The putative bacterial oxygen sensor Pseudomonas prolyl hydroxylase (PPHD) suppresses antibiotic resistance and pathogenicity in Pseudomonas aeruginosa

Pseudomonas aeruginosa is an extracellular opportunistic bacterial pathogen commonly associated with infectious complications in susceptible individuals, such as those with underlying diseases including HIV/AIDS and cystic fibrosis. Antibiotic resistance in multiple strains of P. aeruginosa is a rapidly developing clinical problem. We have previously demonstrated that the oxygen levels at the site of P. aeruginosa infection can strongly influence virulence and antibiotic resistance in this pathogen, although the oxygen-sensing and -signaling mechanisms underpinning this response have remained unknown. In this study, we investigated the potential role of the putative oxygen sensor Pseudomonas prolyl hydroxylase (PPHD) in the control of virulence and antibiotic resistance in P. aeruginosa. We found that a P. aeruginosa strain lacking PPHD (PAO310) exhibits increased virulence associated with increased bacterial motility. Furthermore, PPHD-deficient P. aeruginosa displayed enhanced antibiotic resistance against tetracycline through increased expression of the xenobiotic transporters mexEF-oprN and MexXY. Of note, the effect of the PPHD knockout on antibiotic resistance was phenocopied in bacteria exposed to atmospheric hypoxia. We conclude that PPHD is a putative bacterial oxygen sensor that may link microenvironmental oxygen levels to virulence and antibiotic resistance in P. aeruginosa.

bacterial infection is rapidly diminishing because many of the causative pathogens develop antibiotic resistance.
Pseudomonas aeruginosa, a motile, Gram-negative facultative anaerobe, is an opportunistic pathogen that is responsible for 17% of cases of nosocomial pneumonia and is a primary cause of morbidity and mortality in cystic fibrosis patients (2,3). P. aeruginosa infection is also frequently associated with infections in individuals with HIV/AIDS, urinary tract infections, ventilator-associated lung infection, endocarditis, meningitis, ocular infection, ear infection, and wound/burn/skin infections (2). P. aeruginosa is an extracellular pathogen that produces toxins (e.g. exotoxin A, which causes direct tissue damage) and proteases (e.g. alkaline protease) that degrade neutrophil-derived elastin and consequently help the bacterium to avoid host immunity. P. aeruginosa expresses a type III secretion system that allows the direct injection of bacterial toxins into host cells, siderophores that promote the chelation of iron and virulence factors such as pyocyanin and pyoverdine, which support bacterial metabolism and the production of exotoxins, respectively. Together, these virulence mechanisms promote bacterial colonization of tissues and the development of infectious disease. A key mechanism underpinning the development of virulence in P. aeruginosa is quorum-sensing, which refers to the intercellular communication between individual bacteria in the niche of the infected tissue (4). Several important factors that mediate the dialog between bacteria during quorum sensing are the products of bacterial metabolism creating another link between the bacterial metabolic strategy and the capacity of the pathogen for virulence.
Antimicrobial resistance is a common and developing problem in P. aeruginosa infection caused in part by high expression levels of a number of xenobiotic efflux pumps (5,6). Typically, when it colonizes a host, P. aeruginosa grows in a biofilm that provides a physical barrier to antimicrobial access. In summary, P. aeruginosa is a formidable opportunistic pathogen that is armed with an array of features that support its virulence in human infection.
Hypoxia is an environmental feature that is frequently encountered at sites of infection (5). For example, the cystic fibrosis lung, commonly infected with P. aeruginosa, has been shown to be hypoxic (7). Furthermore, Mycobacterium tuberculosis was shown to reside in a hypoxic environment within tuberculous granulomas in rabbits (8). Biopsies from patients with skin infection showed an elevated level of hypoxia-inducible factor (HIF) 1␣, a marker of tissue hypoxia (9). We previously demonstrated the impact of hypoxia on P. aeruginosa antibiotic resistance, virulence, and infection (6,10,11). However, the nature of the oxygen-sensing pathways that determine these responses remains unknown and is the topic of the current study.
Prolyl hydroxylases (PHDs) 2 are oxygen sensors in metazoans. They are Fe(II)-and 2-oxoglutarate-dependent dioxygenases whose activity depends on molecular oxygen. When sufficient oxygen is available (normoxia), PHDs hydroxylate the hypoxia-inducible factor (HIF) ␣ subunit. The von Hippel-Lindau tumor suppressor protein polyubiquitinates the hydroxylated HIF␣ and thus marks it for proteosomal degradation. Conversely, when oxygen is limited, HIF␣ accumulates and translocates to the nucleus where it transcriptionally activates genes involved in erythropoiesis and angiogenesis (12).
PHDs are therefore primary oxygen sensors in metazoans (13,14). Recently a homolog to the human oxygen sensing enzyme PHD2 was identified in P. aeruginosa and termed Pseudomonas prolyl hydroxylase (PPHD) (13). Subsequently, another bacterial PHD enzyme was discovered in Bacillus anthracis (15). PPHD has a close structural homology to human PHD2 in that it contains an active 2OG oxygenase domain as demonstrated by biochemical and structural analysis (13). Furthermore, the elongation factor EF-Tu has been identified as a substrate for PPHD. However, cross-reactivity with substrates does not occur (i.e. EF-Tu is not a substrate for PHD2, and the oxygen-dependent degradation domain of HIF-1␣ is not a substrate for PPHD). Because of the role of PHDs in mammalian oxygen sensing, it was hypothesized that PPHD might play a role in oxygen sensing in P. aeruginosa (13,16). More recently we and others have shown that PPHD influences P. aeruginosa virulence; however, the mechanisms involved remain to be defined (11,16). In this study, we investigated the role of PPHD as a possible prokaryotic oxygen sensor linking microenvironmental hypoxia to virulence and antibiotic resistance in P. aeruginosa.

