Antibiotic adjuvant activity revealed in a photoaffinity approach to determine the molecular target of antipyocyanin compounds

Abstract

Infections with Pseudomonas aeruginosa are a looming threat to public health. New treatment strategies are needed to combat this pathogen, for example by blocking the production of virulence factors like pyocyanin. A photoaffinity analog of an antipyocyanin compound was developed to interrogate the inhibitor’s molecular mechanism of action. While we sought to develop antivirulence inhibitors, the proteomics results suggested that the compounds had antibiotic adjuvant activity. Unexpectedly, we found that these compounds amplify the bactericidal activity of colistin, a well-characterized antibiotic, suggesting they may represent a first-in-class antibiotic adjuvant therapy. Analogs have the potential to not only widen the therapeutic index of cationic antimicrobial peptides like colistin, but to be effective against colistin-resistant strains, strengthening our arsenal to combat P. aeruginosa infections.

Graphical abstract

Antibiotic-resistant strains of Pseudomonas aeruginosa are an increasingly prevalent cause of secondary infections, where patients who are critically ill, have compromised immune systems or require implanted medical devices are especially at risk.^1–3 In a recent study of hospitalized patients with COVID-19, half of the patients who did not survive had contracted a secondary infection.⁴ The WHO and CDC have both classified resistant strains of P. aeruginosa as looming public health threats that urgently require new treatment strategies.^{2, 3} Blocking the production of virulence factors during an infection is one possible approach to treat infections. Phenazines, redox-active small molecules produced by P. aeruginosa, are an important set of virulence factors. Pyocyanin and other phenazines have detrimental effects on the host including disruption of both the cellular redox balance and the immune response.⁵ These compounds also upregulate additional virulence genes through detection by SoxR and support biofilm development, including optimal redox cycling in anoxic conditions.⁶, ⁷ These attributes make phenazine biosynthesis a potential antivirulence therapeutic target.⁸

Previously, we identified a set of compounds including 1 (Figure 1C) that were potent inhibitors of pyocyanin production, however, their molecular target and mechanism of action are unknown.⁹ Work in the area has primarily described regulation of phenazine production in P. aeruginosa at the transcriptional level through two-component systems as well as quorum-sensing transcriptional regulators like RhlR.^10–14 Quorum sensing is the strategy that bacteria use to coordinate group behaviors, including virulence and biofilm formation, through the production, secretion and detection of small molecule signals.¹⁵ The antipyocyanin activity of compound 1 is not consistent with inhibition of quorum sensing by direct inhibition of RhlR or LasR, based upon assays of heterologous expression of the rhl and las systems in E. coli, nor are the transcript levels consistent with known inhibition of quorum sensing regulators.⁹ In addition, transcript levels of the genes required for pyocyanin synthesis (phzA-G, phzM, phzS) are unaffected by treatment with 1. Ruling out these mechanisms, we hypothesize a post-transcriptional mode of action.⁹ These results imply that inhibitor 1 may be directly binding and inhibiting one of the phenazine biosynthetic enzymes or that the inhibitor is acting on a target that impacts post-transcriptional regulation of phenazine production. In this work, we sought to uncover the molecular target of the compounds by investigation of direct inhibition of the phenazine biosynthetic enzymes as well as through the development of a photoaffinity probe to identify an alternative molecular target.

Figure 1. Phenazine production.

(A) Biosynthesis of pyocyanin in P. aeruginosa. Adapted from references 16-19. (B) Production of phenazines upon treatment with compound 1. The production of PCA and pyocyanin (PYO) in wild-type P. aeruginosa PA14 (WT PA) is inhibited by compound 1, while heterologous PCA production in the recombinant E. coli (EC) strains are not inhibited. WT P. aeruginosa PA14 and the phz-containing E. coli strains were grown in MOPS-LB in the presence of DMSO or 1 (100 μM). PCA and PYO levels in cell-free supernatants were quantified by HPLC. Error bars represent the standard deviation of the mean of 3 replicates across 1 day (EC) or 15 replicates across 5 days (PA). Statistical significance was determined by t tests comparing the DMSO and compound 1 treatments within each strain. ****, p < 0.0001. (C) Inhibitor 1 blocks 99% production of pyocyanin at 100 μM in WT P. aeruginosa PA14 by the UV/Vis assay. Activity data from reference 9.

