Bioinformatics approaches to structure and function of antimicrobial peptides

Antimicrobial resistance within a wide range of infectious agents is a severe and growing public health threat. Antimicrobial peptides (AMPs) are among the leading alternatives to current antibiotics, exhibiting broad spectrum activity. Their activity is determined by numerous properties such as cationic charge, amphipathicity, size, and amino acid composition. Currently, only around 10% of known AMP sequences have experimentally solved structures. To improve our understanding of the AMP structural universe we have carried out large scale ab initio 3D modelling of structurally uncharacterised AMPs that revealed similarities between predicted folds of the modelled sequences and structures of characterised AMPs. Interestingly, two of the modelled peptides predicted to form β‐hairpins lacked the intramolecular disulphide bonds, cation‐π or aromatic interactions that generally stabilise such AMP structures. Moreover, to the best of our knowledge, the first linear αββ fold AMPs lacking intramolecular disulphide bonds were found. In addition to fold matches to experimentally derived structures, unique folds were also obtained. Following the ab initio study, we performed Molecular Dynamics simulations to check the stability of both newly modelled folds and their fold matches, as well as to gain insight into oligomerisation and membrane interactions of selected folds. Significant membrane curvature, thinning and disruption were observed in the presence of AMPs, as well as interdigitation, headgroup separation and phase changes. Asymmetric ripples leading to localised periodic thinning and thickening of the palmitoyloleoyl-phosphatidylethanolamine and palmitoyloleoyl-phosphatidylglycerol mixture (POPE:POPG) bacterial membrane bilayer model were observed. Although experimental and computational studies of AMPs have been previously shown to induce ripples in pure POPG and POPC (palmitoyloleoyl-phosphatidylcholine) lipids and in vivo, to the best of our knowledge, this is the first time AMPs were observed to induce ripples in a POPE:POPG, or similar mixture of lipids. Distribution of charge on the AMP surface was found to be important in differentiating between the membrane interactions observed. Lastly, we employed the support vector classifier method in order to develop a comprehensive model that can classify between AMPs and non-AMPs. To the best of our knowledge, no comprehensive classifiers based on 3D structures have been reported. The main focus hitherto has been on sequence features, and the models used were often very complex and obscure. The test set performance scores of our best 3-feature model were comparable to complex, state-of-the-art sequence-based deep learning models. Developing a simple classifier such as the one reported in this work enhances our understanding of how AMPs work. The results presented in this work give an overview of the range of protein scaffolds that naturally bear antimicrobial activity, give an insight into AMPs' interactions with bacterial membrane and are a step forward in protein design efforts towards better AMPs