APPLICATION OF IN SILICO PROTEIN ENGINEERING METHODS FOR THE OPTIMIZATION OF ANTIMICROBIAL ACTIVITY OF PEA DEFENSIN PSD1

Mycoses are a class of underreported diseases that affect at least one billion people worldwide every year. Due to the emergence of conventional drug-resistant fungal species and strains, there is an increasing need for alternative compounds for the treatment of fungal diseases. One of these alternatives are Antimicrobial Peptides (AMPs). Our research group currently works with a pea AMP termed Pisum sativum defensin 1 (Psd1). Previous studies showed that Psd1 possesses antifungal activity against human pathogenic fungi through a cell membrane translocation mechanism, while simultaneously exhibiting low toxicity toward human cells. Together, these findings already suggest some therapeutical potential of Psd1; however, it possible nowadays to employ computational techniques in order to increase such potential even further. With this in mind, in silico protein engineering approaches were used to design Psd1 analog peptides with increased activity over the wild-type variant. Initially, a Computational Alanine Scanning (CAS) study was performed in order to identify interaction hotspots of Psd1 with a fungal model lipid membrane comprised of phosphatidylcholine (POPC) and fungal glucosylceramide (GlcCer). These hotspots then served as starting points for the creation of a large combinatorial library comprised of 8748 mutant peptides, with each one containing varying degrees of arginine and/or tryptophan enrichment in the hotspots previously identified by CAS. All of the mutant peptides were then submitted to a computational Virtual Screening (VS) experiment, where 1649 peptides were found to have greater binding affinity than wild-type Psd1 in a biomembrane-mimetic anisotropic slab. The top three ranked peptides from the VS and the wild-type peptide were then submitted to Coarse Grained Steered Molecular Dynamics (CG-SMD) simulations coupled with the Umbrella Sampling method. The preliminary resulting potentials of mean force showed that all three mutant peptides exhibited lower adsorption and penetration free energy barriers than the wild-type Psd1 in a fungal model membrane comprised of POPC, ergosterol and GlcCer. This suggests that the three tested analogs might possess a greater antifungal activity than the wild-type variant in vivo, due to the more favorable translocation energetics. As a perspective, the toxicity of the mutant peptides toward human cells will be estimated by carrying out additional CG-SMD simulations with mammalian model membranes comprised of POPC, cholesterol and sphingomyelin.