Initially, we assessed whether compound 1 could be inhibiting one of the biosynthetic enzymes required for the synthesis of pyocyanin. Enzymes PhzA-G convert chorismate to phenazine-1-carboxylic acid (PCA, Figure 1A).^{16, 17} These enzymes are encoded by the nearly identical operons phzA1-G1 (phz1) and phzA2-G2 (phz2).¹⁶ PCA is then converted to pyocyanin by PhzM and PhzS.^{16, 18, 19} In wild-type P. aeruginosa, treatment with compound 1 not only inhibits pyocyanin production by 95%, but PCA levels also decrease by 98% upon treatment with the inhibitor (Figure 1B), suggesting that inhibition occurs upstream of PCA synthesis as it is unlikely that direct inhibition of PhzM and/or PhzS would also decrease PCA levels.

To interrogate direct inhibition of PhzA1-PhzG1 or PhzA2-PhzG2, we constructed heterologous systems where the phz1 or phz2 biosynthetic gene clusters from P. aeruginosa PA14 were expressed in E. coli. A decreased level of PCA production after treatment with inhibitor 1 would support one of the upstream biosynthetic enzymes as the target. However, treatment of either heterologous strain with 1 did not decrease PCA production (Figure 1B), making PhzA-G unlikely targets. The phz1 and phz2 operons also have different post-tranlational regulation in P. aeruginosa.¹⁰ We also investigated mutants lacking either phz1 or phz2, as well as mutants lacking each phz operon as well as the phzH, phzS and phzM genes, which encode the biosynthetic enzymes downstream of PCA.¹⁰ In all cases, we observed reduction of the phenazines produced upon treatment with inhibitor 1 (Figure S1). These results suggest the inhibitors may instead be interacting with a target involved in the joint post-transcriptional regulation of phenazine production of phz1 and phz2.

To identify that molecular target, we next pursued a photoaffinity labeling approach, where incorporation of a photoreactive group leads to the formation of a covalent bond with the small molecule’s binding partner upon irradiation with the appropriate wavelength of light. This strategy is advantageous over a pull-down approach because the biological activity of the photoaffinity labeled (PAL) analog can be assessed in a biological assay to ensure that the modifications do not make the compound inactive.^{20, 21} In addition, PAL analogs can better capture low affinity interactions with target proteins, and the experiment can be conducted in whole cells rather than cell lysate, to better reflect the true environment of the cell.^{20, 21}

In our initial structure-activity studies, we determined that the linker region between the two aryl groups was permissive to changes (4, Figure 2A).⁹ For example, the oxygen in ether 1 could be replaced with a methylene (2) with minimal loss of activity. The addition of a ketone in this region (3) was also tolerated. We designed PAL analog 5 to incorporate a diazirine functional group on the linker in the place of the ketone. The diazirine is only moderately larger than the ketone, so we predicted that analog 5 would maintain antipyocyanin activity and serve as an effective photoaffinity label.

Figure 2. PAL target.

(A) Design of PAL target based on structure-activity relationships. (B) Synthesis of PAL 5. Inhibition activity from the UV/Vis assay. Activity data from reference 9.

