|Home | About | Journals | Submit | Contact Us | Français|
Although it is well-recognized that bacteria respond to environmental stress via global networks, the mechanism by which stress is relayed to the interior of the cell is poorly understood. Here we show that enigmatic toxin/antitoxin systems play a vital role in mediating the environmental stress response. Specifically, the antitoxin MqsA represses rpoS, which encodes the master regulator of stress. Repression of rpoS by MqsA reduces the concentration of the internal messenger 3,5-cyclic diguanylic acid, leading to increased motility and decreased biofilm formation. Furthermore, the repression of rpoS by MqsA decreases oxidative stress resistance via catalase activity. Upon oxidative stress, MqsA is rapidly degraded by Lon protease resulting in induction of rpoS. Hence, we show that external stress alters gene regulation controlled by toxin/antitoxin systems, such that the degradation of antitoxins during stress leads to a switch from the planktonic state (high motility) to the biofilm state (low motility).
Although toxin/antitoxin (TA) systems are ubiquitous in bacterial chromosomes1, their role in cell physiology is controversial. It has been argued that they have no impact on the cell2 or that they may have as many as nine roles3 including addictive genomic debris, stabilization of genomic parasites, selfish alleles, gene regulation, growth control, persister cell formation (persisters are bacteria that are resistant to antibiotics without genetic change4), programmed cell arrest, programmed cell death, and anti-phage measures5. They have also been linked to biofilm formation6. A TA system typically consists of two genes located in an operon which encode a stable toxin that disrupts an essential cellular process and a labile antitoxin that can bind and form a tight complex with the toxin and neutralize its activity1. Although TA systems were first thought to be related to cell death, there have been few clear examples of toxin-mediated cell death in a physiologically relevant situation3; hence, the primary role of these systems has been enigmatic.
Many chromosomal TA systems have been characterized in Escherichia coli (16 so far7) including (listed as toxin/antitoxin) MqsR/MqsA8, MazF/MazE9, RelE/RelB10, ChpB/ChpS11, YoeB/YefM12, YafQ/DinJ13, and YhaV/PrlF14. Although the mechanism of toxicity at the molecular level is slightly different, MqsR8, MazF1, RelE1, ChpB1, YoeB12, and YhaV14 prevent translation by cleaving RNAs; the mode of translation inhibition by YafQ is unclear2.
Of these redundant TA systems, toxin MqsR (motility quorum sensing regulator) (YgiU/B3022)15, 16 and antitoxin MqsA (YgiT/B3021)8 are particularly significant as the genes that encode them are the first locus that upon deletion, decreases the formation of persister cells17, and mqsR is also the most highly induced gene in persister cells as compared to non-persisters4. MqsR/MqsA is also the first TA system found to be induced in biofilms16, the first to be related to quorum sensing15, the first to be related to cell motility15, and the first to be related to biofilm formation15, 16. Furthermore, MqsA is the first antitoxin shown to regulate more than its own transcription as it binds the mqsRA, cspD, mcbR, and spy promoters8, 18. The three dimensional structure of MqsR/MqsA8 revealed that MqsR is an RNase similar to RelE and YoeB and that MqsA binds DNA via its helix-turn-helix (HTH) motif in the C-terminal domain and binds the toxin via its N-terminal zinc-binding domain. MqsR cleaves mRNA at GCU sites7. MqsR/MqsA is also conserved in 40 eubacteria15.
Since the TA pair MqsR/MqsA has been linked to both motility and biofilm formation15, it appears intimately related to how E. coli switches between motile and sessile (i.e., biofilm) growth. The switch between these two fundamental lifestyles is based on the antagonistic regulation of the master regulator of motility, FlhDC, and the master regulator of the stress response, RpoS19, which controls up to 500 genes in E. coli20. Underlying both lifestyles is regulation of the concentration of the second messenger 3,5-cyclic diguanylic acid (c-di-GMP), which controls the switch from motility (low c-di-GMP) to sessility (high c-di-GMP)19 through a variety of actions, such as binding YcgR and acting as a molecular flagellar brake21. c-di-GMP concentrations are modulated directly via de novo synthesis by diguanylate cyclases (proteins with GGDEF motifs) and via degradation by phosphodiesterases (proteins with EAL or HD-GYP motifs)22. Herein we show how extracellular stress is conveyed to FlhDC and RpoS which was previously not understood19.
