Mechanism of azithromycin inhibition of HSL synthesis in Pseudomonas aeruginosa

Pseudomonas aeruginosa is an opportunistic pathogen and a leading cause of nosocomial infections. Unfortunately, P. aeruginosa has low antibiotic susceptibility due to several chromosomally encoded antibiotic resistance genes. Hence, we carried out mechanistic studies to determine how azithromycin affects quorum sensing and virulence in P. aeruginosa. lasI and rhlI single and double mutants were constructed. We then undertook a quantitative approach to determine the optimal concentration of azithromycin and culture time that can affect the expression of HSLs. Furthermore, based on the above results, the effect on quorum sensing was analyzed at a transcriptional level. It was found that 2 μg/mL azithromycin caused a 79% decrease in 3-oxo-C12-HSL secretion during cultivation, while C4-HSL secretion was strongly repressed in the early stages. Azithromycin acts on ribosomes; to determine whether this can elicit alternative modes of gene expression, transcriptional regulation of representative virulence genes was analyzed. We propose a new relationship for lasI and rhlI: lasI acts as a cell density sensor, and rhlI functions as a fine-tuning mechanism for coordination between different quorum sensing systems.

Scientific RepoRts | 6:24299 | DOI: 10.1038/srep24299 a number of multidrug-resistant strains, which make clinical treatment difficult 19 . The resistance mechanisms of PA are very complex; horizontal gene transfer (HGT) is an important mechanism 20 . Azithromycin (AZM), a member of the macrolide class of antibiotics, is used to treat certain bacterial infections, which are primarily caused by gram-positive bacteria but also some gram-negative pathogens. Many clinical and experimental studies have shown the beneficial effects of AZM in patients with diffuse panbronchiolitis and cystic fibrosis, which are associated with PA infection [21][22][23] .
A study by Henkel 24 showed that AZM has the potential to inhibit QS signal molecules and attenuate the virulence of PA. Sub-MIC concentrations of AZM were found to inhibit the production of QS signals, swimming, swarming and twitching motilities, and biofilm formation in vitro. AZM affects QS by interfering with one of the following pathways: biofilm formation 25 , translation due to its interaction with ribosomes 26 , or increase in outer membrane permeability 27 . Moreover, 2 μg/mL AZM (1/64th of the MIC) showed promising results on QS-dependent virulence factor production, biofilm formation, and oxidative stress resistance in PA 28,29 .
The mechanisms of QS regulation have mostly been described qualitatively. However, the time course of HSL synthesis and the peak activity of this process are currently unclear, and a quantitative analysis of AZM-mediated inhibition of QS signals has yet to be performed. In this study, we applied HPLC-MS to assay both C4-HSL and 3-oxo-C12-HSL synthesis and the effect of the lasI and rhlI genes in both lasI and rhlI mutants and wild-type PAO1 following treatment with AZM. Additionally, the effect of AZM on the transcription of toxin, elastase and hemolysin genes was investigated to assess the transcriptional dysfunction due to reduced LasI. We also assessed the expression of the HAQ's downstream effector genes (phnAB) after adding AZM.