The label was synthesized in six steps. 1-Ethynyl-3-fluorobenzene (7, Figure 2B) was deprotonated with lithium bis(trimethylsilyl)amide (LiHMDS) and combined with methyl glutaryl chloride (6). The desired product (8) was isolated in low yield. The double addition product of the 1-ethynyl-3-fluorobenzene into both the acid chloride and the methyl ester was a major byproduct. Despite the low yield, ynone 8 was brought onward in a hydrogenation of the alkyne followed by saponification of the ester to afford keto acid 9. The ketone (9) was transformed into a diazirine in two steps.^{22, 23} Finally, acid 10 was coupled to the aminopyridine head group (11) to furnish the PAL analog (5). To be useful as a photoaffinity probe, PAL analog 5 must maintain antipyocyanin activity. While the PAL analog (5) is not as potent as inhibitor 1 (83% vs. 99% inhibition at 100 μM by the UV/Vis assay), it is still an effective inhibitor of pyocyanin activity, making it a viable photoaffinity probe. Like inhibitor 1, analog 5 also does not affect bacterial growth at concentrations of 100 μM and lower.

We next tested the PAL analog for photolabeling activity using peptides resulting from a trypsin digest of E. coli Ptsl as a negative control to assess non-specific background labeling in an in vitro experiment. The peptides were treated with PAL analog 5 in the presence of 365 nm light for 15 min at 0 °C. Only a low level of crosslinking was observed. Two distinctive fragmentation patterns of the PAL-peptide adduct were identified for use in the P. aeruginosa proteomics study. The major PAL product observed was reaction of the carbene with water. If there is no specific binding interaction, we would also expect to see this result in the P. aeruginosa experiments.

To identify the potential antipyocyanin target, P. aeruginosa was grown for 17 h in the presence of 100 μM PAL analog 5. A portion of the culture was then irradiated with 365 nm light for 15 min at 0 °C to crosslink the PAL analog to the protein(s) to which it was bound. The cells were lysed and the proteins were separated by SDS-PAGE. The gel lane was then excised into 12 pieces. Each piece was treated with trypsin before analysis by Nano-UPLC-MS. SEQUEST and Mascot were used to identify proteins that had been modified with PAL analog 5.^{24, 25} Five high confidence hits were identified (Table 1). PA14_18350 was identified in 58 peptide spectrum matches, and PA14_30820 and PA14_58910 were detected in 2 peptide spectrum matches, while PA14_37040 and PA14_68550 were each found in a single peptide spectrum match.

Table 1.

Photoaffinity protein hits.

Protein	Description	Sequence*	# PSM†
PA14_18350 (ArnA)	Bifunctional UDP-glucuronic acid decarboxylase/UDP-4-amino-4-deoxy-L-arabinose formyltransferase	QLgEELLR	58
PA14_30820	Methyl-accepting chemotaxis transducer	ASDEIaQR	2
PA14_58910	Chromosome partitioning-like protein	IPNVQIVINCLDQTNDSR	2
PA14_37040	Chaperone CupA2	ASVVVTGTR	1
PA14_68550	Probable LysR-type transcriptional regulator	MNIQTFDLNLLR	1

^*A lower-case amino acid in the sequence indicates the site of photolinking.

^†PSM = peptide spectrum matches.

To confirm the potential targets for the source of antipyocyanin activity, we tested a transposon mutant of the gene encoding each protein target in the presence and absence of inhibitor 1 (Figure S2).²⁶ We would expect that a strain lacking the molecular target would be insensitive to inhibitor 1. In addition, we predicted that the mutant strain may display aberrant levels of pyocyanin in control studies. However, all of the tested transposon mutants were sensitive to the inhibitor, suggesting that none of the protein interactions identified were responsible for the antipyocyanin activity.

Although the molecular target of antipyocyanin was not identified in this approach, the identified proteins suggested that PAL analog 5 may be interacting with proteins that are important for colistin resistance. Colistin is a cationic antimicrobial peptide that is used as a last resort therapy for multidrug resistant infections of P. aeruginosa and other Gram-negative bacteria.^{27, 28} To enter the cell, the antibiotic has an initial electrostatic interaction with the negatively-charged lipopolysaccharide (LPS) of the bacterium’s outer membrane, followed by self-promoted uptake through disruption of the outer membrane.^{29, 30} Colistin-resistant P. aeruginosa modifies its outer membrane through enzymes encoded in the arnBCADTEF operon to become less negatively charged, preventing the initial electrostatic interaction with colistin and allowing the bacteria to evade the antibiotic.^31–33 The top hit, PA14_18350 (PA3554), encodes the bifunctional enzyme ArnA, which performs both the first committed step and the third step in the modification of the LPS. Loss or inhibition of ArnA should prevent the modification of the outer membrane.^31–35

