Application of a Computational Clustering Approach (DBSCAN) to Detect and Characterize Lipid Rafts in Lipid Bilayer Molecular Dynamics Simulations

Cellular membranes are dynamic and heterogeneous lipid bilayers composed of various amphiphilic phospholipids, acting as selective barriers. Phospholipids do not distribute randomly within the bilayer but tend to cluster with similar molecules, forming lateral domains (lipid rafts), which are essential for various cellular functions. Certain bioactive molecules, such as antimicrobial peptides (AMPs), exhibit strong interactions with cell membranes which are typically cationic, facilitating their interaction with negatively charged outer membrane leaflets, characteristic of bacterial and cancer cell membranes. AMPs are attracted to anionic lipid-enriched regions, such as lipid rafts. Molecular Dynamics (MD) simulations were employed to investigate the action of various peptides in membrane models mimicking the composition of cancer (composed of PC, PS, and PE phospholipids) and bacterial (PC, PG phospholipids) cell membranes. Phospholipid clustering was characterized using the DBSCAN (Density-Based Spatial Clustering of Applications with Noise) in each simulation trajectory. DBSCAN proved advantageous by not requiring a predefined number of clusters, identifying arbitrary shapes, and detecting outliers. Results from each system were compared across different membrane compositions and AMP types. This study allowed for the identification of patterns, similarities, and differences in lipid organization and peptide-membrane interactions under various conditions. The findings highlight how phospholipid composition and AMP properties influence cluster formation and dynamics, potentially correlating with antimicrobial activities.