Conformations and Dynamics of the Phosphodiester Backbone of a

J. Phys. Chem. B 2007, 111, 4235-4243

4235

Conformations and Dynamics of the Phosphodiester Backbone of a DNA Fragment That Bears a Strong Topoisomerase II Cleavage Site Brigitte Rene´ , Gre´ goire Masliah, Saı1d El Antri,† Serge Fermandjian, and Olivier Mauffret* UMR 8113 CNRS, Laboratoire de Biotechnologies et Pharmacologie Ge´ ne´ tique Applique´ e (Ecole Normale Supe´ rieure de Cachan), Institut GustaVe Roussy, 39 rue Camille Desmoulins, 94805 Villejuif Cedex, France ReceiVed: December 4, 2006; In Final Form: February 2, 2007

The dynamics of the DNA phosphodiester backbone conformations have been studied for a strong topoisomerase II cleavage site (site 22) using molecular dynamics simulations in explicit water and in the presence of sodium ions. We investigated the backbone motions and more particularly the BI/BII transitions involving the  and ζ angles. The consensus cleavage site is adjacent to the phosphate which shows the most important phosphodiester backbone flexibility in the sequence. We infer that these latter properties could be responsible for the preferential cleavage at this site possibly through the perturbation of the cleavage/ligation activities of the topoisomerase II. More generally, the steps pur-pur and pyr-pur are those presenting the highest BII contents. Relations are observed between the backbone phosphodiester BI/BII transitions and the flexibility of the deoxyribose sugar and the helical parameters such as roll. The roll is sequence dependent when the related phosphate is in the BI form, whereas this appears not to be true when it is in the BII form. The BI/BII transitions are associated with water migration, and new relations are observed with counterions. Indeed, it is observed that a strong coupling exists between the BII form and the presence of sodium ions near the adjacent sugar deoxyribose. The presence of sodium ions in the O4′ surroundings or their binding could assist the BI to BII transition by furnishing energy. The implications of these new findings and, namely, their importance in the context of the sequence-dependent behavior of BI/BII transitions will be investigated in future studies.

Introduction Understanding the principles of DNA-protein recognition is of considerable interest because of its role in the expression and regulation of the genetic information. Recent studies have shown that the DNA-protein recognition does not proceed via a simple lock-and-key mechanism. Instead, a significant degree of structural adaptation must occur within the protein, the DNA, or both.1-8 This adaptation is intimately connected with the dynamics of the molecules and determines the binding thermodynamics of the process.2,9,10 Thus, information on the structures and dynamics of the bound and unbound partners are important to collect. Molecular dynamics (MD) simulations are a powerful tool in studying the molecule dynamics, and in recent years a number of such studies have been performed on protein-bound and free DNA molecules.11-13 These simulations have been used to describe several properties of DNA as its sequence-dependent bending capability or its phosphodiester backbone dynamics.14,15 In some cases it has been possible to demonstrate that DNA molecules can achieve a significant degree of preorganization, permitting the formation of stable protein-DNA complexes. A prime example is the catabolite activator protein (CAP)DNA complex in which the extreme bending of the DNA observed in the complex is, for a significant part (40%), intrinsic to the sequence-dependent structure of the free oligomer.2 This preorganization must be seen as dynamical and is adequately * Corresponding author. Phone: 33 1 42 11 50 91. Fax: 33 1 42 11 52 76. E-mail: [email protected]. † Present address: Saı¨d El Antri, Laboratoire de Chimie Bioorganique et Analytique, Faculte´ des Sciences et Techniques, BP 146, Mohamedia, Morocco.

described by the populations of conformers observed during the course of MD simulations. The protein could be able to select out a DNA conformation present in this population, a process described as a conformational capture phenomena.16 In addition, the MD simulations provided a means to observe the role of the water molecules and counterions, contributors to stability and specificity, in the mediation of protein-DNA recognition.17-22 MD simulations achieved on free DNA revealed the impact of water molecules and counterions on the conformational parameters of DNA.11,23-25 For several years, our laboratory has been analyzing the DNA topoisomerase II enzyme and the properties of its DNA substrates. This enzyme controls the topological state of the DNA. The global cleavage/religation reaction consists in the passage of a DNA duplex (G) through another duplex (T) resulting in a control mechanism to regulate the level of supercoiled DNA in cells.26,27 In the conditions of our studies, the enzyme cuts each DNA strand of the G duplex with an invariant stagger of four base pairs, in the presence of ellipticine derivatives the cleavage of one of the two strands generally occurring immediately after a thymine residue.28 Despite the importance of this reaction for the function of topoisomerase II, relatively little is known about the structural requirements leading to recognition and cleavage. To obtain new information on these processes we searched the DNA sequence preferentially recognized and cleaved by the human topoisomerase II.28 We studied here a sequence (site 22) identified in this latter search (Scheme 1). Its structural and conformational properties are investigated using MD simulations in order to pinpoint peculiar features that could be linked to its selection by the topoisomerase II enzyme.

