Signal Synthase-Type versus Catabolic Monooxygenases: Retracing 3-Hydroxylation of 2-Alkylquinolones and Their N-Oxides by Pseudomonas aeruginosa and Other Pulmonary Pathogens

Pseudomonas aeruginosa, Staphylococcus aureus, and Mycobacteroides abscessus are major players in bacterial chronic infections and particularly common colonizers of cystic fibrosis (CF) lung tissue. Whereas S. aureus is an early onset pathogen in CF, P. aeruginosa establishes at later stages. M. abscessus occurs at all stages but has a lower epidemiological incidence.

ABSTRACT

The multiple biological activities of 2-alkylquinolones (AQs) are crucial for virulence of Pseudomonas aeruginosa, conferring advantages during infection and in polymicrobial communities. Whereas 2-heptyl-3-hydroxyquinolin-4(1H)-one (the “Pseudomonas quinolone signal” [PQS]) is an important quorum sensing signal molecule, 2-alkyl-1-hydroxyquinolin-4(1H)-ones (also known as 2-alkyl-4-hydroxyquinoline N-oxides [AQNOs]) are antibiotics inhibiting respiration. Hydroxylation of the PQS precursor 2-heptylquinolin-4(1H)-one (HHQ) by the signal synthase PqsH boosts AQ quorum sensing. Remarkably, the same reaction, catalyzed by the ortholog AqdB, is used by Mycobacteroides abscessus to initiate degradation of AQs. The antibiotic 2-heptyl-1-hydroxyquinolin-4(1H)-one (HQNO) is hydroxylated by Staphylococcus aureus to the less toxic derivative PQS-N-oxide (PQS-NO), a reaction probably also catalyzed by a PqsH/AqdB ortholog. In this study, we provide a comparative analysis of four AQ 3-monooxygenases of different organisms. Due to the major impact of AQ/AQNO 3-hydroxylation on the biological activities of the compounds, we surmised adaptations on the enzymatic and/or physiological level to serve either the producer or target organisms. Our results indicate that all enzymes share similar features and are incapable of discriminating between AQs and AQNOs. PQS-NO, hence, occurs as a native metabolite of P. aeruginosa although the unfavorable AQNO 3-hydroxylation is minimized by export as shown for HQNO, involving at least one multidrug efflux pump. Moreover, M. abscessus is capable of degrading the AQNO heterocycle by concerted action of AqdB and dioxygenase AqdC. However, S. aureus and M. abscessus orthologs disfavor AQNOs despite their higher toxicity, suggesting that catalytic constraints restrict evolutionary adaptation and lead to the preference of non-N-oxide substrates by AQ 3-monooxygenases.

IMPORTANCE Pseudomonas aeruginosa, Staphylococcus aureus, and Mycobacteroides abscessus are major players in bacterial chronic infections and particularly common colonizers of cystic fibrosis (CF) lung tissue. Whereas S. aureus is an early onset pathogen in CF, P. aeruginosa establishes at later stages. M. abscessus occurs at all stages but has a lower epidemiological incidence. The dynamics of how these pathogens interact can affect survival and therapeutic success. 2-Alkylquinolone (AQ) and 2-alkylhydroxyquinoline N-oxide (AQNO) production is a major factor of P. aeruginosa virulence. The 3-position of the AQ scaffold is critical, both for attenuation of AQ toxicity or degradation by competitors, as well as for full unfolding of quorum sensing. Despite lacking signaling functionality, AQNOs have the strongest impact on suppression of Gram-positives. Because evidence for 3-hydroxylation of AQNOs has been reported, it is desirable to understand the extent by which AQ 3-monooxygenases contribute to manipulation of AQ/AQNO equilibrium, resistance, and degradation.

INTRODUCTION

Quinolone-type secondary metabolites have attracted much attention in the context of Pseudomonas aeruginosa quorum sensing and virulence (1). However, a series of reports on their isolation from fungi, plants of the Rutaceae family, marine pseudo- and alteromonads, marine streptomycetes, as well as soil-inhabitant and lung-pathogenic Burkholderia species substantiates their widespread occurrence (2–8). Among the various 2-alkylquinolones (AQs) produced by P. aeruginosa, 2-heptyl-3-hydroxyquinolin-4(1H)-one (the “Pseudomonas quinolone signal” [PQS]) and, to a lesser extent, its precursor 2-heptylquinolin-4(1H)-one (HHQ) contribute to its quorum sensing network, which controls the expression of numerous virulence genes and enables synchronized behavior within a multicellular population (1, 9–11). AQs with other alkyl chain length configurations inhibit growth of various bacteria (2, 4). Another class of metabolites, the 2-alkyl-1-hydroxyquinolin-4(1H)-ones (alkylhydroxyquinoline N-oxides [AQNOs]) act as toxins with no direct role in QS. They are potent inhibitors of electron transfer by competitively inhibiting the quinone binding sites of the cytochrome bc₁ complex, cytochrome b₀, and b₆f (1–3). While 2-heptyl-1-hydroxyquinolin-4(1H)-one (HQNO) may repress growth of potential competitors at low concentrations (11–13), its accumulation can also lead to self-poisoning and autolysis of P. aeruginosa cells (14).

Continuous exposure to antimicrobials exerts evolutionary pressure to susceptible microorganisms (2, 4, 11, 15). As a consequence, several ways to transform or even degrade toxic AQ and AQNO metabolites have evolved in bacteria (16–19). However, to the best of our knowledge, detoxification of AQs or AQNOs by higher eukaryotes has not been reported so far. It is conceivable that AQ/AQNO detoxification and degradation is solely sustained by microorganisms cooccurring with AQ producers, such as P. aeruginosa. Among the strategies to detoxify HQNO, 3-hydroxylation has been observed for Rhodococcus erythropolis, Arthrobacter sp., and Staphylococcus aureus. However, at least for S. aureus, HQNO hydroxylation resulted in only moderate attenuation of toxicity (20). Interestingly, hydroxylation at C-3 of HHQ is the first step of a bacterial AQ degradation pathway encoded by the aqdRABC gene cluster (18, 19, 21). This PQS-inducible AQ degradation pathway was originally identified in Rhodococcus erythropolis strain BG43 (18); however, orthologs of aqdRABC are conserved among several actinobacteria, including various Mycobacteroides abscessus isolates (18, 19, 21, 22). 3-Hydroxylation, catalyzed by the monooxygenase AqdB, activates the quinolone ring for subsequent 2,4-dioxygenolytic ring cleavage by AqdC, which requires a 3-hydroxyquinolin-4(1H)-one type substrate (23) (Fig. 1). Subsequent hydrolysis of the cleavage product N-acylanthranilic acid by the amidase AqdA yields anthranilic acid and a fatty acid, which enter the central metabolism (18, 19, 21).

FIG 1.

3-Monooxygenases in the biosynthesis and degradation of AQs and AQNOs. In P. aeruginosa, biosynthesis of HHQ is mediated by PqsABCDE (51–54). Subsequent hydroxylation of HHQ by PqsH_PA yields PQS (red frame) (24). The latter reaction is also part of a specific AQ degradation pathway comprising the AqdABC enzymes, which is conserved in Mycobacteroides abscessus subsp. abscessus and a series of other actinobacteria (blue frame) (18, 19). Analogous 3-hydroxylations of HHQ and HQNO are performed by Microbulbifer sp. HZ11 (PqsH_HZ11) and S. aureus (Aqm_SA), respectively (brackets) (16, 20). 3-Hydroxylation reactions of Aqm_SA, hydroxylation of AQNOs by PqsH/AqdB enzymes, and subsequent cleavage of PQS-N-oxide by AqdC are hypothetical (dashed arrows).

