Skip Navigation

This Article
Right arrow Abstract Freely available
Right arrow Print PDF (556K) Freely available
Right arrow Supplementary Data
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrow Search for citing articles in:
ISI Web of Science (6)
Right arrowRequest Permissions
Right arrow Commercial Re-use Guidelines
for Open Access NAR Content
Google Scholar
Right arrow Articles by Debin, A.
Right arrow Articles by Svinarchuk, F.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Debin, A.
Right arrow Articles by Svinarchuk, F.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?

Stability of G,A triple helices
Nucleic Acids Research Pages 2699-2707

Stability of G,A triple helices
Introduction
Materials And Methods
   Oligonucleotide preparation
   Aptamer approach
   DMS footprinting
   Estimation of triplex stability
   Influence of magnesium and KCl concentration on triplex stability
   Modeling studies
Results
   Selection of the stable triplexes by the aptamer approach
   Verification of triplex stability
   Influence of `mismatches' in TFOs on triplex stability
   Influence of different substitutions in the TFOs on triplex stability
   Influence of Mg2+ and K+ concentration on triplex stability
   Modeling studies
Discussion
Acknowledgements
References

Stability of G,A triple helices

A. Debin1, C. Laboulais2, M. Ouali2, C. Malvy1, M. Le Bret2, F. Svinarchuk1, 3, *

1CNRS UMR 8532, Institut Gustave-Roussy, 94805 Villejuif, France, 2Equipe de Modélisation Moléculaire, CNRS UMR 8532, Institut Gustave-Roussy, 94805 Villejuif, France and 3Department of Biochemistry, Novosibirsk Institute of Bioorganic Chemistry, 8 Prospect Lavrenteva, Novosibirsk 630090, Russia

Received March 16, 1999; Revised and Accepted May 17, 1999

ABSTRACT

In this work we selected double-stranded DNA sequences capable of forming stable triplexes at 20 or 50°C with corresponding 13mer purine oligonucleotides. This selection was obtained by a double aptamer approach where both the starting sequences of the oligonucleotides and the target DNA duplex were random. The results of selection were confirmed by a cold exchange method and the influence of the position of a `mismatch' on the stability of the triplex was documented in several cases. The selected sequences obey two rules: (i) they have a high G content; (ii) for a given G content the stability of the resulting triplex is higher if the G residues lie in stretches. The computer simulation of the Mg2+, Na+ and Cl- environment around three triplexes by a density scaled Monte Carlo method provides an interpretation of the experimental observations. The Mg2+ cations are statistically close to the G N7 and relatively far from the A N7. The presence of an A repels the Mg2+ from adjacent G residues. Therefore, the triplexes are stabilized when the Mg2+ can form a continuous spine on G N7.

INTRODUCTION

Short homopurine-homopyrimidine regions in DNA have attracted a great deal of attention in connection with their possible role in gene regulation in eukaryotes (1,2). These regions are potential targets for manipulating gene expression, gene-targeted mutagenesis/knockout and inhibition of virus propagation through artificial triple helix formation (3-6). Recognition of DNA by triplex-forming oligonucleotides (TFO) occurs by hydrogen bonding between oligonucleotide bases and purine bases in the major groove of the target DNA duplex. Triple helix formation arises in either of two patterns, termed the pyrimidine motif and the purine motif (7-9). To promote triplex formation with cytosine-containing oligonucleotides, a slightly acidic environment is required. In contrast, the purine motif is pH-independent and has been used far more often for successful in vivo inhibition of transcription. However, the stability of G,A triplexes is highly unpredictable (10,11). The present work was therefore undertaken to identify sequences that form stable triplexes. This was achieved through a novel aptamer approach involving the simultaneous use of a random bank of target duplexes and a random bank of third strand oligonucleotides. This approach allowed us to select about 200 different sequences of 13mer oligonucleotides for their stability in a co-migration assay at 20 or 50°C. These sequences are characterized by a high guanine content. To verify the importance of guanine residues for the stability of the triplexes, some selected sequences were further analyzed by a cold exchange method at a different temperature. Finally, molecular modeling studies were undertaken to understand the role of guanine residues and divalent cations in the stability of the GA triplexes.

MATERIALS AND METHODS

Oligonucleotide preparation

Oligonucleotides were synthesized on an Applied Biosystems 391A DNA synthesizer using the solid phase phosphoramidite procedure. They were precipitated with 10 vol of a 3% solution of LiClO4 in acetone, then the pellets were washed with acetone, dried and dissolved in water. Concentrations were determined spectrophotometrically. Extinction coefficients were calculated from the extinction coefficients for the nucleotides and dinucleotide phosphates according to the equation given in Puglisi and Tinoco (12). For cold exchange experiments, oligonucleotides were purified by electrophoresis in a 20% denaturing polyacrylamide gel. After electrophoresis the oligonucleotides were eluted from the gel in 1 ml of 0.2 M LiClO4 solution for 12 h at 37°C, then precipitated with 10 vol of acetone.

