The bacterial strains and plasmids used in this study are listed in Table 1, and the sequences of the primers are listed in Supplementary Table S1. The E. coli and P. putida KT2440 strains were grown in Luria Bertani broth (LB) medium at 30°C except where indicated. Gentamicin (15 μg/ml) was used for pre-culturing isogenic knockout mutants and for maintaining pEX18Ap-based plasmids, carbenicillin (100 μg/ml) was used to maintain plasmid pMQ70, and kanamycin (50 μg/ml) was used to maintain plasmids pET28b, pHGE, and pHGR01.

To create the in-frame deletion of mqsR, mqsA, and mqsRA in P. putida KT2440, 5′ and 3′ homologous arm segments of 0.9–1.5 kb were generated by PCR from genomic DNA, and each amplicon was then ligated with Gm^R-encoding cassette (from pPS856) into pEX18Ap to produce the replacement plasmids as described previously (Hoang et al., 1998). For example, to delete mqsA, 1334 bp upstream and 962 bp downstream fragments of mqsA ORF were generated by PCR (primers used were listed in Supplementary Table S1) using P. putida KT2440 genomic DNA. PCR product of each amplicon was then ligated with Gm^R-encoding cassette into pEX18Ap which contains the sacB-based counter-selection marker to create the replacement plasmid. Next, the replacement plasmid was transformed into KT2440 by electroporation and the resulting deletion mutants (mqsA gene was replaced by Gm via homologous recombination) were selected on LB plate supplemented with gentamicin (Hoang et al., 1998). Deletion of the mqsR gene and the mqsRA operon were performed following similar steps, and all three deletion mutants were confirmed by PCR and DNA sequencing with primers listed in Supplementary Table S1.

To overexpress mqsR, mqsA, and mqsRA in E. coli and P. putida KT2440 hosts, the full coding region of mqsR, mqsA, and mqsRA were amplified with primer pairs listed in Supplementary Table S1 using P. putida KT2440 genomic DNA as the template. PCR products were purified using a gel extraction kit (Qiagen, Hilden, Germany), digested with EcoRI (or NheI for pHGE-base plasmids) and HindIII, and were purified with a PCR product purification kit (Qiagen). The purified PCR products were ligated into the pMQ70 and pHGE expression plasmids and transferred into E. coli WM3064 and P. putida KT2440, respectively. The correct constructs were verified by DNA sequencing using primer pairs pMQ70-f/r and pHGE-f/r.

For purification of the MqsA protein and MqsR/MqsA complex, the coding region of mqsA and mqsRA were amplified with primer pairs pET28b-mqsA-f/-r to make pET28b-mqsA-His, pET28b-mqsRA-f/pET28b-mqsRA-His-r to make pET28b-mqsRA-CHis (Supplementary Table S1) using P. putida KT2440 genomic DNA as the template, PCR products were purified using a gel extraction kit (Qiagen), and digested with NcoI and HindIII, and were purified with a PCR product purification kit (Qiagen). The purified PCR products were ligated into the pET28b plasmid and transferred into E. coli BL21. The correct constructs were verified by DNA sequencing using primer T7.

For the promoter activity assay, DNA fragments containing 409 nt upstream of the translational start site of mqsR, a 328 nt upstream of the translational start site of PP_3288, and a 357 nt upstream of the translational start site of algU were selected as the promoter regions. pHGR01-PmqsRA-lacZ contains the promoter region of mqsRA, pHGR01-PmqsRA-mqsA-lacZ contains the promoter region and the ORF of the mqsA. Both of the fragments were digested with EcoRI and HindIII and cloned into the promoter-less lacZ-fusion vector pHGR01 digested with the same two enzymes. Constructed pHGR01-PmqsRA-lacZ was verified by DNA sequencing using the primer pair pHGR01-f/-r listed in Supplementary Table S1. pHGR01-PPP_3288-lacZ and pHGR01-PalgU-lacZ were constructed following the same procedure. The resulting plasmids were verified by DNA sequencing and transformed into E. coli WM3064 strain. Mid-log phase (OD₆₀₀ ∼ 0.7) cells of the indicated strains carrying the reporter plasmids were collected by centrifugation and washed with phosphate buffered saline. The cell soluble protein and β-galactosidase activity were determined using previously described protocols (Wu et al., 2011).

