PDB 1aul.pdb.
ABSTRACT
The tripeptide 1,2-dihydro-(3H)-pyrrolo[3,2-e]indole-7-carboxylate (CDPI3) binds to the minor groove of DNA with high affinity. When this minor groove binder is conjugated to the 5'-end of short oligonucleotides the conjugates form unusually stable hybrids with complementary DNA and thus may have useful diagnostic and/or therapeutic applications. In order to gain an understanding of the structural interactions between the CDPI3 minor groove binding moiety and the DNA, we have determined and compared the solution structure of a duplex consisting of oligodeoxyribonucleotide 5'-TGATTATCTG-3' conjugated at the 5'-end to CDPI3 and its complementary strand to an unmodified control duplex of the same sequence using nuclear magnetic resonance techniques. Thermal denaturation studies indicated that the hybrid of this conjugate with its complementary strand had a melting temperature that was 30°C higher compared with the unmodified control duplex. Following restrained molecular dynamics and relaxation matrix refinement, the solution structure of the CDPI3-conjugated DNA duplex demonstrated that the overall shape of the duplex was that of a straight B-type helix and that the CDPI3 moiety was bound snugly in the minor groove, where it was stabilized by extensive van der Waal's interactions.
Long single-stranded DNA and RNA have considerable secondary and tertiary structure that can interfere with the use of oligodeoxyribonucleotides (ODNs) as diagnostic probes or antisense agents (1-3). One approach to overcoming these barriers involves the design of chemically modified ODNs that form unusually stable hybrids or triplexes that can favorably compete with existing secondary structures. Examples include ODNs conjugated to an intercalator (4), peptide nucleic acids (5) and ODNs containing C-5 propynyl (6) or N3'-P5' (7) groups.
ODNs conjugated to a minor groove binder represent another strategy to form stabilized hybrids. (+)-CC-1065 (Fig. 1) is a natural product that binds to and alkylates DNA in the minor groove and exhibits potent antitumor and antibiotic activity (8-9). Recently conjugation of N-3-carbamoyl-1,2-dihydro-3H-pyrrolo[3,2-e]indole-7-carboxylate (CDPI3; Fig. 1), a non-reactive analog of CC-1065, to the ends of various ODNs has been reported (10). This conjugate was shown to increase the melting temperature of DNA hybrids by as much as 44°C when covalently linked to one strand of an A/T-rich duplex. These conjugates also formed stabilized hybrids with complementary RNA targets and with G/C-rich DNA (11). Subsequent experiments demonstrated that CDPI3 covalently linked to the 5'-end of ODNs effectively blocked primer extension by a DNA polymerase (12). Polymerase blockage was abolished when a single mismatch was introduced into the oligomer or when the minor groove binder was either removed or replaced by a 5'-acridine group. More recently CDPI3 conjugates as short as 8-10mers have been used as primers in PCR to amplify DNA with good specificity and efficiency (13). These results and the other properties mentioned above suggest that DNA conjugates of this type may have useful diagnostic and/or therapeutic applications.
Synthesis of the DNA oligomer d(TGATTATCTG) containing a minor groove binding drug, CDPI3, conjugated to the 5'-end of the dT1 residue with a hexamethylene linker and its complementary sequence d(CAGATAATCA) was accomplished as reported earlier, with some modifications (10). Specifically, two methods for preparation of the 5'-CDPI3 conjugates were compared. The first method used a 10 µmol trityl-off synthesis of the 5'-aminohexyl-modified ODN and treatment of the CPG with 5 equiv. (38 mg) CDPI3 activated ester. Ammonia hydrolysis gave a mixture of products that were separated by reverse phase HPLC as described earlier to give 1.54 mg (5% yield) of the desired conjugate. The second method used HPLC-purified 5'-aminohexyl-modified ODN from a 10 µmol scale synthesis (7.4 mg, 2.3 µmol). The triethylammonium form of the ODN was dissolved in 0.5 ml DMSO and treated with 3 equiv. (6.3 mg) CDPI3 activated ester and 50 µl triethylamine. After 22 h the crude conjugate was precipitated with 2% sodium perchlorate/acetone. Purification by reverse phase HPLC gave 3.96 mg product (42% yield from amine-modified ODN). Re-precipitation gave the sodium salt. The purified conjugates from each method were combined and used for further experiments. The unmodified control duplex was purchased from DNAgency Inc. The purity of these sequences was checked by 1H NMR and the control and CDPI3-modified duplexes were prepared by annealing each oligomer with its complementary sequence as described (14).
