physiology and application in molecular biology

Bacteria have developed multiple complex mechanisms ensuring an adequate response to environmental changes. In this context, bacterial cell division and growth are subject to strict control to ensure metabolic balance and cell survival. A plethora of studies cast light on toxin-antitoxin (TA) systems as metabolism regulators acting in response to environmental stress conditions. Many of those studies suggest direct relations between the TA systems and the pathogenic potential or antibiotic resistance of relevant bacteria. Other studies point out that TA systems play a significant role in ensuring stability of mobile genetic material. The evolutionary origin and relations between various TA systems are still a subject of a debate. The impact of toxin-antitoxin systems on bacteria physiology prompted their application in molecular biology as tools allowing cloning of some hard-to-maintain genes, plasmid maintenance and production of recombinant proteins.


InTROducTIOn
Toxin-antitoxin systems emerged in research in mid 80's.A detailed insight into their functions and mechanisms of action has been gained in the last two decades and brought several interesting conclusions as to the importance of such systems for bacterial physiology.The term "toxin-antitoxin system", usually abbreviated as "TA system", comprises a functional element consisting, in most cases, of a biologically active protein molecule and a corresponding inhibitor, whose nature and inhibitory mechanism depend on the system's class affiliation.
Components of such systems are encoded within policistronic operons, often with partially overlapping open reading frames.The systems are widespread among Bacteria as well as Archaea (Mittenhuber, 1999;Gerdes, 2000;Pandey & Gerdes, 2005;Makarova et al., 2009) and evolved to carry out diverse functions.However, their common feature is an enzymatic activity detrimental for the cell metabolism.Such toxic activity has been demonstrated to switch bacterial cells over to a dormant state, leading to cell death during prolonged exposure.In most cases various stress stimuli are responsible for TA system activation.The signalling pathway in such instances is often related to other stress-induced response pathways.Moreover, it is well documented that in some cases the activity of TA systems stabilizes mobile genetics elements, therefore comprising an important mechanism of plasmids maintenance.In the light of the increasing multi-drug resistance among virulent strains, reports on the potential relation between TA systems and modulation of pathogen-host interactions seem to be of utmost importance.

clAssIfIcATIOn Of TOxIn-AnTITOxIn sysTeMs
The biological activity of a toxin comprising a component of a TA systems is usually (but not always) that of an endoribonuclease.Bioinformatic analysis of multiple available sequences of bacterial genetic elements points to multiple novel, putative TA loci and suggests that many of known TA systems, bacterial as well as archaeal, are evolutionarily related (Anantharaman & Aravind, 2003;Hayes & Sauer, 2003;Gerdes et al., 2005;Sevin & Barloy-Hubler, 2007;Makarova et al., 2009;Weaver et al., 2009;Arbing et al., 2010).The classification of TA systems is based on the mechanism of inhibition of the toxin as well as on operon autoregulatory functions.Initially two classes of TA systems were identified (Gerdes & Wagner, 2007), but subsequent discoveries extended the classification to three classes (Blower et al., 2009).Recent studies suggest the existence of yet another type, namely a three-component TA system (Hallez et al., 2010).As immediately visible from the above discussion the field is in a constant and dynamic growth and one may expect that many interesting findings are likely to emerge in the following years.
Class I includes systems in which the antitoxin is an antisense RNA forming duplexes with the toxin mRNA.This leads to inhibition of translation in a process known as RNA interference.Examples of such systems are chromosomally located operons found in Escherichia coli, namely tisAB (Vogel et al., 2004) and symER (Kawano et al., 2007), as well as plasmid loci parB (Gerdes et al., 1986) of E. coli and par of Enterococcus faecalis (Greenfield et al., 2000;Weaver et al., 2009) and a homologous plasmid operon of Staphylococcus aureus (Jensen et al., 2010).Among the mentioned systems toxins have multiple different roles.For example the SymE toxin is an mRNA interferase encoded in the symER operon.The toxin binds ribosomes to exert its activity (Kawano et al., 2007).The TisB toxin, which is encoded in the tisAB operon (Vogel et al., 2004) decreases the protonmotoric force across the bacterial cell membrane and cause subsequent drop in ATP production, which leads to metabolic dormancy (Unoson & Wagner, 2008).Hok toxin, encoded in the parB operon, irreversibly damages the cell membrane (Gerdes et al., 1986).In the latter case the regulation of the toxin level is indirect.RNA interference suppresses expression of the gene mok, which is a regulator of hok gene transcription (Thisted & Gerdes, 1992).