PPHD suppresses virulence in P. aeruginosa
We have previously demonstrated that microenvironmental hypoxia alters virulence in P. aeruginosa (10,11). To determine whether the putative oxygen sensor PPHD may play a role in controlling virulence and antibiotic resistance, we compared WT (PAO1) with a P. aeruginosa strain deficient in PPHD (PAO310). We first confirmed the absence of PPHD mRNA expression in PAO310 by PCR (Fig. 1A). We next investigated the impact of PPHD knockout on bacterial virulence using Galleria mellonella larvae as a model of host infection (17). PAO310 was more virulent in G. mellonella than PAO1 as demonstrated both qualitatively and quantitatively (Fig. 1, B and C, respectively and Fig. S2). A representative image of G. mellonella injected with P. aeruginosa at 24 h is shown in which light-er-colored larvae are viable and darker-colored larvae were melanized, which immediately precedes death. The data in Fig.  1C are presented as LD 50 (the number of bacteria required to kill 50% of the larvae). In summary, PPHD knockout results in increased lethality in G. mellonella, indicating that PPHD suppresses bacterial virulence in P. aeruginosa.
We next investigated the influence of PPHD expression on key determinants of virulence in P. aeruginosa. We first compared bacterial motility and quantified biofilm formation in PAO1 and PAO310 strains. Consistent with the increased virulence observed in Fig. 1, motility of P. aeruginosa (both swimming ( Fig. 2A) and swarming (Fig. 2B)) was increased when PPHD is not present. Furthermore, biofilm mass was also increased in PAO310 (Fig. 2C). Taken together, these data support a role for PPHD in suppressing key determinants of P. aeruginosa virulence.

PPHD influences antibiotic susceptibility of tetracycline antibiotics
We previously demonstrated that microenvironmental hypoxia plays a determining role in antibiotic resistance in P. aeruginosa (6). Therefore, we next investigated the impact of PPHD knockout on antibiotic resistance. Minimal inhibitory concentration of antibiotics from a broad range of antibiotic classes were tested in PAO1 and PAO310 strains of P. aeruginosa. The minimal inhibitory concentration (MIC) values for tetracycline antibiotics (tetracycline, doxycycline, minocycline,

PPHD regulates virulence of P. aeruginosa
and tigecycline) were selectively increased in P. aeruginosa lacking the PPHD gene, whereas susceptibility of other classes of antibiotics was not influenced (Table 1).
Exposing PAO1 to tetracycline (512 g/ml) prior to injection into larvae increased its LD 50 by 2972-fold (Fig. 3). In contrast, for PAO310, tetracycline pretreatment did not affect its LD 50 , further suggesting that the P. aeruginosa lacking PPHD was more resistant to the antibiotic. Taken together, these data indicate that PPHD selectively suppresses tetracycline antibiotic resistance in P. aeruginosa.