Hybrids formed between minor groove binder conjugates or unmodified oligonucleotides and their complements were melted at a rate of 0.5°C/min in 1* PBS (9.2 mM disodium phosphate, 0.8 mM monosodium phosphate, 131 mM NaCl, pH 7.2) on a Lambda 2S (Perkin Elmer) spectrophotometer with a PTP-6 automatic multicell temperature programer. Each oligonucleotide (4 µM) was mixed with sufficient complement to give a 1:1 ratio. Prior to melting samples were denatured at 100°C and then cooled to the starting temperature over a 10 min period. The melting temperatures of the duplexes were determined from the derivative maxima.
All NMR experiments were performed on a Varian 600 MHz Unity Plus NMR spectrometer at a regulated temperature of 20°C. NOESY, DQF-COSY and TOCSY data were acquired in the phase-sensitive mode with 2048 (4096 for the DQF-COSY) complex points in t2 and 512 (1024 for the DQF-COSY) complex points in t1, a relaxation delay of 2.5 s (during which time the residual HDO peak was irradiated) and a spectral width of 6250 Hz. The TOCSY experiment was carried out with a 120 ms mixing time. Four NOESY spectra for the control and CDPI3-conjugated DNA samples with mixing times of 80, 120, 160 and 200 ms were each collected within separate 3 day periods, without removing the sample from the spectrometer. A 200 ms NOESY spectra for the control and the CDPI3-conjugated DNA in H2O were acquired with an excitation sculpting pulse sequence for solvent suppression (15). Selective 180° pulses were applied with a SEDUCE profile to suppress the water resonance. Gradient pulses were applied along the z-axis for 1 ms at 15 and 3 G/cm with 50 µs delays before and after the gradient pulses. These experiments were collected with States-TPPI phase cycling with 2048 complex points in t2 and 1024 complex points in t1, a relaxation delay of 2.5 s and a spectral width of 12 500 Hz.
NMR data were processed using FELIX 95.0 (Molecular Simulations Inc.). NOESY and TOCSY data sets acquired in D2O or H2O were processed with a squared sinebell shifted 90° window function in the direct dimension and a squared sinebell shifted 90° window function in the indirect dimension. Data in both dimensions were zero filled to 2048 real points, Fourier transformed, phased to pure absorption and baseline corrected with a fifth order polynomial. NOE cross-peak volumes were obtained by cross-peak integration using FELIX software. The DQF-COSY data sets were apodized with unshifted sinebells in both dimensions, followed by zero filling to 2048 real points in both dimensions prior to Fourier transformation. 1H resonances are referenced to an external sample of sodium 2,2-dimethyl-2-silapentane-5-sulfonate in identical buffer.
Interproton distances were calculated from initial NOE build-up rates by fitting the cross-peak volumes at 80, 120, 160 and 200 ms mixing times to a linear equation. These rates (Rij) were converted to distances (rij) using the cytosine H5-H6 distance of 2.45 Å as the reference distance (rref), the averaged rate calculated from well-resolved H5-H6 cross-peaks (Rref) and the relationship rij = rref (Rref/Rij)1/6. Cross-peak volumes obtained from the exchangeable protons in the 200 ms NOESY acquired in H2O were also referenced to the H5-H6 interproton distance of 2.45 Å. These distances were given upper and lower bounds according to the following guidelines: for rij < 2.0 Å the errors given were -0.2/+0.2 Å; for 2.0 Å < rij < 2.5 Å they were -0.2/+0.4 Å; for 2.5 Å < rij < 3.3 Å they were -0.3/+0.5 Å; for 3.3 Å < rij < 6.0 Å they were -0.5/+0.7 Å (16,17). All distances involving exchangeable protons were given upper and lower bounds of 0.7 Å. Base pairs were kept hydorgen bonded in the Watson-Crick form by using distance restaints between the bases as previously described (16).
