The Enhancer-Binding Protein MifR, an Essential Regulator of α-Ketoglutarate Transport, Is Required for Full Virulence of Pseudomonas aeruginosa PAO1 in a Mouse Model of Pneumonia

ABSTRACT

The opportunistic human pathogen Pseudomonas aeruginosa PAO1 has an extensive metabolism, enabling it to utilize a wide range of structurally diverse compounds to meet its nutritional and energy needs. Interestingly, the utilization of some of the more unusual compounds often associated with a eukaryotic-host environment is regulated via enhancer-binding proteins (EBPs) in P. aeruginosa. Whether the utilization of such compounds and the EBPs involved contribute to the pathogenesis of P. aeruginosa remains to be fully understood. To narrow this gap, we investigated the roles of the EBPs EatR (regulator of ethanolamine catabolism), DdaR (regulator of methylarginine catabolism), and MifR (regulator of α-ketoglutarate or α-KG transport) in the virulence of P. aeruginosa PAO1 in a pneumonia-induced septic mouse model. Deletion of genes encoding EatR and DdaR had no significant effect on the mortality of P. aeruginosa PAO1-infected mice compared to wide-type (WT) PAO1-infected mice. In contrast, infected mice with ΔmifR mutant exhibited a significant reduction (~50%) in the mortality rate compared with WT PAO1 (P < 0.05). Infected mice with ΔmifR PAO1 had lower lung injury scores, fewer inflammatory cells, decreased proinflammatory cytokines, and decreased apoptosis and cell death compared to mice infected with WT PAO1 (P < 0.05). Furthermore, molecular analysis revealed decreased NLRP3 inflammasome activation in infected mice with ΔmifR PAO1 compared to WT PAO1 (P < 0.05). These results suggested that the utilization of α-KG was a contributing factor in P. aeruginosa-mediated pneumonia and sepsis and that MifR-associated regulation may be a potential therapeutic target for P. aeruginosa infectious disease.

INTRODUCTION

The bacterium Pseudomonas aeruginosa continues to be one of the more challenging pathogens facing healthcare officials and communities. This opportunistic pathogen is a frequent cause of hospital- and community-acquired pneumonia, as well as a common infectious agent in intensive care units (ICUs). Acute infections are more common in hospitalized patients who use ventilators and those infected with P. aeruginosa have a 70% to 80% fatality rate (1).

The metabolic demands of P. aeruginosa during an infection must be fulfilled from nutrients available in the eukaryotic-host environment. Intriguingly, P. aeruginosa relies on an unusual family of transcriptional regulators called enhancer-binding proteins (EBPs) to coordinate the metabolism of several important amino acids, sugars, dicarboxylates, and other organic molecules that are associated with a eukaryotic-host environment and/or infection. For example, the EBP PhhR regulates the catabolism of phenylalanine and tyrosine (2). Both aromatic amino acids are present in the sputum of individuals suffering from cystic fibrosis (CF). Notably, their catabolism has been shown to promote the biosynthesis of the Pseudomonas quinolone signal (PQS), thereby enhancing the QS response and subsequent production of virulence factors (2, 3). Other examples include the EBP GcsR, which regulates the catabolism of the mucin-relevant amino acids glycine and serine (4, 5), and the EBP AauR, an essential regulator in the transport of glutamate and glutamine (6), common proteogenic amino acids that are important amino donors for the cell.

Although the metabolic and/or biological functions have now been defined for the majority of EBPs in P. aeruginosa, the roles or significance of these regulatory proteins in the virulence of this pathogen have only been investigated for a few of them (5, 7, 8). Therefore, in the current study, we focused on three EBPs known to be involved in the utilization of metabolites associated with a eukaryotic-host environment and evaluated their contributions to P. aeruginosa-induced pneumonia and sepsis in mice. The three EBPs of interest were EatR, DdaR, and MifR of P. aeruginosa PAO1. The EBP EatR regulates the catabolism of ethanolamine (9), which is derived from the breakdown of the phospholipid phosphatidylethanolamine. Ethanolamine utilization has been associated with the virulence or host interactions of some human pathogens, such as Salmonella Typhimurium and Listeria monocytogenes (10). Arginine methylation is a common posttranslational modification of eukaryotic proteins (11, 12). The degradation of eukaryotic proteins releases free methylarginines that can be subsequently consumed by various pathogenic microorganisms (13), including P. aeruginosa in which regulation by DdaR is essential (14). The EBP MifR of P. aeruginosa regulates the expression of the PA5530 gene, encoding an α-ketoglutarate (α-KG) transporter (15–17). Interestingly, expression analysis of P. aeruginosa in the CF lung or when exposed to human airway epithelial cells revealed that the PA5530 gene was induced (18, 19), suggesting that α-KG might be a key nutrient for this human pathogen in the host environment. To assess the contributions of the EBPs EatR, DdaR, and MifR in the virulence of P. aeruginosa, a mouse model of pneumonia and sepsis were used. Mice were infected with each respective ΔEBP mutant, and the mortality and lung injury of infected mice were measured. Additional in vitro cytotoxicity assays with a human airway epithelial cell line were conducted to corroborate the results from the in vivo mouse studies.

RESULTS

The ΔeatR, ΔddaR, and ΔmifR mutants exhibited WT levels of virulence factors.

Before testing the virulence of the ΔeatR, ΔddaR, and ΔmifR mutants in a mouse model of pneumonia and sepsis, we first evaluated the EBP mutants for the production of common virulence factors and/or traits, including motility, protease activity, biofilms, pyocyanin, and pyoverdine. Protease activity (Fig. S1A in Supplemental File 1) and motility (Fig. S1B in Supplemental File 1) of the three EBP mutants were similar to the WT. In addition, all three EBP mutants produced pyoverdine (Fig. S2A in Supplemental File 1), pyocyanin (Fig. S2B in Supplemental File 1), and biofilms (Fig. S2C in Supplemental File 1) to levels identical to that of the WT. Collectively, these findings indicated that deletion of the eatR, ddaR, or mifR gene did not negatively affect the production of the aforementioned virulence factors in P. aeruginosa PAO1 when assayed under standard laboratory conditions.