Using a strain deficient in six major TA systems, Δ6 (MazF/MazE, RelE/RelB, ChpB, YoeB/YefM, YafQ/DinJ, and MqsR/MqsA), we provide insights into extracellular stress and both the general stress response and the switch from planktonic growth to biofilm formation. We show that the antitoxin MqsA regulates the RNA polymerase sigma factor σS, which is encoded by rpoS. Specifically, during stress, Lon protease degrades MqsA which leads to induction of rpoS, which in turn increases c-di-GMP, inhibits motility, and increases cell adhesion/biofilm formation.
Our previous whole-transcriptome analysis, in which the impact of the MqsR toxin was investigated, showed that rpoS was induced significantly by the RNase activity of MqsR18. To explore further the relationship between the MqsR/MqsA TA system and the regulation of rpoS under stress conditions, we cultured cells under oxidative stress conditions in which RpoS is crucial for cell survival23, 24 by regulating antioxidant activities such as those of catalase and superoxide dismutase25. We used a genetic background devoid of the major E. coli TA pairs via the Δ5 strain2, which lacks the MazF/MazE, RelE/RelB, ChpB, YoeB/YefM, and YafQ/DinJ TA systems (Supplementary Results, Supplementary Table 1) and the Δ6 strain which also lacks MqsR/MqsA (Δ5 mqsRA). In this way, the impact of antitoxin MqsA could be studied largely in the absence of other antitoxins.
Our hypothesis is that the antitoxin MqsA plays an essential role in mediating the RpoS stress response. To test this hypothesis, we analyzed rpoS transcripts during oxidative stress to observe the effect of MqsA. Under these oxidative stress conditions (20 mM H2O2 for 10 min), due to the complexity of the regulation of rpoS transcription and post-transcriptional modifications of rpoS mRNA upon stress20, a consistent increase (~2-fold) in rpoS mRNA in wild-type cells was detected by qRT-PCR (see Supplementary Table 2 for all of the qRT-PCR data). When the Δ6 cells were exposed to this oxidative stress in the presence of plasmid-expressed MqsA, rpoS mRNA was reduced by 4 ± 1 fold (via qRT-PCR) compared to the empty plasmid control with oxidative stress. Corroborating this result, deleting mqsRA resulted in a 4.5 ± 0.4-fold increase in rpoS mRNA after sec with 20 mM H2O2 (Δ6 vs. the MG1655 wild-type strain); similar results were seen upon deleting mqsRA in the related E. coli strain BW25113. Hence, MqsA directly or indirectly controls rpoS transcription. It was not possible to test directly the impact of deleting the antitoxin gene mqsA on rpoS transcription since deleting mqsA is lethal4, 26 due to the toxicity of MqsR; similar results have been seen with other antitoxins including MazE, ChpS and YefM26. The production of the antitoxin MqsA is not toxic8.
Since rpoS is repressed by MqsA, the genes controlled by RpoS should also be repressed. We found that adrA, which encodes a diguanylate cyclase that controls cellulose synthesis and which is positively regulated by RpoS27, was repressed by 6 ± 1 fold upon producing MqsA from a plasmid in cells under oxidative stress when directly compared to the identical experiment with cells with an empty plasmid and oxidative stress. Similarly, repression of rpoS by MqsA should further decrease c-di-GMP levels by inhibiting three other genes that encode diguanylate cyclases (ydaM, yegE, and yedQ), all of which are up-regulated by RpoS19. As expected, ydaM (3 ± 1 fold), yegE (3 ± 1 fold), and yedQ (2 ± 1 fold) were repressed by MqsA with oxidative stress compared to the empty plasmid with oxidative stress. In addition, csgD, which encodes the regulator for curli and cellulose, should also be repressed by repressing rpoS with MqsA19. As expected, csgD was repressed (3 ± 1 fold) when MqsA was produced from a plasmid in the presence of oxidative stress. Note that bioinformatics analysis showed that the csgD promoter also contains an mqsRA-like palindrome (5’-AACCT TA AGGTT) 78 bp upstream of the transcription initiation site that is used by MqsA to regulate transcription7, 28(Supplementary Fig. 1 and Supplementary Table 3). Therefore, MqsA potentially reduces curli/cellulose production through two pathways: indirectly via rpoS repression and directly by binding the csgD promoter.
In addition, since RpoS is a positive regulator of catalase activity via katG and katE25, repression of rpoS by MqsA should lead to a reduction in transcription of these genes. As expected, transcription of katE (6 ± 1 fold) and katG (4 ± 1 fold) were both repressed when the Δ6 cells were exposed to oxidative stress (20 mM H2O2 for 10 min) in the presence of elevated MqsA compared to oxidative stress with an empty plasmid. Corroborating this result, deleting mqsRA resulted in a 3.5 ± 0.4-fold increase in katE mRNA after 30 sec with 20 mM H2O2 (Δ6 vs. the MG1655 wild-type strain); similar results were seen for the deleting mqsRA in BW25113. Therefore, MqsA represses rpoS either directly or indirectly resulting in repression of the genes positively regulated by RpoS.
