Single-cell evidence for plasmid addiction mediated by toxin-antitoxin systems

Abstract

Toxin-antitoxin (TA) systems are small selfish genetic modules that increase vertical stability of their replicons. They have long been thought to stabilize plasmids by killing cells that fail to inherit a plasmid copy through a phenomenon called post-segregational killing (PSK) or addiction. While this model has been widely accepted, no direct observation of PSK was reported in the literature. Here, we devised a system that enables visualization of plasmid loss and PSK at the single-cell level using meganuclease-driven plasmid curing. Using the ccd system, we show that cells deprived of a ccd-encoding plasmid show hallmarks of DNA damage, i.e. filamentation and induction of the SOS response. Activation of ccd triggered cell death in most plasmid-free segregants, although some intoxicated cells were able to resume growth, showing that PSK-induced damage can be repaired in a SOS-dependent manner. Damage induced by ccd activates resident lambdoid prophages, which potentiate the killing effect of ccd. The loss of a model plasmid containing TA systems encoding toxins presenting various molecular mechanisms induced different morphological changes, growth arrest and loss of viability. Our experimental setup enables further studies of TA-induced phenotypes and suggests that PSK is a general mechanism for plasmid stabilization by TA systems.

Graphical Abstract

Figure 1.

The ccd system induces the SOS response in plasmid-free segregants. (A) Illustration of the mini-F tracking system. Top: The SopB protein from the native partition system of F was fused to the mNeongreen fluorescent protein in the pNF03 mini-F plasmid. Binding of SopB-mNeongreen to the sopC centromere leads to the formation of green fluorescent foci that enables positional tracking of the plasmid. Bottom: Average SopB-mNeongreen fluorescence intensity signal of 115 cells displaying two SopB-mNeongreen foci. (B) Loss of mini-F plasmids in continuous culture. FN042 cells transformed with pNF03 (EV) and pNF03ccd (ccdAB⁺) were grown to exponential phase in MOPS medium containing 0.4% maltose and 15 μg/ml chloramphenicol to promote mini-F retention before being diluted 1000× in the same medium without antibiotic. Cells were diluted 1000× every 12 h and allowed to grow for 10 generations every cycle. The proportions of foci-free cells were counted at indicated time points and a linear regression was fit to the data of three independent replicates to obtain plasmid loss rate per generation (β). (C) Copy number of mini-F vectors. FN042 cells transformed with pNF03 (EV) and pNF03ccd (ccdAB⁺) grown on MOPS medium containing 0.4% maltose were imaged on agarose pads. The numbers of SopB-mNeongreen foci were counted for the indicated number of cells. (D–F) Time-lapse fluorescence microscopy analysis of unperturbed mini-F loss. FN042 cells transformed mini-F vectors pNF03 (EV) or pNF03ccd (ccdAB⁺) imaged by time-lapse fluorescence microscopy every 15 min on agarose pads made with MOPS medium containing 0.4% maltose. (D) Representative micrographs of mini-F loss events. The SopB-mNeongreen fusion that localizes the plasmid is shown in green, while the sulA-bfp fusion that reports the SOS response is shown by the associated color scale. Time 0 corresponds to the last time point where SopB-mNeongreen foci were visible in arrow-indicated cells. Scale bar is 1 μm. Quantification of median sulA-bfp fluorescence (E) and cell size (F) over time in the indicated number of plasmid-free segregants after foci loss. Shaded areas represent interquartile ranges.

Figure 2.

I-SceI-mediated curing reveals PSK. (A) Schematic representation of the plasmid curing system. The SCE1 gene is under the control of the araBAD arabinose-inducible promoter on a low-copy plasmid (pSce), which allows production of the I-SceI restriction enzyme when cells are grown in the presence of arabinose. Plasmid pNF04, which is used to clone TA systems, contains an I-SceI cutting site (I-SceI_CS), which allows its digestion and curing by the restriction enzyme when arabinose is added to the medium. (B) Viability loss by ccd under plasmid curing conditions. Cells transformed with pSce and pNF04 derivatives (pNF04, EV; pNF04ccd, ccdAB⁺; pNF04ccdGE, ccdAB_G100E) were grown to exponential phase in MOPS medium containing 0.4% glucose, 25 μg/ml kanamycin and 20 μg/ml chloramphenicol, serially diluted and spotted on M9 plates containing 20 μg/ml chloramphenicol and either 0.4% glucose and 25 μg/ml kanamycin to promote plasmid retention or 0.1% glucose and 0.3% arabinose to promote plasmid curing. Data represent the mean and standard deviation of three independent experiments.

Figure 3.