Deletion of the mifR gene increased the survival time of mice challenged with P. aeruginosa-mediated pneumonia and sepsis.

To begin to understand the functional relevance of EBPs in the survival and pathogenesis of P. aeruginosa in a host environment, we infected mice with the ΔeatR, ΔddaR, and ΔmifR mutants, as well as WT P. aeruginosa PAO1. As shown in Fig. 1, the survival times of mice infected with the ΔeatR and ΔddaR mutants were statistically similar to the survival times of mice infected with WT. At 36 h postinfection, the survival probability for mice infected with WT P. aeruginosa PAO1 was 50%. A survival probability of 50% was observed at 42 and 39 h for mice infected with the ΔeatR or ΔddaR mutants, respectively, and neither one was statistically different from the WT-infected group. In contrast, the survival probability for mice infected with the ΔmifR mutant did not reach 50% until 48 h postinfection. Most deaths of mice infected with the ΔmifR mutant occurred after 48 h, whereas those infected with WT P. aeruginosa PAO1 were mostly dead before this time. Overall, deletion of a key transcriptional regulator in the utilization of α-KG − but not in the utilization of methylarginines or ethanolamine − increased the survival time of mice when challenged with P. aeruginosa-mediated pneumonia and sepsis.

FIG 1.

Comparison of the survival time of mice. The animals were treated with wild-type P. aeruginosa PAO1 (WT PAO1), ΔddaR PAO1, ΔeatR PAO1, ΔmifR PAO1 (1 × 10⁶ CFU/mouse), or sterile saline (Sham). The results from the Kaplan-Meier survival analysis showed a lower rate of mortality in the mice infected with ΔmifR PAO1 compared with the mice infected with WT PAO1. *, P = 0.033, n = 20.

Mice infected with the ΔmifR mutant showed reduced lung injury compared to those infected with WT P. aeruginosa PAO1.

Having established that the mifR gene contributed to P. aeruginosa PAO1-induced pneumonia and sepsis, we next evaluated the extent of lung injury occurring under such circumstances. It was found that WT P. aeruginosa PAO1 caused severe lung damage in mice (Fig. 2). The injured lung showed many inflammatory cells in the alveolar and interstitial space, accumulation of protein fragments, and alveolar septal thickness (Fig. 2A). In comparison, the lungs of mice infected with the ΔmifR mutant had a moderate number of inflammatory cells, a small amount of accumulated protein fragments, and only some of the alveolar walls were thickened (Fig. 2B). The lung tissue of the mice in the sham group was normal (Fig. 2C). When the extent of lung injury was scored using a 0 to 2 scale, a mean score of 0.459 was calculated for mice infected with the ΔmifR mutant versus a mean score of 0.634 for mice infected with WT P. aeruginosa PAO1 (Fig. 2D).

FIG 2.

Lung histological examination after infection. Sections of lungs stained with hematoxylin-eosin after infection 24 h are shown for mice infected with (A) WT PAO1, (B) ΔmifR PAO1, and (C) Sham group. The lung tissues of the mice of WT PAO1 group (A) showed severe tissue injury, evidenced by large amounts of inflammation cells (black arrows) filled in alveolar spaces and interstitials, proteinaceous debris accumulation, and alveolar septal thickness. (D) Quantitative analysis of lung injury scores indicated that the mice in the ΔmifR PAO1 group showed a significantly lower score of lung injury than the mice in the WT PAO1 group. Data points represent mean values ± 25% quantile (box plot) and the minimum and largest observed values (lower and upper edge). n = 6; *, P < 0.05; ***, P < 0.001. Magnification 200×, scale bar = 100 μm. Pneu, pneumonia.

Reduced numbers of inflammatory cells and CFU were observed in the BALF of mice infected with the ΔmifR mutant versus WT P. aeruginosa PAO1.

We compared the total number of macrophages and neutrophils in the bronchoalveolar lavage fluid (BALF) of mice that had either been infected with the ΔmifR mutant or WT P. aeruginosa PAO1. In both groups, 90% of the observed inflammatory cells in the BALF were neutrophils (Fig. 3A and B). Macrophages were predominantly observed in the BALF of mice in the sham group (Fig. 3C). The total number of alveolar macrophages and neutrophils was significantly lower in the BALF of mice that had been infected with the ΔmifR mutant compared to those infected with WT P. aeruginosa PAO1 (Fig. 3D and E). In addition, the BALF of mice infected with the ΔmifR mutant had a bacterial count of 331.5 CFU ± 129.5 CFU, which was significantly lower than the observed 625.3 CFU ± 252.4 CFU for WT P. aeruginosa PAO1 (Fig. 3F).

FIG 3.

Inflammatory cells and bacteria in BALF from infected mice. Representative images show macrophages and neutrophils in BALF from mice infected with (A)WT PAO1, (B) ΔmifR PAO1, and (C) Sham. The cells from BALF are mounted on slides by the cytospin centrifugation and then stained with a HEMA3 staining kit. The results showed that the (C) Sham group of mice only had macrophages (green arrows), and the (A) WT PAO1 and (B) ΔmifR PAO1 groups of mice contained more than 90% of neutrophils (black arrows). The measured cell counts of (D) macrophages and (E) neutrophils were significantly lower in the BALF from the ΔmifR PAO1 group of mice compared to the WT PAO1 group of mice. (F) Reduced bacteria was observed from the BALF of the ΔmifR PAO1 group of mice compared with the WT PAO1 group of mice. Data points represent mean values ± 25% quantile (box plot) and the minimum and largest observed values (lower and upper edge). n = 6; *, P < 0.05; **, P < 0.01; ***, P < 0.001. Magnification 400×, scale bar = 50 μm. BALF, bronchoalveolar lavage fluid; Pneu, pneumonia.

