PMC

Outline of the Rosetta modeling protocol. This flowchart summarizes the complete protocol for docking small-molecule ligands into comparative models using Rosetta 3.4.

Building loops in comparative models of T4 lysozyme. Loops were rebuilt in comparative models of T4-lysozyme using P22 lysozyme as a template, as detailed in Steps 1–13 of the protocol. (a) The RMSD of C^α atoms between 10,000 models and the native protein (PDB ID: 2ou0) was computed over the full protein (black) and the core residues of T4 lysozyme (gray). The top 10% of models by Rosetta energy are shown here. Generally, a low Rosetta energy correlates with a low RMSD. For comparison, the Rosetta energy for the energy-minimized native crystal structure is shown in red. (b) Five of the lowest-energy models are seen in comparison with the native structure (shown in gray).

Docking MR3 into comparative models of T4 lysozyme. The MR3 ligand was docked into the ten lowest-energy comparative models of T4 lysozyme, as detailed in Steps 17–22 of the protocol. (a) 10,000 binding modes were clustered by RMSD using applications available in the bcl::Commons. The largest five clusters are shown, with the interface_delta score plotted against the RMSD to the native ligand-binding mode (shown in black). Generally, the largest clusters are also those with the lowest RMSD to the native binding mode. (b) The RMSD between 10,000 binding modes and the native binding mode (shown in red) was computed. The top ten percent of models by interface_delta score are shown here. Sub-angstrom binding modes are within the top ten percent of models, but Rosetta also identifies an alternative lower-energy binding mode within the site. (c) The lowest RMSD binding mode (orange) is closer to the native binding mode (gray) compared with the lowest-energy binding mode of the largest cluster (magenta) and the lowest-energy binding mode overall (cyan).

An overview of Rosetta energetic minimization and all-atom refinement via the relax protocol. (a) Simplified energy landscape of a protein structure. The relax protocol combines small backbone perturbations with side-chain repacking. The coupling of Monte Carlo sampling with the Metropolis selection criterion36 allows for sampling of diverse conformations on the energy landscape. The final step is a gradient-based minimization of all torsion angles to move the model into the closest local energy minimum. (b) Comparison of structural perturbations introduced by the repack and minimization steps. During repacking, the backbone of the input model is fixed, whereas side-chain conformations from the rotamer library33 are sampled. Comparison of the initial (transparent yellow) and final (light blue) models reveals conservation of the R135 rotamer but changes to the R11 and E15 rotamers. Minimization affects all angles and changes the backbone conformation.

Criterion for selecting regions for de novo loop building. (a) The target sequence is threaded over the template backbone; the initial structure is shown in beige. There are 12 amino acids from the target sequence that do not have a corresponding amino acid from the template sequence (amino acids 44–55). The resulting alignment produces an insertion into the backbone of the template structure. To rebuild missing density, two anchor points, N- and C-terminal from the missing region, are chosen to remain fixed. The flanking amino acids of the areas of missing density (K43 and G56, highlighted in red) are chosen as the initial anchor points. Rosetta will perform de novo loop building in the area of missing density. (b) The two anchor points are repositioned, allowing enough space to rebuild the 12 amino acids. In addition to the 12-residue insertion, the region highlighted in red will be rebuilt with the de novo loop modeling protocol. (c) During de novo loop rebuilding, secondary structure is also taken into consideration. Target residues 39–50 and 31–33 are both predicted to have secondary structural elements, but the template sequence does not contain secondary structural elements at these positions. Therefore, the loop to be built is extended to include residues 39–50 and 31–33. The final anchor points G28 and K60 are chosen, allowing 31 amino acids to be rebuilt (shown in red).