Class II encompasses a wide range of TA systems.Antitoxins of this class are proteins.The biological activities exhibited by the toxins include transcription inhibition by targeting gyrase function and interference with translation through an mRNA interferase activity, which may or may not rely on ribosome binding.The endoribonucleolytic activity of mRNA interferases is often sequence specific.Table 1 gives a short overview of the class II TA systems and their characteristics.
Class III comprises a single member only.This system is encoded in the toxIN operon of Erwinia carotonovora, a plant pathogen.In this case inhibition of ToxN toxin activity is driven by RNA molecules directly interacting with the toxin molecules (Blower et al., 2009;Fineran et al., 2009).

RelATIOns And sTRucTuRAl sIMIlARITIes AMOng clAss II TA sysTeMs
The evolutionary relationship among class II TA systems is a subject of an open debate.Attention is mainly focused on toxins since there is a substantial sequence and structural variety among the antitoxins.Ten TA families of class II have been described so far (Pandey & Gerdes, 2005;Jorgensen et al., 2009;Van Melderen & Saavedra De Bast, 2009) and for three of them, relBE, parDE and higBA, a phylogenetic relationship based on sequence similarities has been proposed (Anantharaman & Aravind, 2003;Tsilibaris et al., 2007).Strikingly, the toxin of the parDE system is a gyrase inhibitor in contrast to the toxins of the relBE and higBA systems, which are mRNA interferases.A broader analysis of this issue leads to other interesting conclusions.There is no evidence for an evolutionary relation between the ccdAB and parDE systems (Anantharaman & Aravind, 2003) although the toxin of the ccdAB system is also a gyrase inhibitor.However, there is a significant structural similarity between the toxins of the ccdAB and kis/kid (parD) systems (Diago-Navarro et al., 2010), which, similarly to the parDE and relBE or higBA systems, are a gyrase inhibitor and an mRNA interferase, respectively.Other reports point to a structural similarity among the toxins of the ygiUT (mqsRA), relBE and yefM-yoeB systems as well as RNase Sa of Streptomyces aureofaciens (Brown et al., 2009).
Not only among RelE homologues is a similarity with RNase Sa noticeable.Toxins of the ccdAB and kis/kid or mazEF (chpAK) systems are also structurally similar.
This similarity is related to the presence of a β-sheet core in these molecules (Fig. 1).However, this β-sheet core structure is most likely related to the ability to form dimers (Miller, 1989) rather than reflects evolutionary or functional relationships.Structural analysis of mRNA interferases and comparative studies allow the deduction of the mechanism of their endoribonucleolytic activiy (Agarwal et al., 2009;Brown et al., 2009;Diago-Navarro et al., 2010).Tracing evolutionary relations among the TA systems is difficult because of the fast specialisation of TA system components (Arbing et al., 2010).It has been reported that the toxin of the phd/doc system is similar to a virulence factor toxic to eukaryotic host cells (Arbing et al., 2010).Another example is the sequence similarity of toxins of the symER and phd/doc systems to antitoxins of other TA systems -yefM-yoeB (Arbing et al., 2010) and mazEF (Kawano et al., 2007), respectively.
hicAB and ygiUT systems.Binding of toxin-antitoxin complexes to promoter sites is the most common way of direct transcription regulation of TA operons (Fig. 2).Single components also bind the promoters but with a low affinity (Kedzierska et al., 2007;Li et al., 2008) when compared to the toxin-antitoxin oligomers which bind to palindromic sequences within the promoters, which process is enhanced cooperatively (Tsuchimoto & Ohtsubo, 1993;Black et al., 1994;Magnuson et al., 1996;Magnuson & Yarmolinsky, 1998;Marianovsky et al., 2001;Bailey & Hayes, 2009).Moreover, apart form the described primary palindromes, promoter of the mazEF operon contains alternate palindromes.