Mice infected with the ΔmifR mutant exhibited decreased levels of plasma cytokines compared to mice infected with WT P. aeruginosa PAO1.

In a sepsis mouse model induced by pneumonia, the continuous high expression of inflammatory cytokines is an important characteristic (20). Therefore, we measured the levels of three key cytokines, tumor necrosis factor (TNF)-α (Fig. 4A), interleukin (IL)-6 (Fig. 4B), and IL-1β (Fig. 4C), in the plasma of mice that had been infected with the ΔmifR mutant or WT P. aeruginosa PAO1. The mean levels of TNF-α, IL-6, and IL-1β in the plasma of mice infected with WT were 56.1 ±10.7, 484.4 ± 84.1, and 579.4 ± 96.0, respectively (Fig. 4A to C). These levels were significantly greater than those observed in the plasma of mice infected with the ΔmifR mutant, which were 37.3 ± 10.5 for TNF-α, 330.8 ± 81.5 for IL-6, and 345.5 ± 82.6 for IL-1β. The measured cytokine levels from the ΔmifR-infected mice, however, were still higher than those of the sham group. Furthermore, we also analyzed these three proinflammatory cytokine levels in the BALF (Fig. 4D to F). A similar tendency of these cytokines was observed in this study, i.e., the BALF from WT PAO1-infected mice contained higher levels of TNF-α, IL-6, and IL-1β compared to those from ΔmifR mutant-infected mice. The reduced levels of TNF-α, IL-6, and IL-1β in the ΔmifR-infected mice indicated that disruption of α-KG transport could lessen the symptoms of systemic inflammatory response syndrome in the early stage of P. aeruginosa-mediated pneumonia and sepsis.

FIG 4.

Analysis of cytokines in the serum and lung BALF of mice infected with wild-type P. aeruginosa PAO1 (WT PAO1), ΔmifR PAO1, or Sham. Serum inflammatory cytokines (A) TNF-α, (B) IL-6, and (C) IL-1β levels were measured in mice infected with WT PAO1, ΔmifR PAO1, or Sham via ELISA. The levels of serum TNF-α, IL-6, and IL-1β in mice infected with ΔmifR PAO1 were significantly lower than those measured in mice infected with WT PAO1. BALF inflammatory cytokines (D) TNF-α, (E) IL-6, and (F) IL-1β levels showed similar changes in this study. Data points represent mean values ± 25% quantile (box plot) and the minimum and largest observed values (lower and upper edge). n = 6; *, P < 0.05; **, P < 0.01; ***, P < 0.001. Magnification 200×, scale bar = 100 μm.

Damage and cell death were reduced in the lungs of mice infected with the ΔmifR mutant compared to mice infected with WT P. aeruginosa PAO1.

We investigated the expression of the apoptosis-related protein Bcl-2 and cleaved caspase-3 in the lungs of mice that had been infected with the ΔmifR mutant or WT P. aeruginosa PAO1 (Fig. 5A). Bcl-2 and cleaved caspase-3 are inhibitors and activators of apoptosis, respectively. Western blotting showed that the levels of Bcl-2 in the lung tissue of mice infected with WT P. aeruginosa PAO1 were 35% lower than mice infected with the ΔmifR mutant (Fig. 5B). The opposite was observed for levels of cleaved caspase-3 (Fig. 5C). Furthermore, apoptotic cells in the lungs were examined by terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) assay (Fig. 5D to F). The quantitative analysis of apoptotic cells in the lung tissues indicated that there are more apoptotic cells in the lungs from WT PAO1-infected mice compared to ΔmifR mutant-infected mice (Fig. 5G). Collectively, these findings indicated that deletion of the mifR gene of P. aeruginosa PAO1 weakened its ability to cause cell death during acute lung injury.

FIG 5.

Apoptosis in mice infected with wild-type P. aeruginosa PAO1 (WT PAO1), ΔmifR PAO1, or Sham. (A) Representative Western blots for Bcl-2 and cleaved caspase-3, (B) quantification of Bcl-2 expression, and (C) quantification of cleaved caspase-3 expression. The levels of apoptosis-related protein are decreased in the ΔmifR PAO1 group compared to the WT PAO1 group. Representative images showed TUNEL assay in the lung of mice infected with (D) wild-type P. aeruginosa PAO1 (WT PAO1), (E) ΔmifR PAO1, or (F) Sham. (G) The quantitative analysis of apoptotic cells in each HPF demonstrated fewer apoptotic cells in the lung of the ΔmifR PAO1 group compared to the WT PAO1 group. Data points represent mean values ± 25% quantile (box plot) and the minimum and largest observed values (lower and upper edge). n = 6; *, P < 0.05; **, P < 0.01. Pneu, pneumonia.

We next measured the expression of NLRP3 (an indicator of inflammation), as well as the proteins apoptosis-associated speck-like protein containing a caspase recruitment domain (ASC) and gasdermin D (GSDMD), both of which are related to apoptosis and pyroptosis (Fig. 6A). At 24 h postinfection, the expression levels of NLRP3 (Fig. 6B), ASC (Fig. 6C), and GSDMD (Fig. 6D) in mice infected with WT P. aeruginosa PAO1 were significantly higher than those measured in mice that had been infected with the ΔmifR mutant. The expression levels of NLRP3, ASC, and GSDMD in the lung tissue of ΔmifR-infected mice were still greater than that of the sham group.

FIG 6.

NLRP3 pathway and pyroptosis in mice infected wild-type P. aeruginosa PAO1 (WT PAO1), ΔmifR PAO1, or Sham. (A) Representative Western blots for NLRP3, GSDMD, and ASC, (B) quantification of NLRP3 expression, (C) quantification of GSDMD expression, and (D) quantification of ASC expression. Data points represent mean values ± 25% quantile (box plot) and the minimum and largest observed values (lower and upper edge). n = 6; *, P < 0.05; **, P < 0.01. Pneu, pneumonia.