10.1021/jp0683115 CCC: $37.00 © 2007 American Chemical Society Published on Web 03/29/2007

4236 J. Phys. Chem. B, Vol. 111, No. 16, 2007 SCHEME 1: Site 22 Sequencea

a The 19 mer DNA derives from the strongest cleavage site in pBR322 for the enzyme topoisomerase II in the presence of an ellipticine derivative (ref 28). The red arrows indicate the bonds cleaved by topoisomerase II. The T in red indicates the highly conserved thymine (T-1) in the topoisomerase cleavage site.

We focused on characterizing the motions of the DNA backbone. The DNA exhibits conformational polymorphism with backbone substates which rapidly interconvert. X-ray diffraction analysis,29-32 later confirmed by NMR and other spectroscopic approaches,33-36 has revealed the existence of two main backbone states: BI (, trans; ζ, gauche -) and BII (, gauche -; ζ, trans) interconverting freely in solution. The relative proportion of these conformers during a given period of time appears to be sequence dependent according to rules that have yet to be determined. Other motions of significant amplitude which involve the backbone are the C2′-endo-C3′endo equilibrium which has been shown to be coupled to BI/ BII relations37 and the R/γ transition recently described.38,39 Due to the great number of interactions between DNA phosphate groups and protein side chains either direct or water mediated,6,7,40,41 the BI/BII transitions are hypothesized to play a genuine role in DNA-protein recognition.33,42 Several recent MD studies revealed new information on these aspects: BI/BII interconversions occur synchronously to hydrogen-bond breaking and formation,42 play a role in favoring the curvature of DNA in nondirectly contacted regions,43 or disturb the pattern of water molecules at the protein-DNA interface.41 The backbone-specific structural features of the topoisomerase II site were obtained from a 10 ns MD simulation. The cleaved site demonstrates extensive BI/BII exchange which could be linked to its preferential cleavage. More generally, we noted that the dinucleotide steps pyrimidine-purine and purinepurine lay more often in the BII conformation than the purinepyrimidine and pyrimidine-pyrimidine steps.30,31,44 The analysis of hydration of phosphate groups revealed interesting correlations between water migration and DNA backbone conformers. We also found that monovalent cations could stimulate the BII transition by interacting with oxygens of the sugar ring in the minor groove. The two events appear to be correlated. The NaDNA sugar interactions and their connected BII transitions are sequence dependent since the sodium binding sites vary according to the chemical nature of contiguous nucleotides. Experimental Methods Computational Methods. Molecular dynamics simulations were carried out under NPT conditions (303 K, 1 atm) with the suite of programs AMBER 4.1 using the modified version of the Cornell et al. force field parm 98.45,46 Water was simulated with the TIP3P model. All bond lengths, involving hydrogen atoms, were restrained using the SHAKE algorithm (tolerance ) 0.0005),47 and a time step of 2 fs was used. Electrostatic interactions were calculated using the particle mesh Ewald summation method with a 106 Ewald convergence tolerance.48,49 Periodic boundary conditions were used with a rectangular unit cell. The initial box was truncated to achieve a minimum distance of 11 Å beyond all DNA atoms in all directions, resulting in a box size of 49 × 49 × 95 Å3 and 5190 water

