Physical and Functional Interplay between MazF₁^Bif and Its Noncognate Antitoxins from Bifidobacterium longum

ABSTRACT

Bifidobacterium longum strain JDM301, a widely used commercial strain in China, encodes at least two MazEF-like modules and one RelBE-like toxin-antitoxin (TA) system in its chromosome, designated MazE₁F₁^Bif, MazE₂F₂^Bif, and RelBE^Bif, respectively. Bacterial TA systems play an important role in several stress responses, but the relationship between these TA systems is largely unknown. In this study, the interactions between MazF₁^Bif and MazE₂^Bif or RelB^Bif were assessed in B. longum strain JDM301. MazF₁^Bif caused the degradation of tufA^Bif mRNA, and its toxicity was inhibited by forming a protein complex with its cognate antitoxin, MazE₁^Bif. Notably, MazF₁^Bif toxicity was also partially neutralized when jointly expressed with noncognate antitoxin MazE₂^Bif or RelB^Bif. Our results show that the two noncognate antitoxins also inhibited mRNA degradation caused by MazF₁^Bif toxin. Furthermore, the physical interplay between MazF₁^Bif and its noncognate antitoxins was confirmed by immunoprecipitation. These results suggest that MazF₁^Bif can arrest cell growth and that MazF₁^Bif toxicity can be neutralized by its cognate and noncognate antitoxins. These results imply that JDM301 uses a sophisticated toxin-antitoxin interaction network to alter its physiology when coping with environmental stress.

IMPORTANCE Although toxin-antitoxin (TA) systems play an important role in several stress responses, the regulatory mechanisms of multiple TA system homologs in the bacterial genome remain largely unclear. In this study, the relationships between MazE₁F₁^Bif and the other two TA systems of Bifidobacterium longum strain JDM301 were explored, and the interactions between MazF₁^Bif and MazE₂^Bif or RelB^Bif were characterized. In addition, the mRNA degradation activity of MazF₁^Bif was demonstrated. In particular, the interaction of the toxin with noncognate antitoxins was shown, even between different TA families (MazF₁^Bif toxin and RelB^Bif antitoxin) in JDM301. This work provides insight into the regulatory mechanisms of TA systems implicated in the stress responses of bifidobacteria.

INTRODUCTION

Type II toxin-antitoxin (TA) systems are ubiquitous in free-living bacteria and consist of adjacent genes encoding a toxin and antitoxin in a single operon (1, 2). The first gene encodes a relatively labile antitoxin and the second gene encodes a stable toxin (2). Bacterial TA systems are considered stress-responsive elements (3, 4). Under normal conditions, antitoxin proteins are abundant, which can neutralize the action of its cognate toxin. The toxin and its cognate antitoxin interact to form an inactive toxin-antitoxin protein complex. The complex or the antitoxin itself acts as a transcriptional autorepressor of the operon (5). However, in response to adverse growth conditions, the amount of antitoxin decreases and the toxin is released, leading to cell death or growth arrest by the toxin acting on its intracellular target (5–7). Transcription is repressed by the antitoxin alone or the toxin-antitoxin complex upon binding to the palindrome upstream of the operon (8). The antitoxin protein is unstable relative to the toxin, since it is susceptible to cleavage by ClpP or/and Lon proteases (9, 10). When stress conditions lead to an increased expression of proteases, the pool of antitoxins is reduced by proteolysis, leading to a relative increase in module transcription, which results in an excess of toxin (5). The free toxin then acts on its target, resulting in transient growth arrest or cell death if antitoxin synthesis does not recover quickly enough (5, 11, 12).

Some TA systems in Escherichia coli are activated under environmental stress, resulting in cell stasis, after which they can recover under favorable conditions (13, 14). The MazEF module (toxin MazF and antitoxin MazE) is a well-characterized TA system of E. coli that is involved in various stress conditions, such as nutritional stress and antibiotic exposure (15–17). Stress conditions lead to the degradation of the antitoxin (MazE) and the release of the free toxin (MazF). The free MazF prevents translation by cleaving RNAs, resulting in cell death or growth arrest (18–20). The RelBE module (toxin RelE and antitoxin RelB) is another TA system in E. coli. Free RelE can induce global inhibition of translation and the arrest of cell growth by cleaving RNAs (21–23). Among them, the tufA (elongation factor Tu) mRNAs are targets of free RelE and HigB (toxin protein of the TA system HigBA) in E. coli (22–24).

