Intracellular Localization of the Proteins Encoded by Some Type II Toxin-Antitoxin Systems in Escherichia coli

ABSTRACT Bacterial toxin-antitoxin (TA) systems encode a toxin and an antitoxin that counteracts the toxin. Such TA systems are found abundantly on bacterial chromosomes and on extrachromosomal genetic elements. The toxin is always a protein. Based on the nature of the antitoxin (protein or RNA) and on their mode of regulation, they are classified into six groups (I to VI). In the group II TA systems, both the toxin and the antitoxin are proteins, and the gene specifying the antitoxin precedes the gene specifying for the toxin. Here, we studied the intracellular localization in Escherichia coli cells of the proteins specified by the following type II TA modules: mazEF, chpBIK, mqsRA, and rnlAB. We visualized the localization of these proteins by fusing them with the fluorescent protein mCherry using recombinant DNA technology. We used fluorescence microscopy and image analysis software to obtain and quantify protein distribution data. With the exception of the chpBIK TA module, we found that the localization of each toxin-antitoxin complex was different from the localization of the toxin itself. Our results demonstrate clearly that the presence of the antitoxin shifts the localization of its respective toxin toward the middle of the cell, which could contribute to the reduction of cellular toxicity.

interact; and (vi) the antitoxin protein is degraded by a specific bacterial protease (2). Here, we focus on only to type II TA systems.
The first chromosomal TA system discovered, and, to date, the most studied, is the Escherichia coli mazEF module (3). The mazEF module encodes the antitoxin MazE and the toxin MazF (3). The mazEF system is triggered by various stressful conditions (4), which include (i) extreme amino acid starvation, leading to the production of the starvation signaling molecule guanosine-39,59-bispyrophosphate (ppGpp), which is synthesized by the relA-encoded protein (RelA) (3); (ii) inhibition of transcription and/or translation by antibiotics, including rifampin, chloramphenicol, and spectinomycin (5); (iii) DNA damage caused by thymine starvation (6), as well as by mitomycin C, nalidixic acid, and UV irradiation (4); (iv) oxidative stress (H 2 O 2 ) (4); and (v) high temperature (50°C) (4). All of these stressful conditions prevent the expression of the labile protein MazE. When mazEF is not expressed, the existing labile antitoxin MazE is degraded by the ATP-dependent ClpAP serine protease; therefore, there is nothing to interfere with the activity of the stable toxin MazF (3). It has also been shown that MazE and MazF interact in the form of a linear heterohexamer composed of alternating MazF and MazE homodimers (7). MazF has been described as a sequence-specific endoribonuclease that cleaves single-stranded mRNAs at either the 39 or the 59 side of the first A in ACA sequences (8).
Homologous to mazEF is the E. coli chromosomal system chpBIK. This system consists of a pair of genes, chpBI and chpBIK, which encode the antitoxin ChpBI and the toxin ChpBK, respectively (9). The toxin ChpBK consists of 116 amino acid residues, and its sequence shows 35% identity and 52% similarity to that of MazF. Like MazF, ChpBK is a sequence-specific endoribonuclease. It has been reported that, both in vivo and in vitro, ChpBK cleaves mRNAs at either the 59 or the 39 side of the A residue in ACY sequences (where Y represents U, A, or G). This cleavage of mRNAs in the E. coli cells leads to a reduction of protein synthesis (10). However, in contrast to the activation of E. coli MazF, activation of the toxin ChpBK does not cause cell death (10).
Another member of the type II TA systems is the mqsRA toxin-antitoxin module. MqsR, the toxin partner of the mqsRA system, is a sequence-specific endoribonuclease that cleaves GCU sites in mRNA (11). The induction of MqsR causes cell growth arrest; coinduction of the MsqA antitoxin leads to recovered cell growth (11). The observation that the genes specifying biofilm formation lack GCU sequences suggested that the toxin MqsR may have a role in E. coli biofilm formation (11). In fact, such a role has been shown in the soil bacterium Pseudomonas putida (12).
A recently discovered type II TA system is the rnlAB toxin-antitoxin module. The toxin RnlA is part of a larger protein, RNase LS, which is responsible for rapid degradation of mRNAs. RnlA is essential for its endonuclease activity (13). LS RnlA is involved in phage T4 development by cleaving late mRNAs of the phage at 39 to G sites (13). Cells that have a mutation in the RnlA component accumulate high levels of rRNA (14). The RnlA toxin is inhibited by the labile antitoxin RnlB, which is degraded by ClpXP protease and Lon (15).
