Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

A Hoogsteen base pair embedded in undistorted B‐DNA

A Hoogsteen base pair embedded in undistorted B‐DNA 5244±5252 Nucleic Acids Research, 2002, Vol. 30, No. 23 ã 2002 Oxford University Press A Hoogsteen base pair embedded in undistorted B-DNA 1 2 3 4 2 4 5 2 4 5 , , , , , Jun Aishima , Rossitza K. Gitti , Joyce E. Noah , Hin Hark Gan , Tamar Schlick 1 2 , , and Cynthia Wolberger * 1 2 Department of Biophysics and Biophysical Chemistry and Howard Hughes Medical Institute, Johns Hopkins University School of Medicine, Baltimore, MD 21205-2185, USA, Department of Chemistry and Biochemistry, University of Maryland Baltimore County, Baltimore, MD 21250, USA, Department of Chemistry and Courant Institute of Mathematical Sciences, New York University, 31 Washington Place, Room 1021 Main, New York, NY 10003, USA Received July 11, 2002; Revised and Accepted October 3, 2002 PDB no. 1K61 ABSTRACT characterized by hydrogen bonds between the side of the purine base that faces the major groove and the Watson±Crick Hoogsteen base pairs within duplex DNA typically base pairing face of the pyrimidine. In B-DNA, formation of a are only observed in regions containing signi®cant Hoogsteen base pair would require rotation of the purine distortion or near sites of drug intercalation. We base about the glycosidic c bond from the anti to the syn report here the observation of a Hoogsteen base conformation and, in the case of a guanine-cytosine base pair, pair embedded within undistorted, unmodi®ed protonation of the N3 of cytosine. Hydrogen bonds in the B-DNA. The Hoogsteen base pair, consisting of a Hoogsteen base pair are formed between the purine N7 to the syn adenine base paired with an anti thymine base, pyrimidine N3 and either the adenine N6 to the thymine O6 or the guanine O4 to the cytidine N4 (4). is found in the 2.1 A resolution structure of the Hoogsteen base pairs have been observed in several crystal MATa2 homeodomain bound to DNA in a region structures of distorted double-stranded B-DNA. DNA com- where a speci®cally and a non-speci®cally bound plexed with intercalating binding drugs, such as triostin-A, homeodomain contact overlapping sites. NMR contain Hoogsteen base pairs in regions of underwound studies of the free DNA show no evidence of B-DNA (1). Hoogsteen base pairs can also occur in regions of Hoogsteen base pair formation, suggesting that DNA that are highly distorted by bound protein. In the protein binding favors the transition from a structure of TATA-binding protein (TBP) bound to DNA (3), a Watson±Crick to a Hoogsteen base pair. Molecular Hoogsteen base pair is observed in the region of DNA dynamics simulations of the homeodomain±DNA underwinding and intercalation by a phenylalanine side chain complex support a role for the non-speci®cally from TBP. Finally, Hoogsteen base pairs have also been bound protein in favoring Hoogsteen base pair observed at free ends of end-to-end stacked oligonucleotides, formation. The presence of a Hoogsteen base pair in such as in the structure of integration host factor bound to the crystal structure of a protein±DNA complex DNA (5). The presence of distortions or ¯exible end regions in raises the possibility that Hoogsteen base pairs B-form DNA may lower the energy barrier for rotation of the could occur within duplex DNA and play a hitherto purine base about the glycosidic c bond by changing the base unrecognized role in transcription, replication and stacking arrangements or the helical properties of the DNA. The lower energy barrier may allow the purine base to rotate other cellular processes. from the normal anti conformation to the syn conformation, leading to the formation of hydrogen bonds characteristic of a INTRODUCTION Hoogsteen base pair. We report here the observation of a Hoogsteen base pair Most duplex DNA, whether in A-, B- or Z-form, is composed embedded within undistorted B-DNA. The Hoogsteen base of complementary strands that associate solely through pair, formed by a syn adenine base and an anti thymine base, Watson±Crick base pairing. In a small number of DNA occurs within a 2.1 A resolution crystal structure containing structures containing intercalating drugs (1,2) or pronounced four MATa2 homeodomains bound to DNA. The 21 bp DNA protein-induced distortion (3), Hoogsteen base pairs have been used for the crystal structure contains two binding sites for the found embedded in duplex DNA. The Hoogsteen base pair MATa2 homeodomain and was previously used in a crystal geometry, which was ®rst observed in crystal structures of structure of the MATa2 homeodomain±DNA complex (6). In monomeric adenine and thymine base derivatives (4), is *To whom correspondence should be addressed at 7 Johns Hopkins University School of Medicine, Baltimore, MD 21205-2185, USA. Tel: +1 410 955 0728; Fax: +1 410 614 8648; Email: cwolberg@jhmi.edu Present address: Jun Aishima, Lawrence Berkeley National Laboratory, 1 Cyclotron Road, Mail Stop 4-230, Berkeley, CA 94720, USA Downloaded from https://academic.oup.com/nar/article-abstract/30/23/5244/1051861 by Ed 'DeepDyve' Gillespie user on 06 February 2018 Nucleic Acids Research, 2002, Vol. 30, No. 23 5245 Figure 1. (A) Crystal structure of the a2 homeodomain contains four a2 proteins bound to two a2 binding sites in the DNA. Base pair A7-T37 (red box) is a Hoogsteen base pair. Figure produced with SETOR (37). (B) The oligonucleotide duplex used in the crystal structure contains two a2 binding sites (box) with a5¢ overhanging base at each end. (C) The DNA fragment used in the NMR experiments is blunt-ended and contains ®ve fewer base pairs. (D) The DNA fragment used in the molecular dynamics simulations. the current structure, two of the MATa2 homeodomains bind Phasing and re®nement to the MATa2 binding sites, while the other two MATa2 Full details of the X-ray data collection, phasing and proteins are bound non-speci®cally to the DNA (7). re®nement procedures are described elsewhere (7). Initial Simulated annealing omit maps (8) clearly show the trials of molecular replacement (using a model of two a2 unambiguous presence of the Hoogsteen base pair in the proteins bound to DNA from the original structure; 6) on the crystal structure. The Hoogsteen base pair appears to be 2.1 A resolution data set were unsuccessful, resulting in stabilized by an extra inter-base pair hydrogen bond and base distorted DNA during re®nement with Xplor (9) and CNS stacking within the DNA, as well as by contacts made by one (10). Subsequently, electron density maps calculated with of the non-speci®cally bound MATa2 homeodomains with multiple isomorphous replacement phases showed that the the sugar±phosphate backbone of the syn adenine base. None molecular replacement solution was correct. However, two of the observed stabilizing contacts are out of the ordinary, unexpected, additional a2 molecules were visible in the suggesting that Hoogsteen base pairs and other non- bones-skeletonized electron density map (11). After the two Watson±Crick base pairs may occur more commonly within additional a2 molecules were placed in the density, re®nement undistorted DNA than has previously been thought. proceeded smoothly with CNS. For all data to 2.1 A resolution, the ®nal R (12) and R factor are 27.8 and free 22.2%, respectively. MATERIALS AND METHODS Sample preparation Protein and oligonucleotide synthesis and The 16 bp DNA fragment used for the NMR studies is 5 bp puri®cation shorter than the crystallization oligonucleotide and does not The procedure for the puri®cation of the a2 homeodomain and contain base pairs 17:27 to 21:23 (Fig. 1C). Single-stranded DNA are described elsewhere (7). Brie¯y, the a2 homeo- oligonucleotides were puri®ed twice over a Dynamax domain (residues 132±191) was synthesized by the Fmoc solid PureDNA column (Rainin) with an acetonitrile gradient in state peptide synthesis method and puri®ed by C4 reverse 0.1 M triethylamine acetate (pH 7). The trityl group on the phase column (Vydac). Mass spectrometry was used to verify oligonucleotide was removed during the second run by 0.5% the molecular weight of the protein. The DNA was TFA. Peak fractions were pooled, dialyzed into 10 mM synthesized at the HHMI±Keck Biopolymer Facility at Yale triethylamine bicarbonate (pH 7) and annealed. Annealed University and puri®ed with a PureDNA reverse phase column oligonucleotides were quantitated by spectroscopic absorb- (Rainin). Oligonucleotides were combined in a 1:1 molar ratio ance at 260 nm, lyophilized and stored at ±80°C. NMR studies and annealed by heating to 90°C, then cooling overnight. Both were performed on unlabeled DNA samples, redissolved after protein and DNA were quantitated by spectroscopic methods. HPLC into 25 mM sodium phosphate and 25 mM NaCl buffer The DNA concentration was quantitated with the formula (pH 7.2) in D O. DNA samples for exchangeable proton 1OD unit = 50 mg/ml. The protein concentration was detection were transferred in H O by lyophilization to a ®nal 260 2 veri®ed by the Lowry method (Sigma). Protein and DNA were concentration of 2.0 mM DNA in 25 mM sodium phosphate combined at a molar ratio of 1.8:1. and 25 mM NaCl buffer (pH 6.4) in 95% H O/5% D O. 2 2 Downloaded from https://academic.oup.com/nar/article-abstract/30/23/5244/1051861 by Ed 'DeepDyve' Gillespie user on 06 February 2018 5246 Nucleic Acids Research, 2002, Vol. 30, No. 23 Table 1. Twist and rise for the DNA in the crystal NMR spectroscopy structure and standard B-DNA NMR experiments were carried out at 25°C on General Base pair Twist Rise Electric OMEGA PSG and Bruker DMX 600 MHz spectro- meters equipped with x,y,z-shielded gradient triple resonance C2-G42 36.38 3.88 probes. NMR data were processed with NMRPipe (13) and A3-T41 25.81 3.59 T4-A40 45.44 2.43 analyzed with NMRVIEW (14). NOE data (mix time t = G5-C39 26.93 3.09 150 ms) were obtained from 2D NOESY (15,16). Water T6-A38 38.66 3.04 suppression was achieved with WATERGATE and ®eld A7-T37 32.73 3.89 gradient pulses (17) or presaturation pulses during the A8-T36 32.38 3.29 relaxation delay for DNA samples in H O. 2D TOCSY data T9-A35 31.06 3.46 T10-A34 39.12 2.77 were obtained in D O with a 75 ms clean-MLEV-17 mixing C11-G33 44.77 3.22 1 13 period (18±20). H- C HMQC (21) was recorded in natural A12-T32 27.94 3.85 abundance. T13-A31 35.29 3.13 T14-A30 35.50 3.38 Energetics and dynamics calculations T15-A29 39.41 2.98 A16-T28 27.29 3.32 The a2B and a2D proteins were used for the dynamics C17-G27 36.56 3.18 studies. Structures of the DNA alone, a2B±DNA complex A18-T26 30.60 3.44 and a2B±a2D±DNA complex were studied by ®rst solvating C19-G25 38.06 3.35 with water, then neutralizing with sodium