The arnB operon is upregulated by a number of two-component systems.^36–40 A second protein identified in the proteomics study, PA14_30820 (PA2573), is a probable methyl-accepting chemotaxis protein, whose mutant in P. aeruginosa PAO1 is more susceptible to colistin, likely acting at least in part through downregulation of pmrA.⁴¹ PmrA is a response regulator that upregulates the arnB operon in concert with the sensor kinase PmrB.^{37, 42} Disruption of the PmrAB two-component system has also been shown to enhance colistin susceptibility in P. aeruginosa.^{37, 43, 44}

Due to its close structural similarity to PAL analog 5, we investigated the ability of 2 (Figure 2A) to act as an antibiotic adjuvant. Colistin susceptibility of wild-type P. aeruginosa in the presence of 50 μM inhibitor 2 was tested (Figure 3A). In the presence of 2, the amount of colistin required to kill the bacterium is lowered, demonstrating that the inhibitors do negatively impact the bacterium’s ability to survive treatment with colistin. While the static minimum inhibitory concentration of compound 2 was not lower than colistin alone (1 μg/mL), compound 1 was much more potent (0.25 ug/mL). To further investigate the dose-dependence of inhibitor 2, we did a titration of the inhibitor at a constant sublethal colistin concentration (Figure 3B). We found that the compound had an IC₉₅ of 23 μM in the presence of 0.375 μg/mL colistin. Treatment of 2 alone at concentrations up to 100 μM does not affect the growth of the bacterium, supporting the adjuvant theory.

Figure 3. Compound 2 increases susceptibility to colistin.

(A) P. aeruginosa PA14 was treated for 17 h with increasing amounts of colistin in the presence of no added inhibitor (DMSO, circles) or 50 μM inhibitor 2 (triangles). (B) P. aeruginosa PA14 was treated with increasing amounts of inhibitor 2 in the presence of no added colistin (water, circles) or 0.375 μg/mL colistin (triangles). (A-B) Cell viability was tracked by OD600. Error bars represent the standard deviation of the mean of three replicates.

In this work, we pursued the identification of the molecular target of inhibitors including 1, which decrease pyocyanin production in P. aeruginosa. Previous studies suggested that the inhibition was happening in a post-transcriptional manner.⁹ We hypothesized that the small molecules could be inhibiting the biosynthetic enzymes directly or that they could be impacting the synthesis at an uncharacterized post-transcriptional target. By construction of heterologous systems in E. coli with each phz operon, we were able to directly examine the biosynthetic enzymes that produce PCA, a precursor to pyocyanin. While in P. aeruginosa we observe a reduction in both pyocyanin and PCA levels upon treatment with inhibitor 1, in the heterologous strain we saw no decrease in PCA. These results suggest that PhzA-G are not being inhibited directly. The results also argue against direct inhibition of PhzM or PhzS. If either of those enzymes were inhibited, we would expect PCA levels to increase or stay the same because PCA is a precursor to pyocyanin. The potential post-transcriptional target also appears to be involved in joint regulation of both phz operons or flux through both pathways, as P. aeruginosa mutants lacking either phz1 or phz2 still displayed reduced phenazine production after treatment with inhibitor 1.