A starting structure for the CDPI3-conjugated duplex was generated by constructing the DNA duplex in canonical B-type form and then docking the CDPI3 moiety into the minor groove of the duplex using InsightII software (Molecular Simulations Inc.). The structures for both the control and CDPI3 duplexes were refined by restrained molecular dynamics using the program X-PLOR 3.1 (18) as previously described (16). A total of 210 or 175 experimentally determined interproton distance restraints were used in the control and CDPI3 refinements, respectively. At this stage of the calculations an effort was made to eliminate those distances from interproton interactions where contamination from spin diffusion was most likely [e.g. H3'(i)-H8/H6(i+1)]. As described in more detail below, sugar torsion angles were approximated by a qualitative estimation of JH2''-H3' and JH3'-H4' from high resolution DQF-COSY data and from NOESY-determined distances involving sugar protons (19). In addition to the experimentally determined distance and sugar dihedral restraints, the [alpha], [beta], [gamma], [epsilon] and [zeta] backbone torsion angles were restrained to a range covering all right-handed DNAs (16,17). The restrained molecular dynamics procedure was repeated six times from each starting structure, using different random number seeds for each subsequent molecular dynamics run. The structure calculations were carried out in an iterative manner by adding more CDPI3-DNA NOEs to each cycle to resolve potentially ambiguous assignments. Each restrained molecular dynamics structure was further refined using complete relaxation matrix refinement using the RELAX option of X-PLOR as reported earlier (16); a total of 1080 or 840 NOE intensities were used in refinement of the control and CDPI3 structures, respectively. No backbone torsion angles were used in this phase of the calculations. Analyses of helical parameters were performed with the program Dials-and-Windows (kindly provided by Dr G.Ravishanker, Wesleyan University) (20).
In order to understand the mode of binding, stability and sequence specificity of DNA conjugated to CDPI3 we have undertaken a structural study of a 5'-CDPI3 conjugate of the DNA decamer 5'-TGATTATCTG-3' hybridized to its complementary sequence 3'-ACTAATAGAC-5'. Preparation of these minor groove binder-oligodeoxyribonucleotide conjugates has been previously reported by Lukhtanov et al. (10) and was accomplished by reaction of the 2,3,5,6-tetrafluorophenyl ester of CDPI3 with a 5'-aminohexyl phosphate ester. Two methods were compared for conjugating CDPI3 to the amine-modified ODN. Treatment of the CPG-immobilized ODN was less labor intensive, but HPLC separation of the desired product was more difficult. Modification of pre-purified amine-modified ODN was found to give easier isolation, better yields and consumed less of the valuable CDPI3 activated ester. UV melting studies on the conjugated DNA indicated a melting temperature of 61°C for this duplex versus a melting temperature of 30°C for an unmodified control duplex of the same sequence. This large increase in stability of the conjugated duplex was comparable with increases previously observed in other CDPI3-conjugated duplexes (10-13).
The non-exchangeable base protons (purine H8 and pyrimidine H6) and the sugar protons (H1', H2', H2'', H3' and H4') for the control and CDPI3-conjugated duplexes were assigned on the basis of an analysis of through-space distance connectivities in NOESY data sets as a function of mixing time and through-bond connectivities in COSY data sets recorded in D2O buffer. Theexpanded NOESY (400 ms mixing time) contour plot establishing sequential connectivities between the base protons and the sugar H1' protons of the CDPI3-conjugated duplex is plotted in Figure 2. A similar sequential tracing could be made from the base protons to the H2'/H2'' protons (not shown).
Qualitative analysis of the NOE data obtained for both the control and CDPI3-conjugated duplexes revealed the general pattern of NOEH2'(i)-H8/H6(i) >> NOEH2''(i - 1)-H8/H6(i) > NOEH2'(i - 1)-H8/H6(i). This pattern is characteristic of right-handed B-type structures. Analysis of the DQF-COSY data indicated that the control duplex contained a variety of sugar pucker conformations (C1'-exo, C2'-endo and O4'-endo) found in the B-type DNA family. The complete absence of H3'-H4' and H2''-H3' cross-peaks in the DQF-COSY spectrum of the CDPI3-conjugated duplex (data not shown) suggests that the sugar puckers of the modified DNA were primarily in a C2'-endo conformation (21). Based on these qualitative observations, B-type duplexes were used as starting structures for the molecular dynamics structure calculations. In these calculations sugar puckers were restrained with relatively wide bounds (P = 120-160°).
