| Nucleic Acids Research | Pages |
Crystal structure of the MATa1/MAT[alpha]2 homeodomain heterodimer in complex with DNA containing an A-tract
Introduction
Materials And Methods
Crystallization
Structure determination and refinement
Biochemical studies
Results
Overview
DNA conformation
Conserved protein-DNA sequence recognition in the minor groove
Conserved waters at the protein-DNA interface
Minor groove spine of hydration
In vivo and in vitro studies of a1/[alpha]2 heterodimer binding to DNA
Discussion
Structure of the A-tract within the a1/[alpha]2 binding site
Recognition of bases in the minor groove
Role of solvent in mediating protein-DNA interactions
Acknowledgements
References
Crystal structure of the MATa1/MAT[alpha]2 homeodomain heterodimer in complex with DNA containing an A-tract
PDB accession no. 1AKH
ABSTRACT
INTRODUCTION
The MATa1 and MAT[alpha]2 proteins act in concert to bind DNA and regulate transcription in the yeast, Saccharomyces cerevisiae. In a/[alpha] cells, the a1 and [alpha]2 proteins form a heterodimer that binds to sites upstream of haploid-specific genes, causing repression of those genes and thereby determining the a/[alpha] diploid cell type (1,2 and references therein). Both a1 and [alpha]2 are members of the homeodomain superfamily of DNA binding proteins and contact the DNA with their respective homeodomains. Studies with protein fragments have shown that cooperative binding of a1 and [alpha]2 to DNA requires only the a1 homeodomain and a fragment of [alpha]2 consisting of the homeodomain plus a 21 amino acid C-terminal tail that has been shown to be critical for a1/[alpha]2 interactions (3,4). While the [alpha]2 protein alone binds DNA with modest specificity and affinity and the a1 protein binds DNA non-specifically, the a1/[alpha]2 heterodimer binds to its operator sites with a high degree of specificity and affinity (5). The specificity of the interaction of the a1/[alpha]2 heterodimer with its DNA binding sites derives both from the contacts formed between the respective proteins and the DNA and from the precise intrasite spacing requirements imposed by the interactions between the two homeodomain proteins (6).
Previously, we determined the 2.5 Å crystal structure of the a1/[alpha]2 homeodomain heterodimer bound to a 21 bp DNA fragment (7). In the structure of the ternary complex, the a1 and [alpha]2 homeodomains bind in tandem to one face of the DNA. Heterodimer formation is mediated by the C-terminal tail of [alpha]2, part of which folds to form a short amphipathic helix that binds in a hydrophobic patch between helices 1 and 2 of the a1 homeodomain. Both the a1 and [alpha]2 proteins fold into a compact three-helix domain containing a helix-turn-helix DNA binding motif (8) and use residues in the third, DNA recognition helix for specific DNA base recognition. The DNA binding specificity and affinity of the [alpha]2 protein and the a1/[alpha]2 heterodimer are further enhanced by interactions between residues in the N-terminal arm of the [alpha]2 protein and DNA bases in the minor groove. The N-terminal arm of the a1 homeodomain, in contrast, is disordered (7). In addition to the base contacts, there are extensive contacts with the DNA backbone that serve to anchor the respective homeodomains on the DNA.
A key feature of the a1/[alpha]2-DNA complex is the 60° bend induced in the DNA helix by binding of the a1/[alpha]2 heterodimer. The observed DNA bend, which is not found in the structure of the complex formed by the [alpha]2 homeodomain alone bound to DNA (9), brings the a1 and [alpha]2 proteins sufficiently close to one another to form protein-protein contacts. The protein-induced DNA bending observed in the crystal structure correlates well with solution studies, which show that [alpha]2 alone does not bend the DNA whereas the a1/[alpha]2 heterodimer does (10). The DNA bending is smooth and is accomplished without any significant local distortion or kinking of the B-DNA helix. The highest degree of bending is observed between the a1 and [alpha]2 binding sites, where the minor groove of the DNA is at its narrowest and the DNA sequence is AT-rich.
Bound water molecules play an important role both in mediating protein-DNA contacts and in stabilizing the bent conformation of the DNA. The protein side chain-DNA contacts observed in the major and minor grooves of the DNA consist both of direct contacts and of hydrogen bonds mediated by one or more water molecules. A number of important contacts with the DNA sugar-phosphate backbone are also mediated by bound waters. At the center of the DNA fragment, where the DNA bending that facilitates the interaction between a1 and [alpha]2 gives rise to a significantly narrowed minor groove, a spine of hydration is observed that is presumed to help stabilize the bent DNA conformation. The network of water molecules that comprise the spine of hydration also interacts with an N-terminal arm side chain that, in addition, participates in direct DNA base contacts. In all, there are 58 bound solvent molecules that appear to play important roles in the a1/[alpha]2-DNA complex.
