the canonical nuclear localization signals (20,21). In the present study, we used electrophoretic mobility shift assay (EMSA) and isothermal titration calorimetry (ITC) and identified the DBD of tvMyb3 to be the fragment spanning residues 53–180, which encompasses R2, R3 motifs and a C-terminal extension (designated tvMyb353–180). This fragment shows about a 40 times increase in MRE-1 DNA binding ability than the shorter fragment (tvMyb353–161), which encompasses a similar R2R3 DBD of tvMyb1 that was shown in our previous study to interact with MRE-1 DNA strongly. The crystal structure of tvMyb353–180 in complex with MRE-1 DNA was then determined; the C-terminal extension forms a β-hairpin conformation followed by a flexible tail, with missing electron density for the last 10 residues. The roles of C-terminal extension, including the β-hairpin and the C-terminal tail, in DNA binding were further investigated by NMR and site-directed mutagenesis. Our study provides a structural basis for MRE-1 DNA recognition, identifies a novel β-hairpin conformation in the C-terminal extension and suggests a possible regulation of MRE-1 element by acetylation or phosphorylation at the C-terminal extension of the tvMyb3 DBD.

MATERIALS AND METHODS

Sample preparation

The DNA fragments encoding tvMyb353–161 and tvMyb353–180 were cloned into the pET29b (Novagen) vector using NdeI and XhoI restriction sites to obtain a recombinant protein with a C-terminal His-tag. The various mutants of tvMyb353–180 were generated by the QuikChange site-directed mutagenesis protocol (Stratagene) and confirmed by DNA sequencing. The steps for expression and purification of tvMyb3 recombinant proteins are similar to those of tvMyb1 (22). All proteins were expressed in Escherichia coli BL21(DE3). For labeled (15N, 15N/13C and 15N/13C/2H) protein samples, the cells were grown in H2O- or D2O-containing M9 minimal medium containing 15NH4Cl (1 g/l) and/or 13C-glucose (2 g/l) at 37°C. The cells were lysed by microfluidizer, and the proteins were purified by nickel–nitrilotriacetic acid (Ni–NTA) affinity chromatography. Purity and authenticity of the recombinant proteins were verified by SDS–PAGE and mass analysis. Finally, the target protein was dialyzed and concentrated with buffer (20 mM NaH2PO4, 50 mM NaCl, 0.5 mM NaN3 and 10 mM dithiothreitol) at pH 6.0 for study by NMR and circular dichroism (CD). The single-stranded DNAs of MRE-1, MRE-2f and biotinylated DNAs for surface plasmon resonance (SPR) experiments were purchased from MDBio Inc. (Taiwan). The double-stranded DNA was prepared by mixing equal amounts of two complementary deoxynucleotides, heating to 95°C for 10 min and cooling slowly to room temperature.

EMSA

Oligonucleotides were labeled at the 5′-end with [γ-32P] ATP (Amersham Pharmacia Biotech) by T4 polynucleotide kinase as described by the supplier (Promega). All protein samples were incubated with γ-32P-labeled 16-bp MRE-1 or MRE-2f. The mixtures were separated on a 10% acrylamide gel by electrophoresis. The DNA–protein complexes were detected and imaged by use of a Typhoon 9410 variable mode imager.

ITC

ITC experiments were performed at 25°C with a MicroCal iTC200 system (GE Healthcare). The DNA and protein samples were dialyzed in phosphate buffer containing 20 mM NaH2PO4, 150 mM NaCl, 0.5 mM NaN3 and 1 mM dithiothreitol at pH 6.0. The proteins (30 µM) in the calorimetric cell were titrated with 300 µM DNA in the injection syringe. Titration was initiated by one 1-µl injection followed by 19 injections of 2 µl spaced by 180 s intervals. Control experiments involved titrating the DNA samples into buffer. The raw calorimetry data were collected and analyzed by use of the software Origin 7.0 (Microcal Inc., Northampton, MA, USA). The binding isotherms were fitted to the one-site binding model, giving values of the stoichiometry (n) of the interaction, the enthalpy of binding (ΔH) and the association constant (Ka).

Crystallization and data collection

Crystallization trials of the tvMyb353–180/16-bp MRE-1 DNA complex involved the hanging-drop vapor-diffusion method. The tvMyb353–180/DNA complex (1:1 molar ratio) was prepared in a buffer containing 20 mM HEPES, pH 7.0, 50 mM NaCl and 1 mM dithiothreitol. Initial crystals were obtained by use of crystal screen kits (Hampton Research) at 4 mg/ml. These crystals grew in a buffer of 0.1 M sodium cacodylate trihydrate, 0.1 M magnesium acetate tetrahydrate and 10% w/v PEG8000 at pH 6.4. The cryogenic data for the tvMyb353–180/DNA crystal were collected by the use of the BL13B1 beamline of the National Synchrotron Radiation Research Center (NSRRC) in Taiwan. Crystal was flash-frozen in a liquid-nitrogen stream at 100 K. The data were processed and scaled by use of the program HKL2000 (23). The crystal diffracted to 2.9 Å and belongs to the P1 space group with unit-cell parameters, a = 45.81, b = 71.77, c = 87.82, α = 94.68, β = 97.84 and γ = 99.28, and Rmerge 4.6%. The VM was calculated as 2.71 Å3Da−1, corresponding to a solvent content of 45%, containing four molecules per asymmetric unit.

