Strains and plasmid used in this study are listed in Table 1. For the construction of an eno deletion mutant, a 934-bp upstream fragment and a 1207-bp downstream fragment of the eno conding region were amplified by PCR with PAK chromosome as the template and primers shown in Supplementary Table S1. The fragments were cloned into the plasmid pEX18TC (Hoang et al., 1998). Deletion of the eno gene in P. aeruginosa was performed as previously described (Hoang et al., 1998). For the complementation of eno, the eno gene and its native promoter were amplified with primers shown in Supplementary Table S1. The fragments were ligated into pUC18T-mini-Tn7T-Gm. The plasmid was transferred into the eno mutant strain along with the helper plasmid pTNS3 by electroporation as previously described (Choi and Schweizer, 2006). The ahpB and ahpC coding regions were amplified with primers shown in Supplementary Table S1 and ligated into pUCP20, respectively. The plasmid was transferred into the eno mutant by electroporation. To construct the ahpB- and ahpC-lacZ transcriptional fusions, the promoter regions of ahpB and ahpC were amplified with primers shown in Supplementary Table S1 by PCR. The fragments were ligated into the vector pDN19lacZΩ (Li et al., 2013).

All bacterial strains were cultured in Luria broth (LB, 1% Bacto-tryptone, 0.5% yeast extract, 1% NaCl; Oxoid Ltd, USA) at 37°C. Antibiotics were used at the following concentrations: for E. coli, kanamycin 50 μg/ml, gentamicin 15 μg/ml; for P. aeruginosa, carbenicillin 150 μg/ml, gentamicin 50 μg/ml, tetracycline 50 μg/ml. All antibiotics are from BBI Life Science, Shanghai, China.

β-Galactosidase assay was performed as previously described (Miller, 1972) with minor modifications. Briefly, bacteria were cultured overnight and diluted 1:100 in fresh LB medium and grown at 37°C with agitation. When the optical density at 600 nm (OD₆₀₀) reached 2.0, 0.5 ml bacteria were collected by centrifugation and resuspended in 1.5 ml Z buffer (60 mM Na₂HPO₄, 60 mM NaH₂PO₄, 10 mM KCl, 1 mM MgSO₄, 50 mM β-mercaptoethanol, pH 7.0; BBI Life Science, Shanghai, China). One milliliter of the suspension was allocated for OD₆₀₀ measurement. The other 0.5 ml suspension was added with 10 μl chloroform (BBI Life Science, Shanghai, China) and 10 μl 0.1% SDS (BBI Life Science, Shanghai, China), followed by vortex for 10 s. Then 100 μl ONPG (40 mg/ml; Sigma, USA) was added to the mixture and incubated at 37°C. The reaction was stopped by addition of 0.5 ml 1M Na₂CO₃. The time was recorded and OD₄₂₀ was measured. β-Galactosidase activity (Miller units) was calculated as (1000 × OD₄₂₀)/(T × V × OD₆₀₀). T, reaction time (minute); V, bacteria volume (ml).

Infection of mouse was performed as previously described (Sun et al., 2014). Briefly, overnight bacterial culture was diluted 1:100 in fresh LB medium and grown at 37°C with agitation. When the optical density at 600 nm (OD₆₀₀) reached 1.0, bacteria were collected and resuspended in phosphate-buffered saline (PBS) at a concentration of 1 × 10⁹ CFU/ml. Six to eight weeks old female BALB/c mice (Vital River, Beijing, China) were anesthetized by the injection of 100 μl 7.5% chloral hydrate (Sigma, USA) intraperitoneally. Twenty microliter bacterial suspension was then inoculated intranasally to each mouse, resulting in 2 × 10⁷ CFU bacteria per mouse. Twelve hours post-infection (hpi), the mice were sacrificed and lungs were isolated and homogenized in 1% proteose peptone (Sigma, USA), followed by determination of bacterial loads by serial dilution and plating. In the mortality assay, each mouse was infected with 4 × 10⁷ CFU bacteria, and monitored for 6 days. The statistical analysis was performed with the Prism software (Version 6, Graphpad Software, La jolla, USA).

Total RNA was isolated with the RNA prep Pure cell/Bacteria Kit (Tiangen Biotec, Beijing, China). Random primers and the Prime Script Reverse Transcriptase (Takara, Dalian, China) were used to synthesize cDNA. The cDNA was used as the template to detect the relative mRNA levels of indicated genes with specific primers and Fast Start Essential DNA Green Master (Roche, Switzerland). Gene PA1805 was used as the internal control (Son et al., 2007).