In order to learn more about the effect of DNA sequence on the observed DNA bending by the a1/[alpha]2 heterodimer and to verify the importance of bound waters in the a1/[alpha]2-DNA complex, we have determined the structure of the a1/[alpha]2 heterodimer in complex with a 21 bp DNA fragment that differs at four base positions from the fragment used in the original structure determination (7). These base substitutions introduce a polyadenine tract (A-tract) of 5 bp in length into the center of the a1/[alpha]2 binding site. Repeated A-tracts of 5-6 bp in length have been shown to introduce bends in DNA when arranged in helical phase and interspersed with GC-rich sequences (11), although structural studies have shown the A-tracts themselves to be straight (12-14). Our results show that both the protein main chain and the DNA backbone conformations are similar in the two a1/[alpha]2-DNA complexes. Analysis of the A5-tract embedded within the 21 bp fragment shows that, while it shares some structural features with A-tract oligonucleotide structures (12,14), there is a greater degree of bending in the A-tract found in the a1/[alpha]2 binding site that is likely protein induced. In comparing the location of bound water molecules, we find that the spine of hydration in the narrowed minor groove and the water molecules mediating most of the protein-DNA contacts are highly conserved in the two structures, underscoring their importance in binding of the a1/[alpha]2 heterodimer to DNA.
MATERIALS AND METHODS
Crystallization
Purification of the a1 and [alpha]2 proteins, crystallization of the a1/[alpha]2-DNA complex and the method of flash-freezing the crystals have been described (15). In brief, a fragment of the a1 protein containing the C-terminal DNA binding domain residues 66-126 and a fragment of the [alpha]2 protein containing the DNA binding domain with the additional C-terminal tail required for heterodimerization with a1 (residues 128-210) were separately over-expressed in Escherichia coli and purified to >99% homogeneity by cation exchange and hydrophobic interaction chromatography. The DNA oligonucleotide used for crystallization was synthesized by the solid-phase phosphoramidite method on an Applied Biosystems DNA synthesizer and purified by HPLC on a reverse-phase Dynamax-300 PureDNA column (Rainin). The double-stranded oligonucleotide has an AT dinucleotide overhang at the 5[prime]-end of each strand that promotes stacking of the a1/[alpha]2-DNA complex in the crystal. The a1/[alpha]2-DNA ternary complex was crystallized using the hanging drop vapor diffusion method. Crystals were grown by mixing equal volumes of the protein-DNA solution and a precipitating solution containing 100 mM HEPES, pH 7.0, 20 mM CaCl2 and 5 mM Co(NH3)6Cl3. The crystals typically grow in 1 month and attain a maximum size of 0.8 × 0.6 × 0.2 mm. Since the crystals are subject to osmotic shock when soaked in cryoprotectant prior to freezing, crystals were first cross-linked with glutaraldehyde as previously described (15) for the original a1/[alpha]2-DNA ternary complex (7). The crystals were then transferred to a series of PEG 400 solutions of successively increasing concentrations, going from 8 to 35% PEG 400 in nine steps. The MATa1/[alpha]2-DNA ternary complex containing the A-tract DNA crystallizes isomorphously with the original crystals in space group P61 with unit cell dimensions of a = b = 132.2 Å and c = 45.25 Å and with one complex per asymmetric unit. A flash-frozen crystal maintained at -179°C diffracts X-rays to 2.8 Å along the c-axis and to 2.4 Å in perpendicular directions.
Structure determination and refinement
Data were collected at -179°C using a RAXIS-IIC image plate detector (Molecular Structure Corp., TX) mounted on a Rigaku RU-200 rotating anode X-ray generator equipped with a graphite monochromator using CuK[alpha] radiation. The data were processed using the UNIX version of the RAXIS data processing package; statistics are shown in Table 1.