Materials and Methods
Construction of P. aeruginosa PAO1 lasI and rhlI mutant strains. Bacterial strains, plasmids, and media. The bacterial strains and plasmids used in this study are shown in Table 1. PAO1 was a gift from Dr. Zhou Lin (Children's Hospital of Chongqing Medical University). Strains were routinely grown in rich liquid or solid (15 g/L agar) Luria-Bertani medium (LB). The medium was supplemented with ampicillin (Amp, 60 μg/ mL), chloramphenicol (C, 34 μg/mL) or gentamicin (GM, 30 μg/mL). Solid LB medium with 10% sucrose and without NaCl was used to select for plasmid excision from the chromosome in the gene allelic exchange experiments. Omission of NaCl from this medium was previously shown to improve sucrose counter-selection in Escherichia coli (E. coli).
DNA techniques. The rTaq DNA polymerase, restriction endonucleases, DNA-modifying enzymes, Klenow fragment, DNA kination kit and DNA ligation kit used in this study were purchased from TaKaRa. The plasmid extraction kit and DNA gel purification kit were purchased from Guangdong Dongsheng Biotech Corporation. Antarctic phosphatase and T4 DNA ligase were purchased from NEB.
The pGSM-ΔlasI vector was derived from pMD18AB and pCVD442 in several steps. First, the GM sequence was amplified from the pBBR-LuxAB plasmid and was TA cloned into pMD18-GM, and fragments of sacB and mob from pCVD442 were amplified 30 . The GM fragment was phosphorylated using the DNA kination kit and inserted into a SmaI site on pMD18AB to create pG-ΔlasI, and mob was TA cloned into pMD18-mob.
For the construction of pGSM-ΔlasI, a 1.8 kb PCR fragment of sacB was ligated into the XbaI site of pG-ΔlasI to form pGS-ΔlasI, and the mob sequence of a 2.2 kb SmaI/HindIII fragment from pMD18-mob was blunt-ended and ligated into pGS-ΔlasI to construct pGSM-ΔlasI.
An 831 bp XbaI/HindIII fragment from pMD18-GM was sub-cloned into pUC19CD, forming pG-ΔrhlI. For the construction of pGSM-ΔlasI, a 1.8 kb PCR fragment of sacB was ligated into a HindIII site of pG-ΔrhlI, yielding pGS-ΔrhlI, and the mob sequence of a 2.2 kb SmaI/HindIII fragment from pMD18-mob was blunt-ended and ligated into pGS-ΔrhlI to construct pGSM-ΔrhlI.
Plasmids pGSM-ΔlasI and pGSM-ΔrhlI were mobilized by conjugation from E. coli SM10λpir into P. aeruginosa PAO1 to generate a lasI mutant strain (PA-ΔlasI) and an rhlI mutant strain (PA-rhlI), respectively. For the generation of an rhlI and lasI double mutant (PA-ΔlasIrhlI), pGSM-ΔlasI was conjugally transferred from E. coli SM10λpir into a P. aeruginosa PA-ΔrhlI mutation strain. Growth curve. PAO1, PA-ΔrhlI and PA-ΔlasIrhlI were grown in 3 mL of LB medium at 37 °C with shaking. The cultures were simultaneously supplemented with AZM at a final concentration of 2 μg/mL. The cultures were collected after 0, 2, 4, 6, 8, 10, or 12 h of growth and assayed at OD 600 .
Quantification of HSLs by HPLC-MS/MS. After clarification by centrifugation, 1 mL of the cell-free supernatant was extracted three times with an equal volume of ethyl acetate, which was supplemented with 0.2 M acetic acid 33 . Thereafter, the combined organic phase was dried in an N2 stream, redissolved in 1 mL of methanol and stored at − 20 °C to prepare other stock solutions. Twenty microliters of each of these samples was added to 980 ìL of methanol for a 50-fold dilution immediately prior to HPLC-MS/MS analysis. All samples were filtered through 0.22 μm nylon disk filters before HPLC-MS/MS analysis 34 .
The HPLC-separated compounds were detected by electrospray ionization ion trap mass spectrometry (ESI-MS) using a Bruker Esquire-LC spectrometer (Bruker Daltonic, Germany) under positive-ion conditions. The analysis of two structurally distinct HSLs requires fast and selective analysis. The choice of + ESI mode was based upon the greater sensitivity for both HSL analytes; this was relatively straightforward because all the analytes demonstrated protonated [M + H]+ species as the dominant pseudo-molecular ion.
Ten microliters of the sample was injected for HPLC-MS/MS analysis and introduced onto a ZORBAX SB-C18 column (4.6 × 100 mm, 2.1 μm). The HPLC system used 0.3% formic acid in water as mobile phase A, and mobile phase B was 50% acetonitrile in methanol. The gradient profile was as follows: isocratic for 3 min with 10% B, then a further gradient with 90% B over 2 min, followed by 10% B for 5 min, at a flow rate of 0.5 mL/min. All mass spectrometry (MS) experiments were conducted on a 4000 QTRAP hybrid triple-quadrupole linear ion trap mass spectrometer (Applied Biosystems, Foster City, CA, USA) equipped with a TurboIon source used in positive ion electrospray mode. A Windows XP (Microsoft, Redmond, WA, USA) workstation running Analyst (version 1.6) was used for data acquisition and processing.
The working flow rate and gas were as follows: curtain, gas 1 and 2 were nitrogen, 60 L/h, and 60 L/h, respectively. The ion source potential was 5,500 Vm, and the source was held at 550 °C. Quantification was performed using Analyst 1.6 in Quantitate mode. Six-point standard curves were generated for C4-HSL and 3-oxo-C12-HSL, and the standard mixture was reanalyzed after every sixth sample.
Recovery Test. The recovery of analytes from the medium was determined by adding a standard mixture of C4-HSL and 3-oxoC12-HSL at low (10 μg/mL) and high (100 μg/mL) concentrations to 3 mL of LB culture media. Each sample was extracted three times with an equal volume of ethyl acetate. The recovery was calculated by comparing the response ratios of spiked extracted medium with a standard mixture prior to extraction.
Precision was calculated from the relative standard deviation (RSD) of the replicates (n = 5), and accuracy was calculated by direct comparison of mean measured levels of spiked analytes with expected concentrations for unextracted standards.