Rene´ et al. molecules. Thirty-six neutralizing Na cations were placed using a simple energy minimization algorithm (Leap module of AMBER 5.0). The starting structure was the AMBER-generated canonical B-DNA double helix. The energy of the system was minimized for 2250 cycles, using a combination of steepest descent and conjugate gradient algorithms. During the energy minimization, the oligomer conformation and the counterion positions were maintained with harmonic restraints (100 kcal/ mol Å2) with a progressive diminution of the force constants for a final 500 cycles without restraints. The system was then heated to 303 K over 10 ps, the velocities were rescaled up to 150 K, and then the system was coupled to a heat bath, with the weak coupling Berendsen algorithm. During heating, harmonic constraints (25 kcal/mol Å2) were imposed on the atomic positions of the oligomer and of the sodium counterions. After a constant-mass, constant-pressure, and constant-temperature (NVT) equilibration period of 5 ps at 303 K, the simulation was performed under constant-mass, constant-pressure, and constant-temperature (NPT) conditions (303 K, 1 atm), using coupling constants of 0.2 ps. The restraints were relaxed over a period of 25 ps until a free system was achieved, and a final equilibration was carried out for 5 ps before the beginning of the 10 ns production phase. Then a Maxwell distribution of the velocities was assigned to the system, followed by 10 ps of free equilibration. This procedure was repeated before the beginning of the production phase. After 50 and 100 ps of production a Maxwell distribution of the velocities was again assigned. The velocities were redistributed to attenuate the influence on the dynamics of the initial configuration of the system. The production phase lasts 10 ns. Calculations were performed on a Silicon Graphics Origin 200 server. Structural Analysis. We used the following numbering for the DNA sequence: 5′(A1G2C3T4T5A6T7C8A9T10C11G12A13T14A15A16G17C18T19):(A20G21C22T23T24A25T26A27C28G29A30T31G32T33A34A35G36C37T38). CURVES 5.2 was used to analyze DNA structures stored every 1 ps of MD.50 This work refers to the local parameters output by CURVES. The terminal residues were not taken into account. Each parameter was measured 10 000 times. The CARNAL module of AMBER 6.0 was used for the analysis of DNA diluted in water and ions as well as for the structural properties (root mean square deviation (rmsd), hydrogen bond, and distance between atoms). The hydrogenbond criterion was a maximum donor-acceptor distance of 3.5 Å and a minimum donor-proton-acceptor angle of 120°. Results and Discussion We performed MD simulations of the 19 base pair site 22 DNA duplex d(AGCTTATCATCGATAAGCT).(AGCTTATCGATGATAAGCT) fully solvated and neutralized by 36 Na+ ions using AMBER 6 with the parm98 all-atom force field.45 The duplex form was retained throughout the MD trajectory of 10.5 ns, and equilibration was reached after 0.5 ns from the starting point. Data were collected through a 10 ns period (production phase), which allowed a good sampling (atom coordinates were saved and analyzed every 1 ps) of conformational events. Root mean square deviations as a function of time for coordinates of heavy atoms in the 17 central base pairs of the duplex versus the B-DNA canonical structure are presented in Figure 1. It shows that after 500 ps the duplex has completely relaxed from the starting conformation and follows its own dynamics. The superimposition of a symmetrical six base pair portion present in the upper (2-37 and 7-32) and lower (1826 and 26-13) part of an averaged molecule computed on the whole 10 ns range provided an rmsd value of 0.6 Å. This value

Topoisomerase II Site DNA Backbone Conformations

J. Phys. Chem. B, Vol. 111, No. 16, 2007 4237

Figure 1. Root mean square atom positional deviations as a function of simulation time from B-DNA. The production phase begins 500 ps after the equilibration phase.