The previously solved a1/[alpha]2-DNA ternary complex structure containing no solvent molecules and with the central four substituted DNA base pairs omitted was used as a starting model for phasing. The overall positional parameters were adjusted by rigid body refinement using XPLOR (16) in which the [alpha]2, a1 and DNA models were treated as separate fragments. A cycle of simulated annealing refinement was then carried out using a starting temperature of 1800 K, followed by 120 cycles of positional refinement. 2Fo - Fc and Fo - Fc electron density map calculations showed clear electron density for the four omitted base pairs. These nucleotides were then fitted into the electron density map. The side chains of amino acids Arg142 and Thr189 in [alpha]2 and Ser94 in a1, which had been modeled as alanines in the original structures, are ordered in the current structure and are included in the new model; the side chain of amino acid Glu139 in the [alpha]2 protein was disordered and was modeled as alanine. The side chain conformation of Lys77 in the [alpha]2 protein and Gln93, Lys98, Gln102 and Lys105 in the a1 protein have been modified as compared with the original structure. Four rounds of 120 cycles each of positional and individual isotropic temperature factor refinement did not show any misfit residues. The 2Fo - Fc andFo - Fc maps were examined at each stage of the refinement for satisfactory agreement using a Silicon Graphics Indigo2 workstation running the O graphics modeling program (17). Water molecules were included in five rounds of difference map calculation followed by assignment of waters to electron density peaks with [sigma] values >3.5[sigma]. Each assigned water molecule participates in at least two hydrogen bonds and has a temperature factor of <50 Å2. The newly assigned water molecules were subjected to 10 cycles of positional, temperature factor and occupancy refinement. In order to recheck for water molecules that appeared in the previous structure but not in the current one, we re-examined the difference map with successively lower [sigma] cut-offs until reaching a [sigma] value of 2.0; no further water molecules were located. The final model contains 51 water molecules and has an R factor of 20.1% and an Rfree (18) of 30.2% for the 13 030 reflections above 2[sigma](F) to 2.5 Å resolution. The average atomic B factor for the complex is 32.1 Å2. The average B factors for subsets of atoms in the complex are: 24.6 Å2 for the DNA, 36.4 Å2 for the a1/[alpha]2 heterodimer with the exclusion of the C-terminal tail of [alpha]2 and 58.3 Å2 for the C-terminal tail residues. The refinement statistics and final refined protein and DNA geometry are tabulated in Table 1.
Table 1.
| Resolution | 2.4 Å | ||
| Measured reflections | 31 013 | ||
| Unique reflections | 15 656 | ||
| Completeness | 83.6% | ||
| Rsym | 6.6% | ||
| I/[sigma](I) (outer shell) | 10.5 (2.5) | ||
| Refinement statistics | |||
| Resolution range | 6-2.5 Å | ||
| R factor | 20.11% | ||
| Rfree factor | 30.25% | ||
| Refined geometry (r.m.s.) | Overall | Protein | DNA |
| Bond angle (°) | 1.866 | 1.899 | 1.834 |
| Bond length (Å) | 0.014 | 0.016 | 0.012 |
In order to facilitate the comparison between our findings in the current and the previously solved a1/[alpha]2-DNA ternary complex structure, the amino acid residues of the a1 and [alpha]2 homeodomains were renumbered from 0 to 60 according to the numbering scheme established in previous studies of the homeodomain (7,9,19,20). In this numbering scheme, residues 128-151 of [alpha]2 correspond to homeodomain residues 0-23 and residues 155-190 of [alpha]2 correspond to homeodomain residues 24-59. Residues 152-154 of [alpha]2, a three amino acid insertion relative to other homeodomains, are labeled a, b and c. The a1 residues 69-126 have been renumbered 0-57.
Biochemical studies
[beta]-Galactosidase assays. Oligonucleotides corresponding to the a1/[alpha]2 DNA binding sites, CW-1 (TACATGTAATTTATTACATCATA) and CW-2 (TACATGTAAAAATTTACATCATA), were synthesized with 5[prime]-TCGA overhangs and cloned into a SalI site within the promoter region of a CYC1-lacZ fusion in plasmid pAV73 as previously described (21). Plasmids containing the a1/[alpha]2 binding sites were transformed into the diploid yeast strain AJ87 (MATa trp1 leu2 ura3 his4) and lacZ expression was assayed by measuring the [beta]-galactosidase activity (22). For each binding site, [beta]-galactosidase activities were measured on three independent transformants and the values were averaged. The standard deviation for all values was <10%.DNA binding experiments. DNA probes used in the electrophoretic mobility shift assays were synthesized and labeled as described (23). Samples for electrophoretic mobility shift assays were prepared in 20 mM Tris, pH 8.0, 0.1 mM EDTA, 5 mM MgCl2, 10 mg/ml BSA (Fraction V), 5% glycerol, 0.1% NP-40 and 10 mg/ml sheared salmon sperm DNA. Protein dilutions were made in 50 mM Tris, pH 8.0, 1 mM EDTA, 500 mM NaCl, 10 mM 2-mercaptoethanol and 10 mg/ml BSA. Amounts of 5 ml of the [alpha]2 and a1 proteins were added to 40 ml of end-labeled operator fragment diluted with assay buffer until the final NaCl concentration was 100 mM. In the control experiment carried out in the absence of protein, 10 ml of protein dilution