Structure determination and refinement

The crystal structure of the tvMyb353–180/16-bp MRE-1 DNA complex was determined by the molecular replacement method with MOLREP (24). The c-Myb/DNA complex (PDB code: 1H8A) with the highest protein sequence identity of 50% with tvMyb353–180/DNA was selected as the template model. Data between 8.0 and 4.0 Å and a Patterson radius of 25 Å were used for the rotation and translation function calculations. Significant rotation and translation solutions were obtained for the tvMyb353–180/DNA. Structural model building involved XtalView programs (25) and structural refinement was calculated by the use of CNS (26). A bulk solvent mask was calculated to improve the reflection data. The final model for the tvMyb353–180/DNA contains Lys53 to Lys170 in one molecule and Lys53 to Ser169 in the others. A total of 98 water molecules and four DNA molecules were located in the electron density map. The model was refined to a Rfactor of 21.8% for all reflections above 2σ and a Rfree of 27.3% using 5.0% randomly distributed reflections at 2.9 Å. Geometry of the structure checked by a Ramachandran plot with use of PROCHECK (27) showed 80.4% of protein residues in the most favored regions and no residues in disallowed regions. The statistics of data collection and structure refinement are in Table 2.

Table 2.

Data collection and refinement statistics

Crystals	Myb353–180/DNA
Data collection
Space group	P1
Cell dimensions
a (Å)	45.81
b (Å)	71.77
c (Å)	87.82
α (o)	94.68
β (o)	97.84
r (o)	99.28
Wavelength (Å)	1
Resolution (Å)	30–2.9 (3.0–2.9)a
Number of total reflections	88 446
Number of unique reflections	23 110
Completeness (%)	95.5 (87.5)
Redundancy	3.8 (3.5)
Rmerge (%)b	4.6 (22.7)
I/σ (I)	29.46 (5.02)
Refinement
Number of reflections	19 482
Rwork (%)	21.8
Rfree (%)c	27.3
Geometry deviations
Bond length (Å)	0.009
Bond angles (o)	1.6
Mean B-values (Å) / No.
Protein atoms	63.01/3929
DNA	59.82/2600
Water molecules	36.79/98
Ramachandran plot (%)
Most favored	80.4
Additionally allowed	19.6

aValues for the highest resolution shell are shown in parentheses.

bRmerge =Σ|I – | /ΣI, where I = observed intensity and  = average intensity from multiple observations f symmetry related reflections.

cRfree was calculated on the basis of 5% of the total number of reflections randomly omitted from the refinement.

SPR

The real-time sensorgrams of tvMyb353–180 and several mutants interacting with a 21-bp MRE-1 DNA duplex were measured at 25°C on a BIAcore 3000 instrument (GE Healthcare). The 5′-biotinylated MRE-1 DNA duplex dissolved in PBS buffer with a concentration of 0.02 µM was immobilized to the streptavidin SA sensor chip at a flow rate of 10 µl/min. Protein samples were injected into the SA sensor chip immobilized with DNA at a constant flow rate of 30 µl/min for 3 min for association and then the running buffer containing 10 mM HEPES, 150 mM NaCl, 3.4 mM EDTA, pH 7.4, 0.05% (v/v) Tween-20 and 1 mM dithiothreitol was applied at the same rate for 3 min for dissociation. After each injection, the sensor chip was regenerated with 1 M NaCl. For analyzing the relative protein binding, the sensorgrams were evaluated by use of BIAevaluation v4.1 (BIAcore AB, Uppsala, Sweden). All values of the relative binding were the averages from three to five independent experiments.

NMR spectroscopy

NMR samples (unlabeled or 15N, 15N/13C and 15N/13C/2H isotopically labeled proteins) at pH 6.0 in phosphate buffer were prepared. Each sample contained ~0.30 ml of 1–2 mM protein in a Shigemi NMR tube. All NMR spectra were acquired with the use of Bruker AVANCE 600 or 800 MHz spectrometers equipped with a triple (1H, 13C and 15N) resonance cryoprobe, which included a shielded z-gradient at 298 or 310 K. Traditional triple-resonance experiments [HNCO, HN(CA)CO, CBCA(CO)NH and HNCACB] and transverse relaxation-optimized spectroscopy-based triple-resonance experiments were used for backbone resonance assignments of free tvMyb353–180 and tvMyb353–180 in complex with DNA, respectively. Side-chain resonance assignment followed previously published procedures (28). Nuclear Overhauser effect (NOE) data were extracted from 3D 1H-15N NOE spectroscopy (NOESY)-heteronuclear single-quantum coherence (HSQC) spectra with 120 ms mixing time. The weighted chemical shift perturbations (CSPs) for backbone 15N and 1HN resonances were calculated by the following equation: Δδ = [(ΔδHN)2 + (ΔδN/5)2]0.5. 1H-15N steady-state heteronuclear NOE (HX-NOE) experiments involved a 5 ms separated train of 120° 1H saturation pulses and a recycle delay of 5 s (29). To measure residual dipolar couplings (RDCs) of tvMyb353–180 in complex with DNA, the filamentous bacteriophage Pf1 (Asla Biotech Ltd, Latvia) was selected as the orienting medium. 2D 1H-coupled (F1) IPAP 1H-15N HSQC spectra (30) were acquired with 256 complex t1 (15 N) points and 64 scans per t1 increment for both the isotropic and anisotropic samples. The quality factor of RDC values equals RMS(Dcalc − Dobs)/RMS(Dobs), where Dcalc and Dobs are calculated and observed RDC values, respectively. Processing of all NMR spectra involved use of Bruker Topspin 2.0 (Karlsruhe, Germany) or NMRPipe (31), and data were analyzed by use of NMRView (32).

Paramagnetic relaxation enhancement measurements

To observe the intermolecular paramagnetic relaxation enhancement (PRE) effects on tvMyb353–180, an EDTA was conjugated with deoxythymidine (dT-EDTA) and introduced into DNA as described (33). The DNA fragments with dT-EDTA were purchased from Yao-Hong Biotechnology Inc. (Taiwan). Purity and authenticity of individual strands of DNA were verified by chromatography and mass analysis. The double-stranded DNA was prepared by mixing equal amounts of two complementary deoxynucleotides, heating to 95°C for 10 min and cooling slowly to room temperature. The EDTA-conjugated DNA duplex was mixed with equal amounts of metal ions (Mn2+ or Ca2+), and tvMyb353–180 (0.2 mM) was added into the mixture at a protein–DNA ratio of 1:1.5. PRE 1HN-Γ2 measurement involved 2D 1H-15N correlated spectra with a two time-point delay T of 10 ms as described (34). Values of 1HN-Γ2 were calculated from peak intensities for the diamagnetic (chelated with Ca2+) and paramagnetic (chelated with Mn2+) state as described (34).