Here, we studied the intracellular localization of several representatives of the type II TA systems in E. coli cells. In order to visualize the localization of these proteins, we fused them with the fluorescent protein mCherry using recombinant DNA technology. Using fluorescence microscopy techniques and Image-Pro software, we examined the intracellular localization of the protein products of the type II TA systems mazEF, chpBIK, mqsRA, and rnlAB.

RESULTS
The mazEF toxin-antitoxin module. Initially, we studied the intracellular localization of the protein products of the mazEF TA system. In all of the experiments reported here, we exclusively used E. coli MC4100 relA 1 cells. Since the concentrations of the toxin-antitoxin proteins are extremely low, we established their localization by the use of the fluorescent protein mCherry located on the low-copy-number plasmid pBAD18.
The proteins were induced by 10 mM arabinose. At the first stage, we determined the intracellular localization of mCherry itself (unfused). This was done in E. coli MC4100 relA 1 cells with the mcherry gene, located on plasmid pBAD18 and induced by arabinose. As can be seen in Fig. S1 in the supplemental material, mCherry itself is dispersed in the cytoplasm, i.e., it is not specifically localized in E. coli MC4100 relA 1 cells.
To determine the intracellular localization of the protein products of the mazEF toxin-antitoxin module in E. coli cells, we fused each of these products (the toxin mazF, the antitoxin mazE, and the combined mazEF) to the fluorescent reporter mCherry. For this purpose, primers were generated to produce mazF, mazE, and mazEF genes lacking stop codons and fused to a linker of three alanine residues, as described in Materials and Methods. The fused genes thus generated were first amplified by PCR and then cloned into the pBAD18-mcherry plasmid. The primers used for cloning are listed in the supplemental material. The cloning protocols are described in detail in Materials and Methods. The constructed fusions are illustrated schematically in Fig. S2 in the supplemental material.
A series of assays were performed to determine the effect of the overproduced protein products of the fusions on cell viability. E. coli MC4100 relA 1 cells carrying the pBAD18 plasmids with the constructed fusions were grown as described in Materials and Methods. After induction with 10 mM arabinose, cell viability was determined. Figure 1A shows the effect of the overproduced mCherry fusions of the protein products of the mazEF TA module on bacterial viability. Overproduction of the MazF-mCherry fusion caused loss of cell viability (Fig. 1A, lane 6), corresponding to the effect of the overproduced MazF that was not fused to mCherry (Fig. 1A, lane 2). On the other hand, no loss of viability was observed in the case of overproduced antitoxin MazE-mCherry (Fig. 1A, lane 4) or the combined antitoxin-toxin MazEF-mCherry (Fig. 1A, lane 5). Thus, fusing mCherry to the protein products of the mazEF module did not impair their functionality; even after fusion with mCherry, MazF remains toxic and MazE prevents its toxicity. Figure 2A shows the intracellular localization of the toxin MazF fused with mCherry in the E. coli MC4100 relA 1 strain. As can be clearly seen, MazF-mCherry was observed in two distinctly different intracellular localization patterns. There are cells in which MazF-mCherry is located at the poles of the cells (indicated by white arrows). In fact, in 20 of 37 cells of this representative image (which shows one out of 15 similar experiments), MazF-mCherry was localized at the pole of the E. coli cells ( Fig. 2A). On the other hand, in the same representative image, in 17 cells MazF-mCherry is not localized at the pole of the cell (indicated by yellow arrows). In contrast, unlike MazF-mCherry, MazE-mCherry, the antitoxin MazE fused with mCherry, did not form foci but was dispersed homogenously in the cell (Fig. 2B).
In the MazEF-mCherry fusion, the mazF stop codon is not present. Instead, MazF is fused to a linker of 3 alanine residues followed by mCherry. Thus, the toxin MazF is the mCherry-fused protein, and the overproduced antitoxin MazE is combined with MazF. Unexpectedly, we found MazEF-mCherry to be localized completely differently from either MazF-mCherry or MazE-mCherry. While MazF-mCherry was localized mainly at the cell poles ( Fig. 2A) and MazE-mCherry was dispersed throughout the cell (Fig. 2B), MazEF-mCherry was localized mostly in the center of the cell (indicated by the white arrows in Fig. 2C, which represents one of 15 similar experiments).
