| Nucleic Acids Research | Pages |
©1999 Oxford University Press |
NMR-derived solution structure of a 17mer hydroxymethyluracil-containing DNA
Introduction
Materials And Methods
hmU-DNA synthesis
Sample preparation
NMR spectroscopy
Molecular dynamics
Results
Exchangeable protons
Non-exchangeable protons
Structural calculations
Discussion
Acknowledgements
References
NMR-derived solution structure of a 17mer hydroxymethyluracil-containing DNA
Received August 6, 1999; Revised and Accepted September 9, 1999
ABSTRACT Incorporation of 5-(hydroxymethyl)-2[prime]-deoxyuridine into DNA in place of thymine by SPO1, a Bacillus subtilis bacteriophage, allows the viral DNA to bind selectively to transcription factor 1. We have synthesized a TF1-binding site: d(5[prime]-ACCHACHCHHHGHAGGT-3[prime])-d(5[prime]-ACCHACAAAGAGHAGGT-3[prime]) and studied this molecule using NMR spectroscopy. The chemical shifts of exchangeable and non-exchangeable protons were sequentially assigned. Absence of corresponding NOEs in the imino-imino region suggested that the end base pairs did not form Watson-Crick hydrogen bond. Restrained molecular dynamics calculation yielded a family of B-DNA structures whose r.m.s.d. was 0.66 Å (all atoms) for the internal 15 bp. The helical twist was 38.5° per step. The base pairs were situated directly on the helix axis (X-displacement = -0.2 Å). All sugars exhibited C2[prime]-endo puckering with P = 167.3° and [upsilon]max = 38.2°. The OH groups of all hmU bases resided on the 3[prime] side of the base plane and may affect the base orientation relative to the sugar plane as the average [chi] value for all hmU was 4° more positive than that of other nucleosides (258° versus 254°). Positive roll angles ([rho]) and small flanking twists ([omega]) at hmU suggested that the two hmU-A base pair steps open toward the minor grooves.
INTRODUCTION
HmU DNA, of which the oxidation of thymine's methyl group yields 5-(hydroxymethyl)-2[prime]-deoxyuracil (Fig. 1A), occurs naturally in several bacteriophages (1) and serves at preferred-binding sites (over its thymine-DNA counterpart) for phage-encoded proteins such as transcription factor 1 (2) and phage promoter-recognition proteins (3) while disrupting other protein-DNA interactions (4). This pyrimidine base analog is cytotoxic (5), mutagenic (6) and has antiviral activity (7). Its cellular accumulation has been correlated with the aging process (8). A specific hmU-DNA glycosylase has been identified, the function of which is to excise hmU from DNA (9).
Figure 1. hmU-DNA sequence. (A) Structures of hmU and T pyrimidine bases. (B) Sequence of the 17mer duplex. H denotes hmU. From the 5[prime]-end, residues in the top strand are numbered 1->17; those in the bottom strand are labeled 18->34. HA in bold type denotes the beginning and end of the TF1 recognition element.
What structural features of hmU-DNA constitute recognition sites for its specific and cooperative binding to TF1 is of special interest. Many proteins bind DNA by inserting protrusive structural elements such as an [alpha]-helix or a [beta]-hairpin loop into the major groove. Sequence specificity for this binding mode is achieved by multiple contacts between the functional groups of the protein sidechains and the exposed edges of DNA base pairs. Integration host factor, a DBPII, binds DNA by intercalating hydrophobic side chains into the minor groove and in the process severely bend DNA into a U shape (10-12). Since functional groups projecting into the minor groove are limited in providing reliable discriminative elements for recognition, it is thought that bendability at specific sites may contribute to selectivity for specific binding (13). On this note, hmU-DNA was shown to be more deformable than T-DNA and recognition elements for TF1 binding have been suggested by gel retardation studies to include two hmU-A base pair steps separated by an optimal length of 7-9 bp (14,15).
