Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 7-Day Trial for You or Your Team.

Learn More →

How methyl–sugar interactions determine DNA structure and flexibility

How methyl–sugar interactions determine DNA structure and flexibility Downloaded from https://academic.oup.com/nar/article-abstract/47/3/1132/5239025 by Ed 'DeepDyve' Gillespie user on 19 February 2019 1132–1140 Nucleic Acids Research, 2019, Vol. 47, No. 3 Published online 12 December 2018 doi: 10.1093/nar/gky1237 How methyl–sugar interactions determine DNA structure and flexibility Korbinian Liebl and Martin Zacharias Physics Department T38, Technical University of Munich, James-Franck-Str. 1, 85748 Garching, Germany Received September 28, 2018; Revised November 26, 2018; Editorial Decision November 28, 2018; Accepted November 30, 2018 ABSTRACT by mechanical stress and are coupled to the dimensions of minor and major groove (4–7). The distribution of BI/BII The sequence dependent structure and flexibility of substates in DNA can influence protein binding and may the DNA double helix is of key importance for gene also change upon protein binding (8–12). expression and DNA packing and it can be modulated In previous Molecular Dynamics (MD) simulation and by DNA modifications. The presence of a C5 -methyl NMR based studies, the sequence dependent population of group in thymine or the frequent C5 -methylated- BI/BII states and their impact on DNA’s structure has been extensively investigated (13–18). Recently, the sequence de- cytosine affects the DNA fine structure, however, pendence of BI/BII states was systematically analyzed us- the underlying mechanism and steric origins have ing trajectories obtained by the Ascona B-DNA consor- remained largely unexplained. Employing Molecular tium (ABC) based on a database of extensive MD simula- Dynamics free energy simulations that allow switch- tions of DNA oligomers containing all 136 distinct tetranu- ing on or off interactions with the methyl groups cleotide sequences (15,19). It was found that the sequence in several DNA sequences, we systematically iden- dependent formation of unconventional hydrogen bonds tified the physical origin of the coupling between between base and backbone atoms plays a key role in stabi- methyl groups and DNA backbone fine structure. lizing the BII substate (a C8-H8..O3 , between purine base Whereas methyl-solvent and methyl–nucleobase in- and backbone in case of RpR and YpR steps; and a C6- teractions were found to be of minor importance, H6..O3 contact in RpY and YpY steps(15,19)). The for- the methyl group interaction with the 5 neighboring mation of a (base) C-H..O3 (sugar) hydrogen bond con- sugar was identified as main cause for influencing tact perfectly correlated with the formation and percent- age of BII states at dinucleotide steps in DNA. A hierarchy the population of backbone substates. The sterical of bond strengths can be established from the populations methyl sugar clash prevents the formation of uncon- observed in simulations (for each dinucleotide step and se- ventional stabilizing hydrogen bonds between nucle- quence context) that can be used to interpret the observed obase and backbone. The technique was also used sequence-dependent BI/BII propensities (19–21). However, to study the contribution of methyl groups to DNA in principle, all purines and pyrimidines in DNA can form flexibility and served to explain why the presence of such contacts based on the same sterical and geometrical methyl sugar clashes in thymine and methyl-cytosine reasons, hence, the observed correlation does not offer a di- can result in an overall local increase of DNA flexibil- rect sterical explanation. ity. The presence of a methyl group in case of thymine and also the C5-methylation of cytosine has an influence on the DNA backbone structure and on the occurrence of BII INTRODUCTION states. The impact of methyl groups on DNA’s structure has The structure and flexibility of double-stranded DNA and been addressed in several previous studies (22–24), how- the binding of proteins is influenced by the nucleic acid ever, the sterical mechanism by which methyl compounds backbone structure (1–3). One of the most prominent con- influence the DNA backbone has still remained enigmatic. formational polymorphism in DNA is due to two differ- Methyl groups are directed towards the major groove in case ent combinations of the  and  dihedral angles in nu- of thymine and C5-methylated cytosine and are hydropho- cleotides adopting either the canonical BI (/ in the trans/ bic. Consequently, one might expect the altered hydration gauche-) or BII configuration ( / in the gauche-/trans, Fig- pattern in the major groove to be a central feature of DNA ure 1). These substates also contribute to the bimodal distri- methylation (25,26). Furthermore, it has also been stated bution of a base-pair step’s twist, are significantly affected To whom correspondence should be addressed. Tel: +49 89 289 12335; Fax: +49 89 289 12444; Email: [email protected] C The Author(s) 2018. Published by Oxford University Press on behalf of Nucleic Acids Research. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. Downloaded from https://academic.oup.com/nar/article-abstract/47/3/1132/5239025 by Ed 'DeepDyve' Gillespie user on 19 February 2019 Nucleic Acids Research, 2019, Vol. 47, No. 3 1133 MATERIALS AND METHODS Force field variation In total, seven different sequences of 15 base pair (bp) DNA duplexes have been studied. Starting structures were gener- ated using the nab module of the Amber16 package (33). The xleap module of Amber16 was used to generate pa- rameter topology files based on the parmbsc1 force field (34) and the DNA structures were solvated in explicit sol- vent (TIP3P water model (35)) within a rectangular box and a minimum distance of 10 A between DNA and box bound- aries. Potassium ions were added in order to neutralize the systems. For the base-atoms of C5-methylated cytosine we Figure 1. ApT base-pair step with a BI backbone conformation (A) and a BII conformation (B). used the parameters by Rauch et al.(24). These parameters have served as extension to the parm99 force-field. How- ever, updates to the parm99 force-field (e.g. parmbsc1) have that for C5-methylated cytosine BI states are stabilized by a addressed torsional backbone angles and not nonbonded water molecule bridging between the methyl-carbon and a parameters of the base-atoms. Hence, using the parameters phosphate bound oxygen (24). The hypothesis has, however, of Rauch et al.(24) in combination with a nucleic backbone been revised in a follow-up study by Wibowo et al.(27) that according to the parmbsc1 force field gives best compatibil- rather claims that methyl induced BI stabilization is due to ity with the force field description of all other nucleotides. an increased mean water residence time around base atoms. Classical MD force-fields have the form: The repulsion between methyl groups and the sugar ring of the 5 -neighboring base might stabilize the population of 2 2 BI states (28). Indeed, MD simulations by Peguero-Tejada E = k (r − r ) + k (θ − θ ) + b 0 θ 0 and coworkers (23) indicate higher populations of BI states bonds angles for DNA containing thymines compared to analogs with N−1 N uracil instead of thymine. q q i j + V [1 + cos(nφ − γ )] + + Finally, BI states are possibly stabilized through methyl- ij dihedrals i =1 j =i +1 stacking with adjacent bases. Interactions between the thymine-methyl group and a 5 -neighboring base are attrac- 12 6 R R min,ij min,ij tive and might have important implications on the deforma- + − 2 , (1) ij r r bility of DNA (29–31). ij ij In order to investigate the molecular mechanism and sterical origins on how the presence of a methyl group in where the last two terms describe the interactions between thymine and in C5-methylated cytosine influences the ra- all non-bonded atoms and are composed of Coulomb and tio of BI/BII states and DNA deformability we performed van-der-Waals contributions that allow us to specifically comparative MD simulations including or omitting non- modify interactions between pairs of atoms. bonded interactions between nucleobase methyl groups and Besides regular parameter topologies, we also prepared other parts of the DNA and solvent. The simulations reveal topology files with modified force-field descriptions. This that methyl– stacking and interaction of methyl groups includes the modification of the non-bonded interactions with solvent have negligible impact on DNA’s backbone between methyl groups and specific partner groups using structure. However, turning-off interactions between methyl the parmed module of Amber16. In order to eliminate spe- groups and the sugar C2 atom of the 5 -neighboring nu- cific non-bonded interactions, the partial charge on each cleotide and its hydrogen atoms drastically increases the atom of the methyl groups (C7, H71, H72, H73) was re- population of BII states. Besides influencing the BI /BII ra- moved and partial charges on adjacent atoms were redis- tio methyl groups may also influence the intrinsic flexibility tributed in accordance to the charge distribution in the of DNA (32). Using the same technique we also examined de-methylated analogs. In addition, the pair-wise van-der- how methyl–sugar clashes influence structure and flexibility Waals parameters between methyl group and defined part- of the base pairs. Methyl–sugar clashes are predicted to in- ner group were set to zero. Hence, for each sequence three crease the intrinsic bending but significantly reduce DNA’s additional parameter topologies were generated, neglecting local stiffness with the bending stiffness reduced by up to either: ∼40% and the stretching stiffness to ∼35%. Most strik- ing are the calculated effects on the twist-stiffness: For G– T and C–T steps, these clashes reduce the twist stiffness by ∼50–60%. The simulations allow us to explain the ob- • interactions between methyl groups and all water served effects based on sterical effects of the methyl groups. molecules Given the substantial contribution of methyl–sugar clashes • or interactions between methyl groups and the C2 atom to DNA’s stiffness, we suppose that this atomistic effect is and its hydrogens of the 5 -neighboring sugars of significant biological relevance, e.g. for the regulation of • or interactions between methyl groups and 5 neighbored gene expression. bases. Downloaded from https://academic.oup.com/nar/article-abstract/47/3/1132/5239025 by Ed 'DeepDyve' Gillespie user on 19 February 2019 1134 Nucleic Acids Research, 2019, Vol. 47, No. 3 Table 1. Sequences of the studied DNA duplexes DNA sequence abbrev. P(BI) [%] 5 -CGCGCATATACGCGC-3 AT 83.6 5 -CGCGCAUAUACGCGC-3 AU 75.5 5 -CGCGCGCGCGCGCGC-3 CG 73.4 5 -CGCGCGC∗GC∗GCGCGC-3 C*G 78.8 5 -CGCGCAAAAACGCGC-3 AA 88.3 5 -GCGCCTCTCTGCGCG-3 CT 77.2 