Aptamer approach

The following oligonucleotides were used to select stable triplexes. Matrix 1, a random 13mer oligopurine was placed between two amplifying sequences, 5[prime]-CGAACGCCGGTACCA(R13)-CAAG-CTTAGGGGCGC-3[prime]. The flanking sequences contained restriction sites for KpnI and HindIII, respectively, to clone the selected triple helix forming duplex. An oligonucleotide 5[prime]-(R13)-C-biotin was used as a third strand for triplex formation. In another set of experiments we used a modified construct, matrix 2, 5[prime]-CGAACGCCGGTACCAtc(R13)caCAAGCTTA-GGGGCGC-3[prime]. Selection was performed as follows (Fig. 1). The duplex-containing random part was amplified with [[gamma]-32]ATP-labeled primers 5[prime]-CGAACGCCGGTACCA-3[prime] and 5[prime]-GCGCCC-CTAAGCTTG-3[prime] (specific activity 100 Ci/mmol) on either matrix 1 or 2 and purified on a 15% non-denaturing gel. After elution (overnight at 37°C in water), the purified duplex (~200 pmol) was incubated with 2000 pmol of the third strand oligonucleotide in a buffer containing 50 mM NaAc, 20 mM Tris-Ac, pH 7.5, and 10 mM MgAc2 for 24 h at 37°C. An aliquot of 1000 pmol of streptavidin was then added for an additional 30 min and the mixture loaded on a 10% polyacrylamide gel containing 50 mM NaAc, 20 mM Tris-Ac, pH 7.5, and 10 mM MgAc2. Migration was performed in a thermostated Hoeffer minigel apparatus at the temperature of selection (20 or 50°C for different experiments). After visualization of the bands by autoradiography, the band with low mobility was cut out and the DNA eluted in 50 µl of water at 37°C and re-amplified. The procedure was repeated three or four times. After the last round of selection the DNA was cut with KpnI and HindIII restriction enzymes and cloned in the corresponding sites of pBluescript II SK (Stratagene). Plasmids were purified on Qiagen columns, following the manufacturer's recommendations. Sequencing was performed with an Amersham ThermoSequenase cycle sequencing kit with M13-20 and T3 primers.


Figure 1. Design of the experiments to select stable GA triple helices.

DMS footprinting

Preparation of a DNA fragment. The plasmid used in this experiment was obtained by the aptamer approach and was found to contain the sequence 5[prime]-AGGAGGGGGAGGGGG-3[prime], which harbors the 13mer polypurine sequence that we have previously studied (13). To prepare a DNA fragment for modification by DMS, 30 µg of a plasmid containing a polypurine sequence from the c-pim proto-oncogene promoter (14) were cut with ClaI restriction enzyme, 3[prime]-labeled with the Klenow fragment of DNA polymerase I (Eurogentec) in the presence of 50 pmol of [32P]dCTP (3000 Ci/mmol) and digested with XhoI restriction enzyme. The large DNA fragment was then purified on Genclean beads (US Biochemical).

Probing with DMS. The prepared fragment (~0.5 pmol) was incubated with TFOs (100 pmol) in 20 µl of buffer containing 50 mM MOPS, pH 7.2, 50 mM NaAc and 10 mM MgAc2 at 25°C overnight. An aliquot of 2 µl of 5% DMS was then added and the reaction performed for 2 min at 25°C. The reaction was stopped by the addition of 5 µl of a solution containing 50% mercaptoethanol and 0.1 M NaAc. After double precipitation in ethanol, the samples were treated with 50 µl of 10% piperidine at 95°C for 20 min and the cleavage products were separated in a 6% polyacrylamide denaturing gel.

Estimation of triplex stability

Triplex formation. To form the triplex, 0.015 pmol of 5[prime]-32P-labeled purine TFO (specific activity 1000 Ci/mmol) was incubated overnight at 25°C with 1 µg (~0.5 pmol) of the target plasmid DNA in 10 µl of buffer containing 10 mM MgCl2, 50 mM NaCl and 20 mM Tris-Ac, pH 7.5.

Estimation of triplex stability by a cold exchange method. The triplex obtained in this manner was placed on ice and 250 pmol of corresponding unlabeled competitor oligonucleotide were added. The mixture was covered with mineral oil to avoid evaporation and incubated for 30 min at different temperatures, as specified in the figure legends. After incubation, the samples were immediately cooled on ice and loaded on a 1% agarose gel. Electrophoresis was performed in the cold room at 3.5 V/cm for 5 h in the presence of 10 mM MgCl2, 50 mM NaCl and 20 mM Tris-Ac, pH 7.5. During the migration, the buffer temperature was kept below 10°C. The gel was prepared for autoradiography by drying on the glass surface in a heated air flow (60°C). To check that the mobility of the retarded band corresponds to the TFO co-migrating with the target plasmid, the DNA was visualized by brief staining of the dried gel in a 10 µg/ml ethidium bromide solution. The quantity of radioactivity bound to the targeted DNA was determined using a STORM phosphorimager (Molecular Dynamics) with the flow-dried gel and plotted against the incubation temperature. Td is defined as the temperature at which half of the initially bound radioactivity dissociates from the plasmid (and, correspondingly, half of the triplex dissociates).

Influence of magnesium and KCl concentration on triplex stability

To form the triplex, 0.15 pmol of 5[prime]-32P-labeled purine TFO was incubated overnight at 25°C with 10 µg (~5 pmol) of the target plasmid DNA in 20 µl of buffer solution containing either 50 mM NaCl and 20 mM Tris-Ac, pH 7.5, (NaCl buffer) plus 1 mM MgCl2 or 150 mM KCl and 20 mM Tris-Ac, pH 7.5, (KCl buffer) plus 1 mM MgCl2. The stability of the triplex was estimated by a cold exchange method in the presence of varying amounts of magnesium at 37°C. After 30 min incubation, the samples were cooled on ice and 2 µl of 100 mM MgCl2 were added immediately to prevent further triplex dissociation. Electrophoresis and autoradiography were carried out as described above.

Modeling studies

The experimental system was modeled by a canonical ensemble consisting of ions and a single triplex set at the center of a sphere of 50 Å. The A6G7, G4AG5AG2 and (GA)6G triple helices were built up by stacking the GCoGbqt and the AToAbqt triplets from our database (15) using our programs MORCAD (16) and OCL (17). In these triplexes, the sugar of the third strand is the [beta] anomer, both purine strands are mutually anti-parallel and the glycosidic angle is in the anti conformation. Moreover, in these triplexes, the third strand divides the major groove of the duplex into a narrow groove between the two purine strands and a wider groove between the purines of the third strand and the pyrimidines of the duplex. The naked triplexes were minimized using our quasi-Newtonian minimizer MORMIN (16) and the AMBER 4.1 force field parameters (18,19). Details on the force field, algorithm, protocol and ionic concentration determination can be found in the Supplementary Material.