Nuclear magnetic resonance spectroscopy (NMR) has become a standard method to study structure and dynamics of biological molecules (16). Recently established techniques in molecular biology made possible the biosynthesis of isotopically labeled proteins and nucleic acids, a necessary first step for multi-dimensional heteronuclear NMR experiments (17). This approach is not yet feasible for modified DNA such as hmU-containing DNA and, hence, NMR studies of this kind of molecules have remained homonuclear. The first NMR report on hmU-DNA showed that the methylene protons of the hydroxymethyl group of hmU are magnetically not equivalent and that they tend to resonate from 3.5 to 4.5 p.p.m. (18). The orientation of the CH2OH moiety was such that the oxygen atom was found on the 3[prime] side of the base (19).
Previously, a palindromic 16mer hmU-DNA duplex that somewhat resembled the consensus sequence of TF1 was shown to adopt Watson-Crick B conformation in solution (19). In this study, we aim to elucidate a solution structure of a longer piece of hmU-DNA that more closely mimics the central region of a preferred TF1 binding site (Fig. 1B). We report here the complete 1H assignments and a high resolution structure of this 17mer duplex.
MATERIALS AND METHODS
hmU-DNA synthesis
The solid-phase synthesis of 5-hydroxymethyluracil-containing DNA has been described (20). The starting material, 2[prime]-deoxyuridine, was oxidized to yield 5-hydroxymethyl-2[prime]deoxyuridine. Protective moieties were added to all three hydroxyl groups prior to oligo synthesis via standard phosphoroamidite methods. Complete deprotection after synthesis with 80% acetic acid yielded the final products without 5[prime]-end phosphate.
Sample preparation
Each single strand of the lyophilized, HPLC-purified hmU-DNA samples was dissolved directly into 600 µl of NMR buffer (50 mM NaCl, 20 mM KH2PO4, 1 mM NaN3, 90% H2O, 10% D2O, pH 7). The final concentrations of the top (mol. wt = 5289) and bottom (mol. wt = 5301) strands were 1.716 and 1.504 mM, respectively. After NMR studies of the single-stranded (ss)DNA, appropriate aliquots from each sample were mixed together for annealing. The combined volume was reduced to 600 µl to yield a 1.4 mM solution of duplex DNA.
NMR spectroscopy
Data were acquired at 10°C on a Bruker DRX-600 spectrometer with a 5 mm inverse, triple resonance probe equipped with XYZ gradient coils (Bruker). Two-dimensional TOCSY experiments (21) were run with spin-lock mixing times of 30 and 100 ms for a radio frequency field strength of 15 kHz. The water signal was suppressed by using the 3-9-19 pulse sequence with gradients (22) after the MLEV-17 pulse and by high power trim pulses of 2.5 ms at the beginning and end of the spin-lock pulse. Two-dimensional NOESY spectra (23) were obtained with 30, 70, 100, 300, 400 and 600 ms mixing times. The watergate pulse with gradients was used for solvent suppression. The mixing time was varied randomly within 5% to eliminate the appearance of J-cross peaks due to zero-quantum coherence transfer (24). A DQF-COSY experiment was also carried out using a gradient ratio of 1:2 for selection of double-quantum coherence (25). The spectral width was 13 228 Hz with the carrier frequency at the water resonance. All spectra were acquired with 2048 complex points in t2 (4k for DQF-COSY and 600 ms NOESY) and 512 complex points in the t1 dimension (1k for 600 ms NOESY). For each t1 incrementation, 96 scans were averaged with relaxation delays ranging from 1.0 to 3.5 s. In all experiments, quadrature detection in t1 was achieved using TPPI. Time domain data were processed and analyzed using Felix 97.0 (Biosym Technologies Inc.). 1H chemical shifts are reported downfield from internal TSP.
