Comprehensive Functional Analysis of the 18 Vibrio cholerae N16961 Toxin-Antitoxin Systems Substantiates Their Role in Stabilizing the Superintegron

INTRODUCTION

Toxin-antitoxin (TA) systems were discovered in 1983 on plasmid F of Escherichia coli (1) and were shown to be involved in stable plasmid maintenance by postsegregational killing, a mechanism distinct from replication and partition. These systems typically consist of an operon of two genes, which encode a toxin that targets an essential cellular function and an antitoxin that binds to and inhibits the toxin. Toxin activity is regulated through differential stability of the stable toxin and the labile antitoxin; loss of a TA system by the progeny during cell division results in cell death by the action of the stable toxin (for reviews, see references 2 and 3). The antitoxin, in most cases, also acts as a transcriptional autorepressor of the operon, and the degradation of the antitoxin results in the transcriptional activation of the TA operon. Currently there are three types of TA systems, which are classified according to the nature of their antitoxin (4). For the type 1 system, such as tisA-tisB (5) and hok-sok (6), toxins are hydrophobic proteins and antitoxins are small RNAs (sRNAs). Type 2 TAs are the most represented among TA systems, and both their toxins and antitoxins are present as proteins (7). The third type of TA system is currently represented only by the toxIN locus, which was discovered in 2009 on a cryptic plasmid of Erwinia carotovora (8). It consists of a protein-RNA toxin-antitoxin pair, where RNA antitoxin ToxI inhibits the toxicity of toxin ToxN by interfering with its biochemical activity (9). Type 2 TA systems are ubiquitous in prokaryotic genomes and have more recently been identified as genuine components of the chromosomes of most free-living bacteria (7, 10, 11), with up to 88 predicted TAs in the Mycobacterium tuberculosis genome (12). Several hypotheses have been proposed for the biological roles of chromosomal TAs, but these are not yet clear and remain under debate. TAs have been proposed to be physiological or developmental regulators of processes such as programmed cell death (PCD), growth and/or development, and persister cell formation. PCD refers to cell death mediated by an intracellular death program through a cascade of signalization in response to an external stress (13); however, this hypothesis is controversial, as several groups have failed to reproduce the original observations (4). Gerdes proposed that TA loci function as stress response elements and that they help the cells cope with nutrient starvation by modulating their growth (14). In the context of their role in survival, TA systems have been proposed to induce a dormant cellular condition termed persistence (15–17). Persisters are nongrowing, dormant cells exhibiting low levels of translation, and this state could allow them to overcome stress conditions, such as antibiotic treatment (reviewed in reference 18). TAs have also been involved in bacteriophage defense through abortive infection mechanisms, where bacteria commit suicide to prevent phage spreading (19–21, 56). Two roles for TAs regarding genetic stability have also been proposed. The first is as an antiaddiction module, in which TAs can protect the host genome from colonization by an incoming mobile element or from a plasmid carrying a TA from the same functional family by allowing its harmless loss through neutralization of the invading toxin by the chromosomal antitoxin (22). This hypothesis is supported by the fact that some chromosomal TA systems can cross-interact with their plasmidic homologs; i.e., a chromosomally encoded toxin is neutralized by a plasmid-encoded antitoxin or vice versa (for examples, see references 22 to 24). Recently, we have shown that the chromosomally encoded phd-doc system of Vibrio cholerae can interact with the homologous phd-doc system encoded by phage P1 (25). These observations emphasize that chromosomal TA systems may act as “exclusion” systems to protect the bacteria from an excess of invading mobile genetic elements (plasmids, transposons, and phages) carrying identical TA systems (22, 24). In addition, chromosomal TA systems, by interfering with their homologs encoded by mobile genetic elements, could drive the evolution of the latter ones to select the systems having toxins that are no longer recognized by the antiaddiction modules. Finally, similar to their role in plasmid stabilization, TAs have