259595 Forcefield Ptm: Development and Testing of a First Generation AMBER Forcefield for Post-Translational Modifications

Wednesday, October 31, 2012: 9:06 AM
Washington (Westin )
George A. Khoury, Jeff P. Thompson and Christodoulos A. Floudas, Chemical and Biological Engineering, Princeton University, Princeton, NJ

Currently the Protein Data Bank (PDB) [1] contains over 80,000 structures. Over 21,000 structures contain at least one modification, of which ~6,000 contain modifications other than disulfide bridges. There is no general forcefield that can accurately simulate how these ubiquitous chemical alterations affect the micro-environments of the proteins they reside in as well as their interactions with each other.

Therefore, in this work we develop and validate a consistent set of AMBER [2, 3] forcefield parameters for the 27 most-common post-translational modifications [4] (FFPTM). Calculations consistent with the AMBER forcefield were derived using the procedure of Duan et al. [5]. A two-stage RESP-fitting procedure [6] using α-helical and β-strand conformations of dipeptides (ACE-XXX-NME) was used to fit charges to reproduce the quantum mechanically calculated electrostatic potential (ESP) for each PTM. Utilizing Antechamber [7] for atom-type perception, the minimum set of missing Van der Waals, bond, angle, and dihedral parameters were added for each PTM from the General Amber Forcefield (GAFF) [8] to maintain consistency and to minimize the number of new parameters added to define each new modified amino acid type.

The parameters derived in FFPTM were validated using over 10 CPU years of stress-testing via all-atom Langevin molecular dynamics simulations in AMBER for 22 pairs of modified/unmodified systems. Unrestrained local minimizations starting from the initial PDB coordinates for each of the modified proteins yielded an average Cα RMSD of 0.410±0.173Å, comparable to the average Cα RMSD of 0.421±0.163Å for the unmodified structures. For each system, hundreds of local minimizations were performed starting from random initial points created along the MD trajectory. The forcefield parameters yielded an R2 = 0.880 between the average Cα RMSD of the modified structure and its initial point compared to the unmodified structure and its corresponding initial point. Further validation was done via analysis of secondary structure preservation, stability in energies, as well as correlations between the between Cα RMSDs and energetics during the trajectories from the unmodified to the modified structures. Although the initial structures were dissimilar for about 1/5 of the pairs, the average energies during the modified/unmodified structure simulations correlated with R2 = 0.997, whereas average Cα RMSDs to their corresponding minimized PDB structures were correlated with R2 = 0.725, indicating the new parameters did not largely perturb the energetics while the structures dynamically evolved according to the equations of motion. We extend the analysis to validate multiple modifications on a single structure. All modifications parameterized were conformed to the PDB atom-naming convention and were tested using starting coordinates contained in the PDB for each modification for ease of user portability. Insights for applications of the forcefield for blind structure prediction of proteins containing PTMs as well as de novo protein design will also be discussed. Further insights regarding structural implications of post-translational modifications observed as well as PTM Density will be presented.


1.         Berman, H.M., et al., The Protein Data Bank. Nucleic Acids Res, 2000. 28(1): p. 235-42.

2.         Weiner, P.K. and P.A. Kollman, AMBER: Assisted model building with energy refinement. A general program for modeling molecules and their interactions. Journal of Computational Chemistry, 1981. 2(3): p. 287-303.

3.         Cornell, W.D., et al., A Second Generation Force Field for the Simulation of Proteins, Nucleic Acids, and Organic Molecules. Journal of the American Chemical Society, 1995. 117(19): p. 5179-5197.

4.         Khoury, G.A., R.C. Baliban, and C.A. Floudas, Proteome-wide post-translational modification statistics: frequency analysis and curation of the swiss-prot database. Sci. Rep., 2011. 1.

5.         Duan, Y., et al., A point-charge force field for molecular mechanics simulations of proteins based on condensed-phase quantum mechanical calculations. Journal of Computational Chemistry, 2003. 24(16): p. 1999-2012.

6.         Bayly, C.I., et al., A well-behaved electrostatic potential based method using charge restraints for deriving atomic charges: the RESP model. The Journal of Physical Chemistry, 1993. 97(40): p. 10269-10280.

7.         Wang, J., et al., Automatic atom type and bond type perception in molecular mechanical calculations. Journal of Molecular Graphics and Modelling, 2006. 25(2): p. 247-260.

8.         Wang, J., et al., Development and testing of a general amber force field. Journal of Computational Chemistry, 2004. 25(9): p. 1157-1174.





Extended Abstract: File Not Uploaded