Bacteria grow in constantly changing environments that often limit growth or threaten their survival. To adapt or survive in such fluctuating environments, bacteria need to slow down their growth rate by multiple ways including redistribution of metabolic pathways and nutrient transport, shutdown of translation, or alteration of gene expression of components involved in translation, replication of the genome, and division of the cell walls (1). The ability to adapt to unfavorable environments is largely dependent on how bacteria quickly respond to environmental stimuli and modulate gene expression to adjust the bacterium’s growth in a newly given environment.
Toxin-antitoxin systems are one of such genetic elements that directly regulate bacterial growth in response to a variety of cellular stresses including nutrient limitation, SOS response, heat shock, bacteriophage infection, and antibiotic treatment (2, 3). Toxin-antitoxin systems consist of two genes encoding the toxin and its cognate antitoxin, respectively. Toxin proteins arrest bacterial growth by inhibiting DNA gyrase, degrading messenger RNAs, or modifying the ribosomal components, most of which disrupt translation process (4-15). Considering that translation is the most energy-demanding process and global translation efficiency is one of the major limiting factors for bacterial growth rate (16, 17), it is not surprising that the toxins are predominantly involved in inhibiting protein synthesis. Among these toxins, the majority of toxin proteins are endoribonucleases degrading mRNAs (2, 3, 18, 19). The endoribonuclease toxins recognize and cleave defined mRNA sequences depending on its structures and residues at the active sites (20). In this review, we summarize the current understanding of bacterial endoribonuclease toxins focusing on the substrate specificity of the toxins and underlying mechanisms.
Toxin-antitoxin systems are classified into six different types depending on how the antitoxin recognizes and antagonizes the toxin protein (3). For example, in type I toxin-antitoxin system, the antitoxins are antisense RNAs that bind to the translation initiation regions of the toxin genes and inhibit translation of the toxin mRNAs. By contrast, type II antitoxins are proteins and inhibit toxin’s activity by directly binding to the toxin proteins with different molecular ratios (toxin : antitoxin ratio of 2 : 2, or 4 : 2) (18). Type III antitoxins are similar to type I antitoxins in a sense that they are small noncoding RNAs. However, unlike type I antitoxins, type III antitoxin RNAs directly bind to the toxin proteins and inhibit the toxin’s activity instead of blocking translation of the toxin genes by base-pairing (Fig. 1).
RelE is a representative endoribonuclease toxin among bacterial toxin-antitoxin systems. RelE specifically cleaves ribosome-bound mRNAs and has no activity on free mRNAs
Similarly to RelE, YoeB toxin has a broad codon specificity. It cleaves mRNAs between positions 2 and 3 in the UAA stop codon and AAU Asn codon and after the third base of AAA Lys codon (20, 31, 32). It was also reported that YoeB toxin also cleaves between the second and third bases of AAA Lys, CUG Leu, and GCG Ala codons (Table 1) (29). Given that most of mRNA cleavage sites are located close to the AUG initiation codon, it was suggested that YoeB inhibits translation initiation (20). Structural analyses showed that the second nucleotide of mRNA codon in the A-site specifically interacts with YoeB Lys49, while the first and third nucleotides of mRNA codon lack such base-specific interactions (Fig. 2B) (30, 31), which explain the wide range of codon specificity of YoeB.
Like RelE, YafO toxin has endoribonuclease activity when it is bound to ribosome (34). YafO binds to the 50S subunit in the 70S ribosome and induces mRNA cleavage. However, unlike RelE and YoeB that cleave mRNAs at the A-site, YafO cleaves mRNAs 11 to 13 nucleotides downstream of the AUG start codon (34). The cleavage location corresponds to the 3’ end of mRNAs that are protected by the 70S ribosome initiation complex (34), indicating that YafO is a ribosome-associated endonuclease inducing mRNA cleavage outside of the ribosome. Interestingly, YafO-mediated mRNA cleavage requires ribosome binding but not translation
HigB toxin is a ribosome-dependent RelE-family ribonuclease and cleaves mRNAs between the second and third bases at adenosine-rich codons. Interestingly, although HigB preferentially cleaves AAA Lys and ACA Thr codons, HigB cleaves basically any codon containing adenosine (40). An explanation for the unique selectivity of adenosine was provided by structural analyses of HigB toxin bound to AAA or ACA codons in the A-site of the 70S ribosome (41, 42). HigB does not have a specific interaction with the first position in the codon that allows any nucleotide to be recognized. HigB preferentially interacts with A or C at the second position in the codon, thus A or C being the most effective nucleotide for HigB-mediated mRNA cleavage. At the third position, HigB interacts with C1054 in the 16S rRNA to form an adenosine-specific pocket to accommodate adenosine most efficiently. The HigB Asn71 residue is critical to determine the adenosine specificity of the third position (Fig. 2D) (41). Additionally, His54, Asp90, Tyr91, and His92 residues in the HigB toxin were determined to be critical for the endoribonuclease activity of HigB toxin (Fig. 2D) (42).
The RNA substrate specificity of MazF toxin varies in different species.