Twelve hours after infection with indicated P. aeruginosa strains, lungs of the mice were removed and fixed with 10% paraformaldehyde (Sigma, USA), then dehydrated with ethanol (Tian Jin chemical reagent company, Tianjin, China), and embedded in paraffin (BBI Life Science, Shanghai, China). The tissue sections were cut into slices and stained with hematoxylin and eosin (BBI Life Science, Shanghai, China). Images were taken with an Olympus microscope (Version IX71, Tokyo, Japan).

Bacterial cytotoxicity was determined by the lactate dehydrogenase (LDH) release assay. Briefly, HeLa cells (ATCC, USA) were cultured in Dulbecco’s modified Eagle medium (DMEM, Hyclone, USA) with and 2% (vol/vol) heat-inactivated fetal bovine serum (hiFBS, Gibco, Australia) at 37°C with 5% CO₂. Eighteen hours before infection, 1.2 × 10⁵ HeLa cells were seeded into each well of a 24-well plate. Bacteria were grown to an OD₆₀₀ of 1.0, collected by centrifugation, then washed twice and resuspended in PBS. After addition of bacteria to each well, the plate was centrifuged at 1,700 g for 10 min to synchronize the infection. Three hours after the infection, LDH released from the dead cells was measured by the LDH cytotoxicity assay kit (Beyotime, Haimen, China). Cells treated with the cell lysis buffer provided by the kit were used as the control of 100% LDH release. The culture medium without cell was used to set the background LDH level. The cytotoxicity percentage was calculated following the manufacturer’s instruction.

Over night bacterial culture was diluted 1:100 in LB or 1:50 in LB with 5 mM EGTA (BBI Life Science, Shanghai, China) and incubated at 37°C with agitation. After 4 h, the supernatant of each culture was collected by centrifugation. Supernatants collected from equal numbers of bacteria were loaded to a 12% sodium dodecyl sulfate-polyacrylamide gel (SDS-PAGE). Then the proteins were transferred to a polyvinylidene difluoride (PVDF, Millipore, USA) membrane, and probed with a rabbit polyclonal antibody against ExoS (Li et al., 2013) at room temperature for 1 h. The membrane was washed three times with PBS containing 0.2% tween-20 (Tian Jin chemical reagent company, Tianjin, China), followed by incubation with a horseradish peroxidase-conjugated goat anti-rabbit IgG (Millipore, USA) at room temperature for 1 h. The signal was detected with the ECL-plus kit (Millipore, USA).

HL-60 cells (ATCC, USA) were cultured in RPMI 1640 medium (Hyclone, USA) with 10% (vol/vol) heat-inactivated fetal bovine serum (Gibco, Australia) and penicillin G (100 U/ml) and streptomycin (100 μg/ml; Hyclone, USA). The cells were cultured at 37°C with 5% CO₂. Differentiation of the HL-60 cells was conducted as previously described (Chen and Seifert, 2011). Briefly, HL-60 cells were diluted to ∼4.5 × 10⁵ cells/ml and 1.3% dimethylsulfoxide (Sigma, USA) was added to the medium. The cells were then cultured for 6–7 days before use.

The ROS production levels were determined as previously described with slight modifications (Wu and Hsu, 2009). Briefly, differentiated HL-60 cells were washed once with warm Hank’s balanced salt solution (HBSS; Hyclone, USA) and diluted to 7.5 × 10⁴ cells/ml in HBSS containing 100 μM luminol (Sigma, USA) and 5 units per ml horseradish peroxidase (Sigma, USA). Two hundreds microliter cell suspension was added to each well of a 96-well plate, followed by incubation at 37°C for 10 min. Then the cells were infected with wild-type PAK or the Δeno mutant at a multiplicity of infection (MOI) of 30. The ROS levels were measured every 3 min for 4 h with a Luminoskan Ascent Luminometer (Varioskan Flash, Thermo Scientific, USA).

Bacteria were grown to an OD₆₀₀ of 1.0, collected by centrifugation and washed three times with sterile PBS. Then 1 × 10⁷ bacteria of each strain were incubated with 1 × 10⁶ undifferentiated or differentiated HL-60 cells in 200 μl RPMI 1640 medium at 37°C. At indicated time points, the live bacterial numbers were determined by serial dilution and plating. The growth inhibitory rate of each strain was calculated by dividing the live bacterial number incubated with differentiated HL-60 cells by the live bacterial number incubated with undifferentiated HL-60 cells.

