Recognition of GC base pairs by triplex forming oligonucleotides containing nucleosides derived from 2-aminopyridine
Recognition of GC base pairs by triplex forming oligonucleotides containing nucleosides derived from 2-aminopyridineSarah A. Cassidy, Peter Slickers1,+, John O. Trent1, Daniel C. Capaldi2, Peter D. Roselt2, Colin B. Reese2, Stephen Neidle1 and Keith R. Fox*
Division of Biochemistry and Molecular Biology, School of Biological Sciences, University of Southampton, Bassett Crescent East, Southampton SO16 7PX, UK, 1CRC Biomolecular Structure Unit, The Institute for Cancer Research, Cotswold Road, Sutton, Surrey SM2 5NG, UK and 2Department of Chemistry, King's College London, Strand, London WC2R 2LS, UK
Received September 19, 1997;Revised and Accepted November 3, 1997
ABSTRACT
We have attempted to alleviate the pH dependency of triplex recognition of guanine by using intermolecular triplexes containing 2-amino-5-(2-deoxy-d-ribofuranosyl)pyridine (AP) as an analogue of 2'-deoxycytidine (dC). We find that for the [beta]-anomer of AP, the complex between (AP)6T6 and the target site G6A6-T6C6 is stable, generating a clear DNase I footprint at oligonucleotide concentrations as low as 0.25 µM at pH 5.0, in contrast to 50 µM C6T6 which has no effect on the cleavage pattern. This complex is still stable at pH 6.5 producing a footprint with 1 µM oligonucleotide. Oligonucleotides containing the [alpha]-anomer of AP are much less effective than the [beta]-anomer, though in some instances they are more stable than the unmodified oligonucleotides. The results of molecular dynamics studies on a range of AP-containing triplexes has rationalized the observed stability behaviour in terms of hydrogen-bonding behaviour.
The formation of DNA triple helices offers the possibility of designing sequence specific DNA binding agents, which may have therapeutic and experimental uses (1 -4 ). The third strand oligonucleotide binds in the major groove of polypurine tracks of duplex DNA and is held in place by specific hydrogen bonds to substituents on the bases (1 ,2 ). Two types of triple helix have been characterised which differ according to the orientation of the third strand. Pyrimidine-rich oligonucleotides bind parallel to the duplex purine strand forming T-AT and C+-GC triplets (5 -7 ) while purine-rich oligonucleotides bind in an anti-parallel orientation and are characterised by G-GC, A-AT and T-AT triplets (8 -11 ).
One limitation to the formation of parallel (Y-RY) triplexes is that conditions of low pH are necessary for protonation of the third strand cytosine in the C+-GC triplet. The free base has a pKa of ~4.35, though this may be elevated within triplex forming oligonucleotides which may be stable at a pH up to 6.0, depending on the number and location of the cytosine residues. Contiguous C+-GC triplets are especially unstable (12 ), presumably on account of the proximity of the charged bases. A number of cytosine analogues and mimics have been synthesised in attempts to overcome this restriction. 5-Methylcytosine, which has a slightly higher pKa value than cytosine (13 ,14 ) generates triplexes which are more stable at a higher pH, but are still not formed under physiological conditions (12 ). This increase in stability may be a consequence of the extra spine of methyl groups within the DNA major groove (15 ) rather than from changes in the pKa. A carbocyclic analogue of 5-methylcytosine also has a higher pKa than cytosine (4.80 instead of 4.35) and has been shown to enhance triplex formation at elevated pHs (16 ). 6-Oxocytosine (17 ), pseudoisocytosine (18 ) and 8-oxoadenine (19 ,20 ) have also been shown to recognize GC base pairs in a pH-independent fashion, though they not have been widely used in triple-forming oligonucleotides. Other promising dC analogues are [beta]- and [alpha]-amino-5-(2-deoxy-d-ribofuranosyl)pyridines ([beta]- and [alpha]-AP) (Fig. 1 B and C) which have pKas of 5.93 and 6.16 respectively (21 ). Psoralen-linked oligonucleotides containing these modifications have been successfully used to target a portion of the aromatase gene (21 ). These initial studies with AP-containing oligonucleotides suggested that both [beta]- and [alpha]-anomers of this nucleoside analogue could form stable triplets. The possibility of forming complexes with [alpha]-anomers was especially interesting since [alpha]-oligonucleotides are more resistant to serum nucleases. However these studies used psoralen-linked oligonucleotides which form DNA cross-links after UV-irradiation and might promote the formation of less stable complexes. Initial modelling studies also confirmed the possibility of forming a triplet with [alpha]-AP, though this consisted of a single [alpha]-AP-GC triplet within a block of T-AT triplets. Recent studies have also shown that oligonucleotides containing AP and its 3-methyl derivative bind with higher affinity to duplex DNA targets between pH 6 and 8 than 5-methylcytosine containing oligonucleotides (22 ,23 ). This difference was especially pronounced for sequences containing contiguous cytosines.