There are two palindromes in the promoter of the mqsRA operon (5’-TAACCT TTT AGGTTA and 5’-ACCT TTT AGGT), and MqsA as well as the MqsR/MqsA complex bind to the two palindromic sequences to negatively regulate mqsRA7, 28. Each MqsA of the dimer binds to this palindrome via residues Asn97 and Arg101, which make base-specific interactions with eight nucleotides including four from one strand (5’-TAAC) and four from the anti-parallel strand (5’-AGGT)28. In addition, the intervening TTT does not seem to have any effect on MqsA binding affinity28. As with csgD, we identified a similar mqsRA-like palindrome in the promoter region of rpoS (Supplementary Fig. 2), 5’-ACCT TGC AGGT, and thus by direct extension of our work with mqsRA, we hypothesized that MqsA binds directly to the rpoS promoter (PrpoS) and thus controls the transcription of the sigma factor that is necessary for control of the general stress response. To demonstrate direct binding of MqsA to the regulatory sequences of rpoS, electrophoretic mobility shift assays (EMSA) were used to detect the binding of MqsA to the 25 bp fragment of PrpoS that includes this palindrome (from −161 to −137, Supplementary Fig. 2). The EMSA showed that MqsA binds and shifts PrpoS with 10-, 25-, and 50-fold molar excesses of MqsA (Fig. 1a). However, upon mutating the palindrome in PrpoS, 5’-ACCT TGC AGGT, to 5’-ACCT TGC TCAC in PrpoS-M, binding of MqsA to PrpoS was abolished (Fig. 1b). As expected, at the same concentrations, no shifts were observed for the promoter of the negative control gadA (PgadA) (Supplementary Fig. 3a). Furthermore, replacing the key residues of the base-specific interactions of MqsA (N97A and R101A) prevents MqsA from binding the rpoS promoter (Supplementary Fig. 3b). Therefore, MqsA binds directly and specifically to the mqsRA-like palindrome sequence of the rpoS promoter.
To confirm that MqsA directly influences transcription of rpoS, a chromosomal lacZ transcriptional fusion was constructed. A 943 bp region upstream of the rpoS translation initiation site, including the 565 bp mRNA leader sequence and 378 bp upstream of the leader sequence (Supplementary Fig. 2), was fused to translationally-independent lacZ and inserted into the suicide vector pKNOCK-Tc29. The constructed vector was conjugated into Δ6 ΔlacZ (Supplementary Table 1) generating an integrated lacZ reporter driven by the rpoS promoter and an intact copy of rpoS with its native promoter (Fig. 1c); this reporter strain (Δ6 ΔlacZ PrpoS::lacZ), named Δ6 R1 PrpoS, was used to probe the influence of MqsA on rpoS transcription. As expected, when compared to an empty plasmid, there was a 7.5 ± 0.5-fold repression of β-galactosidase activity with mqsA when induced from pCA24N-mqsA with 0.1 mM isopropyl-thio-2-D-galactopyranoside (IPTG) for 2 h, and a 5.4 ± 0.6-fold repression for 4 h (Fig. 1d). Corroborating this result, deleting mqsRA resulted in a 4.2 ± 0.7-fold increase in rpoS transcription in the stationary phase as determined by comparing β-galactosidase activity in reporter strains Δ6 R1 PrpoS vs. Δ5 R1 PrpoS (Supplementary Table 1).
To confirm that MqsA regulates rpoS expression through the identified mqsRA-like palindrome in rpoS, the four nucleotides in the rpoS palindrome that are used for direct base-specific interactions28 were mutated from 5’-AGGT to 5’-TCAC (Supplementary Fig. 2) in pKNOCK-PrpoS-M, and integrated into Δ6 ΔlacZ to form Δ6 R2 PrpoS-M (Fig. 1c). Without the palindrome, MqsA no longer repressed lacZ transcription since similar β-galactosidase activities were found for cells with or without production of MqsA (Fig. 1d). Moreover, the control strain, reporter Δ6 R3 PrpoS, with wild-type PrpoS fused to lacZ but with the mutated palindrome in front of the rpoS gene (Fig. 1c), had 3.0 ± 0.5 (2 h) and 5.8 ± 0.3-fold (4 h) reduced β-galactosidase activity upon production of MqsA (Fig. 1d). Therefore, MqsA directly represses rpoS transcription, and the repression is dependent on the palindrome in the promoter of rpoS.
