Detailed protocol for measuring MDK

The protocol proposed here for measuring tolerance may be automated easily (as we have done), and is a better quantitative indicator of tolerance than existing methods, as will be explained at length in the Results. The protocol exposes populations of bacteria to different concentrations of antibiotics for varied time periods, and the presence or lack of survivors (regrowth) is used to determine the MDK.

Preparation. A 96-microwell plate is filled with antibiotics in concentrations decreasing exponentially with the column. The final column is left antibiotic free, as a control for growth. It should be noted that if the MIC is known (approximately), concentrations below and around it need not be measured. Concentrations should typically reach at least 20× MIC. If the MIC can only be estimated, even up to an order of magnitude, it is advisable to spread concentrations over a wider range, to reach the range where killing is concentration independent.

Bacteria inocula are diluted to the concentration corresponding to the MDK being measured, so that treatment durations longer than the MDK (time) will result in eradication of the population without regrowth. For example: to determine the MDK₉₉, 100 bacteria per well are required; to determine the MDK_99.9, 10³ bacteria should be added to each well; etc. The mean number of bacteria in the inocula, N, is evaluated by serial dilution and plating.

To maintain viability and metabolic state of the bacteria over time, bacteria are stored in NaCl solution at 2–3°C. We verified that these conditions do not alter the level of tolerance in our strains. Note that, because the physiological state of the bacteria is often key to their tolerance, the details of the preparation of the bacterial culture should be carefully determined, depending on the form of tolerance to be measured. For example, tolerance-by-lag, measured in this work, depends on the lag time of the bacteria and therefore on the growth conditions of the culture before the MDK measurement; cultures grown without adequate shaking, or for too short a time, do not reach a uniform metabolic state. Standardized conditions were chosen, to ensure the entire population reaches stationary phase (Appendix A in the Supporting Material). Measurement of tolerance by slow growth would also depend upon the growth conditions before the measurement but, unlike tolerance-by-lag, would be best carried out from a strictly exponential culture, such as a chemostat culture.

Inoculation-incubation cycle. The plate is inoculated, one row at a time, at set time intervals, and returned to incubation and shaking at 750 rpm. If the MDK is not known at all, a wide exponential timescale should be used. Subsequent assays with a narrower range of inoculation times may be required for accurate MDK determination.

Plates should be allowed to reach 37°C before the first inoculation.

Antibiotic wash. Once inoculation and incubation of all rows have been concluded, it is necessary to wash away the antibiotic remains. In the case of ampicillin, β-lactamase is added to all wells at a final concentration of 0.5 unit/mL. Alternatively, two spin-downs are required to reach sub-MIC antibiotic levels in all wells. Each spin-down is 10 min at 1200 g to minimize stress on surviving cells. This procedure is repeated twice, as each spin-down is equivalent to a 10–20-fold dilution, and the highest antibiotic concentration used may be as large as 100× MIC. In this work, surface-treated plates with U-shaped wells were used (Cat. No. 163320; Thermo Fisher Scientific, Waltham, MA). A model No. 5810R centrifuge (Eppendorf, Hamburg, Germany) was used with a multiwell-plate adaptor, and supernatant was discarded manually. The centrifugation was performed at 10°C, for historical reasons.

Final incubation and results. Finally, the plate is incubated overnight with shaking (longer incubation periods may be required for slow-growing strains). Results may be read directly from the plate, as will be discussed, by looking at the high-concentration end, and determining the time at which the treatment becomes too long for the survival of even a single bacterium.