Triplex hydration: nanosecond molecular dynamics simulation of the solvated triplex formed by mixed sequences

Abstract

A theoretical model for the hydration pattern and motion of ions around the triple helical DNA with mixed sequences d(GACTGGTGAC)d(GTCACCAGTC)*d(GACTGGTGAC) in solution, during MD simulation, using the particle mesh Ewald sum method, is elaborated here. The AMBER 5.0 force field has been used during the simulation in solvent. The simulation studies support a dynamically stable atmosphere around the DNA triplex in solution over the entire length of the trajectory. The results have been compared with Hoogsteen triplexes and examined in the context of the observed behaviour of hydration in crystallographic data of duplexes. The dynamical organization of counterions and water molecules around the triplex formed by mixed sequences is described here. It has been observed that cations prefer to bind between two adjoining purines of the second and the third strands. The idea of localized complexes (mobile counterions in unspecific electronegative pockets around the DNA triplex with water molecules) may have important implications for understanding the specificity of the interactions of nucleic acids with proteins and other ligands.

INTRODUCTION

Water participates in all biochemical processes involving folding, recognition or catalysis of nucleic acids. An extensive survey of crystallographic data has led to the view that water is an integral part of nucleic acid structures. In helical duplexes, owing to the periodicity of the contacts between water and each repeating unit, specific hydration patterns are frequently observed. The solvent interactions are the key factor in the conformational variability of nucleic acids (1–5). Although diffraction studies have produced very attractive molecular pictures of DNA hydration, they are limited in two important aspects. First, only the partially ordered water molecules cause discernible diffraction intensities, the (less ordered) majority of the waters in the first solvation layer around DNA remains unrefined and as such are thus invisible. Secondly, localized ions around DNA cannot be easily distinguished from the water oxygen atom (5). Computer simulation techniques, which are free from these two limitations, have been used to complement the experimental findings (6–10).

In recent years, triplex DNA formation by the binding of the oligonucleotides to duplex DNA has attracted much attention as an approach to develop sequence-specific DNA cleavage agents and to suppress the transcription of disease related genes (11–19). Molecular dynamics (MD) simulation techniques have been used to study the conformation of triplexes (20–24) and to study the effect of hydration, but these techniques have thus far been limited mostly to triple helical DNA containing homopurine–homopyrimidine sequences. The semi-empirical quantum mechanical and molecular dynamics methods show that the hydrogen-bonded-DNA triplexes with mixed sequences may also form a stable conformation (25–32). The hydrogen-bonding pattern similar to that observed in mixed sequences (25–30) has been reported in a crystal–structure analysis (33). In order to investigate the role of water in the stabilization of specific structural motifs and in the function of triplexes formed by mixed sequences, we have analysed a fully hydrated (3671 water molecules) and neutralized (27 Na⁺ counterions) 10mer triplex structure using 1.6 ns MD simulation.

The crystal structures reveal only ‘ordered’ water structure (defined as that which can be assigned to residual electron density in crystal structure determination), but in spectroscopic measurements the number of water molecules in the first shell of DNA (defined as waters with oxygen atoms located within 3.0 Å from the DNA surface) are found to be 11–12 waters per nucleotide (34,35). The full complement of water molecules hydrating the DNA triplex has not as yet been experimentally observed. It was found that the stability of the (dT)₁₅(dA)₁₅(dT)₁₅ triplex increased by a factor of 10 on addition of 40 volume percent of either ethylene, methanol, ethanol, dioxane, or dimethylformamide (36). However, the stability of a triplex with a mixed thymine and cytosine third strand in the absence of additives is higher than in the case of a homothymine third strand, and it even slightly decreases with added ethylene glycol (36). The role of dehydration was illustrated by the studies of Compos and Subirana (37) who observed the triplex formation of poly(G)poly(C) in the presence of N-α-acetyl-l-argenine ethylamide. In other studies (38–40), no definite correlation between sample dehydration and efficiency of triplex formation has been reported. The NMR studies (41,42) suggest the presence of immobilized water molecules in all the three grooves of the triple helix formed by homopurines and homopyrimidines. The data from molecular dynamics simulations show a spine of water molecules that bridge the amino groups of cytosine and guanine in the groove between pyrimidine and the third strand of the PyPuPu triplex (23,24).