In this paper we use DNase I footprinting to examine triple helix formation at the sequence G6A6-T6C6 using a series of modified oligonucleotides based on C6T6, in which the cytosine are replaced with 2-aminopyridine residues. Oligonucleotides containing either the [beta]- or [alpha]-anomers of 2-AP were tested. We have previously shown that unmodified CT-containing oligonucleotides show little interaction with this target site, even at pH 5.5. We find that oligonucleotides containing the [beta]-anomer form stable complexes at higher pH than cytosine-containing oligonucleotides. We also report on results from more extensive molecular modelling studies on sequences analogous to those examined by footprinting.
C6T6 was purchased from Oswel DNA Service. Modified oligonucleotides containing 2'-deoxy-5-methylcytidine (MeC) and [alpha]- and [beta]-AP were synthesized as previously reported (21 ) from the appropriate phosphoramidite building blocks in a Applied Biosystems 381A DNA synthesizer. The amino functions in the [alpha]- and [beta]-AP derived phosphoramidites were protected with dichloroacetyl groups, and dichloroacetic anhydride was used instead of acetic anhydride in the capping steps. The abbreviations used for these oligonucleotides, which were designed to interact with the target site G6A6-T6C6, are shown in Figure 1 D. The integrity of the oligonucleotides was confirmed at various times by labelling with [[gamma]- 32P]ATP using polynucleotide kinase, and examining the products on denaturing polyacrylamide gels. In each case >90% of the material ran as the intact 12mers. The naphthylquinoline triplex-binding ligand (24 -26 ) was a generous gift from Dr L.Strekowski, Department of Chemistry, Georgia State University.
The preparation of plasmid pGA1 has been previously described (27 ). This contains the sequence G6A6-T6C6 cloned into the BamH1 site of pUC19. A radiolabelled restriction fragment containing this sequence was prepared as previously described by cutting the plasmid with HindIII, labelling at the 3'-end with [[alpha]-32P]dATP using reverse transcriptase and cutting again with EcoR1. This procedure labels the pyrimidine-containing strand of the triplex target site. The fragment of interest was resolved from the remainder of the plasmid DNA on an 8% non-denaturing polyacrylamide gel. This was eluted from the gel slice, precipitated with ethanol and redissolved in 10 mM Tris-HCl, pH 7.5 containing 0.1 mM EDTA.
DNase I footprinting was performed as previously described (24 ,26 -28 ). An aliquot (1.5 µl) of labelled DNA (~10 nM, dissolved in 10 mM Tris, pH 7.5 containing 0.1 mM EDTA) was mixed with 1.5 µl oligonucleotide, dissolved in an appropriate buffer as indicated in the text. All buffers contained 5 mM MgCl2, necessary to stabilise triple helix formation. This mixture was left to equilibrate at room temperature for at least 1 h before adding 2 µl DNase I (0.1 U/ml dissolved in 20 mM NaCl, 2 mM MgCl2, 2 mM MgCl2). The reaction was stopped after 1 min by adding 4 µl of formamide containing 10 mM EDTA and 0.1% (w/v) bromophenol blue. Samples were boiled for 3 min and cooled rapidly on ice before loading onto 11% polyacrylamide gels containing 8 M urea. Gels of 40 cm length were run at 1500 V for about 2 h, then fixed in 10% (v/v) acetic acid, transferred to Whatmann 3MM paper and dried at 80°C. Dried gels were exposed to autoradiography at -70°C using an intensifying screen. Bands in the autoradiographs were assigned by comparison with Maxam-Gilbert marker lanes specific for G+A.
Initial oligonucleotide models were constructed with G6A6-T6C6 duplexes in an idealised B-DNA conformation. The bases of the third strand were placed in the major groove of the duplex, connected together with sugar-phosphate backbones, and then energy minimised in vacuo with a distance dependent dielectric constant and with distance restraints to force Hoogsteen hydrogen bonding of the bases. The five dodecamer triplex models constructed had third strand of sequence [beta]6T6, [alpha]6T6, [alpha]3[beta]3T6, [beta]3[alpha]3T6 and ([alpha][beta])3T6.