Using distance data, NOE cross-peak volumes and sugar pucker dihedral angle restraints obtained from DQF-COSY and NOESY experiments, structures of the modified and control duplexes were obtained using restrained molecular dynamics and complete relaxation matrix refinement (Fig. 4). The pairwise root mean square (r.m.s.) deviations for the set of control and CDPI3-conjugated structures were 0.54 ± 0.13 and 0.83 ± 0.10 Å respectively. Structural statistics for the starting structures, ensemble of restrained molecular dynamics structures and ensemble of relaxation matrix-refined structures are shown in Table 3. The small deviations from idealized geometry indicate that the calculations have not resulted in any distortion of the covalent structure. Following relaxation matrix refinement there are small increases in the r.m.s. deviation from the experimental distance and dihedral restraints. This is due to the isolated spin approximation used to calculate the distances used in the restrained molecular dynamics calculations. The lack of backbone dihedral restraints during the relaxation matrix refinement procedure leads to the increase in the r.m.s. deviation from the dihedral angle restraints. The R1/6 factors (which reflect the degree of agreement between experimental NOE volumes and the NOE volumes back-calculated from the structures) were reduced from 0.129 and 0.156 in the starting structures to 0.086 ± 0.001 and 0.083 ± 0.001 in the final ensembles of relaxation matrix-refined control and CDPI3-conjugated duplex structures respectively. This reduction in R1/6 factors for the refined structures in comparison with the starting structures demonstrates convergence to structures that are in good agreement with the experimental data.AB
The minor groove binding drugs netropsin (22), distamycin (23) and their structural analogs (24-25) have been reported to bind isohelically into the AT-rich minor groove of the DNA double helix, owing to their crescent-shaped structure. These molecules have planar N-pyrrole moieties and fit edge-on into the minor groove, replacing the spine of hydration and follow the natural curvature of the DNA. This type of molecular recognition is at least partially driven by formation of hydrogen bonds between the drug and the DNA.
In the case of CC-1065 binding of the drug in the minor groove of the DNA is accompanied by opening of the cyclopropyl ring of the drug to form a covalent bond with the N3 position of an adenine base. Using gel electrophoresis, hydroxyl radical footprinting and NMR studies, Hurley and co-workers have extensively characterized the structural consequences of CC-1065 bonding to DNA (26-28). They concluded from these studies that covalent modification of the DNA by CC-1065 results in kinking of the DNA by 17-22° with an overall direction and magnitude of bending similar to those of an A6 tract sequence (27). It was also reported that a narrowing of the minor groove at the binding site occurred as a result of (+)-CC-1065 binding, possibly as a consequence of hydrophobic contacts between the drug and the minor groove of the DNA. This region of narrowing was sandwiched between widened regions (27). Finally, it was proposed that the inherent propensity of a DNA sequence to form a bent structure was related to its preference as a target site for binding of (+)-CC-1065 (28).
In an NMR study on another analog of CC-1065, CPI-CDPI2, Powers and Gorenstein (29) reported the presence of two conformers of the drug-DNA adduct in solution. The major isomer had a kink of 60° at the site of alkylation, similar to the sequence-specific bends reported in several DNA structures (30-31). The general binding mode of the CPI-CDPI2 moiety to the DNA was similar to that reported by Hurley and co-workers for the CC-1065 adduct, which was not surprising since both of these molecules formed an N3 adduct with adenosine. In the CPI-CDPI2 structure, however, bending of the DNA was much greater than that observed in CC-1065-modified DNA structures; furthermore, the minor groove was widened at the site of CPI-CDPI2 binding rather than narrowed, as in the case of CC-1065.
The molecule under study in this paper has two major features that distinguish it from those described above. First, no adduct is formed between the CDPI3 moiety and any group in the minor groove of the DNA. Second, the CDPI3 moiety is attached to one strand of the DNA duplex by a flexible hexamethylene linker. Despite these differences, the structure of the CDPI3-conjugated DNA duplex that we report here has some similarities to the structures of other DNA complexes with minor groove binding agents. Like other crescent-shaped minor groove binders, the CDPI3 moiety fits snugly into the minor groove of the DNA duplex (Figs 4 and 5) even though there are no hydrogen bonds nor adduct formation occurring in the minor groove to stabilize this binding mode. In the minor groove van der Waal's forces appear to provide the major factor in stabilizing the conjugated duplex. The compact binding mode that is observed produces a narrowing of the minor groove at the drug binding site and a widening of the groove distal to the binding site; as mentioned above, a similar observation was reported by Lin et al. (27).
