High‐throughput statistical screening of antiinfective peptides from natural antibacterial protein repertoire: Chemometric prediction, molecular modeling, and susceptibility analysis

Abstract

Antibiotic‐resistant bacterial infection (ARBI) is one of the most serious global public health threats. Antiinfective peptides (AIPs) have been recognized as a promising alternative to traditional antibiotics, which can effectively combat the ARBI in a distinct mechanism. In the current study, we attempt to discover new and potent AIPs from the natural antibacterial protein repertoire. Hundreds of antibacterial proteins with sequence length > 50 amino acids are manually curated from literatures and databases, which are then broken into a large pool of 12‐mer peptide fragments. In the procedure, a high‐throughput statistical screening strategy that integrates machine learning, chemometic prediction, and molecular modeling is employed to computationally identify 8 promising AIP hits from the fragment pool, of which 5 are determined by susceptibility test to possess a moderate or high potency against human pathogenic bacteria (20 μg/mL < minimum inhibitory concentration < 90 μg/mL), while the other 3 have only a low or no antibacterial activity (minimum inhibitory concentration > 100 μg/mL). Conformational analysis characterizes that the active AIPs are almost α‐helical (helical rate > 50%), carry many positive charges (net charge > +3), and exhibit an amphipathic profile. Dynamic simulation of a representative membrane‐AIP interaction reveals that the peptide can fluctuate nearby the membrane surface and use its cationic side chains to directly interact with and tightly bind to the anionic hydrophilic layer of bacterial outer membrane.