| Nucleic Acids Research | Pages |
©1999 Oxford University Press |
Imidazole-imidazole pair as a minor groove recognition motif for T:G mismatched base pairs
Introduction
Materials And Methods
Sample preparation
NMR data collection
Structural refinement
Results
Binding of the drug to DNA containing T:G sequences
Structure of the AR-1-144/GAACTGGTTC complex
Exchangeable proton NMR data consistent with the refined structure
Heterodimer for the recognition of a single T:G mismatch
Discussion
Acknowledgements
References
Imidazole-imidazole pair as a minor groove recognition motif for T:G mismatched base pairs
Received July 12, 1999; Accepted August 28, 1999
ABSTRACT The T:G mismatched base pair is associated with many genetic mutations. Understanding its biological consequences may be aided by studying the structural perturbation of DNA caused by a T:G base pair and by specific probing of the mismatch using small molecular ligands. We have shown previously that AR-1-144, a tri-imidazole (Im-Im-Im) minor groove binder, recognizes the sequence CCGG. NMR structural analysis of the symmetric 2:1 complex of AR-1-144 and GAACCGGTTC revealed that each AR-1-144 binds to four base pairs with the guanine N2 amino group forming a bifurcated hydrogen bond to a side-by-side Im/Im pair. We predicted that the free G-N2 amino group in a T:G wobble base pair can form two individual hydrogen bonds to a side-by-side Im/Im pair. Thus an Im/Im pair may be a good recognition motif for a T:G base pair in DNA. Cooperative and tight binding of an AR-1-144 homodimer to GAACTGGTTC permits a detailed structural analysis by 2D NOE NMR refinement and the refined structure confirms our prediction. Surprisingly, AR-1-144 does not bind to GAATCGGTTC. We further show that both the Im-Im-Im/Im-Py-Im heterodimer and the Im-Im-Im/Im-Im-Im homodimer bind strongly to the CACGGGTC + GACTCGTG duplex. These results together suggest that an Im/Im pair can specifically recognize a single T:G mismatch. Our results may be useful in future design of molecules (e.g. linked dimers) that can recognize a single T:G mismatch with specificity.
INTRODUCTION
Mutations leading to mismatched base pairs play important roles in the genesis of tumor cells. The T:G mismatch is especially important and has been shown to be responsible for most of the common mutations in human ras oncogenes (1). A T:G mismatch is often introduced by spontaneous deamination of 5-methylcytosine or by errors of replication. In vertebrate cells, methylation of cytosine provides an important method for distinguishing active genes from inactive genes. In fact, even though only ~3% of the cytosine nucleotides in human DNA are methylated, mutations in 5-methylcytosine account for about one-third of the single base mutations that have been observed in inherited human diseases. Although a specific T:G mismatch repair system exists in the cell, some T:G mismatches in different sequence contexts could escape being repaired.
A molecular ligand capable of recognizing a specific DNA sequence incorporating a T:G mismatch and actually binding to the site tightly may promote the T:G repair system to recognize and repair the T:G mismatch more efficiently. Therefore, malignant transformation of a cell could be switched off by the ligand.
T:G mismatching is also found in DNA of single-stranded DNA viruses in some important regions such as the replication origin, which often contains stable secondary structures including hairpins and bulges. Therefore, a ligand that can bind precisely to a sequence incorporating a T:G mismatch, thereby inhibiting expression of the viral gene, may be lethal to the virus.
Indeed, designs of molecular agents which can recognize mismatched base pairs have begun to emerge (2,3). We have taken a different approach by using sequence-specific DNA minor groove binders (MGBs) for such studies. A recognition rule for canonical DNA sequences (i.e. Watson-Crick C:G and A:T base pairs) has been proposed recently (4,5) in which combinations of pyrrole (Py), hydroxypyrrole (Hp) and imidazole (Im) units of polyamides can confer excellent specific recognition properties for all four different base pairs, i.e. A:T, T:A, G:C and C:G.
