Mutation of PA4292 in Pseudomonas aeruginosa Increases β-Lactam Resistance through Upregulating Pyocyanin Production

ABSTRACT

Metallo-β-lactamase (MBL)-producing Pseudomonas aeruginosa is increasingly reported worldwide and usually causes infections with high mortality rates. Aztreonam/avibactam is a β-lactam/β-lactamase inhibitor (BLBLI) combination that is under clinical trials. The advantage of aztreonam/avibactam over the currently used BLBLIs lies in its effectiveness against MBL-producing pathogens, making it one of the few drugs that can be used to treat infections caused by MBL-producing P. aeruginosa. However, the molecular mechanisms underlying aztreonam/avibactam resistance development remain unexplored. Here, in this study, we performed an in vitro evolution assay by using a previously identified MBL-producing P. aeruginosa clinical isolate, NKPa-71, and found mutations in a novel gene, PA4292, in the aztreonam/avibactam-resistant mutants. By mutation of PA4292 in the reference strain PA14, we verified the role of PA4292 in the resistance to aztreonam/avibactam and β-lactams. Transcriptomic analyses revealed upregulation of pyocyanin biosynthesis genes among the most overexpressed in the PA4292 mutant. We further demonstrated that pyocyanin overproduction in the PA4292 mutant increased the bacterial resistance to β-lactams by reducing drug influx. These data revealed a novel mechanism that might lead to the development of resistance to aztreonam/avibactam and β-lactams.

INTRODUCTION

Pseudomonas aeruginosa is an opportunistic pathogen that is the predominant cause of chronic lung infections in adults with cystic fibrosis (CF) and chronic obstructive pulmonary disease (COPD) (1–3). β-Lactam antibiotics are the most commonly used drugs to treat P. aeruginosa infections. However, P. aeruginosa strains commonly evolve antibiotic resistance during long-term antimicrobial treatment, which leads to poor clinical outcomes (4). The mechanisms of β-lactam resistance are usually multifactorial, including β-lactamases production, low outer membrane permeability, overexpression of efflux pump systems, and mutations in penicillin binding proteins (PBP) (5). Among these mechanisms, β-lactamase production has emerged as a major public health concern, as the presence of β-lactamase genes on self-transmissible plasmids promotes spreading of the genes through horizontal transfer. Based on sequence similarity, the Ambler classification system divides β-lactamases into four classes, class A (e.g., extended-spectrum β-lactamase, [ESBLs]), class B (metallo-β-lactamases [MBL]), e.g., NDM, IMP, VIM, and SPM, class C (e.g., AmpC), and class D (e.g., OXA) β-lactamases. A successful strategy to preserve efficacies of β-lactams is to combine the antibiotics with β-lactamase inhibitors (6). Currently, the β-lactam/β-lactamase inhibitor combinations (BLBLIs) used in the clinic include amoxicillin/clavulanic acid, ticarcillin/clavulanic acid, ampicillin/sulbactam, piperacillin/tazobactam, meropenem/vaborbactam, ceftazidime/avibactam, ceftolozane/tazobactam, and imipenem/relebactam (7, 8). However, none of these BLBLIs is effective against MBL-producing pathogens. Nowadays, MBL-producing P. aeruginosa is spreading worldwide, and the lack of safe and effective treatment options for these pathogens has led to the mortality rate of more than 45% in patients with severe infections (9).

Aztreonam is a monobactam that targets penicillin-binding protein 3 (PBP-3). It was approved by the U.S. Food and Drug Administration (FDA) in 1986. In 2010, inhalation of aztreonam was approved for chronic suppression therapy in CF patients with P. aeruginosa infection (10, 11) and has been shown to improve pulmonary function and reduce bacterial density (12–14). Aztreonam is intrinsically resistant to hydrolysis by MBLs (15). However, it is susceptible to ESBLs, AmpC β-lactamases and Klebsiella pneumoniae carbapenemase (KPC) carbapenemases (16). Currently, the combination of aztreonam with a β-lactamase inhibitor, avibactam, is under phase 3 clinical trials (17). Avibactam is a non-β-lactam β-lactamase inhibitor that is effective against most class A and C and some class D β-lactamases (18, 19). Thus, the aztreonam/avibactam combination may serve as an effective treatment option against a variety of β-lactamase-producing strains, including those producing MBLs.

