Antimicrobial peptides (AMPs) offer a highly potent alternative solution due to their broad‐spectrum activity and minimum resistance development against the rapidly evolving antibiotic‐resistant pathogens. Herein, to accelerate the discovery process of new AMPs, a predictive and generative algorithm is build, which constructs new peptide sequences, scores their antimicrobial activity using a machine learning (ML) model, identifies amino acid motifs, and assembles high‐ranking motifs into new peptide sequences. The eXtreme Gradient Boosting model achieves an accuracy of ≈87% in distinguishing between AMPs and non‐AMPs. The generated peptide sequences are experimentally validated against the bacterial pathogens, and an accuracy of ≈60% is achieved. To refine the algorithm, the physicochemical features are analyzed, particularly charge and hydrophobicity of experimentally validated peptides. The peptides with specific range of charge and hydrophobicity are then removed, which lead to a substantial increase in an experimental accuracy, from ≈60% to ≈80%. Furthermore, generated peptides are active against different fungal strains with minimal off‐target toxicity. In summary, in silico predictive and generative models for functional motif and AMP discovery are powerful tools for engineering highly effective AMPs to combat multidrug resistant pathogens.
The human sliding clamp Proliferating Cell Nuclear Antigen (hPCNA) interacts with over 200 proteins through a conserved binding motif, the PIP-box, to orchestrate DNA replication and repair. It is not clear how changes to the features of a PIP-box modulate protein binding and thus how they fine tune downstream processes. Here we present a systematic study of each position within the PIP-box to reveal how hPCNA-interacting peptides bind with drastically varied affinities. We synthesized a series of 27 peptides derived from the native protein p21 with small PIP-box modifications, and another series of 19 peptides containing PIP-box binding motifs from other proteins. The hPCNA-binding affinity of all peptides, characterised as KD values determined by surface plasmon resonance (SPR), spanned a 4000-fold range, from 1.83 nM to 7.59 μM. The hPCNA-bound peptide structures determined by X-ray crystallography and modelled computationally revealed inter- and intramolecular interaction networks which correlate with high hPCNA affinity. These data informed rational design of three new PIP-box sequences, testing of which revealed the highest affinity hPCNA binding partner to date, with a KD value of 1.12 nM, from a peptide with PIP-box QTRITEYF. This work showcases the sequence-specific nuances within the PIP-box that are responsible for high-affinity hPCNA binding, which underpins our understanding of how nature tunes hPCNA-affinity to regulate DNA replication and repair processes. Additionally, these insights will be useful to future design of hPCNA inhibitors.
Hybridization of common motifs from known antimicrobial peptides (AMPs) is a promising strategy for designing a new class of antimicrobial agents with improved activity. Herein, the machine learning-generated peptide sequence IIKLLGVLAKFVLGQ-NH2 was utilized for sequence optimization using an alanine scan and alignment with known AMPs to identify the motifs with the highest occurrences. Motifs were then assembled to obtain positively charged peptides with 12 to 18 amino acid lengths and hydrophobicity >60%. The designed peptides exhibit broad-spectrum antimicrobial activity against multiple Gram-negative and Gram-positive bacterial strains, along with excellent pH and plasma stability. The peptides demonstrate high efficiency in eradicating bacterial biofilms with no resistance development over 300 generations against a "superbug" clinical strain of Acinetobacter baumannii. Additionally, AMPs were biocompatible and promoted fibroblast migration, which accelerated artificial wound closure. This study highlights an innovative strategy for engineering highly effective AMPs with significant potential in combating multidrug-resistant pathogens.