In order to establish the preferred localization pattern of the toxin MazF by itself, and of the combined toxin MazF and antitoxin MazE (MazEF) in E. coli MC4100 relA 1 cells, we determined the positions of the foci relative to the nearest pole of the cell. This was done in 250 cells for each of the cases using Image-Pro software. We measured the positions of the foci relative to the nearest pole of the cell and converted these measurements to a percentage of cell length. We divided the cells into groups according to the relative measured positions of the foci. In 62% of the cells, MazF-mCherry foci were localized at the pole (Fig. 3A, lane 1). In the remaining 38% of the cells (Fig. 3A, lanes 2, 3, 4, and 5), MazF-mCherry foci were mostly localized between the pole and the center of the cell.
In contrast to the calculated position of the toxin MazF by itself (Fig. 3A), a completely different pattern of localization was observed when the toxin MazF was These experiments were carried out by the induction of mazEF proteins with 10 mM arabinose. Similar results were obtained by the use of 5 mM arabinose (see Fig. S3 and S4 in the supplemental material), which was the lowest concentration that permitted their visualization.
To support our discovery that the interaction of MazE with MazF causes a shift in the localization of MazF from the poles toward the center of the E. coli cells, we repeated this experiment with MazF-mCherry and MazE generated from different plasmids. We wished to recreate the situation in which the antitoxin is overexpressed. Here, only the toxin is fused with mCherry, and the antitoxin is overexpressed. For this purpose, we cotransformed the pBAD18-mazF-mcherry plasmid with a compatible pBAD33-mazE plasmid into E. coli MC4100 relA 1 cells and examined the localization of  Figure 2D demonstrates the intracellular localization in E. coli MC4100 relA 1 of MazF-mCherry in the presence of MazE, when the latter is derived from another plasmid. As shown once again, here MazF-mCherry is also not localized at the poles. Rather, it is localized in the center of the cells (indicated by white arrows).
The chpBIK toxin-antitoxin module. We asked if the localization patterns of the mazEF TA system protein products are unique, or whether they are similar to those of its homologue, the chpBIK TA system. As we had done for the mazEF system, to examine the intracellular localization of the protein products of the chpBIK TA system, we fused the fluorescent reporter mCherry to the toxin ChpBK, to the antitoxin ChpBI, and to the combined ChpBIK. Constructed fusions are illustrated schematically in Fig. S2 in the supplemental material.
Unlike the TA module mazEF, the TA module chpBIK is not involved in cell death, but rather inhibits E. coli cell growth (10). As shown in Fig. 1B, ChpBK-mCherry induction by 10 mM arabinose inhibits cell growth. On the other hand, there is no effect on cell growth when the antitoxin ChpBI is combined with ChpBK. Thus, the fusion with mCherry did not interfere with the functionality of the protein products of the chpBIK system.
We observed a clear pattern of intracellular localization for the toxin ChpBK-mCherry. ChpBK-mCherry formed foci only at the poles of the cells (Fig. 4A, indicated by white arrows, representing one out of 15 experiments). Similarly, the antitoxin ChpBI-mCherry fusion also formed foci at the poles of the cells (Fig. 4B, indicated by white arrows).
In the toxin-antitoxin ChpBIK-mCherry fusion, the stop codon of chpBK was not present; in its place, ChpBK was fused to a linker of three alanine residues followed by mCherry. Thus, the mCherry-fused protein is the toxin ChpBK. In addition to the overproduction of ChpBK-mCherry, ChpBI was also overproduced, generating a combined ChpBIK product in which ChpBK is fused with mCherry. When the toxin ChpBK was combined with the antitoxin ChpBI, the foci were localized at the cell poles (Fig. 4C, indicated by white arrows; one out of 15 typical experiments).
As described above for the mazEF TA module, we established the cellular localizations of the antitoxin ChpBI, the toxin ChpBK, and the combined toxin-antitoxin ChpBIK by determining the positions of each one relative to the nearest cellular pole. Figure 5 shows the calculated positions of the foci of ChpBK-mCherry, ChpBI-mCherry, and ChpBIK-mCherry in E. coli MC4100 relA 1 cells. As can be seen, nearly all of the foci of the protein products of the chpBIK toxin-antitoxin module have a clear pattern of localization in E. coli cells. The foci are localized at the pole of the cells. This was true in 86% of the cells in the case of the toxin ChpBK-mCherry (Fig. 5A, lane 1), in 88% of the cells in the case of antitoxin ChpBI-mCherry (Fig. 5B, lane 1), and in 86% of the cells in  (Fig. 5C, lane 1). Thus, in stark contrast to our results in the case of the interaction of the antitoxin MazE with the toxin MazF, the interaction of the antitoxin ChpBI with the toxin ChpBK did not cause a shift in the localization of the toxin ChpBK from the poles toward the centers of the cells.