RESULTS

Identification of the DBD of tvMyb3

Previously, we mapped the DBD of tvMyb1 as the fragment from residue 35 to 141 and determined its solution structure as well as its complex model with MRE-1 DNA (22). To identify the tvMyb3 DBD, we generated two protein fragments and examined their binding interactions with the 16-bp DNA containing the MRE-1 of 5′-ATAACGAT-3′ or MRE-2f of 5′-GTATCGTC-3′ by EMSA. The shorter fragment, tvMyb353–161, which includes similar R2R3 motifs as tvMyb135–141 (Figure 1), is not sufficient for binding to both DNAs (Figure 2A). Interestingly, tvMyb353–180 that contains 19 additional C-terminal amino acids and displays similar helical conformation and slightly higher thermal stability than the shorter tvMyb3 (Supplementary Figure S1), exhibits the DNA-binding ability.

Figure 2.

DNA-binding activity of tvMyb135–141, tvMyb353–161 and tvMyb353–180 probed by EMSA and ITC. (A) EMSA experiments were performed using 16-bp oligonucleotides. Three pico mole Myb proteins were incubated with a γ-32P-labeled MRE-1 DNA. Three pico mole tvMyb135-141 and 30 pmol tvMyb3 proteins were incubated with a γ-32P-labeled MRE-2f DNA. All protein–DNA mixtures were separated by 10% PAGE. (B) ITC thermograms for sequential 2-µl additions of 16 bp MRE-1 DNA (300 µM) to tvMyb proteins (30 µM). The lines were fitted to the integrated heats using a single-site binding model.

DNA-binding activity of tvMyb135–141, tvMyb353–161 and tvMyb353–180 probed by EMSA and ITC. (A) EMSA experiments were performed using 16-bp oligonucleotides. Three pico mole Myb proteins were incubated with a γ-32P-labeled MRE-1 DNA. Three pico mole tvMyb135-141 and 30 pmol tvMyb3 proteins were incubated with a γ-32P-labeled MRE-2f DNA. All protein–DNA mixtures were separated by 10% PAGE. (B) ITC thermograms for sequential 2-µl additions of 16 bp MRE-1 DNA (300 µM) to tvMyb proteins (30 µM). The lines were fitted to the integrated heats using a single-site binding model.

To confirm the EMSA results and to quantitate the DNA-binding affinity of the two tvMyb3 fragments, we monitored the protein–DNA interactions at 150 mM NaCl by ITC (Figure 2B). Fitting of ITC thermograms showed that tvMyb3 binds with 1:1 stoichiometry to the specific MRE-1 or MRE-2f DNA. The formation of the tvMyb3/MRE-1 complex is enthalpically driven and that of the tvMyb3/MRE-2f complex entropically driven. The association constant (Ka) value of tvMyb353–180 bound to 16-bp MRE-1 was 1.51 ± 0.20 × 107 M−1 and was ~10 times weaker for MRE-2f DNA (Table 1). Also, the shorter fragment bound to MRE-1 DNA ~40 times weaker than the longer tvMyb3, and the interaction with MRE-2f DNA was too weak to be determined by ITC, which agrees with the EMSA results. Interestingly, with thymine replaced at position 8 by adenine (MRE-2f_T8A), which changes the sequence to 5′-GTC-3′ (the mutated base is underlined and the conserved bases with MRE-1 are in bold italics), the Ka value of tvMyb353–180 bound to MRE-2f_T8A was similar to that of the tvMyb353–180/MRE-1 complex formation, which suggests that tvMyb353–180 recognizes the sequence 5′-TAACG-3′ specifically. The longer fragment, tvMyb353–180, and the MRE-1 DNA were hence selected for complex structure study.

Table 1.

Thermodynamic parameters of Myb DBDs binding to DNAs

Sample	n	Ka (M−1)	ΔG (kcal mol−1)	ΔH (kcal mol−1)	ΔS (cal mol−1 K−1)
MRE-1 (16 bp)
tvMyb135–141	1.01 ± 0.08	1.48 (±0.45) × 108	−11.14	−24.18 ± 0.09	−43.77 ± 0.55
tvMyb353–161	0.98 ± 0.05	3.40 (±0.43) × 105	−7.54	−8.75 ± 0.07	−4.07 ± 0.02
tvMyb353–180	0.99 ± 0.05	1.51 (±0.20) × 107	−9.79	−11.41 ± 0.04	−5.44 ± 0.23
AAAA	1.10 ± 0.01	2.86 (±0.46) × 106	−8.80	−10.85 ± 0.14	−6.87
MRE-1 (21 bp)
tvMyb353–180	1.09 ± 0.00	6.02 (±0.96) × 107	−10.62	−13.00 ± 0.05	−7.99
S169D	1.01 ± 0.00	2.67 (±0.41) × 107	−10.14	−12.93 ± 0.08	−9.38
AAAA	1.01 ± 0.00	5.12 (±0.41) × 106	−9.16	−13.21 ± 0.07	−13.60
MRE-2f_T8A (16 bp)
tvMyb353–180	1.05 ± 0.03	1.17 (±0.59) × 107	−9.60	−6.69 ± 0.08	9.77 ± 1.32
MRE-2f (16 bp)
tvMyb135–141	1.10 ± 0.09	1.17 (±0.14) × 108	−11.01	−23.24 ± 0.24	−41.03 ± 0.96
tvMyb353–161	NDb	NDb	NDb	NDb	NDb
tvMyb353-180	1.03 ± 0.13	1.39 (±0.32) × 106	−8.387	−3.49 ± 0.22	16.4 ± 0.95

aΔG = −RT·lnKa, R = 1.985 cal·K−1 mol−1, gas constant.

bND, not determined due to the small ΔH.