Thus, very little is actually known about triplex hydration at the molecular level. The nature of the casual role of hydration in conformational equilibrium and free energy awaits a detailed explanation. The present study of the dynamical organization of counterions and water molecules around a DNA triplex is an effort in this direction. The implications of the results are of potentially considerable interest in the structural biology of nucleic acid sequences. The presence of ions in sequence-specific pockets in the groove would be expected to have the net effect of mitigating electrostatic repulsion among the various anionic phosphates in the region. This could, in turn, influence the helix morphology that is, axis bending and groove widths.

METHODS

The initial structure of the DNA triple helix was built by using the parameters from the Arnott fibre diffraction model (43). The creation of the initial structure, equilibration and dynamics was performed as described previously (31). The model developed by us was taken for further studies. The particle mesh EWALD method (PME) within AMBER 5.0 (44) has been used in this study. The simulation was performed in the presence of TIP3P water molecules at atmospheric pressure and room temperature. The PME charge grid leads to the size of 60 × 60 × 48 Å. The restrained molecular dynamics was run for over 500 ps during equilibration period, which was followed by five rounds of minimization where the solute restraints were reduced by 5 kcal/mol during each cycle. Finally, the system was heated from 100 to 300 K over 2 ps and then the production run (for 1.6 ns) was initiated.

The computed result was analyzed using the CARNAL, RDPARM/PTRAJ, MDANAL of AMBER 5.0 (44). Standard angles and helicoidal parameters are determined using CURVES-5.3 (45). Nucleic acid residue names are referred to in the text as one-letter codes with a residue number; residue numbers in the 5′ to 3′ directions are 1–10 for the first strand, 11–20 for the second strand and 21–30 for the third strand. Numbers 31–57 represent the residue number of ions in the solution. The same nomenclature is used in the discussion of the results. For the anionic phosphate oxygen atoms, the nomenclature OR/OS instead of OA/OB or O2P/O1P has been used. Average structures from the trajectories were calculated using the CARNAL module of AMBER 5.0 (44). The helicoidal parameters were calculated from the average structure of 1000–1600 ps dynamics run.

Water and ion distribution around triplex molecule was calculated using the PTRAJ module of AMBER 5.0 (44). The spacing was 0.1 Å, while the water density was 0.033 molecules/ų that corresponds to a density of water equal to 1.0 g/ml. The residence time of the water molecules forming the bridging structure has been calculated by the CARNAL module. The number of water molecules forming the hydrogen bonds with the triplex was calculated at various intervals during the dynamics. At any given time, a particular water molecule, which is present in a specific pocket for a certain length of time, forms a hydrogen bond with a specific group of atoms. The time duration for which the particular water molecule is forming a hydrogen bond with the specific group of atoms can be ascertained by looking into the structure at short intervals of time. The atomic fluctuations of the triplex including the cations were calculated with the MDANAL module of AMBER. All the molecular graphics images were produced using the Insight II program of the Biosym package (46). Several macros have been developed to make the AMBER data compatible with the Insight II module of the Biosym package to plot the figures. All simulations were run on Silicon Graphics workstations, Power Indigo2 and Indy in our laboratory at DDU Gorakhpur University, Gorakhpur.

RESULTS AND DISCUSSION

The DNA triplex contains a variety of hydrogen bond donor [base (G) H1/H21/H22, (C) H41/H42, (A) H61/H62 and (T) H3] and acceptor [backbone O3′, OR, OS, O5′; sugars O4′; base (G) N3/N7/O6, (C) N3/O2, (A) N3/N7 and (T) O2/O4] sites. Among them, hydroxyl-, imino- and amino-hydrogen atoms as well as non-protonated N3 and N7 base atoms may establish one-hydrogen bond contacts with the solvent, and simultaneously oxygen atoms may form up to two or three hydrogen bonding contacts with the surrounding water molecules. The water accessibility of these different hydrophilic sites will be discussed, followed by a precise characterization of the residence time of structurally important water molecules.