Although TA systems are distributed widely in free-living bacteria, which can encode more than one TA system, almost all intracellular bacteria are devoid of TA systems, suggesting that these systems are stress-response elements, which are crucial for bacterial survival in fluctuating environmental conditions (16, 25–27). However, genomes of free-living bacteria usually encode many TA system homologs (28, 29). The relationships between these TA systems in the bacterial genome are largely unknown. Recently, multiple toxin-antitoxin systems were reported to cooperate to increase the persister frequency in E. coli (14). Interactions were also found among three RelB-like TA systems and even between different TA families (MazF toxins and VapB antitoxins) in Mycobacterium tuberculosis (30, 31). Nineteen genes of TA systems belonging to the MazEF and RelBE families were found by an in silico analysis of 36 sequenced genomes from several strains of bifidobacteria (32). The whole genome of Bifidobacterium longum strain JDM301, a widely used commercial strain in China, was completely sequenced (33). A total of 11 putative TA systems were found by bioinformatic analysis of the JDM301 genome (10). The JDM301 genome harbors at least two pairs of functional mazEF-like loci (BLJ_811-BLJ_812 and BLJ_864-BLJ_865) and one pair of functional relBE-like loci (BLJ_989-BLJ_990), designated MazE₁F₁^Bif, MazE₂F₂^Bif, and RelBE^Bif, respectively (10, 34, 35). In our previous report, we showed that MazE₁F₁^Bif was activated under acid stress (10). However, the roles of these systems in the stress response of JDM301 remain largely unclear. The relationships between MazE₁F₁^Bif and the other two TA systems were explored in this study.

In this study, the physical and functional interplay between toxin MazF₁^Bif and its noncognate antitoxins was characterized. In addition, the mRNA degradation activity of MazF₁^Bif was shown. In particular, noncognate interactions were found, even between different TA families (MazF₁^Bif toxin and RelB^Bif antitoxin) in B. longum. Interactions with noncognate antitoxins might reduce the toxicity of MazF₁^Bif in vivo. This work provides insight on the interplay between different TA systems in B. longum, which helps the bacterium adapt to harsh environmental conditions.

RESULTS

MazF₁^Bif and its cognate antitoxin, MazE₁^Bif, form a complex.

To show the direct interaction between MazE₁^Bif and MazF₁^Bif, the MazE₁^Bif and MazF₁^Bif genes were both cloned into a single pET28a expression vector under one promoter in accordance with our previous report (10). Thus, only MazE₁^Bif was expressed as a His₆-tagged fusion protein, while MazF₁^Bif was expressed as a Myc-tagged fusion protein. The recombinant proteins were expressed and purified from Ni-nitrilotriacetic acid (Ni-NTA) resin. The purified proteins were subjected to Western blot analysis using anti-His₆ or anti-Myc monoclonal antibodies. The resulting bands observed corresponded to MazE₁^Bif (12.6 kDa) with the His₆ tag and to MazF₁^Bif (14.4 kDa) with the Myc tag (Fig. 1). Thus, His₆-MazE₁^Bif and MazF₁^Bif-Myc were copurified from Ni²⁺-chelating Sepharose resin to show that MazF₁^Bif and its cognate antitoxin, MazE₁^Bif, formed a complex.

FIG 1.

Interaction of Myc-tagged MazF₁^Bif and His₆-tagged MazE₁^Bif recombinant proteins. Recombinant proteins were expressed and purified using Ni-NTA resin. The purified proteins were detected by Western blotting with anti-His₆ (A) and anti-Myc (B) monoclonal antibodies. Recombinant proteins were expressed from IPTG-induced E. coli harboring pET-E₁ or pET-F₁(Myc). M, molecular mass markers; 1, lysate of E. coli harboring pET-F₁(Myc); 2, purified products of E. coli harboring pET-F₁(Myc); 3, purified recombinant proteins from E. coli harboring pET-E₁. (C) MazE₁^Bif-His₆, including the His₆ tag at its N-terminal end. (D) MazF₁^Bif-Myc, including the Myc tag at its C-terminal end. Recombinant proteins were expressed from IPTG-induced E. coli harboring pET-E₁F₁(Myc). Both the MazE₁^Bif-His₆ and MazF₁^Bif-Myc fusion proteins were detected at their expected molecular masses. M, molecular mass markers; 1, eluates of absorbed lysate from uninduced E. coli harboring pET-E₁F₁(Myc); 2, eluates of absorbed lysate from IPTG-induced E. coli harboring pET-E₁F₁(Myc); 3, purified recombinant proteins from IPTG-induced E. coli harboring pET-E₁F₁(Myc).

mRNA degradation by MazF₁^Bif is antagonized by its cognate antitoxin, MazE₁^Bif.