MqsA or the MqsR/MqsA complex with a hexahistidine tag at the C-terminus of MqsA and the MqsR/MqsA complex without any tag were purified using E. coli BL21 with pET28b-mqsA-His, pET28b-mqsRA-CHis, and pET28b-mqsRA. Strains were grown in LB with kanamycin (50 μg/ml) and were induced with 1 mM IPTG at a turbidity of 0.1 at 600 nm for 5 h. Cells were collected and resuspended in lysis buffer [50 mM potassium phosphate buffer (pH 8.0), 300 mM NaCl, and protease inhibitor cocktail (Sigma–Aldrich, USA)]. Samples were sonicated using a Sonic Dismembrator (Ningbo dongzhi, China) at level 2 for 5 min on ice. Ni-NTA resin (Qiagen) was used according to the manufacturer’s protocol. Purified proteins were desalted by passage on disposable Sephadex G-25 pre-packed PD-10 columns pre-equilibrated in 20 mM Tris-HCl buffer (pH 8.0), and the protein concentration was measured by the Bi Yuntian BCA assay kit (Haimen, Jiangsu, China). Tricine–SDS–PAGE was performed as previously described (Baba et al., 2006). A total of 20 μg of protein from each sample was loaded for SDS–PAGE.

Total RNA was isolated as described previously (Ren et al., 2004). The cDNA synthesis was conducted using reverse transcription (Promega, Madison, WI, USA). Total cDNA (50 ng) was used for qRT-PCR using the Step One Real-Time PCR System (Applied Biosystems StepOne Real-Time PCR System, USA). Primers used for qRT-PCR are listed in Supplementary Table S1, and the level of 16S rRNA gene transcript was used to normalize the gene expression data, and fold change of each gene was calculated as described previously (Pfaffl, 2001).

Attached biofilm formation was assayed in 96-well polystyrene plates with 0.1% crystal violet staining as described previously (Pratt and Kolter, 1998). To remove growth effects, biofilm formation was normalized by dividing the maximal bacterial growth (measured by the turbidity at 620 nm) for each strain. The assay was repeated twice with at least three independent cultures.

Electrophoretic mobility shift assays (EMSAs) were performed as previously described (Kim et al., 2010). The DNA probes were amplified by PCR using primers listed in Supplementary Table S1. The PCR products were gel purified and labeled with biotin using the Biotin 3′-end DNA Labeling Kit (Pierce, Rockford, IL, USA). For the binding reactions, biotin-probes (0.05 pmol DNA) were incubated with purified MqsA-His protein for 1 h at room temperature. The DNA binding reactions were performed with the non-specific competitor DNA (poly dI-dC) and NP-40 in buffer containing 10 mM HEPES (pH 7.3), 20 mM KCl, 1 mM MgCl₂, and 5% glycerol in a total volume of 20 μl at 25°C for 1 h. Aliquots of binding reaction mixtures were then electrophoresed on 6% DNA acrylamide retardation gels (Invitrogen, Carlsbad, CA, USA) at 100 V in 0.5 × TBE buffer (10 mM Tris borate at pH 8.3 and 2 mM EDTA) for 90 min. DNA was transferred to a nylon membrane at 390 mA for 40 min, followed by UV crosslinking at 302 nm for 20 min. Chemiluminescence was performed with the LightShift Chemiluminescent EMSA Kit (Thermo Scientific, Rockford, IL, USA) according to the manufacturer’s protocol.