5 -GCGCGTGTGTGCGCG-3 GT 76.4 C* denotes C5-methylated cytosine (on both strands). Simulation setup and equilibration All simulation systems were first energy minimized with the steepest descent method in 2500 steps by using the sander module of the Amber16 package. All subsequent MD sim- ulations were performed with the pmemd.cuda module of the Amber16 package. Initially, the systems were heated up to 300 K in three stages (in 100 K steps). Each stage was simulated for 100 ps and included positional restraints on all non-hydrogen atoms with respect to the B-DNA start- ing conformation. Subsequently, positional restraints were gradually reduced from 25 kcal/(mol A ) to 0.5 kcal/(mol A ) in vfi e consecutive simulations at 300 K and at con- stant pressure of 1 bar (weak coupling with a time constant of 5 ps). The equilibration phase was completed by a 2 ns simulation, during which only the first two base pairs were Figure 2. Modulation of the BI-BII free energy profile at central ApT steps positionally restrained with a small force constant of 0.1 by force field modifications. ( A–C) Illustration of specific force field modi- kcal/(mol A ) which avoids overall rotation of the DNA fications for specifically omitting methyl-solvent interactions (in A), steric in the simulation box. The equilibrated structures served clashes between methyl group and 5 -neighboring C2 -sugar atoms (and connected hydrogens, in B) and turning off methyl group interactions with as input for the production runs for each force field topol- neighboring base atoms (omitting methyl– stacking interactions, illus- ogy, during which we kept the soft restraints on the ter- trated in C). (D) Calculated free energy profiles along the – coordinate minal bases-pairs. Data gathering simulations were carried for the AT-case including all interactions or omitting interactions as in- out for 900-4000 ns. Coordinates were written out every dicated in panels A–C (indicated as different line colors or grey scales in the figure panel). For comparison the free energy profile for the AU case is 5000 steps. Using hydrogen-mass-repartitioning allowed us also shown (green curve). The free energies along the – coordinate were to use a time step of 4 fs. Details on the calculation of free calculated by Boltzmann-Inversion F =−k T · ln(p) and represent the av- energy profiles, calculations of errors and convergence and erage over all dinucleotide steps in the central segment of the AT (or AU) hydrogen bonding as well as conformational deformabilities sequence. The same data (enlarged) is shown in Supplementary Informa- are given in Supporting Information (Sections 1–5). tion Figure S1 including error bars. RESULTS AND DISCUSSION ulations (Figure 2 and Supplementary Information, Section Convergence, Figures S1–S6). DNA backbone substates The calculated populations (probabilities) for the canon- MD simulations were performed on seven DNA duplexes ical BI states for all investigated dinucleotide steps (Table 1) with different central sequences (Table 1) to record struc- emphasize two findings: First, sequences consisting only tural fluctuations including frequent transitions between of central A–T base pairs (AT,AA) show a higher popula- DNA backbone substates to sample substate populations. tion of BI states (by ∼10–15%) than other sequences. Sec- In a given dinucleotide step, a DNA backbone strand can ond, ‘elimination’ of a methyl group (by replacing central either adopt BI or BII configurations that are determined thymines by uracil or C5-methylated cytosine by cytosine) by the  and  dihedral angles (Figure 1): results in a lower occupation of BI states (by ∼5–8%, Ta- ble 1). Thus, the MD simulations indicate that the presence BI :  − ζ< 0 , BII :  − ζ>0(2) of methyl groups at the C5 of pyrimidines in DNA stabilizes the BI backbone states. Similar trends have been observed Transitions between BI and BII substates occur rapidly in other studies (23). relative to the total length of the simulations. Hence, it is In order to understand the molecular mechanism of this possible to directly extract probability distributions along stabilizing effect we performed MD simulations of the AT the  −  coordinate and to extract associated free energies case (central ATATA sequence, Table 1) by turning off non- on the time scale of 900–4000 ns used in data gathering sim- bonded interactions between the C5-methyl groups and se- Downloaded from https://academic.oup.com/nar/article-abstract/47/3/1132/5239025 by Ed 'DeepDyve' Gillespie user on 19 February 2019 Nucleic Acids Research, 2019, Vol. 47, No. 3 1135 Figure 3. Calculated free energy versus – coordinate for single base-pair steps in the AT-case. (A) Free energy simulation of AT-case including all interactions. (B) Simulations omitting thymine-methyl interactions with the C2 sugar atom (and connected hydrogens) of the 5 -neighboring nucleotide. (C) Snapshot of a BI conformation. (D) Normalized density plot of sampled sugar pucker phase of the 3 ring versus – coordinate (indicating low- phase preference in BI and high-phase preference for BII states). (E) Normalized density plot of sampled sugar puckering of 5 ring versus – coordinate (indicating high-phase preference in BI and low-phase preference for BII states). (F) Snapshot of BII conformation with arrows highlighting sugar pucker phase induced shifts. lected subsets of atoms in the system. During the simula- the change in sterical van der Waals interactions (see Sup- tions either all interactions of the C5-methyl group were plementary Information Figures S6–S11). turned off (representing an AU step) or only with the sol- vent, with the 5 neighboring base or with the C2 atom (and Why methyl–sugar clashes trigger BI promiscuity its hydrogens) of the 5 sugar. The associated free energies along the – coordinate were calculated by Boltzmann in- In order to better understand the effect of the steric in- version of the sampled probability distributions and repre- teractions between methyl group and 5 -sugar, we investi- sent the average over all steps in the central DNA sequence gated its influence on individual base-pair steps. Note, that (steps 6–9). Comparison of the AT versus AU cases indi- methyl–sugar clashes can only occur in base-pair steps with cates a negligible impact of the methyl group on the shape a thymine base at the 3 -position (e.g. in ApT but not in of the free energy curve in the BI subspace. However, the TpA steps). Interestingly, the population of BII states in BII regime is of significantly lower free energy resulting ApT steps is even more stabilized through the omission in an increased BII population. The exclusion of all inter- of methyl–sugar clashes than expected from the free en- actions between methyl groups and water molecules (Fig- ergy profiles obtained as averages over the central segments ure 2, purple curve) as well as the exclusion of all interac- (compare Figure 3 with Figure 2). Intriguingly, the omission tions between methyl groups and the 5 -neighboring bases of methyl-5 -sugar interactions leads to a remarkable desta- (Figure 2, yellow curve) also shows much smaller deviations bilization of BII states in the juxtaposed TpA steps. Based from the regular case (Figure 2, blue curve). However, ex- on the calculated free energies along the – coordinate we cluding interactions between C5-methyl groups and the C2 conclude that methyl–sugar clashes destabilize BII confor- atom (and its hydrogens) of the 5 -neighboring sugar results mations of ApT steps by ∼2kcal/mol but at the same time in a large drop of the free energy in the BII regime (Figure 2, the BII state of neighboring steps is stabilized (albeit to a red curve). Hence, neither methyl-base stacking nor interac- lesser degree of ∼0.5 kcal/mol for TpA steps). This finding tions with the solvent are decisive but the steric clashes be- reflects the anti-correlation of DNA base-pair steps, a phe- tween C5-methyl group and 5 -sugar are the main cause of nomenon which has also been subject of previous studies the BI stabilization by the C5-methyl group. It is interesting (18,36). to note that eliminating only the electrostatic interactions of Our simulations suggest a qualitative sterical explanation the methyl group with specific groups had only a small im- of this nearest-neighbor anti-correlation that is illustrated pact on the population of substates but it is dominated by in Figure 3C–F. For an ApT step in the BI configuration, the 3 sugar pucker (T-nucleotide) adopts preferably a lower Downloaded from https://academic.oup.com/nar/article-abstract/47/3/1132/5239025 by Ed 'DeepDyve' Gillespie user on 19 February 2019 1136 Nucleic Acids Research, 2019, Vol. 47, No. 3 Figure 4. Methyl–sugar clashes destabilize the BII subspace in an ApT step. (A) Center of mass distance of thymine methyl group from C2 atom of 5 neighboring sugar versus − coordinate in case of simulations including interactions of methyl groups with the C2 -atom of a 5 -sugar. BII states are populated very rarely. (B) Same as in (A) but switching-off interactions between methyl and sugar that allows for a closer approach of the groups. As a consequence, the BII subspace can also be accessed. (C) Snapshot of BI conformation, taken from simulation including all interactions. (D)Distanceof thymine methyl group vs. H6-O3 hydrogen bonding distance between thymine and 5 -sugar in case of simulations including all interactions (as in A). A close H6–O3 hydrogen bonding distance between thymine and 5 -sugar is not formed under these conditions. (E) Same as in D but interactions between methyl and sugar are switched-off, giving rise to sampling of close H6–O3 hydrogen bonding distances. (F) Snapshot of sampled BII conformation for simulation allowing clashes between methyl and sugar. The dashed red line indicates the unconventional hydrogen bond between thymine’s H6 and the backbone’s O3 atom. pucker phase than the 5 sugar (A nucleotide) to adopt a In a next step, we studied the influence of methyl–sugar well stacked configuration. A BII configuration, on the con- clashes for other sequences (Figure 5). Similar to the ApT trary, forces the 3 sugar to adopt preferably a higher pucker steps (see above), we also find that methyl–sugar clashes are phase that inclines the sugar ring in order to keep a near responsible for blocking H6-O3 hydrogen bonds for GpT, planar stacking geometry of adjacent bases (Figure 3). The CpT and GpC* steps (Supplementary Information Figures opposite is observed for the 5 -sugar. Since this latter sugar S12–S14). Note that we observed a different pattern for TpT adopts the role of a 3 sugar in the consecutive step and it has steps, where H6-O3 bonds are rare even when methyl–sugar been shifted to lower phase by the BII state, the population interactions are switched off. We rather find that H6-O5 of another BII step