YoeB toxin is activated during thermal stress

Type II toxin-antitoxin (TA) modules are thought to mediate stress-responses by temporarily suppressing protein synthesis while cells redirect transcription to adapt to environmental change. Here, we show that YoeB, a ribosome-dependent mRNase toxin, is activated in Escherichia coli cells grown at elevated temperatures. YoeB activation is dependent on Lon protease, suggesting that thermal stress promotes increased degradation of the YefM antitoxin. Though YefM is efficiently degraded in response to Lon overproduction, we find that Lon antigen levels do not increase during heat shock, indicating that another mechanism accounts for temperature-induced YefM proteolysis. These observations suggest that YefM/YoeB functions in adaptation to temperature stress. However, this response is distinct from previously described models of TA function. First, YoeB mRNase activity is maintained over several hours of culture at 42°C, indicating that thermal activation is not transient. Moreover, heat-activated YoeB does not induce growth arrest nor does it suppress global protein synthesis. In fact, E. coli cells proliferate more rapidly at elevated temperatures and instantaneously accelerate their growth rate in response to acute heat shock. We propose that heat-activated YoeB may serve a quality control function, facilitating the recycling of stalled translation complexes through ribosome rescue pathways.


Introduction
In Escherichia coli, prolonged translational arrest often leads to mRNA degradation into the ribosome A site (Hayes and Sauer 2003;Sunohara et al. 2004;Li et al. 2006Li et al. , 2008Garza-S anchez et al. 2008). This A-site mRNA cleavage activity results in truncated A-site codons, which prevent further decoding and produce stalled translation complexes. Such non-productive ribosomes are "rescued" by at least three quality control systems in bacteria (Janssen and Hayes 2012). tmRNA-SmpB mediates the primary ribosome rescue pathway, and this system is found in all eubacteria and some plastids (Tu et al. 1995;Keiler et al. 1996Keiler et al. , 2000Gueneau de Novoa and Williams 2004). tmRNA is a stable RNA with both transfer-RNA and messenger-RNA functions and recycles stalled ribosomes in a process termed trans-translation. During trans-translation, tmRNA-SmpB enters the ribosome A site, and the nascent peptide is transferred to alanine-charged tmRNA. The truncated message is then released from the ribosome and translation resumes using a short open reading frame within tmRNA. In this manner, tmRNA provides a stop codon in trans, thereby allowing normal translation termination and ribosome recycling (Hayes and Keiler 2010). More recently, two alternative ribosome rescue pathways have been identified.
ArfA (alternative rescue factor A) was discovered in a synthetic lethal screen for genes that are essential for the viability of mutants that lack tmRNA (Chadani et al. 2010). ArfA is a small peptide that binds the ribosome and allows release factor-2 to catalyze nascent chain release in the absence of an A-site stop codon (Chadani et al. 2012;Shimizu 2012). YaeJ (ArfB) is a release factor homolog that lacks the canonical stop codon recognition domain (Baranov et al. 2006;Hayes and Keiler 2010). Thus, YaeJ can bind the ribosome A site in the absence of a stop codon and catalyze nascent peptide release (Chadani et al. 2011;Handa et al. 2011;Gagnon et al. 2012;Feaga et al. 2014). Because A-site cleavage produces nonstop transcripts in response to translational pausing, this nuclease activity is thought to function in molecular quality control by facilitating ribosome rescue (Hayes and Sauer 2003;Sunohara et al. 2004).
