The pathogenic bacterium Pseudomonas aeruginosa uses acyl-HSL quorum-sensing signals to regulate genes controlling virulence and biofilm formation. We found that paraoxonase 1 (PON1), a mammalian lactonase with an unknown natural substrate, hydrolyzed the P. aeruginosa acyl-HSL 3OC12-HSL. In in vitro assays, mouse serum-PON1 was required and sufficient to degrade 3OC12-HSL. Furthermore, PON2 and PON3 also degraded 3OC12-HSL effectively. Serum-PON1 prevented P. aeruginosa quorum-sensing and biofilm formation in vitro by inactivating the quorum-sensing signal. Although 3OC12-HSL production by P. aeruginosa was important for virulence in a mouse sepsis model, Pon1-knock-out mice were paradoxically protected. These mice showed increased levels of PON2 and PON3 mRNA in epithelial tissues suggesting a possible compensatory mechanism. Thus, paraoxonase interruption of bacterial communication represents a novel mechanism to modulate quorum-sensing by bacteria. The consequences for host immunity are yet to be determined.
The pathogenic bacterium Pseudomonas aeruginosa uses acyl-homoserine lactone (acyl-HSL) quorum-sensing signals to regulate the expression of virulence genes implicated in infectivity, tissue damage, and the generation of a host inflammatory response [1–5]. Two hierarchically regulated acyl-HSL quorum-sensing molecules are produced and recognized: N-(3-oxododecanoyl)-l-homoserine lactone (3OC12-HSL), and N-butanoyl-l-homoserine lactone (C4-HSL) [6–8]. These same quorum-sensing signals also coordinate the establishment of biofilms [1–3,9]. One well-described example of the importance of quorum-sensing in establishing P. aeruginosa infection is found in the controlled expression of the rhamnolipid group of surfactants. Without coordinated expression of these surfactants by quorum-sensing, P. aeruginosa does not form structured, well-differentiated biofilms [10,11].
Other organisms have evolved to disrupt quorum-sensing signaling to compete with bacteria that use quorum-sensing to organize their behavior; some environmental strains of bacteria secrete lactonases or acylases to degrade acyl-HSL signals [12–16]. Bacteria expressing these anti-acyl-HSL factors have a competitive edge over their quorum-sensing neighbors [16,17]. Broad protection by anti-acyl-HSL factors has been demonstrated in transgenic experiments. AiiA (an acyl-HSL lactonase from a soil-dwelling species of Bacillus) was transgenically expressed in plants and protected them from soft-rot infection by Erwinia carotovora, a quorum-sensing plant pathogen . Previously we reported that some acyl-HSL quorum-sensing signals lose biological activity when exposed to mammalian epithelial cells . Therefore, we hypothesized that mammals, like bacteria, also degrade acyl-HSL quorum-sensing signals. Furthermore, we speculated that this innate activity would confer on the host a competitive advantage over pathogenic bacteria by reducing infection and biofilm formation.
The paraoxonases are a family of mammalian lactone hydrolases with high sequence similarity but distinct substrate specificities and expression patterns [20–22]. PON1 is expressed primarily in the liver and secreted into the serum where it associates with high density lipoprotein (HDL) particles [22–24]. PON3 is also expressed in liver and serum but at a lower level than PON1 [22,25], and PON2 is cell-associated and expressed in several tissues . PON1 protects against organophosphate poisoning  and has anti-oxidative properties ; it is also associated with protection from atherosclerosis, presumably through its ability to hydrolyze oxidized phospholipids and protect serum lipids from oxidative modification [23,28]. Acyl-HSL molecules are substrates for the human paraoxonases (PON), PON1, PON2, and PON3 . Here we report that human and mouse PON1 degrades 3OC12-HSL by hydrolyzing its lactone ring, thereby blocking P. aeruginosa quorum-sensing signaling. (A preliminary report of these findings has been presented ).
Materials and methods
Synthetic 3OC12-HSL was reconstituted in phosphate buffered saline containing magnesium and calcium (PBS) to 10 μM and incubated for 1 h on monolayers of A549 airway epithelial cells or in the absence of cells. 3OC12-HSL was extracted from the medium in acidified ethyl acetate, dried under sterile nitrogen gas, and reconstituted in deuterated chloroform. Proton-nuclear magnetic resonance (NMR) , high-pressure liquid chromatography (HPLC) column, and electrospray mass spectrometry (MS) methods are described in supplemental information.