at this neighboring step is suppressed. hydrogen bonds are decisive for this sequence (Supplemen- The simulations also indicate a qualitative sterical mech- tary Information Figures S15 and S16). For simplicity, we considered the BI/BII population as average over the cen- anism for the BII destabilization (by the methyl group) tral DNA segment, though the same anti-correlation trend in case of the ApT steps as illustrated in Figure 4:The as before is obtained on the base-pair step level (illustrated methyl-5 -sugar sterical interaction locks the bases to a spe- in Supplementary Information Figures S17–S21). Similar cific conformational space in which the backbone prefer- entially adapts BI configurations. Switching off this steri- to the comparison of the AT and AU cases we find that cal hindrance allows both components to come closer to- methyl-5 -sugar clashes destabilize BII states for each inves- gether whereby also the BII space becomes accessible (com- tigated sequence. Indeed, the CT, GT and C*G (methylated pare Figure 4A and B). The population of these states is cytosine) cases show an even stronger increase in BII pop- then stabilized by unconventional hydrogen bonds between ulation upon omission of methyl–sugar interactions than thymine’s H6-atom and the O3 atom of the backbone sugar the AT-sequence (Figure 5). The sequence dependence of (Figure 4D and E). Notably, the existence and correlation to DNA backbone substates has also been studied experimen- the BII population of these unconventional hydrogen bonds tally. Based on NMR experiments, it has been shown that has already been pointed out by Balaceanu et al. (19), how- out of the ten dinucleotide steps, the four steps which con- ever, such hydrogen bonds are sterically possible only at spe- tain a thymine on the 3 position (ApT, GpT, TpT and CpT) cific base pair steps. clearly exhibit the lowest BII population (2). This confirms our results that methyl groups (specifically at the 3 posi- Downloaded from https://academic.oup.com/nar/article-abstract/47/3/1132/5239025 by Ed 'DeepDyve' Gillespie user on 19 February 2019 Nucleic Acids Research, 2019, Vol. 47, No. 3 1137 Figure 5. Calculated free energy profiles versus – coordinate for the AA, CT, GT and C*G sequences (blue lines correspond to simulations including all interactions whereas red lines indicate simulation results omitting interactions of the base methyl group and the 5 neighboring sugar C2 atom and connected hydrogens). tion) destabilize BII backbone states due to methyl–sugar the covariance-matrix of helical parameter fluctuations interactions. recorded during simulations (see Supplementary Informa- tion Section 5). The stiffness parameters were calculated with respect to the averages of the central four base pair Methyl–sugar clashes influence DNA’s global structure and steps. It is important to note, that the harmonic stiffness flexibility model is fully valid on this scale, as the steps superpose to The conformation of DNA’s backbone is strongly coupled single Gaussian distributions (indicating a quadratic under- to the configuration of the base-pairs ( 3,21,37,38). Indeed, lying effective free energy profile) whilst individual steps can the sterical methyl–sugar interactions also have an impact show clear bimodal behavior (Supplementary Information on the overall structure and flexibility of DNA in the present Figure S22). simulations. As relevant parameters we consider the mean Methyl–sugar repulsion is found to cause significant de- twist, stretching and bending of the central DNA segments. creases in twist-, stretch- and bending-stiffness making the Whereas changes in the equilibrium twist and stretch- DNA-molecule overall more flexible. methyl–sugar repul- ing are negligible, the intrinsic bending of the DNA double sion alone significantly decreases the bending-stiffness by helix is markedly increased by methyl–sugar repulsion for ∼40% for the methylated cytosine sequence (C*G) and by most sequences (Figure 6), e.g. for the methylated cytosine ∼20–30% for the GT and CT sequence. The largest changes sequence (C*G) an increase by ∼18% was observed. We cal- are found for the twist-stiffness of the CT and GT sequence, culated the DNA stiffnesses based on a harmonic stiffness where it has been found that methyl–sugar clashes soften by model (assuming an underlying quadratic free energy sur- ∼50–60%. Besides, also the stretching flexibility is enhanced face for twisting, bending and stretching) that has also been through methyl–sugar clashes, with the GT sequence show- used in previous studies (5,6,39–43), ing the largest effect (∼35%). −1 It should be emphasized that these findings do not nec- K = k T · C , (3) essarily indicate that DNA molecules containing methy- where k and T indicate Boltzmann constant and tem- lated cytosines are more flexible than sequences with un- perature, respectively. K denotes the stiffness- and C Downloaded from https://academic.oup.com/nar/article-abstract/47/3/1132/5239025 by Ed 'DeepDyve' Gillespie user on 19 February 2019 1138 Nucleic Acids Research, 2019, Vol. 47, No. 3 than GpC steps, while C*pG steps are overall more flexible than CpG steps. On the level of the whole central segment stretching and bending stiffnesses are slightly lower for the methylated case (C*G case) compared to the CG case (∼5– 7%), but the twisting-stiffness of the C*G case is signifi- cantly higher compared to the unmethylated central CG se- quence (∼20%, Supplementary Information Tables S1-S3). We also checked in how far the charge-reassignment alone of methylated-bases influenced changes in DNA’s structure and flexibility (Supplementary Information Figures S23– S24). Here, our simulations indicate, that such charge ef- fects are relevant for methylated cytosine sequences, but are negligible for canonical sequences. The effect of methyl–sugar clashes to increase flexibility is counterintuitive (restriction to BI substate), however, this is a direct consequence of the influence of the backbone: The twist-distribution within the BI states is broader than that of BII states (Figure 7A). Thus, an increased popula- tion of BII states results in a narrower overall twist distribu- Figure 6. Relative changes in structure and flexibility due to methyl–sugar tion (Figure 7B). The lower variance in twist directly reflects clashes. The first three columns indicate for each sequence the relative a higher stiffness with respect to this mode. Note, that from change in equilibrium twist, stretch and bending of the central segment the – free energy profiles one would expect that the twist- whereas the last three columns represent the change in flexibility (indicated as the change in the calculated stiffness constant). For the latter cases, red distribution of BI states is narrower, however, the backbone entries mean that methyl sugar clashes have a decreasing/softening effect population does not map linearly to the twist variable. In- and blue entries represent an increase of the stiffness. All changes are given deed, similar – configurations can show different twisting as relative to the reference case of reassignment of thymines’ and methy- (illustrated in Figure 7C and D). lated cytosines’ bases, i.e. the entries reveal the influence of the van der Waals interactions between methyl and sugar group. CONCLUSIONS methylated cytosine since we only evaluate the influence of The methyl group in thymine and in C5-methylated cy- methyl–sugar interactions by switching-off interactions be- tosine modulates the BI-BII substate distribution and the tween methyl and sugar group. It emphasizes that DNA’s lo- DNA flexibility. Since DNA methylation plays a key role cal and global deformability is strongly influenced by these in epigenetic regulation of gene expression (44–46), it is interactions. However, van-der-Waals interactions between likely that its influence on DNA deformability is also linked methyl groups and other chemical groups are still included. to its biological function (43,44,47,48). In the presented Previous studies have pointed out that methylated cytosine study, we addressed the correlation between methyl groups sequences are overall stiffer than their regular analogs (43). and BI/BII promiscuity using comparative MD simula- Based on comparative simulations (Supplementary Infor- tions. As a key technique we employed sets of simulations mation Section 7) we obtained the following trend: On the that specifically included or omitted non-bonded interac- level of base pair steps, GpC* steps are significantly stiffer tions between methyl groups and hypothetically important Figure 7. Methyl sugar clashes increase DNA flexibility at the central segment. ( A) Calculated twist distribution of a GpT step in case of simulations including methyl sugar interactions indicating the total cumulative distribution (both sampled BI and BII states) as well as distributions considering BI ◦ ◦ and BII states separately. The distribution at the BI states is shifted and broader (standard deviation  = 7.2 ) than that for BII states ( = 6.2 ). The standard deviation of the cumulative distribution amounts to  = 7.1 .(B) Twist distribution upon removing methyl–sugar interactions: The cumulative ◦ ◦ ◦ twist distribution is narrower since the population of BII is strongly increased ( = 5.7 ). (C) Snapshot of undertwisted GpT step (∼20 )at – −80 ◦ ◦ in Watson and Crick strand. (D) Snapshot of overtwisted GpT step (∼38.5 )at – −80 in both strands. Downloaded from https://academic.oup.com/nar/article-abstract/47/3/1132/5239025 by Ed 'DeepDyve' Gillespie user on 19 February 2019 Nucleic Acids Research, 2019, Vol. 47, No. 3 1139 partner groups. Based on these simulations, we showed that 3. Packer,M.J. and Hunter,C.A. (1998) Sequence-dependent DNA structure: the role of the sugar-phosphate backbone. J. Mol. Biol., the hydrophobicity of methyl groups as well as methyl- 280, 407–420. stacking that had been proposed to cause changes in the 4. Drsa ˇ ta,T., Per ´ ez,A., Orozco,M., Morozov,A.V., Sponer,J. and BI/BII ratio (24,25,27) exhibit only a small influence on the Lankas,F. (2013) Structure, stiffness and substates of the population of BI/BII substates. Switching-off interactions Dickerson-Drew dodecamer. J. Chem. Theory Comput., 9, 707–721. ˇ ˇ 5. Drsata,T. and Lankas,F. (2013) Theoretical models of DNA between methyl group and the C2 atom and its hydrogen flexibility. Wiley Interdiscip. Rev. Comput. Mol. Sci., 3, 355–363. atoms, in contrast, stabilized BII and hence decreased pop- 6. Liebl,K. and Zacharias,M. (2017) Unwinding induced melting of ulation of BI conformations significantly. This trend con- double-stranded DNA studied by free energy simulations. J. Phys. firms results of previous studies ( 23,26,28) and was found Chem. B, 121, 11019–11030. 