Global hydration features

The total solvation density for the DNA triplex molecule is composed of contributions due to water molecules and counterions, which is observed by inspection of the animation of the dynamics motion of the system. In the minor groove, considerable localization has been found, both of water molecules and Na⁺ ions, at various sites. The localization of water molecules and Na⁺ ions is also seen close to the surface of the DNA triplex. A close inspection shows that the ‘spine of hydration’ for the minor groove of the DNA triplex is similar to that for the DNA duplex. In the first shell of water, no particular evidence has been observed for spines of regular filaments in the groove formed between the second and the third strands, but some workers (41,42) have found a fully connected network of water molecules for specific sequences which stabilizes the triplex structure by solvating the complex. The water molecules in the groove between the first and the third strands form the filaments or spines of hydration in the first shell. These water molecules in the grooves may screen repulsive electrostatic interactions between phosphate groups across the narrow groove in the duplex and between the third strand and the duplex strand of the triplex across the Watson and Crick grooves. They also act as bridging polar groups belonging to different strands. The presence of these groups in the specific grooves are sequence specific, therefore the spine of hydration is also sequence specific. In the mixed sequences, the observed spine of hydration is not regular throughout the sequence. The molecular dynamics simulation for the triplex formed by mixed sequences suggests the presence of water molecules bound with NH₂ groups of G, A, and C; O2 group of C and T; N3 of A and G, in the nucleotide sequences.

The calculation of the cylindrical distribution of water molecules with respect to the helical axis of the DNA molecule (8) is well adapted to study the hydration pattern of a DNA triplex. The average number of water molecules in the first shell (3.0 Å) is nearly the same (110–111 per strand) for all the three strands (the end nucleotides were not considered). The boundary of the first solvation shell of the triplex as described by simulation may be defined in terms of the first minimum point in radial distribution functions of the solvent molecules (47) (Fig. W1 in the Supplementary Material). The radial distribution function of water around the phosphate oxygen shows that the distribution starts from ∼1.75 Å. The spine of water associated with the triplex molecule can be visualized from Figure 1, where a snapshot of water molecules located in a 3 Å shell from the DNA surface is shown. The number of water molecules in the first shell of the triplex (defined as waters with oxygen located less than 3.0 Å from the DNA triplex surface) is found to average out at 11.4 waters per nucleotide (including end effect). This number is consistent with estimates from spectroscopic measurements of 11–12 waters per nucleotide in the case of duplexes. The variation of the number of water molecules with time, at distances of 3 and 3.4 Å from the surface of the DNA triplex is shown in Figure W2 in the Supplementary Material. The water distribution around the different nucleotides shows that water molecules are arranged in a chain-like structure in grooves around the triplex and thus the molecule remains fixed by hydrogen bonds in solution. The triplex molecule in solution is behaving both as a donor and an acceptor for the water molecules. The donor protons from the triplex molecule, for example, H22 of all guanines in the triplex, H62 of adenine bases and H41 atom of cytosine bases of the third strand are capable of making hydrogen bonds with the oxygens of water molecules. The HO5′ and HO3′ of end nucleotides of the DNA triplex make hydrogen bonds with a water molecule.

Figure 1.

A snapshot of the DNA triplex with water molecule within 3 Å from the DNA surface. Some of the water molecules are shown as a ball and stick model. Both hydrogens of these molecules are involved in hydrogen bonding with the DNA triplex. The triplex is shown as a stick model in which, for the duplex, stick size is 0.3 while for the third strand, it is 0.2 [key: first strand, red; second strand, violet and third strand, yellow; water, white and ions, blue].

Hydrogens of C and T are not participating in the formation of hydrogen bonds with water molecules, while such binding has been observed for A-DNA (7). It is also observed that O6, N7 and N3 of G; N3 and N7 of A; N4 and O2 of C and O2 and O4 of T are playing a role as acceptors of protons of water molecules around triplets. The distances between the acceptor atoms and corresponding hydrogen atoms involved in a particular hydrogen bond vary from 1.7 to 2.2 Å. The oxygen of sugars and phosphate of nucleotides are also participating in hydrogen bond formation at the minor groove side of the triplex and are forming similar pattern as for a duplex.