The T:G mismatch has been shown to adopt the wobble conformation normally (6-9), although unusual conformations have also been observed (10). Previous studies (5,11) suggested that an Im/Im pair has an ambiguous G:C/C:G sequence specificity and weaker binding properties toward normal base pairs. However, we propose that an Im/Im pair could be a good motif for T:G/G:T recognition since the N2 amino group of the G in a T:G mismatch, unlike in a C:G base pair, is not involved in base pairing with T. Therefore, the free amino group could form two hydrogen bonds, each with one imidazole nitrogen atom of the polyamide. To test this hypothesis, we used high resolution NMR to probe the binding of AR-1-144, (an Im-Im-Im polyamide) and Im-Py-Im (Fig. 1A) to various DNA sequences containing T:G mismatches. Our results show that an Im/Im pair is a good motif for the recognition of a T:G/G:T base pair.
Figure 1. (A) Chemical structures of AR-1-144 and Im-Py-Im. (B) Schematic representation of the homodimeric binding of AR-1-144 to GAACTGGTTC with the numbering of the nucleotide sequences indicated.
MATERIALS AND METHODS
Sample preparation
AR-1-144 was prepared as described previously (12). Im-Py-Im was readily synthesized employing a similar methodology to that published (13,14) and characterized by 1H NMR, electrospray mass spectra and analytical HPLC. Analytical data for Im-Py-Im: 1H NMR (500 Hz, DMSO-d6): d = 10.24 (s, 1H), 10.16 (s, 1H), 9.87 (s, 1H), 9.74 (s, br, 1H), 8.22 (t, J = 6.0 Hz, 1H), 7.49 (s, 1H), 7.42 (s, 1H), 7.36 (d, J = 1.5 Hz, 1H), 7.13 (d, J = 1.5 Hz, 1H), 3.94 (s, 3H), 3.93 (s, 3H), 3.85 (s, 3H), 3.05 (m, 2H), 2.75 (s, 3H), 2.74 (s, 3H), 2.49 (m, 2H, overlapped with the peak of DMSO), 2.02 (s, 3H), 1.88 (m, 2H). ESMS m/z calculated for C23H31N10O4 (M - 1) 511.3, found 511.2. The polyamide stock solutions (30 mM) for NMR studies were prepared by dissolving appropriate amounts of the polyamide in 500 µl of a 90% H2O/10% D2O solution, plus 10 µl of dilute HCl solution for AR-1-144 to increase the solubility.
All DNA oligonucleotides were synthesized on a DNA synthesizer in the Genetic Facility at the University of Illinois, purified by Sepharose G-50 column chromatography and lyophilized. NMR samples were prepared by dissolving the oligonucleotides in 530 µl of phosphate buffer (20 mM sodium phosphate, pH 7.0, 90% H2O/10% D2O). For the drug/DNA titration monitored by 1D 1H NMR, small scale DNA samples (0.5 mM duplex) were prepared. For structural analysis, a large scale GAACTGGTTC sample (1.5 mM duplex) was used to collect 2D NMR data. DNA samples were titrated by stepwise addition of the polyamides, monitored by 1D 1H NMR. A large scale GAATCGGTTC sample (1.9 mM duplex) was also prepared to study its duplex conformation. For the non-exchangeable proton NMR experiments, samples were prepared as described previously (15).
NMR data collection
AllNMR measurements were made on a Varian Inova 750 MHz spectrometer for 2D spectra and on a Varian VXR500 500 MHz spectrometer for 1D spectra at 2°C. 1D 1H NMR spectra in H2O were recorded using the 1-1 pulse sequence (16). The chemical shifts in parts per million (p.p.m.) were referenced to the HDO peak, which is calibrated to 2,2-dimethyl-2-silapentene-5-sulfonate at different temperatures. The phase-sensitive NOESY spectrum of the AR-1-144/GAA-CTGGTTC complex in D2O was recorded with 2048 points in the t2 dimension and 512 complex FIDs in the t1 dimension with 32 scans/FID (17). The recycle delay was 7.2 s with an average T1 of 3.6 s, and the mixing time was 150 ms. TOCSY spectra were also recorded with 2048 points in the t2 dimension and 512 complex FIDs in the t1 dimension with eight scans/FID. The WET (water suppression enhanced through T1 effects) solvent suppression method was used to record the 2D NOESY spectrum of the complex in H2O (18). All NMR data sets were processed with the program FELIX v.1.1 (formerly from Hare Research, Woodinville, WA) or FELIX 98 (Molecular Simulation, San Diego, CA) on Silicon Graphics workstations. The 2D NOESY and TOCSY spectra in D2O and 2D NOESY spectra in H2O were used to assign (by the standard sequential assignment procedure) the resonances of all non-exchangeable and exchangeable protons.