been proposed to stabilize chromosomal regions by preventing accidental deletions, especially when located in unstable segments such as mobile genetic elements (26–28). The mosAT and tad-ata systems have been shown to be responsible for the stabilization of two different integrative conjugative elements (ICEs) by postsegregational killing as in plasmids (28, 29). The insertion of two systems (relBE1 and parDE1) from Vibrio vulnificus stabilized a 165-kb nonessential chromosomal region in E. coli (27). Strikingly, TAs are very common in cassettes of chromosomal integrons, especially those TA systems known to be involved in plasmid stabilization (26, 27, 30), and are present often in high numbers, especially in superintegrons (SIs) (31). Bacterial chromosomes are subject to lateral DNA exchange through genomic parasites such as integrons, conjugative transposons, and bacteriophages. Among these, integrons are very efficient in capturing and rearranging the open reading frames (ORFs) embedded in gene cassettes and converting them to functional genes by ensuring their correct expression. Vibrio SIs gather hundreds of highly diverse gene cassettes (32), with a large majority of ORFs of unknown functions. The analysis of available Vibrionales genomes in 2007 led to the identification of 1,677 gene cassettes (33). Cassettes are generally promoterless and are thought to constitute a silent reservoir of adaptive functions (26, 33). Expression of these cassettes relies on their relative position from the Pc promoter, and any silent cassette can be called on for expression through recombination to an attI or attI-proximal location (for a review, see reference 31). Integron integrase expression, which is commonly controlled by the SOS response (34, 35), governs cassette recombination. Thus, cassette recombination is triggered in times of stress, while under nonstressful conditions, the cassette arrays stay steady and the selective pressure exerted on distal silent cassettes is assumed to be low if not nonexistent. Previous analysis of the V. cholerae N16961 genome had led to the identification of 13 type 2 TA systems, all localized in the SI (7, 26). Interestingly, these potential TA cassettes are among the few cassettes that seem to carry their own promoter (25, 31). The fact that these cassettes are most likely expressed led us to propose that they could play a stabilizing role by preventing cassette loss through homologous recombination between identical attC sites (the cassette recombination site) or between copies of repeated identical cassettes (26, 27). Chromosomal regions such as SIs are not generally essential for bacterial survival but rather encode functions that aid the bacteria in rapidly adapting to a changing environment. In this way, TA systems contribute to maintain the potential genes that can confer a major selective advantage to their host under stress, and these genes may disappear rapidly under steady growth conditions in the absence of TA modules. Previous studies of the V. cholerae TA cassettes of the HigBA, ParDE, and Phd-Doc families gave further support to this role (25, 30, 36). However, these properties do not restrain cassette microevolution and the role of TA systems as antiaddiction modules. Here, we reanalyzed the N16961 genome and identified 5 potential new TA systems in addition to the 13 previously listed. Strikingly, four candidates were also embedded within integron cassettes, while the fifth is related to HipAB and does not seem to be embedded within a mobile genetic element. Among the previously described potential TA cassettes present in the V. cholerae N16961 SI, 3 parDE, 2 higBA, and the phd-doc TA cassettes had already been shown to be functional TA systems (25, 30, 36, 37). In this study, we established the functionality of all 11 remaining candidates, as well as the sedentary HipAB homolog that we identified. Thus, we demonstrate for the first time that all of these 17 SI cassette systems are functional and that they are all expressed in their native context, demonstrating that they all carry their own promoter, a rare property for integron cassettes. Importantly, we show that these systems are very specific. Indeed, we found that even for systems belonging to the same TA family, there is no cross-interaction between noncognate toxins and antitoxins. The roles that these multiple TA cassettes can play in this genomic structure are discussed in light of their lack of cross-interactions and specific pattern of transcription.