The tufA^Bif gene was cloned into the promoter of the arabinose operon of pBAD/HisB to produce tufA^Bif mRNA in E. coli. pACYCDuet-1, pAD-F₁, or pAD-F₁E₁ was transformed into E. coli with pBA-tufA for the coexpression of MazF₁^Bif or MazF₁^Bif and MazE₁^Bif with tufA^Bif mRNA. Quantitative real-time PCR (qRT-PCR) was used to determine whether MazF₁^Bif mediates tufA mRNA degradation in strain JDM301 and whether the activity of MazF₁^Bif is inhibited by MazE₁^Bif. Our results show that the induction of MazF₁^Bif in E. coli decreased tufA^Bif mRNA levels compared with levels when tufA^Bif was transcribed alone, while tufA^Bif mRNA levels increased when MazF₁^Bif was coexpressed with MazE₁^Bif compared with levels in E. coli expressing only MazF₁^Bif, indicating that MazE₁^Bif alleviates the degradation of tufA^Bif mRNA by MazF^Bif (Fig. 2). These results suggest that MazF₁^Bif causes the degradation of tufA^Bif mRNA and that the activity of MazF₁^Bif is alleviated by its cognate, MazE₁^Bif.

FIG 2.

MazF₁^Bif is an mRNA interferase that is inhibited by its cognate antitoxin, MazE₁^Bif. Relative transcript levels of tufA^Bif were determined in E. coli expressing tufA^Bif with pACYCDuet-1, pAD-F₁, or pAD-F₁E₁. The strains were grown with 0.2% arabinose for 2 h to induce tufA^Bif expression. Then, 1 mM IPTG was added to induce MazF₁^Bif or MazF₁^Bif and MazE₁^Bif expression. After 3 h, 200 μg/ml rifampin was added. Samples were collected at the indicated time points after rifampin addition. The levels of tufA^Bif mRNA were monitored by qRT-PCR (normalized to the 16S rRNA transcript level). The values presented are the averages from three independent experiments, and error bars represent the standard deviations. A two-way analysis of variance with Bonferroni posttest was used to obtain P values for each time point: a, P < 0.05 versus pACYCDuet-1; b, P < 0.05 versus pAD-F₁E₁.

MazF₁^Bif physically interacts with its noncognate antitoxin protein.

Plasmid pACYCDuet-1, pAD-F₁E₁, pAD-F₁E₂, or pAD-F₁B was introduced into E. coli to simultaneously express His-tagged MazF₁^Bif and S-tagged antitoxins (MazE₁^Bif, MazE₂^Bif, or RelB^Bif). Subsequently, coimmunoprecipitation was performed to detect the physical interactions between the toxin MazF₁^Bif and each of the three antitoxin proteins, including its cognate antitoxin, MazE₁^Bif, and noncognate antitoxins MazE₂^Bif and RelB^Bif. An anti-His antibody against the His-tagged MazF₁^Bif and an anti-S antibody against the S-tagged antitoxins were used in coimmunoprecipitation experiments. As shown in Fig. 3, noncognate toxin-antitoxin interactions (MazF₁^Bif with MazE₂^Bif and MazF₁^Bif with RelB^Bif) and a cognate toxin-antitoxin interaction (MazF₁^Bif with MazE₁^Bif) were observed by immunoprecipitation. The interaction between the toxin MazF₁^Bif and the antitoxin MazE₂^Bif was only observed by immunoprecipitation using the anti-S antibody. The interaction between the toxin MazF₁^Bif and the antitoxin MazE₁^Bif was also confirmed by immunoprecipitation using only the anti-S antibody. The reason for this is unclear; however, steric hindrance stemming from the presence of the His tag might be responsible (30). Our results demonstrated that toxin MazF₁^Bif and its noncognate antitoxins physically interact with each other, indicating that the noncognate antitoxins of MazF₁^Bif, particularly RelB^Bif, may act in lieu of its cognate antitoxin, MazE₁^Bif, to inhibit toxicity.

FIG 3.

Molecular interactions between MazF₁^Bif and cognate or noncognate antitoxin proteins are confirmed by coimmunoprecipitation assays. Cell lysates or proteins immunoprecipitated with the anti-His₆ or anti-S antibodies were analyzed by immunoblotting using anti-His₆ or anti-S antibodies. M, molecular mass markers; 1, E. coli carrying pACYCDuet-1 (an empty vector) used as the control; 2, E. coli carrying pAD-F₁E₁; 3, E. coli carrying pAD-F₁E₂; 4, E. coli carrying pAD-F₁B. Asterisks indicate the bands corresponding to MazF₁^Bif-His₆, MazE₁^Bif-S, MazE₂^Bif-S, or RelB^Bif-S. The bands corresponding to the heavy chains of the anti-His₆ or anti-S antibody are indicated by arrows.