buffer was added instead of the [alpha]2 and a1 proteins. Reactions were incubated at room temperature for at least 1 h and then half of each reaction was loaded onto a 6% native polyacrylamide gel buffered with 0.5× Tris/borate/EDTA (TBE) and electrophoresed at 200 V for 2 h. Dried gels were exposed on a phosphor screen, scanned on a Molecular Dynamics model 425 phosphorimager and the bands corresponding to bound and free probe were quantitated to calculate the binding affinity.DNA bending assay. Oligonucleotides corresponding to the a1/[alpha]2 DNA binding sites CW-1 and CW-2 were synthesized with 3[prime]-AGCT overhangs and cloned into the SacI site in plasmid pGD579 (10). The circular permutation DNA bending assay (11) was performed as previously described (23). DNA fragments of 430 bp in length containing the a1/[alpha]2 binding site at different positions were generated by cutting the plasmid with appropriate enzymes (BamHI, NheI, HindIII or EcoRI) and filling in 5[prime]-overhangs with [[alpha]-32P]deoxynucleotide triphosphates (dNTP). The labeled fragments were purified and electrophoretic mobility shift assays were performed as described above. Reaction mixtures were electrophoresed on a non-denaturing 4% polyacrylamide gel buffered with 1× TBE. Electrophoresis of DNA in the absence of proteins was carried out on a non-denaturing 6% polyacrylamide gel, stained with ethidium bromide and illuminated with UV light. Apparent DNA bending angles were calculated based on the Thompson and Landy relationship (24).RESULTS
Overview
The structure of the a1/[alpha]2-DNA complex is shown in Figure
A
 B  |
Figure 1. (A) Structure of the MATa1/MAT[alpha]2-DNA ternary complex containing the A5-tract. The full atomic model of the DNA is shown; proteins are depicted as ribbon diagrams. The top monomer corresponds to the MAT[alpha]2 protein and the bottom monomer with shaded helices corresponds to the MATa1 protein. The adenine bases that are part of the A5-tract are marked with the letter A. This figure was generated with the program MOLSCRIPT (40). (B) Sequence of the oligonucleotide used in the current structure. The base pairs that are different from the previous structure are indicated in parentheses.
There is no appreciable variation in the DNA conformation between the current and previously solved a1/[alpha]2-DNA ternary complex structures, which were both determined at 2.5 Å resolution and are of comparable quality. The two DNA structures superimpose with a 0.48 Å root mean square (r.m.s.) difference in phosphorus atom positions. For the [alpha]2 and a1 proteins, comparison of the previous and current structures reveals an r.m.s. difference in C[alpha] atomic positions of 0.36 and 0.16 Å, respectively. As shown in Figure
DNA conformation
The 21 bp oligonucleotide is in the B-DNA conformation throughout, with a mean helical twist of 34.4° and a mean rise per base step of 3.24 Å. In the crystal, complexes are stacked end-to-end, with the DNA forming a pseudo-continuous supercoiled helix that spirals around the crystallographic 61 screw axis. The superhelical spiral has a radius of 63 Å and a pitch of 45.25 Å, the latter corresponding to the length of the c-axis of the crystallographic unit cell. The stacking of adjacent complexes is promoted by Watson-Crick base pairing between the overhanging 5[prime]-ApT at the end of one complex and the complementary unpaired bases at the end of the adjacent complex.
DNA helix parameters. The DNA in the current a1/[alpha]2-DNA ternary complex has essentially the same overall conformation as that of the previously solved a1/[alpha]2-DNA ternary complex (Fig.Figure 2. Comparison of the DNA structures in the current (red) and previous (yellow) MATa1/MAT[alpha]2-DNA complexes achieved by least squares superposition of phosphorus atoms in the two complexes. The top portion of the DNA corresponds to the [alpha]2 binding site and the bottom portion corresponds to the a1 binding site. This figure was generated with the program SETOR (41). Table 2 shows a comparison of the DNA parameters propeller twist, roll, tilt, inclination, and minor groove width from the previously solved MATa1/[alpha]2-DNA ternary complex and the current structure, along with parameters determined for the Nelson (12) and DiGabriele (14) A-tract DNA structures. All of the A-tract global helical parameters shown in Table 2 were extracted from an analysis of the full-length DNA using the program CURVES (25). The average propeller twist along the A-tract is -15.7° for the current structure, as compared with -20.0° for the Nelson A-tract and -17.2° for the DiGabriele A-tract DNA oligonucleotide structures. In all analyzed A-tract structures, the large propeller twist of AT base pairs puts the N6 of adenine within hydrogen bonding distance of the O4 of thymine in the major groove of the adenine tract. The current a1/[alpha]2-DNA complex structure differs from the previously solved ternary complex structure in that the A-tract-containing structure shows continuous bifurcated hydrogen bonds across the AT base pairs whereas the original structure does not, despite the common property of high propeller twist angles. This difference is due in part to DNA sequence differences, since N6···O4 hydrogen bonding can only occur at AA base steps.