In P. aeruginosa, 3-hydroxylation of AQs is catalyzed by the flavoprotein monooxygenase (FPMO) PqsH (24). A PqsH ortholog of the marine gammaproteobacterium Microbulbifer sp. HZ11 likewise hydroxylates HHQ (17). Interestingly, the pqsH_HZ11 gene is located adjacent to a putative AQ biosynthetic gene cluster (Fig. 2). The AqdB proteins of AQ-degrading actinobacteria, as well as the two PqsH orthologs are homologous and belong to group A within FPMOs, sharing a glutathione reductase II fold, a flavin adenine dinucleotide (FAD) cofactor, and NADH/NADPH cosubstrate specificity (18, 19, 24).

FIG 2.

Phylogenetic relationships of monooxygenases with known or putative AQ 3-hydroxylase function. (A) Strains harboring AQ 3-monooxygenases were identified by blastp search of PqsH_PA (PAO1 reference genome) against the RefSeq database (55). Based on sequence similarity or location within putative AQ biosynthetic (green) or catabolic gene clusters (red), enzymes with known or postulated functions as AQ 3-monooxygenases were selected for alignment and subsequent modeling of the tree (ClustalW/MEGA6). Stars indicate existing experimental evidence for AQ 3-hydroxylation. (B) Genetic context of selected AQ 3-monooxygenases from P. aeruginosa (PqsH_PA, blue), Microbulbifer sp. HZ11 (PqsH_HZ11, green), Mycobacteroides abscessus subsp. abscessus (AqdB_MA, purple), and Staphylococcus aureus (Aqm_SA, red). Biotransformation of 10 μM HHQ (C) or HQNO (D) by recombinant E. coli harboring monooxygenases PqsH_PA (blue), PqsH_HZ11, (green), AqdB_MA (purple), or Aqm_SA (red) as resting cell cultures (OD₆₀₀ = 3.0).

P. aeruginosa, Rhodococcus and Arthrobacter spp., S. aureus, and environmental mycobacteria such as M. abscessus can cooccur in many habitats. P. aeruginosa, S. aureus, and M. abscessus moreover are often involved in opportunistic pulmonary infections, especially in patients with cystic fibrosis, and may cocolonize the lung (25–28). In such polymicrobial communities, microorganisms interact through diffusible metabolites. Because 3-hydroxylation of AQs appears to be common to both, signal synthesis as well as degradation of secondary metabolites, AQ 3-monooxygenases should strongly contribute to AQ- and AQNO-based “chemical warfare” in P. aeruginosa-colonized habitats.

AQ and AQNO 3-hydroxylation by different organisms raises the question of whether the involved monooxygenases show detectable idiosyncrasies reflecting specific physiological roles or, alternatively, share similar functional properties. In particular, the pivotal role of AQNOs for the competitiveness of P. aeruginosa prompted us to study their 3-hydroxylation in more detail. We, therefore, investigated four different AQ 3-monooxygenases, focusing on their general catalytic properties and their potential to convert HQNO and HHQ. We also investigated the physiological impact of HQNO hydroxylation in P. aeruginosa. To this end, we assessed the importance of HQNO efflux and AQ/AQNO substrate competition for HQNO conversion in P. aeruginosa and tested the respiratory toxicity of PQS-N-oxide (PQS-NO), the product of HQNO 3-hydroxylation. Furthermore, we analyzed HQNO degradation by mycobacterial catabolic enzymes.

RESULTS

Two clades of AQ 3-monooxygenases.

For a sequence-based comparison of AQ 3-monooxygenases, we used the known signal synthase-type (PqsH) and catabolic (AqdB) enzymes (16, 18, 19, 24) to search databases for homologs. Sequence similarity and analysis of neighboring genes suggested a series of PqsH- and AqdB-type enzymes, a representative selection of which was used for phylogenetic analysis (Fig. 2). The phylogenetic tree revealed relatively close relatedness of AqdB-type enzymes, with the corresponding genes located in (predicted) catabolic gene clusters. PqsH of P. aeruginosa and Microbulbifer sp. HZ11 (termed PqsH_PA and PqsH_HZ11) are more distantly related. Other enzymes originating from potential AQ biosynthetic clusters are more closely related to the PqsH branch than to the AqdBs. While AQs appear to be predominantly produced by Gram-negative bacteria, degradative gene clusters comprising (predicted) aqdB-type genes occur mainly in Gram-positive species. However, we identified a putative AQ 3-monooxygenase encoded in a monocistronic operon without apparent association to a catabolic gene cluster but more closely related to the PqsH branch in Staphylococcus aureus Newman (GenBank accession number WP_000684141) (hereafter referred to as Aqm_SA). Escherichia coli expressing Aqm_SA, similar to recombinant E. coli strains producing known AQ 3-monooxygenases from different genetic environments (Fig. 2B), converted 10 μM HHQ within 60 min (Fig. 2C), confirming the proposed activity of the S. aureus enzyme.

A previous observation of HQNO 3-hydroxylation by R. erythropolis BG43 (21) raised the question of whether AqdB was responsible for the reaction. Activity toward AQNOs, although not reported in P. aeruginosa, could even be a general property of AQ 3-monooxygenases. Indeed, biotransformation assays revealed conversion of HQNO within 120 min by all four enzymes (Fig. 2D). Decrease of HQNO was accompanied by formation of a compound with a mass of 276 (M+H)⁺ and a UV absorption spectrum identical to previously described PQS-NO (21). Reduction of the N-oxide moiety with zinc powder (4) yielded PQS, thus confirming the hydroxyl moiety at position 3 of the AQ backbone.

The observation that the AqdB- as well as PqsH-type enzymes convert HQNO led us to focus on the role of PQS-NO in AQ biosynthesis and degradation pathways: Can PQS-NO, as a product of HQNO hydroxylation by PqsH_PA, also be found in the native host P. aeruginosa, and if so, are there any benefits of the compound for P. aeruginosa in terms of toxicity or signaling? As regards the degradative side, in light of AqdB_MA activity toward HQNO, is the resulting PQS-NO accepted as a substrate by the ring-cleavage PQS dioxygenase AqdC?

HQNO hydroxylation in P. aeruginosa.

Because HQNO is hydroxylated by recombinant PqsH_PA in E. coli, we wondered whether the reaction could also be observed in P. aeruginosa or if any mechanisms existed to shut down PqsH activity toward exogenously added HQNO. Initial experiments were conducted in a P. aeruginosa PAO1 ΔpqsA genetic background (Fig. 3), as the lack of AQ production due to deletion of pqsA facilitates detectability of metabolites with low abundances in HQNO biotransformations.

FIG 3.

Biotransformation of HQNO in P. aeruginosa mutants. P. aeruginosa ΔpqsA (A), ΔpqsAH (B), ΔpqsAH(p:pqsH_PA) (C), ΔpqsAH(p:aqdB_MA) (D), and ΔpqsAH(p:empty) (E) mutant resting cell cultures (OD = 3) were supplemented with 15 μM HQNO (gentamicin and 0.2% arabinose added as required); sampled at indicated intervals; and PQS-NO (red circles), PQS (purple squares), HQNO (blue triangles), and HHQ (green triangles) were quantified by HPLC. Given data are means and standard deviation (SD) of biological duplicates. Notably, HQNO stocks contained 5% HHQ originating from synthesis and decomposition.