MazF₁^Bif inhibits the growth of E. coli, and the inhibition is alleviated by its noncognate antitoxin proteins.

Several growth curves of E. coli strains carrying pACYCDuet-1, pAD-F₁, pAD-F₁E₁, pAD-F₁E₂, or pAD-F₁B in the presence of 1 mM IPTG (isopropyl-β-d-thiogalactopyranoside) were plotted to determine whether the toxicity of the MazF₁^Bif toxin could be inhibited by noncognate antitoxins in vivo. For E. coli strains containing mazF₁^Bif alone, growth inhibition was observed upon IPTG induction compared with that of the cells containing an empty vector (Fig. 4A). Furthermore, the cells coexpressing MazE₁^Bif and MazF₁^Bif grew better than those expressing MazF₁^Bif alone but worse than those containing the empty vector (Fig. 4A). Notably, when the MazF₁^Bif toxin was induced in the presence of noncognate antitoxin MazE₂^Bif or RelE^Bif, growth inhibition was alleviated (Fig. 4B and C), indicating that cell growth inhibition caused by MazF₁^Bif can be rescued by the activity of the noncognate antitoxin MazE₂^Bif or RelE^Bif. These rescue experiments enabled the detection of interactions that may be less stable in vitro. Our results demonstrated interactions between MazF₁^Bif and noncognate antitoxins RelE^Bif and MazE₂^Bif, which act in lieu of MazE₁^Bif to inhibit the activity of MazF₁^Bif.

FIG 4.

Interactions between MazF₁^Bif and its cognate antitoxin, MazE₁^Bif, or noncognate antitoxins affect cell growth. The growth characteristics of E. coli carrying pACYCDuet-1, pAD-F₁, and pAD-F₁E₁ (A), pAD-F₁E₂ (B), or pAD-F₁B (C) were analyzed by measuring absorbance (OD₆₀₀) following induction with 1 mM IPTG. The values presented are the averages from three independent experiments, and error bars represent the standard deviations. A two-way analysis of variance with Bonferroni posttest was used to obtain P values for each time point: c, P < 0.001 versus pACYCDuet-1; f, P < 0.001 versus pAD-F₁E₁, pAD-F₁E₂, or pAD-F₁B.

MazF₁^Bif-induced mRNA degradation is antagonized by noncognate antitoxins.

The results above showed that the MazF₁^Bif toxin associates with the noncognate antitoxin RelE^Bif or MazE₂^Bif to alleviate growth inhibition caused by MazF₁^Bif, implying that the toxic effect of MazF₁^Bif can be antagonized by noncognate antitoxins. To test this hypothesis, E. coli was transformed with pBA-tufA and pACYCDuet-1 (a blank vector), pBA-tufA and pAD-F₁, pBA-tufA and pADuet-F₁E₂, or pBA-tufA and pAD-F₁B. When RelB^Bif or MazE₂^Bif was coexpressed with MazF₁^Bif, the level of tufA^Bif mRNA increased in comparison to the level of tufA^Bif in E. coli expressing MazF₁^Bif alone (Fig. 5). Our results suggest that MazF₁^Bif-induced mRNA degradation is inhibited by the noncognate antitoxins RelE^Bif and MazE₂^Bif.

FIG 5.

Noncognate antitoxin proteins counteract the mRNA interferase activity of MazF₁^Bif. Relative transcript levels of tufA^Bif were determined in E. coli expressing tufA^Bif with pACYCDuet-1, pAD-F₁, and pAD-F₁E₂ (A) or pAD-F₁B (B). The strains were grown with 0.2% arabinose for 2 h to induce tufA^Bif expression. Then, 1 mM IPTG was added to induce MazF₁^Bif, MazF₁^Bif and MazE₂^Bif, or MazF₁^Bif and RelB^Bif expression. After 3 h, 200 μg/ml rifampin was added. Samples were collected at the indicated time points after rifampin addition and the levels of tufA^Bif mRNA were monitored by qRT-PCR (normalized to the 16S rRNA transcript level). The values presented are the averages from three independent experiments, and error bars represent the standard deviations. A two-way analysis of variance with Bonferroni posttest was used to obtain P values for each time point. ***, P < 0.001.

DISCUSSION

The cognate toxin and antitoxin of a TA system are small proteins encoded by two genes organized in one operon (36). The activity of the toxin can be neutralized by forming a protein complex with its cognate antitoxin (30, 37). Our results confirmed that the mazF homologue (BLJ_811) present in the chromosome of the JDM301 strain encodes a toxic protein (MazF₁^Bif), which forms a complex with its cognate antitoxin, encoded by the adjacent mazE gene (BLJ_812) (10).