It is interesting to see here that both the hydrogen atoms of some water molecules are forming hydrogen bonds with atoms from different strands thereby forming bridges between the strands. O2 of T and O4′ of G nucleotides are making a bridging structure with water molecules through hydrogen bonds in the TpG sequence. It is fascinating to see that water molecules are forming a strip-like pattern around the phosphate groups wherein both the hydrogen atoms of water are making a bridging structure with O5′ and OR, O3′ and OS, and OR and OS in some of the nucleotides. N7 of A29 is making a bridging pattern with its OR. Similarly, O4′ of G18 and N3 of A17, O4′ of G5 and O2 of T4 and O4′ of T19 and N3 of G18 are hydrogen bonded via a water molecule. In addition to the intra-strand-bridging pattern, the inter-strand spine patterns through hydrogen bonds are also seen between strands whenever water molecules find a suitable environment. For example, N7 of G18 of the second strand and N4 of C23 of the third strand make a bridging structure during the dynamics. Similar inter-strand bridging has also been observed between N7 of G6 and O2 of T27; N7 of A14 and O6 of G26; O2 of C13 and N3 of A9; O4 of T27 and N7 of A14. The number of water molecules around the donor and acceptor atoms of the DNA bases at a distance of 4 Å, for all the three strands, have been calculated and are shown in Table 1. This table indicates the number of possible hydrogen bonded water molecules with various atoms of the DNA triplex. It is seen that the hydrogen bond formation capability of backbone atoms with water molecules is approximately the same for all the strands. For the base atoms though, the hydrogen bond formation is different and depends on the choice of a particular strand of the triplex. As shown in Table 1, the number of water molecules in different grooves of the triplex is varying and therefore, it is sequence specific.

Table 1. The average number of water molecules around various atoms (for all three strands in the triplex) at a distance of 4.0 Å.

Strand	N1	N3	N6	N7	N9	O3′	O4′	O5′	P
Adenine
1st	0.5	1.4	1.3	3.4	1.3	6.5	2.4	6.1	12.6
	(0.4)	(0.4)	(0.4)	(0.6)	(0.7)	(0.9)	(0.7)	(0.8)	(1.0)
2nd	0.8	2.4	0.8	2.4	2.4	6.5	2.4	6.7	12.8
	(0.5)	(0.6)	(0.3)	(0.4)	(0.4)	(0.8)	(0.7)	(0.8)	(1.1)
3rd	0.8	2.3	2.2	3.3	1.8	6.5	3.0	6.3	12.8
	(0.4)	(0.6)	(0.4)	(0.6)	(0.5)	(0.9)	(0.8)	(0.8)	(1.1)
	Strand	N1	N3	N4	O2	O3′	O4′	O5′	P
Cytosine
1st	1.1	0.4	0.0	1.8	6.5	2.2	6.7	13.4
	(0.5)	(0.2)	(0.0)	(0.7)	(1.1)	(0.8)	(1.0)	(1.5)
2nd	1.6	0.8	1.2	2.3	6.4	2.4	6.1	12.3
	(0.4)	(0.3)	(0.3)	(0.4)	(0.7)	(0.4)	(0.7)	(0.9)
3rd	1.0	1.3	1.6	2.0	6.4	2.4	6.2	12.8
	(0.5)	(0.6)	(0.5)	(0.5)	(1.1)	(0.7)	(1.1)	(1.4)
Strand	N1	N3	O4	O2	Methyl	O3′	O4′	O5′	P
1st	1.4	0.7	0.0	2.3	3.2	5.8	2.2	6.2	12.6
	(0.5)	(0.4)	(0.0)	(0.6)	(0.8)	(0.9)	(0.7)	(0.8)	(1.1)
2nd	1.3	0.7	0.2	2.6	3.6	6.1	2.9	6.3	12.6
	(0.5)	(0.4)	(0.3)	(0.5)	(0.6)	(0.8)	(0.6)	(0.8)	(1.0)
3rd	2.0	0.9	0.4	2.9	3.0	6.3	2.9	6.6	12.7
	(0.6)	(0.4)	(0.2)	(0.6)	(0.7)	(0.9)	(0.7)	(0.7)	(1.3)
Strand	N1	N2	N3	N7	N9	O6	O3′	O4′	O5′	P
1st	0.5	4.2	2.8	1.4	2.1	0.2	6.2	3.1	6.0	12.0
	(0.3)	(0.5)	(0.5)	(0.4)	(0.4)	(0.2)	(0.7)	(0.6)	(0.7)	(0.9)
2nd	0.5	4.0	2.5	1.9	2.0	1.0	6.6	3.3	6.3	12.7
	(0.5)	(0.8)	(0.8)	(0.9)	(0.9)	(0.1)	(1.3)	(1.1)	(1.2)	(1.6)
3rd	0.6	2.8	3.1	2.4	1.8	0.8	6.8	2.7	6.3	12.5
	(0.3)	(0.5)	(0.4)	(0.4)	0.5	(0.3)	(0.7)	(0.5)	(0.7)	(0.9)

