Box 2.

Proposed research questions that could be explored to aid easier antibiotic discovery.

The entry/exit problems

• Methods to measure entry of molecules into (Gram-negative) bacteria are needed—there are some tentative thoughts that mass spectrometry may be a way forward, but the methods are a long way from being truly useful (28) • Develop rules, like an antibacterial version of ‘Lipinsky’s rule of 5′ for cytoplasmic entry into Gram-negative bacteria • Studies of self-promoted uptake, including investigating more closely the surface structure and charge of the outer membrane in Gram-negatives • Investigate more closely which properties that make molecules substrate for efflux pumps

Chemical novelty • Better screening libraries are needed, keeping in mind that antibacterials in general have been found to have differing chemical properties than other pharmaceuticals (29). Maybe virtual screens of compounds, e.g. in PubChem, could be useful as a first step, much like that performed at the University of Dundee for neglected diseases (30) • Create new, more varied libraries—however, diversity can mean many different things, and it needs to be defined how best to achieve diversification • New libraries could also be created by going back to natural products and finding new sources of compounds such as slow-growing micro-organisms, marine bacteria, or products of cryptic pathways. The molecules need to be coupled with whole-cell screens and assays that allow for effective de-replication of already discovered natural products (31)

Avoiding resistance

• Increase fundamental knowledge of under what circumstances (clinically relevant) resistance mutations arise • Finding ways of ‘reversing’ resistance other than beta-lactamase and efflux pump inhibitors, such as perhaps inactivating the Cfr-methylase that gives resistance to a variety of ribosome-directed drugs • Better animal models are needed, to understand better the correlation between in vitro and in vivo resistance data. Tests for efficacy usually involve a loading dose of ∼103–105 bacteria, so resistance events, occurring a frequency of 1 in 107 or less, may rarely be picked up

Other

• Methods to test the usefulness of anti-virulence approaches in vivo, perhaps by using engineered strains that can turn off virulence factors during an infection • Investigate which panel(s) of strains that are the most relevant to test new antibiotics against • Address the lack of structural data for many bacterial proteins