Thermodynamic parameters of Myb DBDs binding to DNAs

aΔG = −RT·lnKa, R = 1.985 cal·K−1 mol−1, gas constant.

bND, not determined due to the small ΔH.

Crystal structure of tvMyb353–180 bound to MRE-1 DNA

The overall structures of tvMyb353–180 bound to a 16-bp MRE-1 DNA were determined at 2.9 Å (Table 2) by X-ray crystallography. The asymmetric unit of the crystal contains four protein molecules in complex with four DNA duplexes (Figure 3A). The N-terminal parts, from residues Lys53 to Ser151, of all four protein structures revealed the classical R2R3 Myb DBD, with the third helices of R2 and R3 motifs intercalating in the major groove of DNA. Interestingly, residues from Ile154 to Leu164 form a β-hairpin conformation, which was not previously observed in Myb proteins. Additionally, the electron density was not observed for the C-terminal 10 or 11 residues, which suggests that they are flexible in the crystals. The superimposition of four protein structures for all backbone atoms shows a high degree of structural similarity, with RMSD of 0.47 ± 0.05 Å for backbone atoms. However, fitting only the backbone atoms of the R2 and R3 motifs (Lys53 to Ser151) showed that the RMSD for this region can be lowered to 0.35 ± 0.03 Å, but the deviation of the C-terminal residues (Lys152 to Ser169) was increased to 0.95 ± 0.19 Å (Supplementary Figure S2), which implies that the structure of the C-terminal fragment in the complex state is relatively more flexible than that of R2 and R3 motifs.

Figure 3.

Crystal structures and interactions of tvMyb353–180 in complex with 16-bp MRE-1 DNA. (A) There are four proteins in complex with four DNA duplexes in the asymmetric unit of the crystal. (B) The representative complex structure, in which, tvMyb353–180 consists of six α-helices, a β-hairpin and a C-terminal tail. The electron density for the C-terminal 10 residues was not observed, and the residues from Glu162 to Lys170 are colored in red. (C) The hydrophobic, electrostatic and hydrogen-bonding interactions between the R3 motif and C-terminal β-hairpin. The involved residues in the C-terminal β-hairpin are labeled. Two O2P atoms of ADE4 and THY5 are shown. The salt bridges are linked with dashed lines in magenta and hydrogen bond in blue. (D) Schematic representation illustrating the contacts between tvMyb353–180 and MRE-1 DNA. The amino acid residues in R2 and R3 motifs are shown with green and blue background colors, respectively. The van der Waals contacts and hydrogen-bonding interactions are indicated by orange and black dotted lines, respectively. The specific AACG base pairs are colored in pink. The phosphates are indicated in yellow circles.

Data collection and refinement statistics

aValues for the highest resolution shell are shown in parentheses.

bRmerge =Σ|I – | /ΣI, where I = observed intensity and  = average intensity from multiple observations f symmetry related reflections.

cRfree was calculated on the basis of 5% of the total number of reflections randomly omitted from the refinement.

Crystal structures and interactions of tvMyb353–180 in complex with 16-bp MRE-1 DNA. (A) There are four proteins in complex with four DNA duplexes in the asymmetric unit of the crystal. (B) The representative complex structure, in which, tvMyb353–180 consists of six α-helices, a β-hairpin and a C-terminal tail. The electron density for the C-terminal 10 residues was not observed, and the residues from Glu162 to Lys170 are colored in red. (C) The hydrophobic, electrostatic and hydrogen-bonding interactions between the R3 motif and C-terminal β-hairpin. The involved residues in the C-terminal β-hairpin are labeled. Two O2P atoms of ADE4 and THY5 are shown. The salt bridges are linked with dashed lines in magenta and hydrogen bond in blue. (D) Schematic representation illustrating the contacts between tvMyb353–180 and MRE-1 DNA. The amino acid residues in R2 and R3 motifs are shown with green and blue background colors, respectively. The van der Waals contacts and hydrogen-bonding interactions are indicated by orange and black dotted lines, respectively. The specific AACG base pairs are colored in pink. The phosphates are indicated in yellow circles.

To select a representative complex structure, we determined the one-bond RDC constant, 1DNH, of tvMyb353–180 in complex with DNA by NMR spectroscopy in solution. With the four crystal structures used as templates, the theoretical 1DNH RDC values can be calculated. For the residues in secondary structural regions, the experimental RDC values fit well to the theoretical values from four protein structures with quality factors between 0.37 and 0.43 (Supplementary Figure S3). However, the experimental RDC values of all residues fit best to the values from the fourth complex structure with a quality factor of 0.65, which suggests that the fourth complex structure in crystalline state is similar to that in solution state.

In the fourth complex structure, tvMyb353–180 consists of six α-helices, a β-hairpin and a C-terminal tail (Figure 3B), with α1 (Glu58 to Asn71), α2 (Trp76 to Ser81), α3 (Pro88 to Leu99), α4 (Pro110 to Lys122), α5 (Trp127 to Leu133), α6 (Asp139 to Ser151), β1 (Ile154 to Thr156) and β2 (Glu162 to Leu164). The three-helix bundles of R2 and R3 motifs are stabilized by hydrophobic and electrostatic interactions. Two hydrogen bonds (Val104 and Gly136, Asn97 and Thr138) and 1 van der Waals interaction between the side chains of Val104 and Arg137 span the R2 and R3 motifs. Notably, the C-terminal fragment makes several contacts with the R3 motif (Figure 3C). Leu164 on β2 takes part in the formation of the R3 hydrophobic core. Arg153 and Asp167 form a salt bridge network with Asp113 on α4, Arg145 on α6 and O2P of THY5. In addition, O′ of Glu162 on β2 constitutes a hydrogen bond with Hη12 of Arg118 on α4. The C-terminal β-hairpin conformation seems to be stabilized by these interactions.