To date, toxins of TA systems include ribonucleases, DNA gyrase poisons, phosphotransferases, and protein kinases (2, 38). MazF has been shown to act by cleaving mRNA, resulting in translation inhibition in E. coli, Mycobacterium tuberculosis, Streptococcus mutans, and Clostridium difficile (18, 39–41). Like other toxin proteins from E. coli, M. tuberculosis, Streptococcus mutans, and Clostridium difficile, the MazF₁^Bif toxin was determined to cause mRNA degradation. Our results suggest that MazF₁^Bif induces the degradation of tufA^Bif mRNA, which may partially account for the inhibition of protein synthesis and cell growth arrest. Cleavage of elongation factor Tu mRNAs is partly responsible for growth inhibition caused by MazF₁^Bif. The MazF₁^Bif toxin was not purified in this study because of its low production level in E. coli. Thus, we performed an in vivo RNase assay with tufA^Bif mRNAs as the substrate. In previous reports, the homologs of tufA^Bif were used to determine the RNase activity of RelE and HigB in E. coli (23, 24). Our previous work showed that MazE₁F₁^Bif is activated through the hydrolysis of MazE₁^Bif (10). Therefore, it was proposed that in response to adverse conditions, the antitoxin MazE₁^Bif is degraded, releasing the toxin MazF₁^Bif to cleave existing transcripts, such as tufA^Bif mRNA. Consequently, cell growth is modulated and stasis may occur, which may help the cells cope with environmental stress (7).

To date, few studies presenting the interaction between different TA modules have been reported. Yang et al. observed that three M. tuberculosis RelE toxins physically interact with the same RelB protein to conditionally regulate RelB binding with promoter DNA (31). Zhu et al. observed noncognate toxin-antitoxin associations, even among different TA families (MazF toxins and VapB antitoxins), in M. tuberculosis (30). Recently, transcriptional cross-activation between toxin-antitoxin systems was found in E. coli (42). Multiple TA systems have been shown to coordinately govern the persister phenotype in E. coli (14). In this study, MazF₁^Bif was shown to physically interact with the noncognate antitoxin RelE^Bif, which belongs to another family of TA systems, or to MazE₂^Bif in strain JDM301. Furthermore, when either antitoxin RelE^Bif or MazE₂^Bif was overexpressed in E. coli, cell growth inhibition conferred by the MazF₁^Bif toxin was alleviated. Thus, the interaction between MazE₁F₁^Bif and RelBE^Bif or MazE₂F₂^Bif was demonstrated. In addition, RelE^Bif and MazE₂^Bif antitoxins were observed to antagonize the degradation of tufA^Bif mRNA by MazF₁^Bif. Interestingly, the TA system MazE₁F₁^Bif is activated under acid stress (10). In addition, the expression levels of ClpP₁^Bif and ClpP₂^Bif proteases responsible for the activation of MazE₁F₁^Bif are also increased significantly during acid stress (10), whereas MazE₂F₂^Bif (data not shown) and RelBE^Bif are not activated under this adverse condition (35). As a major challenge to bifidobacteria, acid stress might reduce the viability and probiotic effects of these bacteria (43). TA systems have been implicated in the acid stress response of E. coli and Streptococcus mutants. In E. coli, the antitoxin MqsA mediates the general stress response in bacteria, including the acid stress response (27). Furthermore, a mutated strain of Streptococcus devoid of TA systems was shown to be more resistant to changes in pH than the wild-type strain (44).

Generally, there is an excess of antitoxin proteins, since the level of gene expression is often proportional to the gene order in a polycistronic message (i.e., the gene encoding the antitoxin precedes the gene encoding the toxin in the TA operon) (45, 46). In other words, when TA systems are inactivated, the antitoxin proteins exist in excess relative to their cognate toxins (47). Thus, it was speculated that there is an excess of antitoxins (MazE₂^Bif and RelB^Bif) in B. longum strain JDM301 under acid stress when TA systems MazEF₂^Bif and RelBE^Bif are all inactivated. When strain JDM301 was subjected to acid stress, the activation of MazE₁F₁^Bif led to the release of free MazF₁^Bif toxin resulting in cell growth inhibition. It is possible that excess noncognate antitoxins MazE₂^Bif and RelB^Bif partially abolish the toxicity of MazF₁^Bif, helping cells to switch more quickly from a state of growth inhibition to one of normal growth when the acid stress is removed. As a common probiotic bacterium, bifidobacteria are added to many types of fermented dairy foods. However, during the industrial process, storage, and passage through the digestive tract of the host, bifidobacteria are subjected to various stresses, such as acid stress. It was implied that the interaction between MazF₁^Bif and MazE₂^Bif or RelB^Bif may facilitate bacterial adaptation to changing environments encountered during the industrial manufacturing process and passage through the digestive tract of the host. However, the industrial environment is probably not an evolutionary driver of TA systems. TA systems in closely related bifidobacterial species show extensive 95% to 100% similarity, which suggests that horizontal gene transport may account for significant portions of the distribution of TA systems among bifidobacteria (45, 48). Given that the gastrointestinal tract (GIT) is a natural environment of Bifidobacterium and a broad variety of bacterial species inhabit the GIT, the main site of horizontal gene transport of bifidobacterial TA modules is the GIT (49, 50). Thus, the harsh conditions in the GIT are the main evolutionary driver of bifidobacterial TA systems. Previously, it was shown that MazEF and RelBE are widely distributed in bifidobacteria (32). Interestingly, among the species, only strain JDM301 and ATCC 15697 have as many TA system genes, while the other bifidobacteria harbor only a few TA systems (48). It was speculated that multiple TA systems may help bifidobacteria to cope with various environmental stresses.