In E. coli there are at least two enzymes, RelE and YoeB, which have ribosome-dependent A-site nuclease activity Feng et al. 2013). RelE and YoeB are encoded by type II toxin-antitoxin (TA) modules together with cognate antitoxins that specifically neutralize nuclease activity. These mRNases have been termed "toxins" because their activities inhibit cell growth and can lead to cell death under some circumstances. In general, toxins are activated under stress or other conditions that prevent continued antitoxin synthesis. Antitoxins are labile to proteolysis and their degradation liberates the cognate toxins to exert growth inhibition activities. The physiological functions of TA systems remain controversial. They have been proposed to play roles in stressresponse, persistence, genomic stability and programmed cell death (Engelberg-Kulka et al. 2005;Gerdes et al. 2005;Magnuson 2007;Tsilibaris et al. 2007;Nariya and Inouye 2008;Maisonneuve et al. 2011). Though RelE and YoeB have the potential to catalyze A-site mRNA cleavage during translational arrest, mutants lacking these enzymes and other known toxins retain A-site cleavage activity (Hayes and Sauer 2003;Sunohara et al. 2004;Garza-S anchez et al. 2009;Janssen et al. 2013). Thus, the identity of the A-site nuclease is unknown, but it remains possible that an unidentified toxin catalyzes cleavage in response to the stress of translational pausing.
Though the A-site nuclease has not been identified, the phenomenon clearly requires the 3 0 -to-5 0 exoribonuclease activity of RNase II (Garza-S anchez et al. 2009;Janssen et al. 2013). In Δrnb mutants, which lack RNase II, prolonged translational arrest produces transcripts that are truncated 12 nucleotides downstream of the ribosome A site (Garza-S anchez et al. 2009;Janssen et al. 2013). This position corresponds to the "toeprint" of the paused ribosome, suggesting that the ribosome physically blocks further mRNA degradation into the A site. However, purified RNase II only degrades mRNA to the +18 position (with respect to the A-site codon) when incubated with translation complexes in vitro (Garza-S anchez et al. 2009). Moreover, the deeply recessed active site of RNase II is incompatible with the ability to directly catalyze A-site mRNA cleavage (Frazao et al. 2006;Zuo et al. 2006). Because RNase II is unable to degrade the A-site codon itself, the enzyme must play an indirect role in A-site mRNA cleavage. One possibility is that RNase IImediated mRNA degradation to the 3 0 -edge of the ribosome is required for subsequent degradation into the A site by an unknown nuclease. This model is supported by experiments showing that secondary structures placed on the 3 0 -side of paused ribosomes are sufficient to inhibit RNase II activity and also block A-site cleavage (Garza-S anchez et al. 2009).
We previously reported that A-site mRNA cleavage is suppressed in response to heat shock at 42°C (Garza-S anchez et al. 2009). This observation suggests that A-site nuclease activity is heat labile, and therefore we revisited these experiments in an effort to identify the enzyme responsible. Although acute heat shock temporarily inhibits A-site mRNA cleavage, the experiments presented herein demonstrate that prolonged growth at elevated temperature actually restores the activity. This heatinduced mRNase activity is dependent on the Lon protease, suggesting the activation of one or more TA modules during thermal stress. Taking a candidate-gene approach, we discovered that the yefM-yoeB TA operon is required for heat-induced A-site mRNA cleavage activity. Remarkably, YoeB mRNase activity remains constant over several hours of culture at 42°C. Cells proliferate rapidly under these conditions, demonstrating that growth arrest is not necessarily concomitant with toxin activation. Moreover, global protein synthesis is not suppressed by the active toxin. Together, these findings show that type II toxins can be activated at low levels in response to environmental stress. We propose that YoeB facilitates the recycling of stalled translation complexes, thereby playing a role in molecular quality control during thermal stress.