We next used a photoaffinity approach in an attempt to directly identify the molecular target. Analog 5 maintained antipyocyanin activity, despite incorporation of a diazirine photoaffinity tag. While we identified a number of high confidence hits, none appeared to be responsible for the antipyocyanin activity based on pyocyanin assays with transposon mutants of the potential targets. Among the hits, PA14_30820 was the most promising as the antipyocyanin target. In the P. aeruginosa PAO1 ortholog, the probable methyl-accepting chemotaxis protein (PA2573) works in concert with PA2572 (PA14_30820) to regulate genes involved in virulence and antibiotic resistance.⁴¹ Mutation of PA2573 was reported to decrease pyocyanin production in P. aeruginosa PAO1 through an unknown mechanism,⁴¹ but we did not observe a similar phenotype in the PA14_30820 mutant in P. aeruginosa PA14 (Figure S2).None of the other proteins in the photoaffinity experiment are known to affect pyocyanin. ArnA (PA14_18350) is involved in the LPS modification pathway. CupA2 (PA14_37040) is a chaperone involved in assembly of the Pel-dependent pellicle.^45–47 Neither PA14_58910 nor PA14_68550 have been characterized, but are predicted to be a chromosome partitioning-like protein and a LysR type transcriptional regulator, respectively.⁴⁸ It is possible that the true antipyocyanin target is at low abundance and could not be detected above our experimental background. We will pursue the incorporation of a bioorthogonal terminal alkyne into the photoaffinity probe for click chemistry and streptavidin pull-down enrichment in future work.

While the desired antipyocyanin target was not found, the hits identified, including ArnA (PA14_18350) and PA14_30820, hinted at a potential off-target effect of the inhibitor on colistin resistance. Indeed, when we assayed P. aeruginosa for colistin susceptibility, we saw increased sensitivity to the antibiotic in the presence of inhibitors 1-2 in colistin-susceptible P. aeruginosa. Others have identified colistin or polymyxin B adjuvants in P. aeruginosa, including those that target PmrAB signaling,⁴³ disrupt the outer membrane^49–53 or function through unknown mechanisms.^54–56 Inhibitors of ArnT have been previously characterized, but no other successful inhibitors of the enzymes encoded by the arnB operon have been identified.^57–59 In addition, no inhibitor of PA14_30820 has been identified, suggesting that compounds 1-2 and analogs could be first-in-class inhibitors of ArnA and/or PA14_30820. Should the inhibitor prove to have an impact on colistin-resistant strains, it would be a promising lead for adjuvant therapies to increase the effectiveness of the drug. High concentrations of colistin during treatment can lead to undesired side effects in patients including nephrotoxicity and neurotoxicity.^{60, 61} Widening the therapeutic index with the use of an adjuvant would be of great utility. The use of inhibitors 1-2 and analogs to increase colistin sensitivity will be pursued in future studies.

Methods

Strains and media.

The E. coli strains were grown with shaking at 37 °C in Luria broth (LB; BD Difco) or MOPS-LB in the presence of 25 μg/mL chloramphenicol (CAM).⁶² P. aeruginosa PA14 strains were grown with shaking at 37 °C in LB or MOPS-LB.⁶² The P. aeruginosa transposon mutants were from the ordered transposon library.²⁶

Construction of pN15C200_phzA1G1.

The phzA1G1 gene cluster was amplified from genomic DNA obtained from Pseudomonas aeruginosa UCBPP-PA14 using primers p001 and p002. Genomic DNA was prepared using the MasterPure Gram Positive DNA Purification Kit (Lucigen). Amplification of the phzA1G1 gene cluster from P. aeruginosa genomic DNA template was performed using Phusion High-Fidelity DNA Polymerase (NEB) and a touchdown PCR protocol with an initial annealing temperature of 65 °C, a ramp rate of −0.5 °C per cycle, and a 6.5 minute extension time. Amplification of promoter P_BfP1E6 from pWW3365 was carried out using primers p003 and p004. Plasmid pN15C200_phzA1G1 was assembled via Gibson assembly (Gibson Assembly Master Mix, NEB) using electrocompetent E. coli S17-1λpir as the host strain and combining the phzA1G1 amplicon (amplified using p001+p002), the PBfP1E6 amplicon (amplified using p003+p004), and the pN15C200 vector digested with BamHI and XhoI. Each DNA fragment was gel purified prior to combining for Gibson assembly. Transformed E. coli cells were allowed to recover at 37 °C for 1 hour prior to plating on LB agar plates supplemented with CAM at 25 μg/mL. Colonies were picked and cultured overnight in LB in the presence of 25 μg/mL CAM and were miniprepped and screened for proper assembly size via restriction digest with KpnI. The plasmid from a clone giving the correct restriction banding pattern was sequenced via sanger sequencing to confirm the entire phzA1G1 insert along with P_BfP1E6 promoter using primers p005-p018. The NEB Turbo strain was generated by transforming competent NEB Turbo cells with the phz1 construct using the 5 minute transformation protocol (C2984) and plating the resulting transformants on LB agar plates supplemented with CAM at 25 μg/mL

Construction of pN15C200_phzA2G2.

