Multiscale modelling of bacterial and viral sanitisation: A combined molecular dynamics and nuclear magnetic resonance study

Abstract

The cell envelope is a pathogen’s first defence against potentially hazardous materials in the immediate environment. Because of this, pathogens evolved to control which chemicals are allowed in and out of the cell selectively. New cell envelope adaptations mean our current sanitisation methods are becoming less effective daily. Lipid compositions, protein mutations, and lipopolysaccharide (LPS) in the outer membrane of Gram-negative bacteria are but a few features that pathogens have developed to help them resist current antimicrobials. Due to the importance of the cell envelope in the survivability of a pathogen, it has become a widespread target of sanitisers, specifically the lipid bilayer or essential proteins. A detailed understanding of these areas is critical if antimicrobial resistance one day renders all of our current methods ineffective. If this were to happen, we must understand why they worked initially, as this will inform our search for the adaptations responsible for resistance. By identifying the areas where these adaptations are found, scientists can engineer new, tailored antimicrobials. These tailored antimicrobials could exploit the properties of the adapted bilayer or new mutations in membrane proteins to induce deformation and disable or destroy them. In this thesis, Chapters 4 and 5 explored the effect of short-chained alcohols and chlorhexidine (CHX) on phospholipid (PL) membranes with and without LPS using a multidisciplinary approach. I found that the application of sanitising alcohols at concentrations as low as 20% had a significant deforming effect on the cell membrane (SaCM) of Staphylococcus aureus (S. aureus), as well as the inner (EcIM) and outer (EcOM) membrane of Escherichia coli (E. coli). Furthermore, the application of short-chain alcohols created disruptions in the headgroup region, which allowed CHX to bind more effectively with PL leaflets. In contrast, this same disruption led to budding in the PL leaflet of the EcOM. I validated the simulation data by performing equivalent nuclear magnetic resonance (NMR) experiments. In this way, simulations added visual context to changes in the spectra obtained from NMR experiments. These experiments assessed whether the prevalent effects in the simulations were replicated in vitro beyond computationally accessible time scales. Said NMR experiments found that the CHX’s binding position and conformation in molecular dynamics (MD) were consistent with those in vitro. In conjunction, these works showed that although short-chained alcohols only underwent transient interactions with the headgroup-tail interface of membranes, this overall aggregation was essential in the initial phases of membrane sanitisation. Furthermore, this allowed CHX to adopt a "c-shape" binding conformation with its termini embedded in the same interface region. II Multiscale modelling of bacterial and viral sanitisation In chapters 6 and 7, the effect of sanitising agents on proteins and the intricacies of protein protein interaction during SARS-CoV-2 host infection were explored in all-atom (AA) and coarse grain (CG) resolution, respectively. By applying a CG resolution to large cross-linking systems, I was able to simulate these systems for very long timescales. On the other hand, the AA system was simulated for far less time because this provided a significantly greater resolution, enabling me to explore specific interactions in further detail. The dynamics of the SARS-CoV-2 spike (S) protein were explored in a two-step approach: firstly, by detailing the impact of sanitising alcohols on the ectodomain (ECD) of the S protein in AA detail and secondly, by simulating cross-linking of different-sized S protein and angiotensin-converting enzyme 2 (ACE2) receptor clusters. Simulating this area in these separate approaches allowed for assessing structural and conformational stability. In chapter 6, it was found that, similarly to alcohol-membrane systems, despite the transience of interactions, the deformation process was less stochastic than it initially appeared. This hypothesis was first formed when assessing the trends in amino acids, which tended to see more interactions with alcohol in the surrounding solution. These amino acids tended to be large and hydrophobic with charged or polar regions, not dissimilar to the headgroup-tail interface of a membrane. The properties of amino acids that interacted more with short-chain alcohol suggested that similar to the membrane, short-chained alcohols were partitioned into hydrophilic-hydrophobic interface regions. These interactions resulted in an overall rigidifying of the trimer. Simulation of S-ACE2 cross-linking in Chapter 7 found that higher coordination were generally less stable but possible. The S trimer’s conformational restrictions were seen to restrict its ability to bind to multiple receptors. Tilting in the neck region and an observed rigid-body rotation in the receptor binding domain (RBD) meant that the protein was significantly less flexible and placed particular conformational requirements on how these proteins bind. Ultimately, this resulted in high-order binding being substantially less favourable and suggested that this would not occur in vitro due to the energetic penalties incurred by deviation from the low-order cross-linking conformation.

More information

Identifiers

ORCID for Russell Minns: orcid.org/0000-0001-6775-2977

ORCID for Phil Williamson: orcid.org/0000-0002-0231-8640

Catalogue record

Export record