A recent study identified the open reading frame of HigB in PA14 and demonstrated its growth inhibitory function (Wood and Wood, 2016). In most type II TA systems, toxin and antitoxin genes form one operon and the antitoxin binds to and represses its own promoter (Wood and Wood, 2016). To test whether higB and higA are in the same operon, we designed a pair of primers annealing to the 5′ end of higB and 3′ end of higA coding region, respectively (Figure 1A), and performed RT-PCR. Total RNA was isolated from PA14 and a higA mutant from the PA14 transponson insertion mutant library (Liberati et al., 2006). A 384-bp PCR product was amplified using cDNA from the higA::Tn mutant (Figure 1B, lane 4), and the size was the same as that when genomic DNA was used as the template (Figure 1B, lane 2). Substantially less PCR product was obtained when cDNA from wild type PA14 was used as the template (Figure 1B, lane 3), suggesting a lower HigB mRNA level. To confirm the transcriptional level of higB and higA, we performed quantitative RT PCR with previously reported PA1769 and proC as internal controls for normalization (Savli et al., 2003; Son et al., 2007). Since HigB might cleave mRNAs and affect the expression of multiple genes, we included the 16S rRNA (PA0668.1) (Ruzin et al., 2007), which might not be a target of HigB, as another internal control. Similar mRNA fold of changes (within 1.5-fold difference) were observed when proC and the 16S rRNA were used as internal controls. Therefore, we used the 16S rRNA as the internal control in this study. At both exponential and stationary growth phases, the mRNA levels of higB and higA in the higA::Tn mutant were higher than those in wild type PA14 (Figures 1C,D). In addition, a previous microarray analysis has demonstrated an up regulation of higB in a higA mutant (Wood and Wood, 2016). In combination, these results suggest that higB and higA are in the same operon, which is negatively regulated by the HigA.

To examine whether HigA binds to the promoter of its own operon, we first determined the transcriptional start site by a 5′ RACE analysis. The start site was located at 29 bp upstream of the start codon for higB (Figure 1E). Of note, we found a palindromic sequence downstream of the −10 region (Figure 1E), which might be the binding site of HigA. Electrophoretic mobility shift assay (EMSA) revealed an interaction between the fragment and HigA, and mutation of the palindromic sequence abolished the interaction (Figure 1F). These results suggest that HigA directly binds to and represses the promoter of the higB-higA operon.

To identify which protease is involved in the degradation of HigA, a C-terminus His-tagged HigA (HigA-His) driven by a tac promoter was introduced into wild type PA14. After 60 min of culture in the presence of IPTG, spectinomycin was added to block protein synthesis, then the stability of HigA-His was monitored. In wild type PA14 and a clpP::Tn mutant, the HigA-His was gradually degraded. However, the protein was stable in a lon::Tn mutant (Figure 1G, Figure S1A), suggesting a role of the Lon protease in the degradation of HigA.

To test the role of HigB-HigA in persister formation, the higA::Tn mutant was examined for a time-dependent killing by ciprofloxacin. Compared to the wild type PA14, the higA::Tn mutant displayed 100-fold higher survival rate, which was restored to the wild type level by complementation with an intact higA gene (Figure 2A). However, a ΔhigB mutant displayed a similar survival rate as the wild type PA14 (Figure 2B), which we suspect might be due to redundant TA systems in P. aeruginosa. It has been demonstrated that sublethal level of ciprofloxacin induces persister formation (Dörr et al., 2009, 2010). Thus, we examined the role of HigB in ciprofloxacin induced persister formation as previously described (Dörr et al., 2009, 2010). Pre-exposure to 0.025 μg/ml (1/10 MIC) ciprofloxacin increased the survival rate of wild type PA14 by approximately 5-fold, suggesting an induction of persister formation (Figure 2B). However, deletion of higB or the higB-higA operon abolished such induction (Figure 2B). The expression of higB and higA was induced by the ciprofloxacin treatment (Figures 2C,D) and overexpression of HigB increased bacterial survival rate by approximately 1000-fold (Figure 2E). In addition, the minimal inhibitory concentration (MIC) of ciprofloxacin was not altered by the mutation of higA or overexpression of higB (data not shown). In combination, these results demonstrate that HigB contributes to persister formation.

RNA-seq analyses were employed to explore the effect of higA inactivation on bacterial global gene expression at both exponential and stationary growth phases. Compared to wild type PA14, expression of 193 genes was altered in the higA mutant at both growth phases (Table S1). Of note, all of the T3SS genes were up regulated in the higA mutant (Table 1), suggesting a regulatory role of the TA system on the T3SS.