The standard deviations are shown in parentheses, below the average values during the last 200 ps dynamics.

Ion solvation and counterion condensation

All the Na⁺ cations were observed both for the restrained dynamics and the unrestrained dynamics to enable us to discuss the diffusion of ions around the DNA. Since the general stability of the DNA triplex is maintained, it indicates that the dynamic stability of the triplex is not dependent on the starting ion configuration. A plot of the calculated cylindrical distribution function g(R) of cations around the phosphates and the running co-ordination number Nc(R) for the PME MD trajectory is presented in Figure 2. An inflection point at 17.5 Å has been observed. The area under this curve indicates that ∼66% of the counterions may be described as condensed (48). The presence of an inflection point at 8.5 Å indicates those counterions which are within the groove region of the triplex whereas the inflection point at 3.75 Å indicates the presence of cations within the first hydration shell. These ions may be bound in any of the three grooves of DNA triplex. An animation of the motion of the system during dynamics was prepared and it was observed that some of the ions were positioned in the TpG, GpA, CpT and CpC pockets during the equilibration process. During the production dynamics however the ions moved away and were replaced by water molecules. Some counterions move to similar positions in other pockets except for the A7pG8 pocket, where the cation hovers in the same region during the whole dynamics. It is to be noted that two cations 41 and 42 moved from the backbone side to the region between the second and third strands where they are stabilized due to the interaction with the N7 atoms of purines. The snapshots of the positions of these ions are shown in Figure 3. Details and their stereo diagrams are available in the Supplementary Material (Fig. W3). The cation 41 (Fig. 3c) positioned itself between the N7 atoms of A17-G18, G21-A22 nucleotides during equilibration and remained there during the whole dynamics. The cation 42 (Fig. 3d) has positioned itself between the N7 atoms of A14, G25 and G26 and has remained there during the whole dynamics run. It is to be noticed that another cation 37 (Fig. 3b) which was initially not positioned in such a pocket is seen during the dynamics and positions itself near the N7 atoms of G11 and A29. It is seen that cations seem to prefer a position between the nucleotides of the second and the third strands whenever the corresponding nucleotides at the 5′ end of both the strands are purine. The average structure of the DNA triplex for the simulation during 1000–1600 ps is shown in Figure 4, in which 10 Na⁺ ions are seen, which are within 5 Å from the DNA surface. During the dynamics, the Na⁺ are observed to exchange positions with the nearby water molecules. The ions are subsequently displaced farther away from these pockets in concert with the motions of a number of diffusing water molecules. It is observed that the Na⁺, which were previously in the first hydration shell, moved towards the DNA backbone. After further dynamics, the Na⁺ are observed to penetrate the second hydration shell of the triplex and this is also shown by the inflection points in Figure 2. The co-ordination of the sodium ions pulls the two bases close together, resulting in a locally anomalous propeller twist angle. The MD predicts that the nature of these structures is not static but dynamic, with significant exchange of ions and waters at favourable sites in the expected grooves of the spine. The hydration state of various sequences in DNA and their affinity for multivalent cations may be relevant to the overall triplex-forming ability. Based on the measurement of ultrasonic velocity during a course of DNA titration with Mg²⁺, Buckin et al. (49) came to the conclusion that Mg²⁺ forms two types of complexes with AT and GC base pairs. Mobile counterions show sequence- and strand-specific patterns during the dynamics as indicated by r.m.s.d. calculations. The relative motion of sodium ions for GpT and TpG sequences was small throughout the DNA triplex but for a few exceptions like T4pG5 and G11pT12. The lowest r.m.s.d. fluctuations observed were 1.48 and 1.94 Å for G18pT19 and T27pG28, respectively. On the other hand, largest motions were observed for PupPu. The ion movements were very high for the end triplets and gradually decrease towards the middle nucleotide of the triplex. The lowest ion fluctuations were observed for the third strand during the dynamics. This is most likely because of the presence of oxygen and nitrogen atoms in the strands, which are not participating in hydrogen bonds of the triplets.