Persistence of P. putida KT2440 was measured by time-dependent killing experiments (Dorr et al., 2010). To test the formation of persister cells in P. putida KT2440, overnight bacterial culture was re-inoculated into fresh LB medium (at a turbidity of 0.1 at 600 nm) and grown to an OD₆₀₀ of 1.0. For strains carrying pHGE-based plasmids, overnight cultures were diluted to a turbidity of 0.1 and grown to a turbidity of 0.5, then 1 mM IPTG was used to induce gene expression for 2 h, and the turbidity was adjusted to 1.0. Then the bacterial cultures were exposed to 1.25 μg/ml ciprofloxacin. Cell viability (CFU/ml) was determined at the time points indicated. Cell survival was determined by drop assay.

MqsR homologs were retrieved from ortholog clusters (K13651) in the KEGG ORTHOLOGY database and the data in a previous constructed phylogenetic tree of MqsR (Makarova et al., 2009). By excluding a few nearly identical sequences from different strains of the same bacterial species, the data set of total 133 genes were used for our phylogenetic analysis. The amino acid sequences were aligned with the MAFFT program (L-INS-i) (Katoh and Standley, 2013) and the poorly aligned sites were deleted by trimAl (with gt 0.8) (Capella-Gutierrez et al., 2009). A final data set of 98 amino acid sites was used for the maximum likelihood (ML) phylogenetic analysis by the W-IQ-TREE (Trifinopoulos et al., 2016). The best-fit substitution model was automatically determined and the reliability of internal branches was tested by 1000 ultrafast bootstrap replicates (Minh et al., 2013) in the W-IQ-TREE web interface. The ML tree was rooted to split the two groups of MqsR described previously (Makarova et al., 2009) and further annotated by the iTOL tool (Letunic and Bork, 2016).

The motifs 5′-ACCT (N)₃ AGGT-3′ were identified using the Biostrings and BSgenome libraries in the R statistical package (version 2.9.2) and Fuzznuc (EMBOSS) (Rice et al., 2000).

Data are presented as means ± SE of three or more independent cultures. Statistical significance was assessed using two-tailed unpaired Student’s t-test.

A total of 143 genes encoding for MqsR homologs were found in various bacterial genomes by searching the KEGG ORTHOLOGY database. As shown in Figure 1, MqsR toxin seems to be divergent in different genera. In consistent with what has been reported earlier (Makarova et al., 2009), MqsR toxin belongs to two phylogenetically distinct groups (Figure 1). Group 1 comprises mostly proteobacterial MqsR toxins, while Group 2 comprises MqsR toxins of Firmicutes, Bacteroidetes, and Chlorobi. MqsR homologs of Proteobacteria basically fall into two separate clades. The Proteobacteria clade 1 mainly comprises MqsRs of α-Proteobacteria and δ-Proteobacteria, and the Proteobacteria clade 2 comprises those of β-Proteobacteria and γ-Proteobacteria. MqsR homologs of P. putida seem more closely related with that of E. coli and Yersinia spp. than other Pseudomonas MqsR toxins.

Putative type II TA systems in the P. putida KT2440 genome were predicted with the web-based tool RASTA-Bacteria (Sevin and Barloy-Hubler, 2007). Two neighboring genes, PP_4205 and PP_4204, encoding two proteins of 98 aa and 133 aa, were identified as a putative TA pair (Figure 2A and Supplementary Figure S1). PP_4205 shares 58 % identity with MqsR in E. coli, and the four amino acid residues (Lys⁵⁶, Gln⁶⁹, Tyr⁸¹, and Lys⁹⁶) which are important for MqsR toxicity of E. coli (Brown et al., 2009) are also conserved in P. putida KT2440 (Figure 2B). PP_4204 shares 45% identity with MqsA in E. coli at the amino acid sequence level, the two residues Asn⁹⁷ and Arg¹⁰¹ that are important for MqsA to bind DNA and for transcription regulation in E. coli (Brown et al., 2009) are also conserved in P. putida KT2440 (Figure 2C).