ions. The systems G20-C24 26.69 3.24 C21-G23 were propagated using a multiple timestepping integrator for Average 34.2 3.29 Langevin dynamics (22,23) as implemented in the CHARMM B-DNA 38.6 3.38 molecular modeling package, v.26a2 (24). The systems were minimized and equilibrated to 300K for 75 ps before starting production runs. The molecular simula- each containing residues 132±191 of the MATa2 protein tions and energy minimization use the Cornell et al. (25) force [homeodomain residues 4±60 as numbered by Qian et al. ®elds as implemented in the CHARMM package. Energy (31)], bound to a 21 bp fragment of duplex DNA that contains minimization of the system was done using the steepest two a2 binding sites. Two of the a2 homeodomains, a2A and descent method and an adopted-basis Newton±Raphson a2B (Fig. 1A), bind DNA at the two a2 binding sites, while protocol in CHARMM. (See 26,27 for further details.) The the other two a2 proteins bind DNA in an atypical fashion and free energy, f, was estimated as f(c) = ±0.59 lnn(c) (kcal/mol), are considered non-speci®cally bound proteins. One of the where n(c) is the number of counts at c (28). non-speci®cally bound homeodomains, a2D, binds DNA in a previously unobserved manner, resulting in a signi®cant Coordinates reorientation of helix 3 relative to the DNA that disrupts The atomic coordinates and structure factors have been normal side chain±major groove contacts. Instead, other side deposited in the Protein Data Bank (accession number 1K61). chains that are not typically involved in homeodomain±DNA interactions are found to mediate base contacts with the major groove. Despite the overall change in the homeodomain RESULTS docking and major groove contacts, there is little change in the A Hoogsteen base pair is seen in a homeodomain/DNA contacts with the sugar±phosphate backbone. The other non- crystal structure speci®cally bound a2 homeodomain, a2C, binds near the junction of two stacked, crystallographically related DNA Like other homeodomains, the MATa2 homeodomain (abbre- fragments. viated as a2) consists of a compact three a-helix domain with The DNA reported in the current crystal structure contains a a ¯exible N-terminal arm that becomes ordered upon binding total of 21 bp (Fig. 1B). Twenty base pairs are contained DNA. Homeodomains make base-speci®c contacts with the within the duplex and one base pair is formed by overhanging major groove of the DNA using residues in the third a-helix, 5¢ bases from adjacent complexes, which stack end-to-end to while the N-terminal arm contacts bases in the minor groove. form a pseudocontinuous DNA helix. The DNA is B-form In addition to base-speci®c contacts, residues in the loop throughout, with sugar puckers, axial rises and twist all between helices 1 and 2, as well as residues in helix 3, contact characteristic of B-DNA (Table 1). the sugar±phosphate backbone. Previous crystal structures of All of the DNA was initially modeled with Watson±Crick the a2 homeodomain bound to DNA (PDB entry 1APL) (6), base pairs. The 2F ± F and F ± F electron density maps ®t the a1±a2 heterodimer bound to DNA (PDB entry 1YRN) o c o c the Watson±Crick DNA model well except at base pair A7- (29) and the MCM1/a2 heterotetramer bound to DNA (PDB T37. The difference density at A7-T37 could only be entry 1MNM) (30) show that the a2 homeodomain has the same structure and makes similar DNA contacts when bound accounted for by the rotation of the A7 base about the torsion to canonical a2 binding sites. angle c to the uncommon syn conformation, yielding an We recently determined a higher resolution structure of the A(syn)´T(anti) Hoogsteen base pair. Simulated annealing omit MATa2 homeodomain bound to DNA (7). This new, 2.1 A maps of the A7-T37 base pair con®rm the presence of the resolution structure was determined by a combination of Hoogsteen base pair at this position (Fig. 2A). To accommo- multiple isomorphous replacement and molecular replace- date the short C1¢±C1¢ distance across the base pair [8.5 A for ment. Brie¯y, the structure contains four a2 homeodomains, this base pair, as opposed to 10.5 A for Watson±Crick base Downloaded from https://academic.oup.com/nar/article-abstract/30/23/5244/1051861 by Ed 'DeepDyve' Gillespie user on 06 February 2018 Nucleic Acids Research, 2002, Vol. 30, No. 23 5247 Figure 2. The Hoogsteen base pair is stabilized by intra-DNA and protein±DNA interactions in the crystal structure of the a2 homeodomain±DNA complex. (A) Simulated annealing omit map of the A7-T37 base pair. The two hydrogen bonds between the N7 of base A7 and N3 of base T37, as well as between the N6 of A7 and the O4 of T37 characterize a Hoogsteen base pair. The same base pair modeled in the Watson±Crick con®guration (green) clearly does not ®t the electron density. Figure prepared with Pymol (38). (B) The A7 N6 group makes a bifurcated hydrogen bond with both T37 and T36. Base stacking inter- actions between T6 and A7, as well as A8 and T9, may contribute to stabilization of the Hoogsteen base pair. (C) The Arg4 residue of the a2D homeodomain packs against the sugar±phosphate backbone at bases A7 and T6. (B) and (C) prepared with VMD (39). pairs (4)], the a and g torsion angles (about bonds P±O5¢ and mistakenly modeled as Watson±Crick base pairs. Crystal C5¢±C4¢, respectively) in the sugar±phosphate backbone are in structures of proteins bound to straight DNA determined from + ± Ê the unusual gauche /gauche conformation about base A7. In crystals diffracting to 2.5 A resolution or better and with normal B-DNA, the a and g torsion angles are in the gauche / available structure factor ®les were examined (PDB entries gauche conformations. There are no changes in the conform- 1FJL, 1B72, 1DP7, 2CGP and 1LAT). 2F ± F and F ± F o c o c electron density maps were generated with CNS (10) and ation of the T37 phosphate backbone a or g torsion angles and viewed in x®t from the XtalView package (33). In no case was there are no other irregularities in the backbone of the DNA. density indicative of a Hoogsteen base pair seen in the DNA. The Hoogsteen base pair seen in this crystal structure appears to be a property of this protein±DNA complex and is In the absence of protein, the DNA in solution does not observed in different crystals grown from independent DNA contain a Hoogsteen base pair syntheses. In addition to the 5-iodouracil derivative crystal used for re®nement, we also examined the simulated Is the Hoogsteen base pair somehow intrinsic to this DNA annealing omit maps of the A7-T37 base pair using data sequence? To answer this question we used NMR spectro- from the native crystal, which extends to 2.4 A resolution. scopy to examine the structure of a 16 bp DNA fragment in Electron density maps calculated from the native data set also solution. This 16 bp DNA fragment (Fig. 1C) is a shorter reveal the presence of the Hoogsteen base pair, showing that version of the crystallized DNA and lacks the 5 bp farthest the 5-iodouracil substitution in the DNA does not affect the from the A7-T37 site, but it is suf®ciently long to retain local formation of the Hoogsteen base pair and that the Hoogsteen interactions that could potentially have induced formation of base pair occurs in different crystals of this protein±DNA the Hoogsteen base pair. In the crystal structure of the complex. In contrast, neither the DNA from the previous a2±DNA complex, only the base pairs immediately adjacent structure of the a1±a2±DNA complex nor the DNA in the to base pair A7-T37 make stabilizing contacts with the other a2 binding site in the current a2±DNA crystal structure Hoogsteen base pair. contains a Hoogsteen base pair. The Hoogsteen base pair We examined base conformations in solution by following therefore appears to be unique to this a2±DNA complex and to the standard sequential connectivities involving the sugar± the A7-T37 position within the DNA. base H1¢±H6/H8 and H2¢/H2¢¢±H6/H8 pathways for all 32 The unexpected observation of a Hoogsteen base pair in the bases using both 2D NOESY (nuclear Overhauser and present structure raised the possibility that the presence of exchange spectroscopy) and TOCSY (total correlation spec- Hoogsteen base pairs may have been missed in previous troscopy) experiments in D O (see Materials and Methods). If structures due to misinterpretation of ambiguous electron the A7 base is in the syn conformation in solution, there should density. We examined several structures of protein±DNA be a strong NOE between A7-H1¢ and A7-H8 as the distance complexes in the Protein Data Bank (32) to determine whether between these two protons is 1.5 A closer than in the anti any structures contained Hoogsteen base pairs that were conformation (34). The entire 2D NOESY-D O spectrum in Downloaded from https://academic.oup.com/nar/article-abstract/30/23/5244/1051861 by Ed 'DeepDyve' Gillespie user on 06 February 2018 5248 Nucleic Acids Research, 2002, Vol. 30, No. 23 Figure 3. NMR studies of the DNA clearly show a Watson±Crick base pair in solution. (A) The NOESY-D O spectrum was easily assignable in the H1¢±H6/ H8 region. The weak A7 H1¢±A7 H8 NOE correlation peak is consistent with A7 in the anti conformation. (B) The NOESY-H O spectrum was assignable in the amino±imino region. The A7 H2-T37 H3 NOE correlation (red box) is strong, consistent with a Watson±Crick base pair at A7-T37. (C) In the NOESY- D O spectrum, the A7-H2 peak is shifted up®eld and is broader than other adenine H2 peaks, consistent with a ¯exible A7 base. the H1¢±H6/H8 (Fig. 3A) and H2¢/H2¢¢±H6/H8 (data not conformation in solution at least 95% of the time, in contrast shown) regions was well resolved and easy to assign. to the syn A7 base conformation that was observed in the The presence of a weak NOE correlation peak between A7- crystal structure of the a2±DNA complex. Furthermore, the H1¢ and A7-H8 indicates that the A7 base is in an anti A7-H2¢/H2¢¢ to A7-H8 NOE correlation peaks are strong, Downloaded from https://academic.oup.com/nar/article-abstract/30/23/5244/1051861 by Ed 'DeepDyve' Gillespie user on 06 February 2018 Nucleic Acids Research, 2002, Vol. 30, No. 23 5249 which is another indication of the presence of a Watson±Crick the Watson±Crick model are also present in the molecular base pair. An A-T Hoogsteen base pair would be characterized dynamics simulations described below. No other a2 proteins by strong imino thymine H3 to adenine H8 NOE cross-peak contact the Hoogsteen base pair. These potentially unfavor- correlations (2), in contrast to the observed strong imino able contacts could favor a Hoogsteen conformation for the thymine H3 to adenine H2 NOE cross-peak correlations that A7-T37 base pair. In contrast, the A29-T15 base pair