RESULTS

Selection of the stable triplexes by the aptamer approach

In order to find target sequences that form stable triplexes with purine TFOs, we developed a new aptamer approach using a random duplex matrix obtained by PCR amplification and a random third strand biotinylated oligonucleotide. In the presence of streptavidin the duplexes hybridizing with the TFO form complexes that were purified by gel retardation at a well-defined temperature. The selected duplexes were re-amplified and subjected to a new round of selection. After two or three rounds of selection the selected duplexes were cloned and sequenced. The experimental design is shown in Figure 1 and the details given in Materials and Methods. To verify that stable triplexes are selected by this approach, preliminary experiments were performed with the sequence 5[prime]-GGGGAGGGGGAGG-3[prime], which is known to form a very stable triplex with its corresponding TFO (13). For this purpose the matrix 5[prime]-CG-AACGCCGGTACCAGGGGAGGGGGAGGCAAGCTTA-GGGGCGC-3[prime] was subjected to one round of selection (Materials and Methods) in the presence of the third strand oligonucleotide 5[prime]-GGAGGGGGAGGGGc-biotin. The formation of a complex between streptavidin and the biotinylated TFO was used to achieve a better separation between the target duplex and the triplex in the co-migration assay. In the absence of streptavidin the mobilities of the duplexes and triplexes were too close to make the selection efficient (Fig. 2A). Oligonucleotide 5[prime]-GG-AGGGGGAGGGGc-biotin formed a low mobility complex with the duplex and streptavidin which could be easily separated from the duplex DNA by a co-migration assay (Fig. 2A). The low mobility band was not observed when the oligonucleotide was incubated either with the duplex DNA alone or with streptavidin alone (Fig. 2A). Because biotinylated TFOs gave the same pattern of guanine protection in DMS footprint experiments (data not shown), we suggest that the low mobility band is composed of a complex of the triple helix DNA with streptavidin. The other minor low mobility bands seen in Figure 2A are interpreted as complexes between streptavidin and more than one triplex. This experiment shows that stable triplexes can be selected by our technique.


Figure 2. Gel-retardation assay at 50°C. (A) With amplified matrix 5[prime]-CGAACGCCGGTACCAGGGGAGGGGGAGGCAAGCTTAGGGGCGC-3[prime] (duplex 1). Lane 1, duplex 1 plus unrelated oligonucleotide 5[prime]-GGAAGGAAAGGGGc-biotin; lane 2, the same plus streptavidin; lane 3, duplex 1 plus oligonucleotide 5[prime]-GGAGGGGGAGGGGc-3[prime]; lane 4, duplex plus TFO 5[prime]-GGAGGGGGAGGGGc-biotin plus streptavidin. (B) With amplified matrix 1 (duplex 2), second round of selection. Lane 1, duplex 2; lane 2, duplex 2 plus streptavidin; lane 3, duplex 2 plus oligonucleotide 5[prime]-(R13)-C-biotin plus streptavidin. S, T and D correspond to the start of migration, positions of the complex (duplex plus TFO-biotin plus streptavidin) and duplex, respectively.

Verification of matrix composition. To make sure that the composition of the initial matrix is random, we sequenced DNA from 40 clones obtained in the same way as for selected sequences but without selection for triplex formation. For this purpose, matrix 1 was amplified, purified on a non-denaturing gel, cut with restriction enzymes HindIII and KpnI, then cloned and sequenced as described in Materials and Methods. The results of this experiment show that matrix 1 is initially enriched in G residues (68.7%) although it was supposed to contain 50% guanines and 50% adenines (see Table S1 in Supplementary Material). We then constructed matrix 2 with an initial composition of 40% G and 60% A. Verification of this matrix actually gave a value of 60.8% for A residues (Table 1). We decided to make the selection with the two matrices independently. Another interesting finding concerns the variations in the length of the triple helix-forming region (Tables 1 and S1). These variations most likely arise during the amplification process and might reflect the difficulties of Taq polymerase in the synthesis of G-rich stretches.


Table 1. Results of selection with matrix 2
Column 1, sequences of the purine strand of the duplex before selection; columns 2 and 3, the same after three and four rounds of selection, respectively, at 20°C; column 4, selected sequences after three rounds at 20°C followed by one round at 50°C. At the end of each column percents of G residues are calculated for all sequences in the column.

Selection with matrix 1. Selection with matrix 1 was performed at the temperatures 20 and 50°C (Fig. 2B). For each temperature, three rounds of selection were performed and about 40 clones per temperature were sequenced (see Table S1 in Supplementary Material). The first conclusion to emerge from this experiment is that the G content of the selected sequences is greater than that of the matrix (Table S1). Moreover, this proportion increased with temperature (76.0% at 20°C and 82.0% at 50°C). Finally, in the selected sequences, no stretch of A residues longer than three was found.

Selection with matrix 2. In this experiment, we first performed three rounds of selection at 20°C, after which 40 clones were sequenced (Table 1). Here again, the proportion of G increased in comparison to non-selected sequences. However, the increase in G content after three rounds of selection was lower in matrix 1 (increase from 68.7 to 80%) than in matrix 2 (increase from 39.2. to 90%). To evaluate the impact of the number of selection cycles on the selected sequences, sequencing was performed after an additional fourth round of selection at 20 or 50°C. This round increased the proportion of G residues to 93.4% at 20°C and to 94.6% at 50°C. This clearly shows that the G content plays a major role in the stability of the G,A triplexes. Surprisingly, one of the sequences selected at 50°C (5[prime]-AGGAAGAGGAAGGA) contains only 50% G residues. The stability at 50°C of this sequence was not further confirmed by the cold exchange method (Table 2). This can be explained either by contamination of the duplex during PCR amplification or by interaction of these oligonucleotides with streptavidin. The exact position of the few adenines remaining in the selected sequences appears to be unimportant for stability so long as they do not lie in a row. Further support for this conclusion was obtained by independent estimation using a cold exchange method for several selected duplexes of different G content and various TFOs (see below).