Structural refinement
The initial model of the symmetrical side-by-side 2:1 complex was constructed by docking two AR-1-144 molecules in an anti-parallel orientation to the minor groove of GAACTGGTTC using MIDAS (University of California), guided by the NOE data and our previous NMR structure of the AR-1-144/GAACCGGTTC complex (11). Then, 10 starting models were generated by following the simulated annealing protocol as described in X-PLOR (19). The root mean square deviations of the 10 starting models from the `average structure' were 1.051 Å for the complex, 1.073 Å for DNA and 0.891 Å for the drug (Table 1).
Table 1. Structural statistics of the 2:1 AR-1-144/d(GAACTGGTTC)2 complexa
| Number of non-exchangeable NOE restraints (all/overlap <30%) | ||||
| Total | 988/281 | |||
| Between DNA and drug | 227/41 | |||
| Inter-nucleotide | 423/83 | |||
| Intra-nucleotide | 299/147 | |||
| Inter-drug | 14/2 | |||
| Intra-drug | 25/8 | |||
| NMR R-factor ([Sigma]|No - Nc|/[Sigma]Nc) | 23.3% | |||
| Structure statistics (r.m.s.d.) | ||||
| NOE distance deviationsb | 0.296 Å | |||
| Bond distance deviations from ideal values | 0.011 Å | |||
| Bond angle deviations from ideal values | 3.30° | |||
| Energy (from X-PLOR) (kcal/mol) | ||||
| Total | 253 | |||
| Chemical | -191 | |||
| NOE | 444 | |||
| Atomic r.m.s.d.c | <SA>s vs [SA]s | <SA> vs [SA] | <SA> vs [SA]e | <SA> vs [SA]r |
| All | 1.051 Å | 0.421 Å | 0.495 Å | 0.505 Å |
| DNA | 1.073 Å | 0.445 Å | 0.520 Å | 0.531 Å |
| Drug | 0.891 Å | 0.230 Å | 0.312 Å | 0.290 Å |
bPairwise distance deviations between the refined structure and NOE-derived values.
cExcludes hydrogen atoms.
The structural refinement was carried out by the procedure SPEDREF (15). The procedure calculates the simulated 2D NOE spectrum using the full matrix relaxation method (from MORASS; 20) on the basis of a molecular model. The volume of all NOE peaks is measured and compared between the experimental and the calculated spectra. The difference between the experimental and the simulated volumes is minimized by a conjugated gradient procedure in X-PLOR to adjust the interproton distances and generate a new starting model. The procedure is iterative and an NMR R-factor is used to monitor the progress of the refinement. The NOE restraints were applied during refinement using the asymmetric biharmonic energy functions according to the inter-proton distances (<2.2 Å; 2.2-3.5 Å; 3.5-4.5 Å; >4.5Å) as described in X-PLOR. Overlapping peaks were deconvoluted as described before (15) to maximize the information retrieved from the experimental data. Use of the full matrix relaxation theory (which takes into consideration spin diffusion) on the basis of a molecular model reduces the need to collect multiple 2D NOESY data at different mixing times. A mixing time of 150 ms was used to obtain a balance between optimal number of NOE crosspeaks and moderate spin diffusion.
The force field parameters for AR-1-144 were generated through a standard database in X-PLOR whose all-atom force field for DNA was used with explicit hydrogen bond potentials. The 10 starting models were each subjected to 40 refinement cycles of NOE-restrained conjugate gradient minimization using X-PLOR. Two-fold symmetry constraints of the complex were also applied in the refinement. The experimental NMR correlation time ([tau]c) was determined to have an optimal value of 5 ns by testing empirically a range of [tau]c values in the NMR refinement.