that is are consistent with Watson±Crick base pairing. related to A7-T37 by pseudo-two-fold symmetry of the a2 The A7-H8 and A7-H2 resonances for the 16 bp DNA binding sites in the DNA adopts the typical Watson±Crick construct were reliably assigned using a combination of the geometry. The A29-T15 base pair is embedded within the 1 13 2D NOESY in H O and natural abundance H- C HMQC in same sequence as the A7-T37 base pair, differing only in D O experiments, which also allowed for the assignment of 12 the absence of a non-speci®cally bound homeodomain out of 16 bp. The A7-H2 assignments were further con®rmed counterpart to a2D. by a strong NOE correlation between A7-H2 and A6-H2 (from Overall, the A7-T37 base pair in the Hoogsteen conform- base pair A6-T38). Also, we observed no detectable A7-H8 to ation has several favorable interactions within the DNA T37-H3 correlation that would be expected in the Hoogsteen around the Hoogsteen base pair, and between the non- base pair. As shown in Figure 3B, the strong NOE correlation speci®cally bound a2D protein and the A7-T37 base pair, observed between A7-H2 and imino T37-H3 is consistent with that would not be present were this base pair in the A-T Watson±Crick base pair formation. This, together with Watson±Crick conformation. Besides the protein±DNA inter- the lack of a strong A7-H1¢ to A7-H8 NOE correlation actions that favor the Hoogsteen base pair, interactions with provides strong evidence for an anti conformation for the A7 adjacent base pairs may help stabilize the Hoogsteen base pair. base in solution. We therefore conclude that the 16 bp DNA In addition to the hydrogen bonds that constitute the fragment itself does not have a propensity to form a Hoogsteen Hoogsteen base pair interaction, the syn conformation of base pair in solution. base A7 and negative roll of base T37 allow the adenine 7 N6 While analyzing the NOESY-H O peaks, we noticed that group to make an additional hydrogen bond with thymine 36 the NOE peaks of some adenine H2 protons were broad and O4 in the adjacent A8-T36 base pair (Fig. 2B). This intra-base shifted up®eld. The adenine H2 proton NOE peaks were pair hydrogen bond does, however, come at the cost of base 1 13 partially assigned with the help of the H- CHMQC pair hydrogen bonds that are lengthened beyond 3.2 A in the spectrum. One of the adenine H2 peaks, the A7-H2 peak, is T6-A38 and A8-T36 base pairs, resulting in only one very broad and clearly shifted up®eld relative to the other H2 hydrogen bond between them instead of the two hydrogen protons by at least 0.2 p.p.m. (Fig. 3C). This broad, upshifted bonds in a normal A-T base pair. The Hoogsteen base pair may peak is characteristic of the adenine H2 proton of TpA base be further stabilized by favorable base stacking interactions doubles. The adenine bases of TpA base doubles often have between the A7 and T6 bases, and the A8 and T9 bases, where high mobility about the glycosidic c bond, resulting in the the six-membered rings of the adenine and the thymine stack broad peak seen in this and other NMR experiments (35). The favorably on one another (Fig. 2B). Thus, several favorable high mobility of the TpA base double at T6-A7 may allow the interactions within the DNA and between the non-speci®cally A7 base to ¯ip more easily than in other base doublets, bound a2D protein and DNA could preferentially stabilize the increasing the probability of forming the Hoogsteen base pair Hoogsteen base pair over the Watson±Crick base pair. upon formation of the a2±DNA complex containing both Energetics calculations and molecular dynamics speci®cally and non-speci®cally bound proteins that is simulations of the protein±DNA system observed in this structure. Molecular dynamics simulations were used to study the Protein±DNA and base±base interactions that may in¯uence of the a2 homeodomains on the stability of the stabilize the Hoogsteen base pair Hoogsteen base pair. The system studied by molecular The fact that the Hoogsteen base pair does not form in the dynamics consisted of either 16 bp DNA alone, the DNA absence of bound protein suggests that the a2 proteins must with the speci®cally bound a2B homeodomain, or DNA with make favorable contacts with the DNA in the crystal that the speci®cally bound a2B and the non-speci®cally bound contribute to the formation of the Hoogsteen base pair, A7- a2D homeodomains. The DNA in the simulations was 1 bp T37. In the crystal structure, Arg4 of the non-speci®cally shorter than the oligonucleotide duplex used in the NMR bound a2D monomer makes van der Waals contacts with the experiments (Fig. 1D). Each system was fully solvated and A7 base and the sugar±phosphate backbone of bases T6 and neutralized with sodium ions. The dynamics simulations were A7 (Fig. 2C). This van der Waals contact may stabilize the run for 1 ns using Langevin dynamics (22,23) implemented in Hoogsteen base pair by preventing the sugar±phosphate CHARMM (24). Fluctuations of the dihedral angles and ± + backbone from moving out to the gauche /gauche conform- distances are used to estimate free energies (not on an absolute ation for the a and g torsion angles. When the A7-T37 base scale; see Materials and Methods), not the relative height of pair is modeled as a Watson±Crick base pair, several the free energy barriers. unfavorable steric clashes are observed between the a2D The presence of the non-speci®c a2D protein minimizes the molecule and the A7 base. In the hypothetical Watson±Crick free energy of the base pair at c =52°, as expected for a base pair model, the A7 C3¢ deoxyribose atom and the a2D Hoogsteen base pair (Fig. 4A). Additionally, the distribution Gly5 main chain N atoms are only 2.2 A from one another. of energies is much sharper in the presence of the a2D protein, Additionally, the C8 atom of an A7 base modeled in the anti with a potential well width of 5° (compared to >20° without conformation would be within 2.8 A of the a2D Asn47 ND2 a2D), indicating that the Hoogsteen base pair is more stable in amino group (Fig. 2C). Many of these unfavorable contacts in the presence of the a2D protein. The large ¯uctuations of the Downloaded from https://academic.oup.com/nar/article-abstract/30/23/5244/1051861 by Ed 'DeepDyve' Gillespie user on 06 February 2018 5250 Nucleic Acids Research, 2002, Vol. 30, No. 23 Figure 4. (A) Energetics of the A7 base calculated from glycosidic angle c in molecular dynamics simulations. The presence of bound a2 proteins stabilizes the c angle at 52°, while free DNA is free to rotate over a broad range. (B) The ¯uctuations of the c angle around the A7 base (red box) decrease signi®cantly when the a2 proteins are bound to the DNA. DNA in the region of the A7 base suggest a possible low complexes (such as triostin A±DNA) with underwound DNA transition energy barrier for ¯ipping the A7 base in the DNA (1), in DNA severely distorted by binding of the transcription in the absence of bound a2 proteins. factor TBP (3) and at the ends of DNA oligonucleotides used Molecular dynamics simulations con®rm many of the in crystallization (5). In all of these cases, the base stacking contacts between the non-speci®cally bound a2D protein energy barrier to forming the Hoogsteen base pair rather than and the DNA in the crystal structure, as well as revealing the more typical Watson±Crick base pair is presumably additional stabilizing contacts that favor the Hoogsteen base lowered by the intercalation of drugs, bending and unwinding pair. The dynamics simulations show that there are many of the DNA, or the presence of free, unpaired bases, all of contacts between the a2D Arg4 residue and the sugar± which have the ¯exibility to allow the ¯ipping of a purine base phosphate backbone at base A7. Additional hydrogen bonds from the normal anti conformation to the syn conformation. not seen in the crystal structure form between a2D Arg4 and To our knowledge, a Hoogsteen base pair has not previously the O3¢ and O4¢ atoms of base A7 during the simulations. been seen in oligonucleotide structures that lack any of these Furthermore, a simulation run with the A7-T37 base pair in the kinds of distortions. Watson±Crick conformation shows that the peptide backbone The crystal structure described here contains both intra-base of the a2D N-terminal arm moves away from the minor pair interactions and protein±DNA interactions that could groove to avoid steric clashes, leading to loss of hydrogen stabilize the Hoogsteen base pair. However, in the absence of bonds and van der Waals contacts between the a2D protein bound a2 proteins, the DNA does not contain a Hoogsteen and DNA. These steric clashes are similar to the clashes base pair, as shown in solution NMR studies of the free DNA. observed when the A7-T37 base pair was modeled in the The NMR studies instead show a broadened A7-H2 peak that Watson±Crick con®guration in the crystal structure. may indicate that the A7 base is free to rotate about the The ¯uctuations of the glycosidic c angle at the A7 base glycosidic c bond. We speculate that the ¯exible A7 base may pair in the molecular dynamics simulations show that the TpA spontaneously ¯ip about the c bond to the unusual syn base doubles, which show characteristic large ¯uctuations of conformation, leading to formation of the Hoogsteen base the c angle, may be contributing to the formation of the pair. Molecular dynamics simulations of the DNA and Hoogsteen base pair. In addition to the broadened peaks protein±DNA systems con®rm the stability of the complex observed at A7-H2 in the NMR experiments, the simulations of a2 proteins bound to DNA containing the Hoogsteen base also show a 16° root mean square ¯uctuation of the A7 c pair. The dynamics simulations also show that the A7 base angle. Furthermore, stacking energy calculations using the indeed has an inherent ¯exibility, with signi®cant ¯uctuations AMBER force ®eld yield low stabilization energies for base about its glycosidic c bond angle. From these experiments, pairs between base pairs T4-A40 and T10-A34 (Fig. 4B). however, we cannot determine whether the Hoogsteen base These low stabilization energies and large ¯uctuations may be pair is a result of the binding of the non-speci®cally bound favorable for ¯ipping the adenine base in solution, and the a2D protein or whether the presence of the Hoogsteen base Hoogsteen base pair may not subsequently revert back to the pair may stabilize the a2D contacts with DNA. Watson±Crick base pair due to the favorable a2D±DNA and It appears that the energetics of the particular con®guration intra-DNA contacts for the Hoogsteen base pair. of proteins and DNA in the present structure gives rise to this unusual base pair, despite the absence of DNA