Conserved protein-DNA sequence recognition in the minor groove
The base pair changes introduced into the a1/[alpha]2 binding site are located in the region of the DNA where the N-terminal arm of the protein forms contacts with bases in the minor groove. Three residues in the N-terminal arm of [alpha]2 contact base pairs in the minor groove: Arg4, Gly5 and Arg7. The bases that are contacted by Arg4 and Gly5 are not affected by these changes and these two residues therefore make analogous contacts in the two ternary complex structures. The DNA sequence changes that were introduced do, however, invert both base pairs contacted directly by Arg7. In the original structure (Fig.
Conserved waters at the protein-DNA interface
A number of the important contacts between the a1/[alpha]2 heterodimer and the DNA are mediated by bound water molecules. Determining the structure of a1/[alpha]2 bound to a second DNA fragment provided an important opportunity to ascertain whether the water-mediated hydrogen bonds identified in the previous structure were, indeed, conserved. A cross-comparison between the 58 water molecules assigned in the previous structure and the 51 water molecules assigned in the current structure indicates that 85% of the previously assigned water molecules are conserved with positional displacements of <1.2 Å.
Table 2.
| Propeller twist (°) | Roll (°) | Tilt (°) | Inclination (°) | Minor groove width (Å | |
| TAa (TA)b | -9.98 (-8.95) | 0.38 (2.19) | 5.26 (4.85) | ||
| 5.26 (1.03) | 2.77 (-2.54) | ||||
| AT (AT) | -6.65 (-11.56) | 3.8 (2.43) | 6.34 (5.92) | ||
| 2.06 (4.60) | -0.69 (0.00) | ||||
| AT (AT) | -22.32 (-19.35) | 2.68 (1.82) | 5.24 (4.85) | ||
| -0.07 (-3.68) | 0.98 (0.48) | ||||
| AT (TA) | -19.71 (-23.43) | 2.17 (-0.01) | 3.60 (3.43) | ||
| -4.63 (3.33) | 1.71 (2.23) | ||||
| AT (TA) | -11.67 (-15.00) | 1.51 (0.93) | 2.24 (2.32) | ||
| -6.12 (-13.11) | 1.07 (1.07) | ||||
| AT (TA) | -18.08 (-15.66) | 2.91 (1.10) | 2.45 (2.51) | ||
| -6.75 (-6.55) | -6.08 (1.01) | ||||
| TA (AT) | -19.29 (-7.68) | 0.44 (3.30) | 3.44 (3.64) | ||
| -1.73 (-2.71) | 0.57 (-3.96) | ||||
| TA (TA) | 5.61 (6.58) | ||||
| -1.73 (-2.71) | |||||
| CGc (GC)d | -1.98 (-12.38) | 0.80 (3.11) | 7.31 (4.18) | ||
| 10.0 (-0.82) | 1.98 (-6.24) | ||||
| AT (AT) | -13.9 (-16.64) | 0.54 (-2.10) | 4.76 (3.24) | ||
| 0.31 (-3.96) | -1.9 (-0.88) | ||||
| AT (AT) | -22.43 (-16.21) | -1.58 (-2.37) | 3.31 (3.25) | ||
| -0.47 (-2.82) | 0.39 (-0.25) | ||||
| AT (AT) | -25.96 (-18.26) | -3.81 (-2.23) | 3.22 (3.31) | ||
| 3.41 (5.21) | 7.72 (2.01) | ||||
| AT (AT) | -22.04 (-19.19) | 0.26 (-0.81) | 3.54 (3.45) | ||
| 2.22 (-0.11) | -3.93 (-1.35) | ||||
| AT (AT) | -17.44 (-19.76) | -2.96 (0.93) | 4.06 (4.67) | ||
| -4.52 (-1.17) | 5.36 (-10.76) | ||||
| AT (AT) | -18.27 (-13.48) | 1.46 (-2.93) | 4.26 (--) | ||
| 7.44 (9.40) | -2.58 (-0.27) |
Both the a1 and [alpha]2 homeodomains use the third [alpha]-helix, known as the recognition helix, for DNA binding. In the original structure, the O[gamma] of Ser51 uses two water-mediated hydrogen bonds to form contacts with the N7 of Ade4 and O4 of Thy5 (Fig.
Figure 3. Stereo view of the superposition of the A-tract in the current structure (yellow) with the A-tract in the oligonucleotide structure of Nelson (12) (cyan). The line marking the global helical axis for each DNA fragment was calculated with the program CURVES (25). The DNA orientation is the same as shown in Figure 2 where the top portion extends towards the [alpha]2 half of the binding site and the bottom portion extends towards the a1 binding site.