7. Oguey,C., Foloppe,N. and Hartmann,B. (2011) Understanding the for each investigated sequence, albeit to different degrees: sequence-dependence of DNA groove dimensions: implications for The strongest changes appeared for sequences including DNA interactions. PLOS ONE, 5, 1–8. CpT and GpT steps or methylated cytosines. Previous stud- 8. R.,W.F., Christine,R. and Michael,T.R.L.K. (2005) M.TaqI facilitates ies indicate a decisive role of the formation of unconven- the base flipping via an unusual DNA backbone conformation. tional base-sugar H6–O3 hydrogen bonds (or H8–O3 in Biopolymers, 79, 128–138. 9. Robertson,J.C. and Cheatham,T.E. (2015) DNA backbone BI/BII case of purines) for stabilizing the BII state (19–21) that is distribution and dynamics in E2 protein-bound environment also found in the present simulations. However, since such determined by molecular dynamics simulations. J. Phys. Chem. B, bonds are in principle possible for every dinucleotide step 119, 14111–14119. the correlation alone does not explain the physical reason 10. Hegde,R.S., Grossman,S.R., Laimins,L.A. and Sigler,P.B. (1992) for the BI/BII sequence dependence. Based on sterical con- Crystal structure at 1.7 A of the bovine papillomavirus-1 E2 DMA-binding domain bound to its DNA target. Nature, 359, siderations it was possible to qualitatively explain the effect 505–512. of the methyl group on the BI/BII ratio. Interestingly, an 11. Djuranovic,D. and Hartmann,B. (2004) DNA fine structure and increased propensity of BII at one step results in a reduced dynamics in crystals and in solution: the impact of BI/BII backbone BII probability at a neighboring step that can be qualita- conformations. Biopolymers, 73, 356–368. 12. Djuranovic,D. and Hartmann,B. (2005) Molecular dynamics studies tively explained by a coupling to the nearest-neighbor sugar on free and bound targets of the bovine papillomavirus type I E2 pucker conformation. Given the pronounced impact of the Protein: the protein binding effect on DNA and the recognition interaction between methyl and sugar group it also influ- mechanism. Biophys. J., 89, 2542–2551. ences the global structure and flexibility of DNA ( 29–31). 13. Madhumalar,A. and Bansal,M. (2005) Sequence preference for While these interactions increase the intrinsic DNA bend- BI/BII conformations in DNA: MD and crystal structure data analysis. J. Biomol. Struct. Dyn., 23,13–27. ing, a decrease in the DNA’s stiffness was observed. For CpT 14. Heddi,B., Foloppe,N., Bouchemal,N., Hantz,E. and Hartmann,B. and GpT rich sequences, the steric clashing between methyl (2006) Quantification of DNA BI /BII backbone states in solution. and sugar group decreases DNA’s torsional rigidity by up Implications for DNA overall structure and recognition. J. Am. to ∼60%. We found as a main sterical reason that the twist Chem. Soc., 128, 9170–9177. 15. Pasi,M., Maddocks,J.H., Beveridge,D., Bishop,T.C., Case,D.A., distribution of states within the BI basin is broader than in Cheatham,T. III, Dans,P.D., Jayaram,B., Lankas,F., Laughton,C. case of sampled states within the BII regime. Consequently, et al. (2014) ABC: A systematic microsecond molecular dynamics since the methyl-5 -sugar interactions stabilize the BI state study of tetranucleotide sequence effects in B-DNA. Nucleic Acids the twist flexibility increases. The comparative simulation Res., 42, 12272–12283. 16. Ben Imeddourene,A., Elbahnsi,A., Guroult,M., Oguey,C., methodology of including or omitting specific non-bonded Foloppe,N. and Hartmann,B. (2015) Simulations meet experiment to interactions could be applied in future studies on other phe- reveal new insights into DNA intrinsic mechanics. PLoS Comput. nomena like phosphate repulsion or the role of hydrogen Biol., 11, 1–27. bonding in structure formation could be tackled. 17. Svozil,D., Kalina,J., Omelka,M. and Schneider,B. (2008) DNA conformations and their sequence preferences. Nucleic Acids Res., 36, 3690–3706. 18. Zgarbova,M., ´ Jurecka,P ˇ ., Lankas,F ˇ ., Cheatham,T.E., Sponer,J. and SUPPLEMENTARY DATA Otyepka,M. (2017) Influence of BII backbone substates on DNA Supplementary Data are available at NAR Online. Twist: A unified view and comparison of simulation and experiment for all 136 distinct tetranucleotide sequences. J. Chem. Inf. Model., 57, 275–287. 19. Balaceanu,A., Pasi,M., Dans,P.D., Hospital,A., Lavery,R. and FUNDING Orozco,M. (2017) The role of unconventional hydrogen bonds in determining BII propensities in B-DNA. J. Phys. Chem. Lett., 8, Deutsche Forschungsgemeinschaft (DFG) [SFB749, 21–28. project C05]; The Leibniz super computer (LRZ) center 20. Dans,P.D., Faustino,I., Battistini,F., Zakrzewska,K., Lavery,R. and is acknowledged for providing supercomputer access Orozco,M. (2014) Unraveling the sequence-dependent polymorphic [grant pr48ko]. Funding for open access charge: Deutsche behavior of d (CpG) steps in B-DNA. Nucleic acids Res., 42, 11304–11320. Forschungsgemeinschaft. ˜ ˇ ˇ 21. Dans,P.D., Danilane,L., Ivani,I., Drsata,T., Lankas,F., Walther,J., Conflict of interest statement. None declared. Pujagut,R.I., Battistini,F., Gelpı,J.L., Lavery,R. et al. (2016) Long-timescale dynamics of the Drew–Dickerson dodecamer. Nucleic Acids Res., 44, 4052–4066. REFERENCES 22. Temiz,N.A., Donohue,D.E., Bacolla,A., Luke,B.T. and Collins,J.R. (2012) The role of methylation in the intrinsic dynamics of B- and 1. Hartmann,B., Piazzola,D. and Lavery,R. (1993) BI-BII transitions in Z-DNA. PLoS ONE, 7, 1–9. B-DNA. Nucleic Acids Res., 21, 561–568. 23. Peguero-Tejada,A. and van der Vaart,A. (2017) Biasing simulations 2. Heddi,B., Oguey,C., Lavelle,C., Foloppe,N. and Hartmann,B. (2009) of DNA base pair parameters with application to propellor twisting Intrinsic flexibility of B-DNA: the experimental TRX scale. Nucleic Acids Res., 38, 1034–1047. Downloaded from https://academic.oup.com/nar/article-abstract/47/3/1132/5239025 by Ed 'DeepDyve' Gillespie user on 19 February 2019 1140 Nucleic Acids Research, 2019, Vol. 47, No. 3 in AT/AT, AA/TT, and AC/GT steps and their uracil analogs. J. 36. Lankas,F ˇ ., Sponer,J., Langowski,J. and Cheatham,T.E. (2003) DNA Chem. Inf. Model., 57, 85–92. basepair step deformability inferred from molecular dynamics 24. Rauch,C., Trieb,M., Wellenzohn,B., Loferer,M., Voegele,A., simulations. Biophys. J., 85, 2872–2883. Wibowo,F.R. and Liedl,K.R. (2003) C5-Methylation of cytosine in 37. Keepers,J.W., Kollman,P.A., Weiner,P.K. and James,T.L. (1982) B-DNA thermodynamically and kinetically stabilizes BI. J. Am. Molecular mechanical studies of DNA flexibility: coupled backbone Chem. Soc., 125, 14990–14991. torsion angles and base-pair openings. Proc. Natl. Acad. Sci. U.S.A., ˇ ˇ ¨ ¨ 25. Warmlander,S., Sponer,J.E., Sponer,J. and Leijon,M. (2002) The 79, 5537–5541. influence of the thymine C5 methyl group on spontaneous base pair 38. Okonogi,T.M., Alley,S.C., Harwood,E.A., Hopkins,P.B. and breathing in DNA. J. Biol. Chem., 277, 28491–28497. Robinson,B.H. (2002) Phosphate backbone neutralization increases 26. Teng,X. and Hwang,W. (2018) Effect of methylation on local duplex DNA flexibility: a model for protein binding. Proc. Natl. mechanics and hydration structure of DNA. Biophys. J., 114, Acad. Sci. U.S.A., 99, 4156–4160. 1791–1803. 39. Olson,W.K., Gorin,A.A., Lu,X.-J., Hock,L.M. and Zhurkin,V.B. 27. Wibowo,F.R., Trieb,M., Rauch,C., Wellenzohn,B. and Liedl,K.R. (1998) DNA sequence-dependent deformability deduced from (2005) The N6-Methyl group of adenine further increases the BI protein-DNA crystal complexes. Proc. Natl. Acad. Sci. U.S.A., 95, stability of DNA compared to C5-Methyl groups. J. Phys. Chem. B, 11163–11168. 109, 557–564. 40. Drsa ˇ ta,T., Kara,M., Zacharias,M. and Lankas,F ˇ . (2013) Effect of 28. Hunter,C.A. (1993) Sequence-dependent DNA Structure: The role of 8-Oxoguanine on DNA structure and deformability. J. Phys. Chem. base stacking interactions. J. Mol. Biol., 230, 1025–1054. B., 117, 11617–11622. ˇ ˇ 29. Parker,T.M., Hohenstein,E.G., Parrish,R.M., Hud,N.V. and 41. Drsa ˇ ta,T., Spack ˇ ova,N ´ ., Jurecka,P ˇ ., Zgarbova,M., ´ Sponer,J. and Sherrill,C.D. (2013) Quantum-Mechanical analysis of the energetic Lankas,F ˇ . (2014) Mechanical properties of symmetric and contributions to  Stacking in Nucleic Acids versus Rise, Twist, and asymmetric DNA A-tracts: implications for looping and nucleosome Slide. J. Am. Chem. Soc., 135, 1306–1316. positioning. Nucleic Acids Res., 42, 7383–7394. 30. Svozil,D., Hobza,P. and Sponer,J. (2010) Comparison of intrinsic 42. Liebl,K., Drsa ˇ ta,T., Lankas,F ˇ ., Lipfert,J. and Zacharias,M. (2015) stacking energies of ten unique dinucleotide steps in A-RNA and Explaining the striking difference in Twist-Stretch coupling between B-DNA duplexes. Can we determine correct order of stability by DNA and RNA: A comparative molecular dynamics analysis. Quantum-Chemical calculations? J. Phys. Chem. B, 114, 1191–1203. Nucleic Acids Res., 43, 10143–10156. 31. Umezawa,Y. and Nishio,M. (2002) Thymine-methyl/ interaction ´ 43. Perez,A., Castellazzi,C., Battistini,F., Collinet,K., Flores,O., implicated in the sequence-dependent deformability of DNA. Nucleic Deniz,O., Ruiz,M., Torrents,D., Eritja,R., Soler-Lpez,M. et al. (2012) Acids Res., 30, 2183–2192. Impact of methylation on the physical properties of DNA. Biophys. 32. Per ´ ez,A., Castellazzi,C.L., Battistini,F., Collinet,K., Flores,O., J., 102, 2140–2148. Deniz,O., Ruiz,M.L., Torrents,D., Eritja,R., Soler-Lopez,M. ´ et al. 44. Pennings,S., Allan,J. and Davey,C.S. (2005) DNA methylation, (2012) Impact of methylation on the physical properties of DNA. nucleosome formation and positioning. Brief. Funct. Genomics, 3, Biophys. J., 102, 2140–2148. 351–361. 33. Case,D., an D.S.Cerutti,R.B., T.E. Cheatham,I., Darden,T., Duke,R., 45. Jones,P.A. and Takai,D. (2001) The role of DNA methylation in Giese,T., Gohlke,H., Goetz,A., Homeyer,N., Izadi,S. et al. (2016) mammalian epigenetics. Science, 293, 1068–1070. Amber 16. University of California, San Francisco. 46. Laird,P.W. and Jaenisch,R. (1996) The role Of DNA methylation in 34. Ivani,I., Dans,P.D., Noy,A., Prez,A., Faustino,I., Hospital,A., cancer genetics aand epigenetics. Annu. Rev. Genet., 30, 441–464. Walther,J., Andrio,P., Goi,R., Balaceanu,A. et al. (2015) 47. Ngo,T.T., Yoo,J., Dai,Q., Zhang,Q., He,C., Aksimentiev,A. and PARMBSC1: A refined force-field for DNA simulations. Nature Ha,T. (2016) Effects of cytosine modifications on DNA flexibility and methods, 13, 55–58. nucleosome mechanical stability. Nat. Commun., 7, 10813. 35. Jorgensen,W.L., Chandrasekhar,J., Madura,J.D., Impey,R.W. and 48. Severin,P.M.D., Zou,X., Gaub,H.E. and Schulten,K. (2011) Cytosine Klein,M.L. (1983) Comparison of simple potential functions for methylation alters DNA mechanical properties. Nucleic Acids Res., simulating liquid water. J. Chem. Phys., 79, 926–935. 39, 8740–8751. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Nucleic Acids Research Oxford University Press