distortions previously observed to be required for Hoogsteen base pair DISCUSSION formation within duplex DNA. A combination of van der Implications for protein±DNA interactions Waals interactions, hydrogen bonds and base stacking inter- actions may allow the stabilization of the A7-T37 Hoogsteen We have observed a Hoogsteen base pair embedded in the structure of undistorted dsDNA. Hoogsteen base pairs have base pair by the a2 proteins in this structure. These base previously been observed in crystal structures of drug±DNA ¯uctuations of the A7 base between syn and anti may happen Downloaded from https://academic.oup.com/nar/article-abstract/30/23/5244/1051861 by Ed 'DeepDyve' Gillespie user on 06 February 2018 Nucleic Acids Research, 2002, Vol. 30, No. 23 5251 2. Gilbert,D.E., van der Marel,G.A., van Boom,J.H. and Feigon,J. (1989) at a very low frequency due to base ¯ipping out of the DNA Unstable Hoogsteen base pairs adjacent to echinomycin binding sites and reinsertion into the DNA. Such base ¯ipping has been within a DNA duplex. Proc. Natl Acad. Sci. USA, 86, 3006±3010. observed and predicted to require an energy of 25 kcal/mol 3. Patikoglou,G.A., Kim,J.L., Sun,L., Yang,S.H., Kodadek,T. and (36), much less than the 100 kcal/mol or greater predicted to Burley,S.K. (1999) TATA element recognition by the TATA box- binding protein has been conserved throughout evolution. Genes Dev., ¯ip the A7 base within DNA (data not shown). One such base 13, 3217±3230. ¯ipping event, a Hoogsteen base pair that appears to require 4. Hoogsteen,K. (1963) The crystal and molecular structure of a hydrogen- stabilization by an a2 protein bound to the DNA, has bonded complex between 1-methylthymine and 9-methyladenine. Acta been observed in the current crystal structure. Because the Crystallogr., 16, 907±916. Hoogsteen base pair is only present in the crystal structure and 5. Rice,P.A., Yang,S., Mizuuchi,K. and Nash,H.A. (1996) Crystal structure of an IHF-DNA complex: a protein-induced DNA U-turn. Cell, 87, not in the DNA alone in NMR experiments, we cannot 1295±1306. determine whether the presence of the Hoogsteen base pair 6. Wolberger,C., Vershon,A.K., Liu,B., Johnson,A.D. and Pabo,C.O. (1991) could in¯uence the binding af®nity of the a2 protein for the Crystal structure of a MAT alpha 2 homeodomain-operator complex DNA. suggests a general model for homeodomain-DNA interactions. Cell, 67, Our observation of a Hoogsteen base pair within otherwise 517±528. 7. Aishima,J. and Wolberger,C. (2002) Crystal structure of the MATalpha2 undistorted B-DNA that is either induced or stabilized by homeodomain-DNA complex with nonspeci®cally bound protein±DNA contacts raises the possibility that Hoogsteen homeodomains. Proteins Struct. Funct. Genet., in press. base pairs could occur within cellular DNA and play a role in 8. Hodel,A., Kim,S.-H. and Brunger,A.T. (1992) Model bias in protein±DNA interactions. The particular con®guration of macromolecular crystal structures. Acta Crystallogr., A48, 851±858. 9. Brunger,A.T. (1992) X-PLOR, Version 3.1. A System for X-ray proteins and DNA reported here is undoubtedly in¯uenced by Crystallography and NMR, 3.84 Edn. Yale University Press, the non-physiological concentrations of a2 protein in the New Haven, CT. crystal drops and does not re¯ect the arrangement of binding 10. Brunger,A.T., Adams,P.D., Clore,G.M., DeLano,W.L., Gros,P., sites found upstream of genes regulated by a2 in vivo. Grosse-Kunstleve,R.W., Jiang,J.S., Kuszewski,J., Nilges,M., Pannu,N.S. Nevertheless, it is possible that the local conditions under et al. (1998) Crystallography & NMR System: a new software suite for macromolecular structure determination. Acta Crystallogr., D54, which the present Hoogsteen base pair forms could be 905±921. duplicated for other proteins at in vivo regulatory sites. The 11. Kleywegt,G.J. and Jones,T.A. (1996) xdlMAPMAN and presence of multiple overlapping binding sites is common in xdlDATAMANÐprograms for reformatting, analysis and manipulation chromosomal DNA and could give rise to a con®guration of of biomacromolecular electron-density maps and re¯ection data sets. proteins analogous to that observed in the crystal. The open Acta Crystallogr., D52, 826±828. 12. Brunger,A.T. (1992) The free R value: a novel statistical quantity for question is whether such an arrangement of proteins either assessing the accuracy of crystal structures. Nature, 355, 472±474. binds preferentially to transiently formed Hoogsteen base 13. Delaglio,F., Grzesiek,S., Vuister,G.W., Zhu,G., Pfeifer,J. and Bax,A. pairs or favors Hoogsteen base pair formation in order to form (1995) NMRPipe: a multidimensional spectral processing system based optimal interactions. Since we were unable to detect measur- on UNIX pipes. J. Biomol. NMR, 6, 277±293. able Hoogsteen base pair formation in free DNA, it was not 14. Johnson,B.A. and Blevins,R.A. (1994) NMRview: a computer program for the visualization and analysis for NMR data. J. Biomol. NMR, 4, possible to assess the energetic contribution of Hoogsteen base 603±614. pair formation by directly comparing the DNA-binding 15. Jeener,J., Maier,B.H., Bachmann,P. and Ernst,R.R. (1979) Investigation af®nity of the a2 homeodomain for sites containing of exchange processes by two-dimensional NMR spectroscopy. J. Chem. Hoogsteen versus Watson±Crick base pairs. However, the Phys., 71, 4546±4553. 16. Macura,S. and Ernst,R.R. (1980) Elucidation of cross relaxation in absence of DNA distortion and the relatively typical array of liquids by two-dimensional NMR-spectroscopy. Mol. Phys., 41, 95±117. protein±DNA contacts suggests that the conditions that favor 17. Piotto,M., Saudek,V. and Sklenar,V. (1992) Gradient-tailored excitation Hoogsteen base pair formation could be replicated in a cellular for single-quantum NMR spectroscopy of aqueous solutions. J. Biomol. context. These observations raise the intriguing possibility that NMR, 2, 661±665. Hoogsteen base pair formation could potentially play a role in 18. Greisinger,C., Otting,G., Wuthrich,K. and Ernst,R.R. (1988) Clean TOCSY for 1H spin system identi®cation in macromolecules. J. Am. the binding of proteins to undistorted B-DNA, although Chem. Soc., 110, 7870. further investigation will be required in order to determine 19. Bax,A. and Davis,D.G. (1985) MLEV-17-based two-dimensional whether this indeed occurs in the cell. homonuclear magnetization transfer spectroscopy. J. Magn. Reson., 65, 355±360. 20. Braunschweiler,L. and Ernst,R.R. (1983) Coherence transfer by isotropic ACKNOWLEDGEMENTS mixingÐapplication to proton correlation spectroscopy. J. Magn. Reson., 53, 521±528. We thank W. Olson for advice and discussions and for 21. Bax,A. and Subramanian,S. (1986) Sensitivity-enhanced two- allowing us to preview the 3DNA program, M. Summers for dimensional heteronuclear shift correlation NMR-spectroscopy. J. Magn. Reson., 67, 565±569. generously allowing the use of equipment for the NMR 22. Schlick,T. (2001) Time-trimming tricks for dynamic simulations: experiments, and C. Garvie, A. VanDemark, A. Ke, splitting force updates to reduce computational work. Structure, 9, N. LaRonde-LeBlanc, P. Minary and R. Campbell for helpful R45±R53. discussions. This work was supported by NSF grant 23. Barth,E. and Schlick,T. (1998) Overcoming stability limitations in MCB9808412 (C.W.). biomolecular dynamics. I. Combining force splitting via extrapolation with Langevin dynamics in LN. J. Chem. Phys., 109, 1617±1632. 24. Brooks,B.R., Bruccoleri,R.E., Olafson,B.D., States,D.J., Swaminathan,S. and Karplus,M. (1983) CHARMM: a program for macromolecular REFERENCES energy, minimization and dynamics calculations. J. Comp. Chem., 4, 1. Wang,A.H., Ughetto,G., Quigley,G.J., Hakoshima,T., van der Marel,G.A., 187±217. van Boom,J.H. and Rich,A. (1984) The molecular structure of a 25. Cornell,W.D., Cieplak,P., Bayley,C.I., Gould,I.R., Merz,K.M., DNA-triostin A complex. Science, 225, 1115±1121. Ferguson,D.M., Spellmeyer,D.C., Fox,T., Caldwell,J.W. and Downloaded from https://academic.oup.com/nar/article-abstract/30/23/5244/1051861 by Ed 'DeepDyve' Gillespie user on 06 February 2018 5252 Nucleic Acids Research, 2002, Vol. 30, No. 23 Kollman,P.A. (1995) A second generation force ®eld for the simulation 32. Berman,H.M., Westbrook,J., Feng,Z., Gilliland,G., Bhat,T.N., of proteins, nucleic acids and organic molecules. J. Am. Chem. Soc., 117, Weissig,H., Shindyalov,I.N. and Bourne,P.E. (2000) The Protein Data 5179±5197. Bank. Nucleic Acids Res., 28, 235±242. 26. Qian,X., Strahs,D. and Schlick,T. (2001) Dynamic simulations of 13 tata 33. McRee,D.E. (1999) XtalView/X®tÐa versatile program for variants re®ne kinetic hypotheses of sequence/activity relationships. manipulating atomic coordinates and electron density. J. Struct. Biol., J. Mol. Biol., 308, 681±703. 125, 156±165. 27. Strahs,D. and Schlick,T. (2000) A-tract bending: insights into 34. Wuthrich,K. (1986) NMR of Proteins and Nucleic Acids. John Wiley & experimental structures by computational models. J. Mol. Biol., 301, Sons, New York, NY. 643±663. 35. McAteer,K. and Kennedy,M.A. (2000) NMR evidence for base dynamics 28. Kottalam,J. and Case,D.A. (1988) Dynamics of ligand escape from the at all TpA steps in DNA [In Process Citation]. J. Biomol. Struct. Dyn., heme pocket of myoglobin. J. Am. Chem. Soc., 110, 7690±7697. 17, 1001±1009. 29. Li,T., Stark,M.R., Johnson,A.D. and Wolberger,C. (1995) Crystal 36. Chen,Y.Z., Mohan,V. and Griffey,R.H. (2000) Spontaneous base ¯ipping structure of the MATa1/MAT alpha 2 homeodomain heterodimer bound in DNA and its possible role in methyltransferase binding. Phys. Rev., to DNA. Science, 270, 262±269. E62, 1133±1137. 30. Tan,S. and Richmond,T.J. (1998) Crystal structure of the yeast 37. Evans,S.V. (1993) SETOR: hardware-lighted three-dimensional solid MATalpha2/MCM1/DNA ternary complex. Nature, 391, 660±666. model representations of macromolecules. J. Mol. Graphics, 11, 31. Qian,Y.Q., Billeter,M., Otting,G., Muller,M., Gehring,W.J. and 134±138, 127±138. Wuthrich,K. (1989) The structure of the Antennapedia homeodomain 38. Delano,W.L. (2002) The PyMOL Molecular Graphics System. Delano determined by NMR spectroscopy in solution: comparison with Scienti®c, San Carlos, CA. prokaryotic repressors. Cell, 59, 573±580. [Erratum (1990) Cell, 61, 39. Humphrey,W., Dalke,A. and Schulten,K. (1996) VMDÐvisual 548.] molecular dynamics. J. Mol. Graphics, 14, 33±38. Downloaded from https://academic.oup.com/nar/article-abstract/30/23/5244/1051861 by Ed 'DeepDyve' Gillespie user on 06 February 2018 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Nucleic Acids Research Oxford University Press

A Hoogsteen base pair embedded in undistorted B‐DNA

Loading next page...