Table 2. Determination of triplex stability (Td) of some selected sequences

Verification of triplex stability

Triplex stability was estimated by a modified cold exchange method (see Fig. 3 and Materials and Methods for details) using duplexes containing 13, 11, 9 and 7 guanines in one strand (Table 2). Corresponding TFOs were chosen according to canonical rules (9). The data obtained clearly show that stability decreases when the number of guanines decreases (Table 2). Moreover, stability depends not only on the content but also on the sequence. Td varied from 28°C in the case of seven G residues in a row to <10°C for the sequences with dispersed G residues (Table 2), in which case triplex formation is undetectable by our approach (Materials and Methods). However, if the sequences are longer the stability for sequences with dispersed G residues can be quantified. For the longer sequences of identical G content: (G)10(A)10 and (GA)10 the respective Td values at 50 and 22°C drastically differ.


Figure 3. Determination of triplex stability for the TFO 5[prime]-GAGGGAGAGGAGG-3[prime]. Target DNA was incubated with the TFO at different temperatures under the conditions described in Materials and Methods and subjected to a co-migration assay. (A) Gel after staining with ethidium bromide. (B) Autoradiograph of the same gel. S, D, D+TFO and TFO correspond to the positions of the start of the migration, target plasmid and target plasmid plus TFO and TFO, respectively. Lanes 1 and 3-8, incubation with the cold competitor TFO at 100 (negative control), 35, 40, 45, 50, 55 and 60°C, respectively; lane 2, without competitor (positive control). (C) Graph built on the basis of the data obtained in the co-migration assay. Td is defined as the temperature at which half of the initially bound radioactivity (A, line 2) dissociates from the plasmid. For this TFO it is equal to 45°C.

Influence of `mismatches' in TFOs on triplex stability

Since the aptamer approach involves selection of the duplex part and we were unable to see selected TFOs it was interesting to further investigate the rules for triplex formation in G-rich sequences. To do so, we studied the stability of triplexes containing one `mismatch' in the TFO. For this purpose we chose the target sequence 5[prime]-GGGGAGGGGGAGG-3[prime] and a set of TFOs containing one C instead of the canonical base (Table 3). Each `mutated[prime] TFO gave a less stable triplex than the canonical TFO. Some additional conclusions can be drawn here. First, guanines once again play the major role in triplex stability. For example, substituting A5 by C decreases Td to 52°C, but substituting any G flanking this A5 decreases Td to 40 (for G4) or to 21°C (for G6) (Table 3). For A11 the corresponding values are 55 (C11), 37 (C10) and 43°C (C12). Secondly, the bases in the middle of the sequence have more impact on triplex stability than the flanking bases (Table 3). To estimate how the mismatched TFOs interact with the target sequence, DMS footprint experiments were performed (Fig. 4). Here it was seen that substituting any guanine by cytosine in the TFOs decreased the guanine protection not only of the target base, but also of the neighboring guanines. This effect was more pronounced for the five central guanines, with the one exception of G7 (Fig. 4). Substitution of A by C had no visible effect on the footprint, suggesting that guanines play the most important role in triplex stability.


Figure 4. Autoradiogram of a 6% polyacrylamide sequencing gel showing the results of DMS footprint experiments with TFOs containing `mismatches' in different positions. Lane T-, without TFO; lane C0, TFO 5[prime]-GGGGAGGGGGAGG-3 (without `mismatch'); lanes C1-C13, TFOs containing corresponding substitution of G or A bases for C.


Table 3. Influence of `mismatches' in the TFO on triplex stability
Triplex stability (the value Td) was determined for the target sequence 5[prime]-GGGGAGGGGGAGG-3[prime] and a set of TFOs containing one C (marked in italic bold) instead of the canonical base. C0, TFO without `mismatch'.

Influence of different substitutions in the TFOs on triplex stability

To check that destabilization of the triplex is not specific for G->C or A->C substitutions in the TFO, we next performed experiments with all four base substitutions in the TFO (Table 4). In accordance with published data, A can be replaced by T without visible effects on triplex stability. Substituting G or C for A destabilized the triplex to the same extent, but even when both A residues were substituted the resulting triplex could be visualized in the co-migration assay. In contrast, replacing G by any base destabilized the triplex to an almost equivalent degree (Table 4).


Table 4. Influence of different substitutions in the TFO on triplex stability Triplex stability (the value Td) was determined for the target sequence 5[prime]-GGGGAGGGGGAGG-3[prime] and a set of TFOs containing all four base substitutions (marked in italic bold) in the TFO instead of the canonical base. C0, TFO without `mismatch'.

Influence of Mg2+ and K+ concentration on triplex stability

To determine the minimal Mg2+ concentration at which the triplexes are still stable, we assayed triplex stability with the TFO 5[prime]-GGGGAGGGGGAGG-3[prime] in NaCl and KCl buffers at different Mg2+ concentrations (Materials and Methods). In NaCl buffer there was no dissociation at Mg2+ concentrations as low as 0.5 mM. The monovalent ion concentration markedly influenced triplex stability since no retarded bands could be visualized at concentrations below 1 mM Mg2+ in the presence of 150 mM K+.