How methyl–sugar interactions determine DNA structure and flexibility

Liebl, Korbinian; Zacharias, Martin
Nucleic Acids Research , Volume 47 (3): 9 – Feb 20, 2019
Download PDF
9 pages

Loading next page...
 
/lp/oxford-university-press/how-methyl-sugar-interactions-determine-dna-structure-and-flexibility-8WKQ2hOKMI

References (52)

Publisher
Oxford University Press
Copyright
© The Author(s) 2018. Published by Oxford University Press on behalf of Nucleic Acids Research.
ISSN
0305-1048
eISSN
1362-4962
DOI
10.1093/nar/gky1237
Publisher site
See Article on Publisher Site

Abstract

Downloaded from https://academic.oup.com/nar/article-abstract/47/3/1132/5239025 by Ed 'DeepDyve' Gillespie user on 19 February 2019 1132–1140 Nucleic Acids Research, 2019, Vol. 47, No. 3 Published online 12 December 2018 doi: 10.1093/nar/gky1237 How methyl–sugar interactions determine DNA structure and flexibility Korbinian Liebl and Martin Zacharias Physics Department T38, Technical University of Munich, James-Franck-Str. 1, 85748 Garching, Germany Received September 28, 2018; Revised November 26, 2018; Editorial Decision November 28, 2018; Accepted November 30, 2018 ABSTRACT by mechanical stress and are coupled to the dimensions of minor and major groove (4–7). The distribution of BI/BII The sequence dependent structure and flexibility of substates in DNA can influence protein binding and may the DNA double helix is of key importance for gene also change upon protein binding (8–12). expression and DNA packing and it can be modulated In previous Molecular Dynamics (MD) simulation and by DNA modifications. The presence of a C5 -methyl NMR based studies, the sequence dependent population of group in thymine or the frequent C5 -methylated- BI/BII states and their impact on DNA’s structure has been extensively investigated (13–18). Recently, the sequence de- cytosine affects the DNA fine structure, however, pendence of BI/BII states was systematically analyzed us- the underlying mechanism and steric origins have ing trajectories obtained by the Ascona B-DNA consor- remained largely unexplained. Employing Molecular tium (ABC) based on a database of extensive MD simula- Dynamics free energy simulations that allow switch- tions of DNA oligomers containing all 136 distinct tetranu- ing on or off interactions with the methyl groups cleotide sequences (15,19). It was found that the sequence in several DNA sequences, we systematically iden- dependent formation of unconventional hydrogen bonds tified the physical origin of the coupling between between base and backbone atoms plays a key role in stabi- methyl groups and DNA backbone fine structure. lizing the BII substate (a C8-H8..O3 , between purine base Whereas methyl-solvent and methyl–nucleobase in- and backbone in case of RpR and YpR steps; and a C6- teractions were found to be of minor importance, H6..O3 contact in RpY and YpY steps(15,19)). The for- the methyl group interaction with the 5 neighboring mation of a (base) C-H..O3 (sugar) hydrogen bond con- sugar was identified as main cause for influencing tact perfectly correlated with the formation and percent- age of BII states at dinucleotide steps in DNA. A hierarchy the population of backbone substates. The sterical of bond strengths can be established from the populations methyl sugar clash prevents the formation of uncon- observed in simulations (for each dinucleotide step and se- ventional stabilizing hydrogen bonds between nucle- quence context) that can be used to interpret the observed obase and backbone. The technique was also used sequence-dependent BI/BII propensities (19–21). However, to study the contribution of methyl groups to DNA in principle, all purines and pyrimidines in DNA can form flexibility and served to explain why the presence of such contacts based on the same sterical and geometrical methyl sugar clashes in thymine and methyl-cytosine reasons, hence, the observed correlation does not offer a di- can result in an overall local increase of DNA flexibil- rect sterical explanation. ity. The presence of a methyl group in case of thymine and also the C5-methylation of cytosine has an influence on the DNA backbone structure and on the occurrence of BII INTRODUCTION states. The impact of methyl groups on DNA’s structure has The structure and flexibility of double-stranded DNA and been addressed in several previous studies (22–24), how- the binding of proteins is influenced by the nucleic acid ever, the sterical mechanism by which methyl compounds backbone structure (1–3). One of the most prominent con- influence the DNA backbone has still remained enigmatic. formational polymorphism in DNA is due to two differ- Methyl groups are directed towards the major groove in case ent combinations of the  and  dihedral angles in nu- of thymine and C5-methylated cytosine and are hydropho- cleotides adopting either the canonical BI (/ in the trans/ bic. Consequently, one might expect the altered hydration gauche-) or BII configuration ( / in the gauche-/trans, Fig- pattern in the major groove to be a central feature of DNA ure 1). These substates also contribute to the bimodal distri- methylation (25,26). Furthermore, it has also been stated bution of a base-pair step’s twist, are significantly affected To whom correspondence should be addressed. Tel: +49 89 289 12335; Fax: +49 89 289 12444; Email: [email protected] C The Author(s) 2018. Published by Oxford University Press on behalf of Nucleic Acids Research. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. Downloaded from https://academic.oup.com/nar/article-abstract/47/3/1132/5239025 by Ed 'DeepDyve' Gillespie user on 19 February 2019 Nucleic Acids Research, 2019, Vol. 47, No. 3 1133 MATERIALS AND METHODS Force field variation In total, seven different sequences of 15 base pair (bp) DNA duplexes have been studied. Starting structures were gener- ated using the nab module of the Amber16 package (33). The xleap module of Amber16 was used to generate pa- rameter topology files based on the parmbsc1 force field (34) and the DNA structures were solvated in explicit sol- vent (TIP3P water model (35)) within a rectangular box and a minimum distance of 10 A between DNA and box bound- aries. Potassium ions were added in order to neutralize the systems. For the base-atoms of C5-methylated cytosine we Figure 1. ApT base-pair step with a BI backbone conformation (A) and a BII conformation (B). used the parameters by Rauch et al.(24). These parameters have served as extension to the parm99 force-field. How- ever, updates to the parm99 force-field (e.g. parmbsc1) have that for C5-methylated cytosine BI states are stabilized by a addressed torsional backbone angles and not nonbonded water molecule bridging between the methyl-carbon and a parameters of the base-atoms. Hence, using the parameters phosphate bound oxygen (24). The hypothesis has, however, of Rauch et al.(24) in combination with a nucleic backbone been revised in a follow-up study by Wibowo et al.(27) that according to the parmbsc1 force field gives best compatibil- rather claims that methyl induced BI stabilization is due to ity with the force field description of all other nucleotides. an increased mean water residence time around base atoms. Classical MD force-fields have the form: The repulsion between methyl groups and the sugar ring of the 5 -neighboring base might stabilize the population of 2 2 BI states (28). Indeed, MD simulations by Peguero-Tejada E = k (r − r ) + k (θ − θ ) + b 0 θ 0 and coworkers (23) indicate higher populations of BI states bonds angles for DNA containing thymines compared to analogs with N−1 N uracil instead of thymine. q q i j + V [1 + cos(nφ − γ )] + + Finally, BI states are possibly stabilized through methyl- ij dihedrals i =1 j =i +1 stacking with adjacent bases. Interactions between the thymine-methyl group and a 5 -neighboring base are attrac- 12 6 R R min,ij min,ij tive and might have important implications on the deforma- + − 2 , (1) ij r r bility of DNA (29–31). ij ij In order to investigate the molecular mechanism and sterical origins on how the presence of a methyl group in where the last two terms describe the interactions between thymine and in C5-methylated cytosine influences the ra- all non-bonded atoms and are composed of Coulomb and tio of BI/BII states and DNA deformability we performed van-der-Waals contributions that allow us to specifically comparative MD simulations including or omitting non- modify interactions between pairs of atoms. bonded interactions between nucleobase methyl groups and Besides regular parameter topologies, we also prepared other parts of the DNA and solvent. The simulations reveal topology files with modified force-field descriptions. This that methyl– stacking and interaction of methyl groups includes the modification of the non-bonded interactions with solvent have negligible impact on DNA’s backbone between methyl groups and specific partner groups using structure. However, turning-off interactions between methyl the parmed module of Amber16. In order to eliminate spe- groups and the sugar C2 atom of the 5 -neighboring nu- cific non-bonded interactions, the partial charge on each cleotide and its hydrogen atoms drastically increases the atom of the methyl groups (C7, H71, H72, H73) was re- population of BII states. Besides influencing the BI /BII ra- moved and partial charges on adjacent atoms were redis- tio methyl groups may also influence the intrinsic flexibility tributed in accordance to the charge distribution in the of DNA (32). Using the same technique we also examined de-methylated analogs. In addition, the pair-wise van-der- how methyl–sugar clashes influence structure and flexibility Waals parameters between methyl group and defined part- of the base pairs. Methyl–sugar clashes are predicted to in- ner group were set to zero. Hence, for each sequence three crease the intrinsic bending but significantly reduce DNA’s additional parameter topologies were generated, neglecting local stiffness with the bending stiffness reduced by up to either: ∼40% and the stretching stiffness to ∼35%. Most strik- ing are the calculated effects on the twist-stiffness: For G– T and C–T steps, these clashes reduce the twist stiffness by ∼50–60%. The simulations allow us to explain the ob- • interactions between methyl groups and all water served effects based on sterical effects of the methyl groups. molecules Given the substantial contribution of methyl–sugar clashes • or interactions between methyl groups and the C2 atom to DNA’s stiffness, we suppose that this atomistic effect is and its hydrogens of the 5 -neighboring sugars of significant biological relevance, e.g. for the regulation of • or interactions between methyl groups and 5 neighbored gene expression. bases. Downloaded from https://academic.oup.com/nar/article-abstract/47/3/1132/5239025 by Ed 'DeepDyve' Gillespie user on 19 February 2019 1134 Nucleic Acids Research, 2019, Vol. 47, No. 3 Table 