HQNO concentrations decreased over time in the ΔpqsA strain but not in the ΔpqsAH double mutant strain (Fig. 3A and B), confirming conversion of HQNO by PqsH_PA in the native host. HQNO also was converted when genes encoding AqdB_MA or PqsH_PA were expressed in the ΔpqsAH double mutant, albeit to different extents (Fig. 3C and D). However, expression levels may have a major influence on the observed transformation kinetics. In the case of the P. aeruginosa ΔpqsAH(p:pqsH_PA) strain, the supplemented HQNO was completely converted to PQS-NO within 6 h. PQS-NO spontaneously decomposes to PQS (the reported half-life of PQS-NO in sterile LB is 2.5 h at 30°C [20]), but only about 2 μM PQS was found in these cultures (Fig. 3C), 1 μM of which originated from conversion of contaminant HHQ. With regard to the spectrum of alkyl quinolone derivatives described for P. aeruginosa (29), the observation that PqsH_PA readily converts HQNO was somewhat surprising.

PQS-NO is a native metabolite of the P. aeruginosa Pqs pathway.

Under laboratory conditions, AQNOs are the most abundant compounds synthesized by the Pqs pathway of P. aeruginosa, reaching concentrations of above 40 μM (29–31). Still, it seems that the main compounds subjected to 3-hydroxylation are the C₇ and C₉ AQ species (24, 29, 31, 32). Upon discovery of HQNO conversion by PqsH, we analyzed in detail the culture extracts of wild-type P. aeruginosa PAO1 and detected a compound matching PQS-NO in absorption characteristics and molecular mass. The compound was low in abundance, and absent in cultures of ΔpqsA, ΔpqsL (pqsL encodes the 2-aminobenzoylacetate N-monooxygenase required for AQNO synthesis), and ΔpqsH mutants (Fig. 4A to C). The compound was also found in cultures of P. aeruginosa strain PA14. We estimate a concentration of up to 1 μM and 2 μM PQS-NO in PAO1 and PA14 cultures, respectively, in contrast to 15 μM and 30 μM PQS in the same respective cultures (data not shown).

FIG 4.

Presence of PQS-NO in P. aeruginosa wild-type cultures and its effect on aerobic respiration. P. aeruginosa strains were inoculated into LB and grown overnight at 37°C and 160 rpm. Extracts of culture samples were analyzed by HPLC-MS. Chromatograms (A), UV-Vis spectra (B), and mass spectra (C) of putative PQS-NO (red) were compared to a PQS-NO reference sample (blue). (D) Effects on aerobic respiration were assayed by adding HQNO or PQS-NO to cell extracts of a P. aeruginosa ΔpqsAH mutant. Aerobic respiration was initiated by addition of NADH, followed by the supplementation of HQNO (gray squares) or PQS-NO (black circles) at final concentrations between 0.004 and 300 μM. Decrease in aerobic respiration was monitored with a Clark electrode and normalized to untreated controls. Error bars reflect standard deviations of two independent experiments, each with two technical replicates.

Since PQS-NO has not been reported before as a native metabolite of P. aeruginosa, we were interested in whether the compound confers a benefit to P. aeruginosa. Addition of PQS-NO to cultures did not lead to increased production of AQs, indicating that the compound, in contrast to HHQ and PQS, does not act as an inducer of the PQS quorum sensing system (20). Even though PQS-NO is less toxic to S. aureus than HQNO, both compounds caused a similar degree of respiratory inhibition to P. aeruginosa (Fig. 4D). The compound hence appears to be a side product of the AQ/AQNO biosynthetic pathway with no obvious biological functionality or advantage for the producer.

HQNO hydroxylation in P. aeruginosa is counteracted by efflux.

The occurrence of only minute amounts of PQS-NO in P. aeruginosa PAO1 cultures made us wonder whether a preference of PqsH for HHQ over HQNO is sufficient for the observed selectivity, or if additional mechanisms, such as substrate competition or efflux, play a role in preventing HQNO hydroxylation. Indeed, inhibition of multidrug resistance (Mex-type) efflux pumps with the inhibitor phenylalanine-arginine β-naphtylamide (PAβN) (33) resulted in an about 3-fold increase of HQNO conversion (Fig. 5A). Intriguingly, inhibition of efflux pumps had a stronger impact on HQNO hydroxylation than on HHQ conversion, which was only increased 1.6-fold upon treatment with PAβN. This observation suggests that AQNOs may be more rigorously exported than AQs, leading to lower cytoplasmic abundance and, hence, lower conversion by PqsH. However, it cannot be assessed to what extent transport of 3-OH-AQNOs is influenced by PAβN and to what extent their transport affects their decomposition to PQS.

FIG 5.

RND-type efflux pumps, but not substrate competition, affect HQNO hydroxylation. (A) For inhibition of Mex-type efflux pumps, the P. aeruginosa ΔpqsA strain was incubated with 100 μg ml⁻¹ PAβN for 45 min at 37°C before 10 μM HQNO or HHQ were added. Conversion was analyzed by HPLC of whole-culture extracts after 90 min. Error bars reflect SD of three biological replicates. Significance is denoted as P values of <0.05 (*) or <0.001 (***). (B) P. aeruginosa ΔpqsAH(p:pqsH_PA) (blue circles) and ΔpqsAH(p:aqdB_MA) (red squares) mutant resting cell cultures (OD₆₀₀ = 3.0) were supplemented with premixed AQs and incubated for 30 min at 37°C. AQs were extracted from whole-culture samples and analyzed by HPLC. Activity toward HQNO was normalized to overall substrate conversion.

To elucidate, if substrate competition or inhibitory effects by high HHQ concentrations limit or abolish PQS-NO formation in vivo, we assayed HQNO conversion in the presence of varying amounts of HHQ in the P. aeruginosa ΔpqsAH(p:pqsH_PA) mutant strain. Concentrations of HHQ were varied between 0.5 μM and 15 μM, a range which can typically be found in wild-type P. aeruginosa cultures. We observed that HQNO conversion was not affected by titration of HHQ. In the presence of equal concentrations of HHQ and HQNO (15 μM each), relative HQNO conversion was 13% (Fig. 5B), which confirms the reaction to be catalyzed in the presence of HHQ in vivo despite presumed preference of PqsH toward HHQ. To check whether differences between degradative and signal synthase type AQ 3-monooxygenases would manifest in this experiment, we repeated the experiment with a ΔpqsAH(p:aqdB_MA) strain. The relative activity of the AqdB-expressing strain toward HQNO as a substrate was similar to that of the PqsH-producing strain, with 19% activity toward HHQ at 15 μM of both HHQ and HQNO (Fig. 5B). Taken together, the experiments indicate that the low level of PQS-NO formed in P. aeruginosa cultures is a result of compartmentalization by efflux of AQNO, coupled with a preference of the enzyme for HHQ, which is however not a distinguishing feature of PqsH_PA but is also observed for the degradative enzyme AqdB_MA.

PQS formation via HQNO plays a minor role in P. aeruginosa cultures.