Bacterial strains and growth conditions
All bacterial strains were derived from E. coli X90 (DE3) and are listed in Table 1. The ΔclpB::kan, ΔhtpG::kan, Δlon::kan, and ΔclpP::kan alleles were obtained from the Keio collection (Baba et al. 2006) and were transferred between strains using bacteriophage P1-mediated transduction (Moore 2011). The ΔdnaK::kan disruption was generated as described (Hayes and Sauer 2003   The two polymerase chain reaction (PCR) products were sequentially ligated to plasmid pKAN to generate pKAN-dnaK. Plasmid pKAN-dnaK was digested with SacI/KpnI and the small fragment was used to delete the dnaK gene by Redmediated recombination (Thomason et al. 2014). Kanamycin-resistant cassettes were removed with flippase (FLP) recombinase to allow the construction of strains carrying multiple gene deletions (Cherepanov and Wackernagel 1995). All mutations were confirmed by locus-specific PCR amplification. Escherichia coli cells were grown in lysogeny broth (LB) media supplemented with the appropriate antibiotics (ampicillin, 150 lg/mL; chloramphenicol, 66 lg/mL; kanamycin, 50 lg/mL; and tetracycline 25 lg/mL). Cells from overnight cultures were resuspended at optical density (OD 600 )~0.05 in fresh LB supplemented with the appropriate antibiotics. Cells were grown to mid-log phase (at the indicated temperatures) with shaking, then flag-(m) ybeL-PP expression was induced with isopropyl b-D-1-thiogalactopyranoside (IPTG) at a final concentration of 1.5 mmol/L. After 30 min, the induced cultures were poured into an equal volume of ice-cold methanol to arrest growth. The effects of Lon, YefM, YoeB, RelB, and r 32 on mRNA processing were determined by expressing the corresponding genes from a plasmid-borne P BAD promoter (Hayes and Sauer 2003). Cells were grown to mid-log phase at the indicated temperatures and flag-(m)ybeL-PP induced with 1.5 mmol/L IPTG. After 30 min, L-arabinose was added to 0.4% and incubation continued for an additional 15 min. Cultures were then poured into an equal volume of ice-cold methanol. Cells were harvested by centrifugation and the cell pellets frozen at À80°C for subsequent RNA isolation. The growth-rate response to acute heat shock was determined in LB medium without antibiotics. Cells were grown for 2.5 h at 30°C in an environmental shaker. Cultures were then split in two, with one culture maintained at 30°C while the other was transferred to a shaking-water bath equilibrated at 42°C. The growth of each culture was monitored during the acute heat shock and for an additional 4.5 h. Long-term growth at various temperatures was performed on LB agar without antibiotics. Cells were grown to mid-log phase in shaking LB medium at 30°C. Culture density was adjusted to OD 600 = 1.0, then subjected to 10-fold serial dilutions in LB medium. Samples (2 lL) from each dilution were spotted onto LB agar and incubated at the indicated temperatures for 15 h.

Immunoblot analysis
Protein was extracted from frozen cells using freeze-thaw in urea lysis buffer (50% urea, 10 mmol/L Tris-HCl [pH 8.0], 150 mmol/L NaCl) and lysates clarified by centrifugation at 15,000 9 g for 15 min (Hayes et al. 2002b). Protein was quantified by the Bradford method and equal amounts of total urea-soluble protein were run on Tristricine sodium dodecyl sulfate (SDS) 10% polyacrylamide gels. Immunoblot analysis was performed as described (Janssen and Hayes 2009), and blots were imaged using a Odyssey â infrared imager (LI-COR, Lincoln, NE, USA). Lon antigen was detected with rabbit polyclonal antisera and IRDye â 680 (LI-COR, Lincoln, NE, USA) labeled anti-rabbit secondary antibodies.