1. Sequences of the studied DNA duplexes DNA sequence abbrev. P(BI) [%] 5 -CGCGCATATACGCGC-3 AT 83.6 5 -CGCGCAUAUACGCGC-3 AU 75.5 5 -CGCGCGCGCGCGCGC-3 CG 73.4 5 -CGCGCGC∗GC∗GCGCGC-3 C*G 78.8 5 -CGCGCAAAAACGCGC-3 AA 88.3 5 -GCGCCTCTCTGCGCG-3 CT 77.2 5 -GCGCGTGTGTGCGCG-3 GT 76.4 C* denotes C5-methylated cytosine (on both strands). Simulation setup and equilibration All simulation systems were first energy minimized with the steepest descent method in 2500 steps by using the sander module of the Amber16 package. All subsequent MD sim- ulations were performed with the pmemd.cuda module of the Amber16 package. Initially, the systems were heated up to 300 K in three stages (in 100 K steps). Each stage was simulated for 100 ps and included positional restraints on all non-hydrogen atoms with respect to the B-DNA start- ing conformation. Subsequently, positional restraints were gradually reduced from 25 kcal/(mol A ) to 0.5 kcal/(mol A ) in vfi e consecutive simulations at 300 K and at con- stant pressure of 1 bar (weak coupling with a time constant of 5 ps). The equilibration phase was completed by a 2 ns simulation, during which only the first two base pairs were Figure 2. Modulation of the BI-BII free energy profile at central ApT steps positionally restrained with a small force constant of 0.1 by force field modifications. ( A–C) Illustration of specific force field modi- kcal/(mol A ) which avoids overall rotation of the DNA fications for specifically omitting methyl-solvent interactions (in A), steric in the simulation box. The equilibrated structures served clashes between methyl group and 5 -neighboring C2 -sugar atoms (and connected hydrogens, in B) and turning off methyl group interactions with as input for the production runs for each force field topol- neighboring base atoms (omitting methyl– stacking interactions, illus- ogy, during which we kept the soft restraints on the ter- trated in C). (D) Calculated free energy profiles along the – coordinate minal bases-pairs. Data gathering simulations were carried for the AT-case including all interactions or omitting interactions as in- out for 900-4000 ns. Coordinates were written out every dicated in panels A–C (indicated as different line colors or grey scales in the figure panel). For comparison the free energy profile for the AU case is 5000 steps. Using hydrogen-mass-repartitioning allowed us also shown (green curve). The free energies along the – coordinate were to use a time step of 4 fs. Details on the calculation of free calculated by Boltzmann-Inversion F =−k T · ln(p) and represent the av- energy profiles, calculations of errors and convergence and erage over all dinucleotide steps in the central segment of the AT (or AU) hydrogen bonding as well as conformational deformabilities sequence. The same data (enlarged) is shown in Supplementary Informa- are given in Supporting Information (Sections 1–5). tion Figure S1 including error bars. RESULTS AND DISCUSSION ulations (Figure 2 and Supplementary Information, Section Convergence, Figures S1–S6). DNA backbone substates The calculated populations (probabilities) for the canon- MD simulations were performed on seven DNA duplexes ical BI states for all investigated dinucleotide steps (Table 1) with different central sequences (Table 1) to record struc- emphasize two findings: First, sequences consisting only tural fluctuations including frequent transitions between of central A–T base pairs (AT,AA) show a higher popula- DNA backbone substates to sample substate populations. tion of BI states (by ∼10–15%) than other sequences. Sec- In a given dinucleotide step, a DNA backbone strand can ond, ‘elimination’ of a methyl group (by replacing central either adopt BI or BII configurations that are determined thymines by uracil or C5-methylated cytosine by cytosine) by the  and  dihedral angles (Figure 1): results in a lower occupation of BI states (by ∼5–8%, Ta- ble 1). Thus, the MD simulations indicate that the presence BI :  − ζ< 0 , BII :  − ζ>0(2) of methyl groups at the C5 of pyrimidines in DNA stabilizes the BI backbone states. Similar trends have been observed Transitions between BI and BII substates occur rapidly in other studies (23). relative to the total length of the simulations. Hence, it is In order to understand the molecular mechanism of this possible to directly extract probability distributions along stabilizing effect we performed MD simulations of the AT the  −  coordinate and to extract associated free energies case (central ATATA sequence, Table 1) by turning off non- on the time scale of 900–4000 ns used in data gathering sim- bonded interactions between the C5-methyl groups and se- Downloaded from https://academic.oup.com/nar/article-abstract/47/3/1132/5239025 by Ed 'DeepDyve' Gillespie user on 19 February 2019 Nucleic Acids Research, 2019, Vol. 47, No. 3 1135 Figure 3. Calculated free energy versus – coordinate for single base-pair steps in the AT-case. (A) Free energy simulation of AT-case including all interactions. (B) Simulations omitting thymine-methyl interactions with the C2 sugar atom (and connected hydrogens) of the 5 -neighboring nucleotide. (C) Snapshot of a BI conformation. (D) Normalized density plot of sampled sugar pucker phase of the 3 ring versus – coordinate (indicating low- phase preference in BI and high-phase preference for BII states). (E) Normalized density plot of sampled sugar puckering of 5 ring versus – coordinate (indicating high-phase preference in BI and low-phase preference for BII states). (F) Snapshot of BII conformation with arrows highlighting sugar pucker phase induced shifts. lected subsets of atoms in the system. During the simula- the change in sterical van der Waals interactions (see Sup- tions either all interactions of the C5-methyl group were plementary Information Figures S6–S11). turned off (representing an AU step) or only with the sol- vent, with the 5 neighboring base or with the C2 atom (and Why methyl–sugar clashes trigger BI promiscuity its hydrogens) of the 5 sugar. The associated free energies along the – coordinate were calculated by Boltzmann in- In order to better understand the effect of the steric in- version of the sampled probability distributions and repre- teractions between methyl group and 5 -sugar, we investi- sent the average over all steps in the central DNA sequence gated its influence on individual base-pair steps. Note, that (steps 6–9). Comparison of the AT versus AU cases indi- methyl–sugar clashes can only occur in base-pair steps with cates a negligible impact of the methyl group on the shape a thymine base at the 3 -position (e.g. in ApT but not in of the free energy curve in the BI subspace. However, the TpA steps). Interestingly, the population of BII states in BII regime is of significantly lower free energy resulting ApT steps is even more stabilized through the omission in an increased BII population. The exclusion of all inter- of methyl–sugar clashes than expected from the free en- actions between methyl groups and water molecules (Fig- ergy profiles obtained as averages over the central segments ure 2, purple curve) as well as the exclusion of all interac- (compare Figure 3 with Figure 2). Intriguingly, the omission tions between methyl groups and the 5 -neighboring bases of methyl-5 -sugar interactions leads to a remarkable desta- (Figure 2, yellow curve) also shows much smaller deviations bilization of BII states in the juxtaposed TpA steps. Based from the regular case (Figure 2, blue curve). However, ex- on the calculated free energies along the – coordinate we cluding interactions between C5-methyl groups and the C2 conclude that methyl–sugar clashes destabilize BII confor- atom (and its hydrogens) of the 5 -neighboring sugar results mations of ApT steps by ∼2kcal/mol but at the same time in a large drop of the free energy in the BII regime (Figure 2, the BII state of neighboring steps is stabilized (albeit to a red curve). Hence, neither methyl-base stacking nor interac- lesser degree of ∼0.5 kcal/mol for TpA steps). This finding tions with the solvent are decisive but the steric clashes be- reflects the anti-correlation of DNA base-pair steps, a phe- tween C5-methyl group and 5 -sugar are the main cause of nomenon which has also been subject of previous studies the BI stabilization by the C5-methyl group. It is interesting (18,36). to note that eliminating only the electrostatic interactions of Our simulations suggest a qualitative sterical explanation the methyl group with specific groups had only a small im- of this nearest-neighbor anti-correlation that is illustrated pact on the population of substates but it is dominated by in Figure 3C–F. For an ApT step in the BI configuration, the 3 sugar pucker (T-nucleotide) adopts preferably a lower Downloaded from https://academic.oup.com/nar/article-abstract/47/3/1132/5239025 by Ed 'DeepDyve' Gillespie user on 19 February 2019 1136 Nucleic Acids Research, 2019, Vol. 47, No. 3 Figure 4. Methyl–sugar clashes destabilize the BII subspace in an ApT step. (A) Center of mass distance of thymine methyl group from C2 atom of 5 neighboring sugar versus − coordinate in case of simulations including interactions of methyl groups with the C2 -atom of a 5 -sugar. BII states are populated very rarely. (B) Same as in (A) but switching-off interactions between methyl and sugar that allows for a closer approach of the groups. As a consequence, the BII subspace can also be accessed. (C) Snapshot of BI conformation, taken from simulation including all interactions. (D)Distanceof thymine methyl group vs. H6-O3 hydrogen bonding distance between thymine and 5 -sugar in case of simulations including all interactions (as in A). A close H6–O3 hydrogen bonding distance between thymine and 5 -sugar is not formed under these conditions. (E) Same as in D but interactions between methyl and sugar are switched-off, giving rise to sampling of close H6–O3 hydrogen bonding distances. (F) Snapshot of sampled BII conformation for simulation allowing clashes between methyl and sugar. The dashed red line indicates the unconventional hydrogen bond between thymine’s H6 and the backbone’s O3 atom. pucker phase than the 5 sugar (A nucleotide) to adopt a In a next step, we studied the influence of methyl–sugar well stacked configuration. A BII configuration, on the con- clashes for other sequences (Figure 5). Similar to the ApT trary, forces the 3 sugar to adopt preferably a higher pucker steps (see above), we also find that methyl–sugar clashes are phase that inclines the sugar ring in order to keep a near responsible for blocking H6-O3 hydrogen bonds for GpT, planar stacking geometry of adjacent bases (Figure 3). The CpT and GpC* steps (Supplementary Information Figures opposite is observed for the 5 -sugar. Since this latter sugar S12–S14). Note that we observed a different pattern for TpT adopts the role of a 3 sugar in the consecutive step and it has steps, where H6-O3 bonds are rare