A previous in vitro evolution study identified chromosomal mutations that contribute to aztreonam resistance in a P. aeruginosa reference strain PAO1 (20). The developed resistance was mainly due to mutations in the PBP3-encoding gene ftsI, as well as mexR and nalD, which encode negative regulators of the multidrug efflux system MexAB-OprM (20). Since aztreonam/avibactam is likely to be used in the treatment of infections caused by carbapenem-resistant strains, especially those producing MBL, we performed an in vitro evolution assay by using an MBL (DIM-2)-producing multidrug-resistant P. aeruginosa clinical isolate, NKPa-71 (21). Besides genes known to be related to aztreonam resistance, we found that mutation in a PA4292 gene increased the bacterial resistance to aztreonam/avibactam and other β-lactam antibiotics. We further elucidated the mechanism of resistance in the PA4292 mutant. Overall, our results revealed a novel gene that is involved in the bacterial resistance to β-lactam antibiotics.

RESULTS AND DISCUSSION

Development of resistance to aztreonam/avibactam by a clinical isolate, NKPa-71.

Development of aztreonam/avibactam resistance was investigated by passaging four biological replicates of NKPa-71 in LB broth containing increasing concentrations of aztreonam/avibactam. After 13 days of evolution, the MICs for the groups reached 128 to 256 μg/mL, whereas no change of MIC was observed for the control group that was evolved in LB (Fig. 1). To verify that the resistance is due to stable mutations, we streaked the population of each of the replicates that reached the highest MICs at the earliest days on LB plates, including replicates no. 1 (day 12), no. 2 (day 13), no. 3 (day 10), and no. 4 (day 13) (Fig. 1). A single colony was isolated from each of the plates and designated 1-12, 2-13, 3-10, and 4-13, respectively. The strains were cultured in LB broth for two sequential passages, followed by MIC determination. The MICs of the strains were the same as those of the corresponding populations (Table 1), indicating that the increased resistance was due to selection of stable mutants.

FIG 1.

Development of resistance to aztreonam/avibactam by clinical isolate NKPa-71. MICs of experimentally evolved mutants to aztreonam/avibactam. Four parallels were passaged in the presence of aztreonam/avibactam for experimental evolution, and one group was passaged in MHB as a control. MICs of aztreonam/avibactam were determined daily.

TABLE 1.

MICs of aztreonam/avibactam (ATM+AVI) and aztreonam (ATM) (μg/mL)^a

Strain	MIC (μg/mL) for:
	ATM+AVI	ATM
NKPa-71	4	4
1-12	128	256
2-13	256	1,024
3-10	128	128
4-13	256	512
PA14	4	4
ΔPA4292	8	8
Δpcm	4	4
ΔcbrA	4	4
ΔPA4292Δpcm	8	8
ΔPA4292ΔcbrA	8	8
ΔPA4292ΔpcmΔcbrA	16	16
PA14/1176_1202del PA4292	8	8
ΔPA4292/pUCP24-PA4292	4	4
PA14/1176_1202delPA4292/pUCP24-PA4292	4	4

^adel, deletion. Data represent results from three independent experiments.

Mutations in the resistant mutants.

To understand the genetic events that contribute to the resistance in the evolved mutants, we sequenced the genomes of the isolated strains. Table 2 shows the mutations in the resistant mutants that were not present in the control group. Mutations in three known aztreonam resistance-related genes were identified in strains 1-12, 2-13, and 4-13, including ftsI, mexR, and nalD (20). Strain 3-10 carries mutations in three genes, cbrA, pcm, and PA14_55770 (PA4292 in PAO1), which encode a two-component regulatory system sensor, an l-isoaspartate protein carboxylmethyltransferase, and a probable phosphate transporter, respectively. Meanwhile, mutations in PA4292 were identified in strains 1-12 and 2-13.

TABLE 2.

Genes mutated in experimentally evolved mutants^a

Mutant strain	Mutated genes	Mutation type
1-12	mexR	G218C
	clpP	T461A
	PA14_55770 (PA4292)	1256_1261del
2-13	mexB	C136A
	nalD	C190T
	PA14_55770 (PA4292)	G1449A
	ftsI	T1571G
	fha1	849_850insCAGCCA
3-10	PA14_55770 (PA4292)	1176_1202del
	cbrA	C406T
	pcm	T356A
4-13	mexR	299delA
	secY	T64C
	clpA	C1879T
	PA14_39110 (PA1965)	T64C; T52G; 55_61del; 40_45del; 67_70del; 33_34insAG; 39_40insAA
	ftsI	T1519G
	PA14_36170 (PA2207)	299_300insAACGCGGGGAGAACAGCG; 302_303insTTCATCGGGGAGTCCTT

^adel, deletion; ins, insertion.