 
/lp/oxford-university-press/a-hoogsteen-base-pair-embedded-in-undistorted-b-dna-ntbr4Mvlow

References (40)

Publisher
Oxford University Press
ISSN
0305-1048
eISSN
1362-4962
DOI
10.1093/nar/gkf661
Publisher site
See Article on Publisher Site

Abstract

5244±5252 Nucleic Acids Research, 2002, Vol. 30, No. 23 ã 2002 Oxford University Press A Hoogsteen base pair embedded in undistorted B-DNA 1 2 3 4 2 4 5 2 4 5 , , , , , Jun Aishima , Rossitza K. Gitti , Joyce E. Noah , Hin Hark Gan , Tamar Schlick 1 2 , , and Cynthia Wolberger * 1 2 Department of Biophysics and Biophysical Chemistry and Howard Hughes Medical Institute, Johns Hopkins University School of Medicine, Baltimore, MD 21205-2185, USA, Department of Chemistry and Biochemistry, University of Maryland Baltimore County, Baltimore, MD 21250, USA, Department of Chemistry and Courant Institute of Mathematical Sciences, New York University, 31 Washington Place, Room 1021 Main, New York, NY 10003, USA Received July 11, 2002; Revised and Accepted October 3, 2002 PDB no. 1K61 ABSTRACT characterized by hydrogen bonds between the side of the purine base that faces the major groove and the Watson±Crick Hoogsteen base pairs within duplex DNA typically base pairing face of the pyrimidine. In B-DNA, formation of a are only observed in regions containing signi®cant Hoogsteen base pair would require rotation of the purine distortion or near sites of drug intercalation. We base about the glycosidic c bond from the anti to the syn report here the observation of a Hoogsteen base conformation and, in the case of a guanine-cytosine base pair, pair embedded within undistorted, unmodi®ed protonation of the N3 of cytosine. Hydrogen bonds in the B-DNA. The Hoogsteen base pair, consisting of a Hoogsteen base pair are formed between the purine N7 to the syn adenine base paired with an anti thymine base, pyrimidine N3 and either the adenine N6 to the thymine O6 or the guanine O4 to the cytidine N4 (4). is found in the 2.1 A resolution structure of the Hoogsteen base pairs have been observed in several crystal MATa2 homeodomain bound to DNA in a region structures of distorted double-stranded B-DNA. DNA com- where a speci®cally and a non-speci®cally bound plexed with intercalating binding drugs, such as triostin-A, homeodomain contact overlapping sites. NMR contain Hoogsteen base pairs in regions of underwound studies of the free DNA show no evidence of B-DNA (1). Hoogsteen base pairs can also occur in regions of Hoogsteen base pair formation, suggesting that DNA that are highly distorted by bound protein. In the protein binding favors the transition from a structure of TATA-binding protein (TBP) bound to DNA (3), a Watson±Crick to a Hoogsteen base pair. Molecular Hoogsteen base pair is observed in the region of DNA dynamics simulations of the homeodomain±DNA underwinding and intercalation by a phenylalanine side chain complex support a role for the non-speci®cally from TBP. Finally, Hoogsteen base pairs have also been bound protein in favoring Hoogsteen base pair observed at free ends of end-to-end stacked oligonucleotides, formation. The presence of a Hoogsteen base pair in such as in the structure of integration host factor bound to the crystal structure of a protein±DNA complex DNA (5). The presence of distortions or ¯exible end regions in raises the possibility that Hoogsteen base pairs B-form DNA may lower the energy barrier for rotation of the could occur within duplex DNA and play a hitherto purine base about the glycosidic c bond by changing the base unrecognized role in transcription, replication and stacking arrangements or the helical properties of the DNA. The lower energy barrier may allow the purine base to rotate other cellular processes. from the normal anti conformation to the syn conformation, leading to the formation of hydrogen bonds characteristic of a INTRODUCTION Hoogsteen base pair. We report here the observation of a Hoogsteen base pair Most duplex DNA, whether in A-, B- or Z-form, is composed embedded within undistorted B-DNA. The Hoogsteen base of complementary strands that associate solely through pair, formed by a syn adenine base and an anti thymine base, Watson±Crick base pairing. In a small number of DNA occurs within a 2.1 A resolution crystal structure containing structures containing intercalating drugs (1,2) or pronounced four MATa2 homeodomains bound to DNA. The 21 bp DNA protein-induced distortion (3), Hoogsteen base pairs have been used for the crystal structure contains two binding sites for the found embedded in duplex DNA. The Hoogsteen base pair MATa2 homeodomain and was previously used in a crystal geometry, which was ®rst observed in crystal structures of structure of the MATa2 homeodomain±DNA complex (6). In monomeric adenine and thymine base derivatives (4), is *To whom correspondence should be addressed at 7 Johns Hopkins University School of Medicine, Baltimore, MD 21205-2185, USA. Tel: +1 410 955 0728; Fax: +1 410 614 8648; Email: cwolberg@jhmi.edu Present address: Jun Aishima, Lawrence Berkeley National Laboratory, 1 Cyclotron Road, Mail Stop 4-230, Berkeley, CA 94720, USA Downloaded from https://academic.oup.com/nar/article-abstract/30/23/5244/1051861 by Ed 'DeepDyve' Gillespie user on 06 February 2018 Nucleic Acids Research, 2002, Vol. 30, No. 23 5245 Figure 1. (A) Crystal structure of the a2 homeodomain contains four a2 proteins bound to two a2 binding sites in the DNA. Base pair A7-T37 (red box) is a Hoogsteen base pair. Figure produced with SETOR (37). (B) The oligonucleotide duplex used in the crystal structure contains two a2 binding sites (box) with a5¢ overhanging base at each end. (C) The DNA fragment used in the NMR experiments is blunt-ended and contains ®ve fewer base pairs. (D) The DNA fragment used in the molecular dynamics simulations. the current structure, two of the MATa2 homeodomains bind Phasing and re®nement to the MATa2 binding sites, while the other two MATa2 Full details of the X-ray data collection, phasing and proteins are bound non-speci®cally to the DNA (7). re®nement procedures are described elsewhere (7). Initial Simulated annealing omit maps (8) clearly show the trials of molecular replacement (using a model of two a2 unambiguous presence of the Hoogsteen base pair in the proteins bound to DNA from the original structure; 6) on the crystal structure. The Hoogsteen base pair appears to be 2.1 A resolution data set were unsuccessful, resulting in stabilized by an extra inter-base pair hydrogen bond and base distorted DNA during re®nement with Xplor (9) and CNS stacking within the DNA, as well as by contacts made by one (10). Subsequently, electron density maps calculated with of the non-speci®cally bound MATa2 homeodomains with multiple isomorphous replacement phases showed that the the sugar±phosphate backbone of the syn adenine base. None molecular replacement solution was correct. However, two of the observed stabilizing contacts are out of the ordinary, unexpected, additional a2 molecules were visible in the suggesting that Hoogsteen base pairs and other non- bones-skeletonized electron density map (11). After the two Watson±Crick base pairs may occur more commonly within additional a2 molecules were placed in the density, re®nement undistorted DNA than has previously been thought. proceeded smoothly with CNS. For all data to 2.1 A resolution, the ®nal R (12) and R factor are 27.8 and free 22.2%, respectively. MATERIALS AND METHODS Sample preparation Protein and oligonucleotide synthesis and The 16 bp DNA fragment used for the NMR studies is 5 bp puri®cation shorter than the crystallization oligonucleotide and does not The procedure for the puri®cation of the a2 homeodomain and contain base pairs 17:27 to 21:23 (Fig. 1C). Single-stranded DNA are described elsewhere (7). Brie¯y, the a2 homeo- oligonucleotides were puri®ed twice over a Dynamax domain (residues 132±191) was synthesized by the Fmoc solid PureDNA column (Rainin) with an acetonitrile gradient in state peptide synthesis method and puri®ed by C4 reverse 0.1 M triethylamine acetate (pH 7). The trityl group on the phase column (Vydac). Mass spectrometry was used to verify oligonucleotide was removed during the second run by 0.5% the molecular weight of the protein. The DNA was TFA. Peak fractions were pooled, dialyzed into 10 mM synthesized at the HHMI±Keck Biopolymer Facility at Yale triethylamine bicarbonate (pH 7) and annealed. Annealed University and puri®ed with a PureDNA reverse phase column oligonucleotides were quantitated by spectroscopic absorb- (Rainin). Oligonucleotides were combined in a 1:1 molar ratio ance at 260 nm, lyophilized and stored at ±80°C. NMR studies and annealed by heating to 90°C, then cooling overnight. Both were performed on unlabeled DNA samples, redissolved after protein and DNA were quantitated by spectroscopic methods. HPLC into 25 mM sodium phosphate and 25 mM NaCl buffer The DNA concentration was quantitated with the formula (pH 7.2) in D O. DNA samples for exchangeable proton 1OD unit = 50 mg/ml. The protein concentration was detection were transferred in H O by lyophilization to a ®nal 260 2 veri®ed by the Lowry method (Sigma). Protein and DNA were concentration of 2.0 mM DNA in 25 mM sodium phosphate combined at a molar ratio of 1.8:1. and 25 mM NaCl buffer (pH 6.4) in 95% H O/5% D O. 2 2 Downloaded from https://academic.oup.com/nar/article-abstract/30/23/5244/1051861 by Ed 'DeepDyve' Gillespie user on 06 February 2018 5246 Nucleic Acids Research, 2002, Vol. 30, No. 23 Table 1. Twist and rise for the DNA in the crystal NMR spectroscopy structure and standard B-DNA NMR experiments were carried out at 25°C on General Base pair Twist Rise Electric OMEGA PSG and Bruker DMX 600 MHz spectro- meters equipped with x,y,z-shielded gradient triple resonance C2-G42 36.38 3.88 probes. NMR data were processed with NMRPipe (13) and A3-T41 25.81 3.59 T4-A40 45.44 2.43 analyzed with NMRVIEW (14). NOE data (mix time t = G5-C39 26.93 3.09 150 ms) were obtained from 2D NOESY (15,16). Water T6-A38 38.66 3.04 suppression was achieved with WATERGATE and ®eld A7-T37 32.73 3.89 gradient pulses (17) or presaturation pulses during the A8-T36 32.38 3.29 relaxation delay for DNA samples in H O. 2D TOCSY data T9-A35 31.06 3.46 T10-A34 39.12 2.77 were obtained in D O with a 75 ms clean-MLEV-17 mixing C11-G33 44.77 3.22 1 13 period (18±20). H- C HMQC (21) was recorded in natural A12-T32 27.94 3.85 abundance. T13-A31 35.29 3.13 T14-A30 35.50 3.38 Energetics and dynamics calculations T15-A29 39.41 2.98 A16-T28 27.29 3.32 The a2B and a2D proteins were used for the dynamics C17-G27 36.56 3.18 studies. Structures of the DNA alone, a2B±DNA complex A18-T26 30.60 3.44 and a2B±a2D±DNA complex were studied