A-site nuclease activity is induced at elevated temperature
We previously reported that A-site mRNA cleavage is suppressed during heat shock (Garza-S anchez et al. 2009), suggesting that the A-site nuclease is thermolabile. Alternatively, abrupt increases in temperature could disrupt protein synthesis temporarily and therefore indirectly affect A-site cleavage. To test this possibility, we revisited these experiments using the previously described flag-(m)ybeL-PP transcript as a reporter of Asite mRNA cleavage (  Hayes et al. 2002a;Hayes and Sauer 2003;Garza-S anchez et al. 2008). A substantial proportion of flag-(m)ybeL-PP transcripts is truncated in the stop codon when expressed in ssrAmutants, which lack tmRNA, but not in ssrA + cells (Fig. 1B, compare lanes 2 and 4). Presumably, mRNA processing also occurs in wild-type cells, but the truncated messages are rapidly degraded once released from the stalled ribosome through tmRNA activity (Hayes and Sauer 2003;Yamamoto et al. 2003). As reported (Garza-S anchez et al. 2009), RNase II (encoded by the rnb gene) is required for this activity because truncated transcripts were not detected in ssrA À Δrnb cells (Fig. 1B, lane 6). We next examined RNA isolated from cells that had been grown at 42°C for 2 h, and unexpectedly found that A-site mRNA cleavage was not suppressed (Fig. 1B, lane 3). In fact, growth at 42°C actually restored A-site cleavage activity to ssrA À Δrnb cells (Fig. 1B  and 7). Another smaller truncated transcript was also detected in ssrA À cells grown at 42°C (Fig. 1B, marked by an arrow in lanes 3 and 7). These observations suggest that a new RNase activity is induced during growth at 42°C. Because tmRNA promotes nonstop mRNA turnover during ribosome rescue (Yamamoto et al. 2003;Richards et al. 2006;Ge et al. 2010), the apparent absence of truncated mRNA in ssrA + cells suggests that cleavage may occur during translation. We tested this hypothesis by introducing an amber stop codon at position Glu28 in the flag-(m)ybeL-PP coding sequence (Fig. 1A). Thus, ribosomes terminate translation at codon 28 in the Glu28Am transcript, rather than the original ochre stop at codon 61. Truncated Glu28Am transcripts were not detected in any of the examined backgrounds (Fig. 1C), indicating that translation to the tandem Pro codons is required for the heat-induced mRNase activity.
YoeB mediates A-site mRNA cleavage at elevated temperature Some type II TA modules encode ribosome-dependent RNases that cleave A-site codons Prysak et al. 2009;Feng et al. 2013). Toxins are typically activated through Lon protease-mediated degradation of antitoxins (Gerdes and Maisonneuve 2012;Brzozowska and Zielenkiewicz 2013); therefore we examined flag-(m) ybeL-PP transcript processing in Δlon cells. We observed less truncated mRNA in Δlon compared to lon + backgrounds even at 37°C ( Fig. 2A, lanes 2 and 4), indicating that Lon influences A-site cleavage during translational pauses. But more importantly, we failed to detect heatinduced mRNase activity in Δlon mutants ( Fig. 2A, lanes  7 and 9). The latter result suggests that a toxin is responsible for heat-induced mRNase activity, which led us to test an E. coli strain that lacks multiple TA genes. The E. coli Δtox(6) strain lacks six validated TA systems encoded by the relBE, mazEF, yefM-yoeB, dinJ-yafQ, chp-BIK and yhaV genes. As reported previously (Garza-S anchez et al. 2009), flag-(m)ybeL-PP transcripts still undergo A-site cleavage in the ssrA À Δtox(6) background at 37°C (Fig. 2B, lanes 2 and 4). However, ssrA À Δtox (6) Δrnb cells did not exhibit heat-induced mRNase activity (Fig. 2B, lanes 7 and 9), strongly suggesting that one (or more) of the deleted toxins is responsible for activity. Further analysis of ssrA À Δrnb strains carrying individual TA gene deletions revealed that ΔyefM-yoeB mutants lack the heat-induced activity (Fig. 2C). The YefM antitoxin specifically binds to YoeB toxin and neutralizes its RNase activity (Cherny et al. 2005;Kamada and Hanaoka 2005;Feng et al. 2013). Therefore, if heat-induced transcript cleavage is mediated by YoeB, then the activity should be specifically blocked by yefM overexpression. We cloned yefM under control of the P BAD promoter and induced expression in cells that coexpress flag-(m)ybeL-PP. In cells grown at 37°C, yefM expression had no discernible effect on mRNA cleavage (Fig. 3A, lanes 2, 3, 6, and 7). However, yefM expression suppressed mRNA cleavage in ssrA À Δrnb cells at 42°C ( Fig. 3A, lane 8 and 9). This suppressive effect was specific because induction of relB, which encodes the antitoxin for RelE toxin, had little effect on heat-induced mRNase activity (Fig. 3A, bottom blot). We then expressed yoeB from a plasmid-borne P BAD promoter to determine whether the toxin cleaves flag-(m)ybeL-PP transcripts. The two major truncated species that accumulate during thermal stress were also produced in response to yoeB induction at 37°C (Fig. 3B, lanes 2 and 4). Thus, the heat-induced mRNA activity can be recapitulated by yoeB expression at lower temperature. In contrast, expression of the entire yefM-yoeB operon from the same plasmid vector did not induce mRNA cleavage (Fig. 3B, lanes 3  and 5). Taken together, these data indicate that YoeB is responsible for the heat-induced mRNase activity.