Molecular dynamics
Angle and distance restraints were derived as described by Pasternak et al. (19 and references therein). Isolated spin-pair approximation was used with a reference distance of 2.46 Å between CH5 and CH6 to sort 759 cross peaks obtained from a 70 ms NOESY spectrum into three groups: strong (1.8->2.5 Å), medium (1.8->3.5 Å) and weak (1.8->6.0 Å). A combination of NOE distances and 3J coupling constants (H1[prime]-H2[prime], H1[prime]-H2[prime][prime], H2[prime]-H3[prime], H2[prime][prime]-H3[prime] and H3[prime]-H4[prime]) measured from a DQF-COSY spectrum were used to extract 184 dihedral restraints for the furanose ring and the [gamma], [epsiv] and [chi] angles. Thirty-six H-bond restraints were added to the internal 15 bp to keep the double helix from falling apart. Initial structures were built in A and B conformations with thymine in place of hydroxymethyluracil. The duplexes were modified and capped before being subjected to a simulated annealing protocol using Discover (MSI). Restrained molecular dynamics was done with AMBER force fields as follows: initial energy minimization using the steepest descent and conjugate gradient algorithms; dynamics with temperature beginning at 200 K and increasing to 800 K in steps of 200 K followed by a 14-step annealing to a final temperature of 100 K; a final energy minimization until a maximum derivative of 0.01 kcal/mol Å was achieved. Water was not included in the calculation but a distance-dependent dielectric constant was used instead. Conformational spaces were explored for 1 ps (1 fs steps) at each temperature. Eighty-nine dynamics trajectories were archived for each final structure and a family of 10 structures was calculated for each starting conformation. Helical parameters were calculated with Newhelix 91 (MSI).
RESULTS
Exchangeable protons
Figure 2 shows the imino proton regions from 1-D spectra of the ss and dsDNAs. Under our experimental conditions (50 mM NaCl, pH 7.0, 10°C), 15 out of 17 base pairs were observed for the duplex (Fig. 2C; see also Fig. 3) indicating that terminal fraying disrupts H-bond formation of the first and last base pairs. The NOE connectivities between the imino protons of adjacent base pairs were sequentially assigned (Table 1 and Fig. 3) (16,26-28). Of the 15 observable base pairs, eight belonged to hmUA (downfield of 13.5 p.p.m.) and seven to GC (upfield of 13 p.p.m.). From these chemical shifts, the resonances of the amino and aromatic protons were identified (Table 2 and Fig. 4). In the downfield hmUA group of Figure 4, the stronger peaks were due to cross relaxation between the H3 protons of hmU and the H2 protons of A within the same base pair. For a few hmUA base pairs, interstrand NOEs were also observed for hmUH3 and AH6 protons from the same base pair. The upfield GC group showed correlation between GH1 and CH4 protons within the same base pair. Wherever a pair of these NOEs were observed from a single GC base pair, the downfield peak was assigned to the CH4 amino proton that formed a Watson-Crick H-bond with the GH1 proton. NOEs between the GH1 protons and the H2 protons of neighboring adenine residues (both intrastrand and interstrand) were also detected.
Figure 2. The 1-D 600 MHz imino proton spectra of the top strand (A), bottom strand (B) and duplex (C). Sample concentrations were 1.7, 1.5 and 1.4 mM for the top, bottom and double strands, respectively. The spectra were acquired at 283 K; the ssDNAs in 50 mM NaCl, 20 mM phosphate, pH 7.0 and the dsDNA in 100 mM NaCl, 40 mM phosphate, pH 7.0.
Figure 3. Sequential NOE connectivity in the imino-imino region of the 17mer duplex. Base pairs 1 and 17 are not detected due to end-fraying. Also not detected is the NOE between base pairs 10 and 11 (arrow). The NOESY spectrum was acquired at 283 K with 600 ms mixing time.
Figure 4. NOEs between imino to aromatic (AH2) and amino protons. Numbers specify the sequentially assigned base pairs. The spectrum was acquired at 283 K, 600 ms mixing time.