Finally, immunofluorescence was used to evaluate the differences in the activation of the NLRP3 inflammasome in the lung tissues of mice infected with WT P. aeruginosa PAO1 (Fig. 7A), the ΔmifR mutant (Fig. 7B) or the sham group (Fig. 7C). Expression of NLRP3 was activated in mice infected with the WT P. aeruginosa PAO1 (Fig. 7A) and to a lesser extent in the ΔmifR mutant (Fig. 7B). The NLRP3 inflammasome was not activated in the sham group (Fig. 7C). Notably, there was ~2-fold greater number of NLRP3-positive cells in the lung tissues of mice infected with WT P. aeruginosa PAO1 versus those infected with the ΔmifR mutant (Fig. 7D).

FIG 7.

The expression of NLRP3 in immunofluorescence experiment in mice infected with (A) wild-type P. aeruginosa PAO1 (WT PAO1), (B) ΔmifR PAO1, or (C) Sham. The white arrows in the images indicate the immunofluorescence NLRP3 expression. (D) Statistical analysis of NLRP3 positive cell number was significantly decreased in the mice infected with ΔmifR PAO1 compared with the mice infected with WT PAO1. Data points represent mean values ± 25% quantile (box plot) and the minimum and largest observed values (lower and upper edge). n = 6; **, P < 0.01; ***, P < 0.001. Magnification 200×, scale bar = 100 μm. Pneu, pneumonia; HPF, high power field.

The ΔmifR and ΔPA5530 mutants displayed decreased cytotoxicity compared to WT P. aeruginosa PAO1.

The results of our animal study demonstrated that the mifR gene was required for the full virulence of P. aeruginosa PAO1. The EBP MifR has not been shown to directly regulate the expression of any known virulence factors (15). Instead, in response to extracellular C₅-dicarboxylates, such as α-KG, MifR activates expression of the PA5530 gene, which encodes for a dicarboxylate-transport protein (15, 17). Because MifR and its target PA5530 gene were essential for the utilization of α-KG in P. aeruginosa, it was speculated that PA5530 might also be necessary for the full virulence of this human pathogen. This was briefly addressed by measuring the cytotoxicity of both ΔmifR and ΔPA5530 mutants towards A549 epithelial cells (Fig. 8). Specifically, the amount of released cytoplasmic lactate dehydrogenase (LDH) from the A549 cells when challenged with the bacteria was used as a measure for cytotoxicity. As shown in Fig. 8A, at 6 h post addition of bacteria, levels of released cytoplasmic LDH were ~50% greater for A549 cells challenged with the WT strain compared to those challenged with the ΔmifR and ΔPA5530 mutants. The cytotoxicity of the ΔmifR and ΔPA5530 mutants was, therefore, lower than that of WT (Fig. 8B). Indeed, a significantly greater percentage of live A549 cells was observed for both the ΔmifR and ΔPA5530 treatments compared to that of WT (Fig. 8C).

FIG 8.

Cytotoxicity of wild-type P. aeruginosa PAO (WT PAO1), ΔmifR PAO1, and ΔPA5530 PAO1 on A549 lung epithelial cells. (A) The trends or changes in LDH activity post addition of bacteria. (B) The percentage of LDH secreted by the three different strains at 6 h post addition of bacteria. Levels of LDH secreted by WT were taken as 100%. (C) The percentage of active A549 cells remaining at 6 h post addition of bacteria. Data points represent mean values ± SEM for (A and B), mean values ± 25% quantile (box plot), and the minimum and largest observed values (lower and upper edge) for (C). n = 3; *, P < 0.05; **, P < 0.01. SEM, standard error of the mean.

To confirm that the reduced cytotoxicity of the ΔmifR and ΔPA5530 mutants was a result of their respective mutations, genetic complementation experiments were conducted. The ΔmifR and ΔPA5530 mutants were transformed with the expression plasmids pJRH08 and pBRL479, which carried the mifR and PA5530 gene, respectively (15). Mutants and WT P. aeruginosa PAO1 harboring empty expression plasmid (pBBR1MCS-5) served as controls. Genetic complementation was initially evaluated by growing the recombinant strains in minimal media in which either glucose or α-KG served as the sole carbon source. In the presence of glucose as the sole carbon source, the final cell densities of all recombinant strains were similar, i.e., optical density at 600 nm (OD₆₀₀) of ~ 1.8 (Fig. 9A). This was not the case when α-KG was the only available carbon source. As shown in Fig. 9A, the ΔmifR and ΔPA5530 mutants harboring empty pBBR1MCS-5 did not grow in a minimal medium with α-KG as the sole carbon source. Both recombinant strains yielded final OD₆₀₀ values <0.1 compared to the OD₆₀₀ of ~0.8 observed for the WT cells carrying pBBR1MCS-5. The growth of the ΔmifR and ΔPA5530 mutants on α-KG was restored to wild-type levels when the PA5530 gene was expressed from the lac promoter of pBBR1MCS-5 (Fig. 9A). Expression of mifR from pBBR1MCS-5 not only rescued the growth of the ΔmifR mutant but enhanced its growth on this dicarboxylate by ~2-fold, resulting in a final OD₆₀₀ of 1.5 (Fig. 9A). Expression of mifR from pBBR1MCS-5 did not recover the growth of the ΔPA5530 mutant on α-KG (Fig. 9A). These results demonstrated successful genetic complementation of the ΔmifR and ΔPA5530 mutants regarding α-KG utilization. Notably, they highlighted the essential nature of these genes in the consumption of this preferred dicarboxylate in P. aeruginosa.

FIG 9.