To confirm the elevated expression of T3SS genes in the higA::Tn mutant, the mRNA levels of exsA, exsC (two positive regulatory genes), pcrV (required for translocation of effector proteins) and exoU (encodes for an effector protein) were examined. Mutation of higA resulted in higher mRNA levels of all of these genes, which were restored to the wild type levels by complementation with a higA gene (Figure 3A, Figure S1B). As the higA::Tn mutant grew slower than the wild type strain, translation of the T3SS genes might be impeded. To test this possibility, we examined PcrV protein levels by immunostaining in strains harboring a mcherry gene driven by the promoter of higB-higA (P_higB-mcherry). Compared to the wild type strain, the higA::Tn mutant expressed higher levels of PcrV and mCherry proteins (Figure S1C). Next, we constructed a C-terminal His-tagged ExoU driven by its native promoter (P_exoU-ExoU-His). Consistent with the above results, the levels of ExoU-His protein in the higA::Tn mutant were higher at both exponential (Figure 3B) and stationary growth phases (Figure S1D).

To test whether the increased expression of T3SS genes leads to higher cytotoxicity, we performed LDH release assay. Compared to wild type PA14, the higA::Tn mutant caused quicker cell death to either macrophages (Raw264.7) (Figure 3C) or epithelial cells (HeLa) (Figure S1E). In addition, when HeLa cells were infected with strains containing the ExoU-His, more ExoU was translocated into the cells by the higA::Tn mutant (Figures S1F,G). Altogether, these results demonstrate that mutation of the higA results in up regulation of T3SS genes and consequently higher cytotoxicity.

HigB functions as a RNase, which is directly inhibited by HigA (Wood and Wood, 2016). Thus, we examined the role of HigB in the expression of T3SS genes and cytotoxicity. First, a ΔhigBΔhigA double mutant was constructed, which displayed similar levels of T3SS gene expression and cytotoxicity as the wild type PA14 (Figures S2A,B). Second, a C-terminal His-tagged HigB driven by a regulatable tac promoter (Ptac-higB-His) or the empty vector was introduced into wild type PA14 (Figure S2C). Addition of IPTG increased the mRNA levels of exsC, exsA, pcrV, and exoU, with the highest levels in the presence of 0.5 mM IPTG. In the presence of 1.0 mM IPTG, the mRNA levels of those genes were reduced, which might be due to strong growth inhibition as a consequence of high level expression of the HigB (Figures 4A–D). To further confirm the expression level of ExoU, we transferred the plasmid carrying P_exoU-ExoU-His into the above strains. Consistently, the protein level of ExoU was increased by the overexpression of HigB (Figure 4E).

Next, we determined the correlation between bacterial cytotoxicity and expression levels of HigB. Bacteria grown in the presence 0.1 mM IPTG displayed the highest cytotoxicity to both Raw264.7 (Figure 4F) and HeLa cells (Figure S2D). However, further increase of the IPTG concentration reduced the cytotoxicity (Figure 4F, Figure S2D), although the mRNA levels of the T3SS genes were higher than or similar to those in the presence 0.1 mM IPTG (Figures 4A–D). Mutation of exsA severely reduced the HigB mediated increase of the T3SS gene expression and cytotoxicity (Figures 4A–F, Figure S2D). These results suggest that HigB promotes bacterial cytotoxicity through the T3SS. However, too high level of HigB might repress the overall bacterial fitness, which impedes the translocation of T3SS effector proteins.

Our results from the higA::Tn mutant and the HigB overexpressing strain demonstrate that activation of HigB increases persister formation and T3SS mediated cytotoxicity. A more important question is whether persister cells harbor higher levels of T3SS proteins and are more cytotoxic than their isogenic vegetative cells.