MqsR toxin turned out to be a ribosome-independent endoribonuclease in
Interestingly, HicA toxin cleaves mRNAs with no sequence specificity (7). Because HicA also cleaves tmRNA at specific AAAC sequences (A^AAC), HicA toxin appeared to be a ribosome-independent endoribonuclease (7). However, the detailed mechanism of RNA recognition by HicA toxin needs to be elaborated.
SymE toxin appeared to have a ribosome-independent ribonuclease activity given that most of the tested mRNAs were cleaved upon SymE induction (70). However, the specific cleavage sites of SymE toxin have yet to be determined. The predicted structure of SymE toxin is unique in a sense that it is similar to AbrB superfamily, a protein fold typically observed in antitoxins such as MazE (70). It would be interesting to understand how the antitoxin-folded SymE toxin recognizes and cleaves RNA substrates.
AbiQ toxin has a sequence-dependent endoribonuclease activity that cleaves an adenine-rich sequence within the
In another plasmid-encoded ToxIN from
Here we summarized the bacterial toxins with endoribonuclease activity. The toxin components are organized in pairs with its cognate antitoxin components. For example, type II toxin-antitoxin systems consist of toxins and its cognate antitoxin pro-teins that are organized as bicistronic operons. The expression of the bicistronic operon is mostly repressed by the antitoxin protein, which has a DNA-binding domain for auto-repressor activity and a toxin-binding domain for neutralizing the toxin’s activity. Stress conditions including amino acid starvation promote the differential degradation of the labile antitoxins by Lon or ClpAP proteases (78), and the removal of antitoxins results in an increase in the expression of the toxin-antitoxin operon (Fig. 1). The molecular ratio between toxin and antitoxin proteins appears to be tightly regulated because most of the antitoxin and toxin genes are bicistronic and translationally coupled. Generally, the antitoxin gene precedes the toxin gene, which is also likely to ensure an appropriate molecular ratio between toxin and antitoxin proteins. However, the
Toxin-antitoxin systems are expressed in response to diverse cellular stresses, including nutrient starvation, stringent response, or exposure to acidic pH (8). In type II toxin-antitoxin systems, these stress conditions preferentially degrade antitoxin proteins by Lon or ClpAP proteases, leading to expression of the toxin genes. In addition to the multiple stress response-mediated antitoxin degradation, SOS response is also suggested to be an inducing signal for several ribonuclease toxin operons includeing the
Ribonuclease toxins cleave mRNAs with a diverse range of substrate specificity. RelE ribonuclease toxin requires ribosome to cleave mRNAs and it cleaves mRNA in a codon-dependent manner but with loose codon specificity (5, 25). YafQ ribonuclease toxin is also ribosome-associated but cleaves mRNAs at mostly AAA Lys codon, showing narrow substrate specificity (36, 37, 41). As an opposite extreme, MazF ribonuclease toxin does not require ribosome and cleaves mRNAs in a codon-independent and sequence-dependent manner (4, 6). Considering such diverse substrate specificity, it is not surprising that the overall sequence similarity of endoribonuclease toxins is low. However, these endoribonuclease toxins have a strikingly similar protein structure depending on toxin types, which raises a question about the factors determining its substrate specificity. Type II endoribonuclease toxins have a common microbial RNase protein fold similar to RNase T1 and RNase Sa2 (Fig. 2) (81-84). SymE type I toxin has a protein fold similar to MazF endoribonuclease (70). Type III endoribonuclease toxins, ToxN and AbiQ, are also homologs of MazF endoribonuclease toxin with additional residues for antitoxin RNA binding (76). And ToxN and AbiQ toxins are structurally similar to each other and superimposable (73, 75, 76). Interestingly, the active site residues of the endoribonuclease toxins are highly variable and thus the amino acid compositions within the active sites appear to determine the substrate specificity. For example, RelE toxin lacks residues required for ribonuclease activity that were found in RNase T1. Instead, 16S rRNA C1054 with the ribosome provides a base required for recognition of specific mRNA codons and its ribonuclease activity (25). By contrast, YafQ ribonuclease toxin harbors active site residues for recognizing mRNAs similar to RNase T1, explaining the ribonuclease activity of the purified YafQ toxin
The biological roles of endoribonuclease toxins were suggested to inhibit bacterial growth in response to multiple stressful conditions such as nutrient starvation, SOS response, and bacteriophage infection (2, 3). Such inhibition of bacterial growth contributes to antibiotic tolerance, persister cell formation, biofilm, colonization in the host, and abortive infection (65, 85-89), some of which are controversial (63, 90, 91). To understand the biological role of the endoribonuclease toxins and the underlying mechanisms, mRNA substrate specificity of each toxin needs to be determined in the context of the bacterium’s niche and physiology.
This work was supported by Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Science, ICT and Future Planning [NRF-2019R1A2C2003460 to E.-J.L.], the Bio & Medical Technology Development Program of the National Research Foundation (NRF) and funded by the Korean government (MSIT) [NRF-2020 M3A9H5104235 to E.-J.L.], and a grant from Korea University.
The authors have no conflicting interests.