Plasmid-based expression of either the mifR or PA5530 gene in the ΔmifR mutant restored its growth on α-KG as the sole carbon source and cytotoxicity towards A549 lung epithelial cells. The ΔmifR and ΔPA5530 mutants of P. aeruginosa PAO1 were transformed with expression plasmids carrying no insert (pBBR1MCS-5), mifR (pJRH08), or PA5530 (pBRL479). Wild-type P. aeruginosa PAO1 (WT PAO1) carrying pBBR1MCS-5 served as the positive control. Recombinant strains were (A) grown in minimal media in which glucose or α-KG served as the sole carbon source or (B) tested for relatively cytotoxicity towards A549 lung epithelial cells in which WT PAO1/pBBR1MCS-5 was equated to 100. As expected, wild-type levels of growth on α-KG and cytotoxicity were observed for the ΔmifR mutant carrying pJRH08 or pBRL479 but not empty pBBR1MCS-5. Recovery of growth on α-KG and cytotoxicity was only observed for the ΔPA5530 mutant harboring pBRL479. Data points represent mean values ± SEM. n = 3; *, P < 0.05; **, P < 0.01. SEM, standard error of the mean.

We next evaluated the genetic complementation of the ΔmifR and ΔPA5530 mutants as it pertained to cytotoxicity or virulence. As shown in Fig. 9B, reduced cytotoxicity (~30% decrease) towards A549 cells was observed for ΔmifR and ΔPA5530 mutants harboring empty pBBR1MCS-5. In contrast, wild-type levels of cytotoxicity were observed for ΔmifR cells expressing mifR or PA5530 from pBBR1MCS-5, whereas full cytotoxicity of the ΔPA5530 mutant was only restored through expression of PA5530 and not mifR (Fig. 9B). These findings suggested that the reduced virulence of the ΔmifR and ΔPA5530 mutants was a direct result of deletions in the mifR and PA5530 genes, respectively, and not from indirect or secondary effects/mutations.

DISCUSSION

The EBP MifR contributed to the virulence of P. aeruginosa PAO1 in a mouse model of pneumonia and sepsis.

The mifR gene of P. aeruginosa PAO1 was required for full virulence in a mouse model of pneumonia and sepsis. It was also necessary for complete cytotoxicity in vitro. The EBP MifR is a response regulator that forms a two-component signal transduction system (TCS) with the histidine kinase MifS (15–17). The MifSR TCS detects or senses extracellular C₅-dicarboxylates, such as glutarate and α-KG, and in response, activates transcription of the PA5530 gene encoding a dicarboxylate-transport protein (15). Notably, micromolar concentrations of α-KG are sufficient to activate the expression of PA5530 in P. aeruginosa PAO1 (15). The MifSR TCS is specific towards C₅-dicarboxylates, meaning that other organic acids and/or dicarboxylates are not acceptable substrates for the induction of PA5530 via MifSR (15, 17). Both the MifSR TCS and its target PA5530 gene are essential for the utilization of α-KG in P. aeruginosa PAO1. α-KG and other TCA intermediates are preferred carbon sources for this pathogen. Apart from PA5530, no other genes, including those encoding virulence factors or proteins involved in α-KG metabolism, have been experimentally shown to be directly regulated by MifSR TCS. Similarly, to this date, no other transcriptional regulator(s) have been found to directly regulate the expression of PA5530.

The reduced virulence of the ΔmifR mutant (Fig. 1), as well as the decreased cytotoxicity observed for both the ΔmifR and ΔPA5530 mutants (Fig. 8), is believed to be the direct result of an inability to utilize α-KG from the eukaryotic-host environment. The death or lysis of eukaryotic cells during infection is expected to release α-KG, which is a TCA intermediate, key biosynthetic precursor, and signaling molecule (21). Metabolite profiling of human cell lines revealed that the intracellular concentration of α-KG varied among them, ranging from 70 to 900 μM (22). The release of α-KG to micromolar concentrations in the immediate environment is more than sufficient to induce the expression of PA5530 via the MifSR TCS in P. aeruginosa (15). Indeed, the induction or upregulated transcription of the PA5530 gene has been previously reported for P. aeruginosa when exposed to eukaryotic cells and/or the environment. For example, a microarray study found that following exposure of 4.0 h to human airway epithelial cells, the PA5530 gene of P. aeruginosa PAO1 was induced by >4-fold (18). The induction of PA5530, however, was not observed at 12 h postexposure, suggesting that α-KG utilization occurred in the early stages of interaction between P. aeruginosa and human airway epithelial cells. In a more recent study, in situ transcriptomics on CF sputum revealed that PA5530 (or PA14_72960) was upregulated in the P. aeruginosa population (19). Interestingly, transport of α-KG was observed to be necessary for the virulence of uropathogenic Escherichia coli (23). This suggests that the utilization of α-KG may have a broader, more influential role in the virulence of human pathogens than previously thought.

Despite the transcriptomic data surrounding PA5530, the presence of α-KG has not been detected in CF sputum (3). The MifSR TCS and its target PA5530 gene allow P. aeruginosa to readily consume extracellular α-KG at concentrations as low as 20 μM (15). Thus, the failure to detect α-KG in CF sputum could be a consequence of the rapid transport/metabolism of this dicarboxylate by P. aeruginosa. Although not detected in CF sputum, α-KG was observed in BALF from both healthy individuals and those suffering from acute respiratory distress syndrome (ARDS) (24). Notably, the mean concentration of α-KG in the BALF of patients suffering from ARDS was significantly greater (>8-fold) than that measured for the healthy group. Metabolomic studies aimed at measuring the levels of extracellular α-KG when P. aeruginosa is exposed to eukaryotic cells are warranted.

It should be noted that the ΔmifR mutant used in this study was indistinguishable from its parental PAO1 strain regarding motility, protease activity, biofilm formation, and the production of pyoverdine and pyocyanin (Fig. S1 and S2 in Supplemental File 1). The reduced virulence observed for the ΔmifR mutant, therefore, cannot be attributed to a general (nonspecific) defect in the production of any of these tested virulence factors. In addition, pyocyanin biosynthesis is strictly dependent on PQS and QS (25–27), so the WT levels of pyocyanin produced by the ΔmifR mutant is an indicator that neither of these signaling systems is disrupted in its background. Nonetheless, a more detailed investigation into the molecular mechanisms surrounding α-KG utilization in the lung or eukaryotic-cell environment is needed to confirm this hypothesis. Defining the roles of the histidine kinase MifS and the dicarboxylate-transporter PA5530 in the virulence of P. aeruginosa is a goal of future work.