Since MqsA represses rpoS which, in turn, results in the repression of the genes related to synthesizing c-di-GMP, it is expected that c-di-GMP concentrations should decrease in the presence of MqsA. As expected, upon overexpressing mqsA from a plasmid in stationary-phase cells, the intracellular c-di-GMP concentration decreased by 1.7 ± 0.4 fold (Fig. 2a). Corroborating this result, the deletion of mqsRA increased c-di-GMP concentration by 3.3 ± 0.6 fold (Δ6/pCA24N vs. Δ5/pCA24N) (Fig. 2a). Hence, MqsA reduces c-di-GMP concentrations.
Since MqsA represses rpoS, we reasoned that the deletion of the six toxin/antitoxin systems should increase cell survival under stress as RpoS would be constitutively induced and the cells would be preconditioned for a stress response. As expected, Δ6 had over 10-fold greater resistance to both oxidative stress (20 mM H2O2 for 10 min) (Fig. 2b) and acid stress (pH 2.5 for 10 min) (Fig. 2c) compared to MG1655 (parent strain). Corroborating this result, overexpression of MqsA reduced cell survivability under oxidative stress by 850 fold compared to Δ6 with the empty plasmid (Fig. 2b) and reduced cell survivability under acid stress by 22 fold (Fig. 2c). This decrease in the resistance to oxidative stress is explained by a 19 ± 2-fold reduction in catalase activity when MqsA is produced in the Δ6 strain vs. the empty plasmid (only trace catalase activity was seen in the MqsA-producing strain). Similarly, there was a 1.8 ± 0.3-fold decrease in catalase activity for Δ5 vs. Δ6. Catalase converts H2O2 to H2O and O2; hence, the reduced ability of the cells to decompose H2O2 when MqsA represses rpoS was demonstrated by a dramatic reduction in oxygen bubbles upon addition of H2O2 to Δ6 (Fig. 2d). Moreover, overproduction of MqsA in the strain that only lacks mqsRA also had greatly reduced oxygen bubbles upon addition of H2O2 (Fig. 2d). Together, these nine sets of results (including the qRT-PCR results with katE and katG) convincingly show that the MqsRA TA system and specifically the antitoxin MqsA regulate resistance to H2O2. Therefore, the MqsRA TA system directly affects the ability of the cell to withstand external stress.
Since c-di-GMP levels are reduced in the presence of antitoxin MqsA, motility should increase due to lower RpoS levels which no longer inhibit expression of flhD, the master regulator of motility19. As expected, production of MqsA in Δ6 increased motility by 3.2 ± 0.3 fold (Fig. 3a). Similarly, it was expected that motility should decrease in the presence of MqsR since elevated toxin levels should increase stress levels in the cell leading to elevated RpoS and repression of flhD. As expected, production of MqsR abolished swimming motility (Fig. 3a) since the viable cells were non-motile (with 0.1 mM IPTG, production of MqsR was not completely toxic as the cell number increased from 2.1 × 104 to 8.5 × 106). In addition, production of MqsA in BW25113 increased cell motility by 4 ± 1 fold (Fig. 3b); therefore, the increase in motility via MqsA is not an artifact of the Δ6 strain.
The increase in motility by MqsA was also tested in an rpoS deletion background, and as expected, MqsA only increased motility slightly (1.2 ± 0.2 fold). This result confirms that MqsA works primarily through rpoS. Critically, the increase in motility upon producing MqsA was also tested in the lacZ reporter strains where it was found that MqsA no longer increases motility in reporter Δ6 R3 PrpoS (Fig. 3c) in which the mutated palindrome lies in the promoter of the chromosomal rpoS gene (Fig. 1c). The other two reporter strains, Δ6 R1 PrpoS and Δ6 R2 PrpoS-M, which both have the wild-type PrpoS for rpoS, are repressed by MqsA and became more motile (Fig. 3c). These results demonstrate that MqsA regulates rpoS transcription and further confirm that MqsA binds the mqsRA-like palindrome in the promoter of rpoS.
Moreover, when RpoS is produced from the expression vector pCA24N-rpoS such that rpoS lacks its native promoter and leader region, production of MqsA no longer increases cell motility (Fig. 3d). This result indicates that MqsA exerts its control directly on rpoS expression and requires its native upstream sequence. This result also eliminates the possibility that RpoS is regulated by MqsA at a posttranslational level since MqsA no longer increases motility once rpoS mRNA is translated.