We have examined the ability of 2-AP to act as a cytosine mimic for recognising GC base pairs in a pH-independent fashion by studying triple helix formation at the target site G6A6-T6C6. We chose this target site since we have previously shown that the cytosine-containing oligonucleotide C5T5 does not form a particularly stable triplex, presumably as a result of the run of contiguous C+-GC triplets. This triplex only forms at relatively high oligonucleotide concentrations, at low pH (<5.5) and low temperatures (4°C). Figure 2 A shows DNase I digestion of a fragment containing this target site, and compares the binding of the various AP-containing oligonucleotides with C6T6 and 5MeC6T6 at pH 5.0. It can be seen that, at this third strand concentration (10 µM), [beta]6 is the only oligonucleotide which shows any interaction, producing a clear footprint covering the entire triplex target site. Neither the C6T6 or 5MeC6T6 show any interaction with the duplex target. In order to examine whether the interaction with [beta]6 extends to higher pH this experiment was repeated at pH 5.5 (Fig. 2 B), pH 6.0 (Fig. 2 C), pH 6.5 (Fig. 2 D) and pH 7.0 (Fig. 2 E). Clear footprints with 10 µM [beta]6 are evident up to pH 6.5, though no protection is evident at pH 7.0. Further studies with higher concentrations of [beta]6 (50 µM, not shown) reveal clear footprints at pH 7.0, but not pH 7.5. These results clearly demonstrate that replacement of dC by [beta]-AP residues dramatically enhances the oligonucleotide binding affinity at pH 5.0 and stabilises the complex at elevated pH. Furthermore this stabilisation is much greater than that achieved with 5-methylcytosine. It should be noted that the experiments performed at pH 6.0 and above used lower ionic strength buffers. Since triplex stability decreases at lower ionic strengths these experiments will underestimate the affinities at higher pHs.
The results presented above clearly show that [beta]6 generates the best complex and that this is still stable at pH 6.5, even though it generates a block of six contiguous AP-GC triplets. We have investigated the binding of [beta]6 to G6A6-T6C6 in more detail by examining concentration dependence of the footprints at pH 5.0 and 6.5. The results are presented in Figure 4 . It can be seen that at pH 5.0, [beta]6 produces a footprint which persists to concentrations as low as 0.25 µM (Fig. 4 A). Since 50 µM of the cytosine-containing oligonucleotide shows no interaction under the same conditions (Fig. 3 A), it appears that replacement of cytosine with [beta]-AP enhances the oligonucleotide binding affinity by >200-fold. A similar experiment at pH 6.5 is presented in Figure 4 B. In this case, [beta]6 generates a DNase I footprints down to a concentration of 1 µM. This represents an increase in stability (compared to the unmodified oligonucleotide at pH 5.0) of at least 50-fold.
Several studies have shown that a naphthylquinoline triplex-binding ligand can enhance the stability of complexes containing T-AT triplets, reducing the oligonucleotide concentration required to generate a footprint by as much as 200-fold. We were therefore interested to see whether this compound could stabilise the complexes formed with oligonucleotides containing [alpha]- and [beta]-AP. Figure 5 A examines the effect of 10 µM of this compound on the interaction of 10 µM AP-modified oligonucleotides with G6A6-T6C6, at pH 5.0. In the absence of the ligand, the only oligonucleotide which shows any interaction at 10 µM is [beta]6 (Fig. 2 ). In the presence of the ligand clear footprints are evident with the oligonucleotides containing cytosine and 5-methylcytosine residues and for [beta]6, ([alpha][beta])3 and [alpha]3[beta]3. No interaction is detected with [alpha]6 or [beta]3[alpha]3, supporting the previous observation that the [beta]-anomers are more effective than the [alpha]-anomers. Since the complexes generated with AP-containing oligonucleotides should be less pH dependent than those involving cytosine or 5-methylcytosine residues, we have investigated whether the ligand could also stabilise these complexes at a higher pH. The results are presented in Figure 5 B, showing the effect of 10 µM naphthylquinoline compound on the binding of 10 µM oligonucleotides to G6A6-T6C6 at pH 6.5. It can be seen that only [beta]6 generates a clear footprint. The binding of this oligonucleotide is not surprising since 10 µM [beta]6 can itself generate a footprint at pH 6.5 (Fig. 2 ). The ligand is unable to induce triplex formation with the remaining oligonucleotides, though some weak interaction with [alpha]3[beta]3 may be indicated by a reduction in cleavage intensity around the triplex target site.
The results of these footprinting experiments are summarised in Table 1 .