Modeling studies

Since TFO composition and monovalent and divalent ion concentrations play a crucial role in triplex stability, we performed modeling studies to seek a correlation between ion distribution, TFO sequence and triplex stability. For this purpose, three triplex structures were chosen: one containing G and A bases and having the highest stability in our studies (G4AG5AG2) and two others with the same GA content but different stability [A6G7 and (GA)6G] (Table 3). The aim was to show that triplex ionic environments may play an important role in their stabilization. As we wanted to simulate a system made of a triplex immersed in a solution of given ion concentrations that would be as close as possible to the experimental conditions, we used a Monte Carlo method. In such a method, the ionic environment is built up by a succession of independent moves (20). An ion is randomly chosen and then randomly moved to a trial position inside a cube of size 2[delta], where [delta] is the maximum step size. The variation of the energy in this trial move is computed. The trial move is accepted or refused according to the Metropolis rule (21). The acceptance rate is the ratio of accepted over total trial moves and should be as close as possible to 0.5 for a good survey of the conformational space. In this work, preliminary calculations showed that the cations stuck to the triplex, so that a density scaled method had to be implemented (22; see also Supplementary Material). The distribution of ions should not and does not depend on the law chosen for [delta] if a very large number of trial moves are generated. However, the choice of the law has a large effect on the reproducibility of the results for a given computer time. Table 5 compares some values obtained after simulation of the distribution of counterions around three triple helices. In all systems, the number of ions shown in columns 2 and 3 was chosen so that the outer bulk cation concentrations fit the experimental values to within 10%. Each system was heated to get four independent starting ion distributions. The system was then cooled down as described in Supplementary Material. The averages and fluctuations of the average energies of each production run are shown in column 4.


Table 5. Monte Carlo results Column 1, triplexes; columns 2 and 3, number of Mg2+ and Na+ ions chosen so that the outer bulk cation concentrations fit the experimental values within 10%; column 4, averages and fluctuations of the average energies of each run. The table shows that with a distance-independent dielectric constant the energy of the ionic environment and the number of Mg2+ and Na+ that must be included in the simulation volume to reproduce the experimental bulk concentrations do not depend on the triplex. In contrast, the results depend not only on the G content but also on the sequence when a sigmoidal dielectric constant is used. The more Mg2+ are stuck to the oligonucleotides, the more Mg2+ should be introduced into the simulation volume to give the same concentration far from the triplex.

In the first three lines the aqueous medium is simulated by a distance-independent dielectric constant equal to 78. As shown in Table 5, the average energies and the number of cations are not statistically different. Therefore, the implicit representation of the solvent with a distance-independent dielectric constant does not reproduce the difference in stability that is experimentally observed. On the other hand, the results obtained with the sigmoidal dielectric constant (23) strongly depend on the sequence of the triple helices. The total number of Mg2+ that must be included in the simulation sphere to reproduce the same bulk concentrations increases with the number of guanines. In contrast, the number of Na+ decreases. This means that Mg2+ cations bind to guanines more strongly than does Na+. As the Mg2+ ions bind, Na+ ions are released to the bulk solution. The energy values show the same trend. The triplex with the higher guanine content has a lower energy. From the comparison of the two triplexes with the same guanine content, it may be concluded that Mg2+ ions have a higher binding affinity to guanine stretches.

It was then interesting to determine where the cations are, which can easily be done by analyzing the radial distribution of the Mg2+ ions from the third strand guanine N7 atoms. The molar local concentrations as defined in Supplementary Material for all the third strand guanine N7 are pooled into one curve CMg-GN7 (r) in Figure 5A. When the triplex has stretches of G residues, the first maximum of CMg-GN7 (r) occurs at r = 2.25 Å. The radial distribution is very different when the G residues are adjacent to A residues: the first peak is located at 2.55 Å and is much lower than for G stretches (Fig. 5A). As far as the adenine N7 atoms are concerned, Figure 5B shows that the first peak of CMg-AN7 (r) is wider, lower and occurs for larger values of r than in the case of guanine. The Mg2+ are less concentrated and lie farther from the adenine N7 atoms. Each ion was followed in each run. In only one case, and for a slightly different bulk concentration, we found a significant concentration of Mg2+ around one of the two adenine N7 of the G4AG5AG2 triplex. This suggests that some Mg2+ ions may bind to the adenine N7 when it is surrounded by G residues. In all our simulations, the Na+ ions are far from the N7 atoms (result not shown). This is expected if the Mg2+ ions have already occupied the best positions.


Figure 5. Mg2+ local molar concentration as a function of the distance r (in Å) from the third strand (A) guanine or (B) adenine N7 atoms for three triplexes: A6G7 (solid line), G4AG5AG2 (dashed line) and (GA)6G (circles).

We also analyzed the radial distribution of Mg2+ using a distance-independent dielectric constant. In this case, the curves have their first maximum at r = 2.65 Å regardless of sequence (result not shown).

DISCUSSION

One of the most intriguing features of the purine TFOs is the existence of very stable triplexes, where the triplex can be more stable than the corresponding duplex (13,24). In spite of many studies the reasons for this are still unclear. Formation of stable triple helices is crucial for the efficiency of in vivo targeted mutagenesis (25) and, moreover, short TFOs (<25 bases) are more efficient in a mutagenic assay (26). We therefore undertook this study to find short sequences (13mer) forming stable triplexes and to understand the nature of this stability. To achieve this goal, we used an aptamer approach, which has already been successfully applied in many fields. In order to find a maximum number of sequences capable of forming stable triplexes we used random sequences as starting target duplex and as starting TFO. The random part of the initial oligonucleotides contained only guanines and adenines, since it is already known that any mismatch (substitution of purine for pyrimidine) in the duplex destabilizes the resulting triplex (9). This allowed us to decrease the initial number of variants from 413 = 67 108 864 in the case of all four bases to only 213 = 8192. For the third strand oligonucleotide we also used TFOs containing purines only, since cytosine destabilizes the R(RY) triplex and adenines and thymines have essentially the same impact on triplex stability (9). To achieve a better separation between the pools of the duplexes and triple helices, the TFOs were biotinylated. The biotin moiety was coupled to the TFOs through a C residue to decrease any interference between triplex formation and its recognition by streptavidin. Streptavidin is necessary to separate the duplex from the complex streptavidin-biotinylated TFO duplex in a gel. Because of this good separation in the presence of streptavidin, only two or three rounds were used to select stable triplexes. The sequences selected were found to obey two rules: (i) they have a high G content; (ii) the G residues lie in stretches. The necessity of a high G content (>54%) to obtain high affinity binding was recently suggested for longer TFOs (27). This second rule is in contrast to Y(YR) triplexes, in which the most stable sequences are CT repeats, since adjacent C bases partially cancel the stability of one another (28). Our conclusions were further supported by determination of triplex stability for some selected sequences using a modified cold exchange method. This approach was chosen because we were unable to detect transition, corresponding to dissociation of the triplexes, by UV spectroscopic temperature-dependent melting studies as described by others (29,30). This method permits the study of rate of dissociation versus temperature. However, no information could be obtained regarding triplex association and competitive parallel duplex (30,31) or quadruplex (32,33) structure formation.