Additional phenotypes related to the reduced c-di-GMP levels should also be influenced by MqsA and MqsR. Our hypothesis predicts that if MqsA represses rpoS, both curli and cellulose production should likewise be repressed since RpoS induces csgD19. Using Congo red, a dye that binds to both cellulose and curli30, we found, as expected, that producing MqsA in the presence of oxidative stress (2 mM H2O2 for 180 min at 30°C) decreased curli/cellulose production by 13 ± 2 fold (Fig. 4a). Furthermore, producing MqsR increased curli/cellulose production by 1.4 ± 0.3 fold compared to the empty plasmid control and by 20 ± 3 fold compared to production of MqsA (Fig. 4a). Moreover, deleting mqsR and mqsA lead to a 2.8 ± 0.3-fold increase in curli/cellulose (Δ6 vs. Δ5) under the same oxidative stress conditions. In contrast, without H2O2, there was only a small reduction in curli/cellulose production when MqsA was produced (2.1 ± 0.6 fold), and no change in this phenotype for MqsR production.
Since MqsA represses rpoS, which leads to reduced c-di-GMP levels and increased motility, deletion of mqsA should also increase biofilm formation due to the increase in c-di-GMP. As expected, the Δ6 strain with mqsRA deleted had 1.6 ± 0.2 fold more biofilm formation than the Δ5 strain after 24 h in minimal medium (Fig. 4b). Corroborating this result, production of MqsA in Δ6 (which reduces c-di-GMP) decreased biofilm formation 2.0 ± 0.4 fold after 24 h; this result also shows that producing MqsA in Δ6 reduces biofilm formation to the level of Δ5 so that the increase in biofilm formation seen with Δ6 was complemented (Fig. 4b). Taken together, these results show that MqsA decreases biofilm formation due to its reduction in c-di-GMP.
To show that oxidative stress regulates proteolysis of MqsA, we performed a Western blot analysis to monitor MqsA levels upon addition of oxidative stress. Rifampin was added to the cultures to stall RNA polymerase to block transcription of the mqsA mRNA31. As expected, MqsA was degraded rapidly upon adding H2O2 (20 mM) (Fig. 5a). Within 10 min, the amount of MqsA was greatly reduced and barely detectable (Fig. 5a), thus the half-time of MqsA under oxidative stress is approximately 1.25 min. However, MqsA was stable without oxidative stress up to 60 min (30 min time point shown in Fig. 5b).
Consistent with these MqsA degradation results, after prolonged (10 min) contact with 20 mM H2O2, there was no difference in rpoS and katE mRNA levels in contrast to an induction with short contact (30 sec) for Δ6 vs. MG1655 since MqsA is rapidly degraded under these conditions and is no longer available to repress rpoS; similar results were seen for the deleting mqsRA in BW25113. The PrpoS::lacZ reporter strains also exhibited this behavior with higher initial promoter activity in Δ6 R1 PrpoS compared to Δ5 R1 PrpoS with H2O2 but this difference gradually disappeared (Fig. 5c) due to degradation of MqsA with longer H2O2 contact time. Western blotting also showed threefold higher RpoS levels upon deleting mqsRA for short H2O2 contact time (0 to 0.5 min) but after long incubations (e.g., 10 min), there was no difference between strains with and without MqsA due to degradation of MqsA (Fig. 5d). The relatively small changes in RpoS protein (Fig. 5d) are expected due to the translational regulation of RpoS synthesis via OxyR-induced OxyS RNA under oxidative stress32, which may offset somewhat increased rpoS transcription. Therefore, both PrpoS activity and RpoS levels are repressed by MqsA but upon oxidative stress, MqsA is degraded and rpoS is derepressed.
We previously determined that the three genes encoding the proteases Lon, ClpX and ClpP are all induced upon oxidative stress (8 ± 1 fold, 6 ± 1 fold and 6 ± 1 fold, respectively)18. To further explore which protease degrades MqsA, the effect of Lon, ClpP and ClpX on MqsA in a motility assay was examined. Producing Lon and MqsA simultaneously significantly reduced the ability of MqsA to increase motility (Fig. 5e). As a control, producing Lon did not affect motility (Fig. 5e). However, in the presence of ClpP, MqsA increased motility. Furthermore, producing ClpX abolished motility completely making it impossible to observe the effect of MqsA. Therefore, the protease Lon degrades MqsA, as shown for antitoxin CcdA33 and antitoxin RelE34, under oxidative stress conditions where lon is induced.