The phzA2G2 construct was assembled in an analogous manner to the phzA1G1 construct described above with the following changes. The phzA2G2 gene cluster was amplified using primers p019 and p002, while promoter P_BfP1E6 was amplified with primers p003 and p020. After Gibson assembly, the plasmid was transformed into chemically competent E. coli NEB Turbo as the host strain. The plasmid from a clone giving the correct KpnI restriction banding pattern was sequenced via sanger sequencing to confirm the entire phzA2G2 insert and P_BfP1E6 promoter using primers p005, p008-p009, p013, p016-p018, and p021-p027.

E. coli phenazine assay.

Cultures of the E. coli strain carrying the phz vector were grown for 18 h in LB with chloramphenicol (34 μg/mL) to exponential phase at 37 °C with shaking at 300 rpm. The cells were centrifuged at 14.1 x 1000 g for 2 minutes, and the pellet was resuspended in MOPS-LB. The cells were pelleted and resuspended two more times. A 5 mL sample of fresh MOPS-LB medium with chloramphenicol (34 μg/mL) was inoculated to an OD600 of 0.05 with the prepared E. coli culture. The subcultures were treated with 100 μM of compound 1 or DMSO in triplicate and grown for 18 h at 37 °C with shaking at 300 rpm. Samples were centrifuged at 14.1 x 1000 g for 15 min, and the supernatants were filtered through 0.22 μm syringe-driven filters.

P. aeruginosa phenazine assay.

Cultures of WT P. aeruginosa PA14 were grown for 17 h in LB to exponential phase at 37 °C with shaking at 300 rpm. The cells were centrifuged at 14.1 x 1000 g for 2 minutes, and the pellet was resuspended in MOPS-LB. The cells were pelleted and resuspended two more times. A 5 mL sample of fresh MOPS-LB medium was inoculated to an OD600 of 0.05 with the prepared P. aeruginosa culture. The subcultures were treated with 100 μM of compound 1 or DMSO in triplicate and grown for 17 h at 37 °C with shaking at 300 rpm. Samples were centrifuged at 14.1 x 1000 g for 15 min, and the supernatants were filtered through 0.22 μm syringe-driven filters.

Phenazine quantification by HPLC.

Samples of 30 μL were loaded onto a Cogent Bidentate C18 reversed-phase column (particle diameter 4μm, 2.1 x 150 mm dimension, 100 Å pore size). HPLC based separation was carried out using Milli-Q water + 0.01 % TFA (v/v) (solvent A) and acetonitrile + 0.01 % TFA (v/v) (solvent B). Samples were eluted using a reversed phase gradient method at a flow rate of 0.4 mL/min; gradient: 0-2 min 5% B; 2-8 min to 20% B; 8-30 min to 90% B; and 30-32 min to 5% B, with 10 min post-equilibration time. Calibration standards were prepared at eight concentration levels for both pyocyanin (10, 25, 50, 75, 100, 150, 200, 250 μg/mL) and phenazine-1-carboxylic acid (10, 25, 50, 100, 200, 300, 400, 500 μg/mL) using initial condition mobile phase (5 %B) as diluent. Wavelength 388 nm was used for phenazine detection.

Generic peptide photoaffinity labeling.