Binding to the latter by a toxin-antitoxin complex manifests in a decrease in the transcription efficiency of the operon (Marianovsky et al., 2001).An exception to the above rule is the prophage P1 zeta-epsilon system (ζε) where the antitoxin serves only as an inhibitor of toxin activity and an additional expression regulator ω is present (de la Hoz et al., 2000), which is similar to recently reported three-component systems homologous to parDE, namely paaR1-paaA1-parE1 and paaR2-paaA2-parE2 (Hallez et al., 2010).Such a way of controlling the cellular levels of TA system components combined with high proteolysis susceptibility of the antitoxin provides the way of tight and environmentally switchable regulation.The instability of the antitoxin in a TA system is a crucial step in the system activation.It is suggested that disordered C-terminal regions of the antitoxin are target for ATP-dependent serine proteases (Kamada et al., 2003).These members of chaperone family are responsible for degradation of misfolded proteins as well as components of signalling pathways (Gottesman, 1996).However, the antitoxin YgiT (MqsA) of the ygiUT (mqs-RA) system is structured throughout its entire sequence, both free and toxin-bound state (Brown et al., 2009).The activity of ATP-dependent proteases stays in a specific relation with the activity of TA systems.In all documented cases only a single protease is responsible for degradation of a particular antitoxin (although the proteases of interest comprise a family of related enzymes) (Van Melderen et al., 1994;Lehnherr & Yarmolinsky, 1995;Aizenman et al., 1996;Christensen et al., 2001;2004;Kawano et al., 2007;Christensen-Dalsgaard et al., 2010;Donegan et al., 2010).Degradation of the antitoxin component leads to subsequent toxin activation and increase in operon transcription in response to a toxin and antitoxin level imbalance.However, a halt of translation, induced for example by antibiotics, acts as another way of toxin activation by causing a drop in the production of labile antitoxin.
The significant influence of the TA systems on bacterial metabolism implies multiple ways of their activity regulation.A well documented mechanism is the relation between the mazEF system of E. coli and locus relA, which codes for ATP:GTP 3′-diphosphotransferase implicated in the synthesis of 3′,5′-guanosine bisphosphate (Justesen et al., 1986;Metzger et al., 1988).The ppGpp molecule is a signal of amino-acid starvation (Cashel, 1975;Gallant et al., 1976).The mazEF locus is located downstream the relA locus (Masuda et al., 1993) and is cotranscribed when relA expression is activated (Aizenman et al., 1996;Christensen et al., 2003;Hazan & Engelberg-Kulka, 2004).A similar neighbourhood pattern of the mazEF and parDE systems is found in genomes of other enteric bacteria such as Shigella and Salmonella (Pandey & Gerdes, 2005).Another example is the SOS system and its relations with various TA systems of E. coli.In this case the activation of SOS system leads to switching on the activity of TA systems including hokE (Fernandez De Henestrosa et al., 2000), yafNO (McKenzie et al., 2003;Christensen-Dalsgaard et al., 2010), tisAB (Vogel et al., 2004;Unoson & Wagner, 2008), symER (Kawano et al., 2007), and yefQ (Motiejunaite et al., 2007).A similar situation was recently reported for another E. coli TA system -yafNO (Singletary et al., 2009).
The activity of TA systems can also be induced by systems responsible for quorum sensing.Such a mechanism has been reported for the mazEF system of E. coli (Kolodkin-Gal et al., 2007).Another noteworthy fact is the possibility of cascade activation of TA systems (Hazan et al., 2001) since the bacteria often carry more than a single TA system within their genome.Activation of a single system which leads to protein synthesis inhibition and subsequent activation of another TA system is plausible.An even more complex relation has been described for the ygiUT (MqsRA) system of E. coli.In this case activation of the TA system is necessary for activation of toxin CspD, whose gene promoter is controlled by the ygiU/ygiT (MqsR/MqsA) complex (Brown et al., 2009;Kim et al., 2010).Furthermore, a cross-regulation has been observed for homologous systems present in the genome (Yang et al., 2010), where toxin-antitoxin complexes of one system bind to regulatory sequences of another TA system operon.

funcTIOns Of clAss II TA sysTeMs
A plasmid maintenance function was initially assigned to several newly discovered plasmid-borne TA systems (Gerdes & Molin, 1986;Saurugger, 1986;Bravo et al., 1988;Tsuchimoto et al., 1988;Gerlitz et al., 1990;Sobecky et al., 1996).Cells that do not inherit a copy of a plasmid upon division do not survive the effect of a stable toxin after degradation of a labile antitoxin.Moreover, a role of multiple TA loci in stabilization of a megaintegron of Vibrio cholerae has been suggested (Pandey & Gerdes, 2005).There is no doubt that TA systems play a role in the phenomenon of mobile genetic element stabilization but operons of many TA systems are also located in the bacterial chromosome.Recent studies report that TA systems are mainly concerned with the regulation of bacterial metabolism rather than simple plasmid maintenance functions.
