A new type V toxin-antitoxin system where mRNA for toxin GhoT is cleaved by antitoxin GhoS

Toxin-antitoxin systems are found in nearly all bacterial chromosomes1, which attests to their importance in cell physiology. Toxin-antitoxin systems are classified as type I if the antitoxin RNA prevents the translation of toxin RNA, type II if the antitoxin protein binds and inhibits the toxin protein and type III if the antitoxin RNA binds and inhibits the protein toxin2. Also, a type IV designation has been proposed recently for a toxin-antitoxin system in which the protein antitoxin interferes with binding of the toxin to its target rather than inhibits the toxin via direct toxin-antitoxin binding3. Toxins inhibit growth (for example, inhibit translation via mRNA degradation), and antitoxins reduce toxin activity; however, antitoxins are generally labile under various stress conditions, which results in toxin activation2. The role of toxin-antitoxin systems in cell physiology, specifically in biofilm formation4,5, persister cell formation6,7 and the general stress response8,9, is becoming more clear. Notably, mRNA endoRNase toxins are becoming recognized as global regulators that alter gene regulation by cleaving specific mRNAs (termed differential mRNA decay)10. For example, upon antibiotic stress, toxin MazF degrades most mRNAs with ACA sequences. However, its activity also results in the preferential synthesis of a subset of small proteins whose mRNAs are not degraded11. As these enriched proteins are necessary both for toxicity and for survival11, MazF acts as a regulatory factor12. The toxin motility quorum sensing regulator (MqsR, YgiU/B3022)4,13 is also a global regulator13,14 and is conserved in 40 eubacteria13. Its specific mRNA endoRNase activity leads to enrichment of mRNAs that code for the stress-associated proteins CstA, CspD, RpoS, Dps and HokD14. In addition, 14 Escherichia coli mRNA transcripts do not contain the MqsR-preferred GCU cleavage site15,16, and 6 of these (pheL, tnaC, trpL, yciG, ygaQ and ralR) are differentially regulated in biofilms17. Another one of these 14 transcripts that lacks GCU sites is yjdO (B4559, renamed here as ghoT for ‘toxinproducing ghost cells’); the protein it encodes is conserved in E. coli and Shigella spp. and has not been previously characterized. As an indicator of its impact on cell physiology through differential mRNA decay, MqsR is the first toxin that, upon inactivation, decreases the formation of persister cells6. Persister cells are a small fraction of bacteria that show tolerance to antibiotics without genetic change18; it is believed that they survive antibiotic treatment by becoming metabolically dormant19. The crucial regulator of MqsR toxicity, antitoxin MqsA (YgiT/B3021)20, is the first antitoxin shown to be a global regulator, as transcription of loci such as rpoS are derepressed upon MqsA degradation during oxidative stress9,14 and have critical roles in bacterial cell physiology during stress. It is well established that endoRNase toxins, including MqsR6 and RelE21, and the kinase HipA21,22 inhibit protein synthesis, which is correlated with the formation of persister cells and, in turn, an increase in multidrug tolerance. Moreover, isolated persister cells also show increased transcription of the toxin genes mqsR23, relE21 and mazF 21. The mechanism (or mechanisms) underlying the increased persistence observed upon expression of these toxins, however, has not been fully characterized. Here we present evidence that the product of one of the transcripts that lacks the primary MqsR GCU site, GhoT, increases persistence and that GhoT-GhoS (YjdO-YjdK) is a new toxin-antitoxin system. We show that toxin GhoT, when produced, leads to both cell death and, in the absence of cell death, an increase in persister cells. Moreover, GhoT-GhoS is to our knowledge the first non–type I chromosomal toxin-antitoxin system that encodes a presumed membrane-lytic protein.