Nature Chemical Biology Publishes Computational Design of Antimicrobial Peptide Nanopores; KDFA2i + 9-NH2 Matches Levofloxacin in Mouse Bacterial Load Reduction
Nature Chemical Biology published in August 2026 a research paper describing computational design of antimicrobial peptide nanopores that self-assemble into pore structures selective for bacterial membranes. The lead candidate compound, designated KDFA2i + 9-NH2, was administered intraperitoneally in a mouse infection model and reduced bacterial load by approximately 2 logs — a magnitude comparable to the reference antibiotic levofloxacin at equivalent doses. The nanopore mechanism differs from the standard cationic amphipathic peptide (CAP) mechanism of most antimicrobial peptides in that KDFA2i-class molecules assemble on the bacterial membrane into discrete transmembrane pores rather than disrupting the membrane through nonspecific electrostatic interactions. The design approach uses molecular dynamics simulations combined with generative chemistry to search sequence space for compounds that both self-assemble into a specific pore geometry and select for bacterial versus mammalian membranes. The result adds to the July 2026 Nature Communications publication of the generative-AI-designed antimicrobial peptide Arcinin (which killed drug-resistant bacteria in a mouse wound model while sparing human cells), continuing the computational-design-first shift in the antimicrobial peptide field.