The `average structure' of the 10 refined structures was energy minimized and further subjected to another 40 cycles of NOE-restrained conjugate gradient refinement and the final refined `average structure' used for further discussion. The NMR R-factor is defined as R-factor = [sum]|No - Nc|/[sum]No, where No and Nc are the experimental and calculated NOE integrals, respectively. The refinement and structural statistics are listed in Table 1. A total of 988 NOE restraints were retrieved from the experimental data by deconvolution. The reliability of the NOE information depends on the degree of overlap; the less overlap the peaks show, the more reliable the NOE information. The number of `reliable' NOEs, which overlap <30%, was 281 (equivalent to ~27 NOEs/nucleotide). This number is comparable to that used by other workers in the field. The chemical shift assignments, final coordinates and NOE constraints have been deposited in the Research Collaboratory for Structural Bioinformatics Protein Data Bank with RCSB ID code RCSB009629 and PDB ID code 1CYZ.
RESULTS
Binding of the drug to DNA containing T:G sequences
We had previously studied the structure of a 2:1 complex of AR-1-144 bound to GAACCGGTTC (11) and the same sequence was used as a control system. Three DNA oligonucleotides, related to the canonical sequence but incorporating a T:G mismatch at different positions, were designed: GAACTGGTTC, GAACGTGTTC and GAATCGGTTC. They were titrated with AR-1-144. The four DNAs are denoted CCGG, CTGG, CGTG and TCGG, respectively.
Comparing the NMR titration profile of GAACTGGTTC (Fig. 2A) with that of GAACCGGTTC (Fig. 2B), it can be seen that all free DNA molecules are absent at a 2:1 drug:DNA ratio with the CTGG sequence. In contrast, with the CCGG sequence, a drug:DNA ratio greater than 3:1 is needed to obtain a fully formed 2:1 drug-DNA complex. Only one set of new resonances appears, suggesting that a symmetrical 2:1 complex is formed with GAACTGGTTC (Fig. 1B). This observation indicates that AR-1-144 indeed binds to GAACTGGTTC with stronger affinity than it binds to GAACCGGTTC; in other words, Im/Im is a motif recognizing T:G/G:T mismatches better than C:G/G:C base pairs.
Figure 2. Titration of AR-1-144 with different DNA decamers with and without a T:G mismatch. The top spectra of each frame are from DNA alone. Different ratios of drug to DNA duplex are indicated. (A) GAACTGGTTC; (B) GAACCGGTTC; (C) GAACGTGTTC; (D) GAACGCGTTC.
When the core sequence was changed from CTGG to CGTG, a different behavior was observed upon titration of AR-1-144 (Fig. 2C). More than one set of resonances arose from the complex, indicating that the binding was no longer symmetrical and/or more than one binding mode existed. The proposal is supported by the observation of both chemical exchange crosspeaks in the aromatic-aromatic region of the TOCSY spectrum and three H5-H6 TOCSY peaks for each cytosine residue (data not shown). Therefore, an asymmetrical and a symmetrical binding mode coexist with similar binding affinities. This may indicate that AR-1-144 binds to CGTG more weakly than to CTGG (vide infra). The exchange rate between these two binding modes is slow on the NMR timescale. The complicated NOESY spectrum of the 2:1 drug/DNA solution, arising from two binding modes of the drug-DNA complex, precludes a complete assignment of the spectrum and subsequent structural analysis.
Surprisingly, when the core sequence was changed to TCGG, AR-1-144 did not bind to the DNA (datanot shown). Since T4 is not expected to play an important role in drug binding (based on the structure of the AR-1-144/GAACCGGTTC complex we solved), we suspect that a T:G mismatch may change the neighboring DNA conformation depending on the sequence context, so that AR-1-144 can no longer bind to the sequence. To address this question, the duplex structure of GAATCGGTTC was determined by NMR (data not shown). However, no unusual conformation was observed in the free duplex, i.e. the two wobble T:G base pairs are embedded in a normal B-DNA not unlike that found in other structures (7-9).
We have also titrated Im-Py-Im with CTGG and CGTG decamer solutions and showed no significant binding of the drug to either decamer (data not shown). The results confirm our proposal that only the Im/Im pair, and not the Py/Im pair, is a good motif for recognition of T:G/G:T wobble base pairs.