Live imaging of ccd-mediated PSK. (A–C) Time-lapse fluorescence microscopy analysis of I-SceI-mediated plasmid curing. FN053 cells were transformed with pSce and either pNF04 (EV, top) or pNF04ccd (bottom) plasmids, grown on MOPS medium containing 0.4% glucose, 25 μg/ml kanamycin and 20 μg/ml chloramphenicol, then loaded on microfluidic chips, perfused with MOPS medium containing 0.4% arabinose and 20 μg/ml chloramphenicol to induce plasmid curing and imaged by time-lapse fluorescence microscopy every 15 min. (A) Representative micrographs of plasmid-free segregants following I-SceI-mediated curing. SopB-mNeongreen is shown in green, the HU-mCherry fusion that localizes the chromosome is shown in red and the sulA-bfp reporter is shown with the associated color scale. A cell surviving the effects of ccd is shown by a white arrow. Scale bar is 1 μm. (B) Quantification of divisions by microscopy after plasmid loss. The number of cumulative divisions for observed plasmid loss events was quantified at each 15 min time point after SopB foci loss, with time 0 as the last time point where foci were detected, and averaged over the indicated number of imaged plasmid loss events (n). Fate of daughter cells from plasmid curing events was classified as dead (failure to form microcolonies) or live (formation of microcolonies). (C) Quantification of sulA-bfp median fluorescence intensity in daughter cells of plasmid-free segregants 4 h after SopB foci loss. (D) Plasmid retention in DSB repair-deficient mutants. Indicated strains transformed with pNF06 (dashed lines) or pNF06ccd (solid lines) were continuously grown in MOPS medium containing 0.4% glucose and sampled at indicated times during exponential phase to measure the proportion of fluorescent cells by flow cytometry. Data represent the mean and standard deviation of three independent experiments. (E) Viability of DSB repair-deficient mutants under plasmid curing conditions. Indicated mutants transformed with pSce and either pNF04 or pNF04ccd were grown to exponential phase in MOPS medium containing 0.4% glucose, 25 μg/ml kanamycin and 20 μg/ml chloramphenicol, serially diluted and spotted on M9 plates containing 20 μg/ml chloramphenicol and either 0.4% glucose and 25 μg/ml kanamycin to promote plasmid retention or 0.1% glucose and 0.3% arabinose to promote plasmid curing. Data represent the geometric mean and standard deviation of three independent experiments.

Figure 4.

Cooperativity between ccd and lambdoid prophages. (A, B) Time-lapse fluorescence microscopy analysis of I-SceI-mediated plasmid curing in lambda lysogens. MG1655 cells were lysogenized with lambda and transformed with pSce and either pNF04 (EV, top) or pNF04ccd (bottom) plasmids, grown on MOPS medium containing 0.4% glucose, 25 μg/ml kanamycin and 20 μg/ml chloramphenicol, spotted on agarose pads with MOPS medium containing 0.4% arabinose and 20 μg/ml chloramphenicol to induce plasmid curing and imaged by time-lapse fluorescence microscopy every 15 min. (A) Representative micrographs of plasmid-free segregants following I-SceI-mediated curing. SopB-mNeongreen is shown in green, with time 0 as the last time point where foci were detected. Scale bar is 1 μm. (B) Quantification of survival and lysis in plasmid-free segregants in non-lysogen cells grown in microfluidics (as in Figure 3A) or lambda lysogens grown on agarose pads as described above. Fate of daughter cells from plasmid curing events was classified as Lysed (disappearance or loss of phase contrast), Nonlysed Dead (failure to form microcolonies) or Live (formation of microcolonies) in the indicated number of cells (n). (C) Production of viral particles by ccd-induced PSK. Lysogens transformed with pSce and either pNF04 (EV) or pNF04ccd were grown to exponential phase in MOPS medium containing 0.4% glucose, 25 μg/ml kanamycin and 20 μg/ml chloramphenicol, and diluted 100× MOPS medium containing 20 μg/ml chloramphenicol and either glucose (Gluc) or arabinose (Ara) as sole carbon source. PFUs were estimated from lysates after 16 h of growth. (D) Plasmid retention in lysogens. Lysogens of indicated phages transformed with pNF06 (dashed lines) or pNF06ccd (solid lines) were continuously grown in MOPS medium containing 0.4% glucose and sampled during exponential phase at indicated times to measure the proportion of fluorescent cells by flow cytometry. Data represent the mean and standard deviation of three independent experiments.

Figure 5.

Conservation of PSK across TA families. (A) Recapitulative table of tested TA systems and their origin. ND: not determined. (B) TA-induced viability loss under plasmid curing conditions. Cells transformed with pSce and pNF04 derivatives containing the indicated TA systems were grown to exponential phase in MOPS medium containing 0.4% glucose, 25 μg/ml kanamycin and 20 μg/ml chloramphenicol, serially diluted and spotted on M9 plates containing 20 μg/ml chloramphenicol and either 0.4% glucose and 25 μg/ml kanamycin to promote plasmid retention or 0.1% glucose and 0.3% arabinose to promote plasmid curing. Data represent the geometric mean and standard deviation of three independent experiments. (C, D) Time-lapse fluorescence microscopy analysis of I-SceI-mediated plasmid curing. FN053 cells were transformed with pSce and either pNF04 (EV, top) or pNF04ccd (bottom) plasmids, grown on MOPS medium containing 0.4% glucose, 25 μg/ml kanamycin and 20 μg/ml chloramphenicol, and then loaded on microfluidic chips perfused with MOPS medium containing 0.4% arabinose and 20 μg/ml chloramphenicol at time 0 to induce plasmid curing. Cells were imaged by time-lapse fluorescence microscopy every 15 min, with time 0 as the last time point where foci were detected. (C) Representative micrographs of plasmid-free segregants following I-SceI-mediated curing. SopB-mNeongreen is shown in green, the HU-mCherry fusion that localizes the chromosome is shown in red and the sulA-bfp reporter is shown with the associated color scale. Scale bar is 1 μm. White arrows show loss of cytosolic content, while blue arrows show blebbing. (D) Quantification of divisions by microscopy after plasmid loss. The number of cumulative divisions for observed plasmid loss events was quantified at each 15 min time point after SopB foci loss and averaged over the indicated number of imaged plasmid loss events (n).