Blood was collected from euthanized mice through cardiac puncture and serum was separated from red blood cells by centrifugation. Synthetic 3OC12-HSL (10 μM) was prepared in reaction buffer (100 mM HEPES, pH 7.4, 1 mM CaCl2) and exposed to PBS or whole serum. EDTA (25 mM), oxindole (1 mM), isatin (1 mM), or normal saline was added to serum to evaluate inhibition of serum activity against 3OC12-HSL. 1% solutions of human serum, wild-type mouse serum, or PON1 knock-out (KO) serum in reaction buffer were compared for 3OC12-HSL degradation activity. All reactions were performed for 30 min at 37 °C.
Purified PON1 studies
Human PON1 was purified from serum as previously described [27,31]. Synthetic 3OC12-HSL (10 μM) was prepared in reaction buffer. Purified PON1 was added to the solution and incubated at 37 °C. Samples of the reaction solution were taken at 15 and 30 min, and 3OC12-HSL concentration determined by bioassay.
Chinese hamster ovary (CHO) cells were cultured as monolayers in plastic dishes. Recombinant adenovirus serotype 5 (Ad5) expressing hPON1, hPON2, hPON3, or GFP was produced by the University of Iowa Vector and Gene Targeting core. CHO cells were infected with recombinant Ad5 by calcium phosphate transfection  and allowed to express for 48 h. Transfection efficiency was evaluated by fluorescence measurement in Ad5-GFP infected cells. Synthetic 3OC12-HSL (10 μM) in PBS was added to transgene expressing cells and incubated at 37 °C, 5.5% CO2 for 30 min. Remaining 3OC12-HSL in the medium was measured by bioassay.
P. aeruginosa quorum-sensing and biofilm formation
The effect of PON1 in serum on P. aeruginosa biofilm formation was studied in vitro by means of a modified Calgary Biofilm Device assay . Initially, a plastic peg was immersed in each well of a 96-well culture plate containing wild-type or lasI mutant PAO1 in 1% tryptic soy broth containing 1 mM CaCl2. After 1 h, the pegs were removed from the PAO1 culture and transferred to wells containing 1% wild-type mouse serum, Pon1-KO mouse serum with or without purified human PON1 (20 μg/mL), or control serum (5% bovine serum albumin in lactated Ringer's solution) with or without purified human PON1. The pegs were incubated for 48 h, followed by a wash in water to remove non-adherent bacteria and staining with 0.5% crystal violet solution. After washing non-bound crystal violet, the stain was eluted in ethanol and quantified spectrophotometrically at A560 to measure biofilm biomass.
All experiments were reviewed and approved by the Animal Care and Use Committee of the University of Iowa. Mice were backcrossed at least 10 generations into C57BL/6 background. Controls were weight, gender, and age matched. Mice were allowed access to food and water ad libitum. Male C57BL/6 mice (8–10 weeks old, weighing 20–25 g) were obtained from Harlan Industries, Inc. (Indianapolis, IN). Overnight cultures of PAO1 or PAO1-lasI− were subcultured to OD600=0.2–0.4. Cultures were washed 3× and concentrated to the desired OD600 in cold PBS. Bacteria were kept on ice until injection. Colony forming units (CFUs) were quantified by standard plate counting assay. Mice were injected intraperitoneally with 0.1 mL of Pseudomonas suspension. Animals were carefully monitored and euthanized when end-point conditions were met (moribund, distressed, and unable to eat or drink).
Quantitative PCR of epithelia
For RT-PCR analysis, first strand cDNAs were synthesized using total RNA isolated from the airway epithelia and the ThermoScript RT-PCR system (Invitrogen). Real time quantitative PCR was then performed using the QuantiTest SYBR Green PCR kit (Qiagen) in a PerkinElmer Applied Biosystems PRISM 7700 Sequence Detector (Foster City, CA). Known copy numbers of cloned cDNA plasmids for mouse PON1, PON2, PON3, and GAPDH genes were included in the assay to serve as copy number standards. The ratio of PON cDNA copy number relative to GAPDH cDNA copy number of each sample was determined. Primers are included in the supplemental information.