We first determined the stability of a number of sequences with different G contents. In agreement with the conclusion drawn from the aptamer approach, stability increased with increasing guanine content. Secondly, the stability of a series of TFOs bearing the same number of differently located guanines (seven) was determined. The most stable of these sequences had all seven guanines in a row followed by the TFO, with three GG pairs and one G separated by two or one A. Thirdly, we determined the stability of the triplex with different `mismatches' in the third strand. Both methods, cold exchange and DMS footprinting, show that G is the most important base in stabilizing the triplex. In accordance with previously published data (reviewed in 34), mismatched triplets in the center cause a greater decrease in stability than at the ends of the triplex.

Recently, another variant of the aptamer approach allowed identification of the consensus duplex DNA sequence recognized by a G/T TFO under conditions favoring purine motif triplex formation (35). Analysis of 47 sequences indicated that recognition between 13 bases on the TFO 3[prime]-end and the duplex DNA was sufficient for triplex formation. The main difference between that study and ours is that we used `random' TFOs (composed of G and A residues); the previous study used a TFO with a fixed sequence. Since the authors used a restriction nuclease protection assay to select for stable triplexes, the selection temperature could not be increased above 37°C and, consequently, the stability of the selected triplexes in these experiments was underestimated.

In order to better understand the cause of triplex stability we simulated the ionic environment around three of them. One of the most difficult tasks in computing is simulating a system containing chemical species at very different concentrations. Here the typical triplex concentration (0.1 µM) is very small as compared to the ion concentrations (10 mM MgCl2 and 50 mM NaCl). The low triplex concentration implies that the smallest representation of the system would consist of one triplex in a cubic box of side length 2520 Å. This cubic box would contain approximately 105 Mg2+ ions, 5 × 105 Na+ ions, 7 × 105 Cl- ions and >5 × 108 water molecules. At present, such systems cannot be treated in a reasonable amount of time and we therefore had to reduce the number of atoms that are explicitly represented.

As the Debye length of a 100 mM NaCl aqueous solution is ~10 Å, we restricted the simulation domain to a sphere of radius 50 Å, centered at the triplex center of mass, and took it to be electrically neutral. As such a domain contains approximately 10 times the number of water molecules that can currently be taken into account simultaneously (36,37), we had to take an implicit representation of the aqueous medium and chose a force field that favors specific interactions. A drawback of such a choice is that the ions may be trapped and exchange slowly with the bulk, preventing thermodynamic equilibrium from being reached in a reasonable computation time or causing apparent plateaux. In the traditional Metropolis method (21), as implemented by Le Bret and Zimm (20), a cube of fixed size is used to constrain the maximum displacement that can occur on a trial move. The size of the maximum displacement is chosen so that the acceptance rate is close to 0.5. However, in this work, such a simple algorithm could not be used because the Mg2+ concentration can be very high locally, as much as 100 M near the polyelectrolyte, while decreasing to the experimental bulk value of 10 mM at the fringe of the simulation sphere. Therefore, a sampling cube of fixed size that efficiently samples configuration space next to the polyelectrolyte will do a poor job far away from it, and vice versa. We modified the sampling law until the acceptance probability for each ion was between 0.3 and 0.7. This problem is frequent and many efforts have been made to accelerate the convergence, for instance by implementing a multi-particle Monte Carlo (38). Here, because of the difficulties encountered with the Mg2+ ions, a batch of four simulations was performed. In each simulation the system was slowly cooled using a rather large fixed maximum displacement. In such runs, the ions that are trapped may experience some rare but very efficient trial moves over an energy barrier and find a better location. A density scaled algorithm is then used to analyze the distribution. Instead of using grand canonical ensembles (39), the numbers of ions in canonical ensembles were varied until the computed bulk concentrations fit the experimental conditions.

We now turn to discussion of the force field. As we were essentially interested in the distribution of ions, the triplex conformation was left rigid and the only contributions to the energy were the Coulombic and the 6-12 Van der Waals potentials (40). When the ions interacted among themselves, we considered their first hydration shell as a supermolecule, as shown for Mg2+ by Bernal-Uruchurtu and Ortega-Blake (41). However, this is certainly wrong when the ions get close to the DNA. For instance, we estimated the distance between Mg2+ and the G N7 as 2.65 Å in the Van Meervelt et al. crystal. We could have chosen parameters to reproduce that distance. However, in the Van Meervelt et al. (42) crystal, the guanine strands are parallel and the Mg2+ lies between the Watson-Crick pyrimidine strand and the TFO and not between the two guanine strands, as in the triplexes studied here. This is why we chose to use the parameters tabulated by Aqvist (19). Finally, in the light of the results presented above, an implicit representation of the solvent through a distance-independent dielectric constant does not yield any sequence specificity.

