Understanding Periperal Membrane Protein Interactions | BPS Thematic Meeting

Understanding Peripheral Membrane Protein Interactions: Structure, Dynamics, Function and Therapy

Wednesday Speaker Abstracts

DESCRIBING PROTEIN CONFORMATIONAL LANDSCAPES AT THE PROTEIN MEMBRANE INTERFACE: IMPLICATIONS FOR DRUG DESIGN Zoe Cournia 1 ; 1 Biomedical Research Foundation, Academy of Athens, Athens, Greece Peripheral and lipid-anchored proteins exert their function at the protein–membrane interface, whereby the intrinsic plasticity of this environment gives rise to highly dynamic conformational landscapes. Characterizing the conformational landscapes of proteins at membrane interfaces can reveal how lipid environments modulate structural flexibility, allostery, and ligand accessibility. Such insights are increasingly critical for rational drug design, enabling the discovery of allosteric and cryptic binding sites unique to membrane-bound conformations. In this talk, we perform unbiased and biased Molecular Dynamics simulations of several peripheral membrane proteins and report the free energy conformational landscapes for their wild-type (WT) and disease-relevant mutant forms. 1,2,3 We also discuss allosteric coupling implications arising from protein-membrane interactions. We study protein dimerization and oligomerization in membranes and show that the structure and energetics of the dimerization process of membrane proteins can be reproduced with reasonable accuracy and throughput. 4 The distinct conformational ensembles that these WT and mutant proteins adopt on the membrane, enable structure-based detection of allosteric pockets at the protein-membrane interface. These findings offer a mechanistic basis for the function of the studied peripheral membrane proteins and provide insights for new therapeutic strategies targeting the protein-membrane interface. Finally, we describe an ensemble machine learning methodology to predict protein-membrane interfaces of peripheral membrane protein 5 and present a drug design pipeline for drugging protein membrane interfaces using the DREAMM (Drugging pRotein mEmbrAne Machine learning Method) web-server https://dreamm.ni4os.eu. 6 We further extend this study to apply Artificial Intelligence techniques in the context of Natural Language Processing (NLP) and show that the accuracy and prediction time for protein-membrane interface analysis can be significantly improved compared to existing methods. 7 References1. Kotzampasi et al, Comput Struct Biotechnol J. 2024,23:3118-3131.2. Kotzampasi and Cournia. Comms Chem., 2025, in press.3. Koukos et al, biorxiv,2025.04.22.6500404. Lamprakis et al. J Chem Theory Comput., 2021, 11;17(5):3088-3102.5. Chatzigoulas and Cournia. Brief Bioinform ,2022, bbab518.6.

Chatzigoulas and Cournia, Bioinformatics, 2022, 38:5449-5451.7. Paranou,Chatzigoulas,Cournia. Advances Bioinf, 2024, 4(1):vbae078

29