In the bacterial survival assay, we noticed lysis of bacterial cells during ciprofloxacin treatment, presumably due to production and release of pyocins (Penterman et al., 2014; Sun et al., 2014). Based on this phenotype, we developed a method to collect persister cell by washing the ciprofloxacin treated bacteria with 0.3M sucrose, which could efficiently remove lysed cell debris. To assess the effectiveness of this method, we treated a wild type PA14 strain containing a gfp gene driven by the higB promoter (P_higB-gfp) with 0.025 μg/ml ciprofloxin for 2 h to induce persister formation. Then the cells were incubated with 0.25 μg/ml ciprofloxacin for 6 h, resulting in a survival rate of 0.01% as determined by plating assay. Such treated bacterial cells were harvested by centrifugation and washed twice with 0.3M sucrose, followed by propidium iodide (PI) staining. As presented in Figure S3A, 93 ± 0.5% collected cells were PI negative, suggesting an effective isolation of persister cells. In addition, bacteria with strong green fluorescence were negative for PI staining, or vice versa (Figure S3A), indicating an up regulation of HigB in the persister cells. To examine the levels of PcrV in the persister cells, the collected bacterial cells were subjected to immunostaining with an anti-PcrV antibody. 79.7 ± 3.7% GFP positive cells were positive for PcrV (Figure S3B). In combination, these results demonstrate elevated levels of HigB and PcrV in persister cells.

Next, we examined the cytotoxicity of persister cells. Persister cells of wild type PA14 were collected as aforementioned and used to infect Raw264.7 cells. However, the persister cells displayed minimal cytotoxicity compared to vegetative cells (Figures S3C,D). We suspected that the 6-h exposure to ciprofloxacin might result in highly dormant cells that are unable to inject T3SS effectors. Therefore, we reduced the treatment time to 30 min, which resulted in 25% bacterial survival rate. As represented in Figure 5A, 83 ± 2.8% cells collected after ciprofloxacin treatment were PI negative. 89 ± 6.6% cells were GFP positive but GFP and PI double positive cell was barely observed (Figure 5A, lower panels), suggesting high levels of HigB in survived cells. Among the cells, 77 ± 10.1% were double positive for GFP and PcrV (Figure 5B, lower panels), whereas bacteria grown in LB were negative for GFP or PcrV (Figure 5B, upper panels). These results suggest that the expression levels of HigB and PcrV were higher in the survived bacterial cells than those in vegetative cells. These surviving bacteria displayed higher cytotoxicity to Raw264.7 cells (Figure 5C). Addition of anti-PcrV antibody, which has been demonstrated to protect cells from T3SS mediated cytotoxicity (Warrener et al., 2014), protected the Raw264.7 cells from killing by the bacteria survived of the ciprofloxacin treatment (Figure 5C).

To exclude the possibility that the macrophages were killed from stimulation by large amount of LPS or other bacterial ligands in the collected bacterial samples, we tested an exsA::Tn mutant strain. After the same ciprofloxacin treatment, the survival rate of the mutant was similar to that of wild type PA14, however, the bacteria displayed minimum cytotoxicity (Figures S3C,D). Next, we incubated wild type bacteria at 50°C for 30 min, which also resulted in 25% survival rate. The surviving bacteria barely caused cell death (Figures S3C,D). In combination, the above results demonstrate that compared to vegetative cells, bacteria that survived the 30 min ciprofloxacin treatment contain higher level of T3SS proteins, which leads to increased cytotoxicity.

To examine the role of HigB in the expression of T3SS genes in bacteria survived ciprofloxacin treatment, we treated either the ΔhigB mutant or the ΔhigBΔhigA mutant with 0.025 μg/ml ciprofloxin for 2 h followed by incubation with 0.125 μg/ml ciprofloxacin for 30 min. The expression levels of HigB and PcrV were examined by fluorescence microscopy as described above. Similar to wild type PA14, the promoter activity of higB was increased in the surviving bacteria (Figures S4, S5). The stronger fluoresce in the ΔhigBΔhigA mutant further confirmed the negative regulatory role of the HigA on the higB-higA operon (Figures S4, S5). The levels of PcrV were significantly lower in the ΔhigB or ΔhigBΔhigA mutant than that in the wild type PA14 (Figures S4, S5). Consistently, the ΔhigB or ΔhigBΔhigA mutant cells survived ciprofloxacin treatment displayed lower cytotoxicity compared to the counterpart of wild type cells (Figure 5D). Therefore, HigB plays an important role in the up regulation of T3SS genes and increased cytotoxicity in survived bacteria.

The bacterial strains used in this study are listed in Table S2. Bacteria were cultured in Luria–Bertani (LB) broth (10 g/l tryptone, 5 g/l Nacl, 5 g/l yeast extract, pH 7.0–7.5) or LB agar (LB broth containing 15 g/l agar) under aerobic conditions at 37°C. When needed, the medium was supplemented with tetracycline (100 μg/ml) (BBI life sciences, Shanghai, China), gentamicin (100 μg/ml) (BBI life sciences), streptomycin (50 μg/ml) (BBI life sciences), or carbenicillin (150 μg/ml) (BBI life sciences) for P. aeruginosa, and ampicillin (100 μg/ml) (BBI life sciences) for E. coli.

