Review 2: "Ribosome Phenotypes Enable Rapid Antibiotic Susceptibility Testing in Escherichia Coli"

RR:C19 Evidence Scale rating by reviewer:

• Strong. The main study claims are very well-justified by the data and analytic methods used. There is little room for doubt that the study produced has very similar results and conclusions as compared with the hypothetical ideal study. The study’s main claims should be considered conclusive and actionable without reservation.

***************************************

Review: The manuscript "Ribosome Phenotypes Enable Rapid Antibiotic Susceptibility Testing in Escherichia coli" by Farrar and colleagues presents an original (and impressively accurate) approach as an alternative to antibiotic susceptibility testing in clinical context. The manuscript is well written and easy to read. 

The main claims of the manuscript are: 

• Using an innovative method using FISH targeting ribosomal RNA, authors observed that ribosomes occupy intracellular space differently when they are antibiotic resistant compared to when they are susceptible. This was tested with 3 different antibiotics from different families: ciprofloxacin targeting DNA replication, chloramphenicol targeting translation elongation and aminoglycosides targeting the ribosome.

• Authors trained a neural network with microscopy images of both sensitive and resistant strains treated with the 3 different antibiotics. 

• The neural network can now predict whether a clinical E. coli isolate is resistant or sensitive to these antibiotics with strong accuracy. The validation with clinically relevant strains is a strength.

The authors present original and powerful method for antibiotic susceptibility testing in clinics. Importantly, this is a rapid and phenotype-based method, thus also considering cell heterogeneity, as opposed to genotype testing. In fact, the genotype does not always correlate with resistance, and can not always accurately predict antibiotic tolerance or persistence.

The method can also have implications in the fundamental study of bacterial response to antibiotics. In my opinion, the authors could have discussed this a little further in the discussion.

Limitations: In the clinical context, the use of this method implies the necessity of imaging technologies and of previously trained neural networks for the tested species. Also one needs to know which species in causing the infection in order to choose the adequate NN trained with images from this species.

Questions:

1. Would this work with antibiotics causing cell lysis like membrane targeting antibiotics such as beta-lactams?

2. Do differences between resistant vs susceptible strains correlate with live vs dead in images, or do living cells show visually different phenotypes in ribosome space occupancy?