A
B
Figure 4. Contacts between Arg7 in the N-terminal arm of the [alpha]2 homeo-domain and bases in the minor groove. (A) Minor groove contacts by Arg7 found in the previously reported ternary complex structure. The DNA backbone is shown as a ribbon drawn through the C3[prime] atoms of the ribose at each base; water molecules are shown in red. Arg7 donates a hydrogen bond to both the N3 of Ade35 and the O2 of Thy11. In addition, the NH2 helps to stabilize a minor groove spine of hydration that extends towards the center of the DNA fragment. (B) Contacts between Arg7 in the N-terminal arm of the [alpha]2 homeodomain and bases in the minor groove for the ternary complex reported here. Note the reversal of the AT bases as compared with the original structure. In this structure the NH2 of Arg7 makes bidentate contacts with the O2 of Thy35 and N3 of Ade11. In addition, the NH1 of Arg7 is located 3.4 Å from the N3 of Ade11. The NH2 of Arg7 in the new structure also participates in the stabilization of a minor groove spine of hydration. This figure was generated with the program SETOR (41).
: A
 : B  |
Figure 5. Water-mediated hydrogen bonds between helix 3 of [alpha]2 and bases in the major groove. (A) Previous structure. (B) Current structure. Water molecules that are conserved in both structures are depicted as red spheres and waters that are not found in both structures are shown as blue spheres; broken lines indicate hydrogen bonds. Figures 5 and 6 were generated with the program MAXIMAGE (42)
A
 B  |
Figure 6. Water-mediated hydrogen bonds between helix 3 of a1 and bases in the major groove. (A) Previous structure. (B) Current structure. Water molecules that are conserved in both structures are depicted as red spheres and waters that are not found in both structures are shown as blue spheres; broken lines indicate hydrogen bonds.
The a1 and [alpha]2 proteins form a large set of contacts with the sugar-phosphate backbone of the DNA that presumably contribute to the stability of the protein-DNA complex. One critical set of contacts formed by a1 and [alpha]2 with the DNA backbone is mediated by Arg53, an invariant homeodomain residue (8) that was observed in the original structure to mediate an identical set of contacts with the DNA backbone by both the a1 and [alpha]2 homeodomains. In addition to direct hydrogen bond and salt bridge contacts between the side chain of Arg53 and DNA phosphates, the side chain NH1 donates a hydrogen bond to a bound water molecule that simultaneously donates a hydrogen bond to phosphate 3 and accepts a hydrogen bond from the peptide NH of Leu26. The same set of interactions mediated by a bound water in an analogous position is found in both the original a1 and [alpha]2 half-complexes and is completely conserved in the new ternary complex structure reported here.
Minor groove spine of hydration
At the center of the DNA fragment, a zig-zag spine of hydration is found in the narrowed minor groove similar to that observed in the Dickerson dodecamer (26) and in the A-tract structure of DiGabriele (14). In the original a1/[alpha]2-DNA structure, a total of five waters are bound in the minor groove (Fig.
Figure 7. The spine of hydration in the minor groove for the current (red) and the previously (yellow) determined structures. The figure shows hydrogen bonds between the water molecules and DNA bases only; additional hydrogen bonds with the O4[prime] of the ribose rings are not shown in order to reduce the complexity of the illustration. This figure was generated with the program SETOR (41).
In vivo and in vitro studies of a1/[alpha]2 heterodimer binding to DNA
Experiments were carried out to determine the effect of the two DNA sequences used in the structural studies on the DNA binding and bending of the a1/[alpha]2 heterodimer and on the ability of a1/[alpha]2 to repress mRNA transcription in yeast. The results indicate that a1/[alpha]2 has very similar affinity for the two DNA sequences. In an in vivo assay measuring repression of [beta]-galactosidase expression by a1/[alpha]2, the presence of the A-tract-containing a1/[alpha]2 binding site upstream of the [beta]-galactosidase gene resulted in 2-fold greater repression as compared with reporter constructs containing the a1/[alpha]2 binding site used in the original structure determination (Fig.