Plasmids used in this study are listed in Table S2. For DNA manipulation, standard protocols or manufacture instructions of commercial products were followed. Chromosomal gene mutations were generated as described previously (Hoang et al., 1998).

Total RNA was isolated from bacteria at indicated time points with an RNeasy Minikit (Tiangen Biotech, Beijing, China). The cDNA was synthesized from total RNA using random primers and PrimeScript Reverse Transcriptase (TaKaRa, Dalian, China). Specific Primers (Table S2) were used for quantitative RT PCR. For quantitative RT PCR, cDNA was mixed with 4 pmol of forward and reverse primers and SYBR Premix Ex Taq™ II (TaKaRa) in a total reaction volume of 20 μl. The results were determined using a CFX Connect Real-Time system (Bio-Rad, USA).

The transcriptional start site of the higB-higA operon was determined by 5′ RACE (rapid amplification of cDNA ends). The cDNA was synthesized from total RNA using primer higA-R and higB-R. cDNA was purified with a DNA Clean kit (Sangon Biotech, Shanghai, China) and tailed with poly (dC) using terminal deoxynucleotidyl transferase (TaKaRa), then amplified by PCR. The obtained PCR product was ligated into a T-vector (TaKaRa), then sequenced.

Electrophoretic mobility shift assay (EMSA) was performed as described with minor modification (Sun et al., 2014). Briefly, a 38-bp DNA fragment corresponding to sequence up-stream of higB start codon or the 38-bp DNA fragment with palindrome sequence scrambled as a negative control was synthesized. DNA fragments (300 ng) were incubated with 0, 4 or 20 nM purified recombinant HigA protein at 30°C for 30 min in a 20-μl reaction (10 mM Tris-HCl, pH 7.6, 4% glycerol, 1 mM EDTA, 5 mM CaCl₂, 100 mM NaCl, 10 mM-β-Mercaptoethanol). Samples were loaded onto a 8% native polyacrylamide gel in 0.5 × Tris-borate-EDTA (TBE) buffer (0.044 M Tris base, 0.044 M boric acid, 0.001 M EDTA, pH 8.0) that had been prerun for 1 h, electrophoresed on ice at 90 V for 2 h followed by DNA staining in 0.5 × TBE containing 0.5 μg/ml ethidium bromide. Bands were visualized with a molecular imager ChemiDoc™ XRS + (Bio-Rad).

Overnight culture of wild type PA14, a clpP::Tn or lon::Tn mutant harboring pMMB67EH-higA-His plasmid was sub-cultured in fresh LB broth to an OD₆₀₀ of 0.5, then induced with 1 mM IPTG for 1 h, followed by treatment with 50 μg/ml streptomycin. Bacteria were collected at 0, 0.5, 1, 2, 3, 4, and 5 h, boiled in 1 × SDS loading buffer, then subjected to SDS-PAGE. Proteins was transferred onto a PVDF membrane and incubated with mouse anti-His antibody (1:2000) (Millipore, USA) at room temperature for 1 h. After washing with 1 × phosphate buffered saline (1 × PBS: 274 mM NaCl, 5.4 mM KCl, 20 mM Na₂HPO₄, 4 mM KH₂PO₄, pH 7.4) for four times, the membranes were incubated with a horseradish peroxidase-conjugated goat anti-mouse IgG (1:2000) (Promega, USA) at room temperature for 1 h. Signals were detected with the ECL-plus kit (Millipore) and visualized with a Bio-Rad molecular imager ChemiDoc™ XRS+.

Persistence of P. aeruginosa was measured by time-dependent killing experiments (Dörr et al., 2010). To test the persistence level induced by sublethal level of ciprofloxacin, overnight bacterial culture was sub-cultured in fresh LB broth and grown to an OD₆₀₀ of 0.4 with or without 0.025 μg/ml ciprofloxacin. Then the bacterial cultures were exposed to 0.25 μg/ml ciprofloxacin. To test the effect of higA mutation or higB overexpression on persister formation, indicated strains were grown to an OD₆₀₀ of 0.4, followed by treatment with 0.25 μg/ml ciprofloxacin. At indicated time points, the live bacterial number was determined by serial dilution and plating. The plate was incubated at 37°C for 24 h before colony counting.