. Summary of the binding of AP-modified oligonucleotides to the G6A6-T6C6 target site under a range of experimental conditions
Successful binding of the oligonucleotide is indicated by a tick ([radic]), while a cross (*) represents no evidence of binding. W, indicates a possible weak interaction.
The MD simulations resulted in stable structures for all five [alpha]/[beta] combinations generated. (A full analysis of the structures in terms of their energetics and conformations will be presented elsewhere.) None of the structures showed any significant distortions from acceptable geometric features. All are B-type helices with an average helical twist of 30-32°, and an average rise of 3.4-3.5 Å. Stereo plots of the averaged structure of [beta]6T6 and [alpha]6T6 are presented in Figures 6 and 7 , respectively.
The results presented in this paper demonstrate that substitution of dC by [beta]-AP dramatically enhances the stability of parallel triplexes. Unmodified C6T6 shows no interaction with G6A6-T6C6 at pH 5.0, even at a concentration of 50 µM. However, replacement of the dC with [beta]-AP residues allows triplex formation at oligonucleotide concentrations as low as 0.25 µM, representing an increase in binding affinity of >200-fold. Moreover, the AP nucleoside extends the recognition of GC base pairs to higher pH, thereby relieving the pH-dependency of the parallel binding motif. In contrast to the unmodified oligonucleotide, the [beta]-AP-containing oligonucleotide shows significant interaction at pH 6.5 at which clear DNase I footprints are generated at oligonucleotide concentrations as low as 1 µM. Triplex formation is extended to neutral pH with 50 µM [beta]6. The position of the [beta]6 footprint, terminating exactly at the edge of the triplex target site, provides evidence that the [beta]-aminopyridines are actively contributing to the integrity of the complex and that the footprint is not merely caused by a strong interaction of the T-AT triplets.
The [beta]-AP nucleoside differs from a number of other nucleoside analogues which have been designed to recognise GC base pairs such as 8-oxoadenine and derivatives (19 ,20 ) and a pyrazolopyrimidinone derivative (12 ), in that it is an isostructural analogue of cytosine. This ensures minimal structural distortion of the third strand backbone, allowing the formation of a continuous YYR triplex.
One of the first modifications of cytosine involved its methylation at the C-5 position (13 ,14 ). Although this derivative has a slightly higher pKa, the enhancement in triplex stability probably results from favourable hydrophobic effects generated by the formation of a spine of methyl groups in the major groove (15 ). Our results are consistent with these reports, showing that the binding of C6T6 is increased when the cytosine residues are replaced with 5MeC. Substitution with [beta]-aminopyridine imparts even greater stability than 5MeC: the complex with 10 µM [beta]-AP is stable up to pH 6.5, while 10 µM 5MeC-oligonucleotide does not generate a DNase I footprint at any pH. These results demonstrate that the binding of cytosine to GC base pairs is best improved by enhancing the degree of N-3 protonation (by increasing the pKa), rather than attaching a hydrophobic methyl group to the C-5 position.
The susceptibility of natural [beta]-oligonucleotides to nuclease attack may be overcome by using synthetic [alpha]-anomers (34 ,35 ). Many studies have compared the binding of oligonucleotides consisting of entirely [alpha]- or [beta]-nucleotides. Due to the change in anomeric configuration, the orientation of an [alpha]-oligonucleotide is expected to be reversed in comparison to [beta]-oligomers. However oligonucleotides consisting entirely of [alpha]-T are reported to bind in the same parallel orientation as the [beta]-d(T)n oligonucleotide, presumably via reverse-Hoogsteen hydrogen bonds (32 ).
Bates et al. (21 ) were the first group to investigate the triplex binding of oligonucleotides containing a mixture of [alpha]- and [beta]-anomers. They reported that oligonucleotides containing a mixture of [beta]-thymidine and [alpha]-AP residues bind in a parallel orientation with respect to the duplex purine strand. Our results confirm this orientation since oligonucleotides containing various combinations of [alpha]- and [beta]-aminopyridines bind to G6A6-T6C6 but show no interaction with A6G6-C6T6. Molecular modelling studies (21 ) suggested that [alpha]-anomers of AP can be accommodated within an otherwise [beta]-anomeric triplex, with minimal structural perturbation. The results presented in this paper demonstrate that oligonucleotides containing [alpha]-AP form less stable parallel triplexes than [beta]-AP. Comparing the activity of [alpha]6 and [beta]6, we find that the [beta]-AP-oligonucleotide is consistently more stable than the corresponding [alpha]-AP-oligonucleotide.
