Balanced Solvent Model for Intrinsically Disordered and Ordered Proteins

Abstract

Intrinsically disordered proteins (IDPs) have no fixed three-dimensional (3D) structures under physiological conditions, with the content being about 51% in human proteomics. IDPs are associated with many human diseases, such as cancer, diabetes, and neurodegenerative diseases. Because IDPs do not crystallize and have diverse conformers, traditional experimental methods such as crystallization and NMR can hardly capture their conformation ensemble and just provide average structural characters of IDPs. Therefore, molecular dynamics (MD) simulations become a valuable complement to the experimental data. However, the accuracy of molecular dynamics simulation for IDPs depends on the combination of force fields and solvent models. Recently, we released an environment-specific force field (ESFF1) for IDPs, which can well reproduce the local structural properties (such as J-coupling and secondary chemical shifts). However, there is still a large deviation for the radius of gyration (R g). Therefore, a solvent model combined with ESFF1 is necessary to capture the local and global characters for IDPs and ordered proteins. Here, we investigated the underestimation or overestimation of the solvent interaction for four solvent models (TIP3P, TIP4P-Ew, TIP4P-D, OPC) under ESFF1 and found the important ε parameter of the solvent model to play a key role in scaling R g. A near-linear relationship between the simulation R g and the ε parameter was used to develop the new solvent model, named TIP4P-B. The results indicate that the simulated R g with TIP4P-B is in better agreement with the experimental observations than the other four solvent models. Simultaneously, TIP4P-B can also maintain the advantages of the ESFF1 force field for the local structural properties. Additionally, TIP4P-B can successfully sample the conformation of ordered proteins. These findings confirm that TIP4P-B is a balanced solvent model and can improve sampling R g performance for folded proteins and IDPs.

RNA-Specific Force Field Optimization with CMAP and Reweighting

“…Adjusting the energy of one pair of torsions alone might not be sufficient to improve the accuracy of the RNA force field. 33 In future, it is necessary to further test RNA folding and kink-turn motif, adjust the parameters of the χ angles, optimize the solvent model, 65 and use enhanced sampling methods to improve the performance of the force field.…”

Quantitative prediction of ensemble dynamics, shapes and contact propensities of intrinsically disordered proteins

“…An alternative and more principled approach is to improve the MD force fields themselves enabling them to increasingly accurately predict experimental data in a way that is fully transferrable between protein systems, both ordered and disordered. This premise has led to a recent proliferation of protein force field developments [32][33][34][35][36][37] and new explicit water models [38][39][40] specifically geared toward the improved representation of disordered proteins. In a significant development, residue-specific force fields have been introduced.…”

Enhancing Conformational Sampling for Intrinsically Disordered and Ordered Proteins by Variational Autoencoder

“…We split 50,000 conformations of each protein into two parts, the first part (e.g., the first 10% of the conformations) as a training set and the latter as a test set for model training. Such a splitting approach is based on the feature of MD simulation that the first part often contains more extended and diverse conformations, while the latter part contains relatively compact ones [ 13 ]. By training the model with the first part, the model could learn enough diverse conformations, and a similar training process can be accomplished on a rather short MD trajectory.…”

“…The detailed simulation conditions for these IDPs and structured proteins are listed in Table 2 , which are consistent with corresponding experimental conditions. All simulations are conducted in AMBER18 [ 28 ] with ESFF1 force field [ 29 ] for protein and TIP3P [ 30 ] solvent model, whose initial structures are the same as those in the work of Mu et al [ 13 ]. As the calculated radius of gyration for MD trajectories agreed with experimental values, these test systems were converged and correctly captured the features of disordered proteins [ 13 , 15 ].…”