MATERIALS AND METHODS

RESULTS

DISCUSSION

In a previous analysis in 2005, Pandey and Gerdes identified 13 TA operons in the V. cholerae N16961 genome. Strikingly, their analysis revealed that all of these were embedded within cassettes of the superintegron (7). Since then, TA systems have gained even more attention, as they have been the subject of multiple analyses and many new systems have been described (11). This, together with the fact that the role of these genetic modules is highly debated, led us to undertake a reanalysis of the toxin-antitoxin system content of V. cholerae genomes. We uncovered 5 novel TA modules in N16961, 4 of which were carried within integron cassettes, giving a total of 17 TA cassettes (Fig. 1). These cassettes are highly conserved among V. cholerae strains and isolates (see Table S2 in the supplemental material). The single TA system that we identified and characterized which is not carried in the SI encodes a toxin of the HipA family and its antidote. This toxin has been shown to inactivate the glutamyl-tRNA synthetase through phosphorylation and was identified through its major role in persister formation in E. coli (15, 16).

We have demonstrated the functionality of the 17 V. cholerae TA cassettes and explored their cross-interactions. Among these, 10 correspond to well-defined TA families, and six of these cassettes have been previously characterized to various extents. These are the three ParDE, the two HigBA, and the single Phd-Doc cassettes (25, 30, 36, 42). Two of the three ParDE family cassettes correspond to a duplication of the same cassette (parDE1 and parDE3) separated by 43 kb in N16961. The genes of both ParDE cassettes, parDE1-3 and parDE2, are in opposite orientation compared to the majority of cassettes whose expression relies on the Pc promoter (Fig. 1) (44). The ParE toxin is a gyrase inhibitor that induces DNA degradation and activates the SOS response to DNA damage, resulting in inhibition of cell division and reducing cell viability. In agreement with the recent findings of Yuan and coworkers (36, 42), our results confirm that parDE1-3 and parDE2 encode functional toxins whose toxicity was completely blocked by their cognate antitoxins. We further confirmed no cross-interaction between the three members of Par family (36, 42). As in Caulobacter crescentus, which carries three ParDE systems (45), transcription of the two functional paralogous parDE loci is differentially regulated, as parDE1-3 is highly induced during exponential growth phase, while parDE2 is induced only 1.6-fold (Table 1), supporting the idea that these paralogous systems act as distinct functional units (45). The two HigBA cassettes, higBA1 and higBA2, carry operons oriented as classical cassette genes, i.e., in the same direction as transcription from the Pc promoter (Fig. 1). Both HigB toxins have been shown to inhibit the global rate of translation and preferentially cleave translated mRNA in vivo (30). Our results are similar to those published previously (30, 37) and confirm that both of the V. cholerae higBA loci encode functional toxins and that their toxicity is neutralized only by expression of their cognate antitoxins. As for parDE, we found that the two paralogs are differently regulated during growth (Table 1).

We characterized in this study, in terms of both functionality and expression, the remaining 11 TA cassettes. The V. cholerae N16961 SI contains 4 relBE-homologous cassettes, two of which (relBE2 and relBE4) are identical. As with ParDE cassettes, the relBE genes are in opposite orientation compared to canonical integron cassettes. RelE toxins are translation inhibitor cytotoxins that cleave mRNAs from their 5′ ends, leading to a rapid translational shutdown that is lethal or inhibitory to the host cells (46). We show here that all the relBE loci of V. cholerae N16961 encode functional toxins and that the activity of these toxins is completely counteracted by their cognate antitoxins. We also observed that no cross-interactions exist between the noncognate RelE and RelB proteins. In contrast to the ParDE and HigBA TA systems, the four paralogs RelBE are similarly regulated during growth (Table 1).

The seven other TA systems in the superintegron are either hybrid or novel systems. There is one phd-parE-like (VCA0422/0423), one stbD-relE-like (VCA0477/0478), and one relB/parE-like (VCA0488/0489) system according to the TA search software described previously (11) and four systems with ribbon-helix-helix (RHH) transcription factor-like antitoxins (VCA0319 [its duplication in VCA0482], VCA0333, and VCA0486 [see Table S3 in the supplemental material]) associated with toxins that are either N-acetyltransferases (VCA0318/0481, VCA0332, and VCA0487) or conserved proteins of unknown function (VCA0332). We showed that all are expressed and that their transcription, except for VCA0486-0487, is induced during exponential growth (Table 1). We demonstrate that all but one of these TA loci encode functional toxins-antitoxins. The nonfunctional cassette of N16961, VCA0422-0423, carries a toxin (VCA0423) that has been inactivated through an illegitimate recombination with another cassette, which exchanged the 3′ part of the carried gene, leading to a new C-terminal part for the original VCA0423. By chance, we found that this was a recent event, as several V. cholerae strains, including 569B and O395, carry the original VCA0422-0423c cassette, for which we demonstrated toxin functionality.

We tested all possible cross-interactions between the different antitoxins and toxins that could lead to toxin inhibition and found that all the antitoxins block only their cognate toxin activity, even when TAs belonged to the same family (Table 2). The 17 identified TA cassettes are scattered throughout the 176-cassette-long array of the N16961 superintegron (Fig. 1). The first TA cassette corresponds to cassette 18 in the array, and the last is located in position 174, 2 cassettes before the end cassette (see the annotated V. cholerae N16961 genome of the MicroScope platform at the French Genoscope for a precise mapping [https://www.genoscope.cns.fr/agc/microscope/mage/]). As mentioned above, these cassettes are highly conserved among V. cholerae strains and isolates (see Table S2 in the supplemental material).

