Page - 47 - in Charge Transport in DNA - Insights from Simulations

3.4MicrohydratedDNA Table3.3: Number ofwatermolecules in themicrohydrated systems. system watersper A5 A9 A13 ratio to full phosphate basepair G5 G9 G13 hydration Dry1 24 ca. 43 384 576 768 1/10 Dry2 12 ca. 21 192 288 384 1/20 Dry3 6 ca. 11 96 144 192 1/40 Dry4 3 ca. 5 48 72 96 1/80 To keep the water molecules nearest to the backbone, the distance between the O-atoms of thewatermolecules and the P-atoms of the backbonewasmeasured andthenearestwatermoleculeswerekept. Thisway, fourdifferentmicrohydration stateswere created. These obtained systems contained from 24 down to 3water moleculesperphosphate. SeeTable3.3 foranoverviewof thesystemsnamingand theamountofwater left in the simulationbox. This corresponded to the removal of ca. 9/10, 19/20, 39/40 and 79/80 of water fromthefullyhydratedsystem,respectively. TheappearanceoftheDNAhydration shells constructed in thiswaycanbe inferred from3.6. These microhydrated systems in vacuo were simulated in an NVT ensemble. Pi- lot simulations showed some unwanted effects when using cut-offs to deal with the electrostatics in the simulation of these clusters. Since Gromacs is not able to calculate Ewald inmulti-threaded environment, periodic boundary conditions were applied tomakeuseof theparticle-meshEwaldmethod. Toavoidundesired boundary effects the box sizewas increased here to (11,5 nm)3. To obtain stable MDsimulations of thesemolecular clusters, itwas necessary to decrease the time stepto1 fs. Again, theparm99/BSC0 forcefieldwasused, inspiteof the fact that it hadbeendeveloped for simulations in condensedphase. In theparameterization, the atomic charges are overestimatedby 15–20%systematically, compensating for themissingdescriptionof electronic polarization effects. Thismaybe regardedas anerror. However, the aim is to characterize effects of robustnature, andso this is most likelynotan issue. Indeed, itwasshownthat the interactionsbetweennucleic acidbuildingblocks aredescribedwell.[120,121] 47