by ®rst solvating C19-G25 38.06 3.35 with water, then neutralizing with sodium ions. The systems G20-C24 26.69 3.24 C21-G23 were propagated using a multiple timestepping integrator for Average 34.2 3.29 Langevin dynamics (22,23) as implemented in the CHARMM B-DNA 38.6 3.38 molecular modeling package, v.26a2 (24). The systems were minimized and equilibrated to 300K for 75 ps before starting production runs. The molecular simula- each containing residues 132±191 of the MATa2 protein tions and energy minimization use the Cornell et al. (25) force [homeodomain residues 4±60 as numbered by Qian et al. ®elds as implemented in the CHARMM package. Energy (31)], bound to a 21 bp fragment of duplex DNA that contains minimization of the system was done using the steepest two a2 binding sites. Two of the a2 homeodomains, a2A and descent method and an adopted-basis Newton±Raphson a2B (Fig. 1A), bind DNA at the two a2 binding sites, while protocol in CHARMM. (See 26,27 for further details.) The the other two a2 proteins bind DNA in an atypical fashion and free energy, f, was estimated as f(c) = ±0.59 lnn(c) (kcal/mol), are considered non-speci®cally bound proteins. One of the where n(c) is the number of counts at c (28). non-speci®cally bound homeodomains, a2D, binds DNA in a previously unobserved manner, resulting in a signi®cant Coordinates reorientation of helix 3 relative to the DNA that disrupts The atomic coordinates and structure factors have been normal side chain±major groove contacts. Instead, other side deposited in the Protein Data Bank (accession number 1K61). chains that are not typically involved in homeodomain±DNA interactions are found to mediate base contacts with the major groove. Despite the overall change in the homeodomain RESULTS docking and major groove contacts, there is little change in the A Hoogsteen base pair is seen in a homeodomain/DNA contacts with the sugar±phosphate backbone. The other non- crystal structure speci®cally bound a2 homeodomain, a2C, binds near the junction of two stacked, crystallographically related DNA Like other homeodomains, the MATa2 homeodomain (abbre- fragments. viated as a2) consists of a compact three a-helix domain with The DNA reported in the current crystal structure contains a a ¯exible N-terminal arm that becomes ordered upon binding total of 21 bp (Fig. 1B). Twenty base pairs are contained DNA. Homeodomains make base-speci®c contacts with the within the duplex and one base pair is formed by overhanging major groove of the DNA using residues in the third a-helix, 5¢ bases from adjacent complexes, which stack end-to-end to while the N-terminal arm contacts bases in the minor groove. form a pseudocontinuous DNA helix. The DNA is B-form In addition to base-speci®c contacts, residues in the loop throughout, with sugar puckers, axial rises and twist all between helices 1 and 2, as well as residues in helix 3, contact characteristic of B-DNA (Table 1). the sugar±phosphate backbone. Previous crystal structures of All of the DNA was initially modeled with Watson±Crick the a2 homeodomain bound to DNA (PDB entry 1APL) (6), base pairs. The 2F ± F and F ± F electron density maps ®t the a1±a2 heterodimer bound to DNA (PDB entry 1YRN) o c o c the Watson±Crick DNA model well except at base pair A7- (29) and the MCM1/a2 heterotetramer bound to DNA (PDB T37. The difference density at A7-T37 could only be entry 1MNM) (30) show that the a2 homeodomain has the same structure and makes similar DNA contacts when bound accounted for by the rotation of the A7 base about the torsion to canonical a2 binding sites. angle c to the uncommon syn conformation, yielding an We recently determined a higher resolution structure of the A(syn)´T(anti) Hoogsteen base pair. Simulated annealing omit MATa2 homeodomain bound to DNA (7). This new, 2.1 A maps of the A7-T37 base pair con®rm the presence of the resolution structure was determined by a combination of Hoogsteen base pair at this position (Fig. 2A). To accommo- multiple isomorphous replacement and molecular replace- date the short C1¢±C1¢ distance across the base pair [8.5 A for ment. Brie¯y, the structure contains four a2 homeodomains, this base pair, as opposed to 10.5 A for Watson±Crick base Downloaded from https://academic.oup.com/nar/article-abstract/30/23/5244/1051861 by Ed 'DeepDyve' Gillespie user on 06 February 2018 Nucleic Acids Research, 2002, Vol. 30, No. 23 5247 Figure 2. The Hoogsteen base pair is stabilized by intra-DNA and protein±DNA interactions in the crystal structure of the a2 homeodomain±DNA complex. (A) Simulated annealing omit map of the A7-T37 base pair. The two hydrogen bonds between the N7 of base A7 and N3 of base T37, as well as between the N6 of A7 and the O4 of T37 characterize a Hoogsteen base pair. The same base pair modeled in the Watson±Crick con®guration (green) clearly does not ®t the electron density. Figure prepared with Pymol (38). (B) The A7 N6 group makes a bifurcated hydrogen bond with both T37 and T36. Base stacking inter- actions between T6 and A7, as well as A8 and T9, may contribute to stabilization of the Hoogsteen base pair. (C) The Arg4 residue of the a2D homeodomain packs against the sugar±phosphate backbone at bases A7 and T6. (B) and (C) prepared with VMD (39). pairs (4)], the a and g torsion angles (about bonds P±O5¢ and mistakenly modeled as Watson±Crick base pairs. Crystal C5¢±C4¢, respectively) in the sugar±phosphate backbone are in structures of proteins bound to straight DNA determined from + ± Ê the unusual gauche /gauche conformation about base A7. In crystals diffracting to 2.5 A resolution or better and with normal B-DNA, the a and g torsion angles are in the gauche / available structure factor ®les were examined (PDB entries gauche conformations. There are no changes in the conform- 1FJL, 1B72, 1DP7, 2CGP and 1LAT). 2F ± F and F ± F o c o c electron density maps were generated with CNS (10) and ation of the T37 phosphate backbone a or g torsion angles and viewed in x®t from the XtalView package (33). In no case was there are no other irregularities in the backbone of the DNA. density indicative of a Hoogsteen base pair seen in the DNA. The Hoogsteen base pair seen in this crystal structure appears to be a property of this protein±DNA complex and is In the absence of protein, the DNA in solution does not observed in different crystals grown from independent DNA contain a Hoogsteen base pair syntheses. In addition to the 5-iodouracil derivative crystal used for re®nement, we also examined the simulated Is the Hoogsteen base pair somehow intrinsic to this DNA annealing omit maps of the A7-T37 base pair using data sequence? To answer this question we used NMR spectro- from the native crystal, which extends to 2.4 A resolution. scopy to examine the structure of a 16 bp DNA fragment in Electron density maps calculated from the native data set also solution. This 16 bp DNA fragment (Fig. 1C) is a shorter reveal the presence of the Hoogsteen base pair, showing that version of the crystallized DNA and lacks the 5 bp farthest the 5-iodouracil substitution in the DNA does not affect the from the A7-T37 site, but it is suf®ciently long to retain local formation of the Hoogsteen base pair and that the Hoogsteen interactions that could potentially have induced formation of base pair occurs in different crystals of this protein±DNA the Hoogsteen base pair. In the crystal structure of the complex. In contrast, neither the DNA from the previous a2±DNA complex, only the base pairs immediately adjacent structure of the a1±a2±DNA complex nor the DNA in the to base pair A7-T37 make stabilizing contacts with the other a2 binding site in the current a2±DNA crystal structure Hoogsteen base pair. contains a Hoogsteen base pair. The Hoogsteen base pair We examined base conformations in solution by following therefore appears to be unique to this a2±DNA complex and to the standard sequential connectivities involving the sugar± the A7-T37 position within the DNA. base H1¢±H6/H8 and H2¢/H2¢¢±H6/H8 pathways for all 32 The unexpected observation of a Hoogsteen base pair in the bases using both 2D NOESY (nuclear Overhauser and present structure raised the possibility that the presence of exchange spectroscopy) and TOCSY (total correlation spec- Hoogsteen base pairs may have been missed in previous troscopy) experiments in D O (see Materials and Methods). If structures due to misinterpretation of ambiguous electron the A7 base is in the syn conformation in solution, there should density. We examined several structures of protein±DNA be a strong NOE between A7-H1¢ and A7-H8 as the distance complexes in the Protein Data Bank (32) to determine whether between these two protons is 1.5 A closer than in the anti any structures contained Hoogsteen base pairs that were conformation (34). The entire 2D NOESY-D O spectrum in Downloaded from https://academic.oup.com/nar/article-abstract/30/23/5244/1051861 by Ed 'DeepDyve' Gillespie user on 06 February 2018 5248 Nucleic Acids Research, 2002, Vol. 30, No. 23 Figure 3. NMR studies of the DNA clearly show a Watson±Crick base pair in solution. (A) The NOESY-D O spectrum was easily assignable in the H1¢±H6/ H8 region. The weak A7 H1¢±A7 H8 NOE correlation peak is consistent with A7 in the anti conformation. (B) The NOESY-H O spectrum was assignable in the amino±imino region. The A7 H2-T37 H3 NOE correlation (red box) is strong, consistent with a Watson±Crick base pair at A7-T37. (C) In the NOESY- D O spectrum, the A7-H2 peak is shifted up®eld and is broader than other adenine H2 peaks, consistent with a ¯exible A7 base. the H1¢±H6/H8 (Fig. 3A) and H2¢/H2¢¢±H6/H8 (data not conformation in solution at least 95% of the time, in contrast shown) regions was well resolved and easy to assign. to the syn A7 base conformation that was observed in the The presence of a weak NOE correlation peak between A7- crystal structure of the a2±DNA complex. Furthermore, the H1¢ and A7-H8 indicates that the A7 base is in an anti A7-H2¢/H2¢¢ to A7-H8 NOE correlation peaks are strong, Downloaded from https://academic.oup.com/nar/article-abstract/30/23/5244/1051861 by Ed 'DeepDyve' Gillespie user on 06 February 2018 Nucleic Acids Research, 2002, Vol. 30, No. 23 5249 which is another indication of the presence of a Watson±Crick the Watson±Crick model are also present in the molecular base pair. An A-T Hoogsteen base pair would be characterized dynamics simulations described below. No other a2 proteins by strong imino thymine H3 to adenine H8 NOE cross-peak contact the Hoogsteen base pair. These potentially unfavor- correlations (2), in contrast to the observed strong imino able contacts could favor a Hoogsteen conformation for the thymine H3 to adenine H2 NOE cross-peak