Figure 2.

Variation of cylindrical distribution function g(R) and running coordination number Nc(R) with distance of the Na⁺ ions from the triplex surface around the DNA helical axis.

Figure 3.

A snapshot of the position of counterions during simulation. The ions are shown in blue, water in white and all the bases are in different colours. (a) Position of 33rd ion in the T7pG8 pocket of the backbone. (b) Position of 37th ion near the N7 atom of A29 residue of third strand in the G28pA29 pocket. (c) Position of 41st ion close to the N7 atoms of A17, G18 of second strand and G21, A22 of third strand. (d) Position of 42nd ion near the N7 atoms of A14 of second strand and G25, G26 of third strand.

Figure 4.

Average structure of the DNA triplex for 1000–1600 ps simulation. The Na⁺ ions within 5 Å distance are also shown. The duplex and ions are shown in CPK, while the third strand is shown in the stick model. All are in different colours (1st, first strand; 2nd, second strand and 3rd, third strand).

Stability of base triplets

It is apparent that all the strands are intact and bind to each other during the unrestrained dynamics simulation. The average structure for the last 600 ps simulation is shown in Figure 4. The DNA parameters (31) show that the conformation of the duplex of the structure so formed lies between the A- and B-form of DNA. The total angle measured between the local axis vectors of the first and the last helical axis segments is 16.5°. The widths and depths of the major and the minor grooves of the duplex in the triplex structure have been calculated. In general, it is seen that the depths are nearly identical for A- and B-forms of DNA while their widths differ, the major groove being wider for B-DNA while the minor groove is wider for A-type DNA. A comparison of the fibre-diffraction and MD average structure shows that the major groove width increases from about 2.7 Å (A-DNA) and 11.7 Å (B-DNA) to ∼17 Å in the duplex of this triplex structure. The minor groove shows a reduction in width from 11.0 Å (A-DNA) to 5.5–7.5 Å. In the case of fibre diffraction data (43), the major and minor groove widths were 13.9 and 8.9 Å, respectively. The molecular dynamics simulation on a homopurine–homopyrimidine triplex, d(TC)₅d(GA)₅d*d(CT)₅, shows that the major and the minor groove widths change to 9.8 and 10.7 Å, respectively (23). These groove widths clearly indicate the sequence dependence of the DNA triplex. This is likely to affect the binding properties of drugs with triplexes in the minor groove. Analysis of the trajectory yielded reasonable average hydrogen bonding distances between the bases of the third strand and the DNA duplex. The variation in the hydrogen bond lengths for the base-pairs and base triplets are shown in Figure W4 in the Supplementary Material and the corresponding average length is shown in Table W1 also in the Supplementary Material. As shown in Figure W4, the 1.6 ns simulation yielded reasonable hydrogen bond distances, although significantly larger compared with the standard Watson–Crick hydrogen bond lengths, which persisted throughout the simulation trajectory. The average minimized structures of the base triplets formed by nucleotide bases are shown in Figure 5.

Figure 5.

The structure of base triplets from the average minimized structure of the DNA triplex.

