pQSAR-guided Genetic Evolution of Peptide Ligands Toward Improved Binding Capability to Rheumatic Immune-Related Prk Polo-box Domain

Human Prk kinase plays an important role in the maintenance of inflammatory stability and has been established as an attractive target of diverse rheumatic immune diseases. The kinase possesses a highly conserved polo-box domain (PBD) in its C-terminal noncatalytic region, which exhibits a wide specificity across its various substrates to regulate the kinase-mediated substrate phosphorylation. In this study, peptide quantitative structure-activity relationship (pQSAR) was systematically modeled based on a gene ontology-enriched set of Prk PBD-binding peptide data, which was structurally characterized using global descriptors and statistically correlated with peptide affinities via both linear and nonlinear machine learning regressions. The optimal pQSAR predictor was then employed to guide the genetic evolution (GE) of a random peptide group toward improved potency to Prk PBD domain, from which a number of hits were measured to have moderate or high affinity. The resulting Pm7 peptide was determined as a good binder that can effectively interact with the domain at sub-micromolar level, which was then used as a parent template to further derive other promising candidates by combining systematic residue mutagenesis and structure-based rescoring. Two Pm7-derived counterparts Pm72 and Pm74 were rationally designed to exhibit a satisfactory binding profile to Prk PBD domain, with affinity improved by 3.1-fold and 4.6-fold relative to their parent Pm7. Structural and energetic analyses revealed that the potent Prk PBD-binding sequences can be divided into a N-terminal polar section, a middle positively charged anchor and a C-terminal hydrophobic section, thus conferring both stability and specificity to the domain-peptide recognition and association.