correlations that A7-T37 base pair. In contrast, the A29-T15 base pair that is are consistent with Watson±Crick base pairing. related to A7-T37 by pseudo-two-fold symmetry of the a2 The A7-H8 and A7-H2 resonances for the 16 bp DNA binding sites in the DNA adopts the typical Watson±Crick construct were reliably assigned using a combination of the geometry. The A29-T15 base pair is embedded within the 1 13 2D NOESY in H O and natural abundance H- C HMQC in same sequence as the A7-T37 base pair, differing only in D O experiments, which also allowed for the assignment of 12 the absence of a non-speci®cally bound homeodomain out of 16 bp. The A7-H2 assignments were further con®rmed counterpart to a2D. by a strong NOE correlation between A7-H2 and A6-H2 (from Overall, the A7-T37 base pair in the Hoogsteen conform- base pair A6-T38). Also, we observed no detectable A7-H8 to ation has several favorable interactions within the DNA T37-H3 correlation that would be expected in the Hoogsteen around the Hoogsteen base pair, and between the non- base pair. As shown in Figure 3B, the strong NOE correlation speci®cally bound a2D protein and the A7-T37 base pair, observed between A7-H2 and imino T37-H3 is consistent with that would not be present were this base pair in the A-T Watson±Crick base pair formation. This, together with Watson±Crick conformation. Besides the protein±DNA inter- the lack of a strong A7-H1¢ to A7-H8 NOE correlation actions that favor the Hoogsteen base pair, interactions with provides strong evidence for an anti conformation for the A7 adjacent base pairs may help stabilize the Hoogsteen base pair. base in solution. We therefore conclude that the 16 bp DNA In addition to the hydrogen bonds that constitute the fragment itself does not have a propensity to form a Hoogsteen Hoogsteen base pair interaction, the syn conformation of base pair in solution. base A7 and negative roll of base T37 allow the adenine 7 N6 While analyzing the NOESY-H O peaks, we noticed that group to make an additional hydrogen bond with thymine 36 the NOE peaks of some adenine H2 protons were broad and O4 in the adjacent A8-T36 base pair (Fig. 2B). This intra-base shifted up®eld. The adenine H2 proton NOE peaks were pair hydrogen bond does, however, come at the cost of base 1 13 partially assigned with the help of the H- CHMQC pair hydrogen bonds that are lengthened beyond 3.2 A in the spectrum. One of the adenine H2 peaks, the A7-H2 peak, is T6-A38 and A8-T36 base pairs, resulting in only one very broad and clearly shifted up®eld relative to the other H2 hydrogen bond between them instead of the two hydrogen protons by at least 0.2 p.p.m. (Fig. 3C). This broad, upshifted bonds in a normal A-T base pair. The Hoogsteen base pair may peak is characteristic of the adenine H2 proton of TpA base be further stabilized by favorable base stacking interactions doubles. The adenine bases of TpA base doubles often have between the A7 and T6 bases, and the A8 and T9 bases, where high mobility about the glycosidic c bond, resulting in the the six-membered rings of the adenine and the thymine stack broad peak seen in this and other NMR experiments (35). The favorably on one another (Fig. 2B). Thus, several favorable high mobility of the TpA base double at T6-A7 may allow the interactions within the DNA and between the non-speci®cally A7 base to ¯ip more easily than in other base doublets, bound a2D protein and DNA could preferentially stabilize the increasing the probability of forming the Hoogsteen base pair Hoogsteen base pair over the Watson±Crick base pair. upon formation of the a2±DNA complex containing both Energetics calculations and molecular dynamics speci®cally and non-speci®cally bound proteins that is simulations of the protein±DNA system observed in this structure. Molecular dynamics simulations were used to study the Protein±DNA and base±base interactions that may in¯uence of the a2 homeodomains on the stability of the stabilize the Hoogsteen base pair Hoogsteen base pair. The system studied by molecular The fact that the Hoogsteen base pair does not form in the dynamics consisted of either 16 bp DNA alone, the DNA absence of bound protein suggests that the a2 proteins must with the speci®cally bound a2B homeodomain, or DNA with make favorable contacts with the DNA in the crystal that the speci®cally bound a2B and the non-speci®cally bound contribute to the formation of the Hoogsteen base pair, A7- a2D homeodomains. The DNA in the simulations was 1 bp T37. In the crystal structure, Arg4 of the non-speci®cally shorter than the oligonucleotide duplex used in the NMR bound a2D monomer makes van der Waals contacts with the experiments (Fig. 1D). Each system was fully solvated and A7 base and the sugar±phosphate backbone of bases T6 and neutralized with sodium ions. The dynamics simulations were A7 (Fig. 2C). This van der Waals contact may stabilize the run for 1 ns using Langevin dynamics (22,23) implemented in Hoogsteen base pair by preventing the sugar±phosphate CHARMM (24). Fluctuations of the dihedral angles and ± + backbone from moving out to the gauche /gauche conform- distances are used to estimate free energies (not on an absolute ation for the a and g torsion angles. When the A7-T37 base scale; see Materials and Methods), not the relative height of pair is modeled as a Watson±Crick base pair, several the free energy barriers. unfavorable steric clashes are observed between the a2D The presence of the non-speci®c a2D protein minimizes the molecule and the A7 base. In the hypothetical Watson±Crick free energy of the base pair at c =52°, as expected for a base pair model, the A7 C3¢ deoxyribose atom and the a2D Hoogsteen base pair (Fig. 4A). Additionally, the distribution Gly5 main chain N atoms are only 2.2 A from one another. of energies is much sharper in the presence of the a2D protein, Additionally, the C8 atom of an A7 base modeled in the anti with a potential well width of 5° (compared to >20° without conformation would be within 2.8 A of the a2D Asn47 ND2 a2D), indicating that the Hoogsteen base pair is more stable in amino group (Fig. 2C). Many of these unfavorable contacts in the presence of the a2D protein. The large ¯uctuations of the Downloaded from https://academic.oup.com/nar/article-abstract/30/23/5244/1051861 by Ed 'DeepDyve' Gillespie user on 06 February 2018 5250 Nucleic Acids Research, 2002, Vol. 30, No. 23 Figure 4. (A) Energetics of the A7 base calculated from glycosidic angle c in molecular dynamics simulations. The presence of bound a2 proteins stabilizes the c angle at 52°, while free DNA is free to rotate over a broad range. (B) The ¯uctuations of the c angle around the A7 base (red box) decrease signi®cantly when the a2 proteins are bound to the DNA. DNA in the region of the A7 base suggest a possible low complexes (such as triostin A±DNA) with underwound DNA transition energy barrier for ¯ipping the A7 base in the DNA (1), in DNA severely distorted by binding of the transcription in the absence of bound a2 proteins. factor TBP (3) and at the ends of DNA oligonucleotides used Molecular dynamics simulations con®rm many of the in crystallization (5). In all of these cases, the base stacking contacts between the non-speci®cally bound a2D protein energy barrier to forming the Hoogsteen base pair rather than and the DNA in the crystal structure, as well as revealing the more typical Watson±Crick base pair is presumably additional stabilizing contacts that favor the Hoogsteen base lowered by the intercalation of drugs, bending and unwinding pair. The dynamics simulations show that there are many of the DNA, or the presence of free, unpaired bases, all of contacts between the a2D Arg4 residue and the sugar± which have the ¯exibility to allow the ¯ipping of a purine base phosphate backbone at base A7. Additional hydrogen bonds from the normal anti conformation to the syn conformation. not seen in the crystal structure form between a2D Arg4 and To our knowledge, a Hoogsteen base pair has not previously the O3¢ and O4¢ atoms of base A7 during the simulations. been seen in oligonucleotide structures that lack any of these Furthermore, a simulation run with the A7-T37 base pair in the kinds of distortions. Watson±Crick conformation shows that the peptide backbone The crystal structure described here contains both intra-base of the a2D N-terminal arm moves away from the minor pair interactions and protein±DNA interactions that could groove to avoid steric clashes, leading to loss of hydrogen stabilize the Hoogsteen base pair. However, in the absence of bonds and van der Waals contacts between the a2D protein bound a2 proteins, the DNA does not contain a Hoogsteen and DNA. These steric clashes are similar to the clashes base pair, as shown in solution NMR studies of the free DNA. observed when the A7-T37 base pair was modeled in the The NMR studies instead show a broadened A7-H2 peak that Watson±Crick con®guration in the crystal structure. may indicate that the A7 base is free to rotate about the The ¯uctuations of the glycosidic c angle at the A7 base glycosidic c bond. We speculate that the ¯exible A7 base may pair in the molecular dynamics simulations show that the TpA spontaneously ¯ip about the c bond to the unusual syn base doubles, which show characteristic large ¯uctuations of conformation, leading to formation of the Hoogsteen base the c angle, may be contributing to the formation of the pair. Molecular dynamics simulations of the DNA and Hoogsteen base pair. In addition to the broadened peaks protein±DNA systems con®rm the stability of the complex observed at A7-H2 in the NMR experiments, the simulations of a2 proteins bound to DNA containing the Hoogsteen base also show a 16° root mean square ¯uctuation of the A7 c pair. The dynamics simulations also show that the A7 base angle. Furthermore, stacking energy calculations using the indeed has an inherent ¯exibility, with signi®cant ¯uctuations AMBER force ®eld yield low stabilization energies for base about its glycosidic c bond angle. From these experiments, pairs between base pairs T4-A40 and T10-A34 (Fig. 4B). however, we cannot determine whether the Hoogsteen base These low stabilization energies and large ¯uctuations may be pair is a result of the binding of the non-speci®cally bound favorable for ¯ipping the adenine base in solution, and the a2D protein or whether the presence of the Hoogsteen base Hoogsteen base pair may not subsequently revert back to the pair may stabilize the a2D contacts with DNA. Watson±Crick base pair due to the favorable a2D±DNA and It appears that the energetics of the particular con®guration intra-DNA contacts for the Hoogsteen base pair. of proteins and DNA in the present structure