Previously, it was reported that five TA systems had no influence on the general stress response2; however, the most important TA system, MqsR/MqsA, was not considered in this analysis nor was biofilm formation considered. Subsequently, we have shown that TA systems are important for biofilm formation using MqsR/MqsA15, 16 and the E. coli Δ5 strain6, which lacks five TA systems. In addition, two other groups have confirmed our initial result of the importance of TA systems in biofilm formation35, 36. Clearly, the role for TA systems is expanding rapidly. In addition, we have found stress increases E. coli biofilm formation (e.g., oxidative, acid, low-temperature, and heavy metal stress)37 and this phenomenon appears to hold for other bacteria as well. In the current study, we link TA systems to both biofilm formation and the general stress response by showing that the antitoxin MqsA represses rpoS which, in turn, alters the level of the secondary messenger c-di-GMP thereby regulating motility and cell adhesins. Hence, for the first time, we link TA systems to the secondary messenger c-di-GMP and to RpoS. This constitutes a novel role for antitoxins: direct regulation of the general stress response.
RpoS is important for biofilm formation since its inactivation prevents mature biofilm formation by inducing motility-related genes and by repressing colanic acid synthesis (a biofilm exopolysaccharide)38. Many RpoS-regulated genes are also expressed in a temporal manner in biofilms39. Furthermore, RpoS enhances resistance to antibiotics40. Both of these earlier results can now be explained by the direct interaction we have discovered between MqsA and rpoS: increasing MqsA represses rpoS which reduces the ability of the cells to withstand stresses like antibiotics, oxidative stress, and acid stress. In addition, the repression of rpoS by MqsA also decreases c-di-GMP and increases motility and reduces biofilm formation. Moreover, for mature biofilms, rpoS expression was greater for cells in the outer regions of the biofilm where mqsA was repressed41; this provides another line of evidence that MqsA and rpoS are related. Our work also serves to discern the mechanism by which the MqsR/MqsA TA pair was first related to motility and fhlD15; i.e., MqsR/MqsA are related indirectly to motility via RpoS and c-di-GMP rather than via the direct control of flhD transcription by MqsA.
As the production of proteins from plasmids should be interpreted with caution, we corroborated many of the results obtained from expressing MqsA from a plasmid by also observing the effects of deleting mqsRA. c-di-GMP levels, response to oxidative stress and acid stress, curli/cellulose production, biofilm formation, rpoS mRNA levels, and katE mRNA levels were all shown to be altered by deleting mqsRA. In addition, we utilized a strain which has a single copy of lacZ transcribed from the rpoS promoter in the chromosome as a reporter to investigate the regulatory effects of MqsA on rpoS transcription (while leaving the wild-type rpoS system intact) and found that MqsA binds directly to the rpoS promoter region and represses rpoS transcription (Fig. 1). When the rpoS palindrome was mutated to disrupt the sequence-specific DNA binding, MqsA binding to the promoter is abolished (Fig. 1b) and no longer represses rpoS transcription (Fig. 1d). The dependence of MqsA on this palindrome was also demonstrated by the inability of MqsA to increase motility when the palindrome in the native rpoS promoter was mutated (Fig. 3c). Since the palindrome lies upstream (144 bp) of the major transcription start site of rpoS, further investigation is needed to elucidate whether the repression of MqsA involves cooperative binding (two smaller mqsRA-type palindromes are also present in PrpoS).
A schematic of our understanding of how the antitoxin MqsA plays a key role in mediating the general stress response and biofilm formation is shown in Figure 6. Note regulation of RpoS is complex and includes regulation at the level of transcription, translation, and protein stability/activity20; hence, Figure 6 is a simplification and shows primarily the features important for MqsA. Our results show clearly that MqsA is degraded in the presence of oxidative stress (Fig. 5a,b); this degradation leads to derepression of rpoS transcription (Fig. 5c) and an increase in rpoS mRNA, and that production of Lon prevents MqsA from increasing motility (Fig. 5e); hence, the protease Lon degrades MqsA upon stress. These results are consistent with our previous results indicating MqsR is not toxic in a lon deletion background18. Hence, upon oxidative stress, Lon is induced and degrades MqsA like other antitoxins, rpoS is derepressed, and the cell directs transcription toward stress-related genes which includes increasing c-di-GMP concentrations and catalase activity. It appears that MqsA may also regulate directly other stress-related genes such as csgD, rnc, and sspA as identified by the palindrome search (Supplementary Table 3). In addition, upon oxidative stress, transcriptional dual regulator OxyR induces katG42 and the regulatory RNA OxyS32 which inhibits translation of the rpoS message43. Thus, RpoS must be tightly controlled since cells that are stress-resistant grow more slowly than those not utilizing RpoS because resources are directed away from σ70-related genes that are used for growth44. RpoS controls over 70 genes important for resistance to oxidative stress, UV-radiation, heat shock, hyper-osmolarity, low pH, and ethanol44, so the impact of TA systems may be involved in diverse stresses.
