A Linear Pentapeptide Is a Quorum-Sensing Factor Required for mazEF-Mediated Cell Death in Escherichia coli

See allHide authors and affiliations

Science  26 Oct 2007:
Vol. 318, Issue 5850, pp. 652-655
DOI: 10.1126/science.1147248


mazEF is a toxin-antitoxin module located on many bacterial chromosomes, including those of pathogens. Here, we report that Escherichia coli mazEF-mediated cell death is a population phenomenon requiring a quorum-sensing molecule that we call the extracellular death factor (EDF). Structural analysis revealed that EDF is a linear pentapeptide, Asn-Asn-Trp-Asn-Asn. Each of the five amino acids of EDF is important for its activity.

Programmed cell death (PCD) is generally associated with eukaryotic multicellular organisms (1, 2). However PCD systems have also been observed in bacteria (311). One of these systems is mediated by the toxinantitoxin module (mazEF) located in many bacterial chromosomes (1113) and mainly studied in Escherichia coli (3, 10, 11). E. coli mazF encodes a stable toxin, MazF (3), which is a sequence-specific endoribonuclease that preferentially cleaves single-stranded mRNAs at ACA sequences (14). mazE encodes a labile antitoxin, MazE, that counteracts the action of MazF (3). E. coli mazEF is a stress-induced toxin-antitoxin module. Thus, any stressful condition that prevents the expression of mazEF will lead to a reduction of MazE levels in the cell, permitting the MazF toxin to act. Such stresses include the transient inhibition of transcription and/or translation by antibiotics such as rifampicin, chloramphenicol, and spectinomycin, as well as DNA damage caused by thymine starvation, mitomycin C, nalidixic acid, and ultraviolet irradiation (15, 16).

We previously suggested that E. coli mazEF-mediated cell death is a population phenomenon (11). Here we confirm that E. coli mazEF-mediated cell death was dependent on the density of the bacterial population (fig. S1). Adding rifampicin for a short period to inhibit transcription led to mazEF-mediated cell death at densities of 3 × 108 or 3 × 107 cells/ml, but not at 3 × 105 or 3 × 104 cells/ml (fig. S1). Consequently, we examined whether the supernatant of a dense culture could restore mazEF-mediated cell death in a diluted culture. To this end, we added the supernatant of a dense culture to a diluted culture and then induced mazEF-mediated cell death by addition of rifampicin (Fig. 1A), chloramphenicol (fig. S2A), or trimethoprim (fig. S2B). We concluded that mazEF-mediated cell death requires an “extracellular death factor” (EDF). We observed EDF activity in E. coli cultures during logarithmic growth but not during stationary growth (fig. S3). This finding correlates with our previous results showing that E. coli mazEF-mediated cell death occurs during exponential phase but not during stationary phase (15). Here, we show that stationary-phase resistance to PCD results from a lack of EDF activity.

Fig. 1.

(A) A supernatant of a dense culture can restore mazEF-mediated cell death to a diluted culture. E. coli MC4100relA+ (wild type, WT) and MC4100relA+ΔmazEFmazEF) were grown logarithmically (17). At a density of 2.5 × 108 cells/ml, samples were either not diluted (dense) or diluted to a density of 3 × 104 cells/ml in prewarmed M9 medium (M9) or in a prewarmed supernatant of a dense culture (SN) (17). The samples were incubated without shaking at 37°C for 10 min and for another 10 min with rifampicin (10 μg/ml). CFU, colony-forming units. (B) Effects of various pHs on the SN. SN was incubated at pH 3, 5, 7, 9, or 11 for 2 hours and titrated to pH 7. Its ability to restore mazEF-mediated cell death to a diluted culture was determined as in (A). (C) Milliabsorbance at 220 nm, determined during elution from the HPLC column of the purified supernatant (fig. S5). (D) EDF activity plotted as a function of elution time from the HPLC column.