To explore the function of this TA pair in P. putida KT2440, we first tested the toxicity of the two-gene cassette. The coding regions of the two genes were cloned into the pHGE plasmid to construct pHGE-PP_4204 (pHGE-mqsA) and pHGE-PP_4205 (pHGE-mqsR) using genomic DNA of P. putida KT2440 as template. When transformed into P. putida KT2440, cells expressing MqsR using pHGE-mqsR exhibited a notable decrease in cell growth and viability as shown by the reduction in turbidity at 600 nm (OD₆₀₀) and colony forming units (CFUs) (Figures 2D,E). In contrast, expression of MqsA using pHGE-mqsA did not affect cell growth in P. putida KT2440 (Figures 2D,E).

To further determine whether the neighboring protein MqsA can neutralize the toxicity of MqsR, we constructed the pHGE-mqsRA plasmid to co-express MqsA and MqsR in P. putida KT2440 cells. As expected, MqsA completely neutralized the toxicity of MqsR (Figures 2D,E). Moreover, co-expression of MqsA with MqsR effectively inhibited cell death caused by MqsR overproduction (Figures 2D,E). To further test the toxicity of these two proteins in a different host, we constructed and transferred pMQ70-based constructs into E. coli WM3064 host. As expected, MqsR was toxic while MqsA neutralized the toxicity of MqsR in E. coli host (Figure 2F). Therefore, PP_4205 and PP_4204 in P. putida KT2440 consist of a type II TA pair.

To determine whether mqsR and mqsA are co-transcribed, we designed a forward primer in the first gene mqsR (mqsR-RT-f) and a reverse primer in the second gene mqsA (mqsA-RT-r) (Supplementary Figure S1) and conducted a reverse transcription polymerase chain reaction (RT-PCR) assay. As shown in Supplementary Figure S2, a single band of 707 bp was amplified using cDNA synthesized from total RNA as the template, and the PCR product was sequenced to be the region among the two primers, indicating mqsR and mqsA form a single operon (Supplementary Figure S2, lane 3). As controls, the same band was detected using genomic DNA (Supplementary Figure S2, lane 2) as the template but not for total RNA (Supplementary Figure S2, lane 4). This result was also supported by qRT-PCR results as the same levels of transcripts were determined for mqsR and mqsA gene in the normal growing P. putida KT2440 cells (Supplementary Table S2). Collectively, these results demonstrate that mqsR and mqsA are co-transcribed.

Direct protein-protein interaction between toxin and antitoxin in vivo is a typical feature of type II TA systems (Brown et al., 2011). To determine whether MqsR binds to MqsA, we performed a pull-down assay using pET28b-mqsRA-CHis to co-express a C-terminus hexahistidine tagged (His-tagged) MqsA along with an untagged toxin MqsR (Yao et al., 2015; Guo et al., 2016). Affinity purification using Ni-NTA agarose beads and subsequent Tricine–SDS–PAGE revealed that another small protein was pulled down in addition to His-tagged MqsA (15.51 kDa) (Figure 3A, lane 4), and this small protein was verified to be MqsR (11.13 kDa) toxin by mass spectrometry (Supplementary Table S3). We also constructed pET28b-mqsRA to express untagged MqsR and untagged MqsA, and neither of them could bind to Ni-NTA beads (Figure 3B, lane 4), suggesting MqsR was purified due to its interaction with MqsA-His. Hence, we showed that inhibition of MqsR toxicity by antitoxin MqsA was most likely due to the direct interaction between the two proteins.