PQS-NO stability greatly varies depending on solvent and molecular environment. While the compound is comparably stable in acidic methanol, its decomposition to PQS is detectable within minutes in dimethyl sulfoxide (DMSO) or LB (at room temperature). Therefore, we investigated whether a portion of the PQS generated in P. aeruginosa cultures may originate from abiotic decomposition of PQS-NO.

We synthesized isotope-labeled HQNO containing a ¹³C-homoaromatic ring, yielding (¹³C₆)-HQNO for supplementation of P. aeruginosa cultures (for details on synthesis, see Fig. S1 in the supplemental material). For the experiment, we used a P. aeruginosa ΔpqsL strain, which is incapable of AQNO synthesis but competent of producing AQs. Cultures were supplemented with (¹³C₆)-HQNO, and +6-Da shifts were monitored and quantified by liquid chromatography-mass spectrometry (LC-MS) analysis of extracts of culture samples. In accordance with our previous results, we detected ¹³C₆-PQS-NO as a product of ¹³C₆-labeled HQNO. Unexpectedly, ¹³C₆-PQS shares of total PQS only reached 3%. Inhibition of Mex-type efflux pumps with PAβN resulted in slightly increased levels of ¹³C₆-PQS-NO (see Fig. S3 in the supplemental material), but ¹³C₆-PQS shares were even less (1%). Although absolute ¹³C₆-PQS varied, this tendency was reflected in two independent experiments (see Fig. S2 in the supplemental material). These data led us to conclude that PQS formation out of HQNO, although occurring in P. aeruginosa, plays an only minor role with regard to the total PQS amounts formed. Because ¹³C₆-PQS-NO, and moreover the ¹³C₆ phenyl moiety, depleted over time, we hypothesize that, before decomposing to PQS, the compound may bind and react with one or more proteins or other components of P. aeruginosa, leading to the formation of adducts, which would be untraceable in ethyl acetate extracts of cultures.

Fate of PQS-N-oxide in the mycobacterial AQ degradation pathway.

In the bacterial degradation of AQs, hydroxylation by AqdB-like monooxygenases is a prerequisite for subsequent cleavage of the heterocyclic ring by AqdC-type dioxygenases (19, 22, 23). When PQS-NO was reacted with purified AqdC of M. abscessus (AqdC_MA), an additional compound with an indistinct absorption band around 280 nm was detected, whose mass of 279 Da corresponded to the mass of the expected PQS-NO cleavage product N-(N-hydroxy)-octanoyl anthranilic acid (calculated mass, 279.15 Da) (Fig. 6A and B). When ¹³C₆-labeled PQS-NO was converted by AqdC_MA, the assay yielded an identical cleavage product and showed the expected mass shift of +6 Da of the labeled homoaromatic ring (see Fig. S3). The presence of the N-hydroxyl moiety in the cleavage product was further confirmed by reduction with zinc powder, leading to a loss of 16 Da and a UV spectrum characteristic of N-acyl-anthranilates. These observations indicate that microorganisms capable of aqdRABC-based AQ degradation can detoxify AQNOs by hydroxylation and subsequent ring cleavage.

FIG 6.

Conversion of PQS-NO by AqdC_MA and LC-MS-based analysis of decomposition products. Samples were reacted with AqdC_MA or heat inactivated AqdC (mock) at 25°C for 1 h and 1,000 rpm. Metabolites were extracted with ethyl acetate and analyzed by LC-ESI-MS. (A) LC chromatograms of reacted sample extracts at 358 nm and 280 nm. (B) Mass and UV spectra of the prominent reaction product (N-(N-hydroxy)-octanoylanthranilic acid). (C) Scheme of HQNO conversion by AQ 3-monooxygenases (gray) and subsequent PQS-NO cleavage by AqdC (black).

General preference of AQ 3-monooxygenases for AQs over AQNOs.

S. aureus is particularly susceptible to the antibiotic effects of AQNO metabolites produced by P. aeruginosa, and it has been demonstrated before that AQNOs play a role in the interaction of both species in natural and probably clinical habitats (1, 5). Despite a slight inhibition of growth, the impact of AQs on S. aureus and most other Gram-positives is less pronounced (34, 35). Hydroxylation of HQNO by S. aureus led to a decrease of the metabolite’s toxicity, and subsequent (abiotic) reduction of PQS-NO to PQS may result in further detoxification (20, 34). From a physiological point of view, it seems plausible that an inactivating enzyme should preferably convert the more toxic AQNO rather than the AQ substrate. The identification of Aqm_SA as monooxygenase active toward HHQ and HQNO allowed us to analyze its substrate preference compared to PqsH- and AqdB-type enzymes.

AQ/AQNO-3-monooxygenases are notoriously difficult to purify and, due to instability, loss of the FAD cofactor, and probably the peculiarity of utilizing a very hydrophobic substrate, exhibit low activity in vitro. We therefore used cell-free protein extracts of recombinant E. coli strains expressing the respective monooxygenase genes for the analysis of AQ/AQNO conversion kinetics. Similar to the signal synthase-type enzymes PqsH_PA and PqsH_HZ11, Aqm_SA and the degradative monooxygenase AqdB_MA showed a preference for AQs over AQNOs (Fig. 7A to D). Moreover, we compared alkyl chain length preferences among the AQ 3-monooxygenases in an assay, where HHQ as well as 2-pentylquinolin-4(1H)-one (PHQ) and 2-nonylquinolin-4(1H)-one (NHQ) competed as substrates for each respective enzyme. PqsH_PA, PqsH_HZ11, and AqdB_MA shared a similar conversion pattern favoring HHQ and NHQ over PHQ (Fig. 7E). Notably, Aqm_SA showed preferential conversion of NHQ.

FIG 7.

Substrate preferences of AQ-3-monooxygenases. Cell-free extract of recombinant E. coli containing PqsH_PA (A), Aqm_SA (B), PqsH_HZ11 (C), or AqdB_MA (D) was adjusted to a final protein concentration of 200 μg ml⁻¹ and assayed for activity in the presence of 400 μM NADH, 2 mM ethanol, 5 U alcohol dehydrogenase, and either 10 μM HHQ or 10 μM HQNO at 30°C and 1,000 rpm. Additionally, recombinant cell extracts were assayed for conversion of 10 μM PHQ, HHQ, and NHQ, which were simultaneously added to reaction mixtures containing the above-mentioned supplements (E). The reactions were stopped within steady state (between 7 and 30 min) by vortexing with ethyl acetate and analyzed by HPLC. Given data and means were obtained in two independent experiments with 2 to 3 technical replicates (A to D) or two independent experiments, respectively (E).

Comparative steady state kinetics of PqsH_HZ11 with HQNO and HHQ.

For a more detailed analysis of the kinetics and substrate preferences, purification of an AQ 3-monooxygenase was highly desirable. However, recombinant Strep-tagged proteins were mainly present in the insoluble fractions, partly as aggregates and partly attached to the membrane fraction. When expressing PqsH_SA and AqdB_MA as fusion proteins with maltose binding protein (MBP), still the majority of protein was found in the membrane fraction, suggesting that both enzymes have high intrinsic affinity for membranes or membrane surfaces. In view of their physiological function, i.e., hydroxylating hydrophobic, predominantly membrane-bound substrates, a cellular location attached to the inner side of the cytoplasmic membrane would make sense and facilitate access to both the cellular pool of NADH and the substrates. We therefore attempted to purify Strep-tagged proteins out of membrane fractions of the cell lysates. However, only PqsH_HZ11 retained catalytic activity after solubilization of membrane fractions with β-dodecyl maltoside (DDM) and purification via Strep-Tactin affinity chromatography. Solubilization of PqsH_HZ11 was achieved with 4% DDM, which underlines its strong affinity for membranes. As shown in SDS-polyacrylamide gels of the purified protein (Fig. 8A), only minor (presumably chaperone-associated) impurities are present in the preparation, which indicates that most of the protein is properly folded.