Given that the codon bias and GC content of B. longum and E. coli are different and that chromosomally encoded MazF₁^Bif and antitoxin proteins are heterologously produced by the double promoter plasmid, the heterologous expression of toxin and antitoxin proteins may be influenced. Similarly, in our previous reports, MazE₁^Bif did not thoroughly abolish the toxicity of MazF₁^Bif when MazE₁^Bif and MazF₁^Bif were jointly expressed under the control of one promoter, while MazE₂^Bif and RelE^Bif completely neutralized their cognate toxins MazF₂^Bif and RelE^Bif, respectively (10, 34, 35). We speculate that in the original host, MazF₁^Bif is inhibited by the compensatory actions of noncognate antitoxins under normal conditions, as MazE₁^Bif did not completely abolish the toxicity of MazF₁^Bif. On the other hand, apart from a species-specific pattern in codon usage, there are also considerable differences among genes in many species (51). Our results may be due to a larger deviation in codon bias between B. longum and E. coli, leading to a disproportionately lower translation efficiency of MazE₁^Bif than of MazF₁^Bif. Further studies are needed to gain a deeper insight into the role of interactions among TA system families in regulating B. longum cell growth. Overall, these results provide molecular insight regarding the interactions of TA systems implicated in the bifidobacterial stress response and can serve as a foundation for future studies of TA systems in their natural host, B. longum.

MATERIALS AND METHODS

Bacterial strains and culture conditions.

A summary of the bacterial strains used in this study is shown in Table 1. JDM301 was cultured anaerobically in MRS (Difco) supplemented with 0.05% (wt/vol) l-cysteine-HCl at 37°C for 14 to 16 h. The DH5α and BL21(DE3) strains of E. coli were each cultured aerobically in LB medium on a rotary shaker (220 rpm) at 37°C or cultured on LB agar plates. When needed, the culture medium was supplemented with 50 μg/ml kanamycin, 35 μg/ml chloramphenicol, 100 μg/ml ampicillin, or 200 μg/ml rifampin for E. coli. IPTG was added at a final concentration of 0.5 mM or 1 mM to induce the expression of toxin and antitoxin proteins in E. coli. Additionally, 0.2% (wt/vol) arabinose was added to induce the transcription of tufA^Bif mRNA driven by the promoter of the arabinose operon (P_BAD).

TABLE 1.

Strains and plasmids used in this study

Strain or plasmid	Description	Source or reference
E. coli BL21(DE3)	General expression strain	Tiangen Biotech
B. longum JDM301	Commercial strain	33
pET28a	Expression vector with strong T7 promoter, His tag	Novagen (USA)
pET-E1	pET28a derivative carrying the mazE1Bif gene, His tag	This study
pET-F1(Myc)	pET28a derivative carrying the mazE1F1Bif gene, Myc tag	This study
pET-E1F1(Myc)	pET28a derivative carrying the mazE1F1Bif gene, His tag and Myc tag	10
pBAD/HisB	Expression vector with PBAD promoter, His tag	Invitrogen
pBAD-tufA	pBAD/HisB derivative carrying the tufABif gene, His tag	This study
pACYCDuet-1	Expression vector for coexpressing two target genes with T7 promoter, His tag and S tag	Novagen
pAD-F1	pACYCDuet-1 derivative carrying the mazF1Bif gene	This study
pAD-F1E1	pACYCDuet-1 derivative carrying the mazF1Bif gene and mazE1Bif gene	This study
pAD-F1E2	pACYCDuet-1 derivative carrying the mazF1Bif gene and mazE2Bif gene	This study
pAD-F1B	pACYCDuet-1 derivative carrying the relBBif gene and mazF1Bif gene	This study

Construction of plasmids.