PA4292 affects bacterial resistance to aztreonam/avibactam and other β-lactam antibiotics.

To investigate the roles of PA4292, pcm, and cbrA in aztreonam/avibactam resistance, we knocked out each of the genes in a wild-type reference strain, PA14. Deletion of pcm or cbrA did not affect the bacterial resistance to aztreonam/avibactam or aztreonam alone (Table 1). However, deletion of PA4292 or reconstitution of the mutated PA4292 gene of strain 3-10 in PA14 stably increased the MICs of both aztreonam/avibactam and aztreonam by 2-fold, which was restored to wild-type level by complementation with a PA4292 gene driven by its native promoter (Table 1). Simultaneous deletion of pcm and cbrA in the ΔPA4292 mutant further increased the MIC by 2-fold, indicating a facilitating role of pcm and cbrA in the resistance (Table 1). We further examined whether PA4292 plays a role in the bacterial resistance. In addition, the ΔPA4292 mutant was more resistant to other β-lactam antibiotics, including ceftazidime, cefepime, carbenicillin, and the aztreonam-ceftazidime/avibactam combination that has exhibited therapeutic advantages against multiple MBL-producing pathogens (Table 3) (22–24). In combination, these results demonstrate an important role of PA4292 in β-lactam resistance.

TABLE 3.

MICs of β-lactam antibiotics^a

Antibiotic	MIC (μg/mL) for strain:
	PA14	ΔPA4292	ΔPA4292/pUCP24-PA4292	phzC1::Tn	Δphz2	ΔPA4292 phzC1::Tn	ΔPA4292 Δphz2
Aztreonam/avibactam	4	8	4	2	4	4	4
Aztreonam	4	8	4	2	4	4	4
Ceftazidime	1	2	1	1	1	2	1
Cefepime	0.5	1	0.5	0.5	0.5	1	0.5
Carbenicillin	64	128	32	32	64	64	64
Aztreonam-ceftazidime/avibactam	2/1	4/2	2/1	1/0.5	2/1	2/1	2/1

^aThe MIC of aztreonam-ceftazidime/avibactam is expressed as the concentrations of aztreonam/ceftazidime; avibactam was fixed at 4 μg/mL. The synergistic ratio of aztreonam to ceftazidime was assessed by checkerboard assays. Aztreonam: ceftazidime = 2:1, FICI = 0.5. Data represent results from three independent experiments.

PA4292 affects β-lactam resistance through pyocyanin.

To understand the mechanism of PA4292-mediated resistance to β-lactam antibiotics, we compared the transcriptomic profiles between wild-type PA14 and the ΔPA4292 mutant. In total, 196 and 113 genes were up- and downregulated, respectively, in the ΔPA4292 mutant (Table S2). hasAp, PA2092, PA2783, arnE, lecA, and pcrR were highly upregulated (>64-fold) in the ΔPA4292 mutant (Table S3). Among those genes, PA2092 and arnE have been shown to be involved in bacterial resistance to tobramycin and polymyxins, respectively (25, 26). Meanwhile, mutation of PA4292 increased the expression of pyocyanin biosynthesis genes, including phzS and genes in the phzA1-G1 (phz1) and phzA2-G2 (phz2) operons. Compared to the genes in the phz1 operon, those in the phz2 operon were upregulated to a greater extent (Fig. 2A), which was verified by reverse transcription-quantitative PCR (qRT-PCR) (Fig. 2B). In agreement with the gene expression pattern, the ΔPA4292 mutant produced a larger amount of pyocyanin than the wild-type PA14 (Fig. 2C). Complementation with a PA4292 gene restored the expression of phzB1 and phzB2 and the production of pyocyanin in the ΔPA4292 mutant (Fig. 2B and C).

FIG 2.

PA4292 affects bacterial resistance to β-lactam antibiotics through pyocyanin. (A) Genes that were alternatively expressed in the ΔPA4292 mutant. phzB1, phzC1, phzE1, phzG1, and phzS are highlighted in red. phzA2, phzB2, phzC2 and phzG2 are highlighted in green. FC, fold change. (B) mRNA levels of phzB1 and phzB2 genes in the indicated strains were determined by real-time qPCR. (C) Pyocyanin concentration of wild-type PA14, the ΔPA4292 mutant, and the complemented strain. Data represent the mean ± standard deviation of the results from three samples. *, P < 0.05; **, P < 0.01; ***, P < 0.001, by Student’s t test.