Acyl-HSL and quantitative acyl-HSL assay
Chemically synthesized 3OC12-HSL was obtained from Vertex Pharmaceuticals (Coralville, IA). 3OC12-HSL was measured by quantitative bioassay as described  using Escherichia coli MG-4 (pKDT17) [6,35].
P-values of bioassay data were calculated using Student's t test in Microsoft Excel software. Survival data was analyzed by log rank test using online software (http://bioinf.wehi.edu.au/software/russell/logrank/).
OC12-HSL is degraded through lactone ring hydrolysis
We previously found that the signaling activity of the 3OC12-HSL quorum-sensing molecule was lost when exposed to cultured human airway and other epithelial cells. To characterize the structural changes associated with this loss of activity, three independent methods were used: NMR, HPLC, and MS.
First, analysis of the NMR spectrum of 3OC12-HSL showed that peaks shifted after exposure to A549 airway epithelial cells (Fig. 1A). These shifts reflect structural changes previously reported to correspond to lactone ring hydrogens in 3OC12-HSL . Second, HPLC on a C18 column was used to determine whether exposure to A549-cells did in fact hydrolyze the 3OC12-HSL lactone ring. Since it is known that high pH hydrolyzes acyl-HSL lactone rings , we compared the elution times of untreated 3OC12-HSL to 3OC12-HSL exposed to high pH or A549-cells. Fig. 1B shows that while untreated 3OC12-HSL eluted at about 10 min, 3OC12-HSL exposed to either high pH or A549-cells eluted from the column identically – at about 6 min. These data show that both treatments make 3O12-HSL less hydrophobic, consistent with hydrolysis of the lactone ring. Finally, MS showed a mass ion at 298 that corresponded to the molecular weight of the unhydrolyzed form (Fig. 1C). After incubation in alkaline conditions, hydrolysis of the lactone ring was confirmed by the appearance of a mass ion of m/z 316 (Fig. 1C); this represents the molecular weight of 3OC12-HSL plus one water molecule. The MS of 3OC12-HSL inactivated by A549-cells also contained the prominent peak at m/z 316. Taken together, these data indicate that A549 human epithelial airway cells degrade the P. aeruginosa 3OC12-HSL quorum-sensing signal by hydrolyzing its lactone ring.
Previous studies have shown that homoserine lactone rings hydrolyzed at high pH can be recircularized in acidic solutions . Fig. 1D shows that pH 1.5 had no effect on active 3OC12-HSL, but after hydrolysis at pH 10, exposure to pH 1.5 completely restored biological activity. Likewise, pH 1.5 treatment fully restored the signaling activity of 3OC12-HSL degraded by A549-cells. These data suggest that A549-cells degrade 3OC12-HSL by hydrolysis.
Serum lactonase PON1 degrades 3OC12-HSL
Since the mammalian Pon genes encode paraoxonases with lactonase activity, we hypothesized that they may be responsible for degrading the 3OC12-HSL signal. Both PON1 and PON3 are found in serum so we tested whether serum degrades 3OC12-HSL. As with exposure to A549-cells, 3OC12-HSL incubated with wild-type serum is rapidly inactivated (Fig. 2A); this inactivation is completely reversed by incubation at low pH (Fig. 2A). HPLC and MS (Fig. 2B) further confirmed that, similar to A549-cells, serum also contains an activity that hydrolyzes the lactone ring.
To characterize the activity that degrades 3OC12-HSL in serum, we treated human serum with PON inhibitors: EDTA , oxindole, and isatin ; all blocked degradation of 3OC12-HSL (Fig. 2C). Since PON1 is the most abundant paraoxonase found in serum, and mice with targeted disruptions of the Pon1 gene (Pon1-KO mice) have been previously generated, we assayed serum from Pon1-KO mice to more directly test whether PON1 is the enzyme that degrades 3OC12-HSL . Unlike serum from wild-type mice, serum from mice lacking Pon1 exhibited impaired degradation of 3OC12-HSL (Fig. 2D). Furthermore adding back PON1, purified from human serum, was sufficient to degrade 3OC12-HSL in vitro. Fig. 2E shows that the activity of 3OC12-HSL was inversely related to the amount of purified PON1 added to the reaction mixture. Thus, in vitro PON1 is both required and sufficient to degrade 3OC12-HSL in mouse serum.