A-T-A triplets

In the DNA triplex under study, it is observed that A-T-A triplets are present near both ends of the sequences. The hydrogen bonding distances between A1: N6–A22: N1 and A9: N6–A29: N1 show that stable hydrogen bonds are formed with average bond distances of 3.0 (0.1) and 3.1 (0.3) Å, respectively. The other hydrogen bonds formed by this triplet are between T: O4–A: N6, the average distances being 3.3 (0.4) and 3.3 (0.4) Å, respectively for T19: O4–A22: N6 and T12: O4-A29: N6. The average distance between A: N6 of the first while third strand is 3.4 Å and the corresponding oscillation for A9-T12-A29 triplet is shown in Figure W4-B(a).

T-A-T triplets

These triplets are present at the fourth and the seventh positions in the sequence and are stabilized by the hydrogen bonds formed between T4-A17-T24 and T7-A14-T27. The variations in the corresponding hydrogen bond distances during the simulation are shown in Figure W4-B(b). The simulation yielded reasonable hydrogen bonding lengths between T4: O4–T24: N3 [3.20 (0.32) Å] and between T7: O4–T27: N3, [3.1 (0.2) Å], which persisted throughout the trajectory. The other hydrogen bonds formed by the third strand between A17: N6–T24: O4 and A14: N6–T27: O4 have average distances of 3.3 (0.5) and 3.1 (0.3) Å, respectively and these are also maintained during the simulation. These hydrogen bond lengths are notably longer than the standard hydrogen bond length and are certainly weaker than the Watson–Crick hydrogen bonds shown in Figure W4-B(c, d).

G-C-G triplets

In such triplets, the third strand may form three/four hydrogen bonds with the DNA duplex. The hydrogen bonding distances between the donors and acceptors have been analysed throughout the simulation and corresponding distances are plotted in Figure W4-C. The average hydrogen bonding distances are reasonable and persist throughout the simulation trajectory. The strong hydrogen bond is formed between G: O6 of the first strand and G: N1 of the third strand and corresponding variations are shown in Figure W4-C(a). The average hydrogen bond lengths between these atoms lie between 2.9 and 3.0 Å, as shown in Table W-1. The middle G-C-G triplet yielded reasonable hydrogen bond distances between G: O6 of the first strand and G: N2 of the third strand. As shown in Table W-21 in the Supplementary Material, the corresponding values for these triplets are ∼3.2 and 3.1 Å, respectively, while the outer G-C-G triplets have shown extended hydrogen bonding distances as shown in Figure W4-C(b). The hydrogen bonding distances between G: N7 of the first strand and G: N2 of the third strand have reasonable average values between 3.04 and 3.3 Å for all the four triplets as shown in Table W-1 and these persist throughout the simulation trajectory [Figure W4-C(d)]. These distances are notably longer (∼3.2 Å) for the first and the fifth triplets and reasonable for the sixth (3.2 Å) and the eighth (3.0 Å) triplets. The other possible hydrogen bonds for G-C-G triplets are between C: N4 of the second strand and G: O6 of the third strand. The hydrogen bonding distance for these bonds are also reasonable but longer than the usual hydrogen bond lengths. The fluctuations in the hydrogen bond lengths are shown in Figure W4-C(e). These hydrogen bond lengths affect the distances between G: N7 of the first strand and G: N1 of the third strand which fluctuates during simulation. Its variation is plotted in Figure W4-C(c) in the Supplementary Material.

C-G-C triplets

C-G-C triplets occupy the third and tenth position in the DNA triplex. As mentioned earlier this triplet is stabilized by the formation of two or three hydrogen bonds. Reasonable hydrogen bonding distances between donors and acceptors have been observed during the simulation and these persist throughout the simulation trajectory. The average hydrogen bonding lengths between C3: N4–C23: N3 and C10: N4–C30: N3 are 3.0 (0.2) and 3.1 (0.2) Å, respectively, while the distances between G18: O6–C23: N4 and G11: O6–C30: N4 are 3.3 (0.3) and 3.3 (0.5) Å, respectively. The C: N4 from the first strand can also form a hydrogen bond with C: O2 of the third strand. The respective lengths for C3-G18-C23 and C10-G11-C30 triplets are 4.0 (0.4) and 3.3 (0.4) Å. The atom N7 of guanine of the second strand is in the major groove from where the third base is interacting. The distances between G: N7 of the second strand and C: N4 of the third strand (Fig. W4-D) are 5.4 (0.4) and 4.6 (0.6) Å, respectively. Strong hydrogen bonds between G:N7 of the DNA duplex and C:N4 of the third strand are generally observed in case of triplexes formed by homopurine–homopyrimidine sequences.