To summarize, the main features of the force field that yield sequence specificity are a sigmoidal dielectric constant and ions that are naked with respect to DNA but otherwise hydrated. We now compare our results with the data in the literature. As described above, we found that the Mg2+ cations accumulate close to G N7 when the G residues belong to stretches and, in that case, form a kind of spine. This result is essentially similar to the Na+ spine described by Weerasinghe et al. (43) at much higher salt and triplex concentrations. This analogy would not be surprising if the counterions that are in the immediate vicinity of the polyelectrolyte were thought to be bound. Using an explicit representation of the water molecules, Weerasinghe et al. (43) found 2.4 Å for the distance from the N7 where the Na+ radial distribution has its first maximum. This value is incompatible with a fully hydrated ion. Since Mg2+ has a smaller radius than Na+, according to Aqvist (19), the Mg2+ local concentration should have its first maximum at a shorter value of r. The value r(N7-Mg2+) = 2.25 Å is reasonable. When a G is adjacent to A, the maximum of the corresponding curve CMg-GN7 (r) occurs at a larger value. This is interpreted as a repulsion of the Mg2+ ions from the G N7 by the amino group of the adenines. Of course, in a stretch of A residues, the Mg2+ spine cannot form because of the amino group at the adenine carbon C6.

Comparing the results obtained with the various sequences, the formation of the Mg2+ spine seems to be correlated with a computed lower energy of the system and with the experimental stability. The more Mg2+ bound, the more stable the triplex. Our results show that Mg2+ binds to G N7 atoms. The negative potential of the N7 lone pair is even deeper because of the nearby presence of the O6 atom on the same residue. On an adenine, the positive potential created by the amino group acts in the other direction, preventing aggregation of Mg2+ and destabilizing the triplex. In a stretch of G residues, the negative potential is decreased by the proximity of the N7 and O6 atoms of the adjacent G residues and this favors the binding of Mg2+. Therefore, the stability does not depend only on the global G content of the TFO but also on the sequence. The formation of the stabilizing Mg2+ spine may be interrupted by other mechanisms. For instance, in the simulations presented here, the Na+ ions were expelled far from the triplex. We conjecture that if we decreased the Mg2+ bulk concentration, the Na+ ions could chase them, leading to destabilization. The results obtained here can probably be generalized to other alkaline earth cations playing an important role in triplex association (10,44,45). In addition to the electrostatic and Van der Waals interactions as used here, the transition metal cations may generate polarization or quantum effects that could play a critical role in triplex stability (46), but this requires further investigation.

See supplementary material in NAR Online (107 KB PDF file).

ACKNOWLEDGEMENTS

We thank L. Pritchard for critical reading of the manuscript. M. Le Bret is grateful to the ARC for funding by grant 2080. This work was supported by a SIDACTION research fellowship to F.S. and MENRT fellowships to A.D. and C.L.

REFERENCES

1. Beasty, M. and Behe,M.J. (1988) Nucleic Acids Res., 16, 1517-1528.MEDLINE Abstract

2. O'Neill, D., Bornschlegel,K., Flamm,M., Castle,M. and Bank,A. (1991) Proc. Natl Acad. Sci. USA, 88, 8953-8957. MEDLINE Abstract

3. Maher, L.J. (1996) Cancer Invest., 14, 66-82. MEDLINE Abstract

4. Majumdar, A., Khorlin,A., Dyatkina,N., Lin,F.L., Powell,J., Liu,J., Fei,Z., Khripine,Y., Watanabe,K.A., George,J., Glazer,P.M. and Seidman,M.M. (1998) Nature Genet., 20, 212-214. MEDLINE Abstract

5. Vasquez, K.M. and Wilson,J.H. (1998) Trends Biochem. Sci., 23, 4-9. MEDLINE Abstract

6. Wang, G., Seidman,M.M. and Glazer,P.M. (1996) Science, 271, 802-805. MEDLINE Abstract

7. Moser, H.E. and Dervan,P.B. (1987) Science, 238, 645-650. MEDLINE Abstract

8. Le Doan, T., Perrouault,L., Praseuth,D., Habhoub,N., Decout,J.L., Thuong,N.T., Lhomme,J. and Helene,C. (1987) Nucleic Acids Res., 15, 7749-7760. MEDLINE Abstract

9. Beal, P.A. and Dervan,P.B. (1991) Science, 251, 1360-1363. MEDLINE Abstract

10. Malkov, V.A., Voloshin,O.N., Soyfer,V.N. and Frank-Kamenetskii,M.D. (1993) Nucleic Acids Res., 21, 585-591. MEDLINE Abstract

11. Cooney, M., Czernuszewicz,G., Postel,E.H., Flint,S.J. and Hogan,M.E. (1988) Science, 241, 456-459.

12. Puglisi, J.D. and Tinoco,I.,Jr (1989) Methods Enzymol., 180, 304-335. MEDLINE Abstract

13. Svinarchuk, F., Paoletti,J. and Malvy,C. (1995) J. Biol. Chem., 270, 14068-14071. MEDLINE Abstract

14. Svinatchuk, F., Bertrand,J.-R. and Malvy,C. (1994) Nucleic Acids Res., 22, 3742-3747. MEDLINE Abstract

15. Piriou, J.M., Ketterlé,Ch., Gabarro-Arpa,J., Cognet,J.A.H. and Le Bret,M. (1993) Biophys. Chem., 50, 323-343.

16. Le Bret, M., Gabarro-Arpa,J., Gilbert,J.C. and Lemaréchal,C. (1991) J. Chim. Phys., 88, 2489-2496.

17. Gabarro-Arpa, J., Cognet,J.A.H. and Le Bret,M. (1992) J. Mol. Graphics, 10, 166-173.

18. Cornell, W.D., Cieplak,P., Bayly,C.I., Goulg,I.R., Merz,K.M.,Jr, Ferguson,D.M., Spellmeyer,D.C., Fox,T., Caldwell,J.W. and Kollman,P.A. (1995) J. Am. Chem. Soc., 117, 5179-5197.