Efflux pump mexEF-OprN is up-regulated in the absence of PPHD
We next investigated the possible mechanisms underpinning the effects of PPHD knockout on antibiotic resistance. A key determinant of antibiotic resistance in P. aeruginosa is the level of antibiotic efflux via RND efflux pumps (18,19). Indeed, we previously demonstrated that the expression of these pumps is increased in hypoxia, thereby contributing to hypoxia-induced antibiotic resistance (6). We next investigated whether RND efflux pumps contribute to increased tetracycline resistance in PAO310. To do this we determined the MIC of tetracycline in the presence of an efflux pump inhibitor (Phe-Arg-␤-naphtylamide dihydrochloride). Blocking the RND efflux pumps made PAO310 more susceptible to tetracycline (MIC was lowered from 256 to 64 g/ml), indicating that the absence of PPHD contributes to antibiotic resistance by increasing tetracycline extrusion via efflux pumps (Fig. 4A). We quantified the expression of the linker protein gene of four RND efflux pumps. MexE (part of mexEF-OprN) was increased in PAO310 (PPHD knockout mutant), whereas mexA and mexC were unchanged ( Fig. 4B and Fig. S1). Of note, the induction of mexX with a subinhibitory concentration of tetracycline in PAO1 (WT P. aeruginosa) was lost in PAO310 (PPHD knockout strain) (Fig. S1). These data suggest that the absence of PPHD increases mexEF-OprN expression in P. aeruginosa.

Hypoxia increases resistance to tetracycline antibiotics in P. aeruginosa
PPHD is a prolyl hydroxylase, homolog to human PHD2. PHD2 hydroxylates the target HIF-1␣ protein in metazoan cells in an oxygen-dependent manner. The hydroxylation activity of PHD2 can therefore be inhibited by low oxygen concentrations (hypoxia) (20). To investigate whether hypoxia mimics the effects of PPHD knockout on tetracycline antibiotic susceptibility, we tested whether exposure to hypoxia (1% oxygen) altered antibiotic resistance in a similar way to PPHD knockout. Similar to PPHD knockout (albeit to a lesser degree), hypoxia increased tetracycline, doxycycline, and minocycline MICs in PAO1 by ϳ2-fold (tetracycline, 32-64 g/ml; doxycycline and minocycline, 16 -32 g/ml), respectively (Fig. 5A). To test the potential clinical relevance of our observations, we also investigated tetracycline susceptibility in clinical P. aeruginosa isolates (derived from patient infections) under hypoxic conditions. Consistent with the effects observed in PAO1, two of three acute clinical strains demonstrated a 2-fold increase in MIC in hypoxia (Fig. 5B). In summary, exposure to hypoxia phenocopiestheimpactofPPHDknockoutontetracyclineresistance in P. aeruginosa, thereby supporting the potential role for PPHD as a bacterial oxygen sensor.

Discussion
Hypoxia is frequently a prominent microenvironmental feature at sites of infection. We have previously shown that exposure of the opportunistic pathogen P. aeruginosa to a hypoxic microenvironment is a key determinant of pathogen virulence and antibiotic resistance. The mechanism whereby hypoxia mediates its influence on these parameters of pathogen behavior is the topic of this study. Developing our understanding of how the microenvironment at the site of infection affects pathogen virulence and antibiotic resistance will help identify new avenues of therapeutic intervention in P. aeruginosa infection.
Recently the prolyl hydroxylase PPHD was identified in P. aeruginosa and hypothesized to play a role in oxygen sensing (13). PPHD is a structural homolog of human PHD2, which is an important regulator of the hypoxic response in mammals (12). PPHD was shown to hydroxylate the prokaryotic elongation factor EF-Tu on a proline residue (13).
For this study we used the PPHD knockout mutant PAO310 to determine its influence on virulence and antibiotic resistance. PAO310 was more virulent than the WT (PAO1) in the G. mellonella model (Fig. 1). This is consistent with a previous