Since pyocyanin has been shown to be involved in stress responses (27), we investigated the contributions of phz1 and phz2 operons to the β-lactam resistance in the ΔPA4292 mutant. A phzC1::Tn mutant was obtained from the PA14 transposon insertion mutant library (28), in which we deleted the PA4292 gene, resulting in a phzC1::TnΔPA4292 double mutant. In addition, we constructed a Δphz2 mutant by deleting the phz2 operon and a Δphz2 ΔPA4292 double mutant. Transposon insertion in the phzC1 gene reduced the MICs of aztreonam/avibactam for both PA14 and the ΔPA4292 mutant by 2-fold, indicating that the phz1 operon might play an equal role in the β-lactam resistance in the two strains. Deletion of the phz2 operon did not affect the MICs of aztreonam/avibactam, aztreonam, ceftazidime, cefepime, carbenicillin, and aztreonam-ceftazidime/avibactam for PA14 but reduced the MICs for the ΔPA4292 mutant by 2-fold (Table 3). These results indicate that upregulation of the phz2 operon might play a major role in enhancing β-lactam resistance in the ΔPA4292 mutant.

Overproduction of pyocyanin in the ΔPA4292 mutant reduces the influx of β-lactam antibiotics.

Overexpression of MexAB-OprM and AmpC has been shown to increase resistance to aztreonam and ceftazidime/avibactam in P. aeruginosa (29), which might also lead to resistance to aztreonam/avibactam. Real-time PCR results revealed similar expression levels of mexA in wild-type PA14 and the ΔPA4292 mutant in the absence and presence of aztreonam/avibactam (see Fig. S1 in the supplemental material). In the absence of aztreonam/avibactam or aztreonam, the expression levels of ampC were similar in wild-type PA14 and the ΔPA4292 mutant. The presence of aztreonam/avibactam induced the ampC expression in a dose-dependent manner. However, the ampC mRNA levels in the ΔPA4292 mutant were lower than those in wild-type PA14 at the corresponding aztreonam/avibactam concentrations (Fig. 3A). Although 1 μg/mL aztreonam was unable to induce the ampC expression, 8 μg/mL aztreonam increased the ampC expression, with a lower mRNA level in the ΔPA4292 mutant (Fig. 3A). These results excluded the involvement of MexAB-OprM or AmpC in the enhanced resistance of the ΔPA4292 mutant. However, given that expression of ampC is induced by disruption of murein biosynthesis, these results indicated an alleviated inhibition of PBP3 by aztreonam in the ΔPA4292 mutant. In agreement with the ampC expression patterns, aztreonam/avibactam or aztreonam treatment resulted in cell elongation of PA14, whereas the ΔPA4292 mutant retained a regular rod shape (Fig. 3B, Fig. S2). Deletion of the phz2 operon did not affect the morphological change in wild-type PA14 but restored the antibiotic-induced elongation in the ΔPA4292 mutant (Fig. 3B, Fig. S2). Collectively, these results indicate that overproduction of pyocyanin by the ΔPA4292 mutant might reduce the periplasmic amount of aztreonam/avibactam or aztreonam.

FIG 3.

Mutation of PA4292 reduces the influx of β-lactam antibiotics. (A) mRNA levels of the ampC gene were determined by real-time PCR. The bacteria were induced in LB with or without aztreonam/avibactam or aztreonam alone. (B) Statistical analysis of bacterial morphology. The indicated strains were incubated with or without aztreonam/avibactam or aztreonam alone, followed by observation with microscopy. (C) Influx of ceftazidime. ampC overexpressing strains were incubated with ceftazidime. Influx of ceftazidime was represented by measuring the drop of OD₂₆₀ after 1 h. Data represent the mean ± standard deviation of the results from three samples. *, P < 0.05; **, P < 0.01; ****, P < 0.0001, by Student’s t test.

We then measured β-lactam influx by using a ceftazidime degradation assay as previously described (30). Ceftazidime was incubated with ampC-overexpressing strains; thus, the rate of ceftazidime hydrolysis by periplasmic AmpC correlates with the drug influx rate. Compared to PA14, the ΔPA4292 mutant degraded ceftazidime at a lower rate, which was accelerated by deletion of the phz2 operon (Fig. 3C), indicating reduction of β-lactam influx by pyocyanin overproduction in the ΔPA4292 mutant.

Conclusions.