Structure of the AR-1-144/GAACTGGTTC complex
Allresonances of the 2:1 complex of AR-1-144 and GAACTGGTTC were assigned using 2D NOESY and TOCSY data. The sequential assignment in the aromatic-H1[prime] region is shown in Figure 3A. By comparing the NOESY spectrum of AR-1-144/GAACTGGTTC with that of AR-1-144/GAACCGGTTC, we note that many important intermolecular NOE crosspeaks are conserved in both complexes, e.g. ARH1-C4H1[prime] (peak a), ARH1-T5H1[prime] (C5H1[prime] in the CCGG complex) (peak b) and ARH5-2-G7H3[prime] (peak c) (Fig. 3A). These NOEs, plus those between the dimethylamine tail and DNA, indicate that AR-1-144 binds symmetrically in the DNA minor groove from residue 4 to residue 8 in the AR-1-144/GAACTGGTTC complex (Fig. 1B). The 10 starting models and 10 refined structures of the AR-1-144/GAACTGGTTC complex are shown in Figure 4. The refinement statistics are summarized in Table 1.
Figure 3. (A) Expanded regions of the observed and simulated non-exchangeable 2D NOESY spectrum of the 2:1 AR-1-144/GAACTGGTTC complex in D2O showing the sequential assignment pathway and some drug-DNA crosspeaks. Selected crosspeaks are: (a) ARH1-C4H1[prime]; (b) ARH1-T5H1[prime]; (c) ARH5-2-G7H3[prime]. (B) Expanded regions of the observed exchangeable proton 2D NOESY spectrum of the 2:1 AR-1-144/GAACTGGTTC complex in 90% H2O/10% D2O. The sequential imino proton connectivity of the DNA is shown by solid lines with arrows. DNA amino protons involved in crosspeaks are illustrated by dashed lines.
Figure 4. Ten starting (A) and refined (B) structures of the homodimer of AR-1-144 binding in the minor groove of GAACTGGTTC. The refined structures are shown in stereoscopic view.
When we examine the tandem T:G mismatch region, interesting features consistent with our hypothesis are observed. Thymine and guanine form a stable wobble base pair, consistent with the observation that T5H3 and G6H1 are well detected in H2O, indicative of both imino protons being hydrogen bonded. T5 moves into the major groove, whereas G16 stays essentially at the same position as that in the AR-1-144/GAACCGGTTC complex (Fig. 5A and B). The N2 amino group of G16, unlike that in a C:G base pair, is not involved in base pairing with T5 and forms two hydrogen bonds, each with one imidazole nitrogen of the Im/Im pair. However, in order to maintain the stacking between the two AR-1-144 molecules, the imidazole ring on the T5 side cannot go all the way into the cavity generated by T5 moving into the major groove. Therefore, the angle of the hydrogen bond between the imidazole nitrogen on the T5 side and one of the G amino protons is not optimized.
Figure 5. (A) An Im/Im pair recognizes a G:C base pair in the refined AR-1-144/GAACCGGTTC complex of Yang et al. (11). (B) An Im/Im pair recognizes a G:T base pair in the refined AR-1-144/GAACTGGTTC complex from this work. (C) A conserved H2O is observed bridging the G:T mismatch in the crystal structure of Hunter et al. (8). (D) Our hypothesis of using an Im/Hp pair to discriminate a G:T from a T:G mismatch.
Despite such a non-ideal hydrogen bonding interaction, the Im/Im pair forms two hydrogen bonds with the G:T base pair in the AR-1-144/GAACTGGTTC complex, instead of one weak bifurcated hydrogen bond with the G:C base pair in the AR-1-144/GAACCGGTTC complex. The binding affinity between drug and DNA is thereby increased significantly in the AR-1-144/GAACTGGTTC complex, the hydrogen bonding network of which is illustrated in Figure 6. Comparing it with that of the AR-1-144/GAACCGGTTC complex (11), it is clear that the remaining hydrogen bonds are essentially conserved.
Figure 6. Schematic representation of the homodimeric binding model of AR-1-144 to the minor groove of GAACTGGTTC. Hydrogen bonds (<3.3 Å) and potential hydrogen bonds (<3.6 Å) are indicated as dashed and dotted lines, respectively. The values of the hydrogen bond distances (Å) are measured between non-hydrogen atoms.