The mqsRA toxin-antitoxin module. Having found that the protein products of the mazEF and chpBIK TA modules had different intracellular localization patterns, we  Figure 6A demonstrates the intracellular localization of the toxin MqsR fused with mCherry in the E. coli MC4100 relA 1 strain. As shown, two main intracellular localization patterns of MqsR were observed. First, in one subpopulation of the cells, MqsR-mCherry was located at the middle of the cell (indicated by yellow arrows). In another subpopulation of the cells, the toxin MqsR was localized between the pole and the middle of the cell (indicated by white arrows). Figure 6B represents the intracellular localization of the antitoxin MqsA fused with mCherry. In contrast to the toxin MqsR, MqsA-mCherry was dispersed homogenously in the cells.
Unlike the mazEF and chpBIK TA modules, in which the gene for the antitoxin precedes the gene for the toxin, in the mqsRA TA module, the mqsR toxin gene precedes the mqsA antitoxin gene. To recreate the conditions under which the toxin is fused Intracellular Localization of TA System Proteins in E. coli ® with mCherry and is expressed in the presence of the overproduced antitoxin, we needed to use a different approach than the one described above for the mazEF and chpBIK modules. We cloned the mqsA antitoxin gene to pBAD33 and transformed this newly constructed plasmid into E. coli MC4100 relA 1 cells harboring the pBAD-18-mqsR-mcherry plasmids. The cloning protocol is described in details in Materials and Methods. Figure 6C shows the intracellular localization of the toxin MqsR-mCherry in the presence of the overproduced antitoxin MqsA in E. coli MC4100 relA 1 cells. As shown, there are still two main subpopulations of the cells. In the first, the toxin MqsR is located at the middle of the cell (indicated by yellow arrows). In the second, the toxin MqsR forms foci between the pole and the middle of the cell (indicated by white arrows). Determining the relative distribution of the foci allowed us to clarify the difference between the localization of the toxin MqsR by itself or in the presence of the overproduced antitoxin MqsA (Fig. 7). As shown, the majority (54%) of the MqsR foci were localized between the pole and the middle of the cells (Fig. 7A, lane 2), 34% of the foci were localized in the middle of the cells (Fig. 7A, lane 3), and only 12% were localized at the poles of the cells (Fig. 7A, lane 1). In contrast, when the toxin MqsR was expressed in the presence of the overproduced antitoxin MqsA, most (57%) of the foci were localized the middle of the cells (Fig. 7B, lane 3), 38% between the pole and the middle of the cells (Fig. 7B, lane 2), and only 5% of the foci were localized at the poles of the cells (Fig. 7B, lane 1). Thus, in the presence of the overexpressed antitoxin MqsA, in part of the cells, the toxin MqsR is moved toward the middle of the cell. The rnlAB toxin-antitoxin system. Another type II TA system that we investigated was the rnlAB toxin-antitoxin module. As described above, we fused mCherry to the toxin RnlA and to the antitoxin RnlB. The primers used for cloning are listed in the supplemental material. The cloning protocols are described in detail in Materials and Methods. The constructed fusions are illustrated schematically in Fig. S2 in the supplemental material. Figure 8A demonstrates the intracellular localization of the toxin RnlA in E. coli MC4100 relA 1 cells. As can be seen, the toxin RnlA has a clear pattern of cellular localization. RnlA forms foci at the poles of most of the cells (indicated by white arrows). Similarly, the antitoxin RnlB fused with mCherry also formed foci at the poles in most of the cells (Fig. 8B, indicated by white arrows).
As in the mqsRA TA module, the rnlA toxin gene precedes the rnlB antitoxin gene, so we repeated the protocol used for the mqsRA module for the rnlAB system. We cloned the rnlB antitoxin gene into pBAD33 and transformed this newly constructed plasmid into cells harboring pBAD-18-rnlA-mcherry plasmids. The localization of the toxin RnlA expressed in the presence of the overproduced antitoxin RnlB reveals two major subpopulations of the cells. In one, the toxin RnlA is localized at the poles of the cells (Fig. 8C, indicated by white arrows), and in the second, the toxin RnlA is localized  (Fig. 8C indicated by yellow arrows). This shift was even more noticeable after we quantified the relative distributions of the RnlA-mCherry foci. The toxin RnlA (Fig. 9A, lane 1) and the antitoxin RnlB (Fig. 9B, lane 1) were mostly localized at the poles of the cells. In contrast, when the toxin RnlA was expressed in the presence of the overproduced antitoxin RnlB, a significant part (40%) of the foci were found not at the poles but rather toward the middle of the cells (Fig. 9C, lanes 2 and 3). Only 60% of the foci are localized at the poles of cells (Fig. 9C, lane 1). Thus, in the case of the rnlAB module, as in the cases of the mazEF and mqsRA modules, the presence of the antitoxin causes the movement of the toxin from the pole toward the center of the cells.