Bacteria at an OD₆₀₀ of 1.0 were collected and washed for three times with sterile PBS. Then the bacteria were diluted to 2 × 10⁷ CFU/ml in PBS and incubated with or without 10 mM H₂O₂ at 37°C for 15 min. The live bacterial numbers were determined by serial dilution and plating. The survival rate was calculated by dividing the live bacterial number with H₂O₂ treatment by the live bacterial number without H₂O₂ treatment.

All animal experiments complied with Chinese national guidelines on the use of animals in research. The protocol was approved by the institutional animal care and use committee of the college of life sciences of Nankai University with a permit number: NK-04-2012.

To evaluate the role of enolase in P. aeruginosa pathogenesis, we utilized a murine acute pneumonia model as previously described (Sun et al., 2014). Six weeks old female BALB/c mice were infected intranasally with 2 × 10⁷ wild type PAK or an enolase deletion mutant (Δeno). Twelve hours post-infection, lungs were isolated and homogenized. Bacterial loads were determined by serial dilution and plating. Compared to the wild type strain, the number of the Δeno mutant was significantly lower (Figure 1A). For the complementation, an eno gene driven by its native promoter was cloned into pUC18T-mini-Tn7T-Gm and inserted into the chromosome (Choi and Schweizer, 2006). As shown in Figure 1A, complementation with an eno gene fully restored the bacterial load in the lung, indicating a role of enolase in bacterial growth in the lung. When, we grew the bacteria in LB, we noticed that the Δeno mutant grows more slowly than wild type PAK (Supplementary Figure S1A). After 12 h in vitro growth, the bacterial number of the Δeno mutant was ∼70% of that of the wild type PAK or the complemented strain (Supplementary Figure S1B). Given that there was ∼10⁴-fold difference in the bacterial load between wild type PAK and the Δeno mutant infected mice, it is likely that factors other than slow growth contribute to the reduced bacterial number in vivo. To examine the role of enolase in virulence, we monitored the mortality rate in the acute pneumonia model. Infection with wild type PAK or the complemented strain resulted in 82.5% mortality rate, whereas no mouse died after infection with the Δeno mutant (Figure 1B). Furthermore, lungs from mice at 12 hpi were subjected to H&E staining. Infection with wild type PAK resulted in severe occlusion with neutrophil infiltration, which was significantly milder in the Δeno mutant infected lungs (Figure 1C). Consistently, lower mRNA levels of inflammatory cytokines, including IL-1β, IL-6, and TNF-α were detected in the lungs infected with the Δeno mutant compared to those in the wild type PAK infected lungs (Figure 1D). Therefore, enolase is required for bacterial virulence in the acute pneumonia model.

In the mouse acute pneumonia model, neutrophils are rapidly recruited to the lung after infection and play a major role in the defense against bacteria (Wu et al., 2012; Ziltener et al., 2016). Induction and delivery of T3SS effector into neutrophils inhibit the bactericidal effects of those cells, enabling the bacterial colonization and dissemination (Diaz and Hauser, 2010; Howell et al., 2012; Rangel et al., 2015). Previously, we found that PNPase is required for the expression of the T3SS genes in the mouse acute pneumonia model (Chen et al., 2016). Since both enolase and PNPase are components of the RNA degradosome, they may share common regulatory targets. Thus, we examined the effect of eno mutation on the activity of the T3SS. Surprisingly, the expression and secretion of ExoS were similar between the Δeno mutant and wild type PAK upon growth in calcium depleted LB medium, which is a typical in vitro T3SS inducing condition (Figure 2A), and the bacterial cytotoxicity were similar between wild type PAK and the Δeno mutant (Figure 2B). We further examined the expression levels of T3SS genes during infection. Bacteria were isolated from bronchoalveolar lavage fluid (BALF) of infected mice. The mRNA levels of T3SS genes exsC and pcrV were determined by qRT-PCR with previously reported PA1805, PA1769, rpsL, and the 16S rRNA PA0668.1 as internal controls for normalization (Savli et al., 2003; Ruzin et al., 2007; Son et al., 2007; Sun et al., 2014). Similar mRNA fold of changes (within 1.2-fold difference) were observed between these internal controls. Therefore, we used the PA1805 as the internal control in this study. As shown in Figure 2C, the expression levels of exsC and pcrV were similar between wild type PAK and the Δeno mutant. In combination, these results suggest that mutation of the eno does not affect the expression of T3SS genes.