Several notable variances were observed, however, between our results and those obtained in related studies. First, only one conformer of the CDPI3-conjugated DNA duplex was evident from the NMR data (in Fig. 2 cross-peaks involving proton H15 of the CDPI3 moiety may show evidence of slow exchange); this was in contrast to the two conformers that were observed for the CPI-CDPI2-DNA adduct (29). Indeed, the sugar puckers in the CDPI3-conjugated duplex were all C2'-endo, whereas the control duplex exhibited a variety of pseudorotation angles, suggesting that binding of the CDPI3 moiety places additional constraints on mobility of the sugar-phosphate backbone. Second, neither the structure of the control non-conjugated duplex nor the structure of the conjugated DNA duplex is noticeably bent, indicating that binding of the CDPI3 moiety does not induce bending in this DNA sequence. Most adducts of DNA and minor groove binders have been reported to result in bending of the duplex, although a lack of bending was also observed in the structure of a DNA-CPI-lexitropsin adduct determined by Lown and co-workers (32). Finally, neither the Hurley nor the Gorenstein groups used any drug-DNA intermolecular NOE cross-peaks in their structure refinement analyses, whereas 12 intermolecular cross-peaks were employed in the calculations used to generate the structure of the conjugated duplex shown in Figure 4.
It has been known for some time that a bulky amino group at position 2 in guanine in the minor groove, sterically interferes with binding of minor groove binding agents to DNA sequences containing G·C base pairs (33-34). The sequence employed in our study is definitely A/T-rich, yet it does have a G·C base pair in the CDPI3 binding site. While we observe some close van der Waal's contacts between the amino group of the G2 residue and atoms of the CDPI3 moiety, this lone G·C base pair certainly does not preclude CDPI3 binding. It is plausible that the attachment of the drug to the DNA by a flexible hexamethylene linker can explain this and some of the other features of the CDPI3-conjugated duplex. The flexibility of this linker permits the CDPI3 moiety to adjust its position to fit snugly in the minor groove despite the presence of the 2-amino group on G2. Formation of DNA adducts by CC-1065 and CPI-CDPI2 upon binding in the minor groove may be accompanied by conformational stress on the DNA backbone that may promote bending of the helix. The lack of bending in CDPI3-conjugated DNA and its very high melting temperature may be at least partially due to the absence of conformational stress that is permitted by attaching the CDPI3 moiety to the DNA strand by a flexible linker (35).
Nucleic Acids Research
Pages
Introduction
Materials And Methods
Sample preparation
Thermal denaturation studies
NMR experiments
Structure calculations
Results And Discussion
NMR resonance assignments and chemical shift analysis
Solution structure of the CDPI3-conjugated DNA duplex
Comparison with other minor groove binders
References
REFERENCES
This page is run by Oxford University Press, Great Clarendon Street, Oxford OX2 6DP, as part of the OUP Journals
Comments and feedback: www-admin{at}oup.co.uk
Last modification: 23 Jan 1998
Copyright© Oxford University Press, 1998.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  R. Luderer, A. Verheul, and W. Kortlandt Rapid Detection of the Factor V Leiden Mutation by Real-Time PCR with TaqMan Minor Groove Binder Probes Clin. Chem., April 1, 2004; 50(4): 787 - 788. [Full Text] [PDF]
|
|
|
|
 |
|
 I. V. Kutyavin, S. G. Lokhov, I. A. Afonina, R. Dempcy, A. A. Gall, V. V. Gorn, E. Lukhtanov, M. Metcalf, A. Mills, M. W. Reed, et al. Reduced aggregation and improved specificity of G-rich oligodeoxyribonucleotides containing pyrazolo[3,4-d]pyrimidine guanine bases Nucleic Acids Res., November 15, 2002; 30(22): 4952 - 4959. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
I. V. Kutyavin, I. A. Afonina, A. Mills, V. V. Gorn, E. A. Lukhtanov, E. S. Belousov, M. J. Singer, D. K. Walburger, S. G. Lokhov, A. A. Gall, et al. 3'-Minor groove binder-DNA probes increase sequence specificity at PCR extension temperatures Nucleic Acids Res., January 15, 2000; 28(2): 655 - 661. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||