gives rise to this unusual base pair, despite the absence of DNA distortions previously observed to be required for Hoogsteen base pair DISCUSSION formation within duplex DNA. A combination of van der Implications for protein±DNA interactions Waals interactions, hydrogen bonds and base stacking inter- actions may allow the stabilization of the A7-T37 Hoogsteen We have observed a Hoogsteen base pair embedded in the structure of undistorted dsDNA. Hoogsteen base pairs have base pair by the a2 proteins in this structure. These base previously been observed in crystal structures of drug±DNA ¯uctuations of the A7 base between syn and anti may happen Downloaded from https://academic.oup.com/nar/article-abstract/30/23/5244/1051861 by Ed 'DeepDyve' Gillespie user on 06 February 2018 Nucleic Acids Research, 2002, Vol. 30, No. 23 5251 2. Gilbert,D.E., van der Marel,G.A., van Boom,J.H. and Feigon,J. (1989) at a very low frequency due to base ¯ipping out of the DNA Unstable Hoogsteen base pairs adjacent to echinomycin binding sites and reinsertion into the DNA. Such base ¯ipping has been within a DNA duplex. Proc. Natl Acad. Sci. USA, 86, 3006±3010. observed and predicted to require an energy of 25 kcal/mol 3. Patikoglou,G.A., Kim,J.L., Sun,L., Yang,S.H., Kodadek,T. and (36), much less than the 100 kcal/mol or greater predicted to Burley,S.K. (1999) TATA element recognition by the TATA box- binding protein has been conserved throughout evolution. Genes Dev., ¯ip the A7 base within DNA (data not shown). One such base 13, 3217±3230. ¯ipping event, a Hoogsteen base pair that appears to require 4. Hoogsteen,K. (1963) The crystal and molecular structure of a hydrogen- stabilization by an a2 protein bound to the DNA, has bonded complex between 1-methylthymine and 9-methyladenine. Acta been observed in the current crystal structure. Because the Crystallogr., 16, 907±916. Hoogsteen base pair is only present in the crystal structure and 5. Rice,P.A., Yang,S., Mizuuchi,K. and Nash,H.A. (1996) Crystal structure of an IHF-DNA complex: a protein-induced DNA U-turn. Cell, 87, not in the DNA alone in NMR experiments, we cannot 1295±1306. determine whether the presence of the Hoogsteen base pair 6. Wolberger,C., Vershon,A.K., Liu,B., Johnson,A.D. and Pabo,C.O. (1991) could in¯uence the binding af®nity of the a2 protein for the Crystal structure of a MAT alpha 2 homeodomain-operator complex DNA. suggests a general model for homeodomain-DNA interactions. Cell, 67, Our observation of a Hoogsteen base pair within otherwise 517±528. 7. Aishima,J. and Wolberger,C. (2002) Crystal structure of the MATalpha2 undistorted B-DNA that is either induced or stabilized by homeodomain-DNA complex with nonspeci®cally bound protein±DNA contacts raises the possibility that Hoogsteen homeodomains. Proteins Struct. Funct. Genet., in press. base pairs could occur within cellular DNA and play a role in 8. Hodel,A., Kim,S.-H. and Brunger,A.T. (1992) Model bias in protein±DNA interactions. The particular con®guration of macromolecular crystal structures. Acta Crystallogr., A48, 851±858. 9. Brunger,A.T. (1992) X-PLOR, Version 3.1. A System for X-ray proteins and DNA reported here is undoubtedly in¯uenced by Crystallography and NMR, 3.84 Edn. Yale University Press, the non-physiological concentrations of a2 protein in the New Haven, CT. crystal drops and does not re¯ect the arrangement of binding 10. Brunger,A.T., Adams,P.D., Clore,G.M., DeLano,W.L., Gros,P., sites found upstream of genes regulated by a2 in vivo. Grosse-Kunstleve,R.W., Jiang,J.S., Kuszewski,J., Nilges,M., Pannu,N.S. Nevertheless, it is possible that the local conditions under et al. (1998) Crystallography & NMR System: a new software suite for macromolecular structure determination. Acta Crystallogr., D54, which the present Hoogsteen base pair forms could be 905±921. duplicated for other proteins at in vivo regulatory sites. The 11. Kleywegt,G.J. and Jones,T.A. (1996) xdlMAPMAN and presence of multiple overlapping binding sites is common in xdlDATAMANÐprograms for reformatting, analysis and manipulation chromosomal DNA and could give rise to a con®guration of of biomacromolecular electron-density maps and re¯ection data sets. proteins analogous to that observed in the crystal. The open Acta Crystallogr., D52, 826±828. 12. Brunger,A.T. (1992) The free R value: a novel statistical quantity for question is whether such an arrangement of proteins either assessing the accuracy of crystal structures. Nature, 355, 472±474. binds preferentially to transiently formed Hoogsteen base 13. Delaglio,F., Grzesiek,S., Vuister,G.W., Zhu,G., Pfeifer,J. and Bax,A. pairs or favors Hoogsteen base pair formation in order to form (1995) NMRPipe: a multidimensional spectral processing system based optimal interactions. Since we were unable to detect measur- on UNIX pipes. J. Biomol. NMR, 6, 277±293. able Hoogsteen base pair formation in free DNA, it was not 14. Johnson,B.A. and Blevins,R.A. (1994) NMRview: a computer program for the visualization and analysis for NMR data. J. Biomol. NMR, 4, possible to assess the energetic contribution of Hoogsteen base 603±614. pair formation by directly comparing the DNA-binding 15. Jeener,J., Maier,B.H., Bachmann,P. and Ernst,R.R. (1979) Investigation af®nity of the a2 homeodomain for sites containing of exchange processes by two-dimensional NMR spectroscopy. J. Chem. Hoogsteen versus Watson±Crick base pairs. However, the Phys., 71, 4546±4553. 16. Macura,S. and Ernst,R.R. (1980) Elucidation of cross relaxation in absence of DNA distortion and the relatively typical array of liquids by two-dimensional NMR-spectroscopy. Mol. Phys., 41, 95±117. protein±DNA contacts suggests that the conditions that favor 17. Piotto,M., Saudek,V. and Sklenar,V. (1992) Gradient-tailored excitation Hoogsteen base pair formation could be replicated in a cellular for single-quantum NMR spectroscopy of aqueous solutions. J. Biomol. context. These observations raise the intriguing possibility that NMR, 2, 661±665. Hoogsteen base pair formation could potentially play a role in 18. Greisinger,C., Otting,G., Wuthrich,K. and Ernst,R.R. (1988) Clean TOCSY for 1H spin system identi®cation in macromolecules. J. Am. the binding of proteins to undistorted B-DNA, although Chem. Soc., 110, 7870. further investigation will be required in order to determine 19. Bax,A. and Davis,D.G. (1985) MLEV-17-based two-dimensional whether this indeed occurs in the cell. homonuclear magnetization transfer spectroscopy. J. Magn. Reson., 65, 355±360. 20. Braunschweiler,L. and Ernst,R.R. (1983) Coherence transfer by isotropic ACKNOWLEDGEMENTS mixingÐapplication to proton correlation spectroscopy. J. Magn. Reson., 53, 521±528. We thank W. Olson for advice and discussions and for 21. Bax,A. and Subramanian,S. (1986) Sensitivity-enhanced two- allowing us to preview the 3DNA program, M. Summers for dimensional heteronuclear shift correlation NMR-spectroscopy. J. Magn. Reson., 67, 565±569. generously allowing the use of equipment for the NMR 22. Schlick,T. (2001) Time-trimming tricks for dynamic simulations: experiments, and C. Garvie, A. VanDemark, A. Ke, splitting force updates to reduce computational work. Structure, 9, N. LaRonde-LeBlanc, P. Minary and R. Campbell for helpful R45±R53. discussions. This work was supported by NSF grant 23. Barth,E. and Schlick,T. (1998) Overcoming stability limitations in MCB9808412 (C.W.). biomolecular dynamics. I. Combining force splitting via extrapolation with Langevin dynamics in LN. J. Chem. Phys., 109, 1617±1632. 24. Brooks,B.R., Bruccoleri,R.E., Olafson,B.D., States,D.J., Swaminathan,S. and Karplus,M. (1983) CHARMM: a program for macromolecular REFERENCES energy, minimization and dynamics calculations. J. Comp. Chem., 4, 1. Wang,A.H., Ughetto,G., Quigley,G.J., Hakoshima,T., van der Marel,G.A., 187±217. van Boom,J.H. and Rich,A. (1984) The molecular structure of a 25. Cornell,W.D., Cieplak,P., Bayley,C.I., Gould,I.R., Merz,K.M., DNA-triostin A complex. Science, 225, 1115±1121. Ferguson,D.M., Spellmeyer,D.C., Fox,T., Caldwell,J.W. and Downloaded from https://academic.oup.com/nar/article-abstract/30/23/5244/1051861 by Ed 'DeepDyve' Gillespie user on 06 February 2018 5252 Nucleic Acids Research, 2002, Vol. 30, No. 23 Kollman,P.A. (1995) A second generation force ®eld for the simulation 32. Berman,H.M., Westbrook,J., Feng,Z., Gilliland,G., Bhat,T.N., of proteins, nucleic acids and organic molecules. J. Am. Chem. Soc., 117, Weissig,H., Shindyalov,I.N. and Bourne,P.E. (2000) The Protein Data 5179±5197. Bank. Nucleic Acids Res., 28, 235±242. 26. Qian,X., Strahs,D. and Schlick,T. (2001) Dynamic simulations of 13 tata 33. McRee,D.E. (1999) XtalView/X®tÐa versatile program for variants re®ne kinetic hypotheses of sequence/activity relationships. manipulating atomic coordinates and electron density. J. Struct. Biol., J. Mol. Biol., 308, 681±703. 125, 156±165. 27. Strahs,D. and Schlick,T. (2000) A-tract bending: insights into 34. Wuthrich,K. (1986) NMR of Proteins and Nucleic Acids. John Wiley & experimental structures by computational models. J. Mol. Biol., 301, Sons, New York, NY. 643±663. 35. McAteer,K. and Kennedy,M.A. (2000) NMR evidence for base dynamics 28. Kottalam,J. and Case,D.A. (1988) Dynamics of ligand escape from the at all TpA steps in DNA [In Process Citation]. J. Biomol. Struct. Dyn., heme pocket of myoglobin. J. Am. Chem. Soc., 110, 7690±7697. 17, 1001±1009. 29. Li,T., Stark,M.R., Johnson,A.D. and Wolberger,C. (1995) Crystal 36. Chen,Y.Z., Mohan,V. and Griffey,R.H. (2000) Spontaneous base ¯ipping structure of the MATa1/MAT alpha 2 homeodomain heterodimer bound in DNA and its possible role in methyltransferase binding. Phys. Rev., to DNA. Science, 270, 262±269. E62, 1133±1137. 30. Tan,S. and Richmond,T.J. (1998) Crystal structure of the yeast 37. Evans,S.V. (1993) SETOR: hardware-lighted three-dimensional solid MATalpha2/MCM1/DNA ternary complex. Nature, 391, 660±666. model representations of macromolecules. J. Mol. Graphics, 11, 31. Qian,Y.Q., Billeter,M., Otting,G., Muller,M., Gehring,W.J. and 134±138, 127±138. Wuthrich,K. (1989) The structure of the Antennapedia homeodomain 38. Delano,W.L. (2002) The PyMOL Molecular Graphics System. Delano determined by NMR spectroscopy in solution: comparison with Scienti®c, San Carlos, CA. prokaryotic repressors. Cell, 59, 573±580. [Erratum (1990) Cell, 61, 39. Humphrey,W., Dalke,A. and Schulten,K. (1996) VMDÐvisual 548.] molecular dynamics. J. Mol. Graphics, 14, 33±38. Downloaded from https://academic.oup.com/nar/article-abstract/30/23/5244/1051861 by Ed 'DeepDyve' Gillespie user on 06 February 2018

Journal

Nucleic Acids ResearchOxford University Press

Published: Dec 1, 2002

There are no references for this article.