FIG 8.

Purification and enzyme kinetics of PqsH_HZ11. (A) Coomassie-stained SDS gel of PqsH_HZ11 affinity purification. Determination of kinetic constants of PqsH_HZ11 for HQNO (blue) and HHQ (red) (B) and NADH (green) (C). Enzyme activity was measured in a coupled enzyme assay involving PqsH_HZ11 and AqdC_MA. AqdC_MA rapidly depletes reaction products, enabling spectroscopic determination of NADH consumption.

The activity of PqsH proteins toward HHQ can be determined in an endpoint fluorescence-based assay (24). To establish a continuous spectrophotometric assay for AQ 3-monooxygenases, we used the 3-hydroxy-AQ dioxygenase AqdC_MA as a coupling enzyme. Combining the two reactions allowed us to overcome the issue of spectral overlaps between NADH (λ_max at 340 nm) and PQS (λ_max at 337 nm) (Fig. 6C). We performed steady-state kinetics with PqsH_HZ11 and obtained a k_cat of 0.21 s⁻¹ and a K_m of 0.71 μM (Fig. 8B) for HHQ. Thus, PqsH_HZ11 operates with a catalytic efficiency (0.29 μM⁻¹ s⁻¹) similar to that of MBP-PqsH of P. aeruginosa as reported by Whiteley and coworkers (0.45 μM⁻¹ s⁻¹) (24). PqsH_HZ11 uses NADH as an electron donor with a catalytic efficiency of 3.0 × 10⁻³μM⁻¹ s⁻¹ (k_cat = 0.30 s⁻¹; K_m = 97.5 μM) in our experiment. Similarly, the catalytic efficiency of MBP-PqsH_PA reported by Whiteley and colleagues with NADH was within the same order of magnitude (1.67 × 10⁻³μM⁻¹ s⁻¹) (24).

In order to likewise determine the kinetics of HQNO hydroxylation by PqsH_HZ11, we first assessed the kinetics of PQS-NO cleavage by AqdC_MA. The catalytic efficiency of AqdC_MA with PQS-NO as substrate was at least 0.072 s⁻¹μM⁻¹ (k_cat = 5.2 s⁻¹; K_m = 71.7 μM) (see Fig. S4 in the supplemental material). Since undesired PQS contaminations due to instability of PQS-NO might have affected actual PQS-NO concentrations, we estimate an efficacy even slightly higher than represented by this measurement and concluded that reactivity was sufficient for usage in a coupled assay to test PqsH_HZ11 HQNO hydroxylation. Following the decrease of the combined absorbances of HQNO and NADH at 326 nm, we determined a catalytic efficiency for HQNO of 0.016 s⁻¹μM⁻¹, (k_cat = 0.097 s⁻¹; K_m = 5.9 μM). Taken together, kinetic data clearly demonstrate that AQ substrates strikingly are converted 3-fold faster than AQNO substrates.

DISCUSSION

P. aeruginosa not only regulates the expression of virulence factors via AQ-based quorum sensing, but antibiotically active AQs and AQNOs also contribute to the potential of P. aeruginosa to successfully colonize different niches and hosts and to repress other species in the microbial community (11, 36). While PQS is a potent quorum sensing signal molecule, it is less toxic than other metabolites of the AQ biosynthetic pathway, in particular the heptyl- and nonyl-AQ and -AQNO congeners (34). 3-Hydroxylation of AQs by P. aeruginosa and by rivalling organisms is therefore a key determinant for AQ-based chemical competition in P. aeruginosa-colonized habitats. For P. aeruginosa itself, slow hydroxylation of HHQ, sufficient for activation of quorum sensing circuity, is most desirable, which is the behavior that has been observed in various media and lifestyles (11, 24, 37, 38). AQ-converting bacteria, which cooccur with P. aeruginosa, face the problem that, on the one hand, 3-hydroxylation of HHQ and HQNO to reduce their toxicity is desirable but, on the other hand, accumulation of PQS has to be avoided in order to keep the P. aeruginosa virulence response as low as possible. This aspect is particularly relevant for bacteria harboring an AQ 3-monooxygenase but no PQS-cleaving dioxygenase, such as S. aureus (20, 39).

Our study addressed the question of whether the specific requirements of each of these players are reflected by the biochemical properties of their respective AQ 3-monooxygenases. Unfortunately, biochemical analyses were impeded by the fact that most proteins were unstable and difficult or impossible to purify in active form. However, biotransformations and enzyme assays using cell extracts suggested that all AQ 3-monooxygenases, regardless of their (presumed) physiological role, showed comparable catalytic activities and similar preferences toward AQ substrates. Relatively slow turnover and a low K_m for HHQ were observed for PqsH (24) as well as for PqsH_HZ11 (this work). Since low conversion rates were observed in cell-free extracts for all of the tested enzymes, slow turnover of both AQ and AQNO substrates as well as preference of AQs over AQNOs may be common features of these enzymes. In particular, the low efficiency of degradative enzymes is unlikely to reflect an adaptation to the respective physiological requirement but rather constitutes the inherent limitation of the enzymatic reaction. The highly hydrophobic structure of the AQ scaffold likely causes a major energy barrier for the release of the reaction product, which may set a limit to k_cat. A low K_m value for HHQ and the membrane-attached localization of the enzyme may be a requirement for effectual access to the substrate, which mainly resides in the membrane (1).

The observation that all orthologous enzymes tested follow the same substrate preference, i.e., HHQ over HQNO, was quite unexpected. One might have assumed that degradative enzymes from Gram-positives such as S. aureus, which is highly sensitive to HQNO (11, 12, 20, 36), should be better adapted to convert AQNO substrates, especially as HHQ hydroxylation to PQS indirectly promotes P. aeruginosa virulence and, thereby, toxin production (34, 39). Notably, S. aureus was not only shown to convert HQNO in vivo (20), but its presence potentially compensated for a lack of PqsH activity of P. aeruginosa cystic fibrosis (CF) isolates when grown in coculture (39). Our data suggest that not only iron availability but also hydroxylation of HHQ by the PqsH/AqdB-like ortholog of S. aureus itself have the potential to impact the interplay between S. aureus and P. aeruginosa. The fact that Aqm_SA prefers NHQ over congeners with shorter alkyl chains may reflect an adaptation to preferentially attenuate toxicity of the most toxic compound (40). Strikingly, such an adaptation was not observed regarding selectivity of AQNO versus AQ substrates.

Although the composition of AQ family metabolites produced by P. aeruginosa has been subject to thorough analyses, 3-hydroxy-AQNOs had not been reported previously in suspension cultures, which had led to the assumption that PqsH may be highly specific for AQ substrates. Our data confirm that AQs are intrinsically better substrates for the 3-monooxygenases, possibly due to a lower degree of nucleophilicity of the carbon-3 position in AQNOs and thus lower susceptibility for attack of the hydroxylating flavin C4a-hydroperoxide (41). Alternatively, or additionally, poor solubility of 3-hydroxy-AQNO may limit product release and thus the turnover rate.

