GFP fluorescence assay

Strains were grown overnight at 28 °C to saturation in MOPS minimal medium (⁷²), supplemented with 1.12 mM K₂HPO₄, 0.2% glucose, 0.1% casamino acids, 50 μg / mL thiamine, with Cm and Km. The next day cultures were diluted 1:100 into the same medium, and 300 μL added to wells of a black-walled, clear-bottomed 96-well plate (ThermoFisher {"type":"entrez-nucleotide","attrs":{"text":"M33089","term_id":"334008","term_text":"M33089"}}M33089) for incubation in a Cytation 1 multi-mode plate reader (Agilent BioTek). Plate lids were treated with 10% Triton X-100 in ethanol to prevent fogging (⁷³). Incubation in the plate reader was at 37 °C using continuous double-orbital shaking at the most vigorous setting. GFP signal was collected using a gain of 45 and read height of 10.25 mm. GFP and OD₆₀₀ were read every 20 minutes for 3 hours, at which time aTc was added from a 50x stock in 25% EtOH. The plate was returned to incubation, and readings collected every 10 minutes for 1 hour. Tc was added from a 50x stock in 25% EtOH, and datapoints collected every 10 minutes for 1 hour, which marked the end of the experiment. We found it necessary to perform anhydrotetracycline and tetracycline additions sequentially (a single draw and multiple dispenses) from a multichannel electronic pipet (Eppendorf) while leaving the 96-well plate in the holder of the reader. Although this method caused sample carry-over between wells, we considered such effects were likely negligible and indeed none were ever observed. The alternative method of sequential dispenses, with tip changes to prevent carry-over, required the plate to sit for several minutes outside the machine causing substantial lags in both growth and GFP-expression kinetics (not shown). In total, sequential injection of aTc or Tc took no longer than 45 seconds. Preliminary experiments showed plate “edge” effects to be significant and 1% NaCl, or media blanks, were placed in columns 1 and 12. For three independent experiments, four biological replicates of each of the four strains were arrayed into five groups of two columns each. The “no addition” control was represented twice, as technical replicates, in both columns 2 + 3 and 10 + 11. The treatment groups: aTc, aTc / Tc, and Tc resided in columns 4 + 5, 6 + 7 and 8 + 9 respectively, such that addition of the aTc or Tc occurred in adjacent columns.

Preliminary experiments showed MOPS media, unlike LB, possessed low intrinsic fluorescence. However, even when grown in MOPS, E. coli displayed a strong, density-dependent, autofluorescence in the GFP channel (Fig. S6). Therefore, after subtracting media-specific blanks, the GFP autofluorescence signal observed for the WT control (lacking GFP) was used to normalize the signal from experimental strains by subtraction of the per-timepoint GFP signal (not by interpolation as used in Fig. S6). Additionally, the normalization was performed only within treatment groups (no treatment, aTc, aTc / Tc, Tc), to account for treatment-specific growth rate differences, and only within independent experiments. OD₆₀₀ data were adjusted by subtraction of the media blank. Following normalization, the data were combined to generate summary statistics.