Toxin activity leads primarily to bacterial metabolic dormancy that can be abolished at initial stages (Nystrom, 1999;Pedersen et al., 2002;Keren et al., 2004;Gerdes et al., 2005;Suzuki et al., 2005;Buts et al., 2005;Lewis, 2005;Inouye, 2006;Schumacher et al., 2009;Fu et al., 2009;Kasari et al., 2010), which contrasts with ear-figure 2. Binding of toxin-antitoxin complex to regulatory sequences leads to autorepression of TA operon expression Prokaryotic toxin-antitoxin systems lier suggestions that this activity leads to immediate cell death (Aizenman et al., 1996;Hazan & Engelberg-Kulka, 2004;Engelberg-Kulka et al., 2005).There are examples of such systems whose major role is to kill the cells, but this is only true in some specialized situations.A good example are formation of fruiting bodies of Myxococcus xanthus (Nariya & Inouye, 2008) or defence against phage infection in lactic acidic bacteria (Forde & Fitzgerald, 1999).The question whether TA system activity leading to death of selected cells in a colony is a manifestation of an altruistic or other mechanism is currently a topic of discussion (Aizenman et al., 1996;Forde & Fitzgerald, 1999;Nystrom, 1999;Lioy et al., 2006).
A flexible response of a bacterial cell to stress conditions seems to be the major function of most TA systems.A reversible metabolic dormancy caused by their activation allows a bacterial cell to survive detrimental conditions.This phenomenon provides clear advantages in the case of starvation (Christensen et al., 2001;Jorgensen et al., 2009) as well as heat, osmotic and freeradicals-induced stress (Pedersen et al., 2002;Senn et al., 2005).Moreover, TA systems can contribute to the formation of persistent cells during an exposure to antibiotics (Falla & Chopra, 1998;Keren et al., 2004;Dorr et al., 2010;Kasari et al., 2010).The mechanism of described phenomenon is straightforward in the case of drugs acting as transcription (eg.rifampicin) or translation (eg.chloramphenicol, doxycyclin, spectinomycin, eritromycin) inhibitors when the decay of the labile antitoxin causes the toxin activation.Paradoxically, antibiotics that are gyrase inhibitors (quinolone antibiotics) can act in a way similar to the ccdAB TA system, in which the toxin is a gyrase inhibitor.In this case binding of the inhibitor to an open gyrase-DNA complex induces DNA nicks (Drlica & Zhao, 1997;Jiang et al., 2002), which is followed by SOS-system activation (Little & Mount, 1982;Karoui et al., 1983;Bailone et al., 1985).The same mechanism is proposed for homologues of parDE system (Hallez et al., 2010).The described sequence of events leads to increased genetic diversity of a colony and may contribute to persisters formation (Couturier et al., 1998) in the same way as do quinolone antibiotics (Drlica & Zhao, 1997).
The activity of TA systems can also modulate the behaviour of a bacterial colony.An increase in the expression of genes related to cell motility and structural genes of flagella has been reported for the ygiUT (MqsRA) system (Gonzalez Barrios, 2006).In turn the hipAB system is implicated in biofilm formation providing multi drug resistance (Lewis, 2007;2008).TA systems can modulate formation of a biofilm over time (Kim et al., 2009).In line with that, a recent report indicates elevated expression of TA systems in bacterial cells building a biofilm (Mitchell et al., 2010).
A precise control over pathogenesis progression has been demonstrated for mRNA interferases exhibiting sequence specificity.This specificity allows for molecular evolution of target gene sequences.The mRNA interferases of the mazEF-mt3 and mazEF-mt7 systems are able to specifically recognize pentanucleotide sequences.In both cases a statistically significant representation of genes implicated in pathogenesis was found among genes containing underrepresented number of the recognized sequences (Zhu et al., 2008).Such genes are resistant to the interferase activity and thereby are expected to be expressed even when the TA system is activated.A similar relation was found for the sraP gene of S. aureus.This gene, coding for a protein responsible for ad-hesion to platelets (Siboo et al., 2005), is characterized by a statistically significant overrepresentation of the sequence recognized by the mRNA interferase of the maz-EF Sa TA system (Zhu et al., 2009), hence its expression is suggested to be primarily turned off upon TA system activation.Additionally, the mentioned TA system may potentially be implicated in pathogenesis progression in yet another way.Downstream of the mazEF Sa locus a sigB locus is located (Kullik et al., 1998;Gertz et al., 1999;Ferreira et al., 2004).The sigB-encoded alternative subunit σ B of the RNA polymerase is responsible for global transcription regulation of virulence factors, comprising one of the most important staphylococcal systems of gene regulation responsible for pathogenesis (Wu et al., 1996).In stress conditions the sigB locus is coexpressed with mazEF Sa (Senn et al., 2005;Fu et al., 2007;Donegan & Cheung, 2009).However, any potential functional relation demands further investigation since the elevated expression of sigB locus does not necessarily lead to a direct increase in the level of σ B subunit (Senn et al., 2005).Among other pathogenic strains also Bacillus anthracis possesses a TA system of the mazEF family, namely a pemIK module (Agarwal et al., 2007;2009).Recently a pemIK homologue located in a plasmid of an avian strains of S. aureus has been documented (Lowder et al., 2009;Bukowski et al., 2010).In this system the toxin is a sequence-specific endoribonuclease which targets a tetranucleotide sequence.Bioinformatic analysis of the occurrence of the recognized sequence in the coding sequences of the S. aureus genome elucidated a potential relation of the system with virulence factor regulation (Bukowski et al., 2010).