As stated above, in addition to PON1 mammals express two other lactone ring hydrolases – PON2 and PON3 [40,41]. To test whether these proteins similarly degrade 3OC12-HSL we constructed recombinant Adenoviruses expressing human Pon genes. CHO cells were transfected with adenoviruses expressing human PON1, PON2, or PON3 (or GFP as a control) and after 48 h 3OC12-HSL was added to the culture medium and assayed for degradation. After 60 min of incubation the supernatant from cells transfected with both hPON1 and hPON3 had degraded nearly 80% of the added 3OC12-HSL; in media from cells transfected with PON2 nearly 100% of the 3OC12-HSL was degraded (Fig. 3). In comparison, little change was effected by cells expressing GFP. This indicates that all three mammalian paraoxonases degrade 3OC12-HSL with PON2 being the most active.
PON1 blocks P. aeruginosa biofilm formation
Previous reports show that biofilm formation is reduced or altered in mutant P. aeruginosa that cannot synthesize 3OC12-HSL [9,42]. Based on these findings, we predicted that since serum degrades 3OC12-HSL it would also impair P. aeruginosa biofilm formation. We used a previously described in vitro system in which P. aeruginosa biofilms grow on plastic pegs immersed in medium . Fig. 4B shows that 0.25% wild-type mouse serum reduced biofilm formation by 75% while a 0.25% solution of BSA (artificial serum) only reduced bilofilm formation by 10%. In contrast, serum from Pon1-KO mice showed minimal effects, only reducing P. aeruginosa biofilm formation by 15%. Furthermore, adding purified human PON1 to artificial serum, or to Pon1-KO mouse serum, inhibited biofilm formation similar to wild-type mouse serum (Fig. 4A). These data show that adding PON1-containing serum to culture media interferes with the ability of P. aeruginosa to form biofilm.
To confirm that PON1 inhibits biofilm formation by disrupting P. aeruginosa quorum-sensing signaling, we examined biofilm formation by the lasI mutant strain of P. aeruginosa that does not produce 3OC12-HSL. Fig. 4B shows that, as expected the lasI mutant formed less biofilm than wild type P. aeruginosa in artificial serum. However, unlike wild-type P. aeruginosa, lasI mutant bacteria responded the same to both wild-type and Pon- KO mouse serum. These results suggest that serum reduces the ability of P. aeruginosa to form biofilms in vitro because PON1 degrades the 3OC12-HSL quorum-sensing signal.
Quorum-sensing by P. aeruginosa in the murine peritonitis/sepsis model
P. aeruginosa quorum-sensing modulates virulence in several in vivo models of infection. In mice, these include acute and chronic lung infection, and burn infection models [43–45]. To address the role of paraoxonases in host defense mechanisms, we used a mouse model of P. aeruginosa peritonitis/sepsis. To test the effect of 3OC12-HSL quorum-sensing on P. aeruginosa virulence in this model, wild-type mice were infected with the LD50 (1.9 × 107–3.8 × 107 CFU) of PAO1 (a wild-type strain of P. aeruginosa) or an equal dose of the lasI mutant strain (Fig. 5A). While 55% of the mice infected with the PAO1 strain died after 100 h, less than 10% those infected with the lasI mutant died, suggesting that 3OC12-HSL is important for causing mortality. These findings are consistent with C. elegans and other mouse models that show decreased virulence in infections with quorum-sensing deficient strains of P. aeruginosa [15,43–45].
Pon1 deficient mice are paradoxically protected from quorum-sensing dependent P. aeruginosa infection
Our results and those of others suggest that disrupting P. aeruginosa quorum-sensing signaling decreases mortality in the infected host [44,46]. We therefore hypothesized that Pon1-KO mice would be more sensitive to PAO1 infection. Contrary to our prediction, at 50 h post-infection 100% of the Pon1-KO mice survived as compared to 50% of wild-type mice (Fig. 5B). These results implied that disrupting Pon1 actually protects the mice from succumbing to infection. We then compared the outcome of infection by the lasI mutant, which cannot synthesize quorum-sensing signals. We first determined the LD50 of lasI mutant P. aeruginosa in wild type mice (6.3 × 107 CFU). When infected with this dose of the lasI mutant strain, wild-type and Pon1-KO mice survived at the same rate (Fig. 5C). These results demonstrate that the paradoxical difference in mortality between wild-type and Pon1-KO mice infected with PAO1 is related to 3OC12-HSL.
