ABSTRACT
Human mitogen-activated protein kinase (MAPK) family members JNK and p38 are two homologous protein-serine/threonine kinases but play distinct roles in the pathological process of neurological disorders. Selective targeting of JNK over p38 has been established as a potential therapeutic approach to epilepsy and other nervous system diseases. Herein, we describe an integrated in vitro-in silico protocol to rationally design kinase-peptide interaction specificity based on crystal structure data. In the procedure, a simulated annealing (SA) iteration optimization strategy is described to improve peptide selectivity between the two kinases. The optimization accepts moderate compromise in peptide affinity to JNK in order to maximize the affinity difference between peptide interactions with JNK and p38. The structural basis, energetic properties and dynamic behavior of SA-improved peptides bound with the peptide-docking sites of JNK and p38 kinase domains are investigated in detail using atomistic molecular dynamics (MD) simulations and post binding free energy analysis. The theoretical findings and computational designs are then confirmed by fluorescence polarization assays. Using the integrated protocol we successfully obtain three decapeptide ligands, namely RLHPSMTDFL, RAKLPTSVDY and KPSRPWNLEI, that exhibit both potent affinity to JNK (K = 8.0, 5.4 and 12.1 µM, respectively) and high selectivity for JNK over p38 (K/K = 9.2, 17.9 and 6.3 fold, respectively). We also demonstrate that a JNK-over-p38 selective peptide should have a positively charged N-terminus, a polar central region and a negatively charged C-terminus, in which a number of hydrophobic residues distribute randomly along the peptide sequence. In particular, the residue positions 1, 6 and 9 play a crucial role in shaping peptide selectivity; the presence of, respectively, Arg, Thr and Asp at the three positions confers high specificity to kinase-peptide interactions.