Table 1. Chemical shifts of the imino protons (in p.p.m.)a
| Base pairs | [delta] of imino H |
| A1-T34 | ND |
| C2-G33 | 12.86 |
| C3-G32 | 12.86 |
| H4-A31 | 13.89 |
| A5-H30 | 13.66 |
| C6-G29 | 12.75 |
| H7-A28 | 14.13 |
| C8-G27 | 12.54 |
| H9-A26 | 14.24 |
| H10-A25 | 14.12 |
| H11-A24 | 13.78 |
| G12-C23 | 12.47 |
| H13-A22 | 13.67 |
| A14-H21 | 13.91 |
| G15-C20 | 12.75 |
| G16-C19 | 12.53 |
| T17-A18 | ND |
Table 2. Proton chemical shifts of the 17mer duplex hmU-DNAa
| H1[prime] | H2[prime]/H2[prime][prime] | H3[prime] | H4[prime] | H5[prime]/H5[prime][prime] | A/G-H8 C/H/T-H6 |
A-H2; T-CH3 C-H5;H-CH2 |
G-H2; A-H6 C-H4 |
|
| A1 | 6.19 | 2.76/2.86 | 5.03 | 4.64 | 4.46/4.32 | 8.37 | 7.61 | ND |
| C2 | 5.77 | 2.21/2.43 | 4.64 | 4.46 | 4.32/3.87 | 7.25 | 5.19 | 8.06, 6.24 |
| C3 | 5.81 | 2.36/2.68 | 4.85 | 4.38 | 3.69/3.55 | 7.35 | ND | 8.38, 6.73 |
| H4 | 5.61 | 2.35/2.69 | 4.84 | 4.39 | 3.94/3.94 | 7.26 | 3.69/3.54 | NA |
| A5 | 6.00 | 2.51/2.87 | 5.03 | 4.38 | 4.14/4.03 | 8.21 | 7.33 | 7.04, 6.26 |
| C6 | 5.88 | 2.06/2.76 | 4.83 | 4.25 | 4.18/4.15 | 7.51 | 5.55 | 7.89, 6.19 |
| H7 | 5.81 | 2.17/2.55 | 4.85 | 4.39 | 4.19/4.15 | 7.47 | 3.89/3.80 | NA |
| C8 | 6.11 | 2.12/2.16 | 4.89 | 4.21 | 4.11/4.06 | 7.57 | 5.61 | 8.23, 6.81 |
| H9 | 5.95 | ND | 4.96 | ND | ND | 7.28 | ND | NA |
| H10 | 6.04 | 2.18/2.55 | 4.96 | 4.39 | 4.22/4.14 | 7.47 | 3.69/3.55 | NA |
| H11 | 5.94 | 2.14/2.47 | 4.95 | 4.38 | 4.19/4.19 | 7.58 | 4.04/4.04 | NA |
| G12 | 5.55 | 2.22/2.45 | 4.96 | 4.47 | 4.19/4.14 | 8.37 | NA | 7.60, ND |
| H13 | 5.36 | 2.76/2.76 | 4.96 | 4.39 | 4.22/4.14 | 7.47 | 3.69/3.55 | NA |
| A14 | 5.88 | 2.22/2.45 | 4.86 | 4.25 | 4.18/4.15 | 8.05 | 7.37 | 6.73, ND |
| G15 | 6.25 | 2.75/2.87 | 4.96 | 4.28 | 4.19/4.19 | 8.28 | NA | 6.24, 6.19 |
| G16 | 5.55 | 2.54/2.61 | 4.97 | 4.38 | 4.08/4.08 | 8.06 | NA | 7.04, ND |
| T17 | 6.24 | 2.50/2.70 | 4.86 | 4.55 | 4.21/4.07 | 7.37 | 1.43 | ND |
| A18 | 6.16 | 2.68/2.84 | 5.02 | 4.43 | 4.08/4.08 | 8.35 | ND | ND |
| C19 | 5.63 | 2.15/2.48 | 4.74 | 4.42 | 4.08/4.08 | 7.21 | 5.25 | 7.95, 6.24 |
| C20 | 5.54 | 2.49/2.87 | 4.95 | 4.17 | ND | 7.59 | 5.61 | 8.37, 6.94 |
| H21 | 5.93 | 1.99/2.77 | 4.96 | 4.38 | 4.14/4.14 | 7.35 | 3.90/3.75 | NA |
| A22 | 5.98 | 2.24/2.69 | 4.84 | 4.38 | 4.14/4.03 | 8.22 | 7.33 | 6.94, 6.75 |
| C23 | 5.27 | 2.15/2.49 | 4.96 | ND | ND | 7.32 | ND | 8.13, 6.57 |
| A24 | 5.69 | 2.54/2.77 | 4.99 | 4.31 | 4.16/4.16 | 8.12 | 7.03 | ND |
| A25 | 5.70 | 2.16/2.62 | 4.74 | 4.29 | 3.95/3.95 | 8.04 | 7.29 | 7.06, ND |
| A26 | 5.82 | 2.53/2.77 | 4.99 | 4.38 | 4.17/4.17 | 7.96 | 7.28 | 7.07, 5.96 |
| G27 | 5.93 | 2.21/2.55 | 4.84 | 3.90 | 3.75/3.75 | 7.93 | NA | ND |
| A28 | 5.96 | 2.36/2.83 | 4.55 | 4.38 | 3.79/3.79 | 8.06 | 7.48 | 7.35, 6.78 |
| G29 | 5.47 | 2.39/2.65 | 4.94 | 4.33 | 3.69/3.54 | 7.95 | NA | 6.76, ND |