YoeB is activated during heat-shock stress Christensen et al. 2004 have shown that overproduced Lon inhibits cell growth largely by activating YoeB. These findings indicate that YefM is particularly susceptible to proteolysis and suggest that thermal activation may be due to increased Lon levels. To address this possibility, we first confirmed that truncated flag-(m)ybeL-PP transcripts accumulate in cells that overexpress lon at 37°C (Fig. 4A, lane 2). This mRNA processing was not observed when lon was induced in ΔyefM-yoeB cells (Fig. 4A, lane 4). We also determined that protease activity is required for this effect, because truncated messages were not detected when catalytically inactive Lon(Ser679-Ala) was overproduced (Fig. 4A, lane 3) (Botos et al. 2004). Having established that increased Lon is sufficient to activate YoeB in our system, we then examined endogenous Lon levels by immunoblot. Given that Lon is considered to be a heat-shock protein (Phillips et al. 1984), we were surprised to find that Lon antigen levels were remarkably constant in cells cultured for 2.5 h at 30°C, 37°C, and 42°C (Fig. 4B). Lon levels were also unchanged regardless of ssrA or rnb genetic background (Fig. 4B). We further examined whether Lon levels increase transiently in response to temperature up-shift, but observed no significant increase over the first few minutes of heat stress (Fig. 4B, lower blot). These data show that Lon overproduction can activate YoeB, but this mechanism does not account for activation at elevated temperature.
Heat shock activates the transcription of several genes that facilitate adaptation to thermal stress. We reasoned that an additional heat shock factor may collaborate with Lon to degrade YefM and therefore account for activation at high temperature. To test this hypothesis, we induced the heat-shock regulon through ectopic expression of rpoH, which encodes the r 32 heat-shock transcription factor (Nonaka et al. 2006;Guisbert et al. 2008). Expression of rpoH at 37°C induced an mRNase activity that is indistinguishable from that observed in cells grown at 42°C (Fig. 4C, lanes 3 and 5). This activity was dependent on lon and yefM-yoeB, but not clpP (Fig. 4C, lanes 8, 10, 12, and 14). We tested whether Lon increases in response to rpoH expression and found that protease levels remained constant (Fig. 4D). Because YoeB is activated at lower temperature through induction of the heat-shock regulon, we tested whether known ATP-dependent chaperones facilitate activation. ClpB, HtpG, and DnaK are all induced during heat shock and have been implicated in the degradation and/or refolding of proteins at high temperature (Sherman and Goldberg 1992;Mogk et al. 1999;Thomas and Baneyx 2000;Huang et al. 2001). Therefore, we examined flag-(m)ybeL-PP transcripts expressed in ssrA À Δrnb cells that carry additional deletions of the clpB, htpG, or dnaK genes. The ΔclpB and ΔhtpG mutations had no effect on transcript processing at either 37°C or 42°C, but the ΔdnaK mutant showed increased mRNase activity even at 37°C (Fig. 5A). In fact, yet another truncated flag-(m)ybeL-PP transcript accumulated in ssrA À Δrnb ΔdnaK cells (Fig. 5A, indicated by lowest  arrow in lanes 6 and 7). This additional truncation product presumably arises from YoeB-mediated cleavage at an upstream codon within the flag-(m)ybeL message. dnaK mutations are known to induce the heat-shock regulon at low temperature (Straus et al. 1990), suggesting that YoeB is activated in ΔdnaK cells by the same mechanism that underlies toxin activation during ectopic rpoH expression.