As mentioned above, we have shown that all of these cassettes are expressed and carry their own promoter, an extremely rare situation for integron cassettes. This rare property guarantees that they are transcribed independently of their location inside the cassette array. Indeed, integron cassettes seldom carry a promoter, and their expression occurs when they are recombined at, or close to, the attI site, which carries a strong promoter, Pc (44, 47). This dynamic expression model, which allows the expression of adaptive genes only when needed, is considered to be central for integron evolutionary success (31). In the V. cholerae N16961 SI, as for other integrons, this constraint impacts the orientation of the cassette ORFs; among the 188 consecutive ORFs of its SI, only 27 are in opposite orientation relative to the Pc promoter. Strikingly, among these, 18 correspond to genes carried in 9 of the 17 TA cassettes. It is interesting to note that in this configuration, even if transcription extends beyond the cassette boundary, it will not express the surrounding genes, as they are carried on the opposite, complementary, strand. The SI TA expression was found to be higher during exponential growth than during entry into stationary phase, except for two systems (Table 1). It was previously suggested that chromosomal TA loci function as stress response elements, as they help the cells cope with nutrient starvation (14). Indeed, transcription of TA loci was found to be strongly increased during nutrient starvation in Escherichia coli (48–50) and during heat shock in Sulfolobus solfataricus (51). This has also been observed for the two higBA cassettes, whose transcription is induced during amino acid starvation (30). All but two antitoxins are predicted to carry a structural domain known to mediate interactions with DNA (see Table S3 in the supplemental material), suggesting that they can regulate the expression of the TA operons under certain conditions. Our results advocate for a major role of these 17 TA cassettes is the stabilization of the SI cassette arrays, as previously proposed when we characterized the first two TA cassettes (26, 27). Chromosomal TA loci may act to stabilize their neighboring DNA, mobile genetic elements, and large genomic islands because the loss of any given TA locus (e.g., by slippage during replication or illegitimate recombination) will result in a cessation of de novo antitoxin production, toxin activation, and arrest of cell growth and/or viability. This hypothesis is reinforced by the absence of cross talk between the SI TAs we observed (Table 2).

This possible role better explains the presence of 17 functional TA loci grouped in the SI of V. cholerae (Fig. 1), as the V. cholerae SI is tightly packed with genes (one gene per 600 bp and at least 176 attC sites) (26). The V. cholerae attC sites, called VCR, are 123-bp-long specific sequences which show more than 90% identity in 149 out of the 176 VCRs (26). Thus, any recombination between repeated copies of cassettes that can result in a large-scale deletion will be counterselected by one of the 14 unique TA loci, as far as the TA is located in that fragment. The V. cholerae TA loci are likely to constrain the emergence of daughter cells carrying deletions within the SI, which may otherwise survive if there were no functional TA loci and thus contribute to the stable perpetuation of the V. cholerae genome. The facts that all these TA cassettes are expressed from their own promoter and that 15 out 17 are more expressed during exponential growth, which is when the replication is active, support this stabilization role. Indeed, recombination and slippage, which would lead to accidental cassette loss, are more frequent during active replication than when bacteria reach stationary phase and stop replication. However, it is likely that the TA cassettes could also have additional roles. These cassettes can protect their host genome from colonization by incoming mobile elements or plasmids carrying TAs from the same functional families by allowing their harmless loss through neutralization of the invading toxin by the chromosomal antitoxin (22, 25). These TAs can also play a role in bacteriophage defense through abortive infection mechanisms, as already demonstrated in other bacteria (19–21, 56). Vibrio cholerae ecology is intricately linked to phage infection. Indeed, phage predation has been reported to be a factor that influences seasonal epidemics of cholera in Bangladesh (52). Furthermore, phage infections have been shown to occur in the human host, leading both to a decreased load of bacteria (53) and to an alteration of the virulence potential of V. cholerae shed from cholera patients (54).

We observed that when ectopically expressed in E. coli, the efficiencies of the different tested toxins on cell viability vary to a large extent. However, all lead to the formation of a fraction of cells that, if not dividing, survive toxin activity and are in a state very similar to persistence (see Fig. S2 in the supplemental material). Persisters are nongrowing, dormant cells having low levels of translation, and their state may be related to viable-but-nonculturable (VBNC) conditions, which have been commonly observed in Vibrio species, including V. cholerae (for a review, see reference 55). The molecular mechanisms underlying this phenomenon have not yet been identified, but those TAs may have similar properties in V. cholerae and play a key role in persister formation, even when carried in SI cassettes, as found by the groups of Lewis and Hill for E. coli TA systems (16, 17). Indeed, in E. coli, the overexpression of different TA loci enriched for persister cells and ectopic expression of TA loci dramatically increased the number of cells exhibiting multidrug resistance (16–18).

The quasiabsence of TA loci from the V. cholerae core genome is puzzling, especially as we identified TAs in other vibrios both inside and outside their integrons in their genome (31) (Table 3). The presence of expressed TAs within the integron cassettes may have allowed the deletion of TAs located outside the SI due to TA cross-interactions, given the possibility that these cassettes could fulfill the physiological role(s) played by chromosomal TAs. This hypothesis now needs to be tested.

Supplementary Material