Structural basis for MRE-1/MRE-2 r recognition by tvMyb353–180

In the representative tvMyb353–180/DNA complex structure, the recognition helices (helices 3 and 6) insert into the major groove of the MRE-1 DNA duplex. All interactions between protein and DNA are illustrated in Figure 3D. The side chains of Lys89, Glu93 in helix 3 and Asn140, Lys143 and Asn144 in helix 6 form specific hydrogen bonds with the bases of GUA9, CYT8, ADE7, GUA8′ and ADE6, respectively (Supplementary Figure S4). Moreover, several residues, including Gln74, Gln90, Arg92, Arg94 and His98 in the R2 motif and Trp108, Ser125, Trp127 and Ser128 in the R3 motif contact with the phosphate groups of the DNA backbone in the major groove. Additionally, the backbone atoms of Gly54, Pro55 and Phe56 interact with the phosphate groups of ADE7 and CYT8 in the minor groove. However, there are no direct contacts between DNA and the β-hairpin or the C-terminal tail. The nearest distance is 4.42 Å between Nη1 of Arg168 and O1P of GUA3.

In the preceding result of identification of the tvMyb3 DBD, the DNA-binding affinity of tvMyb353–180 is 40 times stronger than that of tvMyb353–161. Therefore, the observation of no interaction between DNA and the C-terminal residues of tvMyb353–180 in the X-ray structure is a surprise. The roles of the C-terminal fragment, including the β-hairpin, the C-terminal tail and the missing residues, in DNA binding need further investigation as described below.

NMR investigations of the C-terminal fragment

We observed no electron density for the C-terminal 10 residues ‘K171RKAADVPKK180’ in the crystal structure of the tvMyb353–180/DNA complex, which may due to its high flexibility. However, the DNA length in the crystal may be too short for contacting with the last 10 residues. Hence, in addition to the 16-bp DNA, 21-, 25- and 30-bp DNAs were titrated into 15N-labeled tvMyb353–180 to characterize their interactions by NMR spectroscopy. 2D 1H-15N HSQC spectra showed backbone amide resonances of several residues in complex with 21-bp DNA significantly different from those in complex with 16-bp DNA (Figure 4A, left panel), and longer DNA did not cause substantial further perturbations in these residues (Figure 4A, right panel), which suggests that the 21-bp MRE-1 sequence is suitable for tvMyb353–180/DNA complex formation.

Figure 4.

NMR investigations on tvMyb353–180/DNA complexes. (A) The overlaid 2D 1H-15N HSQC spectra of tvMyb353–180 in complex with 16-bp (red, left panel), 21-bp (blue, left and right panels) and 25-bp (purple, right panel) MRE-1 DNAs. (B) The overlay of the 1H-15N HSQC spectra of free tvMyb353–180 (black) and in complex with 21-bp MRE-1 DNA (blue) at 310 K. The residues with large chemical shifts are indicated. (C) Weighted CSPs for backbone 15N and 1HN resonances of tvMyb353–180 in complex with 16-bp (red) or 21-bp (blue) DNA as calculated by the following equation: Δδ = [(ΔδHN)2 + (ΔδN/5)2]0.5. (D) 1H-15N heteronuclear NOEs of tvMyb353-180 in free form (black circles), and in complex with 16-bp (red circles) or 21-bp (blue circles) MRE-1 DNAs. The secondary structures are highlighted in (C and D).

NMR investigations on tvMyb353–180/DNA complexes. (A) The overlaid 2D 1H-15N HSQC spectra of tvMyb353–180 in complex with 16-bp (red, left panel), 21-bp (blue, left and right panels) and 25-bp (purple, right panel) MRE-1 DNAs. (B) The overlay of the 1H-15N HSQC spectra of free tvMyb353–180 (black) and in complex with 21-bp MRE-1 DNA (blue) at 310 K. The residues with large chemical shifts are indicated. (C) Weighted CSPs for backbone 15N and 1HN resonances of tvMyb353–180 in complex with 16-bp (red) or 21-bp (blue) DNA as calculated by the following equation: Δδ = [(ΔδHN)2 + (ΔδN/5)2]0.5. (D) 1H-15N heteronuclear NOEs of tvMyb353-180 in free form (black circles), and in complex with 16-bp (red circles) or 21-bp (blue circles) MRE-1 DNAs. The secondary structures are highlighted in (C and D).

The backbone resonance assignments of tvMyb353–180 in the free state and in complex with 16- and 21-bp DNAs were achieved by several 3D triple-resonance experiments (Supplementary Figure S5). Backbone resonances of residues in the R2 and R3 motifs (Lys53 to Ser151) undergo similar CSPs in complex with 16- and 21-bp DNAs (Figure 4B). Most of the largely perturbed residues (Δδ> Δδaverage + SD = ~0.5 ppm) in the R2R3 fragment are located at the N terminus, the loop connecting helixes 1–3, the loop connecting R2 and R3 motifs, and helix 6 (Figure 4C). One additional strong CSP was observed for Ser128 located in helix 5. For the residues located at the C-terminal fragment (Lys152 to Lys180), the complex sample with 21-bp DNA shows significantly larger CSP values than those with 16-bp DNA, which suggests that these residues have stronger interactions with the longer DNA than the shorter one.