Exchangeable proton NMR data consistent with the refined structure
Additional information can be obtained from the exchangeable proton NMR data. All six DNA imino proton resonances of the AR-1-144/GAACTGGTTC complex were detected at 2°C, including the terminal G1 imino (Fig. 2A). Figure 3B shows the exchangeable proton 2D NOESY data of the imino-imino and imino-amino crosspeak regions. The sequential imino proton connectivity can be traced from G1 on one end to G6 in the center without interruption, suggesting that the duplex is mostly intact. The NOEs between the G imino protons and the C amino protons across the G:C base pairs and the T imino protons and AH2 protons across the A:T base pairs establish the Watson-Crick pairing at all base pairs except the T5:G16 mismatch.
Note that the G6N2/G7N2 amino protons are detected as well-resolved peaks, suggesting that the rotation about the C2-N2 bond in these two guanines is slow, consistent with the fact that the G N2 amino group is involved in the hydrogen bonding interaction with the Im nitrogen of AR-1-144 (Fig. 6). Finally, all exchangeable proton NOEs between AR-1-144 and DNA in the H2O spectrum are consistent with our refined structure, supporting the validity of the structure.
Heterodimer for the recognition of a single T:G mismatch
Self-complementary DNA sequences were used for most of the studies with AR-1-144 to simplify the analysis. Therefore, a tandem T:G/G:T mismatch was recognized in the sequence GAACTGGTTC by two Im/Im pairs. However, to recognize a single T:G mismatch is more biologically relevant since a tandem T:G/G:T mismatch is seldom found in vivo.
Taking together our results so far, we have designed a new system to recognize a single T:G mismatch. Specifically, we tested a new molecule, Im-Py-Im and used it together with AR-1-144 (Im-Im-Im) to form a side-by-side heterodimer in the DNA minor groove. A non-self-complementary DNA octamer duplex C1A2C3G4G5G6T7C8 + G9A10C11T12C13G14T15G16 has been designed with the following criteria. First, a single T:G mismatch is placed in the center of the duplex. Second, both single-strand DNAs do not form either duplex or hairpin by themselves. Finally, there is no alternative favorable binding site in the duplex.
The results show that Im-Im-Im can form a stable 2:1 complex with the DNA duplex (Fig. 7A), which suggests that the Im/Im pair can also specifically recognize a single T:G mismatch. However, Im-Py-Im alone does not bind to this octamer DNA (data not shown). In the C1A2C3G4G5G6T7C8 + G9A10C11T12C13G14T15G16 duplex, the G4:C13 base pair can be recognized by an Im/Im pair, as suggested by our earlier study (11). However, the binding affinity of an Im/Im pair for a G:C base pair is weaker than for an Im/Py pair. This is borne out by the following experiment. When Im-Py-Im is titrated into a Im-Im-Im/DNA solution, a new set of peaks other than that of the 2:1 Im-Im-Im/DNA complex appears (Fig. 7B), indicating that a 1:1:1 Im-Im-Im/Im-Py-Im/DNA ternary complex is formed. The peaks from the ternary complex abolish all the peaks from the 2:1 Im-Im-Im/DNA complex almost completely, suggesting that the 1:1:1 Im-Im-Im/Im-Py-Im/DNA complex is more stable than the 2:1 Im-Im-Im/DNA complex.
Figure 7. (A) Titration of AR-1-144 (III) with the C1A2C3G 4 G5G6T7C8 + G9A10C11T12C13G14T15G16 duplex. The spectra show a stable homodimeric binding of AR-1-144 to the duplex. (B) When Im-Py-Im (IPI) is added to the III/DNA solution, a more stable IPI/III/DNA ternary complex is formed. A new set of peaks other than that of the 2:1 III/DNA complex and the free DNA appears, abolishing all the peaks from the III/DNA complex almost completely. The barely remaining resonances from the 2:1 III/DNA complex are marked by asterisks.
DISCUSSION
Our results show that AR-1-144 binds to GAACTGGTTC through a tight and single binding mode, whereas it has multiple binding sites with GAACGTGTTC. This indicates that TGG is a better binding target for Im-Im-Im than GTG. A similar property has been observed on binding of AR-1-144 to GAACCGGTTC and GAACGCGTTC (Fig. 2B and D), i.e. Im-Im-Im binds tighter to CGG than to GCG (11). Our interpretation of such an observation is as follows. Since the peptide bond is a rigid linker, it gives the AR-1-144 molecule an essentially fixed curvature. When a pyrimidine is sandwiched between two bulky guanine residues, the contact between this pyrimidine and the second Im unit is less favorable. This situation becomes more serious for T (paired with G) than for C due to the wobble T:G base pairing. However, because of the additional hydrogen bond between an Im/Im pair and a T:G base pair, AR-1-144 still binds better to CGTG than to CGCG, as proved by the NMR titration spectra (Fig. 2C and D).