Statistical analysis.
All experiments were carried out in triplicate to validate the reproducibility of the experiments. The results were analyzed statistically using repeated measures analysis of variance with SAS 9.2 software to calculate P values. P < 0.05 was taken as statistically significant.

Results
The AZM MIC for P. aeruginosa PAO1 was determined to be 128 mg/mL, the same as described previously. Therefore, 2 μg/mL AZM (1/64th of the MIC) was added to all four strains in these experiments.

HPLC-MS/MS Analysis.
Using the experimental conditions described in the Materials and Methods section, the analogues of C4-HSL and 3-oxo-C12-HSL were separated. Active fractions were located in two single peaks (C4-HSL and 3-oxo-C12-HSL) with retention times of 5.09 min and 7.0 min, respectively. Two compounds with these retention times were obtained from P. aeruginosa PAO1 supernatants but were absent from PA-ΔlasIrhlI supernatants.
The MS parameters (precursor and product ions used for MRM transitions and corresponding optimized voltages) are listed in Table 2, which shows that common product ion fragments could be used for individual families of analytes, i.e., m/z 102 for HSLs. Our detection limits were 0.05 ng/mL and 50 ng/mL. Extraction of the supernatants was performed as described in previously, and detection and quantification of N-acylhomoserine lactones was performed by HPLC-MS/MS. Recovery Test. The bacterial culture media potentially containing HSLs are complex matrices containing components of growth media and bacterial exoproducts, which could affect the specificity and sensitivity of HPLC-MS/ MS detection. To examine the influence of the medium on the extraction, we added a standard mixture of HSLs to the culture media at either low (10 μg/mL) or high (100 μg/mL) concentrations for extraction.
Precision was calculated from the relative standard deviation (RSD) of the replicates (n = 5), and accuracy was calculated by direct comparison of mean measured levels of spiked analytes with expected concentrations for unextracted standards.
Effect of mutation and AZM on P. aeruginosa growth. As shown in Fig. 1, bacterial growth curves indicated that there were no differences between PAO1 and the mutants deficient in lasI, rhlI and lasI/rhlI in terms of growth at 0, 2, 4, 6, 8, 10 and 12 h. Based on the data, 0-8 h is the exponential phase, followed by the early stationary phase. Growth comparisons to PAO1 were performed using paired t-test analyses (P > 0.05). The results indicate that QS mutations, i.e., PAO1 deficient in lasI, rhlI and lasI/rhlI, had almost no influence on P. aeruginosa growth. Moreover, 2 μg/mL of AZM did not affect the growth of the strains compared with untreated controls (P > 0.05) (Fig. 1).

Effect of lasI and rhlI mutations on QS signals.
No C4-HSLs were detected in P. aeruginosa PAO1 in the early growth phase, which was between 2-4 h (Fig. 2). However, 3-oxo-C12-HSL was detected at 6 h, during