All plasmids used in this study are listed in Table 1. PCR primers and restriction sites used are shown in Table 2. As previously reported (10), the intact TA locus (mazE₁F₁^Bif) was amplified and cloned into pET28a to yield pET-E₁F₁(Myc), which encodes an N-terminal His₆-tagged MazE₁^Bif and a Myc-tagged MazF₁^Bif under the control of one promoter. Thus, the native mazE^Bif-mazF^Bif gene organization was kept intact to coexpress MazF₁^Bif and MazE₁^Bif under the control of one promoter. The full-length tufA^Bif gene was amplified and cloned into pBAD/HisB to yield pBA-tufA (35). The mazF₁^Bif gene was placed under the control of the T7 promoter-1 of pACYCDuet-1, to yield pAD-F₁. The mazE₁^Bif, mazE₂^Bif, and relB^Bif genes were amplified and subcloned under the control of the T7 promoter-2 of pAD-F₁, resulting in pAD-F₁E₁, pAD-F₁E₂, and pAD-F₁B, respectively. All the genes were amplified using JDM301 genomic DNA as the template, which was extracted from mid-log-phase cultures grown in MRS broth.

TABLE 2.

Oligonucleotide primers used in PCR and qRT-PCR

Gene	Assay	Direction	Primer sequencea	Length (bp)	Source or reference
mazF1Bif	PCR	Sense	CCGGAATTCAATGAAACGAGGTGAGATT	324	This work
		Antisense	CCCAAGCTTTCACAGCAAACCTAGAAC
mazF1Bif (primer mazF1Bif-2)	PCR	Sense	TTAAGAAGGAGATATACCATGGAAATGAAACGAGGTGAGATTCG	354	This work
		Antisense	CTCGAGTGCGGCCGCAAGCTTTTACAGATCCTCTTCTGAGATGAGTTTTTGTTCCAGCAAACCTAGAACCCG
mazE1Bif	PCR	Sense	GGAAGATCTAATGAGCATACAGATTGCCA	246	This work
		Antisense	CCGCTCGAGCCTCGTTTCATCGGACAT
mazE1Bif (primer mazE1Bif-2)	PCR	Sense	GGATCCATGAGCATACAGATTGCCA	246	This work
		Antisense	CTCGAGCCTCGTTTCATCGGACAT
mazE2Bif	PCR	Sense	GGAAGATCTTATGGCTATCAAGGAGAAGG	294	This work
		Antisense	CCGCTCGAGGTCTTCATCATCGTCCCA
relBBif	PCR	Sense	GGAAGATCTATTGTCTTACGTAATGATGTTGG	300	This work
		Antisense	CCGCTCGAGGATTCCCAACGAATCGAA
mazEBif	qRT-PCR	Sense	GCTGCGTTGTTTAAGGAGAC	94	10
		Antisense	ACCCACGGTAATCATTGAAC
tufABif	qRT-PCR	Sense	ATCCGTCCGACCCAGACC	123	54
		Antisense	CTCGACATCCTCACGGCC
16S	qRT-PCR	Sense	TACGGGAGGCAGCAG	191	55
		Antisense	ATTACCGCGGCTGCTGG

^aRestriction sites for XhoI, EcoRI, BglII, and HindIII incorporated into the primers are in boldface, and the sequence for the Myc tag is underlined.

Assessment of toxin and antitoxin activities in E. coli.

E. coli strains carrying pACYCDuet-1 (a blank vector), pAD-F₁, pAD-F₁E₁, pAD-F₁E₂, or pAD-F₁B were grown in LB broth with IPTG (1 mM) to induce gene expression. Growth curves of the E. coli carrying the corresponding vectors were determined by measuring the OD values at 600 nm (OD₆₀₀) to assess the effects of the toxin and antitoxin on cell growth. Cultures of the E. coli strains were initially grown in LB overnight with 35 μg/ml chloramphenicol and then transferred into fresh LB using 1% inoculum in the presence of 1 mM IPTG (IPTG was added to the LB at 0 h). Samples were taken at different time points, and the optical density at 600 nm was determined for each.