We then checked the backbone rigidity of tvMyb353–180 in the absence and presence of DNA by 1H-15N HX-NOE measurements (Figure 4D). HX-NOE values of almost all tvMyb353–180 residues were significantly increased in the presence of DNA; especially the N- (Lys53 to Thr57) and C-terminal residues (Ile154 to Lys180), which are highly flexible in the absence of DNA, adopt a more rigid conformation on binding to DNA. The average HX-NOE values of the R2 and R3 motifs (Lys53 to Ser151) in the absence and presence of 16- and 21-bp DNAs were 0.78 ± 0.08, 0.84 ± 0.07 and 0.85 ± 0.08, respectively. However, the average values of the C-terminal fragment (Lys152 to Lys180) were increased from 0.40 ± 0.26 for the free state to 0.52 ± 0.18 for the complex with 16-bp DNA and 0.57 ± 0.18 for binding to 21-bp DNA. Although the C-terminal fragment of tvMyb353–180 was still flexible, the amide pico-second to nano-second time scale motions indeed are stabilized by the extension of DNA length.

The results from NMR CSPs and backbone HX-NOE values reveal that the interactions between DNA and the C-terminal fragment, including the last 10 residues missing in the X-ray structure, were increased when the DNA length was extended from 16 to 21 bp. So we tried to crystallize the tvMyb353–180/21-bp MRE-1 DNA complex sample, but the resolution of resulting crystals was too low to determine the structure. Instead, we used intermolecular PRE to investigate the interactions between DNA and the C-terminal region of tvMyb353–180. We purchased the 21-bp DNA duplex containing the EDTA-derivatized deoxythymidine (dT-EDTA) at the desired site. Values of PRE of amide protons (1HN-Γ2) for tvMyb353–180 were determined from a two time-point measurement in both the paramagnetic (chelated with Mn2+) and diamagnetic (chelated with Ca2+) states (33,34). The residues exhibiting 1HN-Γ2 > 5 s−1 are located at the N terminus, the loop connecting the R2 and R3 motifs, helix 6 and the C-terminal tail (Figure 5A). A structural model of tvMyb353–180 in complex with a 21-bp DNA duplex was generated by extending the 16-bp DNA in the crystal structure with a standard B-form conformation. We mapped the intermolecular PRE 1HN-Γ2 values on the structural model (Figure 5B), and found that most of the PRE 1HN-Γ2 values correlated well with the distances measured in the model structure, with 1HN-Γ2 between 14 and 10 s−1 corresponding to the distance from 18 to 25 Å and 1HN-Γ2 between 10 and 5 s−1 corresponding to the distance from 24 to 29 Å (35). The C-terminal residue Arg168 exhibits the largest 1HN-Γ2 value of 13.3 s−1, which suggests that it is close to the DNA although it does not form any direct contact with DNA in the crystal structure. Also, the PRE values of the C-terminal 12 residues, like Ser169 and Lys170, which should be in regions closest to Mn2+ if they formed stable contacts with the extended DNA, are all >5 s−1 but smaller than that of Arg168, which suggests that these residues are close to DNA but their mobility reduces the observed PRE values.

Figure 5.

Intermolecular PRE for tvMyb353–180/DNA complex and DNA binding activities of tvMyb353–180 mutants. (A) EDTA labeled site on 21-bp MRE-1 DNA (in red) and intermolecular PRE 1HN-Γ2 values obtained with EDTA chelating with Mn2+. (B) The intermolecular PRE values are mapped on a structural model of tvMyb353–180 in complex with a 21-bp DNA duplex, which is generated by extending the 16-bp DNA in the crystal structure with standard B-form conformation. EDTA and Mn2+ are shown in magenta. The residues exhibiting 1HN-Γ2 values >10 s−1 are in red, 5–10 s−1 in orange and <5 s−1 in blue. The missing C-terminal 10 residues are indicated with one-letter abbreviations. (C) DNA binding activities of tvMyb353–180 mutants probed with SPR. The relative DNA binding activities of mutants are calculated from SPR responses during 15 s before the end of protein injection.

Intermolecular PRE for tvMyb353–180/DNA complex and DNA binding activities of tvMyb353–180 mutants. (A) EDTA labeled site on 21-bp MRE-1 DNA (in red) and intermolecular PRE 1HN-Γ2 values obtained with EDTA chelating with Mn2+. (B) The intermolecular PRE values are mapped on a structural model of tvMyb353–180 in complex with a 21-bp DNA duplex, which is generated by extending the 16-bp DNA in the crystal structure with standard B-form conformation. EDTA and Mn2+ are shown in magenta. The residues exhibiting 1HN-Γ2 values >10 s−1 are in red, 5–10 s−1 in orange and <5 s−1 in blue. The missing C-terminal 10 residues are indicated with one-letter abbreviations. (C) DNA binding activities of tvMyb353–180 mutants probed with SPR. The relative DNA binding activities of mutants are calculated from SPR responses during 15 s before the end of protein injection.

NMR studies shows that the C-terminal extension is flexible but sensitive to the length of interacted DNA, suggesting that it is involved in DNA binding. The increased intermolecular PRE values of C-terminal missing residues further imply that these residues have transient interactions with DNA.

Mutations on the C-terminal fragment

From the CSPs, backbone HX-NOE and intermolecular PRE, the C-terminal fragment (Lys152 to Lys180) was found to be highly involved in DNA binding, although it does not form stable contacts with DNA in the crystal structure. To further validate the roles of the C-terminal residues in DNA binding, we generated various mutants of tvMyb353–180 and monitored their interactions with DNA by SPR at 150 mM NaCl (Figure 5C and Supplementary Figure S6). The mutants R153A and L164A, which disrupt the electrostatic and hydrophobic interactions between the β-hairpin and R3 motif, show a significant decrease in DNA binding ability. Interestingly, the mutants that change one positively charged residue at the C terminus to alanine bind to DNA as strongly as does the wild-type protein. However, the mutants AA1 (K170A, K171A) and AA2 (R172A, K173A) show notable decreases, and AAAA (K170A, K171A, R172A, K173A) exhibits severely reduced DNA binding activity.