It has been shown that the type II antitoxin alone or the TA complex binds to its promoter and negatively regulates the transcription of the TA operon (Yamaguchi et al., 2011). Analysis of the promoter region of mqsRA operon revealed that it contains a 17 bp palindrome 5′-TTAACCT GGA AGGTTAA-3′, and the palindrome is overlapped the ribosome-binding site (RBS) for mqsR (Figure 4A). To check whether MqsA binds to the promoter region, we first performed EMSA using purified MqsA protein (Supplementary Figure S3) and a PCR product of 130 bp that includes the promoter region of the TA operon (Figure 4A and Supplementary Figure S1). MqsA specifically bound to the mqsRA promoter region in a concentration-dependent manner (Figure 4B). To determine whether MqsA binds to the palindromic sequence, we performed EMSA using a mutant DNA probe with the palindrome disrupted, and results showed that MqsA did not bind to the mutant DNA probe (Figure 4C). When the MqsR/MqsA complex purified using pET28b-mqsRA-CHis was used for EMSA, the TA complex could not bind and shift the mqsRA promoter (Figure 4D). This result is consistent with the previous work in E. coli which demonstrated that the MqsR/MqsA complex was unable to bind DNA when MqsR and MqsA are present at 1:1 ratio (Brown et al., 2013). These in vitro results suggest that the autoregulation of the mqsRA operon relies on the ratio of the toxin protein to the antitoxin protein present in the cell. Thus, similar to MqsR in E. coli, MqsR in P. putida KT2440 does not function as a transcriptional co-repressor, but blocks the antitoxin MqsA to bind DNA when they are present at approximately 1:1 ratio in the cell.

Moreover, a promoter activity assay using a transcriptional fusion was employed to study the autoregulation of the MqsR/MqsA TA pair in vivo. We cloned the promoter region of mqsRA into the pHGR01 plasmid to create the pHGR01-PmqsRA-lacZ reporter plasmid. To examine the effect of MqsA, we also fused the promoter region of mqsRA and the coding region of mqsA to the same promoterless lacZ reporter gene to create pHGR01-PmqsRA-mqsA-lacZ (Figure 4E, up panel). The promoter activity was decreased from 63 ± 4.0 miller units (MU) in the E. coli WM3064 cells carrying the pHGR01-PmqsRA-lacZ plasmid to 8 ± 0.1 MU in cells harboring the pHGR01-PmqsRA-mqsA-lacZ (Figure 4E, left panel), suggesting that the presence of MqsA repressed the promoter activity of pHGR01-PmqsRA-lacZ. In addition, we also overexpressed mqsA using pMQ70-mqsA in the reporter strain WM3064/pHGR01-PmqsRA-lacZ. The promoter activity reduced from 66.0 ± 5.0 MU to 3.0 ± 0.2 MU after overexpressing MqsA in the E. coli WM3064 cells. However, when TA complex was overproduced by pMQ70-mqsRA, the promoter activity was not significantly reduced in the reporter strain (Figure 4E, right panel). Therefore, MqsA represses transcription of the mqsRA operon, and the presence of MqsR blocks the activity of MqsA to bind DNA when it is co-expressed.

To probe the physiological function of MqsR/MqsA in P. putida KT2440, we constructed in-frame deletion mutants of mqsR, mqsA, and mqsRA in P. putida KT2440. The deletion mutants were verified by PCR (Figure 5A) followed by DNA sequencing to make sure no other mutations occurred in the neighboring gene. Deletion of the toxin gene or the antitoxin gene did not significantly affect cell growth (Figure 5B). Since TA systems are related to biofilm formation (Wang and Wood, 2011), attached biofilm formation was tested for the deletion mutants using the 96-well polystyrene plate assay. Attached biofilm formation was reduced 8.0 ± 0.1-fold, 6.3 ± 0.1-fold, and 6.8 ± 0.1-fold after 2 h for the mqsR, mqsA, and mqsRA deletion mutants compared to the wild-type KT2440 strain, respectively (Figure 5C). Moreover, overexpressing mqsA via pHGE-mqsA in the mqsA deletion mutant strain increased biofilm formation and overexpressing mqsRA via pHGE-mqsRA also increased biofilm formation (Figure 5D). We did not perform complementation study for mqsR due to the high toxicity of MqsR when overexpressed in P. putida KT2440.