A
 B  |
C
 |
Figure 8. (A) Electrophoretic mobility shift assay measuring the binding affinity of a1/[alpha]2 for the original DNA sequence, TACATGTAATTTATTACATCATA (CW-1), and for the sequence used in the present study, TACATGTAAAAATTTACATCATA (CW-2). Lanes 1 and 6 are free probe and the [alpha]2 proteins were added to a final concentration of 6 × 10-11 M. (B) Repression by a1/[alpha]2 of [beta]-galactosidase reporter gene in vivo by binding to CW-1 and CW-2. (C) The circular permutation bending assay measures the DNA bending by a1/[alpha]2. The diagram shows the position of the a1/[alpha]2 binding site within each 430 bp fragment used as probe. DNA bending, either intrinsically sequence-directed or protein-induced, remains an important topic in the study of nucleic acids and gene regulation. The protein-induced DNA bending observed in the a1/[alpha]2-DNA complex is required in order for the a1 and [alpha]2 homeodomains to contact simultaneously their respective binding sites in the DNA as well as form the protein-protein interactions that stabilize the heterodimer (7). In the structure determination reported here, we examined the effect of four base pair substitutions at the center of the DNA fragment where there is a high degree of bending, a significantly compressed minor groove at the center of the bend and a correspondingly wide major groove on the opposite face. The DNA sequence changes, which did not change the overall base composition (Fig. The base pair changes in the a1/[alpha]2 binding site have the additional effect of introducing an A-tract of 5 bp in length that extends from the end of the [alpha]2 binding site through the center of the DNA bend that lies between the a1 and [alpha]2 binding sites. While solution studies have firmly established the role played by A-tracts in intrinsic DNA bending (11), a detailed structural explanation for the source and location of the bend has remained elusive. In the crystal structures of A-tract DNA in the absence of protein that have been reported (12-14), the A-tracts themselves have been shown to be quite straight. All of these structures show the A-tract to exhibit features of high propeller twisting of the base pairs, bifurcated hydrogen bonds between bases in the major groove and, in the case of the DiGabriele A-tract structure (14), a minor groove spine of hydration similar to that first seen in the Dickerson dodecamer (26). These structural features are thought to contribute to an overall stiffness of the A-tract itself. In all of the A-tract structures determined in the absence of protein, bending occurs at the junction between A-tract and non-A-tract DNA. Our findings from the current MATa1/[alpha]2-DNA ternary complex structure indicate that protein-protein interactions can induce modest bending within an A-tract. This bending occurs close to the end of the A-tract DNA sequence (Fig. Despite the presence of a bend within the A-tract embedded in the a1/[alpha]2 binding site, the other structural features of the A-tract are very similar to those observed in previous structures of A-tract-containing oligonucleotides. These include the presence of a large base pair propeller twist accompanied by bifurcated hydrogen bonds, narrow minor groove width and a minor groove spine of hydration. With the exception of the bifurcated hydrogen bonding, these features are not unique to the homopolymeric A5-tract in the present structure, as the same properties are observed in the previously solved ternary complex structure containing an A2T3 sequence in place of the A5-tract. The minor groove spine of hydration, which is found in both a1/[alpha]2-DNA ternary complex structures appears most likely to be a consequence of the presence of a narrowed minor groove, highly propeller twisted bases and an additional hydrogen bonding interaction with Arg7 of [alpha]2. The bound waters may also serve to stabilize the overall bent conformation of the DNA. The sequence changes introduced into the a1/[alpha]2 binding site invert two of the base pairs directly contacted in the minor groove by Arg7 in the N-terminal arm of [alpha]2. The affected base pairs, A10-T35 and A11-T34 in the present structure, are both T-A base pairs in the original structure. As noted by Seeman and colleagues (27), base pair reversals do not affect the relative disposition of hydrogen bond acceptors in the minor groove and should therefore not have a significant effect on base recognition in the minor groove. We find that Arg7 in the [alpha]2 N-terminal arm donates hydrogen bonds to acceptors located in analogous positions in the two structures and exhibits the predicted insensitivity to base pair reversal. The conservation of these contacts is reflected in the negligible difference in the affinity of the a1/[alpha]2 heterodimer for the two sites (Fig. The N-terminal arm of [alpha]2 forms extensive contacts with base pairs in the minor groove, a feature common not only to the two a1/[alpha]2-DNA ternary complex structures but to all otherhomeodomain-DNA complex structures that have been previously reported (9,20,29-32). In contrast, the N-terminal arm of the a1 homeodomain is not structured in either the current a1/[alpha]2-DNA ternary complex structure or in the previous a1/[alpha]2-DNA structure (7), despite the absence in either case of