Method to parameterize bacterial interactions with antimicrobial peptide (AMP) coated surfaces

Abstract

Antibiotic resistance is a growing concern and the World Health Organization classifies it amongst the top 10 threats to human health. AMPs were initially used to describe naturally derived host defense peptides and now the definition has also been extended to synthesised peptides. They have been growing in use and popularity as an alternative to antibiotics, especially because they are believed not to not easily give rise to resistance. Models exist to explain their working in bulk since they mostly have been used in bulk so far. None of the models are suitable for a broad range of AMPs and especially for surface bound AMPs. Surface immobilized AMPs could find application in implants, wound patches and other medical applications. Hence it is important to find a method to unravel their mechanism. Image analysis and machine learning are rapidly growing fields that could aid in this endeavour. This master thesis project evaluates the combination of microscopy and image analysis as a viable alternative to existing analytical methods in studying bacterial interactions with AMP coated surfaces. Microfluidic chips serve as the medium to mount and observe bacterial interactions through a microscope because of the importance of flow in naturally occurring conditions. Plots of bacterial area against time, obtained from processing microscopy data shows a decrease in individual size of bacteria over time at a faster rate on AMP immobilized surfaces than when the antimicrobials were in solution as in the control experiments. The average intensity of bacteria, calculated as the intensity of bacterial fluorescence divided by the area of the bacteria, against time remained rather constant. Other parameters, like tendency of bacteria to divide or cluster can be similarly calculated from the data gathered. Further, correlating the structure of the used AMP to the observed effect on bacteria can help design better AMPs. Hence, the method proves to be promising. The future of the method as an analytical tool relies on smarter, improved algorithms and clean, high quality input data