DISCUSSION
Although bacterial TA systems have been studied extensively (1,2,(16)(17)(18), until now, the localization of their protein products has not yet been investigated at all. Here, we examined the intracellular localization in E. coli of the protein products of four type II TA systems: mazEF, chpBIK, mqsRA, and rnlAB. Our observed data from fluorescence microscopy images and our calculated data are summarized in Table 1.
Intracellular localization of the protein products of the mazEF toxin-antitoxin module in E. coli cells. We observed that antitoxin MazE causes a shift in the localization of the toxin MazF from the pole to the middle of the cell.
Intracellular localization of the protein products of the chpBIK toxin-antitoxin module in E. coli cells. In contrast to the results for the mazEF module, each of the protein products of the chpBIK TA module, namely the toxin ChpBK, the antitoxin ChpBI, and the combined toxin-antitoxin ChpBIK, is localized at the cell poles.
Intracellular localization of the protein products of the mqsRA toxin-antitoxin module in E. coli cells. The toxin MqsR is localized either in the region between the pole and the middle of the cell or in the middle of the cells. The antitoxin MqSA is dispersed homogenously in the cells. When the toxin MqsR was expressed in the presence of the antitoxin MqsA, MqsR moved toward the middle of the cells.
Intracellular localization of the protein products of the rnlAB toxin-antitoxin module in E. coli cells. Both the toxin RnlA and the antitoxin RnlB are localized mostly at the poles of the cells. When the toxin RnlA was expressed in the presence of the antitoxin RnlB, although most of the RnlA molecules remained at the pole of the cells, a significant part of the foci had been moved from the poles toward the middle of the cells.
One could expect that the proteins which belong to the same group of the TA systems (type II) would have the same pattern of the intracellular localization. We were surprised to discover that each of the TA modules studied here had its own pattern of localization for the toxin, the antitoxin, and the toxin-antitoxin complex. However, with the exception of the chpBIK TA module, for each TA module, the localization of the toxin-antitoxin complex was different from the localization of the toxin itself. The presence of the antitoxin causes a shift in the localization of the toxin from the pole toward the center of the cells. Thus, we suggest that our results reveal that the antagonistic effect of antitoxins on their cognate toxins might not be based only on their direct structural interactions but may also be caused by changing the intracellular localization of the toxin.
Future experiments should provide explanation of why the toxins tend to migrate toward the poles and the reason for the shift toward the center of the cell by the antitoxin. Does this have anything with the endoribonuclease activity? Interaction with actin-like cytoskeleton proteins? Interaction with ClpXP? In the case of the mazEF proteins, they may interact with mazEF-specific components, such as the peptide EDF (NNWNN) (19) and components of the stress-induced translation machinery (STM) (20) (see Fig. S5 in the supplemental material).

MATERIALS AND METHODS
Primers for cloning were obtained from Integrated DNA Technologies (IDT, Hudson, NH). Other materials and suppliers were as follows: PrimeSTAR HS DNA polymerase from TaKaRa, Japan; T4 ligase,  (21) and either pBAD18-mcherry (22,23) or PBAD33 (22) plasmids in which we constructed the protein-mCherry fusions or cloned the genes of the antitoxins (see below).