Next, we compared the impact of neutrophils on the Δeno mutant and wild type PAK. The bacteria were incubated with differentiated HL-60 (designated as dHL-60 hereafter) and undifferentiated HL-60 in RPMI-1640 medium. Compared to wild type PAK, the Δeno mutant was more susceptible to the dHL-60 mediated growth inhibition (Figure 3A). A major bactericidal mechanism of neutrophils is production of ROS (Arai et al., 2001; Alalwani et al., 2009). As shown in Figure 3B, dHL-60 generated large amount of ROS upon encountering PAK or the Δeno mutant. Therefore, we suspected that the Δeno mutant is more susceptible to oxidative stresses. Indeed, in a disk diffusion assay, H₂O₂ caused bigger inhibition zone on the Δeno mutant than that on the wild type PAK (Figure 3C). And treatment with H₂O₂ resulted in significant lower survival rate of the Δeno mutant (Figure 3D). Complementation with an eno gene restored the bacteria tolerance to H₂O₂ (Figure 3D). These results suggest that enolase is involved in the bacterial tolerance to oxidative stresses.

In P. aeruginosa, the chromosomally encoded catalases (KatA and KatB), and alkyl hydroperoxide reductases (AhpB and AhpC) play important roles in the bacterial tolerance to oxidative stresses (Hassett et al., 1992; Ma et al., 1999). Thus, we examined whether enolase affects the expression of those genes. In wild type PAK, treatment with H₂O₂ induced the expression of katA, katB, ahpB, and ahpC. In the eno mutant, similar expression levels of katA and katB were observed (Figures 4A,B), except for the level of katA in the absence of H₂O₂, which was higher than that in wild type PAK (Figure 4A). However, the mRNA levels of ahpB and ahpC in the Δeno mutant were 20- and 2-fold lower, respectively, in the presence of H₂O₂ (Figures 4C,D). To further confirm the expression levels of ahpB and ahpC, we constructed transcriptional fusions of ahpB promoter (P_ahpB) or ahpC promoter (P_ahpC) with lacZ reporter gene. In the presence of H₂O₂, the expression levels of ahpB-lacZ and ahpC-lacZ were reduced by ∼45 and 35% in the Δeno mutant, respectively (Figures 4E,F).

In P. aeruginosa, OxyR activates the expression of katA, katB, ahpB, and ahpC in response to oxidative stresses (Heo et al., 2009). Since the expression levels of katA and katB were similar between wild type PAK and the Δeno mutant in the presence of H₂O₂, we suspect that the expression and function of OxyR are normal in the eno mutant. Indeed, the mRNA levels of OxyR were similar between wild type PAK and the eno mutant with or without H₂O₂ treatment (Supplementary Figure S2A). In addition, expression of prpL, toxA, and rgsA, under the control of OxyR, was not affected by the mutation of eno (Supplementary Figures S2B–D). These results suggest that enolase affects the expression of ahpB and ahpC independent of the OxyR.

Next, we examined the expression levels of katA, katB, ahpB, and ahpC in bacteria during mouse lung infection. At six hpi, bacteria were collected from BALF, followed by RNA extraction and qRT-PCR. The mRNA levels of katA and katB in the Δeno mutant were slightly higher than those in wild type PAK (Figures 5C,D), however, the ahpB and ahpC mRNA levels were lower in the Δeno mutant (Figures 5A,B).

The in vitro and in vivo results shown above demonstrate defective expression of ahpB and ahpC in the Δeno mutant, which might be the cause of reduced tolerance to oxidative stresses. To test this further, we overexpressed the two genes individually or together in the Δeno mutant and examined the bacterial survival rates after H₂O₂ treatment. As shown in Figure 6A, overexpression of ahpB but not ahpC in the Δeno mutant restored the survival rate. Compared to ahpB alone, co-overexpression of ahpB and ahpC only slightly increased the bacterial survival rate. Consistently, overexpression of ahpB but not ahpC in the Δeno mutant restored the bacterial growth in the presence of dHL60 (Figure 6B).

In the mouse acute pneumonia model, overexpression of ahpB in the Δeno mutant increased the average bacterial load by ∼10-fold. However, overexpression of ahpC had no effect on the bacterial load (Figure 6C). In addition, overexpression of ahpB in the Δeno mutant did not alter the bacterial growth rate in LB medium (Supplementary Figure S1). Therefore, these results suggest that down regulation of ahpB is the major cause of decreased tolerance to H₂O₂ and the reduced bacterial load of the Δeno mutant.