| H30 | 5.36 | 2.23/2.23 | 4.96 | 4.39 | 4.22/4.14 | 7.46 | 3.69/3.55 | NA |
| A31 | 6.26 | 2.23/2.46 | 4.85 | 4.27 | 3.79/3.79 | 8.08 | 7.33 | 6.94, 6.26 |
| G32 | 5.36 | 2.43/2.49 | 4.84 | 4.32 | 3.87/3.87 | 8.37 | NA | 6.94, 6.25 |
| G33 | 5.55 | 2.35/2.69 | 4.86 | 4.55 | 3.79/3.79 | 8.28 | NA | 5.95, ND |
| T34 | 6.24 | 2.24/2.52 | 4.86 | 4.37 | 4.08/4.08 | 7.63 | 1.43 | NA |
Non-exchangeable protons
Figure 5 shows the (n, n+1) NOEs between the base (H8/H6) and sugar (H1[prime]) protons. The cross peaks between A/G H8 to sugar H1[prime] are better resolved than those between H6 of C/T/hmU and sugar H1[prime]. Aromatic protons from adjacent nucleotides of the same strand showed NOEs which demonstrated nicely the sequential connectivity between adjacent nucleotides and, hence, confirmed the assignment using H8/H6-H1[prime] NOEs (data not shown).
Figure 5. Sequential NOE connectivity in the base-sugar region of the 17mer duplex. For simplicity, only the top strand is traced. X denotes peaks that can only be observed at lower threshold. Horizontal lines depict intrastrand, internucleotide NOEs between the base protons (H8 or H6) of residue n to the sugar H1[prime] protons of residue n+1. Vertical lines connect intrastrand, intranucleotide NOEs between the base (H8/H6) and the sugar H1[prime] protons. The spectrum was acquired at 283 K, 600 ms mixing time.
Other sugar protons were straightforwardly assigned and are listed in Table 2. The methylene protons from hmU residues showed up in the expected range of 3.4-4.1 p.p.m. (18). When possible, the HM1 proton signal was assigned downfield to that of HM2 according to the method reported by Pasternak et al. (19). The chemical shifts of H2[prime] (higher field) and H2[prime][prime] were assigned with the assumption that the 17mer duplex adopted a B conformation. A number of NMR criteria have been established to access DNA conformation in aqueous solution. Among these are the observation that in B-DNA, (i) adjacent adenine residues display NOEs between their H2 protons; (ii) the C2[prime]-endo conformation of the ribose ring dictates that the coupling constant 3J2´,3´must be greater than 3J2´´,3´and therefore, only one cross peak between H2[prime] and H3[prime] is observed in a through-bond correlation spectrum (16); and (iii) the ratio of the NOEs intensities ([tau] < 50 ms) between H8/H6:H3[prime] should be several times that of H8/H6:H2[prime] (1:1 for A-DNA) (28). Evidence for these, together with the fact that a closely related 16mer (19) was shown to be a B-DNA led us to assume that our 17mer also adopted a B conformation. Table 2 lists the chemical shifts of the non-exchangeable protons as well as those from the amino groups. The H5[prime]/H5[prime][prime] resonances were not stereospecifically assigned.