Because PqsH_PA, in spite of its preference for HHQ, showed noticeable activity toward HQNO, we wondered why hardly any PQS-NO is observed in P. aeruginosa cultures. We found that PQS-NO production was constant even when cultures were exposed to equimolar concentrations of HHQ, which suggests that the presence of HHQ as a competing substrate does not suffice to suppress HQNO hydroxylation. In contrast, inhibition of multidrug efflux pumps led to an increase in PQS-NO levels, which argues toward a model in which efficient compartmentalization of AQ pathway products suppresses AQNO hydroxylation by the intracellular enzyme. However, the overlapping substrate spectra of RND-type efflux pumps, as well as the nonselectivity of the inhibitor PAβN, preclude assignment of a particular efflux pump to the export of HQNO (33, 42). Interestingly, while Lamarche and Déziel demonstrated MexEF-OprN to export HHQ (43) and MexG, a subunit of MexGHI-OprD, was shown to bind PQS (44), there are no reports on HQNO efflux. Based on catalytic properties of PqsH and efflux pump inhibitor effects, we hypothesize that AQNO export probably constitutes the main mechanism by which P. aeruginosa averts autotoxic effects of the compound family.

We considered that endogenously produced PQS-NO, which is prone to reductive degradation, may also be an alternative intermediate for formation of PQS. However, only trace amounts of PQS were formed out of isotopically labeled, exogenously supplemented HQNO in P. aeruginosa cultures, ruling out that HQNO, via PQS-NO, is an additional precursor of PQS. The inhibitory effect of exogenously supplied PQS-NO on the respiration of P. aeruginosa was similar to that of HQNO, and it also had no apparent effect on quorum sensing. Taken together, AQNO hydroxylation in P. aeruginosa seems to be a side reaction that has no obvious beneficiary effect.

For microorganisms capable of degrading AQs via 3-hydroxylation and subsequent ring cleavage, HQNO hydroxylation by AqdB_MA raised the question of whether PQS-NO is a substrate of the PQS dioxygenase AqdC. Cleavage of the 3-hydroxyquinolone ring by AqdC is accompanied by CO release and should result in full loss of biological activity. AqdC_MA indeed catalyzed PQS-NO conversion to N-(N-hydroxy)-octanoylanthranilic acid.

Catalytic activity of PqsH_PA was studied in detail by Whiteley and colleagues, using an MBP-fusion protein for purification and in vitro assays (24). The authors made use of the solubilizing potential of MBP to purify PqsH out of the soluble fraction of overproducing E. coli and tested catalytic activity in a discontinuous assay based on the fluorometric quantitation of PQS. Our analysis revealed that among the tested orthologs, PqsH_HZ11 was accessible to purification via solubilization and Strep-Tactin affinity purification out of the membrane fraction of E. coli. Since AqdC_MA showed high catalytic activity toward both 3-hydroxy-AQ and -AQNO compounds, we could take advantage of the enzyme for the development of a continuous spectrophotometric assay which monitors NADH oxidation over time. An excess of dioxygenase eliminated 3-hydroxy products and therefore background absorption by these compounds. This setup greatly reduced the experimental effort for determination of Michaelis-Menten parameters. It moreover enabled the kinetic analysis of HQNO monooxygenation and could be used for analysis of further AQ congeners, as long as cleavage by AqdC is possible. In addition, unrelated substrates or inhibitors could be assayed with HHQ as a competitive substrate.

In summary, our data on four different AQ 3-monooxygenases suggest that all of the enzymes tested, regardless of their origin and physiological context, exhibit similar catalytic properties and prefer HHQ over HQNO as substrate. Neither the monooxygenase of S. aureus, an organism highly susceptible to HQNO, nor the “signal synthase” PqsH of P. aeruginosa, have evolved toward divergent specificities for either AQNO or AQ substrates. We identified PQS-NO as a minor but native metabolite of the Pqs biosynthetic pathway. P. aeruginosa appears to favor PQS (over PQS-NO) production by efficient export of HQNO.

MATERIALS AND METHODS

Cloning of plasmids and construction of unmarked deletion mutants.

The pHERD30T plasmid (45) was modified for expression of target proteins as N-terminal Strep-tag II fusion proteins. For this purpose, a synthetic sequence of 69 nucleotides (BioCat, Heidelberg, Germany) (5′-ATGGGTAGCAGCTGGAGCCATCCGCAGTTTGAAAAAGAAAATCTGTATTTTCAGAGCAGCAGTGCCGGC-3′) encoding a Strep-tag II and a tobacco etch virus (TEV)-protease cleavage site was inserted in frame of the NcoI site of pHERD30T by a PCR-based restriction-free method described by van den Ent and Löwe (46), yielding pHERDT30T_Strep.

Chromosomal DNA from P. aeruginosa PAO1, S. aureus strain Newman and Microbulbifer sp. HZ11 was purified (NucleoSpin soil minikit; Macherey-Nagel, Düren, Germany). A synthetic aqdB gene, based on WP_005090569 of Mycobacteroides abscessus subsp. abscessus (DSM 44196) and optimized for codon usage of E. coli (BioCat), was used as a template for aqdB_MA cloning. Flavoprotein monooxygenase genes were amplified with primers listed below (Table 1), generating homologous sequence overlaps for the desired pHERD30T_Strep insertion site. Target genes were inserted into plasmids by PCR amplification of respective products using pHERD30T_Strep as a template as described previously (46).

TABLE 1.

List of primers

Primer no.	Description	Sequence
1	Upstream_fwd	TAAACGACGGCCAGTGCCAGAAGCCTGCAAATGGCAG
2	Upstream_rev	ACAGCCTGAAGACAGAACGTTCCCTCTTC
3	Downstream_fwd	ACGTTCTGTCTTCAGGCTGTGGGGGTGAACC
4	Downstream_rev	GCTCGGTACCCGGGGATCCTCGGATCACCGCCCAGCGC
5	pqsH(p01)_fw	TTTTCAGAGCAGCAGTGCCGGCACCGTTCTTATCCAGGGGG
6	pqsH(p01)_rev	TACCGAGCTCAACCTTGTCGACCTACTGTGCGGCCATCTC
7	aqmSA(p02)_fw	TTTTCAGAGCAGCAGTGCCGGCAAGATAGCAATTATAGGTGCAG
8	aqmSA(p02)_rev	GAGCTCAACCTTGTCGACTTATTTTTCTTTCGATTTATATAAGAATTTAG
9	pqsHHZ11(p03)_fw	GTATTTTCAGAGCAGCAGTGCCGGCAAAACGGACATCGCGATCATAGG
10	pqsHHZ11(p03)_rev	GGTACCGAGCTCAACCTTGTCGACTCAGTCTGTCTTCTCTGTGTGCTTG
11	aqdBMA(p04)_fw	GTATTTTCAGAGCAGCAGTGCCGGCTCTTCTGGTCACGCGGAAGTTG
12	aqdBMA(p04)_rev	GGTACCGAGCTCAACCTTGTCGACTCACGCCGCAGAACGTTTTTTG

