Computational protocol: A Comprehensive Model for the Recognition of Human Telomeres by TRF1

Similar protocols

Protocol publication

[…] MD calculations were performed on the TRF1–TAGGGTTA complex using Protein Data Bank file 1W0T as a starting point. The crystal structure is composed of a 19-nucleotide section of DNA with two TRF1 proteins bound to adjacent binding sites. Models were constructed for both wild-type DNA and two mutations: G6C and T8A (G6C: TTAGGG → TTAGGC, T8A: TTAGGG → TAAGGG). Mutation choices were based on apparent discrepancies between the crystal structure and mutagenesis work: G6C resulted in dramatic loss of binding activity and T8A at T8 reportedly had no effect. In both cases, the opposite effect was predicted by the crystal structure while NMR did not adequately explain such a severe drop-off in affinity. The models were prepared for MD simulation using the WHATIF web interface to build in any missing atoms and identify protonation states. They were then explicitly solvated (approximately 11,500 waters per model) in a 10-nm box of TIP3P water using TLEAP in AMBER 10. Sodium counterions were added for overall charge neutrality, and periodic boundary conditions were applied with the following box dimensions: 63 nm × 66 nm × 71 nm. Bonds to hydrogen were constrained using SHAKE to permit a 2-fs time step, and the particle mesh Ewald algorithm was used to treat long-range electrostatic interactions. The non-bonded cutoff was set at 12.0 Å. Systems were energy minimised using a combination of steepest descent and conjugate gradient methods. MD calculations were carried out with the PMEMD module of AMBER 10 in conjunction with the FF99 Barcelona force field, which is specifically customised for nucleic acids. The FF99 Stony Brook force field was used for the protein. Each system was equilibrated and heated over 100 ps to 300 K and positional restraints were gradually removed. A Berendsen thermostat and barostat was used throughout for both temperature and pressure regulation. Production MD runs of 3 × 50 ns replicates for each system were obtained from randomised starting velocities. Since each simulation features two independent copies of the protein in complex with its cognate (or mutated) DNA sequence (which we refer to as site A and site B), in this way, a total of 300 ns of conformational space exploration was obtained for each mutation. Exact protocols for the minimisations, equilibrations, heating, and production runs are delineated in the Supplementary Material. During calculations, a snapshot was saved every 2 ps. RMSDs (a) and PCA (b) were used to assess the equilibration and reproducibility of each replicate by employing an in-house code: PCAZIP. Contour map outlines were generated from two-dimensional histograms of the first two PCA projections using matplotlib. Subdivided into a 50 × 50 grid, the outline is drawn around all grid bins that have a data point occupancy ≥ 0.025% of the total points. PCA revealed instability in two of the six binding-site replicates and these were discarded before further analysis. RMS clustering of the trajectory frames was carried out using the MMTSB toolset with the radius set to 2 Å and maxerr set to 1. Bonds were identified and quantified using PTRAJ in AMBER together with VMD. Hydrogen bond length cutoff was set at 3.5 Å including water-mediated contacts, and groups participating in hydrophobic effects were defined as non-polar, non-bonded contacts < 5 Å. DNA helical parameters were derived using Curves +. PTRAJ was also used to map hydration density using the “Grid” function and the protocol for this is described in the Supplementary Material. Structural alignments were performed and RMSDs were calculated using PyMOL from selected structures representing the most highly populated clusters. […]

Pipeline specifications

Software tools WHATIF, AMBER, matplotlib, VMD, PyMOL
Applications Miscellaneous, Protein structure analysis
Organisms Homo sapiens