is comparable to that of Young et al. (0.72 Å) observed for a similar six base pairs superimposition obtained on a dodecamer.51 The system appears correctly equilibrated particularly considering the sequence is not perfectly palindromic. Note that the convergence of the commonly used DNA parameters within 5 ns has been recently reported from a 60 ns MD study on a dodecamer.52 In the following we will describe the motions affecting the DNA backbone, specifically those occurring in the surrounding of the highly conserved thymine (T-1) in the topoisomerase cleavage site (site 22) (Scheme 1). Backbone Conformation. The largest fluctuations we observed in the backbone concern the torsion angles  and ζ. These are interconversions between the BI and BII forms, and recent studies underline the reversible character of these transitions.39 The value of the -ζ difference allows easy identification of the BI (-ζ around -90°) and BII (-ζ around +90°) forms. Figure 2 contains the -ζ difference as a function of time for all the nonterminal residues. The diversity of the patterns is in agreement with the known observation that BI/BII transitions vary considerably with the nature of the step.31 Global analysis of data provides an average of nearly 20% of BII conformation in each strand, which is consistent with experimental results31,33-35 (Table 1). Also in accordance with previous experiments and calculations, the highest BII rates are assumed by the pur-pur steps G12pA13 (71%), A15pA16 (66%), G28pA29 (66%), G31pA32 (68%) and the pyr-pur steps T24pA25 (54%) and C8pA9 (41%).13,14,30-35,37,44,45,53-55 However, the pyr-pur step C11pG12 and its symmetry-related step C27pG28 display only low BII conformation contents. Influence of Sequence Context on BII Conformation. In our MD simulations the three steps exhibiting the highest BII content are GpA steps with BII populations of 65-70%. This is true for G12pA13 (in strand 1) and for its symmetry-correlated G28pA29 (in strand 2) as well as for G31pA32 (in strand 2). The three steps are flanked by pyr residues on both sides: the context is CGAT and TGAT. The reverse ApG step in the AAGC context shows very low BII population, while the purpur step ApA surrounded by a pyr3′ and pur5′ (TAAG) displays 40-65% of BII conformation. These data together with previous findings31 reveal the high sensitivity of pur-pur steps’ conformations to sequence context. These results also prove to be consistent with the data of Dixit et al.39 who systematically investigated the impact of the flanking sequences (at the tetranucleotide level) on the flexibility of the various dinucleotides. When the dinucleotide is preceded by a purine, these authors concluded that the flexibility is reduced for the purpur steps.

Figure 2. Substate interconversion for the different steps of -ζ angles as a function of time. The BI state corresponds to a value of -ζ near -90° and the BII state to an -ζ value near +90°.

Experimental data, issued from NMR and X-ray crystallography, have shown that the pyr-pur steps CpG, CpA, TpG, TpA frequently adopt the BII conformation.30,33,35 Here the occurrence of the BII form in these steps appears dependent on the neighbors, namely, a pyr-pur step did not exhibit high BII content if the neighbor is a pur-pur step with a high BII content. In contrast, the CpA (C8pA9), flanked by a pyr-pyr step (T7pC8) and a pur-pyr step (A9pT10) both of which exhibiting

4238 J. Phys. Chem. B, Vol. 111, No. 16, 2007

Rene´ et al.

TABLE 1: BII State Content of the Nonterminal Stepsa step

BII content (in %°)

G2pC3 C3pT4 T4pT5 T5pA6 A6pT7 T7pC8 C8pA9 A9pT10 T0pC11 C11pG12 G12pA13 A13pT14 T14pA15 A15pA16 A16pG17 G17pC18

11.6 0.0 0.0 9.9 17.8 6.7 46.0 2.7 23.8 4.1 71.1 0.5 8.1 66.5 5.0 44.4

G21pC22 C22pT23 T23pT24 T24pA25 A25pT26 T26pC27 C27pG28 G28pA29 A29pT30 T30pG31 G31pA32 A32pT33 T33pA34 A34pA35 A35pG36 G36pC37

30.4 0.0 0.0 54.2 0.0 22.8 10.4 66.8 0.0 5.4 68.0 0.2 12.5 36.8 14.6 3.5

a The backbone was assumed to be BII if  was gauche - (240360°) and ζ was trans (120-240°).

low BII content, displays a high BII content. The two pyr-pur steps T24pA25 and C8pA9 are characterized by a similar BI and BII content (∼50%), while the pur-pur steps rich in BII forms exhibit more than 65% of the BII form. In addition, the comparison of the -ζ variations versus time shows that C8pA9 is characterized by an instability in both the BI and the BII states, while the BI state has been shown to exhibit a longer mean lifetime relative to that of the BII state.39 This behavior seems particularly interesting as the C8pA9 step is immediately adjacent to the phosphate cleavage site (T7pC8). We speculate that this feature could affect the ligation/religation activity of topoisomerase II. The dynamics of BI/BII conversions have been shown to play an important role in the DNA recognition and cleavage processes.6,31 In P site the two cleaved steps T7pC8 (strand 1) and C27pG28 (strand 2) express low BII contents (∼7% and 10%, respectively), but both have adjacent 3′ steps, C8pA9 and G28pA29, with high BII contents (46% and 67%, respectively). Correlated BI/BII Transitions and Ring Dynamics. Previous studies strongly suggest that phosphates and deoxyribose rings are submitted to concerted motions which take place on similar timescales.37 Plots of -ζ (deg) as a function of the pseudorotation angle for the pur-pur step G12pA13 and the pyr-pur step C8pA9 (presented in Figure 3) illustrate the link that exists between the phosphate conformation and the 5′ deoxyribose ring pucker. These results clearly demonstrate that BII is linked with the C2′-endo conformation of the sugar ring, while BI allows a wider range of puckers (from C3′-endo to C2′-endo). It is also important to note that C8pA9 with nearly equal BI and BII contents is the step which reveals the higher backbone flexibility