The role of EatR, DdaR, and other EBPs in the virulence of P. aeruginosa.

Unlike MifR, the EBPs EatR and DdaR were not required for the virulence of P. aeruginosa PAO1 in a mouse model of pneumonia and sepsis (Fig. 1). The EBP EatR positively regulates the expression of the PA4022-eat-eutBC operon, encoding proteins that are essential in the catabolism of ethanolamine (9). Some data suggests that the animal intestine is a rich source of ethanolamine, both in its free form as well as being bound in the phospholipid phosphatidylethanolamine (10, 28). This might explain why the virulence of some intestinal human pathogens is affected by ethanolamine catabolism (10) while the virulence of P. aeruginosa PAO1 in acute lung injury is not.

Our findings also suggested that the catabolism of methylarginines regulated via DdaR was not crucial for the virulence of P. aeruginosa PAO1 in acute lung injury. Eukaryotic proteins are the only known major source of methylarginines, and interestingly, several pathogenic microorganisms possess enzymes called dimethylarginine dimethylaminohydrolases that hydrolyze methylarginines to yield citrulline and methylated amines as products (13). For P. aeruginosa, the liberated citrulline is a source of nitrogen and ATP (14). Neither citrulline nor methylated amines are a suitable carbon source for P. aeruginosa. The limited nutritional value of methylarginines might, therefore, account for the dispensable nature of DdaR in the virulence of P. aeruginosa PAO1. In other words, the more preferable nitrogen- and ATP-yielding substrates are available for P. aeruginosa during acute lung injury, thereby rendering methylarginine catabolism a nonlimiting factor.

There is a total of twenty-two EBPs encoded in the genome of P. aeruginosa PAO1 (29). The biological functions of twenty of the EBPs have either been experimentally described or strongly predicted (2, 4, 6, 9, 14, 15, 29–38). Some of these EBPs directly regulate pathogenic traits, such as motility, alginate, and type IV secretion systems (30, 33, 36, 38). However, the primary functions for thirteen of these EBPs are directed towards the metabolism of P. aeruginosa. The EBPs CbrB and NtrC serve as master regulators for the assimilation of carbon and nitrogen, respectively, so one might expect that deletion of either one would negatively impact the virulence of P. aeruginosa. While this is true for an ΔntrC mutant (8), transposon mutation of cbrB did not affect the virulence of P. aeruginosa PA14 in a mouse model (7). Deletion of the cbrB gene in P. aeruginosa PAO1 did reduce its growth rate in synthetic CF sputum, but its final cell densities were comparable to that of wild-type cells (39).

Of the remaining eleven EBPs associated with the metabolism of P. aeruginosa, only four of them have been investigated in terms of virulence. This includes the three described in the current study, EatR, DdaR, and MifR. The fourth one, GcsR, responds to intracellular glycine and activates the transcription of a glycine-cleavage system that aids in the metabolism of both glycine and serine (4, 5). Glycine is a key component of mucin and serves as a precursor in the biosynthesis of hydrogen cyanide in P. aeruginosa (40, 41). When C. elegans was challenged with a ΔgcsR mutant, the survival time of this worm decreased by >60% compared to those challenged with the parental PAO1 strain (5). The enhanced killing observed for the ΔgcsR mutant was hypothesized to be the result of elevated levels of hydrogen cyanide generated from the intracellular pool of nonmetabolized glycine.

Lastly, EBPs involved in the metabolism of P. aeruginosa that have not been investigated regarding its virulence include the C₄-dicarboxylate-transport regulator DctD, the glutamate-transport regulator AauR and the aromatic-amino acid catabolism regulator PhhR (2, 6, 35). These EBPs are vital for the utilization of these compounds, which have been detected in CF sputum and/or human airway fluids (24, 42). The EBP AcoR is predicted to regulate the catabolism of acetoin while the EBPs SfnR1 and SfnR2 regulate the catabolism of dimethylsulfide (29, 37). Some data suggests that acetoin might be an important interspecies metabolite between Staphylococcus aureus and P. aeruginosa, whereas dimethylsulfide produced by P. aeruginosa serves as a sulfur source for the fungal pathogen Aspergillus fumigatus (43, 44). Both S. aureus and Aspergillus fumigatus are frequently isolated as copathogens in P. aeruginosa-related infections in cystic fibrosis (45, 46). Whether the EBPs AcoR, SfnR1, and SfnR2 are crucial for such interactions in the diseased state is under investigation.

MATERIALS AND METHODS

Bacterial strain and mutants.

Bacteria used in the study consisted of wild-type (WT) P. aeruginosa PAO1 and the isogenic mutants ΔeatR, ΔddaR, ΔmifR, and ΔPA5530 (9, 14, 15, 47). The bacteria were grown at 37°C in Luria-Bertani broth (LB) or on a solid medium comprised of LB-agar. Peptone broth and Kings B medium were used for assaying the production of pyocyanin and pyoverdine, respectively, and were prepared as previously described (48).

Analysis of virulence factors.

All analyses were done in triplicate (n = 3) for each strain. For assessing motility, single colonies of each strain or mutant were stabbed into motility test agar plates (Neogen), which were then incubated at 25°C for 48 h. Protease activity was examined by growing each strain or mutant on skim milk agar (HiMedia) at 25°C for 48 h. To evaluate the production of pyocyanin or pyoverdine, bacteria were first grown in 2.0 mL of LB in a 16 × 125 mm culture tube at 37°C at 200 rpm for 18 h. The LB-grown seed cultures were used at 1% (v/v), i.e., 20 μL, to inoculate 2.0 mL of peptone broth (pyocyanin production) or Kings B medium (pyoverdine production) in a 16 × 125 mm culture tube, and the inoculated cultures were grown at 37°C at 200 rpm for 18 h. Cells were removed via centrifugation and subsequent passage through a 0.2 μm syringe filter. The absorbance at 690 or 405 nm was measured for the cell-free supernatants to determine the relative levels of pyocyanin and pyoverdine, respectively (48). For assessing biofilm formation, LB-grown seed cultures of the bacteria were used at 1% (v/v) to inoculate fresh LB. The inoculated LB cultures were aliquoted (100 μL) into a 96-well flat bottom polystyrene microplate (BD Falcon) that was subsequently incubated at 25°C for 48 h. Staining and quantification of the biofilms were done exactly as described (49).