Construction of the plasmids. (i) pBAD18-mCherry cloning. The genes encoding the proteins MazE, MazF ChpBI, ChpBK, MqsR, MqsA, RnlA, and RnlB (lacking stop codons) were amplified by PCR from E. coli MC4100 relA 1 genomic DNA. For this purpose, we used the primers that introduced EcoRI (forward primer) and SacI (reverse primer) restriction sites and a gene sequence encoding a three-alanine linker. We used primers that introduced EcoRI (forward primer) and SacI (reverse primer) restriction sites and a gene sequence encoding a linker of three alanine residues. The resulting amplicons were digested with EcoRI and SacI and ligated into pBAD18-mcherry plasmids. The following DNA fusions were constructed: mazE-mcherry, mazF-mcherry chpBI-mcherry, chpBK-mcherry, mqsR-mcherry, mqSAmcherry, rnlA-mcherry, and rnlB-mcherry. In the protein products of these DNA fusions, mCherry was fused to the C terminus of the protein of interest through a three-alanine linker. In the case of the combined antitoxin and toxin MazEF-mCherry fusion, the gene sequence of the mazEF operon (lacking the stop codon of mazF) was amplified by PCR and cloned into pBAD18-mcherry plasmid as described above. The following protein products were generated: MazE (by itself without mCherry) and MazF-mCherry, where mCherry was fused to the C terminus of the toxin MazF through a three-alanine linker. In the case of the combined antitoxin and toxin ChpBIK-mCherry fusion, the gene sequence of the chpBIK operon (lacking the stop codon of chpBK) was amplified by PCR and cloned into the pBAB18mcherry plasmid as described above. The following protein products were generated: ChpBI and ChpBK-mCherry. mCherry was fused to the C terminus of the toxin ChpBK through a three-alanine linker. The primers for cloning and constructed plasmids are listed in the supplemental material.
(ii) pBAD33 cloning. In addition to the constructions using pBAD18, gene sequences of mazE, mqsA, and rnlB (lacking stop codons) were amplified by PCR from E. coli MC4100 genomic DNA using the primers that introduced SacI (forward) and XbaI (reverse) sites and a three-alanine linker. The resulting amplicon was cloned into the pBAD33 plasmid. The primers for cloning and constructed plasmids are listed in the supplemental material.
Growth conditions. Cloned plasmids were transformed into the E. coli MC4100 relA 1 strain using electroporation. Single colonies were grown overnight in 10 ml of LB medium with 100 mg/ml ampicillin at 37°C and 220 rpm. A 200-ml aliquot of overnight culture was subcultured (1:20 dilution) in 10 ml of LB in the presence of 100 mg/ml ampicillin. The cells were grown at 37°C and 220 rpm to an optical density at 600 nm (OD 600 ) of 0.4. To express protein-mCherry fusions, 10 mM arabinose was added to the cell culture, and the cells were incubated at 37°C for 4 h.
Slide preparation. Cells were harvested by centrifugation at 14,000 rpm for 2 min and washed twice with phosphate-buffered saline (PBS). For labeling, cell membrane cells were resuspended with 20 ml of FM-143FX dye. Then, for fixation, 100 ml of 4% formalin was applied for 15 min, and the cells were washed twice with PBS (pH 7.2) and resuspended in 10 ml of PBS and 10 ml of polyvinyl alcohol (PVA). Aliquots (10 ml) of resuspended cells were transferred to microscope slides, covered with coverslips, and sealed with nail polish. Slides were dried and stored at 220°C.
Fluorescent microscopy. Protein-mCherry fusions were examined using a Zeiss LSM 710 laser scanning microscope with a Plan-Apochromat 63Â/1.40 oil lens. mCherry fluorescence was measured as excitation/emission at 561/578 to 696 nm. Images were recorded and analyzed using ZEN 2012 (Zeiss) and Image-Pro Plus 6.0 software (Media Cybernetics, Inc.).
Viability assays. Single colonies of the E. coli MC4100 relA 1 strain were grown overnight in 10 ml of M9 medium with 100 mg/ml ampicillin at 37°C, 220 rpm. A 200-ml aliquot of overnight culture was subcultured (1:20 dilution) in 10 ml of LB in the presence of 100 mg/ml ampicillin. The cells were grown at 37°C and 220 rpm to an OD 600 of 0.4. Protein-mCherry fusions were expressed from pBAD18-mcherryderived plasmids induced by 10 mM arabinose. MazF was expressed from PSA1 plasmid induced by 1 mM isopropyl-b-D-thiogalactopyranoside (IPTG). After 4 h of incubation at 37°C the samples were centrifuged at 14,000 rpm for 2 min. The supernatants were removed, and the pellets were resuspended in 0.5 ml of prewarmed M9 medium. The samples were serially diluted in prewarmed M9, plated on prewarmed LB plates, and incubated at 37°C. The percent survival was determined by dividing the number of colonies obtained from the "experiment" sample by the number of colonies obtained from the MC4100 relA 1 sample used as the control.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. TEXT S1, DOCX file, 0.01 MB.