Although an Im/Im pair is a good motif for a T:G/G:T mismatch, it cannot discriminate T:G from G:T. Such discrimination may be achieved by another MGB motif. An important hint comes from the crystal structures of T:G wobble base pairs. Interestingly, a conserved H2O molecule has been found bridging the T and G bases in the minor groove of DNA in all three forms (i.e. B, A and Z) (6-8; Fig. 5C). This observation suggests that an Hp unit could be a good candidate for the recognition of a T:G mismatch from the T side, because the hydroxy group in Hp may fulfill the role of the bridging water (Fig. 5D). In other words, we propose that the Hp/Im pair may be a good motif for a T:G mismatch but not for G:T and, conversely, Im/Hp is good for G:T but not for T:G. Further experiments will be carried out to test this hypothesis.
In summary, our NMR results provide evidence that an Im/Im pair shows strong binding affinity for sequences containing a T:G wobble base pair. Therefore, while an Im/Im pair may not have ideal sequence recognition properties towards Watson-Crick G:C base pairs, it may be used in certain specifically designed MGBs to probe non-canonical DNA structures. For example, one may design a linked Im-Py-Im/Im-Im-Im dimer attached to a duocarmycin headpiece which may be used to recognize and cleave a specific sequence containing a T:G mismatch. Those studies are underway and some of the results have been presented elsewhere (21).
ACKNOWLEDGEMENTS
We thank Dr Howard Robinson for the SPEDREF structure refinement package and Mr Dong Guo for the discussion on the biological relevance of T:G mismatches. This work was supported by American Cancer Society grant RPG-94-014-05 to A.H.-J.W., NIH grant R15 CA56901-01) to M.L., and a Grant-in-Aid for Priority Research from the Ministry of Education, Science, Sports, and Culture, Japan to H.S. The Varian VXR500 NMR spectrometer at UIUC was supported in part by NIH shared instrumentation grant 1S10RR06243. The 750 MHz NMR spectra were obtained in the Varian Oxford Instrument Center for Excellence in NMR Laboratory. Funding for this instrumentation was provided in part from the W.M. Keck Foundation, the National Institutes of Health (PHS 1 S10RR10444-01), and the National Science Foundation (NSF CHE 96-10502).
REFERENCES
*To whom correspondence should be addressed. Tel: +1 217 244 6637; Fax: +1 217 244 3181; Email: ahjwang{at}life.uiuc.edu
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:
|
|
 |
|
  M. Bahr, V. Gabelica, A. Granzhan, M.-P. Teulade-Fichou, and E. Weinhold Selective recognition of pyrimidine-pyrimidine DNA mismatches by distance-constrained macrocyclic bis-intercalators Nucleic Acids Res., September 1, 2008; 36(15): 5000 - 5012. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K. L. Buchmueller, A. M. Staples, P. B. Uthe, C. M. Howard, K. A. O. Pacheco, K. K. Cox, J. A. Henry, S. L. Bailey, S. M. Horick, B. Nguyen, et al. Molecular recognition of DNA base pairs by the formamido/pyrrole and formamido/imidazole pairings in stacked polyamides Nucleic Acids Res., February 9, 2005; 33(3): 912 - 921. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
E. R. Lacy, B. Nguyen, M. Le, K. K. Cox, C. O'Hare, J. A. Hartley, M. Lee, and W. D. Wilson Energetic basis for selective recognition of T{middle dot}G mismatched base pairs in DNA by imidazole-rich polyamides Nucleic Acids Res., April 2, 2004; 32(6): 2000 - 2007. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
E. R. Lacy, K. K. Cox, W. D. Wilson, and M. Lee Recognition of T{middle dot}G mismatched base pairs in DNA by stacked imidazole-containing polyamides: surface plasmon resonance and circular dichroism studies Nucleic Acids Res., April 15, 2002; 30(8): 1834 - 1841. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||