Core Concepts
A deep learning-based framework can efficiently generate novel antimicrobial peptides with potent activities against both bacteria and viruses.
Abstract
The study presents a deep learning-based framework for the de novo design of antimicrobial peptides (AMPs) that can inhibit a broad spectrum of pathogens, including both bacteria and viruses.
The framework consists of a generative adversarial network (GAN) generator and an antimicrobial activity regressor called AMPredictor. The GAN generator learns the hidden patterns of known AMPs and generates novel peptide sequences, while the AMPredictor model predicts the antimicrobial activity (minimum inhibitory concentration, MIC) of the generated peptides using a graph convolutional network.
The authors discovered three bifunctional AMPs (P001, P002, and P076) that exhibited potent antimicrobial and antiviral activities. Experimental validation showed that:
P076 is a highly potent bactericide with an MIC of 0.21 μM against multidrug-resistant Acinetobacter baumannii, outperforming the clinically relevant antibiotic polymyxin B in terms of safety and efficacy.
P002 broadly inhibited five enveloped viruses, including CHIKV, HTNV, DENV-2, HSV-1, and SARS-CoV-2, with low cytotoxicity.
The membrane-acting mechanisms of these AMPs were investigated through experimental assays and molecular dynamics simulations, revealing their abilities to bind and disrupt bacterial and viral membranes.
The study demonstrates the feasibility of using deep learning to uncover novel AMPs with simultaneous antimicrobial and antiviral activities, providing a promising approach to combat a wide range of drug-resistant infections.
Stats
The minimum inhibitory concentrations (MICs) of the three peptides against various bacterial strains ranged from 0.20 to 15.18 μM.
The half-maximal effective concentrations (EC50) of the three peptides against different viruses ranged from 0.37 to 2.67 μM.
The 50% cytotoxic concentrations (CC50) of the three peptides on mammalian cell lines were above 30 μM.
The concentration of P076 to induce 50% death in mice was 80 mg/kg, approximately 3-fold higher than that of polymyxin B (26 mg/kg).
Quotes
"P076 is a highly potent bactericide with the minimal inhibitory concentration of 0.21 μM against multidrug-resistant A. baumannii, while P002 broadly inhibits five enveloped viruses."
"Our study provides feasible means to uncover sequences that simultaneously encode antimicrobial and antiviral activities, thus bolstering the function spectra of AMPs to combat a wide range of drug-resistant infections."