even when methyl–sugar been shifted to lower phase by the BII state, the population interactions are switched off. We rather find that H6-O5 of another BII step at this neighboring step is suppressed. hydrogen bonds are decisive for this sequence (Supplemen- The simulations also indicate a qualitative sterical mech- tary Information Figures S15 and S16). For simplicity, we considered the BI/BII population as average over the cen- anism for the BII destabilization (by the methyl group) tral DNA segment, though the same anti-correlation trend in case of the ApT steps as illustrated in Figure 4:The as before is obtained on the base-pair step level (illustrated methyl-5 -sugar sterical interaction locks the bases to a spe- in Supplementary Information Figures S17–S21). Similar cific conformational space in which the backbone prefer- entially adapts BI configurations. Switching off this steri- to the comparison of the AT and AU cases we find that cal hindrance allows both components to come closer to- methyl-5 -sugar clashes destabilize BII states for each inves- gether whereby also the BII space becomes accessible (com- tigated sequence. Indeed, the CT, GT and C*G (methylated pare Figure 4A and B). The population of these states is cytosine) cases show an even stronger increase in BII pop- then stabilized by unconventional hydrogen bonds between ulation upon omission of methyl–sugar interactions than thymine’s H6-atom and the O3 atom of the backbone sugar the AT-sequence (Figure 5). The sequence dependence of (Figure 4D and E). Notably, the existence and correlation to DNA backbone substates has also been studied experimen- the BII population of these unconventional hydrogen bonds tally. Based on NMR experiments, it has been shown that has already been pointed out by Balaceanu et al. (19), how- out of the ten dinucleotide steps, the four steps which con- ever, such hydrogen bonds are sterically possible only at spe- tain a thymine on the 3 position (ApT, GpT, TpT and CpT) cific base pair steps. clearly exhibit the lowest BII population (2). This confirms our results that methyl groups (specifically at the 3 posi- Downloaded from https://academic.oup.com/nar/article-abstract/47/3/1132/5239025 by Ed 'DeepDyve' Gillespie user on 19 February 2019 Nucleic Acids Research, 2019, Vol. 47, No. 3 1137 Figure 5. Calculated free energy profiles versus – coordinate for the AA, CT, GT and C*G sequences (blue lines correspond to simulations including all interactions whereas red lines indicate simulation results omitting interactions of the base methyl group and the 5 neighboring sugar C2 atom and connected hydrogens). tion) destabilize BII backbone states due to methyl–sugar the covariance-matrix of helical parameter fluctuations interactions. recorded during simulations (see Supplementary Informa- tion Section 5). The stiffness parameters were calculated with respect to the averages of the central four base pair Methyl–sugar clashes influence DNA’s global structure and steps. It is important to note, that the harmonic stiffness flexibility model is fully valid on this scale, as the steps superpose to The conformation of DNA’s backbone is strongly coupled single Gaussian distributions (indicating a quadratic under- to the configuration of the base-pairs ( 3,21,37,38). Indeed, lying effective free energy profile) whilst individual steps can the sterical methyl–sugar interactions also have an impact show clear bimodal behavior (Supplementary Information on the overall structure and flexibility of DNA in the present Figure S22). simulations. As relevant parameters we consider the mean Methyl–sugar repulsion is found to cause significant de- twist, stretching and bending of the central DNA segments. creases in twist-, stretch- and bending-stiffness making the Whereas changes in the equilibrium twist and stretch- DNA-molecule overall more flexible. methyl–sugar repul- ing are negligible, the intrinsic bending of the DNA double sion alone significantly decreases the bending-stiffness by helix is markedly increased by methyl–sugar repulsion for ∼40% for the methylated cytosine sequence (C*G) and by most sequences (Figure 6), e.g. for the methylated cytosine ∼20–30% for the GT and CT sequence. The largest changes sequence (C*G) an increase by ∼18% was observed. We cal- are found for the twist-stiffness of the CT and GT sequence, culated the DNA stiffnesses based on a harmonic stiffness where it has been found that methyl–sugar clashes soften by model (assuming an underlying quadratic free energy sur- ∼50–60%. Besides, also the stretching flexibility is enhanced face for twisting, bending and stretching) that has also been through methyl–sugar clashes, with the GT sequence show- used in previous studies (5,6,39–43), ing the largest effect (∼35%). −1 It should be emphasized that these findings do not nec- K = k T · C , (3) essarily indicate that DNA molecules containing methy- where k and T indicate Boltzmann constant and tem- lated cytosines are more flexible than sequences with un- perature, respectively. K denotes the stiffness- and C Downloaded from https://academic.oup.com/nar/article-abstract/47/3/1132/5239025 by Ed 'DeepDyve' Gillespie user on 19 February 2019 1138 Nucleic Acids Research, 2019, Vol. 47, No. 3 than GpC steps, while C*pG steps are overall more flexible than CpG steps. On the level of the whole central segment stretching and bending stiffnesses are slightly lower for the methylated case (C*G case) compared to the CG case (∼5– 7%), but the twisting-stiffness of the C*G case is signifi- cantly higher compared to the unmethylated central CG se- quence (∼20%, Supplementary Information Tables S1-S3). We also checked in how far the charge-reassignment alone of methylated-bases influenced changes in DNA’s structure and flexibility (Supplementary Information Figures S23– S24). Here, our simulations indicate, that such charge ef- fects are relevant for methylated cytosine sequences, but are negligible for canonical sequences. The effect of methyl–sugar clashes to increase flexibility is counterintuitive (restriction to BI substate), however, this is a direct consequence of the influence of the backbone: The twist-distribution within the BI states is broader than that of BII states (Figure 7A). Thus, an increased popula- tion of BII states results in a narrower overall twist distribu- Figure 6. Relative changes in structure and flexibility due to methyl–sugar tion (Figure 7B). The lower variance in twist directly reflects clashes. The first three columns indicate for each sequence the relative a higher stiffness with respect to this mode. Note, that from change in equilibrium twist, stretch and bending of the central segment the – free energy profiles one would expect that the twist- whereas the last three columns represent the change in flexibility (indicated as the change in the calculated stiffness constant). For the latter cases, red distribution of BI states is narrower, however, the backbone entries mean that methyl sugar clashes have a decreasing/softening effect population does not map linearly to the twist variable. In- and blue entries represent an increase of the stiffness. All changes are given deed, similar – configurations can show different twisting as relative to the reference case of reassignment of thymines’ and methy- (illustrated in Figure 7C and D). lated cytosines’ bases, i.e. the entries reveal the influence of the van der Waals interactions between methyl and sugar group. CONCLUSIONS methylated cytosine since we only evaluate the influence of The methyl group in thymine and in C5-methylated cy- methyl–sugar interactions by switching-off interactions be- tosine modulates the BI-BII substate distribution and the tween methyl and sugar group. It emphasizes that DNA’s lo- DNA flexibility. Since DNA methylation plays a key role cal and global deformability is strongly influenced by these in epigenetic regulation of gene expression (44–46), it is interactions. However, van-der-Waals interactions between likely that its influence on DNA deformability is also linked methyl groups and other chemical groups are still included. to its biological function (43,44,47,48). In the presented Previous studies have pointed out that methylated cytosine study, we addressed the correlation between methyl groups sequences are overall stiffer than their regular analogs (43). and BI/BII promiscuity using comparative MD simula- Based on comparative simulations (Supplementary Infor- tions. As a key technique we employed sets of simulations mation Section 7) we obtained the following trend: On the that specifically included or omitted non-bonded interac- level of base pair steps, GpC* steps are significantly stiffer tions between methyl groups and hypothetically important Figure 7. Methyl sugar clashes increase DNA flexibility at the central segment. ( A) Calculated twist distribution of a GpT step in case of simulations including methyl sugar interactions indicating the total cumulative distribution (both sampled BI and BII states) as well as distributions considering BI ◦ ◦ and BII states separately. The distribution at the BI states is shifted and broader (standard deviation  = 7.2 ) than that for BII states ( = 6.2 ). The standard deviation of the cumulative distribution amounts to  = 7.1 .(B) Twist distribution upon removing methyl–sugar interactions: The cumulative ◦ ◦ ◦ twist distribution is narrower since the population of BII is strongly increased ( = 5.7 ). (C) Snapshot of undertwisted GpT step (∼20 )at – −80 ◦ ◦ in Watson and Crick strand. (D) Snapshot of overtwisted GpT step (∼38.5 )at – −80 in both strands. Downloaded from https://academic.oup.com/nar/article-abstract/47/3/1132/5239025 by Ed 'DeepDyve' Gillespie user on 19 February 2019 Nucleic Acids Research, 2019, Vol. 47, No. 3 1139 partner groups. Based on these simulations, we showed that 3. Packer,M.J. and Hunter,C.A. (1998) Sequence-dependent DNA structure: the role of the sugar-phosphate backbone. J. Mol. Biol., the hydrophobicity of methyl groups as well as methyl- 280, 407–420. stacking that had been proposed to cause changes in the 4. Drsa ˇ ta,T., Per ´ ez,A., Orozco,M., Morozov,A.V., Sponer,J. and BI/BII ratio (24,25,27) exhibit only a small influence on the Lankas,F. (2013) Structure, stiffness and substates of the population of BI/BII substates. Switching-off interactions Dickerson-Drew dodecamer. J. Chem. Theory Comput., 9, 707–721. ˇ ˇ 5. Drsata,T. and Lankas,F. (2013) Theoretical models of DNA between methyl group and the C2 atom and its hydrogen flexibility. Wiley Interdiscip. Rev. Comput. Mol. Sci., 3, 355–363. atoms, in contrast, stabilized BII and hence decreased pop- 6. Liebl,K. and Zacharias,M. (2017) Unwinding induced melting of ulation of BI conformations significantly. This trend con- double-stranded DNA studied by free energy simulations. J. Phys. firms results of previous studies ( 23,26,28) and was found Chem. B, 121, 11019–11030. 