Table 2. Selected precursor and product ion m/z values, retention times and mass spectrometer parameters used for HSL analytes.
the exponential phase of growth, and secretion was maintained until 12 h; the concentration ranged from 89.6 ng/ mL to 172.8 ng/mL (Fig. 3). A comparison of the different time points for the secretion of 3-oxo-C12-HSL showed no statistically significant differences, P > 0.05. The secretion of C4-HSL increased with time, with a maximum concentration of 512.2 ng/mL, which was stable from 10 h to 12 h, i.e., in the stationary phase. (A comparison of the different time points for the secretion of C4-HSL showed statistically significant differences, P < 0.05.) For PA-ΔrhlI, C4-HSL was not detected, but 3-oxo-C12-HSL production was similar to that of PAO1; there was no significant difference between the two strains (P > 0.05). For PA-ΔlasI, 3-oxo-C12-HSL was not detected, and C4-HSL levels were below the detection limits. For PA-ΔlasIrhlI, there was no C4-HSL or 3-oxo-C12-HSL secretion. In the QS system, the lasI gene product directs the formation of the diffusible extracellular signal 3-oxo-C12-HSL, and the rhlI product catalyzes the synthesis of C4-HSL 35 ; therefore, the las system positively regulates the expression of rhlI 36 .
Effect of AZM on QS signals. At (Fig. 2). With the addition of 2 μg/mL AZM, the HSL levels were noticeably lower than in the untreated controls, although there was no significant difference between the two groups (P > 0.05).

Effect of AZM on QS genes.
Real-time polymerase chain reaction (RT-PCR) was used to assess representative genes from the AHL and HAQ signaling systems. The experiment was conducted after 6 h of growth with or without 2 μg/ml AZM (Fig. 4). Selected genes regulated by lasR (aprX, toxA and lasA) and rhlR (rhlA and rhlB), genes of the HAQ system (phnA and phnB) and a repressor of lasI and rhlI, qscR, were assayed for expression in PAO1, PA-ΔrhlI, PA-ΔlasI and PA-ΔlasIrhlI. The expression of the genes is reported as a ratio of the target to a reference (rpoD).
The reduction in lasI expression in PAO1 was observed to correlate with the protein expression; 3-oxo-C12-HSL was reduced by 65% with AZM, matching the 79% reduction in protein levels under the same conditions. In the PAO1 strain, although lasI (3-oxo-C12-HSL) levels were reduced, as assessed by HPLC-MS/ MS analysis, the expression of other QS genes was induced after addition of AZM. Additionally, the expression of other virulence genes, except toxA, showed a stronger repression in PAO1 compared to the mutants when AZM was not added. In the PA-ΔrhlI, PA-ΔlasI and PA-ΔlasIrhlI strains, the expression of virulence genes was induced with addition of AZM. Among the mutants, repression of virulence genes of the AHL system was stronger in PA-ΔlasI than in PA-ΔrhlI. Additionally, the expression of phnA and phnB from the HAQ system was greater in PA-ΔrhlI (and PAO1) compared to PA-ΔlasI.