Aztreonam/avibactam is a BLBLI that can be used to treat infections caused by MBL-producing pathogens. In this study, we found that mutation in PA4292 in an MBL-producing P. aeruginosa clinical isolate increased resistance to aztreonam/avibactam. Deletion of PA4292 in the reference strain PA14 enhanced pyocyanin production, which increased resistance to aztreonam/avibactam and other β-lactams. PA4292 encodes a probable phosphate transporter that might localize on the cytoplasm membrane (31); however, the physiological role of PA4292 remains to be studied. Deletion of PA4292 increased the expression of the phz1 and phz2 operons, with a greater effect on the phz2 operon. The phz1 and phz2 operons encode nearly identical sets of enzymes that are involved in pyocyanin synthesis (32). In PA14, the expression of phz1 and phz2 operons is higher in planktonic and biofilm-associated cells, respectively, indicating a differential regulation (32). Further studies are warranted to understand the regulatory mechanism of PA4292 on the two operons. Pyocyanin is an electrochemically active metabolite that might alter membrane permeability and subsequent drug influx. In addition, pyocyanin is involved in multiple biological activities, such as mediating cell signaling (33), facilitating development and organizing structure of biofilm (34, 35), promoting iron acquisition (36), inhibiting other microorganisms (37), and killing host cells (38). Therefore, overproduction of pyocyanin might promote persistence during infection by enhancing antibiotic resistance and production of virulence factors. Overall, our results revealed a novel gene PA4292 that affects β-lactam resistance through pyocyanin.

MATERIALS AND METHODS

Bacterial strains, plasmids, and primers.

The bacterial strains, plasmids, and primers used in this study are listed in Table S1. The strain NKPa-71 was isolated from the sputum sample of a patient (21). Bacteria were cultured in Luria-Bertani (LB) broth (10 g/L tryptone, 5 g/L yeast extract, 5 g/L NaCl, pH 7.4) at 37°C with shaking at 200 rpm.

Antibiotics were used at the following concentrations: for E. coli, 10 μg/mL tetracycline and 10 μg/mL gentamicin; for P. aeruginosa, 50 μg/mL tetracycline and 50 μg/mL gentamicin. Chromosomal gene mutations were generated as described previously (39). The plasmid containing PA4292 and its upstream promoter region (pUCP24-prom-PA4292) was transferred into the ΔPA4292 strain, and the transformant was selected on LB agar containing 50 μg/mL gentamicin.

MIC determination.

MICs were determined using broth microdilution in accordance with the Clinical and Laboratory Standards Institute guidelines (CLSI 2018). Each well of a 96-well microtiter plate was filled with 100 μL Mueller-Hinton broth (MHB) with serially diluted concentrations of individual antibiotics. Avibactam was used at a fixed concentration of 4 μg/mL (5). Bacterial culture (100 μL; 2 × 10⁵ CFU) was inoculated into each well and incubated at 37°C for 20 h. The MICs were recorded as the lowest antibiotic concentrations in wells where no visible growth was present. Each strain was tested in triplicate.

Checkerboard assay.

Efficacies of the combinations of aztreonam and ceftazidime/avibactam were determined using a 96-well plate as described previously (40). The concentration of each antimicrobial was started at 2× MIC and then serially diluted in a 2-fold manner. Bacterial inoculation and incubation were performed as described in the MIC test. The combinatory effects were determined by calculating the fractional inhibitory concentration (FIC) index (FICI) as follows: FICI = (MIC of drug A, tested in combination)/(MIC of drug A, tested alone) + (MIC of drug B, tested in combination)/(MIC of drug B, tested alone). Synergy was defined as an FICI of ≤0.5; no interaction was identified with an FICI of >0.5 but ≤4; antagonism was defined as an FICI of >4 (41).

Experimental evolution to select for aztreonam/avibactam-resistant mutants.

The continuous aztreonam/avibactam evolution experiment was performed as previously described (21) with five repeats in LB containing increasing concentrations of aztreonam with avibactam at a fixed concentration (4 μg/mL). Overnight cultures of NKPa-71 (10 μL) were inoculated into four tubes containing 2, 4, 8, and 16 μg/mL aztreonam/avibactam in 1 mL LB, which corresponds to 0.5×, 1×, 2×, and 4× the MIC, respectively. After 24 h of aerobic culture at 37°C, cells were obtained from the highest concentration of aztreonam/avibactam that allowed bacteria to grow to an optical density at 600 nm (OD₆₀₀) of ≥2. Then 10 μL of the culture was inoculated into 1 mL of fresh medium containing aztreonam/avibactam at increased concentrations, e.g., 1×, 2×, 4×, and 8× the MIC. The serial passaging was repeated for 13 days until the MIC of the evolved strains rose to 128 μg/mL or higher. Meanwhile, another repeat passage in LB was used as a control.