clAss II TA sysTeMs As BIOTechnOlOgIcAl TOOls
Two of the best-described TA systems have found application in molecular biology, namely ccdAB and mazEF.The former is used as a factor for positive selection of transformants, primarily in E. coli strains (Bernard et al., 1994).Such systems, which are commercially available (e.g.StabyCloning TM and StabyExpress TM , Delphi Genetics SA), are based on CcdB toxicity against gyrase and allow one-step selection of transformants ensuring stable vector plasmid maintenance (Fig. 3).This idea was originally developed by Szpirer and Milinkovitch (2005) followed by other efforts to develop a more complex system allowing increased production of recombinant protein (Stieber et al., 2008).figure 3. ccdAB system components as tools for positive selection during cloning The mazEF system has been adapted for single protein production (SPP) systems.The initial idea uses MazF toxin to trigger bacteriostasis and bacterial protein shutdown.The recombinant gene lacks the ACA sequences, recognized by the MazF interferase, therefore upon induction of MazF expression production of the recombinant protein of interest is continued almost exclusively.Moreover, bacteriostasis allows for culturing of the transformed strains in lower medium volumes than in traditional methods (Suzuki et al., 2005;2007).This idea has been successfully applied for protein production for NMR studies in 150-fold concentrated cultures, which allowed significant cost saving on isotopes (Mao et al., 2009;Schneider et al., 2009).Recently the SPP system based on MazF activity was extended with the capability for induction of protein production using particular amino acids.MazF mutants with histidine or tryptophan substitution were used in histidine or tryptophan auxotrophs, respectively.After transferring cells to the medium enriched in isotopes but lacking one of these amino acids the production of MazF is still provided.Subsequent addition of the amino acid induces exclusive production of the recombinant protein, since production of host proteins is blocked by the toxic action of MazF.Therefore, this approach allows not only single protein production but also high-efficiency isotope-labelling of the target protein (Vaiphei et al., 2010).
TA systems are successfully used also in studies on eukaryotic cells.Recently a report concerning the usage of mazEF system in studies on HIV virus was published (Chono et al., 2010).Further possible applications have already been suggested, such as TA-based contamination control in fermentation processes (Kristoffersen et al., 2000), antibacterial drug development (Engelberg-Kulka et al., 2004;Moritz & Hergenrother, 2007;Lioy et al., 2010), selectable elimination of cells in cell cultures, tissue cultures and whole organisms (de la Cueva-Mendez et al., 2003) or stable plasmid maintenance without antibiotic pressure (Wladyka et al., 2010).

cOncludIng ReMARKs
Results collected so far give a complex but concise image of the role of TA systems in bacterial physiology.Their functions range far beyond stabilization of mobile genetic elements.Metabolic dormancy induced by the systems seems a general but adequate response to various stress stimuli coming from the environment.Endoribonucleases, also termed mRNA interferases, are the most common group among the toxic components of various TA systems.Their activity leads to bacteriostasis through the inhibition of translation, which enables survival during starvation or antibiotic exposition.Further specialisation of interferases in selective sequence recognition allowed some genes to escape from expression suppression or, conversely, become exceptionally sensitive to a particular TA system.These phenomena are suggested to play a significant role in pathogen-host interaction and pathogenesis progression by modulation of biofilm formation and interactions with host proteins or coupling with other pathogen invasion-facilitating systems.
The relations among the ten families of class II TA systems are difficult to untangle.These TA systems are spread throughout the two huge domains of Archaea and Bacteria.Beside clear relationships, it seems that the similar way of acting and regulation of various groups of TA systems are due to convergence.Components of such systems could have evolved divergently from unrelated groups of genes to create autoregulated operons coding for pairs of toxic protein and its inhibitor.
The physiological functions of the TA systems became a base for their successful applications as molecular biology tools, both in industry and research.Primarily they facilitate maintenance of plasmid vectors and transformant selection, but also effective overexpression of recombinant proteins.The potential application of TA systems in antibiotic therapy cannot be omitted as it is known that TA systems induce bacteriostasis, whose prolongation results in bacterial cell death.With the growing knowledge of TA systems new useful applications are expected to be developed.