To assess the photocrosslinking efficiency of the compound 5, 1 μL of 10 μM solution of 5 was combined with 5 μL of a model peptide (tryptic peptide from Ptsl of E. coli) at 25 μM in a glass insert. The sample was irradiated with 120 Watt, 365 nm UV light at 0 °C for 15 min. UV-irradiation of (1) only the peptide and (2) the peptide and compound 3 were used as controls. In addition, three non-irradiated samples were used as controls: (1) peptide, (2) peptide and 5, and (3) peptide and 3. The samples were then analyzed by MS and tandem MS by Nano-UPLC-MS.

P. aeruginosa photoaffinity labeling.

P. aeruginosa PA14 was grown at 37 °C in 5 mL LB on a culture rotator for 17 hours in the presence of 100 μM compound 5, 100 μM compound 3, or the same volume of DMSO. A 1 mL aliquot of each bacterial culture was irradiated with 120 Watt, 365 nm UV light at 0 °C for 15 minutes. As additional controls, samples of the P. aeruginosa treated with compounds 5 or 3 and the untreated control were also analyzed that were not subjected to the UV treatment. The cultures were then pelleted, and the supernatant was removed. The cell pellets were dissolved in 10X sample buffer and DTT. The cells were than lysed at 100 °C. The cell lysate was separated by SDS PAGE gel electrophoresis. The gel was washed and subsequently stained with Coomassie. The gel bands were cut and diced appropriately for MS analysis. Specifically, the gel lane of interest (cell lysate of compound 5 cross-linked to P. aeruginosa) was cut into 12 slices, and the proteins were digested with trypsin. Nano-UPLC-MS experiments were performed for 3 hours on the digested peptides of each slice. After MS and tandem MS data acquisition, SEQUEST²⁴ and Mascot²⁵ search engines were used to identify high-scoring proteins that compound 5 could have covalently modified based on the MS and MS/MS of the peptides, in which the adduct was defined as a +366 Da post-translational modification mass shift parameter.

Pyocyanin absorbance assay.

The protocol of O’Loughlin and coworkers⁶³ was used with the following modification. Pyocyanin production is dependent on cell density, so the data were normalized for optical density at 600 nm.

Colistin susceptibility assay.

Overnight cultures of P. aeruginosa PA14 were backdiluted into fresh LB to an OD600 of 2.5 x 10⁻³. Using polypropylene 96 well plates (Corning 3879), freshly prepared colistin was assayed from 0.125-1.00 μg/mL, taking care to minimize transfers of colistin-containing solutions. The potential antibiotic adjuvant was tested at 50 μM. After 17 h of aerobic growth at 37 °C, 300 rpm, the absorbance at 600 nm was recorded.

Static minimum inhibitory concentration assay.

Overnight cultures of P. aeruginosa PA14 were backdiluted into fresh LB to 5 x 10⁵ CFU/mL. Using polypropylene 96 well plates (Corning 3879), freshly prepared colistin was assayed from 0.125-2.00 μg/mL, taking care to minimize transfers of colistin-containing solutions. The potential antibiotic adjuvant was tested at 50 μM, with DMSO alone in the negative control. The assay was done in triplicate. The plate was sealed with Glad Press-N-Seal and incubated for 17 h at 37 °C. The minimum inhibitory concentration was determined by visual inspection of the lowest concentration of colistin that resulted in wells with no visible growth.

Inhibitor titration assay.

Overnight cultures of P. aeruginosa PA14 were backdiluted into fresh LB to an OD600 of 2.5 x 10⁻³. Using polypropylene 96 well plates (Corning 3879), the potential antibiotic adjuvant was assayed with a 3-fold dilution starting with a 250 μM concentration. The concentration of colistin was held constant at 0.375 μg/mL, using a freshly prepared colistin stock and minimizing transfers of colistin-containing solutions. After 17 h of aerobic growth at 37 °C, 300 rpm, the absorbance at 600 nm was recorded.

Abbreviations Used

PCA

phenazine-1-carboxylic acid

PYO

pyocyanin

PAL

photoaffinity labeled

WT

wild-type

PA

Pseudomonas aeruginosa

EC

Escherichia coli

PSM

peptide spectrum matches