Structural calculations
Various conformers ranging from A- to B-DNA were used as starting structures for restrained molecular dynamics calculation. A family of 20 structures converged to an r.m.s.d. of 0.66 Å for all atoms of the internal 15 bp (no violation > 0.3 Å or 5°). The average structure was that of a right-handed B-DNA (Fig. 6). The helical parameters were calculated using the NEWHELIX91 program (implemented through Insight 97, MSI) and yielded an average twist ([Omega]) of 38.5° corresponding to 9.3 residues per turn and a rise (Dz) of 3.1 Å. The base pairs sat directly on the helix axis as indicated by an x-displacement of only -0.2 Å (29). If the C1[prime]-C1[prime] distance between neighboring bases was taken to be 5.6 Å for a fully extended helix (30) then an observed Dxyz of 5.3 Å would suggest that the backbone chain of hmU-DNA was ~92% extended. It is also informative to further classify a double helix into either a BI or BII conformation since the C3[prime]-O3[prime]-P angle in the former structure does not impose additional restraints on base stacking but the latter promotes base unstacking as it turns perpendicular into the helix axis normal plane (31). The trans and gauche- backbone torsion angles of [epsiv] and [zeta] (Table 3) measured from the resulting structures put the calculated hmU-DNA into the BI subcategory. All sugars exhibited C2[prime]-endo puckering with an average pseudorotation angle P of 167.3° and a maximum out-of-plane pucker [upsilon]max of 38.2° (the average torsion angle [upsilon]2 was -37°) (32). The glycosyl angles [chi] were all within the anti region (244°->265°). The OH group of an hmU base might affect its orientation relative to the sugar plane as the average [chi] for all hmU was 4° more positive than that of other nucleosides (258° versus 254°).
Figure 6. Stereo view of the average structure from a family of 10 structures calculated from an initial A-DNA duplex. The resulting B-DNA structures converge to an r.m.s.d. of 0.66 Å (all atoms) for the internal 15 bp.
Table 3. Backbone torsional angles (degrees) in the 17mer hmU-DNAa
| [alpha] | [beta] | [gamma] | [delta] | [epsiv] | [zeta] | [chi] | [rho] | [omega] | |
| C2 | 295 | 174 | 47 | 152 | 179 | 252 | 248 | -1.3 | 43.9 |
| C3 | 288 | 186 | 58 | 148 | 184 | 244 | 253 | -3.0 | 36.9 |
| H4 | 266 | 179 | 76 | 152 | 186 | 238 | 258 | 15.5 | 42.9 |
| A5 | 299 | 181 | 52 | 151 | 181 | 251 | 248 | -10.6 | 38.4 |
| C6 | 279 | 182 | 65 | 148 | 180 | 248 | 251 | -6.9 | 40.1 |
| H7 | 301 | 187 | 45 | 148 | 182 | 249 | 253 | 3.8 | 39.2 |
| C8 | 291 | 180 | 62 | 148 | 181 | 246 | 253 | -6.4 | 39.8 |
| H9 | 290 | 182 | 58 | 146 | 184 | 245 | 258 | 0.0 | 40.2 |
| H10 | 291 | 177 | 64 | 146 | 183 | 249 | 259 | 1.5 | 38.8 |