any aspect of DNA structure or crystal packing that would preclude the binding of the N-terminal arm of a1 in the minor groove in a manner observed in all other homeodomain complex structures. That the N-terminal arm of a1 may indeed lack a role in DNA binding is supported by in vitro studies of a1/[alpha]2 binding to DNA, which show that deletion of residues in the a1 N-terminal arm does not diminish the affinity of the a1/[alpha]2 heterodimer for DNA (M.R.Stark, T.Li, C.Wolberger and A.D.Johnson, submitted for publication). These results are surprising in light of the sequence similarity between the N-terminal arm residues of a1 and other members of the homeodomain superfamily (33). Further study of the DNA binding properties of a1 will be required in order to learn why its manner of binding to DNA differs from other homeodomain proteins. Water molecules have been shown to play an important role in mediating contacts between proteins and DNA (34-36). Solution NMR studies of the Antennapedia homeodomain identified the presence of extensive hydration at the homeodomain helix 3-major groove interface (29), while X-ray crystallographic studies of the paired class (31), even-skipped (30) and a1/[alpha]2 homeodomains (7) have revealed the positions of bound water molecules that mediate hydrogen bonds between homeodomain side chains and the DNA. These contacts have been postulated to play an important role in the specificity of the interaction of homeodomains with their cognate binding sites, possibly by improving the complementarity of the protein-DNA interaction interface (37). In the study of a1/[alpha]2 bound to an altered DNA site reported here, we have examined the degree to which the positions of bound water molecules located in the previous structure of the a1/[alpha]2-DNA ternary complex are conserved in a second, independent structure determination. While the great majority of the bound waters previously identified can be located in nearly identical positions in the current structure determination, there are differences in overall conservation of waters at the [alpha]2-DNA and a1-DNA interfaces that are involved in base-specific contacts. At the [alpha]2 helix 3-DNA interface, the two bound waters that were previously observed to mediate hydrogen bonds between Ser50 and two bases (Fig. In contrast, the hydration in the major groove of the a1-DNA half-complex is less highly conserved in the two ternary complexes. In addition to the direct hydrogen bonds between Asn51 and Arg55 and functional groups in the major groove and van der Waals contacts between Ile50 and Met54 and the DNA bases, a network of five water molecules is found in the original a1/[alpha]2-DNA structure at the interface between helix 3 of a1 and the DNA. The waters in this network form several hydrogen bonds to DNA bases as well as to Arg46 (Fig. We thank H. Berman, D. Crothers and Z. Shakked for helpful discussions. This work was supported by grant MCB-93045267 from the National Science Foundation (C.W.), grant GM-49265 from the National Institutes of Health (A.K.V.) and by a fellowship from the David and Lucile Packard Foundation (C.W.). Coordinates have been deposited in the Brookhaven Protein Data Bank with accession code 1AKH.
DISCUSSION
Structure of the A-tract within the a1/[alpha]2 binding site
Recognition of bases in the minor groove
Role of solvent in mediating protein-DNA interactions
ACKNOWLEDGEMENTS
REFERENCES
This article has been cited by other articles:
This page is run by Oxford University Press, Great Clarendon Street, Oxford OX2 6DP, as part of the OUP Journals
Comments and feedback: www-admin{at}oup.co.uk
Last modification: 3 Dec 1998
Copyright©Oxford University Press, 1998.
CiteULike 
Connotea 
Del.icio.us What's this?
A. Anbanandam, D. C. Albarado, C. T. Nguyen, G. Halder, X. Gao, and S. Veeraraghavan
Insights into transcription enhancer factor 1 (TEF-1) activity from the solution structure of the TEA domain
PNAS,
November 14, 2006;
103(46):
17225 - 17230.
[Abstract]
[Full Text]
[PDF]
J. R. Mathias, S. E. Hanlon, R. A. O'Flanagan, A. M. Sengupta, and A. K. Vershon
Repression of the yeast HO gene by the MAT{alpha}2 and MATa1 homeodomain proteins
Nucleic Acids Res.,
December 14, 2004;
32(22):
6469 - 6478.
[Abstract]
[Full Text]
[PDF]
Z. Moravek, S. Neidle, and B. Schneider
Protein and drug interactions in the minor groove of DNA
Nucleic Acids Res.,
March 1, 2002;
30(5):
1182 - 1191.
[Abstract]
[Full Text]
[PDF]
J. R. Mathias, H. Zhong, Y. Jin, and A. K. Vershon
Altering the DNA-binding Specificity of the Yeast Matalpha 2 Homeodomain Protein
J. Biol. Chem.,
August 24, 2001;
276(35):
32696 - 32703.
[Abstract]
[Full Text]
[PDF]
This Article 
Abstract
Print PDF (596K)
Alert me when this article is cited
Alert me if a correction is posted
Services
Email this article to a friend
Similar articles in this journal
Similar articles in ISI Web of Science
Similar articles in PubMed
Alert me to new issues of the journal
Add to My Personal Archive
Download to citation manager
Search for citing articles in:
ISI Web of Science (18)
Commercial Re-use Guidelines
for Open Access NAR Content
Google Scholar
Articles by Li, T.
Articles by Wolberger, C.
Search for Related Content
PubMed
PubMed Citation
Articles by Li, T.
Articles by Wolberger, C.
Social Bookmarking
What's this?