Our preliminary characterization of EDF revealed that it is sensitive to extreme pH (Fig. 1B), high temperatures (80° to 100°C) (fig. S4A), and proteinase K (fig. S4B). For chemical characterization we purified EDF from a large volume of a supernatant of an E. coli mid-exponential phase culture grown in a minimal medium. The supernatant was collected and fractions were separated on a C-18 SepPak cartridge (fig. S5) (17). Active fractions were purified by high-performance liquid chromatography (HPLC), and EDF activity comigrated with a single peak with an elution time of 20 min (Fig. 1, C and D). To avoid damaging EDF in acidic conditions (Fig. 1B), we performed electrospray ionization mass spectrometry (ESI-MS) at neutral pH and obtained a peak of 661 daltons (Fig. 2A). This peak was not observed during standard MS analysis at pH 2.5 (fig. S6). Fragmentation (MS/MS) analysis of the material from this 661-dalton peak revealed that EDF is a linear peptide with the amino acid sequence Asn-Asn-Trp-Asn-Asn (NNWNN) (Fig. 2, B and D). The four Asn residues in EDF are vulnerable to deamidation under acidic conditions (18) normally used for ESI-MS.

Fig. 2.

The chemical nature of EDF identified by mass spectrometry. The chemical composition of the purified peak found to have EDF activity was carried out by ESI-MS (QT of 2 Micromass instrument). (A) Peaks ranging from 200 to 662 MW, among them one of 661 MW (marked by an arrow). (B) The MS/MS spectrum of the 661-MW peak revealed a peptide with the amino acid sequence NNWNN. (C) The chemically synthesized EDF-NNWNN can restore mazEF-mediated cell death to a diluted culture. E. coli MC4100relA+ cells were grown as in Fig. 1A. At a density of 2.5 × 108 cells/ml, samples were either not diluted (dense) or diluted to 3 × 104 cells/ml in prewarmed M9 medium (diluted culture) or in a prewarmed M9 applied with chemically synthesized EDF (2.5 ng/ml) (diluted culture + EDF). Samples were incubated with rifampicin as in Fig. 1A. (D) Structure of EDF as determined by nuclear magnetic resonance analysis (table S1).

To test whether the NNWNN peptide is indeed EDF, we chemically synthesized an identical peptide and tested it for biological activity. When added to a diluted E. coli culture, the synthetic peptide enabled mazEF-mediated cell death when induced in different E. coli strains by rifampicin (Fig. 2C and fig. S7) and by several other stressful conditions (fig. S8). At a wide range of concentrations (2.5 to 200 ng/ml), the killing activity of the chemically synthesized EDF was mazEF-dependent; virtually no EDF activity was observed in a mazEF knockout strain (Fig. 3). We further confirmed the mazEF dependence of EDF by transforming the mazEF knockout strain with a plasmid harboring the mazEF module. The presence of this module on the plasmid completely restored the killing activity of EDF (Fig. 3). Note that at higher concentrations of EDF (>200 ng/ml), we observed a reduction of viability even in the mazEF knockout strain (Fig. 3). We assume that at high concentrations, EDF acts less specifically, possibly by inducing some other PCD systems or by inactivating other essential components.

Fig. 3.

The response to chemically synthesized EDF is mazEF-dependent. E. coli MC4100relA+ (WT), MC4100relA+ ΔmazEFmazEF), and MC4100relA+ ΔmazEF pKK223mazEFmazEF pKK223mazEF) were grown as in Fig. 1A. Strain MC4100relA+ΔmazEF pKK223mazEF was grown in M9 applied with ampicillin (100 μg/ml). At a density of 2.5 × 108 cells/ml, samples were diluted in pre-warmed M9 to 2.5 × 104 cells/ml and various concentrations of chemically synthesized wild-type EDF were applied. Samples were incubated with rifampicin (10 μg/ml) as in Fig. 1A.

To determine the role of each residue in EDF activity, we prepared five synthetic peptides, each with a Gly replacing one of the amino acids in the natural EDF sequence, and examined their biological activity in the wild-type strain. Changing the first or fifth amino acid abolished EDF killing activity, whereas changing the second, third, or fourth amino acid led to a moderate reduction in killing activity (Fig. 4A). Thus, each amino acid in EDF is important for its activity, with the N- and C-terminal residues being the most critical. A similar hierarchy of amino acid importance was obtained when we examined whether mutant EDF molecules are able to inhibit wild-type EDF activity (fig. S9). EDFs mutated at the terminal amino acids (1 and 5) were efficient inhibitors of EDF activity (fig. S9); however, mutated EDF peptides in which glycine replaced the amino acids in positions 2, 3, or 4 inhibited EDF activity only when the concentration of wild-type EDF was low. These results (Fig. 4A and fig. S9) indicated that amino acids 1 and 5 had roles similar to each other, as did amino acids 2 through 4.

