Skip to main content

Molecular dynamics simulation of the membrane binding and disruption mechanisms of antimicrobial peptides

Final Report Summary - AMP-MEMBRANE MD (Molecular dynamics simulation of the membrane binding and disruption mechanisms of antimicrobial peptides)

This project aimed at a comprehensive comparative study of the action of a set of antimicrobial peptides by molecular dynamics simulations. Antimicrobial peptides are a relevant test subject, as they are a promising new class of antibiotics with the potential of turning the tables on the current state of bacterial resistance. A good understanding of their mechanism of action is the first step to a successful strategy at improving their efficacy and lower side-effects.

In this project a set of methods were employed to computationally simulate molecular-level detail of the interaction between the antimicrobial peptides and their primary targets, the bacterial cell membranes. Our systematic approach was based on a set of peptides that had been designed as part of a combinatorial chemistry effort. Peptides from this group (a 'library') are quite similar in amino acid sequence, but even small variations generate quite different behaviors, with some peptides from the library being quite active against bacteria, where others mostly attack mammalian-type cells, therefore being more toxic. This variety provided ground for a comparative approach, whereby the factors governing each type of behavior would be obtainable from the study of the action of each kind of peptide.

The planned large-scale comparative approach would be quite computer-intensive, to a point that the timescale of the project would be insufficient. To that end, it was decided from the beginning to focus on simplified simulation models. In this case a coarse-grain model was used; such a model trades computer cost for a small degree of accuracy by simulating groups of atoms as a single entity (effectively increasing the 'grain' of the simulated detail). Of course, to make sure the detail that is smoothed out has indeed little relevance to the process under study, one must carefully check whether the obtained coarse-grain model reproduces to an acceptable degree prior experimental knowledge about the system. One of the peptides in the library, codenamed BP100, was particularly suited to this test, as there is quite detailed experimental information on its mechanism of cell membrane disruption.

The generation of the simulation model of BP100 was partly successful: the coarse-grained framework developed and used in the host group is not yet able to reproduce some details that are relevant for a successful recreation of the behavior of BP100. Notably, cell membranes thus modeled sometimes present an excessively high resistance to being penetrated by polar moieties such as the antimicrobial peptides -- this being an effect that has meanwhile been independently identified by others [1]. This limitation severely affects the simulation of a peptide such as BP100 because membrane crossing and pore formation have been shown to be linked to the triggering of its activity [2], and these are processes that require peptide internalization in the membrane.

A final model has indeed been obtained where a porating behavior of BP100 is observed. This result, however, seems to stem not from the appropriateness of the coarse-grained model but rather from the use of a high peptide concentration. Such concentrations have indeed physiological meaning, but the high positive charge that the peptides sport introduces artifacts at these densities. Because so far, in the absence of these artifact-inducing conditions, no suitable model of BP100 could be constructed, our efforts were redirected to the improvement of the coarse-grained simulation framework. These efforts were essentially centered on the development of multiscale schemes, by which a full-detail and a coarse-grain representation of the system can be simultaneously simulated. Such multiscale systems enjoy a reduced computational cost, as they are partly simplified, but still feature the atomistic details that might be essential for a proper study of the relevant processes.

The antimicrobial peptide alamethicin was a successful test subject for the multiscale methodologies: for the first time the formation of pores by this peptide could be observed without requiring a fully atomistic-detailed system. Further exploration of the mechanism of alamethicin poration is now taking place. A similar strategy is currently being applied to study BP100-membrane interaction, by providing atomistic detail to the membrane, in the hope of circumventing the above-mentioned issues. In addition, the work on the subject of multiscaling has also warranted the publication of a book chapter [3].

Work on drug-membrane interactions was then successfully extended to the study of the anti-tumor drug doxorubicin, through a partnership with the Dutch Cancer Institute. As part of this large-scale collaborative effort important insight has been reached on the details of membrane translocation by this drug, and of mechanisms to enhance it.

A second collaboration was established with the group at the Department of Genetics of the University Medical Center Groningen. Molecular dynamics simulations were successfully employed to assess structural characteristics of the membrane interactions of the neuropeptide Dynorphin A. This work is relevant to the determination of the impact of mutations in the sequence of the peptide in the development of spinocerebellar ataxia. The results provided important molecular-level insight into the effect of specific mutations, the consequences of which are currently being investigated.

The work carried out under this project reached several important results. These fall in the categories both of method development and of method application, and attest to the thorough training undergone by the researcher. Methodological developments on multiscaling will have a wide impact on the field of simulation, beyond the immediate application to the interaction of antimicrobial peptides with membranes. On the other hand, application results, such as those obtained with alamethicin or doxorubicin not only advance knowledge pertaining to the respective mechanisms, allowing for rational improvement of these and related drugs, but also confirms the relevance and importance of simulation methodologies.


[1] J. Chem. Theory Comput., 2011, 7 (9), pp 2981-2988
[2] Biophys. J., 2009, 96 (5), pp 1815-1827
[3] N. Goga, A. Rzepiela, M.N. Melo, A.H. de Vries, A. Hadar, A.J. Markvoort, S. Nedea, and H.J.C. Berendsen in Advances in Planar Lipid Bilayers and Liposomes, Ales Iglic (ed.), 2012, Elsevier, Amsterdam, pp 139-170

Related documents