The association constants (Ka) of the AAAA mutant bound to 16- and 21-bp MRE-1 DNAs at 150 mM NaCl were measured by ITC (Table 1). For 21-bp DNA, the affinity between DNA and the AAAA mutant (5.12 ± 0.41 × 106 M−1) was ~12 times weaker than that of the wild-type protein (60.2 ± 9.6 × 106 M−1). For 16-bp DNA, the reduction in association constant was ~5-fold for the mutant protein. In addition, the Ka value of the S169D mutant bound to 21-bp DNA was about half of that of the wild-type protein (Table 1). The interactions between DNA and S169D, AA1, AA2 or AAAA mutants were all weaker than that between DNA and wild-type protein, so acetylation at the Lys residue or phosphorylation at the Ser residue in the C-terminal fragment will lower the binding affinity between tvMyb353–180 and DNA.

Thus, the structural integrity of the R3 motif and the β-hairpin and the positively charged residues at the C terminus are critical for DNA binding.

The β-hairpin conformation is preserved in free tvMyb353–180

According to 3D structure comparison by the Dali server, the C-terminal β-hairpin conformation of tvMyb353–180 is novel as compared with other Myb-like DBD. Hence, determining whether this unique structure is formed only in the DNA-bound form or is preserved in the free form is of interest. From backbone HX-NOE experiments of free tvMyb353–180 (Figure 4D), the residues in the β-strands, especially Leu164 and Leu165, exhibit backbone HX-NOE values larger than those of N- and C-terminal residues, which suggests that this region adopts a conformation more stable than a random coil. As well, 15N-edited NOESY-HSQC spectra revealed several long-range NOEs between these C-terminal residues (Arg153 to Leu165), like the NOEs between Hα of Thr156 and HN of Lys161–Ile163 (Figure 6A). The backbone NOE connectivity among these residues suggests that this fragment adopts a hairpin conformation in free tvMyb353–180 (Figure 6B).

Figure 6.

The C-terminal fragment of tvMyb353–180 forms a β-hairpin conformation without DNA. (A) Strip plots from 15N-edited NOESY-HSQC spectrum of free tvMyb353–180, in which NOE cross-peaks between Hα of Thr156 and HN of Asn157–Ile163 are connected by a black line. (B) NOE connectivity of residues from Arg153 to Leu165. Strong, medium and weak NOEs are indicated by solid-thick, solid and solid-thin arrows, respectively. The dashed arrows indicate the ambiguous NOEs, which cannot be clearly identified because of overlapping.

The C-terminal fragment of tvMyb353–180 forms a β-hairpin conformation without DNA. (A) Strip plots from 15N-edited NOESY-HSQC spectrum of free tvMyb353–180, in which NOE cross-peaks between Hα of Thr156 and HN of Asn157–Ile163 are connected by a black line. (B) NOE connectivity of residues from Arg153 to Leu165. Strong, medium and weak NOEs are indicated by solid-thick, solid and solid-thin arrows, respectively. The dashed arrows indicate the ambiguous NOEs, which cannot be clearly identified because of overlapping.

DISCUSSION

The C-terminal β-hairpin extension is unique and crucial for MRE-1 DNA recognition

Myb protein, a member of the transcription factor family, represents a functional DBD that is conserved in a wide range of eukaryotic systems. c-Myb protein was first identified in vertebrates and found to contain a three-repeat R1R2R3 DBD. Each repeat of c-Myb forms a three-helix bundle conformation, but only the third helices of R2 and R3 motifs recognize specific DNA in the major groove (13). In plants, most of the double-stranded telomere-binding proteins contain only one Myb-like domain. However, this Myb-like domain alone is not sufficient for telomere binding. A C-terminal extension was found to be highly involved in DNA binding (15,16). Structural studies revealed that instead of adopting the typical fold of Myb DBD, the Myb-like domain of plant double-stranded telomere-binding protein folds into a four-helix tetrahedron structure, and the fourth helix, which is formed by the C-terminal extension, participates in the formation of a hydrophobic core and stabilizes the whole DBD structure for telomere binding (17–19). We determined the structure of tvMyb353–180 in complex with MRE-1 DNA and showed that the C-terminal extension forms a β-hairpin conformation, which was demonstrated by NMR to participate in DNA binding and by site-directed mutagenesis to be crucial for DNA binding. Also, NMR studies indicate that the C-terminal β-hairpin conformation is formed in the DNA-bound state as well as in free protein. The unique features of this C-terminal β-hairpin conformation are described below.

First, like the fourth helix in the Myb-like domain of plant double-stranded telomere-binding protein, the C-terminal β-hairpin conformation of tvMyb353–180, although not making any direct contact with DNA, plays a role in stabilizing the overall structure and is critical for DNA binding. This observation was confirmed by determining the thermal stability, whereby tvMyb353–180 is 3°C more stable than the shorter fragment tvMyb353–161, and by the analysis of the mutants R153A and L164A, which disrupt the electrostatic and hydrophobic interactions between the β-hairpin and R3 motif, showing a significant decrease in DNA binding ability. The one exception is the mutant D167A, which, although breaking the salt bridge networks among Asp167, Arg153 and Arg145 (Figure 3C), showed an increase in DNA binding activity. The mutant D167A may reduce the repulsion forces between Oδ1 of Asp167 and O2P of ADE4, which is 4.4-Å apart in the crystal structure, and the salt bridge networks can be supplemented by the interactions between Nη2 of Arg145 and O2P of THY5, which is 4.1-Å apart in the crystal structure.

