Supporting Information
Immersive Virtual Reality in Computational Chemistry: applications to the analysis of QM and MM data Andrea Salvadori, Gianluca Del Frate, Marco Pagliai, Giordano Mancini* and Vincenzo Barone Scuola Normale Superiore, Piazza dei Cavalieri 7, 56125, Pisa, Italy *Correspondence to:
[email protected]
1 System setup and simulation of DOX-DNA Simulations were carried out with Gromacs 4.6.5.[1] The PDB crystallographic structure 1D11[2] containing one daunomycin molecule and a single palindromic DNA sequence of six nucleotides, was chosen as starting system. Daunomycin was modified adding a hydroxil group on the methyl-ketone side chain using GaussView [3] to obtain doxorubicin (DOX). Considering its palindromic sequence, the nucleic acid was duplicated, rotated of approximately 180° and moved with the aim of reproducing the H-bond pattern between complementary nucleobases. The so-built double-stranded system was protonated and immersed in a cubic box of roughly 16500 TIP3P [4] water molecules, adding 0.1 NaCl to achieve neutrality. The final system was composed by 50160 atoms put in a rectangular box of 493 nm3. DNA was described using the AMBER99SB-ILDN force field. [5] DOX topology was taken from a previous work.[6] To ensure the stability of the terminal base pair in the intercalation site we added additional weak harmonic bonds [7] between the CG base pair above DOX; such bonds (of 50 kJ mol −1 each) were modeled in order to resemble the hydrogen bond coupling between the two nucleobases. Long range electrostatic interactions were accounted by means of the Particle Mesh Ewald method (PME). [8] A cut-off of 14 Å was used for short range electrostatic Van der Waals interactions. LINCS [9] was used to constrain bond lengths and angles. The system was initially minimized by means of the steepest descent algorithm. Relaxation of solvent molecules and counter ions to 300 K was initially performed keeping solute atoms restrained to their initial positions with a force constant of 1000 kJ mol−1 nm−2 ), for 5.0 ns in a NPT (using the Parrinello-Rahman barostat [10]) ensemble and using an integration time step of 2.0 fs. Then, the system was carried again to 0 K and progressively heated to 300 K in steps of 50 K. Starting from the last conformation of the equilibration step, DOX was pulled away from the intercalation site by the application of a harmonic potential of 125 kJ mol−1 nm−2 between the centers of mass (COMs) of DOX and of the four nucleobases delimiting the intercalation site. A pull rate of 10 nm ns −1 was used. The pulling process was allowed in each dimension. When a distance of 17.5 Å was reached, the DOX molecule was assumed to be completely separated from the DNA bundle. Along this
1
path, system configurations were taken every 0.6 Å of COMs separation and used as starting point for the umbrella sampling, for a total of 25 simulation windows. Afterwards, a 200 ps equilibration run (at 300 K and 1 bar pressure) was performed for each selected window, according to the protocol reported by Lekmul et al. [11] In each window 10 ns of MD was performed, for a total of 250 ns. Instantaneous atomic forces along the umbrella sampling trajectories, together with the mutual displacements along the reaction coordinate, were stored every 2 ps. Results were analyzed using the WHAM [12] method to compute the potential of mean force (PMF) profile along the predefined reaction coordinate.
Figure 1: Comparison of the chemical structure of the daunomycin and doxorubicin compounds and starting crystal structure of 1D11. (a) Scheme of the daunomycin and doxorubixin compounds; the hydrophobic anthraquinone moiety is shown in black while the hydrophilic aminosugar part is in blue. (b) Top view doxorubicin drug, shown using a ball and stick representation with standard colours and omitting apolar hydrogen atoms. The position of the center of mass (COM), used in the umbrella sampling, is shown as a cyan atom relative to the center of ring B (see dashed line). (c-d) Top and side view of the starting structure (1D11) after the modeling of doxorubicin drug; the drug is intercalated between the two CG pairs. DNA backbone is shown either blue (5-3) or red (3-5), with the ending part of the ribbon thinner (C3) or thicker (C5); nucleotides are depicted using filled polygons omitting the detailed atomic structure.
2
References [1] Sander Pronk, Szilrd Pll, Roland Schulz, Per Larsson, Pr Bjelkmar, Rossen Apostolov, Michael R. Shirts, Jeremy C. Smith, Peter M. Kasson, David van der Spoel, Berk Hess, and Erik Lindahl. GROMACS 4.5: a high-throughput and highly parallel open source molecular simulation toolkit. Bioinformatics, 29(7):845–854, 2013. [2] Andrew H. J. Wang, Giovanni Ughetto, Gary J. Quigley, and Alexander Rich. Interactions between an anthracycline antibiotic and DNA: molecular structure of daunomycin complexed to d(CpGpTpApCpG) at 1.2-Å resolution. Biochemistry, 26(4):1152–1163, 1987. [3] Gaussian. GaussView Version 5. http://www.gaussian.com/g_prod/gv5.htm. [4] William L. Jorgensen, Jayaraman Chandrasekhar, Jeffry D. Madura, Roger W. Impey, and Michael L. Klein. Comparison of simple potential functions for simulating liquid water. The Journal of Chemical Physics, 79(2), 1983. [5] Kresten Lindorff-Larsen, Stefano Piana, Kim Palmo, Paul Maragakis, John L. Klepeis, Ron O. Dror and David E. Shaw. Improved side-chain torsion potentials for the Amber ff99SB protein force field. Proteins: Structure, Function, and Bioinformatics, 78(8):1950–1958, 2010. [6] Marta Olszowka, Rosario Russo, Giordano Mancini, and Chiara Cappelli. A computational approach to the resonance Raman spectrum of doxorubicin in aqueous solution. Theoretical Chemistry Accounts, 135(2):1–15, 2016. [7] Giordano Mancini, Ilda D’Annessa, Andrea Coletta, Giovanni Chillemi, Yves Pommier, Mark Cushman, and Alessandro Desideri. Binding of an Indenoisoquinoline to the Topoisomerase-DNA complex induces reduction of linker mobility and strengthening of protein-DNA interaction. PLoS ONE, 7(12):1–10, 12 2012. [8] Tom Darden, Lalith Perera, Leping Li, and Lee Pedersen. New tricks for modelers from the crystallography toolkit: the particle mesh Ewald algorithm and its use in nucleic acid simulations. Structure, 7(3):R55 – R60, 1999. [9] Berk Hess, Henk Bekker, Herman J. C. Berendsen, and Johannes G. E. M. Fraaije. LINCS: A linear constraint solver for molecular simulations. Journal of Computational Chemistry, 18(12):1463–1472, 1997. [10] M. Parrinello and A. Rahman. Polymorphic transitions in single crystals: A new molecular dynamics method. Journal of Applied Physics, 52(12), 1981. [11] Justin A. Lemkul and David R. Bevan. Assessing the stability of Alzheimers amyloid protofibrils using molecular dynamics. The Journal of Physical Chemistry B, 114(4):1652–660, 2010. [12] Shankar Kumar, John M. Rosenberg, Djamal Bouzida, Robert H. Swendsen, and Peter A. Kollman. The weighted histogram analysis method for free-energy calculations on biomolecules. I. The method. Journal of Computational Chemistry, 13(8):1011–1021, 1992.
3