Figure 3. Plot of the -ζ angle related to the phosphate backbone vs the phase of the sugar ring located in 5′ to the phosphate for the 10 000 conformers sampled during the production phase: for the G12pA13 (a) and the C8pA9 (b) steps.

and C9 is the sugar which displays the highest proportion of C3′-endo conformer (Table 2). NMR studies of Isaacs and Spielmann37 have suggested that a high BII content is associated with a 5′ deoxyribose in the N-conformational range (The N (“north”) range encompasses a phase angle of 0 ((90°) which includes the C3′-endo puckering). This observation conflicts with crystallographic results of Schneider et al.40 which stipulate that the BII conformer locks the C2′-endo conformation and the BI conformer allows a greater sugar flexibility. Our MD results (Table 2), however, are consistent with X-ray data. They show that the BII conformer constrains the 5′ sugar in the C2′-endo conformation regardless of the step considered. Moreover, the step C8pA9 presents a similar content of BI and BII forms; thus, the largest flexibility while the C8 sugar is the more flexible sugar with 5.9% of north form. In contrast, the purpur steps with high BII content do not exhibit 5′ deoxyribose dynamics. Relation between BI/BII Transitions and Helicoidal Parameters. Previous theoretical studies have demonstrated that BI/BII transitions are accompanied by variations of helical parameters such as the twist, the roll, and the X-disp.31,32 Among these, the roll is the helical parameter most sensitive to variations in the backbone conformation. Figure 4, parts a and b, illustrates the relations between the roll variations and the BI/BII (-ζ) transition for two distinct steps: one exhibiting high BII content but weak 5′ deoxyribose dynamics (G12pA13) and another with close BI and BII contents and high 5′ deoxyribose dynamics (C8pA9). A mean roll value of ∼ -4° is associated to BII conformations, while the BI conformations G12pA13 and C8pA9 are associated with a mean roll value of ∼0° and ∼12° to the BI, respectively. This suggests that the influence of the sequence context on roll is stronger with the BI conformation than with the BII one. Plots of roll values along the site 22 sequence for BI and BII conformers either treated separately (Figure 5, parts a (a1, a2) and b (b1, b2)) or mixed together

Topoisomerase II Site DNA Backbone Conformations

J. Phys. Chem. B, Vol. 111, No. 16, 2007 4239

TABLE 2: C3′-endo Content of the Nonterminal Residuesa residue

C3′-endo content (in %°)

G2 C3 T4 T5 A6 T7 C8 A9 T0 C11 G12 A13 T14 A15 A16 G17

0.1 1.4 1.2 1.7 2.3 0.4 5.9 1.2 0.4 4.7 0.8 1.9 0.5 0.5 0.3 0.1

G21 C22 T23 T24 A25 T26 C27 G28 A29 T30 G31 A32 T33 A34 A35 G36

0.1 1.8 0.6 2.9 1.1 0.2 4.0 0.4 0.9 0.1 1.5 4.6 3.0 3.2 3.1 0.2

a

The sugar was assumed to be in the C3′-endo if the sugar pucker phase is between -1° and 34°.