Genome sequencing.

Genome sequencing of the evolved resistant mutants was performed by Azenta Life Sciences, Suzhou, China. Bacterial genomic DNA was purified with a DNA purification kit (Tiangen Biotec, Beijing, China) and then randomly fragmented to <500 bp by sonication (Covaris S220). After end treatment and adaptor ligation, the DNA fragments were purified and amplified with PCR for six cycles. The PCR products were cleaned up using DNA clean beads, validated using a Qsep100 instrument (Bioptic, Taiwan, China), and quantified using a Qubit 3.0 fluorometer (Invitrogen, Carlsbad, CA, USA). Then libraries with different indices were multiplexed and loaded onto a HiSeq/NovaSeq instrument (Illumina, San Diego, CA, USA) or an MGI2000 instrument according to manufacturer’s instructions (MGI, Shenzhen, China).

Sequences of adaptors, PCR (PCR) primers, content of N bases more than 10%, and bases of quality lower than 20 were removed using Cutadapt (V1.9.1). The Burrows-Wheeler MEM algorithm (BWA; V0.7.17) was used to map clean data to the NKPa-71 genome (NCBI SRA no. SRR16282264) (21), and Picard (V1.119) was used to remove duplications. The Unified Genotyper calls single-nucleotide variations and insertion-deletion (SNV/InDel) with GATK (V3.8.1) software. Annotations for SNV/InDel were performed using Annovar (V21 Apr 2018). Genomic structure variation analysis was performed with BreakDancer and CNVnator. The whole-genome sequencing data were deposited in NCBI (PRJNA827270).

Real-time PCR.

Total bacterial RNA was isolated using the RNAprep pure cell/bacteria kit (Tiangen, Biotec, Beijing, China), and cDNA was synthesized with random primers and PrimeScript reverse transcriptase (TaKaRa, Dalian, China). Then the cDNA was mixed with specific forward and reverse primers (Table S1) and SYBR Premix Ex Taq II (TaKaRa). The quantitative real-time PCR (RT-PCR) was performed with a CFX Connect real-time system (Bio-Rad, USA). The 30S ribosomal protein gene rpsL was used as the internal control.

Bacterial morphology observation.

Bacteria were grown to an OD₆₀₀ of 1.0. The bacteria were kept growing in LB or treated with aztreonam (16 μg/mL)/avibactam (4 μg/mL) or aztreonam (16 μg/mL) alone at 37°C for 15 min. Then the bacterial morphology was observed with light microscopy. The length of each individual bacterium was measured with CellSens Dimension (Olympus, Japan).

Transcriptome sequencing and data analysis.

Total RNA was extracted from PA14 and ΔPA4292 at an OD₆₀₀ of 0.8 using an RNAprep pure cell/bacteria kit (Tiangen Biotec, Beijing, China). Two replicates were prepared for each strain. Sequencing and analysis services were performed by Azenta Life Sciences (Suzhou, China) as previously described (42). The RNA-seq raw data were deposited in the NCBI (BioProject no. PRJNA831365).

Influx of ceftazidime.

Influx of ceftazidime was determined by measuring its hydrolysis by intact P. aeruginosa cells as described previously (30). The ampC gene was amplified from PA14 and cloned into pUCP24, on which the transcription of ampC was driven by an exogenous tac promoter (Table S1). The resultant plasmid was transferred into the indicated strains to achieve equal ceftazidime hydrolysis capacity. The bacteria were grown in LB at 37°C to an OD₆₀₀ of 1.0. Then the bacteria were washed with PBS and incubated with 64 μg/mL ceftazidime. Hydrolysis of ceftazidime was determined by measuring the decrease of the OD₂₆₀ after 1 h with a luminometer (Varioskan Flash; Thermo Scientific).

Measurement of pyocyanin concentration.

The pyocyanin concentration was determined as described previously (43). Briefly, 1 mL of supernatant from each bacterial culture was extracted with 0.5 mL of chloroform. Then 0.4 mL of solution from the lower organic phase was reextracted with 0.3 mL of 0.2 N HCl. The concentration of pyocyanin (μg/mL) was calculated by multiplying the OD₅₂₀ by 17.072.