Persister assays.

The presence of persister cells in bacterial cultures is usually detected using biphasic kill curves in which the addition of a bactericidal antibiotic is followed first by rapid killing of regular cells and then by a second, slower phase in which persister cells are killed (1, 2). Persister cells can then be quantified by comparing the levels of survivors obtained with, e.g., different mutant strains after a given time of antibiotic treatment. Experimental procedures similar to the one used here had also been described in our previous work and in reports of numerous studies by others in the field (5, 8, 9, 11). These studies showed that antibiotic treatment in LB medium and different minimal media for exponentially growing bacteria of ca. 1 to 5 × 10⁸ CFU/ml results in biphasic killing with merely persisters surviving after 5 h. Similarly, we demonstrated biphasic killing under these conditions for the M9 minimal medium that we used throughout this work (Fig. S2).

Overnight cultures were inoculated from single colonies into 3 ml of LB or M9 medium and grown for ca. 16 h in plastic culture tubes (catalog no. 62.515.006; Sarstedt; the lid was taped at the 13-ml mark with its lower end to ensure uniform aeration of replicates). For persister assays based on a single growth time point as described in Fig. 1A, overnight cultures were diluted at 1:100 back into LB or M9 medium in Erlenmeyer flasks and agitated at 37°C in a water bath shaker until they were in the mid-exponential-growth phase (ca. 1 × 10⁸ CFU/ml to 5 × 10⁸ CFU/ml; reached in our setup after ca. 2 to 2.5 h in LB medium or 2.5 to 3 h in M9 medium). At that point, cultures were treated with lethal concentrations of different antibiotics (100 µg/ml ampicillin, 1 or 10 µg/ml ciprofloxacin, or 7.5 µg/ml gentamicin) for 5 h in plastic culture tubes under conditions of rigorous agitation. In parallel, CFU counts per milliliter of the cultures were determined by plating serial dilutions on LB agar plates. After antibiotic treatment, bacterial pellets of 1.5-ml samples were washed once in 1 ml of sterile phosphate-buffered saline (PBS), resuspended in 100 µl of sterile PBS, and serially diluted in sterile PBS. Samples (10 µl) were spotted on LB agar plates to quantify antibiotic-tolerant survivors. Agar plates were incubated at 37°C for at least 24 h, and CFU counts per milliliter were determined from spots containing 10 to 100 bacterial colonies. The fraction of persister cells was calculated as the ratio of the CFU counts per milliliter after and before antibiotic treatment.

In order to study the dynamics of antibiotic-tolerant cells throughout the different growth phases of E. coli (as outlined in Fig. 2A), we performed an experiment as described above but adjusted the time between subculturing and antibiotic treatment from 0 h (i.e., direct inoculation into fresh medium containing antibiotic) to 8 h. For experiments with the (p)ppGpp-deficient strain E. coli K-12 MG1655 relA::FRT spoT::cat(207), the M9 growth medium was supplemented with an amino acid mixture instead of Casamino Acids (see above) and the experiment was extended to 10 h. Though the relA spoT mutant and E. coli K-12 wild-type strain have roughly the same growth rate in this medium, the final CFU count per milliliter of the (p)ppGpp-deficient mutant is significantly lower, which regularly results in experimental artifacts due to the appearance of suppressor mutants during overnight cultures (33). We therefore set up overnight cultures in LB medium (where this effect is much less pronounced) and adjusted the inoculum of both the mutant and wild-type strains to ca. 10⁶ CFU per milliliter in order to account for the roughly 3-fold-higher CFU per milliliter of wild-type overnight cultures.