(Figure 5, part a (a3) and b (b3)) have been obtained for both strand 1 and strand 2. Separate analysis of BI and BII conformers provides a better illustration of the roll sequence dependence. With the BI states (Figure 5, parts a (a1) and b (b1)), the roll distribution clearly appears sequence dependent and roll values are always positive. The pyr-pur steps have the largest roll values, and the pur-pyr steps have the lowest ones, while the other steps adopt intermediary roll values. In the BII state (Figure 5, parts a (a2) and b (b2)), the rolls are generally negative and display a pattern that appears sequence independent. In comparison with BI phosphates, BII phosphates are shifted toward the minor groove. This entails a narrowing of the minor groove which renders the roll more negative. Conversely, in the BI form it is the major groove which is reduced, leading to a more positive roll. In considering sugar puckers and helicoidal parameters, which attest conformational variability, values associated with BII conformers are only weakly scattered. In contrast to the values associated with BI conformers, these values are not sequence dependent. These characteristics have also been noted by Djuranovic and Hartmann.56 Thus, the occurrence of BII forms in the backbone phosphodiester cannot be simply considered as a source of structural diversity, as it appears to freeze the helical parameters, as well as sugar conformations, in a limited range of values. Influence of Hydration on the Backbone Conformation. We studied the relations between the molecule hydration and the phosphodiester backbone conformation. We focused on analyzing the hydration of the unesterified backbone oxygens O1P and O2P and of the acceptor and donor atoms of the bases located in the major and minor groove. The hydration was

Figure 4. Plot of the -ζ angle related to the phosphate backbone vs the phase of the sugar ring located in 5′ to the phosphate for the 10 000 conformers sampled during the production phase: for the G12pA13 step (a) and C8pA9 step (b).

estimated from the number of hydrogen bonds to water molecules per picosecond and was determined separately for BI and BII conformers. The data are reported in Table 3 and reveal a large difference in the hydration of the O1P oxygen according to the backbone conformation (BI or BII), the hydration of the O1P in the BII form being lower. The plot reporting BII percentages for the different steps versus the level of hydration shows the strong relationship existing between the two parameters (Figure 6). Such a strong relation (R ) -0.95) does not exist for O2P in the major groove although this latter seems to be slightly more hydrated in the BII than in the BI form (Table 3). Concerning the binding of water molecules to OP, the neighbor O3′ could strengthen the binding of some water molecules but we find, as others authors, that the number of water molecules bound to O3′ is low (3 times less numerous than those bound to O1P). We may also note that the duration time of bonding to water is rather short and does not exceed 200 ps. The results of hydration for the base atoms are also reported in Table 3. The hydration of O4 (of thymines) and O6 atoms (of guanines), both located in the major groove, have been computed according the backbone conformation of the adjacent phosphate (in 5′ relative to the base; note that similar results are obtained when the phosphate conformation in 3′ of the base is considered). We found a slight difference of 0.1 and 0.15 for O4 and O6 atoms (major groove), respectively, in favor of a better hydration when the adjacent phosphate is in the BII form. The others base positions N7, N6, N4, N3, O2, and N2 oriented either in the major or the minor groove do not display significant hydration difference according to the conformation of adjacent phosphates. Overall, there is a clear relationship between the hydration of the unesterified O1P group that protrudes in the minor groove and the BII conformation: the lower is the hydration the higher

4240 J. Phys. Chem. B, Vol. 111, No. 16, 2007

Rene´ et al.

Figure 5. Plot of the mean roll values (averaged on the different conformers sampled) vs the sequence for strand 1 (a) and strand 2 (b). The color code is related to the nature of the step: red for purine-pyrimidine steps, green for pyrimidine-pyrimidine steps, yellow for purine-purine steps, and blue for pyrimidine-purine steps. In (a1) and (b1) only the conformers in the BI state are computed; in (a2) and (b2) only the conformers in the BII state are computed; in (a3) and (c3) all the conformers are taken into account regardless of their phosphate backbone conformation.

TABLE 3: Hydration of the Unesterified Oxygen and the Different Acceptor and Donors Atoms Located at the Edges of the Bases According the State of the Phosphate Conformationa O1P O2P

N7

O6

N6

N4

O4

N3

O2

N2

BI 3.60 3.40 1.19 1.27 1.08 1.20 1.14 1.02 1.09 0.80 BII 2.88 3.76 1.19 1.37 1.11 1.18 1.29 1.03 1.10 0.80 a The value is the mean number of hydrogen bonds of water molecules to the indicated atom during a period of 1 ps.

is the BII content. However, there is no coupling between the organization of water molecules at the edges of bases and those close to phosphates as no correlation seems to exist between the level of hydration of bases and phosphate conformation. When we detect a lower hydration in the minor groove from analysis of the phosphate hydration, the base hydration in the same groove did not seem to be diminished accordingly. Via crystal structures, the distribution of water molecules near phosphate groups has been thoroughly examined by Schneider et al.40 They were able to conclude that for purines the distribution of water molecules is similar around the BI and BII phosphates. In their analysis, Castrignano et al.57 did not separate the BI and BII conformers; thus, only results relative

