Shining a light on antibiotic selection: optimisation of live/dead fluorescence spectroscopy for rapid antimicrobial susceptibility testing

Abstract

Antibiotic resistance is a serious threat to public health. The empiric use of the wrong antibiotic occurs due to urgency in treatment combined with slow, culture-based diagnostic techniques. Inappropriate antibiotic choice can promote the development of antibiotic resistance and can result in increased patient morbidity and mortality. We propose to use live/dead spectroscopy as a rapid alternative to culture-based techniques through application of the LIVE/DEAD® BacLightTM Bacterial Viability Kit. We have developed a spectroscopic device (Optrode) to measure fluorescence from SYTO 9 and propidium iodide (PI) stained cells that can be used to enumerate the bacterial load. We propose a procedure using the Optrode that will take bacteria in a clinical sample, challenge with a panel of antibiotics, and measure live/dead ratios to determine the best bactericidal choice. Initial investigations revealed that the experimental parameters outlined in the kit instructions do not generate results that model cell viability with the best possible fit. Using calibration data, we optimised the analytical parameters for our application, including selection of wavelength integration ranges for fluorescence emissions and the formula to calculate % live. Following this, we applied the optimised methodology to detect live and dead Escherichia coli in populations challenged with a range of antibiotics. Using this set up we were able to determine the suitability of the system to detect the activity of lytic and non-lytic antibiotics, and the necessity of the pre-staining washing step in an effort to simplify the process. We show that samples of antibiotic-challenged culture do not require washing before staining and that washing may cause loss of biological information from the sample. Antibiotic killing of E. coli was verified by plate counting. Killing was detected by the Optrode in near real-time when E. coli was treated with lytic antibiotics, ampicillin and polymyxin B, and stained with SYTO 9 and PI. However, the same assay system was not sensitive to the action of non-lytic antibiotics, chloramphenicol and ciprofloxacin.