The modelling results indicate that the [alpha]-AP residues are as readily accommodated within a parallel triplex as the isomorphous [beta]-nucleosides. The constitutional modification at C1' is compensated by changes of the sugar pucker phase angle and the torsion about the glycosidic bond. (We have to bear in mind that the [alpha]-anomer is not a mirror image of the [beta]-anomer, since the ribose ring has several chiral centres.) The preferred sugar conformation of the [alpha]-AP residues is C1'-endo and the torsion angle about the glycosidic bond is in the negative syn range. This compensates for the altered configuration at the C1' atom and allows one to integrate [alpha]-nucleotides into a parallel triple helix. Previous studies have reported that the binding of [alpha]-oligonucleotides to duplex DNA is fairly weak, requiring the additional attachment of a stabilising acridine group to the end of the oligonucleotide (32 ,36 ). We can infer from our modelling studies that this is primarily caused by steric restrictions. The major differences between the five simulated structures are seen in the behaviour of the hydrogen bonds between first strand guanine and third strand AP bases. The number of triplets which have a canonical Hoogsteen arrangement of hydrogen bonds approximately accords with the relative stabilities found by footprinting (Table 1 ). It should, however, be borne in mind that the 5-fold differences in concentrations correspond to only a small difference in binding free energy.
Oligonucleotides including a mixture of [alpha]-AP and [beta]-AP nucleosides [i.e. ([alpha][beta])3, [alpha]3[beta]3 and [beta]3[alpha]3], bind more strongly than [alpha]6 under some conditions. These differences are most pronounced under triplex-stabilising conditions. At 4°C, [alpha]3[beta]3 and [beta]3[alpha]3 show strong interaction at 10 µM, with a weaker interaction detected for [alpha]6 and ([alpha][beta])3. Furthermore, at pH 5.0 the naphthylquinoline triplex-binding ligand induces the binding of ([alpha][beta])3 and [alpha]3[beta]3 but not [alpha]6 and [beta]3[alpha]3. In general it seems that structures containing a run of only three [alpha]-AP residues are tolerated more readily than complexes which include a run of six [alpha]-nucleosides. This supports the proposal that introduction of several consecutive [alpha]-AP residues into an otherwise [beta]-anomeric triplex induces some destabilisation. The higher stability of [alpha]3[beta]3 than [beta]3[alpha]3 can be attributed to the presence of nine contiguous [beta]-residues in the former.
We have previously shown that the triplex-binding ligand enhances the stability of 10mer TC-containing oligonucleotides by >100-fold. However the compound was unable to shift triplex formation to physiological pH, presumably due to diminished N-3 protonation on the third strand cytosine residues. Although this ligand induces the binding of ([alpha][beta])3 and [alpha]3[beta]3 at pH 5.0 it does not potentiate the binding of ([alpha][beta])3 at pH 6.5, though some weak interaction is detected with [alpha]3[beta]3. In other words stabilisation of the six T-AT triplets is insufficient to facilitate formation of the adjoining stretch of ([alpha]-AP+)-GC triplets at pH 6.5.
34 Morvan, F., Rayner, B., Imbach, J.-L., Thenet, S., Bertrand, J.-R., Paoletti, J., Malvy, C. and Paoletti, C. (1987) Nucleic Acids Res., 15, 3421-3437.MEDLINE Abstract
35 Cazenave C., Chevrier, M., Thuong, N.T., and Hélène, C. (1987) Nucleic Acids Res., 15, 10507-10521.MEDLINE Abstract
36 Thuong, N.T., Asseline, U., Roig, U., Takasuga, M. and Hélène, C. (1987) Proc. Natl. Acad. Sci. USA., 84, 5129-5133.MEDLINE Abstract
*To whom correspondence should be addressed. Tel: +44 1703 594374; Fax: +44 1703 594459; Email:krf1@soton.ac.uk +Visiting scientist from The Institute of Molecular Biotechnology, Beutenbergstrasse II, D-07745 Jena, Germany
D. A. Rusling, V. E. C. Powers, R. T. Ranasinghe, Y. Wang, S. D. Osborne, T. Brown, and K. R. Fox Four base recognition by triplex-forming oligonucleotides at physiological pH
Nucleic Acids Res.,
May 23, 2005;
33(9):
3025 - 3032.
[Abstract][Full Text][PDF]
T. G. Uil, H. J. Haisma, and M. G. Rots Therapeutic modulation of endogenous gene function by agents with designed DNA-sequence specificities
Nucleic Acids Res.,
November 1, 2003;
31(21):
6064 - 6078.
[Abstract][Full Text][PDF]
Table 1
CiteULike 
Connotea 
Del.icio.us What's this?