PA14 and the higA::Tn mutant were cultured in LB broth at 37°C and harvested at log phage (OD₆₀₀ of 0.8–1.0) and stationary phase (OD₆₀₀ of 2.5–3.0). Total RNA was extracted with an RNeasy Protect Bacteria Mini Kit with on-column DNase I digestion (Qiagen, Shanghai, China). A Turbo DNA-free vigorous protocol was used for a second round of DNase treatment (Ambion). 16S, 23S, and 5S rRNA were removed using the Ribo-Zero Magnetic Kit (Bacteria) (Epicentre).

Gene expression analysis was conducted via Illumina RNA sequencing (RNA-Seq technology). RNA-Seq was conducted for 3 biological replicates of each sample. The rRNA-depleted RNA was fragmented to 150–200 bp in sizes, then first and second strand cDNA were synthesized, followed by end repair and adapter ligation. After 12 cycles of PCR enrichment, the quality of the libraries was assessed using a Bioanalyzer (Agilent Technologies). The libraries were sequenced using an Illumina HiSeq 2500 platform with a paired-end protocol and read lengths of 100-nt.

The sequencing data was analyzed using the method described previously (Chua et al., 2014). Sequence reads were mapped onto PA14 reference genome (NC_008463) using a CLC genomics Workbench 8.0 (CLC Bio-Qiagen, Aarhus, Denmark). The count data of expression values were then analyzed using a DESeq package of R/Bioconductor. The differentially expressed genes were identified by performing a negative binomial test using the DESeq, with the cut-off of fold-change larger than 2. The raw sequence reads were normalized by dividing with size factors, then Log₂ (N + 1) transformed.

Bacteria with or without ciprofloxacin treatment were cytocentrifuged onto glass slides and fixed with 4% paraformaldehyde at room temperature for 30 min. Then bacteria were washed with 1 × PBS three times and permeabilized with 0.2% Triton X-100 in 1 × PBS at room temperature for 5 min. After washed twice with PBS, the bacteria were incubated with rabbit anti-PcrV serum (1:50) in PBSG (1 × PBS containing 0.1% gelatin) at 37°C for 1 h. The cells were washed twice with PBSG and incubated with the secondary antibody, green or red-conjugated goat anti- rabbit IgG (1:100) (Thermo Fisher Scientific, USA) in PBSG at 37°C for 1 h. To determine the viability, bacteria were stained with 1 μg/ml PI in 1 × PBS at room temperature for 15 min after 0.5 or 6 h ciprofloxacin treatment. Then cells were analyzed by a BX53 fluorescence microscope (Olympus, Japan).

Raw264.7 cells and HeLa cells were cultured in DMEM medium with 10% fetal bovine serum (FBS) at 37°C in 5% CO₂, and 95% air, supplemented with 1% penicillin/streptomycin and ciprofloxacin (10 μg/ml). Overnight bacterial culture was sub-cultured in fresh LB broth to OD₆₀₀ of 0.8 before infection. Bacteria were washed once and resuspended in 1 × PBS. Raw264.7 and HeLa cells were infected with bacteria at a multiplicity of infection (MOI) of 10 or 40, respectively, in DMEM medium without FBS and antibiotics. At the end of incubation, lactate dehydrogenase (LDH) present in the supernatant was measured using the LDH cytotoxicity assay kit (Beyotime, Haimen, China). Cells treated with LDH release agent C0017-1 were used as a control of total release (100% LDH release). The background level (0% LDH release) was determined with DMEM medium. The percentage of cytotoxicity was calculated following the manufacturer's instruction.

HeLa cells were infected with strains containing P_exoU-exoU-His at an MOI of 40. 1.5 h post infection, the cells were washed 3 times with 1 × PBS and lysed with 0.25% Trion-X 100. The Cell lysates were subjected to 10% SDS-PAGE. Proteins were transferred onto a PVDF membrane. The protein amounts of actin and ExoU were determined by Western blot analysis using mouse anti-His antibody or rabbit anti-β actin antibody (1:2000) (Cell Signaling Technology, USA).

Overnight bacterial cultures were sub-cultured in fresh LB broth to OD₆₀₀ of 0.4 with 0.025 μg/ml ciprofloxacin, then the bacterial cultures were exposed to 0.125 μg/ml ciprofloxacin. Bacteria with or without ciprofloxacin treatment were washed with 1 × PBS, then added to 10⁴ Raw264.7 cells in 200 μl culture medium with various concentrations (0, 1:100, 1:1000) of either normal rabbit IgG or rabbit anti-PcrV IgG. Each mixture was incubated at 37°C for 3.5 h. Cytotoxicity was measured by LDH release assay as described above.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.