To test the role of MqsR/MqsA in persistence, persister cell assay was performed for the mqsA or mqsR mutant strains using a high dose of ciprofloxacin. As shown in Figure 6A, deletion of mqsA increased the formation of persister cells 52.7 ± 4.2-fold, while deletion of mqsR or mqsRA did not significantly affected persister cell formation. We further complemented MqsA in the ΔmqsA strain via pHGE-mqsA, and results demonstrated that overproduction of MqsA inhibited the formation of persister cells 219.7 ± 10.2-fold compared to the ΔmqsA/pHGE strain (Figure 6B). Since the deletion of mqsA would activate the expression of mqsR, we conducted qRT-PCR to check the expression level of mqsR in the mqsA deletion mutant strain. As shown in Supplementary Table S2, the transcription of the toxin gene mqsR was induced 2.0 ± 0.1-fold in the ΔmqsA strain compared to the wild-type strain. In addition, expression of mqsR was repressed 5.6 ± 0.2-fold in the ΔmqsA/pHGE-mqsA strain compared to the ΔmqsA/pHGE-mqsA strain. Since there was no difference in persister cell formation between the ΔmqsR and ΔmqsRA strain, and deletion of mqsR did not affect mqsA expression as shown in the ΔmqsR strain (Supplementary Table S2), MqsR should be mainly responsible for the increased persistence detected in the mqsA deletion mutant and also responsible for the reduced persistence in the mqsA complementation strain in P. putida KT2440.

We have previously demonstrated that MqsA in E. coli can bind to the promoter of rpoS and csgD via the MqsA-like palindrome 5′-ACCT (N)₃ AGGT-3′ where N is any nucleotide. We found that although MqsR and MqsA in E. coli K-12 and in P. putida KT2440 share medium level of identify at the amino acid sequence level, the palindrome where MqsA binds are the same in these two strains. However, different from E. coli K-12, there is no MqsA-like palindrome in the gene encoding the stationary sigma factor RpoS in P. putida KT2440, and moreover, the master regulator CsgD for curli and cellulose production is not found in P. putida KT2440. Therefore, we conducted a whole-genome search to identify intergenic regions with 5′-ACCT (N)₃ AGGT-3′ in P. putida KT2440. As shown in Table 2, the MqsA-like palindrome was found in the upstream region of another three genes in P. putida KT2440, including algU, PP_3288, and nadB.

AlgU is one of the ECF sigma factors in Pseudomonas, which is required for full resistance of P. aeruginosa to oxidative stress and responsible for alginate biosynthesis (Martinezsalazar et al., 1996; Tart et al., 2005; Bazire et al., 2010). PP_3288 is predicted to be a universal stress family protein but its exact function has not been elucidated. NadB is L-aspartate amino acid oxidase and its activity is required for nicotinamide adenine dinucleotide (NAD) biosynthesis in both aerobic and anaerobic conditions in E. coli (Seifert et al., 1990). We first conducted qRT-PCR to check the expression of these three genes in the mqsA deletion mutant strain as compared to the wild-type P. putida KT2440 strain, and the transcription of the three genes were significantly induced (about twofold to threefold) as compared to the wild-type strain. In addition, overexpressing mqsA via pHGE-mqsA repressed the expression of these genes about threefold (Figure 7A). Next, we performed a promoter activity assay using a transcriptional fusion to study the regulation of the MqsA with algU in vivo. We cloned the promoter region of algU into the pHGR01 plasmid to create the reporter plasmid pHGR01-PalgU-lacZ. The promoter activity was decreased from 216.5 ± 2.3 MU to 75.2 ± 5.2 MU after overexpressing mqsA via pMQ70-mqsA in the E. coli WM3064 that harbored the reporter plasmid (Figure 7B). However, when the TA complex MqsR/MqsA was overproduced via pMQ70-mqsRA, the promoter activity was not significantly changed. Similar results were obtained for the PP_3288 gene. The promoter activity was decreased from 109.6 ± 5.1 MU to 8.8 ± 0.1 MU after overexpressing mqsA in the E. coli WM3064 that harbored the pHGR01-PPP_3288-lacZ plasmid, and when the MqsR/MqsA complex was overproduced by pMQ70-mqsRA, the promoter activity was not significantly reduced (Figure 7B). Collectively, these results suggest that MqsA regulates more than its own locus in P. putida KT2440, and this activity is also blocked by MqsR when it is co-expressed.