A computational study on a globular protein and an intrinsically disordered protein
Abstract
We present molecular dynamics (MD) simulations of two protein targets for drug design: Triosephosphate isomerase (TIM) and Methyl CpG binding protein 2 (MeCP2). First, we studied three TIM proteins: TcTIM, TbTIM and a chimeric protein (Mut1). The first two are homologous enzymes with high sequence similarity, albeit different biophysical parameters. The chimeric protein has TbTIM’s sequence and 13 single point mutations, which are sufficient to obtain TcTIM-like behaviour in reactivation experiments. We analyzed the residue interaction networks observed in the all-atom MD simulations, as well as their electrostatic interactions and the impact of simulation length on them. A conserved salt bridge between catalytic residues Lys 14 and Glu 98 was observed in all three proteins, but key differences were found in other interactions concerning the catalytic amino acids. Although TcTIM forms less hydrogen bonds than TbTIM and Mut1, its hydrogen bond network spans almost the entire protein, connecting the residues in both monomers. Some of these interactions appeared only after the first microsecond of the simulation, and convergence in the number of hydrogen bonds was only reached during the last of the 3 μs of the simulation. Second, we performed MD simulations of the methyl DNA binding domain (MBD), which is the only domain in MeCP2 with an available structure. After characterizing its structure both in solution and in the presence of a surface in order to compare with high-speed atomic force microscopy experiments (HS-AFM), we built the rest of the protein structure by ab initio modelling using Modeller. This model was simulated in both all-atom and coarse-grained force fields. Two main conformations were sampled in the coarse-grained simulations: a globular structure similar to the one observed in the all-atom force field and a two-globule conformation. A similar two-globule conformation has been observed in the HS-AFM experiments. Our results are in good agreement with available experimental data. They predicted 4.1% of α-helical content, the experimental result is 4%. Finally, we compared the model predicted by AlphaFold to our Modeller model. Together, these simulations represent the first attempt to characterize the structure and dynamics of the full-length MeCP2 protein.