In accord with this model, we found that lon and yefM-yoeB are both required for increased mRNase activity in the ΔdnaK background (Fig. 5B, lanes 3 and 4). Additionally, immunoblot analysis showed that Lon protease levels do not increase dramatically in ΔdnaK mutants (Fig. 5C). Collectively, these results demonstrate that YoeB is activated under three conditionshigh temperature, rpoH expression and dnaK mutationthat induce the heat-shock regulon.

Thermal activation of YoeB does not inhibit global protein synthesis
Previous work has shown that YoeB cleaves ompA and lpp transcripts in E. coli (Winther and Gerdes 2009;Zhang and Inouye 2009). Therefore, we tested whether YoeB cleaves these endogenous mRNAs in response to elevated temperature. We were unable to detect truncated lpp transcripts using probes to the 5 0 -and 3 0 -untranslated regions (Fig. 6A). Similarly, the ompA transcript was not cleaved during culture at 42°C (data not shown). We also considered the possibility that YoeB may preferentially cleave heat-shock transcripts and provide a mechanism to finetune their translation. However, we did not detect yoeBdependent cleavage in grpE and ibpB transcripts in response to growth at 42°C (Fig. 6B). Although smaller ibpB transcripts were detected at 42°C in ssrA À cells, these fragments accumulated to similar levels in both yoeB + and ΔyoeB backgrounds (Fig. 6B). These data are seemingly at odds with the observation that flag-(m)ybeL-PP reporter transcripts are efficiently cleaved under the same growth conditions. We hypothesize that YoeB activity is focused on the reporter transcript due to a combination of overexpression and inefficient translation termination. We have shown previously that the entire pool of release factor-1 (RF-1) is sequestered on paused ribosomes during ybeL-PP overexpression (Janssen and Hayes 2009). RF-1 depletion then causes other ribosomes to stall at the ybeL-PP termination codon with unoccupied A sites (Janssen and Hayes 2009). Because YoeB must compete with translation factors to gain access to its A-site codon substrate, we propose that these latter ribosomes with unoccupied A sites are preferentially targeted by the nuclease.

Cell growth is not inhibited during YoeB activation
The stress-response model of TA function postulates that environmental stress activates mRNase toxins to temporarily inhibit translation while transcription is redirected to stress-response genes (Gerdes et al. 2005). Because YoeB is activated at high temperature, we asked whether acute heat shock leads to temporary growth arrest as predicted by the stress-response model. We monitored the growth of wild-type yoeB + cells and observed no inhibi-  tion of growth during the transition from 30°C to 42°C (Fig. 7A). Moreover, ΔyefM-yoeB mutants grew along the same trajectory as yefM-yoeB + cells during the heat-shock treatment (Fig. 7A). These results indicate that YoeB activity is not sufficient to arrest growth, consistent with the absence of detectable cleavage within endogenous messages at 42°C. tmRNA facilitates ribosome recycling after YoeB-mediated A-site cleavage, so we also examined the response of ssrA À cells to heat shock. Although ssrA À mutants grow more slowly than ssrA + cells at 30°C and 42°C, the ΔyefM-yoeB mutation had little effect in this background (Fig. 7B). Because YoeB activity persists over at least 2 h of culture at 42°C, we examined whether the toxin influences cell growth over longer time scales. We spotted yefM-yoeB + and ΔyefM-yoeB cells onto LB agar for overnight incubation at temperatures ranging from 30°C to 46°C, but observed no differences in cell growth in various ssrA and lon backgrounds (Fig. 8). In fact, the most prominent growth defects were observed with ssrA À cells at 46°C and Δlon cells at 30°C. The ssrA mutation reduced cell growth at 46°C about 10-fold in the yefM-yoeB and lon backgrounds (Fig. 8). Additionally, Δlon cells grew more slowly than lon + cells on solid medium at 30°C, and this effect was even more pronounced in the ssrA À background (Fig. 8). Together, these results indicate that tmRNA and Lon contribute to fitness at high and low temperatures, but YefM and YoeB do not confer a discernible advantage under these conditions.