Unmarked deletion mutants were constructed by mobilization of a suicide vector pEX18AP (47) into P. aeruginosa (recipient) via biparental mating with E. coli ST18 (donor) (48) with a protocol modified by Panasia et al. (49). Briefly, upstream and downstream regions of the pqsA gene (PA0996) (length of homology regions, 245 and 274 nt, respectively) were amplified by PCR with primers 1 and 2 (upstream) and 3 and 4 (downstream) (Table 1). Additionally, sequence overlaps for each other fragment and for cloning into linearized pEX18AP plasmids were added at the 5′ terminus of the upstream fragment and the 3′ terminus of the downstream fragment, respectively. pEX18AP was linearized by restriction with BamHI and HindIII. Linearized vector and PCR fragments were linked at their target sites using NEBuilder (New England BioLabs), yielding the pEX18AP::ΔpqsA plasmid. The pEX18AP::ΔpqsA was introduced into E. coli ST18 by heat shock transformation. P. aeruginosa PAO1 ΔpqsA and ΔpqsAH mutants resulted from conjugation of the recombinant donor with P. aeruginosa PAO1 wild-type cells or the P. aeruginosa ΔpqsH strain (recipients). P. aeruginosa cells containing pEX18AP::ΔpqsA were selected on carbenicillin 400 μg ml⁻¹, while E. coli was negatively selected for aminolevulenic acid auxotrophy. The plasmid was removed from recombinant P. aeruginosa by consecutive turns of selection on 7% sucrose to remove the pEX18AP plasmid containing the suicide sacB gene. Plasmid-free colonies were screened for mutant genotypes by PCR amplification and sequencing of target regions. Phenotypes were confirmed by high-performance liquid chromatography (HPLC)-based AQ analysis.

AQ biotransformations.

Biotransformations with recombinant E. coli were carried out as described previously (19) with minor modifications. Precultures of recombinant E. coli LMG194 strains were inoculated into 50 ml of fresh LB containing 40 μg ml⁻¹ gentamicin to an optical density at 600 nm (OD₆₀₀) of 0.05 and incubated at 37°C to an OD₆₀₀ of 0.5 to 0.6. Cells were cooled to 30°C, and expression was induced by addition of 0.2% arabinose at an OD₆₀₀ of 0.5 to 0.7. Cultures were incubated overnight (12 to 16 h) and harvested (4,500 × g, 10 min at room temperature). Cell pellets were resuspended and adjusted to a final OD₆₀₀ of 3.0 in fresh LB containing antibiotics and 0.2% arabinose. AQs were added as indicated, and cells were incubated at 30°C and shaking. Samples of 1 ml were taken and immediately extracted twice with 500 μl acidified ethyl acetate (0.1% acetic acid). Organic phases were dried in vacuo, and samples were either stored at −80°C or immediately analyzed by HPLC.

Biotransformations with P. aeruginosa cells were carried out similarly. Unless otherwise stated, P. aeruginosa cultures were incubated overnight in 50 ml LB at 37°C and shaking. A total of 40 μg ml⁻¹ gentamicin was added as required for selection. Heterologous expression was induced by addition of arabinose to a final concentration of 0.2% at an OD₆₀₀ of 0.5 to 0.8 prior to overnight incubation at 37°C. Cells were pelleted at room temperature (6,000 × g, 10 min) and adjusted to an OD₆₀₀ of 3.0 in fresh LB with gentamicin and 0.2% arabinose as required. In the case of efflux pump inhibitor experiments, 100 μg ml⁻¹ PAβN (dissolved in DMSO) or an equal volume of DMSO was added, and cultures were incubated for 45 min at 37°C. AQs were added as indicated and biotransformation and AQ sampling were carried out as described above.

Analytical methods.

Routine analysis of AQ/AQNO metabolites was performed by HPLC using an Agilent 1100 series HPLC system coupled to a diode array detector (Agilent, Santa Clara, CA) and either a 4 × 150 mm (5 μm) or a 3 × 150 mm (3 μm) Eurospher II C₁₈ column (Knauer, Berlin) with respective flow rates of 0.8 or 0.6 ml/min. Separation of metabolites was achieved using a 30/70 to 21/79 acetic acid (0.1%)/methanol gradient or 10/90 citric acid (0.1%)/methanol over a time span of either 15 or 30 min. Chromatogram processing, data analysis, and visualization were carried out using MATLAB R2019a (The MathWorks, Natick, MA) software. HPLC-MS analysis was conducted with a Dionex UltiMate 3000 ultrahigh-performance liquid chromatography (UHPLC) system (Thermo Scientific, Waltham, MA) coupled to an amaZon speed electrospray ionization (ESI) mass spectrometer (Bruker Daltonics, Bremen, Germany) and a 3 × 150 mm (3 μm) Eurospher II C₁₈ column (Knauer, Berlin, Germany).

Bacterial de novo synthesis of AQNOs with recombinant Pseudomonas putida.

(¹³Phenyl)-HQNO and PQS-NO were produced by bacterial biotransformation out of (¹³phenyl)-anthranilic acid (Merck) and anthranilic acid, respectively (see Fig. S1 in the supplemental material).

For production of (¹³phenyl)-HQNO, a preculture of recombinant P. putida KT2440 carrying plasmids with pqs biosynthesis genes pqsABCDL (P. putida KT2440[pBBR1MCS2::pqsABCD][pME6032::pqsL]), grown overnight at 30°C in LB, was diluted to an OD₆₀₀ of 0.05 in M9 medium with 0.4% glucose as a carbon source. Cultures were grown at 30°C with shaking and 0.5 mM isopropyl-β-d-thiogalactopyranoside (IPTG) was added at an OD₆₀₀ of 0.5. Expression was carried out overnight at 30°C. Cells were harvested (6,000 × g, 10 min) and washed in 0.9% NaCl. Cells were resuspended to an OD₆₀₀ of 0.5 in M9 medium containing 0.4% glucose, 0.5 mM IPTG, 0.25 mM (¹³phenyl)-anthranilic acid, 0.5 mM octanoic acid, kanamycin (50 μg ml⁻¹), and tetracycline (40 μg ml⁻¹). Cultures were incubated for 6 h at 30°C at 160 rpm shaking. AQs and AQNOs were extracted with acidified (0.1% acetic acid) ethyl acetate. Organic phases were dried in vacuo and subjected to preparative HPLC (Agilent 1100 series [a G1315B diode array detector and a G1364C analytical fraction collector]). Samples were separated by a reverse-phase C₁₈ column (Eurospher II 100-5 C₁₈, 250 × 8 mm) using two isocratic steps of 40/60 0.1% aqueous citric acid/methanol (0.1% citric acid), followed by 10/90 citric acid (0.1%)/methanol with 0.1% citric acid with a flowrate of 2.8 to 3.0 ml/min. Impurities in pooled (¹³phenyl)-HQNO fractions were removed by isocratic separation 25/75 citric acid 0.1%/methanol (0.1% citric acid) (Eurospher II 100-5 C₁₈, 250 × 8 mm) with a flow rate of 2.6 ml/min (t_ret(¹³phenyl)-HQNO, 16.8 min). Methanolic fractions were dried as described above, yielding a white powder that was dissolved in DMSO and stored at −20°C for further use. The sample contained 98% pure (¹³phenyl)-HQNO as determined by HPLC/ESI-MS.