19. Aqvist, J. (1990) J. Phys. Chem., 94, 8021-8024.

20. Le Bret, M. and Zimm,B.H. (1984) Biopolymers, 23, 271-285. MEDLINE Abstract

21. Metropolis, N., Rosenbluth,A.W., Rosenbluth,M.N., Teller,A.H. and Teller,E. (1953) J. Chem. Phys., 21, 1087-1092.

22. Gordon, H. and Goldman,S. (1989) Mol. Simulat., 3, 213-225.

23. Lavery, R., Sklenar,H., Zakrzwska,K. and Pullman,B. (1986) J. Biomol. Struct. Dyn., 3, 989-1014. MEDLINE Abstract

24. Pilch, D.S., Levenson,C. and Shafer,R.H. (1991) Biochemistry, 30, 6081-6087. MEDLINE Abstract

25. Wang, G., Levy,D.D., Seidman,M.M. and Glazer,P.M. (1995) Mol. Cell. Biol., 15, 1759-1768. MEDLINE Abstract

26. Wang, G. and Glazer,P.M. (1995) J. Biol. Chem., 270, 22595-22601. MEDLINE Abstract

27. Perkins, B.D., Wilson,J.H., Wensel,T.G. and Vasquez,K.M. (1998) Biochemistry, 37, 11315-11322. MEDLINE Abstract

28. Roberts, R.W. and Crothers,D.M. (1996) Proc. Natl Acad. Sci. USA, 93, 4320-4325 MEDLINE Abstract

29. Lacoste, J., Francois,J.C. and Helene,C. (1997) Nucleic Acids Res., 25, 1991-1998 MEDLINE Abstract

30. Noonberg, S.B., Francois,J.C., Praseuth,D., Guieysse-Peugeot,A.L., Lacoste,J., Garestier,T. and Helene,C. (1995) Nucleic Acids Res., 23, 4042-4049. MEDLINE Abstract

31. Cheng, A.J., Wang,J.C. and Van Dyke,M.W. (1998) Antisense Nucleic Acid Drug. Dev., 8, 215-225. MEDLINE Abstract

32. Cheng, A.J. and Van Dyke,M.W. (1997) Gene, 197, 253-260 MEDLINE Abstract

33. Olivas, W.M. and Maher,L.J. (1995) Biochemistry, 34, 278-284. MEDLINE Abstract

34. Gowers,D. M. and Fox,K.R. (1999) Nucleic Acids Res., 27, 1569-1577. MEDLINE Abstract

35. Hardenbol, P. and Van Dyke,M. (1996) Proc. Natl Acad. Sci. USA, 93, 2811-2816. MEDLINE Abstract

36. Ouali, M., Bouziane,M., Ketterlé,C., Gabarro-Arpa,J., Auclair,C. and Le Bret,M. (1996) J. Biomol. Struct. Dyn., 13, 835- 853.

37. Ketterlé,C., Gabarro-Arpa,J., Ouali,M., Bouziane,M., Auclair,C., Helissey,Ph., Giorgi-Renault,S. and Le Bret,M. (1996) J. Biomol. Struct. Dyn., 13, 963-977. MEDLINE Abstract

38. Jayaram, B., Swaminathan,S., Beveridge,D.L., Sharp,K. and Honig,B. (1990) Macromolecules, 23, 3156-3165.

39. Jayaram, B. and Beveridge,D.L. (1991) J. Phys. Chem., 95, 2506-2516.

40. Weiner, S.J., Kollman,P.A., Case,D.A., Singh,U.C., Ghio,C., Alagona,G., Profeta,S.,Jr and Weiner,P. (1984) J. Am. Chem. Soc., 106, 765-784.

41. Beral-Uruchurtu, M.I. and Ortega-Blake,I. (1995) J. Chem. Phys., 103, 1588-1598.

42. Van Meervelt, L., Vlieghe,D., Dautant,A., Gallois,B., Précigoux,G. and Kennard,O. (1995) Nature, 374, 742-744. MEDLINE Abstract

43. Weerasinghe, S., Smith,P.E., Mohan,V., Cheng,Y.K. and Montgomery Pettitt,B. (1995) J. Am. Chem. Soc., 117, 2147-2158.

44. Blume, S.W., Lebowitz,J., Zacharias,W., Guarcello,V., Mayfield,C.A., Ebbinghaus,S.W., Bates,P., Jones,D.E., Trent,J., Vigneswaran,N. and Miller,D.M. (1999) Nucleic Acids Res., 27, 695-702. MEDLINE Abstract

45. Floris, R., Scaggiante,B., Manzini,G., Quadrifolio,F. and Xodo,L.E. (1999) Eur. J. Biochem., 260, 801-809. MEDLINE Abstract

46. Potaman, V.N. and Soyfer,V.N. (1994) J. Biomol. Struct. Dyn., 11, 1035-1040. MEDLINE Abstract

*To whom correspondence should be addressed at: Laboratoire Hematopoièse et Cellules Souches, INSERM U 362, PR 1, Institut Gustave Roussy, rue Camille Desmoulins, 94805 Villejuif Cedex, France. Tel: +33 1 42 11 5468; Fax: +33 1 42 11 52 40; Email: fedorsvi{at}igr.fr
Correspondence concerning modeling studies may be directly addressed to M.L.B. at mlebret{at}lbpa.ens-cachan.fr Present address: A. Debin, Biovector Therapeutics, chemin du Chêne Vert, 31650 Labège, France

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.
Add to CiteULike CiteULike   Add to Connotea Connotea   Add to Del.icio.us Del.icio.us    What's this?


This article has been cited by other articles:


Home page
 Nucleic Acids ResHome page
E. M. McGuffie and C. V. Catapano
Design of a novel triple helix-forming oligodeoxyribonucleotide directed to the major promoter of the c-myc gene
Nucleic Acids Res., June 15, 2002; 30(12): 2701 - 2709.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Print PDF (556K) Freely available
Right arrow Supplementary Data
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrow Search for citing articles in:
ISI Web of Science (6)
Right arrowRequest Permissions
Right arrow Commercial Re-use Guidelines
for Open Access NAR Content
Google Scholar
Right arrow Articles by Debin, A.
Right arrow Articles by Svinarchuk, F.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Debin, A.
Right arrow Articles by Svinarchuk, F.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?