Discussion
The results presented here reveal temperature-induced mRNase activity in E. coli cells. Several observations argue that YoeB is responsible for this nuclease activity. First, transcripts must be actively translated in order to be cleaved. This is in accord with evidence that YoeB binds the ribosome and only cleaves translated messages in vivo (Christensen-Dalsgaard and Gerdes 2008;Feng et al. 2013). Second, temperature-induced mRNase activity requires Lon, consistent with the well-established role for this protease in antitoxin degradation (Gerdes et al. 2005;Tsilibaris et al. 2006;Gerdes and Maisonneuve 2012;Goeders and Van Melderen 2014). Third, expression of yoeB at lower temperatures recapitulates heat-induced mRNase activity; and yefM overexpression is sufficient to suppress mRNase activity at elevated temperature. Finally, ΔyefM-yoeB mutants do not exhibit heat-induced mRNase activity. Collectively, these results indicate that growth at elevated temperatures results in Lon-mediated degradation of YefM and concomitant activation of YoeB. These findings reinforce the conclusions of several previous studies showing that environmental stresses activate TA modules. Gerdes and colleagues have shown that RelE, MazF and other E. coli toxins become activated in response to amino acid starvation (Christensen et al. 2001;. Additionally, the E. coli dinJ-yafQ operon contains a LexA box in its promoter, and its transcription is de-repressed in response to DNA damage (Prysak et al. 2009). Thus, our data are broadly consistent with a stress-response function and show that temperature stress is yet another environmental stimulus that activates TA modules. Although YoeB activity becomes manifest at higher temperatures, the mechanism by which YefM is degraded preferentially over other antitoxins is not clear. Van Melderen and colleagues reported that gratuitous overproduction of Lon is sufficient to activate YoeB, but other known E. coli toxins were activated to a much lesser degree (Christensen et al. 2004). Similarly, the data presented here suggest that YoeB is the only toxin activated in response to elevated temperature. Together, these observations indicate that YefM is particularly sensitive to Lon activity. Although Lon is considered a heat-shock protein and lon transcription is upregulated during heat shock (Goff et al. 1984;Phillips et al. 1984), we find that Lon levels are essentially the same in cells grown at different temperatures. This argues that increased Lon concentration cannot account for thermal activation. It is also possible that YefM and YoeB dissociate in response to increased temperature, though it appears that YefM-YoeB is no more thermolabile than other TA complexes (Cherny et al. 2005). Moreover, we find that YoeB is activated at 37°C by inducing the heat-shock regulon through ectopic rpoH expression. This latter observation raises the possibility that an additional heat-shock factor may collaborate with Lon to degrade YefM. Indeed, there are reports of ATP-dependent chaperones working in concert with Lon to specifically degrade misfolded and damaged proteins (Sherman and Goldberg 1992;Huang et al. 2001). However, the ClpB, HtpG, and DnaK chaperones are not required for the thermal activation of YoeB.
In fact, ΔdnaK mutants show constitutive activation of YoeB even at low temperature. This finding is consistent with recent data from the Chien and Laub laboratories showing that Caulobacter crescentus ΔdnaK mutants have increased Lon protease activity (Jonas et al. 2013). They find that unfolded proteins stimulate Lon activity allosterically. Because proteins tend to aggregate and misfold during thermal stress, the findings of Jonas et al. provide an explanation for increased protease activity in the absence of Lon overexpression. Their model may also account for YoeB activation in response to r 32 overproduction. DnaK binds to r 32 (Gamer et al. 1992;Liberek et al. 1992), and therefore overproduced transcription factor could potentially sequester the chaperone, preventing it from refolding other client proteins. Thus, super-physiological levels of r 32 could paradoxically lead to more unfolded proteins and increased Lon activity.  Figure 8. YoeB confers no growth advantage at elevated temperatures. Escherichia coli X90 cells of the indicated genetic backgrounds were adjusted to OD 600 = 1.0, then serial diluted in LB medium and spotted onto LB agar for overnight growth at the indicated temperatures. and Gram-positive bacteria (Nieto et al. 2007;Kumar et al. 2008;Yoshizumi et al. 2009;Sevillano et al. 2012), it will be of interest to determine whether thermal activation is conserved in other species.