These unexpected findings suggest that other protective factors compensated for the loss of PON1 in knockout mice. We hypothesized that the expression of Pon2 or Pon3 could be increased in the Pon1-KO mouse. Our hypothesis was confirmed by semi-quantitative PCR of airway epithelia from Pon1-KO mice: both PON2 and PON3 mRNA levels were doubled in the Pon1-KO mice as compared to the wild-type mice (Fig. 5D). Therefore, increased tissue expression of PON2, PON3, or both in Pon1-KO mice may degrade 3OC12-HSL more effectively in vivo thereby protecting the KO mice from sepsis caused by P. aeruginosa.
Previously we reported that the P. aerugionosa quorum-sensing molecule 3OC12-HSL loses signaling activity after incubation with primary airway epithelia and other cultured cells . In this study we identify the mechanism of acyl-HSL degradation as lactone ring hydrolysis. This is consistent with findings in prokaryotes showing that quorum-sensing inhibitor systems degrade the lactone ring of acyl-HSLs [12–14,16,47–49], such as was found for the quorum-sensing inhibitor AiiA from the bacteria Bacillus sp.
We found that, like airway cells, human serum degrades 3OC12-HSL through lactonolysis. We therefore suggested that a likely candidate for this activity is PON1 – a lactonase abundant in serum. Serum from Pon1-KO mice was impaired in degrading this quorum-sensing signal. The importance of serum to host defense cannot be understated, as it is a vital medium for transport of immune cells and defense factors to sites of infection. The discovery that both serum-PON1 and purified PON1 reduced P. aeruginosa biofilm formation showed that PON1 can significantly impact a complex process that relies heavily on quorum-sensing [9,11,50].
Both pathogen and host contribute to microbial pathogenicity, [reviewed in Casadevall and Pirofski ]. In this model, both a weak and an overzealous host response can enhance virulence. Thus, elevated 3OC12-HSL concentrations in the Pon1-KO mouse may provoke different responses from both the P. aeruginosa and the host defense mechanisms. Similar to findings in other animal models of infection [43–45], causing sepsis in mice using a quorum-sensing deficient strain of P. aeruginosa did not kill them as effectively as sepsis caused by wild-type bacteria. These results confirmed the hypothesis that attenuating quorum-sensing signaling reduces bacterial virulence. In apparent discord with these findings, however, Pon1 deficient mice were resistant to P. aeruginosa infection, surviving infection more frequently. Quorum-sensing appears to be involved since a 3OC12-HSL-deficient mutant strain of P. aeruginosa killed both Pon1-KO and wild-type mice with the same frequency. Both PON2 and PON3, although not present in serum, were increased in the PON1-KO mice epithelia, as compared to wild-type mice. PON2 is expressed ubiquitously, including in endothelial cells and bloodstream monocytes [26,40]. It is possible that increased acyl-HSL lactonase activity in these cells compensates for the absence of PON1 in serum. However, to test this hypothesis, it will be necessary to generate a triple knockout animal that lacks endogenous paraoxonase expression. Thus, based on our data we cannot conclude that modulation of P. aeruginosa quorum-sensing by PON1 is a host defense mechanism.
Although paraoxonases are conserved from C. elegans to mammals , their physiological substrates are not known [20,21]. Our data, in conjunction with previous studies [29,52], propose that PON1 may have evolved to degrade bacterial acyl-HSLs. Hence, a quorum-sensing blockade by PON1 and other members of the family may mediate a number of bacterial biofilm, virulence- and inflammation-dependent processes in host organisms.
We thank Jamie Kesselring for excellent assistance. We also thank Mike Welsh, Kate Excoffon, Pradeep Singh, Roberto Sanches de Leon, Blaise Boles, Yannick Lequette, Mark Urbanowski, William Kearney, and Lynn Teesch for their generous assistance and helpful discussions. Also, the University of Iowa In Vitro Models Core, Molecular Analysis Facility, and NMR core greatly contributed to this work. Supported by NIH Grants HL61234 (J.Z.), HL30568 (A.J.L.), and 5 P42 ES04696 (C.E.F.), and NIGM Grant GM59026 (E.P.G.).
Appendix A Supplementary data
Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.femsle.2005.09.023.