Antimicrobial combination therapies : a network perspective

Abstract

The growing number of resistant strains and biofilm-related infections emerging in healthcare settings and in the general community is a major biomedical concern. Currently, antimicrobial studies are revisiting the potential of old products and looking for new products with alternative modes of action. Most notably, antimicrobial peptides (AMPs) are receiving a lot of attention because of the widespread availability, multiple mechanisms of action, non-specific molecular targets, and anti-biofilm capabilities. Considering that most of the results obtained in these studies lay in scientific literature, and manual curation is time and resource consuming, the development of bioinformatics approaches for the systematic screening of the literature is of obvious interest. In particular, the reconstruction of drug interaction networks reflecting in vitro and in vivo results is considered useful to identify the most promising candidates for the development of alternative antimicrobial therapies, such as antimicrobial combinations. Such networks can aid in profiling and interpreting the activity of AMPs and the added value of antimicrobial combinations, and thus, help exploit their potential. As a first contribution to this line of analysis, this work presents a novel network reconstruction for results obtained by AMP-drug combinations in fighting Pseudomonas aeruginosa infections. This network contains information about strains, combination methodologies, mode of growth, compound description (with drug and AMP database cross-linking) and quantification values (MICs, FICs, log reduction, etc.). So far, the network comprises 239 combinations, such that 83 % of the interactions pair an AMP with a non-AMP compound (antibiotics, enzymes, etc.), mainly traditional antibiotics. The majority (82 %) of the studies focused on the use of combinations on planktonic cells, and surprisingly enough, only 3 % of the studies tested the combination in biofilms. Furthermore, the network is dominated by a small number of highly connected nodes, namely the peptides colistin and polimyxin B. These are the products that are more often tested in antimicrobial combinations. The network is publicly available, and may be further explore using graph-based analysis tools. Hopefully, this will be a valuable resource to the design of new experiments, unveiling different mechanisms of action and helping in the prediction of new combinations.