PPHD regulates virulence of P. aeruginosa
study showing higher virulence of PAO310 in a mouse pneumonia model (16). Although this result may appear contradictory to the previous observation that hypoxia reduces virulence in P. aeruginosa, this is not necessarily the case. If one assumes that PPHD acts in a similar manner to the PHDs in eukaryotic cells (i.e. to repress the effector (HIF) in normoxia), then knocking out the PPHD should indeed mimic hypoxia. However, if in prokaryotes, PPHD activates an antivirulence effector in normoxia, then knocking out PPHD would have the opposite effect to hypoxia (i.e. to increase pathogenicity). Although this is conjuncture at this stage, the full answer to this complex question will require knowledge of the biological roles of PPHD targets in P. aeruginosa and how their hydroxylation affects pathogenicity and antibiotic resistance. Our current studies aim to identify the target(s) for PPHD in P. aeruginosa, and we hope in the future to be able to answer this vitally important question.
Resistance to tetracycline antibiotics was selectively increased in PAO310. This was associated with a significant increase in mexE mRNA (Table 1 and Figs. 3 and 4). MexE is part of the RND efflux pump mexEF-oprN. An increase in tetracycline resistance caused by overexpression of mexEF-oprN was reported before (21), although it should be noted that whether tetracycline is a major substrate of mexEF-oprN remains controversial (22,23). We conclude that PPHD influences efflux pump expression and thus changes antibiotic susceptibility of P. aeruginosa for tetracycline antibiotics.

PPHD regulates virulence of P. aeruginosa
PHD2, the mammalian homolog of PPHD is a key oxygen sensor linked to the HIF pathway (24). We tested whether PPHD inhibition by hypoxia in WT P. aeruginosa (PAO1) has influence on antibiotic susceptibility toward tetracycline antibiotics in a similar way to PPHD knockout. Resistance to tetracycline antibiotics was 2-fold higher in PAO1 at 1% O 2 when compared with 21% O 2 (Fig. 5). This is similar to the effect seen when PPHD is absent, albeit to a lesser extent.
In conclusion, we provide evidence that PPHD activity decreases virulence of P. aeruginosa and increases antibiotic susceptibility to tetracycline antibiotics by altering the expression of the RND efflux pumps mexEF-oprN and MexXY. Therefore, P. aeruginosa PPHD may represent a new therapeutic target in reducing virulence and antibiotic resistance in P. aeruginosa infection.
Because PPHD mutation correlates with some of the effects we have reported hypoxia to elicit in P. aeruginosa, these data raise the intriguing possibility that PPHD may provide a molecular link between tissue oxygen levels and bacterial pathogenicity in infectious disease. However, further evidence is required to assess both the oxygen-sensing activity of PPHD and its molecular targets in linking microenvironmental oxygen levels and bacterial pathogenesis and antibiotic resistance.

Determination of virulence using the in vivo G. mellonella model
Sixth-instar larvae of G. mellonella (Livefoods direct) were used for the in vivo virulence determination. Bacteria were grown to an A 600 of 0.6 and diluted to A 600 of 0.1 in PBS (Fisher Chemical). Serial dilutions up to 10 Ϫ7 in PBS were prepared and plated on agar plates in triplicates to determine the bioburden. 20 l of each dilution was injected into the hemocoel through the last pro leg (BD microfine U100 insulin syringe; Becton, Dickinson and Company). For each dilution 10 larvae were used and incubated in Petri dishes (Greiner Bio-One) on filter paper (Fisher). 10 larvae were injected with PBS as control. After 24-h incubation, larval death was assessed by the lack of movement upon stimulation. At least three independent experiments were conducted. The data are presented as LD 50 values.