Figure 6. Hydration of the O1P oxygen, described as the number of waters hydrogen-bonded to this oxygen per picosecond, as a function of the BII state percentage for the different steps of the molecule.

to mixed BI and BII conformations can be compared. For both O1P (3.44 vs 3.8) and O2P (3.48 vs 3.66) our values are similar to those of the latter analysis. And this despite the use of the previous AMBER 95 force field (parm95) by Castrignano et al.,57 which generally yields different BII contents from parm98.45 Results of Flader et al.58 and Winger et al.54 who used the same parm98 set of parameters also corroborate our results. Note also that the lower hydration associated with BII

Topoisomerase II Site DNA Backbone Conformations conformers has been observed in calculation on Ets-DNA complexes (enhancers of transcription). In this case, mutations of the wild-type sequence lead to a higher BII content, a decrease of water content at these loci, and a destabilization of the protein-DNA complexes.41 The hydration of bases has been extensively studied using X-ray data59,60 and MD simulations.11,17,23,25,57,58,61-63 The level of hydration we found is globally similar to those found in these studies. As Flader et al.,58 we did not detect a clear difference for the hydration of the bases when the associated BI and BII conformers were considered. Interestingly, the N7 of purines in the two pyr-pur steps C8pA9 and T24pA25, which both undertake frequent BI/BII interconversions, displays a particularly low level of hydration (not shown). Additional data will be needed to have a better understanding of the relation between the flexibility of nucleic acids and their hydration. However, several studies have brought interesting insights in these topics.23,64 Dynamic Behavior at the Cleavage Site. The pyr-pur step C8pA9 differs from the others, as it shows both frequent BI/ BII transitions and a significant duration of the lifetime in each substate. In contrast, a step such as T33pA34 shows frequent BI/BII transitions but the BII state appears unstable. The step C8pA9 is in the BII form 46% of the timesclose to a perfect equilibrium between the two substates. The step T24pA25 also displays similar contents of each substate. However, the high BII content is not related to frequent incursions in the BII state but to the fact it remains in this state for an exceptionally long period (more than 2 ns). As previously mentioned, this step is contiguous to T7pC8 which is cleaved by the topoisomerase II and contains the consensus thymine residue present in at least one of the cleavage site strands28 (Scheme 1). It is tempting to correlate the motions affecting the C8pA9 step and the presence of a preferred cleavage site at the contiguous T7pC8 step. Recent data28,42 suggest that BI/BII substates exert a direct role in protein sequence-specific recognition by modifying the electrostatic field in the environment of B-DNA. Their roles have also been shown to be of importance for the water organization at the proteinDNA interface.19,41,65 The analysis of BI/BII transitions, although important for a better understanding of DNA properties, is not an end in itself. Above all, the BI-BII equilibrium provides a useful picture of the sequence-dependent malleability of the phosphodiester backbone and its impact on complex formation. In this respect, the backbone plays a crucial role in the indirect readout of DNA sequences by proteins. For topoisomerase II, which accommodates the DNA in its active site mainly via contacts with the phosphodiester backbone (and the minor groove), the binding process certainly relies heavily on an indirect readout mechanism. In the topoisomerase I-DNA complex: the base-specific contacts are very few but are replaced by a large number of DNA backbone contacts that freeze and orient the DNA in adequate conformation in the complex.66 Here, no doubt that the site termed 22 fulfills all the structure/ dynamics conditions to become a strong binding site of topoisomerase II. The pattern of evolution versus time of dynamics reveals the occurrence of large motions in the phosphodiester backbone, especially at the step immediately adjacent to the scissile step (CpA). The selection of a cleavage site by the enzyme requires a good detection of these characteristics and is a complex process involving the following events: DNA recognition and formation of the complex, forward cleavage reaction, and then religation. Many studies show clearly

J. Phys. Chem. B, Vol. 111, No. 16, 2007 4241

Figure 7. Representation of the studied molecule with an indication of the steps presenting the highest BII content (red stars) and of the residues presenting the greatest numbers of oxygen O4′-to-sodium ion short distances (