| H11 | 289 | 180 | 63 | 150 | 188 | 237 | 262 | 14.6 | 43.4 |
| G12 | 299 | 182 | 44 | 151 | 179 | 255 | 246 | -10.4 | 37.3 |
| H13 | 287 | 184 | 60 | 150 | 189 | 235 | 261 | 7.4 | 40.1 |
| A14 | 298 | 178 | 50 | 153 | 182 | 250 | 251 | -3.9 | 38.6 |
| G15 | 294 | 189 | 45 | 150 | 182 | 246 | 251 | -2.8 | 41.4 |
| G16 | 289 | 185 | 56 | 147 | 241 | 288 | 257 | ND | ND |
| C19 | 268 | 173 | 69 | 153 | 193 | 233 | 256 | ||
| C20 | 266 | 176 | 71 | 150 | 183 | 242 | 248 | ||
| H21 | 323 | 185 | 20 | 151 | 181 | 247 | 253 | ||
| A22 | 291 | 183 | 61 | 149 | 185 | 242 | 256 | ||
| C23 | 287 | 178 | 63 | 151 | 184 | 242 | 258 | ||
| A24 | 286 | 183 | 58 | 149 | 183 | 248 | 258 | ||
| A25 | 317 | 191 | 21 | 157 | 183 | 242 | 265 | ||
| A26 | 295 | 181 | 47 | 154 | 185 | 241 | 254 | ||
| G27 | 312 | 184 | 27 | 157 | 184 | 244 | 257 | ||
| A28 | 295 | 182 | 48 | 154 | 180 | 252 | 254 | ||
| G29 | 285 | 188 | 56 | 147 | 181 | 247 | 257 | ||
| H30 | 280 | 179 | 68 | 149 | 180 | 241 | 259 | ||
| A31 | 261 | 179 | 82 | 149 | 190 | 258 | 255 | ||
| G32 | 234 | 176 | 98 | 154 | 196 | 223 | 260 | ||
| G33 | 289 | 173 | 54 | 148 | 183 | 272 | 244 |
DISCUSSION
Until now, only two hmU-DNA structures have been reported (18,19). One of these, a 16mer palindrome that mimics part of the TF1 binding site was shown to be a right-handed B-DNA. This hmU-DNA molecule is sharply bent when bound to TF1 as suggested by molecular modeling (33). Data from gel retardation studies indicate that it is the sequence-dependent local flexibility of hmU-DNA that provides specificity for TF1 (14). We have synthesized a non-palindromic 17mer hmU-DNA duplex that more closely resembled the cognate site for TF1 than the published palindromic 16mer and calculated its structure in solution to atomic resolution. Examination of this structure reveals insights into the specificity and affinity of hmU-DNA for TF1.
For the single strands (Fig. 2A, top strand, and B, bottom strand), the fact that there are peaks at all in the imino proton region implies that these ssDNAs adopt some forms of secondary structures which protect several of their imino protons from exchanging too quickly with the solvent. Inspection of the nucleotide sequences brings up two possibilities: each single strand may form a duplex with itself via self-complementary base pairing and leave a mismatched bubble of 5 bp in the center, or it can fold back upon itself to obtain a hairpin structure with a double-stranded stem of 6 bp and a loop of five bases. Since it was pointed out that, under low salt concentrations (20-50 mM NaCl), a hairpin structure might be the predominant species whereas a duplex conformation was favorable at elevated ionic strength (200 mM NaCl) (34), we think it is more likely that the ssDNAs adopt a hairpin structure.