Second, the C-terminal tail, which is mostly not observed in the complex structure, is also important for DNA binding, which suggests that the DNA recognition by tvMyb3 may be regulated by post-translational modification at these residues. Our NMR studies demonstrated that the C-terminal tail, although flexible, has some interactions with the extended DNA. As well, the AA1, AA2 and AAAA mutants, which replaced the positively charged residues in the C-terminal tail to alanine and S169D, which introduces a negative charge to mimic phosphorylation on Ser residue, all show a significant decrease in DNA binding affinity. In the complex structure, the third helix of the R3 motif contacts with DNA, and the C-terminal basic residues are restricted by the β-hairpin to be in a position next to the third helix of the R3 motif, where these basic side chains may have transient interactions with the DNA phosphate backbone or the minor groove. According to this view, the interaction mode shows similarity with the C-terminal fragment of another DBD, the winged helix or the forkhead box (termed FOX). The FOX motif consists of two wings (W1 and W2), three α-helices (α1, α2 and α3) and three β-strands (β1, β2 and β3), arranged in order of α1–β1–α2–α3–β2–W1–β3–W2. Like the C-terminal half of tvMyb353–180, the second and third helices of FOX adopt a helix-turn-helix conformation, β2 and β3 forms an anti-parallel β-sheet and the loop or wing W2 extends from β3 to the C terminus of the DBD. In DNA recognition by canonical FOX proteins, the third helix makes direct contact with DNA in the major groove and the wing W1 interacts with DNA in the minor groove (36). In the FOXO3a/DNA complex structure, the C-terminal W2 basic residues forms hydrogen bonds with the DNA phosphate backbone (37). For FOXO1 protein, the flexible C-terminal fragment is not observed in the complex structure (38). However, the C-terminal coil is necessary for DNA binding in both FOXO3a and FOXO1 proteins, and their DNA binding activities are attenuated by phosphorylation of Ser or acetylation of Lys residues in the C-terminal coil (37,38). According to our study of tvMyb3 and its similarity with FOX protein, MRE-1 element recognition by tvMyb3 protein may be regulated by phosphorylation or acetylation of the C-terminal residues.

Structural comparison of the R2R3 DBD with other Myb proteins

The tvMyb3 protein exhibits high amino acid sequence identity with human c-Myb and tvMyb1 in the R2R3 DBD (Figure 1). The structure of c-Myb DBD in complex with DNA has been reported (13,39). In our previous study, we determined the solution structure of free tvMyb135–141 and proposed the tvMyb135–141/MRE-1 DNA complex model using CSPs, residual dipolar couplings, DNA specificity data and data-driven macromolecular docking by HADDOCK (22). We found that the orientation between R2 and R3 motifs of tvMyb1 greatly changes on binding to DNA to recognize the DNA major groove through a number of contacts involving residues in two recognition helices. A superimposition of the six helices from tvMyb3 R2R3 DBD with those from tvMyb1 and c-Myb showed high similarity in backbone structure, with RMSD of 0.91 and 1.79 Å for Cα atoms, respectively (Supplementary Figure S7), so the three R2R3 DBDs may adopt a similar architecture of protein backbone to recognize the DNA major groove.

For the residues in two recognition helices, tvMyb3 protein shows 83% amino acid sequence identity with human c-Myb. The R2R3 DBD of c-Myb recognizes the base sequence 5′-TAACNG-3′ specifically and that of tvMyb3 binds to the MRE-1 element containing 5′-TAACGA-3′ with high specificity. High similarities in both protein and DNA sequences suggest that the protein–DNA interactions of tvMyb3 and c-Myb are similar. Indeed, close examination of the two protein–DNA complex structures reveals that most of the residues identified for DNA recognition in tvMyb3 can also be found in the c-Myb/DNA complex (Figure 3D and Supplementary Figure S8A). One difference is that Lys89 in helix 3 forms specific hydrogen bonds with the bases of GUA9 in the tvMyb353–180/DNA complex structure, but the corresponding Lys residue (Lys128) in c-Myb does not contact with the corresponding GUA (GUA21) but rather with GUA22. Also, the residues Trp108, Gly54 and Gln74 in tvMyb3 contact with the DNA phosphate backbone, and these interactions are not conserved in the c-Myb/DNA complex.

The amino acid sequence identity for the residues in two recognition helices between tvMyb3 and tvMyb1 is only 57%. tvMyb1 recognizes the whole MRE-1 element containing 5′-TAACGAT-3′, whereas tvMyb3 binds only to 5′-AACG-3′. The examination of two protein–DNA interactions reveals that three specific hydrogen bonds with bases and four hydrogen bonds with a phosphate backbone in the tvMyb353–180/DNA complex structure can be found in the tvMyb135–141/DNA complex model (Figure 3D and Supplementary Figure S8B). However, two Lys residues in tvMyb3, Lys89 and Lys143, which recognize bases specifically, are Arg residues in tvMyb1 (Arg71 and Arg125) and form hydrogen bonds with bases of ADE10 and ADE11′ and the phosphate backbone of CYT9′, respectively, in the complex model. In addition, the specific interactions of Asn69 and Gln72 identified in the tvMyb135–141/DNA complex are not found in the tvMyb3 complex structure. The corresponding residue of Asn69 in tvMyb3 is Ser87, which does not recognize the base of ADE10, perhaps because of its shorter side chain. However, without the crystal structure of the tvMyb135–141/DNA complex, confirming the interactions observed in the model is not easy. The comparison of tvMyb1 and tvMyb3 complexes suggests that the strategy that combined the free form structure with CSPs, residual dipolar couplings, DNA specificity data and data-driven macromolecular docking can generate a model with reliable protein backbone architecture and some conserved protein–DNA interactions. This method is an alternative way to obtain the structural basis of protein–DNA interactions if the crystal structure is not available.

ACCESSION NUMBER

The atomic coordinates of tvMyb353–180/16-bp MRE-1 DNA complex were deposited in the Protein Data Bank under accession code 3zqc.

SUPPLEMENTARY DATA

Supplementary Data are available at NAR Online.

FUNDING

Academia Sinica and the National Science Council, Taiwan, ROC (NSC 95-2320-B-001-040-MY2 to C.C. and NSC98-2311-B-001-009-MY3 to C.-D.H.). Funding for open access charge: Academia Sinica.

Conflict of interest statement. None declared.