7. Oguey,C., Foloppe,N. and Hartmann,B. (2011) Understanding the for each investigated sequence, albeit to different degrees: sequence-dependence of DNA groove dimensions: implications for The strongest changes appeared for sequences including DNA interactions. PLOS ONE, 5, 1–8. CpT and GpT steps or methylated cytosines. Previous stud- 8. R.,W.F., Christine,R. and Michael,T.R.L.K. (2005) M.TaqI facilitates ies indicate a decisive role of the formation of unconven- the base flipping via an unusual DNA backbone conformation. tional base-sugar H6–O3 hydrogen bonds (or H8–O3 in Biopolymers, 79, 128–138. 9. Robertson,J.C. and Cheatham,T.E. (2015) DNA backbone BI/BII case of purines) for stabilizing the BII state (19–21) that is distribution and dynamics in E2 protein-bound environment also found in the present simulations. However, since such determined by molecular dynamics simulations. J. Phys. Chem. B, bonds are in principle possible for every dinucleotide step 119, 14111–14119. the correlation alone does not explain the physical reason 10. Hegde,R.S., Grossman,S.R., Laimins,L.A. and Sigler,P.B. (1992) for the BI/BII sequence dependence. Based on sterical con- Crystal structure at 1.7 A of the bovine papillomavirus-1 E2 DMA-binding domain bound to its DNA target. Nature, 359, siderations it was possible to qualitatively explain the effect 505–512. of the methyl group on the BI/BII ratio. Interestingly, an 11. Djuranovic,D. and Hartmann,B. (2004) DNA fine structure and increased propensity of BII at one step results in a reduced dynamics in crystals and in solution: the impact of BI/BII backbone BII probability at a neighboring step that can be qualita- conformations. Biopolymers, 73, 356–368. 12. Djuranovic,D. and Hartmann,B. (2005) Molecular dynamics studies tively explained by a coupling to the nearest-neighbor sugar on free and bound targets of the bovine papillomavirus type I E2 pucker conformation. Given the pronounced impact of the Protein: the protein binding effect on DNA and the recognition interaction between methyl and sugar group it also influ- mechanism. Biophys. J., 89, 2542–2551. ences the global structure and flexibility of DNA ( 29–31). 13. Madhumalar,A. and Bansal,M. (2005) Sequence preference for While these interactions increase the intrinsic DNA bend- BI/BII conformations in DNA: MD and crystal structure data analysis. J. Biomol. Struct. Dyn., 23,13–27. ing, a decrease in the DNA’s stiffness was observed. For CpT 14. Heddi,B., Foloppe,N., Bouchemal,N., Hantz,E. and Hartmann,B. and GpT rich sequences, the steric clashing between methyl (2006) Quantification of DNA BI /BII backbone states in solution. and sugar group decreases DNA’s torsional rigidity by up Implications for DNA overall structure and recognition. J. Am. to ∼60%. We found as a main sterical reason that the twist Chem. Soc., 128, 9170–9177. 15. Pasi,M., Maddocks,J.H., Beveridge,D., Bishop,T.C., Case,D.A., distribution of states within the BI basin is broader than in Cheatham,T. III, Dans,P.D., Jayaram,B., Lankas,F., Laughton,C. case of sampled states within the BII regime. Consequently, et al. (2014) ABC: A systematic microsecond molecular dynamics since the methyl-5 -sugar interactions stabilize the BI state study of tetranucleotide sequence effects in B-DNA. Nucleic Acids the twist flexibility increases. The comparative simulation Res., 42, 12272–12283. 16. Ben Imeddourene,A., Elbahnsi,A., Guroult,M., Oguey,C., methodology of including or omitting specific non-bonded Foloppe,N. and Hartmann,B. (2015) Simulations meet experiment to interactions could be applied in future studies on other phe- reveal new insights into DNA intrinsic mechanics. PLoS Comput. nomena like phosphate repulsion or the role of hydrogen Biol., 11, 1–27. bonding in structure formation could be tackled. 17. Svozil,D., Kalina,J., Omelka,M. and Schneider,B. (2008) DNA conformations and their sequence preferences. Nucleic Acids Res., 36, 3690–3706. 18. Zgarbova,M., ´ Jurecka,P ˇ ., Lankas,F ˇ ., Cheatham,T.E., Sponer,J. and SUPPLEMENTARY DATA Otyepka,M. (2017) Influence of BII backbone substates on DNA Supplementary Data are available at NAR Online. Twist: A unified view and comparison of simulation and experiment for all 136 distinct tetranucleotide sequences. J. Chem. Inf. Model., 57, 275–287. 19. Balaceanu,A., Pasi,M., Dans,P.D., Hospital,A., Lavery,R. and FUNDING Orozco,M. (2017) The role of unconventional hydrogen bonds in determining BII propensities in B-DNA. J. Phys. Chem. Lett., 8, Deutsche Forschungsgemeinschaft (DFG) [SFB749, 21–28. project C05]; The Leibniz super computer (LRZ) center 20. Dans,P.D., Faustino,I., Battistini,F., Zakrzewska,K., Lavery,R. and is acknowledged for providing supercomputer access Orozco,M. (2014) Unraveling the sequence-dependent polymorphic [grant pr48ko]. Funding for open access charge: Deutsche behavior of d (CpG) steps in B-DNA. Nucleic acids Res., 42, 11304–11320. Forschungsgemeinschaft. ˜ ˇ ˇ 21. Dans,P.D., Danilane,L., Ivani,I., Drsata,T., Lankas,F., Walther,J., Conflict of interest statement. None declared. Pujagut,R.I., Battistini,F., Gelpı,J.L., Lavery,R. et al. (2016) Long-timescale dynamics of the Drew–Dickerson dodecamer. Nucleic Acids Res., 44, 4052–4066. REFERENCES 22. Temiz,N.A., Donohue,D.E., Bacolla,A., Luke,B.T. and Collins,J.R. (2012) The role of methylation in the intrinsic dynamics of B- and 1. Hartmann,B., Piazzola,D. and Lavery,R. (1993) BI-BII transitions in Z-DNA. PLoS ONE, 7, 1–9. B-DNA. Nucleic Acids Res., 21, 561–568. 23. Peguero-Tejada,A. and van der Vaart,A. (2017) Biasing simulations 2. Heddi,B., Oguey,C., Lavelle,C., Foloppe,N. and Hartmann,B. (2009) of DNA base pair parameters with application to propellor twisting Intrinsic flexibility of B-DNA: the experimental TRX scale. Nucleic Acids Res., 38, 1034–1047. Downloaded from https://academic.oup.com/nar/article-abstract/47/3/1132/5239025 by Ed 'DeepDyve' Gillespie user on 19 February 2019 1140 Nucleic Acids Research, 2019, Vol. 47, No. 3 in AT/AT, AA/TT, and AC/GT steps and their uracil analogs. J. 36. Lankas,F ˇ ., Sponer,J., Langowski,J. and Cheatham,T.E. (2003) DNA Chem. Inf. Model., 57, 85–92. basepair step deformability inferred from molecular dynamics 24. Rauch,C., Trieb,M., Wellenzohn,B., Loferer,M., Voegele,A., simulations. Biophys. J., 85, 2872–2883. Wibowo,F.R. and Liedl,K.R. (2003) C5-Methylation of cytosine in 37. Keepers,J.W., Kollman,P.A., Weiner,P.K. and James,T.L. (1982) B-DNA thermodynamically and kinetically stabilizes BI. J. Am. Molecular mechanical studies of DNA flexibility: coupled backbone Chem. Soc., 125, 14990–14991. torsion angles and base-pair openings. Proc. Natl. Acad. Sci. U.S.A., ˇ ˇ ¨ ¨ 25. Warmlander,S., Sponer,J.E., Sponer,J. and Leijon,M. (2002) The 79, 5537–5541. influence of the thymine C5 methyl group on spontaneous base pair 38. Okonogi,T.M., Alley,S.C., Harwood,E.A., Hopkins,P.B. and breathing in DNA. J. Biol. Chem., 277, 28491–28497. Robinson,B.H. (2002) Phosphate backbone neutralization increases 26. Teng,X. and Hwang,W. (2018) Effect of methylation on local duplex DNA flexibility: a model for protein binding. Proc. Natl. mechanics and hydration structure of DNA. Biophys. J., 114, Acad. Sci. U.S.A., 99, 4156–4160. 1791–1803. 39. Olson,W.K., Gorin,A.A., Lu,X.-J., Hock,L.M. and Zhurkin,V.B. 27. Wibowo,F.R., Trieb,M., Rauch,C., Wellenzohn,B. and Liedl,K.R. (1998) DNA sequence-dependent deformability deduced from (2005) The N6-Methyl group of adenine further increases the BI protein-DNA crystal complexes. Proc. Natl. Acad. Sci. U.S.A., 95, stability of DNA compared to C5-Methyl groups. J. Phys. Chem. B, 11163–11168. 109, 557–564. 40. Drsa ˇ ta,T., Kara,M., Zacharias,M. and Lankas,F ˇ . (2013) Effect of 28. Hunter,C.A. (1993) Sequence-dependent DNA Structure: The role of 8-Oxoguanine on DNA structure and deformability. J. Phys. Chem. base stacking interactions. J. Mol. Biol., 230, 1025–1054. B., 117, 11617–11622. ˇ ˇ 29. Parker,T.M., Hohenstein,E.G., Parrish,R.M., Hud,N.V. and 41. Drsa ˇ ta,T., Spack ˇ ova,N ´ ., Jurecka,P ˇ ., Zgarbova,M., ´ Sponer,J. and Sherrill,C.D. (2013) Quantum-Mechanical analysis of the energetic Lankas,F ˇ . (2014) Mechanical properties of symmetric and contributions to  Stacking in Nucleic Acids versus Rise, Twist, and asymmetric DNA A-tracts: implications for looping and nucleosome Slide. J. Am. Chem. Soc., 135, 1306–1316. positioning. Nucleic Acids Res., 42, 7383–7394. 30. Svozil,D., Hobza,P. and Sponer,J. (2010) Comparison of intrinsic 42. Liebl,K., Drsa ˇ ta,T., Lankas,F ˇ ., Lipfert,J. and Zacharias,M. (2015) stacking energies of ten unique dinucleotide steps in A-RNA and Explaining the striking difference in Twist-Stretch coupling between B-DNA duplexes. Can we determine correct order of stability by DNA and RNA: A comparative molecular dynamics analysis. Quantum-Chemical calculations? J. Phys. Chem. B, 114, 1191–1203. Nucleic Acids Res., 43, 10143–10156. 31. Umezawa,Y. and Nishio,M. (2002) Thymine-methyl/ interaction ´ 43. Perez,A., Castellazzi,C., Battistini,F., Collinet,K., Flores,O., implicated in the sequence-dependent deformability of DNA. Nucleic Deniz,O., Ruiz,M., Torrents,D., Eritja,R., Soler-Lpez,M. et al. (2012) Acids Res., 30, 2183–2192. Impact of methylation on the physical properties of DNA. Biophys. 32. Per ´ ez,A., Castellazzi,C.L., Battistini,F., Collinet,K., Flores,O., J., 102, 2140–2148. Deniz,O., Ruiz,M.L., Torrents,D., Eritja,R., Soler-Lopez,M. ´ et al. 44. Pennings,S., Allan,J. and Davey,C.S. (2005) DNA methylation, (2012) Impact of methylation on the physical properties of DNA. nucleosome formation and positioning. Brief. Funct. Genomics, 3, Biophys. J., 102, 2140–2148. 351–361. 33. Case,D., an D.S.Cerutti,R.B., T.E. Cheatham,I., Darden,T., Duke,R., 45. Jones,P.A. and Takai,D. (2001) The role of DNA methylation in Giese,T., Gohlke,H., Goetz,A., Homeyer,N., Izadi,S. et al. (2016) mammalian epigenetics. Science, 293, 1068–1070. Amber 16. University of California, San Francisco. 46. Laird,P.W. and Jaenisch,R. (1996) The role Of DNA methylation in 34. Ivani,I., Dans,P.D., Noy,A., Prez,A., Faustino,I., Hospital,A., cancer genetics aand epigenetics. Annu. Rev. Genet., 30, 441–464. Walther,J., Andrio,P., Goi,R., Balaceanu,A. et al. (2015) 47. Ngo,T.T., Yoo,J., Dai,Q., Zhang,Q., He,C., Aksimentiev,A. and PARMBSC1: A refined force-field for DNA simulations. Nature Ha,T. (2016) Effects of cytosine modifications on DNA flexibility and methods, 13, 55–58. nucleosome mechanical stability. Nat. Commun., 7, 10813. 35. Jorgensen,W.L., Chandrasekhar,J., Madura,J.D., Impey,R.W. and 48. Severin,P.M.D., Zou,X., Gaub,H.E. and Schulten,K. (2011) Cytosine Klein,M.L. (1983) Comparison of simple potential functions for methylation alters DNA mechanical properties. Nucleic Acids Res., simulating liquid water. J. Chem. Phys., 79, 926–935. 39, 8740–8751.

Journal

Nucleic Acids Research – Oxford University Press

Published: Feb 20, 2019

There are no references for this article.