Although speculative, our model has some other implications for cell physiology. Since the antitoxin MqsA is degraded rapidly upon stress (Fig. 5) and mqsRA is induced upon stress18, it is expected that the RNase activity of MqsR would be increased during stress; this enhanced MqsR activity may serve to rapidly direct the cell toward the translation of newly-transcribed, stress-related transcripts by degrading the older σ70-related mRNA. The duration of MqsR RNase activity, from a burst to sustained RNase activity, would dictate whether the cell responds to the stress by rapidly producing new proteins necessary to withstand stress while forming a biofilm (higher RpoS activity during stress would elevate c-di-GMP levels and increase biofilm formation) or whether the cell becomes dormant (i.e., a persister cell) in a biofilm and avoids the stress by not metabolizing. Hence, our model suggests that persister cell formation may be viewed in part as an extreme example of the general stress response mediated by the MqsR/MqsA TA system. In support of this model, dormant cells have been generated by inducing the general stress response after DNA damage45, and persister cell formation for fluoroquinolone antibiotics is dependent on the general stress response46. In addition, mqsR is induced in biofilms16, is the most induced gene in persister cells4, and the number of persister cells decreases upon deletion of mqsR17.
Our current results indicate that the formerly ambiguous TA systems, especially MqsR/MqsA, are key regulators of gene activity by controlling in part the induction of the stress response, which, in turn, leads to enhanced biofilm formation and reduced motility. Therefore, TA systems have a broad and an important impact on cell physiology by influencing such developmental cascades as the switch from the planktonic to biofilm cells.
The bacterial strains and plasmids are listed in Supplementary Table 1. Luria-Bertani (LB) at 37°C was used except where indicated.
P1 transduction was used to transfer the ΔmqsRA KanR mutation from BW25113 mqsRA17 to Δ52 to form Δ6. Removal of mqsRA relBE, dinJ, yefM, mazEF, and chpB in Δ6 was verified by PCR using primers shown in Supplementary Table 4. P1 transduction was also used to remove lacZYA from Δ5 and from Δ6 using BW25993 DE(lacZYA)51447.
943 bp upstream of the rpoS start codon was amplified using primers PrpoS-XhoI and PrpoS-BamHI. To form pKNOCK-PrpoS (Supplementary Fig. 4), the PCR product was cloned into the suicide vector pKNOCK-Tc29 at the XhoI and BamHI sites upstream of a lacZ cassette that includes the native lacZ translational initiation region cloned into the NotI site. The PrpoS::lacZ reporter was integrated as a single copy into the chromosome of Δ5 ΔlacZ and Δ6 ΔlacZ leaving an intact copy of rpoS with its native promoter. Δ5 R1 PrpoS and Δ6 R1 PrpoS (Supplementary Fig. 4), were verified by PCR (Fig. 1c, nlpD-F/pKNOCK-R and pKNOCK-F/rpoS-R) and by DNA sequencing.
The mqsRA-like palindrome in the PrpoS::lacZ fusion was mutated using PCR with template pKNOCK-PrpoS and primers PrpoS-M-F and PrpoS-M-R to form pKNOCK-PrpoS-M (Supplementary Fig. 4). Conjugating this vector into Δ6 ΔlacZ led to two strains: Δ6 R2 PrpoS-M with mutated palindrome upstream of lacZ and the native promoter rpoS unchanged and Δ6 R3 PrpoS with mutated palindrome upstream of rpoS and the wild-type palindrome upstream of lacZ gene (Fig. 1c). Correct integration was verified by PCR (Fig. 1c, Set 1: nlpD-F and pKNOCK-R, and Set 2: pKNOCK-F and rpoS-R) followed by DNA sequencing at both the lacZ and rpoS promoter regions.
β-galactosidase activity48 was determined with strains grown in low-salt LB medium (LB with 0.5 g/l NaCl) to a turbidity at 600 nm of ~1.0, then IPTG (0.1 mM) was added to induce production of MqsA via pCA24N-mqsA. For experiments with 20 mM H2O2, strains were grown to turbidity ~3.0.