Cytotoxicity assays.

Cytotoxicity assays were conducted in triplicate (n = 3) for each strain or mutant. Single colonies were inoculated in 5.0 mL of LB and incubated at 37°C at 250 rpm for 18 h. Bacteria were collected via centrifugation and subsequently diluted in MEM to yield ~ 4.0 × 10⁷ CFU mL⁻¹. These bacterial suspensions served as the inocula for the cytotoxicity assays.

The human lung adenocarcinoma epithelial cell line A549 was purchased from ATCC (no. CCL-185) and was cultured in 6-well plates in minimal essential medium (MEM) supplemented with 10% (v/v) fetal bovine serum and 1% (w/v) l-glutamine (Gibco by Life Technologies, Thermo Fisher Scientifc, USA). When the number of A549 cells in the 6-well plate reached ~90% of the plate area, they were treated with 3 mL of the bacterial suspensions. The bacterial-treated A549 cells were grown at 37°C and 5% CO₂ (the multiplicity of infection [MOI] at this time was 20). At 1.0, 3.0, and 6.0 h after the addition of bacteria, 50 μL of the medium from bacterial-treated A549 culture was collected for the analysis of LDH activity according to the protocol of Cytotox 96 nonradioactive cytotoxicity kit (Promega, lot no. 0000461938). Cytotoxicity percentage was equal to experimental LDH release (OD490)/maximum LDH release (OD₄₉₀) (50). In addition, the percentage of viable cells was counted with a cell counter.

Genetic complementation experiments.

The mifR and PA5530 genes were previously cloned under the lac promoter of the broad-host-range plasmid pBBR1MCS-5 to yield pJRH08 and pBRL479, respectively (15). The pJRH08, pBRL479, and empty pBBR1MCS-5 plasmids were introduced into the ΔmifR and ΔPA5530 mutants via electroporation (15), and wild-type P. aeruginosa PAO1 carrying pBBR1MCS-5 was used as the positive control. Recombinant strains in triplicate (n = 3) were subjected to two sets of genetic complementation experiments. First, using a previously described microplate assay (14), recombinant strains were grown at 37°C with shaking in minimal medium (22 mM KH₂PO₄, 42 mM Na₂HPO₄, 8.6 mM NaCl, 18.7 mM NH₄Cl, 1.0 mM MgSO₄, and 5.0 μM FeSO₄, pH 7.0) supplemented with 30 mg L⁻¹ gentamicin and either 20 mM glucose or α-KG as the sole carbon source (15). At 24 h postinoculation, the absorbance at 600 nm or optical density (OD₆₀₀) was measured for each bacterial culture. Second, recombinant strains were subjected to cytotoxicity assays performed exactly as described earlier except that the media used for culturing the recombinant strains was supplemented with 30 mg L⁻¹ gentamicin.

Mice and maintaining conditions.

We used wild-type FVB/N, which was originally purchased from the Jackson Laboratory (Bar Harbor, ME). All mice were kept pathogen-free in the animal core facility at the State University of New York (SUNY) Upstate Medical University with a standard12 h/12 h light/dark time cycle, the stable controlled temperature at 24°C, and regular mice feed and water. Both male and female WT mice (8 to 12 weeks old, ~25 g body weight [bw]) were used for the study (20, 51). Mice were infected with wild-type P. aeruginosa PAO1 (WT PAO1), or each P. aeruginosa PAO1 mutant (1 × 10⁶ CFU/mouse), and Sham mice were treated with the same volume of saline to replace bacterial solution (52). All relevant animal experiments are in strict compliance with the National Institutes of Health and the Animal Research: Reporting of In Vivo Experiments (ARRIVE) guidelines and have been approved by the State University of New York Upstate Medical University Institutional Animal Care and Use Committee (IACUC number 380).

P. aeruginosa-induced pneumonia and sepsis model.

This analysis included wild-type P. aeruginosa PAO1 and the ΔeatR, ΔddaR, and ΔmifR mutants. Bacteria were grown in LB at 37°C at 200 rpm to late-log growth at which time 1.0 mL was collected via centrifugation and subsequently suspended in cold, buffered saline to yield an absorbance of 0.6 at 600 nm. The bacteria were diluted an additional 40-fold in buffered saline, and 50 μL of the diluted bacterial suspension was injected into the mouse. The procedures of anesthesia, surgery and intratracheal inoculation of bacteria were performed as described previously (53).

Collection and analysis of mice bronchoalveolar lavage fluid.

Mice were anesthetized by intraperitoneal injection of ketamine/thiazide (90 mg kg⁻¹ ketamine, 10 mg kg⁻¹ thiazide) and then sacrificed by bleeding. A 1.0 mL syringe was used to collect blood from the inferior vena cava of the mouse. The collected blood was centrifuged at 3000 rpm for 10 min, and the supernatant was subjected to an enzyme-linked immunosorbent assay (ELISA; see below). Lung tissue and BALF were collected as previously described (20, 51). To determine the number of bacteria present in the BALF from each mouse, the collected BALF (100 μL) was diluted 1000-fold, and in triplicate (n = 3),100 μL of this dilution was plated on LB-agar. Following an incubation period of 18 h at 37°C, the CFU was counted, averaged, and reported as total CFU per lung (54). The quantification of inflammatory cells present in mice BALF has been described (51). Briefly, the total number of cells present in each BALF was determined with a hemocytometer. Next, the cells in each BALF were collected by centrifugation, suspended in 1.0 mL of cold buffered saline, and finally mounted to a slide by a cytospin centrifuge (Hettich ROTOFIX32 A) at 1,000 rpm for 3 min. The mounted or fixed slides were stained with HEMA 3 Cell Staining kit (Fischer Scientific catalog no. 122911), and the macrophages/monocytes and polymorphonuclear neutrophils (PMN) were identified and quantified with a Nikon Eclipse TE2000-U research microscope (Nikon, Melville, NY).