DNA bendability at specific sites has been considered to be an important recognition element for DNA-binding proteins (35). For type II DNA-binding proteins, this mode of interaction was shown to involve a DNA whose structure includes three segments, the central of which was 9 bp long, separated by two kinks and interaction of hydrophobic side chains in the minor grooves at the kinks would introduce a 160° bend in DNA (11,12). The cognate site for TF1 was also identified to include two hmU-A base pair steps separated by 9 bp (15). It was argued that the local flexibility provided by these hmU-A base pair steps served as a specific flag for TF1 recognition and for its preferred binding over thymine-containing DNA (14). It should be noted that previous biophysical studies suggested similar flexibilities between hmU-DNA and T-DNA (18,19,36). We focus our analysis on those local helix parameters of hmU-DNA which may have contributed to its interaction with TF1. Namely, twist angles ([omega]) along the z-axis and roll angles ([rho]) along the y-axis of 2 bp steps since they strongly influence the orientation of potential H-bond donors and acceptors along the floors of the grooves and thus play a major role in the process of recognition (29). All roll angles were found to be negative except for those of hmU, especially hmU4 and hmU13 (Table 3). Positive [rho] occurs when the angle between base pairs opens toward the minor groove (37,38). In addition, we found that the helical twists for hmU were consistently of higher values (average [omega] = 40.7°) than others (average [omega] = 37.2°) and that for the two `recognized' hmU-A steps, the measured twist angles were 36.9°-42.9°-38.4° for the sequence 5[prime]-CHAC-3[prime] and 37.3°-40.1°-38.6° for 5[prime]-GHAG-3[prime] (Table 3). It was pointed out that small flanking twists allow the central step to open up (29). Finally, it is interesting to note that the nearest-neighbor model for helix propagation using thermodynamics parameters placed TA at the bottom of the duplex stability list (GC = CG > GG > CA = GT = GA = CT > AA > AT > TA) (39,40). If this trend preserves its order for hmU-DNA, then the weakest points of our duplex would coincide with those noticed by Grove et al. (14,15) for their TF1 cognate site. Collectively, these data suggest the propensity of 5[prime]-hmU-A-3[prime] base pair steps to accommodate binding elements of proteins in the minor groove. The relative orientation of the hmU's CH2OH group to the potential H-bond donor NH2 of its 3[prime]-adenosine neighbor may contribute to binding specificity via formation of weak hydrogen bonds (hmU4:O5-A5:H62 distance = 3.4 Å and O5°H-N angle = 121°) which could be strengthened after TF1 binding. Inspection of the calculated structures (Fig. 6) showed that all OH groups of the hmU's hydroxymethyl moiety lay on the 3[prime] side of the base plane. A similar observation was made by Pasternak et al. for their 16mer hmU-DNA duplex (19).
In summary, close examination of the 17mer hmU-DNA duplex revealed: (i) a 92% extended B-DNA whose mean rotation was 38.5° per step (9.3 bp/turn); (ii) all sugar moieties adopted a C2[prime]-endo puckering; (iii) all glycosyl angles were anti; (iv) all OH groups of hmU's hydroxymethyl lay on the 3[prime] side of the base plane; and (v) all roll angles were negative except for those of hmU (most positive when it is followed by a purine base on the 3[prime] side) which meant that these base pairs opened up in the minor grooves. Base unstacking at the hmU-A steps may play a role in the recognition process for TF1 as suggested by Grove et al. (14,15). The high resolution structure of the complex between TF1-G15/I32 and the 17mer hmU-DNA is being calculated and this work should shed more light into their mechanisms of binding.
ACKNOWLEDGEMENTS
We thank Dr Anne Grove and Dr Laura Pasternak for their insightful discussions and Dr John Wright for his help with the NMR experiments. This work was supported in part by grants from the NIH GM40635-09 for D.R.K. and from the Italian M.U.R.S.T. and C.N.R. for L.M.
REFERENCES
*To whom correspondence should be addressed. Tel: +1 619 534 2760; Fax: +1 619 534 0202; Email: drk{at}chem.ucsd.edu Present address: Antonietta Pepe, Department of Chemistry, University of Basilicata, Potenza 85100, Italy
This page is run by Oxford University Press, Great Clarendon Street, Oxford OX2 6DP, as part of the OUP Journals
Comments and feedback: jnl.info{at}oup.co.uk
Last modification:
Copyright© Oxford University Press, 1999.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  Z. Bukowy, J. A. Harrigan, D. A. Ramsden, B. Tudek, V. A. Bohr, and T. Stevnsner WRN Exonuclease activity is blocked by specific oxidatively induced base lesions positioned in either DNA strand Nucleic Acids Res., September 1, 2008; 36(15): 4975 - 4987. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||