The hydrogen bonding patterns of the base triplets in the present case (Figs 5 and W4; Table W1) show that the third strand is strongly bonded with the first strand. The major groove has one proton donor (C:N4 or A:N6) and two proton acceptor (G:O6, G:N7 or T:O4, A:N7) sites. The Watson–Crick side of the bases of the third strand interact with the major groove side of the DNA duplex. The sequence of the duplex also affects the interaction of third bases of the triplet, which may hover in the major groove due to attractions of various groups. As a result of this, the third base may shift in the major groove (in the base-pair plane) to form suitable hydrogen bonds with the base pair of the DNA duplex. The shift in the bases however should lie within a certain limit so that it is suitably accommodated in the specific range (28) and isomorphism is maintained. These shifts are compensated by minor variations in the local conformational parameters.

Structural role of water bridges

Water bridges are recurrent motifs in nucleic acid hydration. In a systematic survey of crystal structures, water bridges occurring between backbone, sugars and bases have been described and their structural role has been emphasized. The high-resolution structure of the RNA duplex revealed a regular column of water molecules linking all adjacent OR atoms (50). Water bridges have also been observed in Z-DNA (51) and B-DNA simulations (6–10,52). For helical conformations, energy minimization studies have reproduced some water molecules with one-water bridges as seen by X-ray crystallography (7).

From this set of MD simulations, a number of structurally important water bridges have been characterized. Among them, the most frequently involved adjacent OR atoms are not participating in regular water bridges. Such bridges are specific to A-DNA and RNA and do not occur in regular B-DNA structure (4). This again indicates that the triplex structure is close to the B-DNA structure. As mentioned above, several intra-strand spine and inter-strand spine bridges of water molecules have been observed.

The dynamical stability of some of the water bridges is about 500 ps for our MD trajectory but no water molecule forms bridges for the total dynamics simulation. By NMR methods, residence times of the order of a nanosecond have been proposed for structurally important water molecules which are trapped in nucleic acid hydration pockets (53,54). However, the upper limit of residence time is not known with precision and may be dependent on the conformation adopted by the nucleic acid as well as the sequence. Our simulation proposes both an estimation of the residence time of water molecule in hydration pockets and a description of the interactions stabilizing these particular solvent molecules. The average number of water molecules around specific atoms at a distance of 4 Å is shown in Table 1. The actual number of water molecules around a specific atom governs the hydrogen bonding pattern of water with the corresponding groups. The localized atoms of the DNA create specific pockets and are responsible for the interaction of water and ions. The overall local charges on a specific pocket play an important role in the interaction of water and ions and the specific interacting group may hover in the same pocket for a rather longer period as a result of which the residence period of such a group increases. Residence time of the water molecules in a specific hydration pocket is sequence specific and depends on the regular pattern of the conformation; such conformations are possible in the case of homopyrimidine–homopurine sequences (41,42).

The occurrence of several water molecules displaying a long hydrogen bond contact time with solute atoms but not involved in one-water bridges seems to suggest that with the passage of time, the particular water molecule involved in the formation of a one-water bridge may be replaced by another water molecule. This is likely since a slight change in the local conformation of the DNA triplex with time may bring another water molecule into a more suitable position for the formation of the bridge. This mechanism of involving several water molecules participating in making a one-water bridge may be important in maintaining the stability of the DNA triplex. Such multiple water bridges, although frequently observed in crystal structures, have not yet been investigated for triplexes.

SUPPLEMENTARY MATERIAL

Supplementary Material is available at NAR Online.