Lung histopathological analysis.

The lung tissues of the mice were fixed with 10% (w/v) formalin for 24 h and embedded in the paraffin block. About 5-μm sections were stained with hematoxylin and eosin (H&E) as described previously (55). Slides were examined under a Nikon Eclipse TE2000-U research microscope (Nikon, Melville, NY) and digital photographs were taken, acute lung injury was evaluated as described in our previous publication (55).

Terminal deoxynucleotidyl transferase dUTP nick end labeling assay.

TUNEL analysis was performed as described in our recent publication (51). In brief, formalin-fixed, paraffin-embedded, 5-μm slides of lung tissue were incubated at 60°C for 1 h, then deparaffinized. Slides were fixed with 4% paraformaldehyde for 15 min and treated with proteinase K for permeabilization. After that, slides were treated and incubated with a Click-iT Plus TUNEL reaction cocktail at 37°C for 30 min. Slides were mounted by fluoroshield mounting medium with DAPI (ab104139, Abcam, Cambridge, MA) to visualize cell nuclei. Apoptotic cells were quantified at each high-power field (×200 magnification) by counting the number of TUNEL-positive cells in 10 randomly chosen fields.

Western blotting.

Western blotting was performed as described in our previous publication (55). In brief, frozen lung tissue was homogenized in radioimmunoprecipitation assay (RIPA) buffer containing a mixture of protease and phosphatase inhibitors. The homogenized lung samples were centrifuged, and the resulting supernatants were assayed for total protein via a micro-BCA assay kit (Thermo Scientific, Rockford, IL). Next, each sample (30 μg of total protein) was dissolved by reduction, electrophoresed on a 12% SDS-polyacrylamide gel, and then transferred to a PVDF membrane (Bio-Rad, Hercules, CA, USA). The blot was blocked in Tris-buffered saline/0.1% Tween20 (TBST) containing 5% skim milk for 1.0 h at room temperature to remove the nonspecific binding and subsequently incubated with the primary antibodies listed below at 4°C overnight with shaking. Subsequently, the blot was incubated with an enzyme-labeled secondary antibody (Bio-Rad, Hercules, CA) at room temperature for 1 hour, detected with Pierce ECL Western blotting substrate/detection reagent (Thermo Scientific, Rockford, IL), and exposed with X-ray film (Pierce Biochemicals, FL). Image J software version 1.48 (Wayne Rasband, NIH, Bethesda, MA) was used to scan and quantify the film. The main antibody used for western blotting, ASC (67824s Santa Cruz Biotechnology, Dallas, TX), BCL-2 (3498s Santa Cruz Biotechnology, Dallas, TX), Cleaved caspase-3 (9661s, Santa Cruz Biotechnology, Dallas, TX), NLRP3 (ma5-20838, Thermo Fisher Scientific, Rockford, Illinois), GSDMD (ab209845, Abcam, Cambridge, MA). β-actin antibody (ab16039, Abcam, Cambridge, MA) was used as an internal control. In some experiments, the blot was stripped to remove the antibody, incubated in stripping buffer (Thermo Fisher Scientific, Rockford, IL) at 50°C for 30 minutes, and then tested again with a second primary antibody. Sham mouse lung tissue was used as a positive control.

Cytokine analysis by ELISA.

Cytokines were measured by ELISA as described previously (51). Serum and BALF IL-6, IL-1β, and TNF-α levels of three groups of mice were determined using a mouse ELISA kit according to the manufacturer's instructions (Thermo Fisher Scientific, Rockford, IL).

Immunofluorescence analysis.

As previously mentioned (20), paraffin-embedded lung sections were performed with immunofluorescence staining. In short, the slides were incubated at 60°C for 1 hour, then deparaffinized with xylene, and then subjected to a graded series of ethanol treatments. The slides were then recovered with citrate buffer antigen (1×, pH 6.0, Sigma-Aldrich, St. Louis, MO) at boiling point for 10 minutes, and then immersed in 0.02% Triton X-100 (T9284, Sigma-Aldrich, MO) at the boiling point), which lasted 45 minutes. After blocking with 10% donkey serum (ab7475, Abcam, Cambridge, MA) for 1 hour at room temperature, the slides were immunostained with the first antibodies listed below at 4°C overnight, and then with donkey anti-rabbit IgG Alexa 488 (ab150073, at room temperature (protected from light), the secondary antibody was conjugated to the secondary antibody for 1 h. DAPI (ab104139, Abcam, Cambridge, MA) slides were used to visualize the nucleus. In a fluorescence microscope (Nikon, the image was viewed and digitized under Melville, NY). The antibody used for immunofluorescence staining was NLRP3 (PA5-20838, Thermo Fisher Scientific, Rockford, IL).

Statistical analysis.

SigmaStat Software (version 3.5) was used for statistical analysis. Significant differences between samples and/or groups were determined through one-way analysis of variance (ANOVA) or t test in Fig. 2 to 9. The Kaplan-Meier method was used for analyzing the survival times of infected mice in Fig. 1. For all comparisons, P < 0.05 was considered statistically significant.

Ethics statement.

All animal experiments and protocols were conducted in accordance with the guidelines of the Institutional Animal Care and Use Committee (IACUC), SUNY Upstate Medical University, those of the National Institutes of Health guidelines on the use of laboratory animals, and those of ‘ARRIVE’ on the use of laboratory animals.