In vivo cleavage of the tufA^Bif mRNA.

pBA-tufA^Bif and pACYCDuet-1 (a blank vector), pBA-tufA^Bif and pAD-F₁, pBA-tufA and pAD-F₁E₁, pBA-tufA and pAD-F₁E₂, or pBA-tufA and pAD-F₁B were transformed into E. coli to determine whether MazF₁^Bif causes the degradation of tufA^Bif mRNA and whether MazF₁^Bif toxicity is inhibited by its cognate antitoxin, MazE₁^Bif, or noncognate antitoxins MazE₂^Bif and RelB^Bif. The plasmids used are listed in Table 1, and primers for qRT-PCR of tufA^Bif mRNA are listed in Table 2. Cells transformed with the corresponding plasmids were grown at 37°C on a rotary shaker. When the cultures reached an OD₆₀₀ of 0.5, 0.2% l-arabinose was added to the medium to induce the transcription of tufA^Bif mRNA. After incubating at 28°C for 2 h, 1 mM IPTG was added to induce the expression of toxin (and antitoxin) proteins for 3 h. Then, 200 μg/ml rifampin was added to halt transcription. At different time points, 2-ml aliquots were taken. qRT-PCR was performed to determine the level of tufA^Bif mRNA. The transcription of 16S rRNA genes in E. coli was evaluated as an internal control.

Quantitative real-time PCR.

Total RNA was extracted from E. coli strains carrying pBA-tufA and pACYCDuet-1 (a blank vector), pBA-tufA and pAD-F₁, pBA-tufA and pADuet-F₁E₁, pBA-tufA and pAD-F₁E₂, or pBA-tufA and pAD-F₁B and was treated with DNase (Roche, Basel, Switzerland). For use in qRT-PCR, cDNA was generated from total RNA (2 μg) using a Superscript III first strand synthesis RT-PCR kit (Invitrogen, Carlsbad, California, USA) according to the manufacturer's protocol. Primers targeting tufA^Bif were designed for detecting mRNA expression by qRT-PCR. The primers used are shown in Table 2. The reaction was performed with an ABI 7500 system (Applied Biosystems, Branchburg, New Jersey, USA) using the following conditions: 95°C for 10 min, followed by 40 cycles at 95°C for 15 s and 60°C for 1 min. Calculations were performed using the 16S rDNA gene as an internal standard. The 2^-ΔΔCT method was used to determine the relative gene expression (52).

Protein purification and Western blot analysis.

pET-E₁, pET-F₁(Myc), and pET-E₁F₁(Myc) were transferred into E. coli to express the His-tagged MazE₁^Bif and/or Myc-tagged MazF₁^Bif recombinant proteins. Cells transformed with the corresponding plasmids were grown at 37°C on a rotary shaker. When the cultures reached an OD₆₀₀ of 0.5, IPTG (at a final concentration of 0.5 mM) was added to induce the expression of recombinant fusion proteins (His-tagged MazE₁^Bif and Myc-tagged MazF₁^Bif) for 5 h. The cells were pelleted and disrupted on ice by sonication, and the soluble or insoluble fraction was recovered by centrifugation. The recombinant proteins were purified by affinity chromatography using Ni-NTA resin (Qiagen, Hilden, Germany) according to the manufacturer's protocol. The purified proteins were fractionated by 15% SDS-PAGE and transferred to a nitrocellulose membrane for detection by Western blot analysis. Anti-His and anti-Myc monoclonal antibodies were used to detect His-tagged MazE₁^Bif and Myc-tagged MazF₁^Bif, respectively.

Coimmunoprecipitation assay.

E. coli strains carrying pACYCDuet-1 (a blank vector), pAD-F₁E₁, pAD-F₁E₂, or pAD-F₁B were grown in LB until an OD₆₀₀ of 0.5 was reached. After that, IPTG (at a final concentration of 1 mM) was added to the medium to induce protein expression for 5 h at 28°C. The cells were pelleted, and the proteins were isolated and immunoprecipitated as described previously (53) with a few modifications. Protein extracts (400 μg) were incubated with anti-His (His-probe [G-18], sc-804; Santa Cruz Biotechnology) or anti-S (MB2016; Bioworld Technology) antibodies on a rotator overnight at 4°C. The immunoprecipitated proteins were incubated with protein A/G agarose (P2012; Beyotime Biotechnology) for 4 h at 4°C. The immunoprecipitated complexes were dissolved in 2× SDS gel loading buffer (at a final concentration of 2% [wt/vol] SDS, 0.1% [wt/vol] bromophenol blue, 10% glycerol, and 50 mM Tris/HCl, pH 6.8). Then, the samples were incubated at 95°C for 5 min and subjected to SDS-PAGE. Extracted proteins and immunocomplexes were analyzed by immunoblotting using anti-His or anti-S monoclonal antibodies. A horseradish peroxidase-labeled goat anti-rabbit IgG antibody was used as a secondary antibody (sc-2054; Santa Cruz Biotechnology). Chemiluminescence signals were visualized with an enhanced chemiluminescence (ECL) reagent (Thermo Scientific, Rockford, Illinois, USA) and were exposed to film.