Fig. 4.

NNWNN is the optimal molecule for EDF activity. (A) Each of the five amino acids is important for EDF activity. (B) The importance of size and external Asn residues for EDF activity. Chemically synthesized EDF (marked by NNWNN-EDF) or its modified derivatives were added at various concentrations to diluted cultures of E. coli MC4100relA+. No EDF was added to a control culture (marked “None”). Samples were incubated with rifampicin as in Fig. 1A.

We further examined several characteristics of EDF required for its activity. Using synthetic peptides (Fig. 4B), we found that (i) the tripeptide NWN does not have EDF activity, whereas the heptapeptide NNNWNNN has partial activity; (ii) the presence of the same amino acid at external positions 1 and 5 of EDF seems to be important, and EDF activity was only partially reduced with Gly instead of Asn at these positions; and (iii) the presence of an amide at the external positions probably has a role in EDF activity, as the replacement of Asn by Gln (Q) at either end of the pentapeptide (QNWNN or NNWNQ)—that is, a substitution that carries an amide and is structurally related to Asn—led to only a partial reduction in EDF activity. We conclude that NNWNN appears to be the optimal sequence for EDF function.

Using database analysis, we searched the E. coli genome for DNA sequences corresponding to the amino acid sequence NNWNN. To our surprise, only five open reading frames predicted peptide similarity to NNWNN (fig. S10). The deletion of only two genes prevented the production of an active EDF (fig. S11): zwf encoding NNWDN (D = Asp) and ygeO encoding NNWN. The zwf product, carrying the sequence NNWDN, may be the precursor of EDF, and a subsequent amidation step may generate the full NNWNN sequence. Amidation may occur either before or after the cleavage of the precursor by one of E. coli proteases. Our results indicated that Asn synthetase A (19) is involved; deleting the gene asnA prevented production of active EDF, whereas deleting asnB (encoding Asn synthetase B) did not (fig. S11). In addition, our results revealed that the product of ygeO gene is also involved in the generation of EDF (fig. S11).

Bacteria communicate with one another via a variety of quorum-sensing signal molecules, or autoinducers, which have been found to be involved in bioluminescence, virulence, biofilm formation, sporulation, mating, and competence for DNA uptake, among other responses (2025). Quorum sensing provides a mechanism for bacteria to monitor each other's presence and to modulate gene expression in response to population density. Our results show that mazEF-mediated cell death is a quorum-sensing process in which the EDF peptide is the autoinducer. The cellular component(s) directly interacting with EDF are currently under investigation, as are the specific stage(s) in the mazEF-mediated death network that is affected. The quorum-sensing process involved in mazEF-mediated cell death is of interest for two reasons: (i) No other peptide besides EDF, to our knowledge, has been reported to be involved in quorum sensing in E. coli, and (ii) EDF appears to be a type of peptide distinct from those known to be involved in quorum sensing among Gram-positive bacteria, because EDF is synthesized from an enzyme (Zwf) (23, 24) and because it is involved in bacterial PCD.

Increasing experimental evidence indicates that bacteria seldom behave as isolated organisms, and in nature they more often exist as communities capable of intercellular communication and concerted social behavior (21, 25, 26). We previously suggested that bacterial PCD is yet another manifestation of bacterial multicellularity (11) and have shown that mazEF prevents the spread of phage infection (27). Here we report that E. coli mazEF-directed bacterial death is mediated by an extracellular penta-peptide (EDF) and that bacterial PCD appears to depend on cell-to-cell communication. When challenged by stressful conditions that trigger mazEF-mediated cell death, the bacterial population can act like a multicellular organism in which a subpopulation of cells dies and releases nutrients (11) and/or signaling molecules, and/or clears phages (27), thereby permitting the survival of the bacterial population as a whole. In addition, the death of a subpopulation may enable biofilm formation by the release of component(s) providing the biofilm matrix (28). Finally, on a practical level, the ability to chemically synthesize an identical peptide carrying EDF activity may be a lead for a new class of antibiotics that specifically trigger bacterial cell death.

Supporting Online Material

Materials and Methods

SOM Text

Figs. S1 to S11

Table S1


References and Notes

View Abstract

Navigate This Article