Bacterial Strains
PAO310 was extracted from a saturating library of sequencedefined transposon insertion mutants. This approach generated over 30,000 defined P. aeruginosa mutants, of which PAO310 was one (28). To our knowledge, only PPHD is different between PAO310 and PAO1.

Bacterial culture conditions
P. aeruginosa was grown in Luria broth (Sigma) or on tryptic soy agar (Sigma). For hypoxic culture, broth was preincubated at 1% oxygen in a hypoxia chamber (Coy Laboratory) prior to inoculation.

Motility
Motility was assessed by investigating swimming and swarming. Motility plates were inoculated with a sterile pipette tip (1-10 l; Star-lab) in the center. After 24 h of incubation at 37°C, the zone of motility was measured.

Biofilm
Overnight cultures were diluted to A 600 of 0.05. Per condition 1 ml of diluted bacteria suspension was incubated in duplicate in a 12-well plate (Greiner Bio-One) for 24 h. For quantification the biofilm was washed two times with 1 ml of PBS, stained with 0.5 ml of 0.1% crystal violet solution (Sigma) for 15 min at room temperature, and washed again three times with 1 ml of PBS. Crystal violet stain was eluted from biofilm bacteria with 1 ml of ethanol (96%, v/v, Sigma), and absorbance at 595 nm was measured with the CLARIO star spectrometer (BMG

PPHD regulates virulence of P. aeruginosa
Labtech). Dishes with medium only were treated the same way and served as negative control.

Antibiotic susceptibility testing
MIC of P. aeruginosa was determined with several methods. For MIC testing with the VITEK 2 system (Bio Merieux), AST-N352 cards were used according to the manufacturer's instruction. E-test (Bio Merieux) were used according to the manufacturer's instructions. Micro broth dilutions were performed using Sensititre GNX2F susceptibility plate (Thermo Scientific) and with freshly prepared plates using tetracycline hydrochloride (Sigma), doxycycline hydrochloride (Sigma), minocycline hydrochloride (Sigma), and tigecycline hydrate (Sigma). Therefore, serial dilutions of the respective antibiotic in cation adjusted Mueller Hinton broth (Becton, Dickinson and Company) double-concentrated were prepared, and 50 l were transferred into a 96-well plate (Greiner Bio-One). P. aeruginosa from a freshly grown tryptic soy agar plate were used to prepare a suspension with the same turbidity as a 0.5 McFarland standard. 100 l of this suspension was added to 11.5 ml of Mueller Hinton broth (Becton, Dickinson and Company). 50 l were added to the wells with the double-concentrated antibiotics and incubated for 18 -24 h at 37°C. Broth without antibiotic served as growth control. To determine the influence of RND efflux pumps on antibiotic susceptibility, Phe-Arg-␤-naphtylamide dihydrochloride (20 g/ml, Sigma) was added. Furthermore, antibiotic susceptibility testing was performed at 1% oxygen.

DNA and RNA extraction
DNA was extracted from overnight cultures with the Wizard genomic DNA purification kit (Promega) according to the manufacturer's instructions. For RNA extraction overnight cultures of P. aeruginosa were diluted to an A 600 of 0.001 and grown statically for 24 h. The bacteria were harvested and processed as described elsewhere (25). 1 g of RNA was reversed transcribed to cDNA with Superscript II (Invitrogen).

Polymerase chain reaction
To test for the presence of the PPHD gene, PPHD was amplified of genomic DNA with the following primer: forward, 5Ј-TGA AAA ACG GCC AGT AGC GCG CAT TGA TAC TCC TT-3Ј, and reverse, 5Ј-CAG GAA ACA GCT ATG ACC CAC GAT CAA GGT CTG GGG TC-3Ј. 16S rRNA was amplified as control. Primers for 16S were published previously (26).

Statistical analysis
The data are presented as means Ϯ S.E. for parametric data and as median for nonparametric data for at least three independent experiments. Statistical analysis was carried out with two-tailed unpaired Student's t test, one-way analysis of variance with Tukey's multiple comparison test, the Mann Whitney test or the Kruskal-Wallis method with Dunne's multiple comparison test. p values of Ͻ0.05 were considered statistically significant.