Overnight cultures were diluted to a turbidity of 0.05 and grown to a turbidity of 0.5, then 1 mM IPTG was used to induce mqsA for 2 h. Cells were centrifuged and resuspended in LB to a turbidity of 1.0 and exposed to either 20 mM H2O2 for 10 min or pH 2.5 for 10 min.
c-di-GMP was quantified using HPLC48. Strains were grown for 2.5 h, then 0.5 mM IPTG was added to induce mqsA for 15 h. c-di-GMP (BIOLOG Life Science Institute) was used as a standard and to verify the c-di-GMP peak via spiking.
Cell motility was examined on motility agar plates (1% tryptone, 0.25% NaCl, and 0.3% agar). Curli/cellulose production was quantified by the Congo-red binding assay performed at 30°C30 in the presence of oxidative stress (2 mM H2O2 for 180 min). IPTG (0.1 mM) was added in both assays to induce mqsA and mqsR via the pCA24N-based plasmids. Catalase activity was quantified by a colorimetric assay49 using dicarboxidine/lactoperoxidase to detect the remaining H2O2.
Biofilm formation was assayed without shaking using 0.1% crystal violet staining16 in 96-well polystyrene plates after 24 h using M9 glucose (0.2%) medium with 1 mM of IPTG to induce mqsA (initial turbidity of 0.05).
After isolating RNA17 using RNAlater™ (Ambion), 50 ng of total RNA was used for qRT-PCR using the Power SYBR® Green RNA-to-CT™ 1-Step Kit and the StepOne™ Real-Time PCR System (Applied Biosystems). Primers were annealed at 60°C, and rrsG was used to normalize the data. To investigate the regulation of promoters by MqsA under oxidative stress conditions, overnight cultures of Δ6/pCA24N-mqsA and Δ6/pCA24N were re-grown to a turbidity of 0.5, then 1 mM IPTG was added for 2 h to induce mqsA, and after diluting to turbidity ~1.0, the cells were exposed to 20 mM H2O2 for 10 min. To investigate the rpoS and katE mRNA changes, overnight cultures were inoculated into LB low salt medium with an initial turbidity of 0.2 and grown until a turbidity ~3.0. After diluting to a turbidity ~ 1.0, cells were exposed to 20 mM H2O2 for 30 sec and 10 min.
Complimentary oligos (25-mers) labeled with biotin at the 3’ end corresponding to the wild type mqsRA-like palindrome and the corresponding mutated palindrome of the rpoS promoter (Supplementary Table 4) were purchased, solubilized, and annealed as previously described28. For EMSA binding reactions, biotin-PrpoS was incubated with purified MqsA8 either with or without unlabeled PrpoS DNA for 20 min at room temperature. Samples were run on a 6% DNA retardation gel (Invitrogen) at 100V in 0.5× TBE for 75 min. DNA was then transferred to a nylon membrane at 390 mA for 45 min then UV cross-linked at 302 nm. Chemiluminescence was performed with the LightShift Chemiluminescent EMSA Kit (Thermo Scientific) and samples were detected with a CCD imager (Typhoon 9410 Imager).
For MqsA, strains were grown to a turbidity of 0.1, then 0.5 mM ITPG was added to induce mqsA. When the turbidity reached 1, 200 µg/ml rifampin was added to inhibit transcription, and 20 mM H2O2 was added. After various times, samples were processed with 1 mM phenylmethylsulfonyl fluoride and protease inhibitor cocktail (Sigma-Aldrich), and the Western blot was performed with primary antibodies raised against a His tag (Cell Signaling Technology) and horseradish peroxidase-conjugated goat anti-mouse secondary antibodies (Millipore). For RpoS, strains were grown until a turbidity ~3.0, were exposed to 20 mM H2O2, and anti-RpoS monoclonal antibody (Neoclone) was used.
The motifs 5’-ACCT (N)2–4 AGGT were identified using the Biostrings and BSgenome libraries in the R statistical package (version 2.9.2) and Fuzznuc (EMBOSS)50.
Data are presented as means ± s.e. of three or more independent cultures. Statistical significance was assessed using two-tailed unpaired Student’s t-test.
This work was supported by grants from the NIH to T.W. and W.P. and by an NSF-CAREER award to R.P. We are grateful for the Keio and ASKA strains provided by the Genome Analysis Project in Japan. T.W. is the T. Michael O’Connor Endowed Professor at Texas A & M University.
Author Contributions. T.W. conceived the project, T.W., X.W., R.P., W.P, and M.B. designed the experiments. X.W., S.H., Q.M., B.B., M.P., and Y.K. performed the experiments. M.B. and A.T. performed the bioinformatics. All authors discussed the results and commented on the manuscript.
Competing financial interests. The authors declare no competing financial interest.