Despite growing global concern for antibiotic resistance, development of new therapeutics is slow. While many antibiotics merely slow bacterial growth, membrane-active peptides (MAPs) can actively kill bacteria by disrupting their essential lipid bilayers. However, MAPs on the market are reserved for severe multi-drug-resistant infections due to their narrow therapeutic window, i.e., off-target toxicity to human cells. The combinatorial space of peptide sequences is astronomical, complicating efforts to establish predictive relationships between peptides’ sequences and their effects on lipid bilayer membranes. Due to the excellent scattering contrast between lipids and peptides, neutron reflectometry (NR) of solid-supported lipid bilayers has emerged as a structural technique of choice for quantifying essential features of the membrane/peptide complex, such as bilayer remodeling and peptide density distributions. The NIST ROADMAP project aims, by creating a corpus of NR measurements of peptides on a variety of lipid membrane compositions, to establish sequence-function relationships for MAPs. Here, we report on two key components of this effort. First, we demonstrate the reliable, automated formation of lipid membranes of the desired compositions, and optimization of the bilayer formation protocol using a bespoke liquid handling robot and autonomous phase space exploration of the parameter space (e.g., flow rate, lipid composition, and buffer composition). The structure of the supported lipid bilayer can be understood in a framework combining collective physicochemical forces and hydrogen bonding. Second, we show first measurements of bacterial and mammalian membrane models before and after exposure to naturally occurring and AI-generated peptide sequences. These early experiments have already begun to reveal patterns between sequence motifs and lipid membrane structure that can be used to generate novel antimicrobial peptides.
Mitchell et al. (Sun,) studied this question.