Discussion
In this study, we constructed lasI and rhlI single and double mutant strains, which had a complete deletion of the coding domain. A selective and rapid method for the simultaneous analysis of the two main HSLs in PA was also developed. Levels of 3-oxo-C12-HSL in PAO1 and PA-ΔrhlI showed a significant decrease following addition of AZM, while the accumulation of C4-HSL was not greatly influenced by AZM in the PAO1 strain. At the gene level, addition of AZM was observed to increase the expression of virulence genes in all backgrounds. The las and rhl systems are hierarchically connected and regulate the timing and production of multiple virulence factors 37 . The las system positively regulates the expression of both rhlR and rhlI. When a strain is deficient in lasI, rhlI gene expression is repressed 38 . Because of this, no 3-oxo-C12-HSL was detected for PA-ΔlasI, and  Because PA-Δ lasI and PA-Δ rhlI did not produce 3-oxo-C12-HSL, the data are not shown.+ AZM indicates strains treated with azithromycin (2 μg/mL). The results are presented as the mean ± SD obtained from three independent experiments. ⭐ denotes PAO1 compared with 2 μg/mL AZM treatment, with statistically significant differences. ♦ denotes PA-Δ rhlI compared with 2 μg/mL AZM treatment, with statistically significant differences.
In the early growth phase (2-4 h) of the PAO1 strain, there were no HSLs detected; 3-oxo-C12-HSL was first detected at 6 h. 3-Oxo-C12-HSL was first secreted in the exponential growth phase and was maintained until the early stationary phase. These results are in agreement with a previous report on PAO1, in which 3-oxo-C12-HSL at a concentration of 0.8 μmol/L was detected at 70.5 h 39 . Moreover, in this study, our detection limit was 0.05 ng/mL (0.17 nmol/L), more senstive than previous reports of 0.2 μmol/L 40 . Previous studies demonstrated that the expression of genes of the las system (lasR and lasI), which is affected by 3-oxo-C12-HSL, remains fairly constant throughout cultivation 41 . In contrast, the secretion of C4-HSL increased over time and reached a maximum at 12 h. In another study, the secretion of C4-HSL reached the maximum concentration at 42.5 h 42 .
At 2 μg/mL AZM, a concentration far below the MIC, the secretion of HSLs was inhibited through the blockade of QS. We observed that 2 μg/mL of AZM could decrease 3-oxo-C12-HSL secretion by nearly 79% and led to a noticeable reduction throughout the cultivation. The addition of 2 μg/mL AZM had a different effect on C4-HSL secretion, as it decreased C4-HSL secretion by 76% at t = 6 h and by 35% at t = 8~10 h, but secretion was not repressed at t = 12 h. Based on these results, we conclude that 2 μg/mL of AZM reduced 3-oxo-C12-HSL secretion during cultivation, but only repressed C4-HSL secretion in the early stages 43 . In this study, both an appropriate model system with a known time course involving autoinducers as well as AZM inhibition of QS signals in PA were clearly demonstrated. The resistance mechanisms of PA are very complex 44,17 , and HGT has been shown to be an important mechanism 41 . Conjugation, one of the most important mechanisms of HGT in the environment, involves the direct transfer of genetic material from cell to cell, usually in the form of plasmids or transposons 45,46 . It has been reported that HSLs have a positive effect on the conjugation of Ti plasmid transfer in Agrobacterium Addition of AZM results in reduced HSLs at the protein level, while at gene level, AZM derepresses virulence genes. Transcriptional regulation of representative genes regulated by lasR and rhlR agreed broadly with previous findings. Stronger repression was observed in the wild-type strain compared to the mutants, and among the mutants, QS and virulence gene expression were observed only in the PA-ΔrhlI strain. Our results validated the hierarchy of las activation on the rhl system 47 . Without addition of AZM, all the virulence genes except toxA were repressed, with a stronger effect observed in the wild-type than in the mutants. This could be due to the phase in which the gene expression study was performed. toxA has been shown to be produced mainly during the early exponential phase of growth when grown aerobically 48 . As observed by Linares et al. 13 , sub-inhibitory concentrations of antibiotics could be the reason for the increased expression of QS and virulence genes (Fig. 4a). Another explanation for this could be that addition of AZM and the subsequent reduction in lasI expression (relative mRNA levels were 99.1 and 34.8, respectively, with AZM and without AZM) can be an inactivating signal, which has been previously shown to enhance virulence 49,32 . We hypothesize a novel regulatory circuit where lasI acts as an initial sensor of the optimal cell density required for activating virulence genes, while activated rhlI acts as a repressor of all the other QS systems (Fig. 5). In support of this idea, in the current study we observed that phnAB expression is reduced in PA-ΔlasI but not in PA-ΔrhlI (and PAO1). It has been shown previously that the las and rhl systems regulate quinolone signals, where lasI provides information to the cells about the QS activation, and rhlI suppresses the HAQ system. One of the regulators of las and rhl system is the quorum-sensing control repressor (qscR) protein, which functions as a las-rhl antagonist, and lasI is known to positively influence qscR. In the present study, low levels of qscR were observed in the absence of lasI and rhlI. Expression of qscR was 6 times higher in the wild-type compared to the mutant strains (Fig. 4), suggesting a feedback mechanism dependent on the concentration of lasI and rhlI via the regulation of qscR. The relationship between QS and virulence is complex. It is known that QS systems activate virulence and that AZM interferes with the QS activation by preventing virulence. Because AZM targets ribosomes, the effect of transcriptional regulation on QS can be inferred as a consequence of the reduction of the autoinducer.
The findings here provide basic data about the time course of AZM and the AZM effect on the synthesis of HSL. These results can further our understanding of antibiotic resistance and the interrelationship between QS and virulence. Here, we propose that QS and virulence can subsist individually, and regulated by the growth phase and the environment.