Early biochemical studies showed that purified YoeB has intrinsic RNase activity in vitro (Kamada and Hanaoka 2005;Christensen-Dalsgaard and Gerdes 2008); but in vivo, the toxin only cleaves actively translated messages that are bound to the ribosome (Christensen-Dalsgaard and Gerdes 2008;Zhang and Inouye 2009;Feng et al. 2013). YoeB binds to the ribosome A site, where it cleaves the A-site codon to produce a truncated nonstop mRNA (Kamada and Hanaoka 2005;Feng et al. 2013). Although YoeB is generally accepted to be a ribosome-dependent A-site nuclease, there are conflicting reports about which stage of translation is affected. Inouye and colleagues have shown that YoeB acts during translation initiation (Yoshizumi et al. 2009;Zhang and Inouye 2009), whereas Gerdes and colleagues report YoeB-dependent cleavage in termination codons (Winther and Gerdes 2009). Both studies monitored the same endogenous lpp and ompA transcripts, but the experimental approaches differed. Zhang & Inouye overproduced YoeB from a plasmid vector, whereas the Gerdes group detected endogenous YoeB activity in response to ectopic VapC toxin overexpression (Winther and Gerdes 2009;Zhang and Inouye 2009). Our data show temperature-induced cleavage at two sites, one in the stop codon and another at an unidentified upstream position. Both cleavages are not observed when codon Glu-28 is mutated to an amber stop, indicating that the upstream cleavage site is well downstream of the start codon. Thus, our results are closer to those of Winther et al. and we conclude that endogenous YoeB activated at elevated temperature does not affect translation initiation.
The stress-response model of TA function postulates that activated toxins facilitate changes in gene expression in response to environmental stress Gerdes et al. 2005). The classic example is RelE activation in response to starvation. In this model, amino acid starvation slows protein synthesis, leading to depletion of RelB antitoxin due to its intrinsic instability. RelE then acts as an A-site nuclease to cleave translated messages and inhibit translation. The resulting stalled ribosomes are recycled by tmRNA-SmpB, which provides a burst of amino acids through proteolysis of ssrA-tagged nascent chains . The starved cells then alter transcription to express genes for amino acid biosynthesis and a new steady-state is eventually achieved. In principle, YoeB could function in the same manner to accelerate mRNA decay so that new heat-shock transcripts are translated more rapidly. However, we were unable to detect YoeB-dependent cleavages within endogenous messages. YoeB must compete with translation factors for access to the ribosome A site, and therefore it is more likely to cleave mRNA in the context of arrested ribosomes. Because heat shock does not inhibit global translation, YoeB activity is presumably limited to a small population of ribosomes that encounter difficulties during protein synthesis. In contrast, amino acid starvation stress not only activates RelE toxin, but also blocks protein synthesis to produce stalled ribosome complexes for RelE-mediated mRNA cleavage. Another discrepancy between our results and the stress-response model is the absence of growth arrest upon YoeB activation. In fact, E. coli cells show no growth lag in response to temperature up-shift, but rather accelerate their growth rate immediately. Moreover, YoeB activity is detectable over several hours of growth at elevated temperature. These data demonstrate that toxins can be activated at low levels, suggesting that they may actually promote more efficient protein synthesis under stress conditions. Heat shock and temperature stress has been shown to damage translation complexes and disrupt protein synthesis (Korber et al. 2000;Jiang et al. 2009). In our model, YoeB activity would perform a quality control function to help recycle translation complexes that stall stochastically during thermal stress. This is still adaptation to environmental stress, yet the proposed function is distinct from previously established roles for toxins in growth arrest and persistence (Maisonneuve et al. 2011;Gerdes and Maisonneuve 2012;Maisonneuve and Gerdes 2014).