For production of PQS-NO, overnight cultures of P. putida carrying plasmids encoding the pqsABCDLH genes (P. putida KT2440[pBBR1MCS2::pqsABCD][pME6032::pqsL][pHERD30T::pqsH]) were diluted to an OD₆₀₀ of 0.05 in fresh LB containing kanamycin (50 μg ml⁻¹), tetracycline (20 μg ml⁻¹), and gentamicin (40 μg ml⁻¹). Cultures were grown at 30°C, and 0.2% arabinose was added at an OD₆₀₀ of 0.5 followed by addition of 1 mM IPTG after another 30 min. A total of 250 μM anthranilic acid and 500 μM octanoic acid were added, and expression was carried out overnight at 30°C. Metabolites were extracted as described above and subjected to preparative HPLC (Agilent 1100 series [G1315B diode array detector and a G1364C analytical fraction collector, Eurospher II 100-5 C₁₈, 250 × 8 mm]). PQS-NO eluted after 10.4 min during isocratic separation with 22/78 citric acid (0.1%)/methanol (0.1% citric acid) at a flow rate of 2.8 ml/min. Solvents were removed as described above. The identity of PQS-NO was confirmed by HPLC/ESI-MS and the purity was 95%.

Oxygen consumption assays.

Overnight cultures of the P. aeruginosa ΔpqsAH strain, grown in LB, were harvested and resuspended in buffer containing 20 mM HEPES pH 7.8 and 75 mM Na₂SO₄. P. aeruginosa cells were disrupted by sonication, and insoluble debris was removed by centrifugation (10 min, 12,000 × g, 4°C). Cell extract supernatants of the P. aeruginosa ΔpqsAH strain were adjusted to a protein concentration of 2 mg ml⁻¹ in 20 mM HEPES pH 7.8 and 75 mM Na₂SO₄ and transferred to a 1-ml cell of a Clark electrode. Respiration was initiated by addition of 1 mM NADH, and linear oxygen consumption established within 20 s. HQNO or PQS-NO were added to final concentrations between 0.004 and 300 μM, and decrease in respiration was further followed. Linear rates of oxygen consumption were normalized to solvent-treated controls and expressed as a function of AQNO concentrations. Nonlinear dose-response curves were plotted with MATLAB R2019a (The MathWorks, Natick, MA).

Expression and purification of enzymes.

E. coli LMG194(pHERD30T::pqsH_HZ11) was inoculated into 2× yeast extract-tryptone (YT) broth (16 g/liter tryptone, 10 g/liter yeast extract, 5g/liter NaCl, with 0.5% glycerol [vol/vol]) and grown to an OD₆₀₀ between 2 and 3 at 37°C and 160 rpm. Cultures were diluted to an OD₆₀₀ of 0.05 fresh 2× YT (0.5% glycerol) in 3 liters and grown to an OD₆₀₀ of 0.6. The temperature was decreased to 18°C, and expression was induced with 0.2% arabinose after a further 30 min (OD₆₀₀ = 0.7 to 0.8). Expression was carried out overnight (15 to 16 h at 18°C, 120 rpm).

Cells were harvested and resuspended in lysis buffer (20 mM Tris-HCl pH 7.8, 150 mM NaCl, 1 mM EDTA, 3 μM FAD, lysozyme 0.5 mg ml⁻¹) and sonicated on ice. Cell debris was removed by centrifugation (12,000 × g, 4°C), and the soluble fraction was subjected to a second centrifugation step (14,000 × g, 4°C). Membrane fractions were pelleted from supernatants by centrifugation at 100,000 × g (4°C) for 90 min. Membrane pellets were either frozen in liquid nitrogen and stored at −80°C or directly resuspended in solubilization buffer (20 mM Tris-HCl pH 7.8, 150 mM NaCl, 1 mM EDTA, 3 μM FAD, 4% DDM) (10 ml/g membrane pellet). Membranes were homogenized by 20 strokes with a Dounce homogenizer. Solubilization of proteins was supported by 2 min of mild sonication (Bandelin Sonopuls; 3-min turns of 1 s pulse and 2 s break [approximately 1,400 kJ]) followed by stirring on ice overnight. Solubilized proteins were separated from membranes by centrifugation at 100,000 × g (4°C) for 90 min, and the supernatant was diluted to a final DDM concentration of 1% (wt/vol). The solution was filtered (0.45-μm pore) and subjected to affinity chromatography (Strep-Tactin XT Superflow, high-capacity resin). Immobilized protein was washed with 20 to 30 column volumes of wash buffer (20 mM Tris-HCl pH 7.8, 150 mM NaCl, 1 mM EDTA, 3 μM FAD, 0.02% DDM) and eluted by consecutive applications of 0.5 column volumes of elution buffer (20 mM Tris-HCl pH 7.8, 150 mM NaCl, 1 mM EDTA, 3 μM FAD, 0.02% DDM, 50 μM biotin). Protein-containing fractions were pooled and desalted (5 ml or 20 ml HiTrap desalt; GE Healthcare) against 20 mM Tris-HCl pH 7.8, 150 mM NaCl, 1 mM EDTA, 3 μM FAD, 0.02% DDM and concentrated. Aliquots were frozen in liquid nitrogen and stored at −80°C.

AqdC_MA was heterologously expressed and purified via immobilized metal ion affinity chromatography as previously described (19, 22).

Activity assays with recombinant cell extract supernatants.

Recombinant E. coli LMG194 strains producing N-terminally Strep-tagged fusion proteins were harvested by centrifugation and resuspended in buffer (20 mM HEPES, 75 mM Na₂SO₄, 1 mM EDTA, 2 μM FAD) at 4 ml/g cell pellet. Cells were lysed by ultrasonication as described above, and the lysate was centrifuged (30,000 × g, 30 min, 4°C). Protein concentrations were quantified by Bradford assay (50). Protein aliquots were frozen in liquid nitrogen and stored at −20°C until further use.

Cell extract supernatants were gently thawed on ice for preparation of assay mixtures. Each reaction mixture contained 20 mM HEPES pH 7.8, 75 mM Na₂SO₄, 2 μM FAD, 400 μM NADH, 5 U alcohol dehydrogenase (Thermo Scientific), 2 mM ethanol, 10 μM HHQ or HQNO, and 200 μg ml⁻¹ recombinant cell extract supernatant. Reactions were started by addition of NADH and HHQ or HQNO to prewarmed reaction mixtures. Vials were incubated at 30°C and 1,000 rpm, and reactions were stopped within steady state (after 7 min for HHQ-containing samples and 21 min for HQNO-containing samples). AQs/AQNOs were extracted twice with ethyl acetate as described above. Solvents were dried in vacuo, and substrate conversion was analyzed by HPLC.

Coupled enzyme assay.

For each reaction mixture, AQs or AQNOs were added from DMSO stocks (final concentrations of DMSO 1%) to 20 mM HEPES pH 7.8, 75 mM Na₂SO₄, 1.25 μM FAD, 250 μM NADH, 10 μg ml⁻¹ AqdC_MA, and reactions were started by addition of 6.9 μg ml⁻¹ PqsH_HZ11. NADH oxidation was monitored at 340 nm for reaction mixtures with HHQ (ε(NADH) = 6,200 M⁻¹cm⁻¹) or at 326 nm (ε(NADH) = 5,783 M⁻¹cm⁻¹, ε(HQNO) = 4,689 M⁻¹cm⁻¹) for HQNO kinetics, respectively. Turnover rates were plotted against substrate concentrations. Michaelis-Menten curves were fitted by nonlinear regression.