Quorum-Sensing Antagonistic Activities of Azithromycin in Pseudomonas aeruginosa PAO1: a Global Approach

Abstract

The administration of macrolides such as azithromycin for chronic pulmonary infection of cystic fibrosis patients has been reported to be of benefit. Although the mechanisms of action remain obscure, anti-inflammatory effects as well as interference of the macrolide with Pseudomonas aeruginosa virulence factor production have been suggested to contribute to an improved clinical outcome. In this study we used a systematic approach and analyzed the impact of azithromycin on the global transcriptional pattern and the protein expression profile of P. aeruginosa PAO1 cultures versus those in untreated controls. The most remarkable result of this study is the finding that azithromycin exhibited extensive quorum-sensing antagonistic activities. In accordance with the inhibition of the quorum-sensing systems, virulence factor production was diminished and the oxidative stress response was impaired, whereas the type III secretion system was strongly induced. Moreover, P. aeruginosa motility was reduced, which probably accounts for the previously observed impaired biofilm formation capabilities of azithromycin-treated cultures. The interference of azithromycin with quorum-sensing-dependent virulence factor production, biofilm formation, and oxidative stress resistance in P. aeruginosa holds great promise for macrolide therapy in cystic fibrosis. Clearly quorum-sensing antagonist macrolides should be paid more attention in the management of chronic P. aeruginosa infections, and as quorum-sensing antagonists, macrolides might gain vital importance for more general application against chronic infections.

The lung of cystic fibrosis (CF) patients has a unique susceptibility to chronic Pseudomonas aeruginosa infection, which is still the major cause of morbidity and mortality in these patients (4, 24, 36). Although in past decades antibiotic therapy has greatly increased life expectancy, only limited therapeutic options are available, and chronic P. aeruginosa infection is rarely eradicated (5, 20, 37). Hence, there is a is an urgent need to develop alternative treatment regimens to improve lung function and thus the prognosis of the disease. Although the principles concerning therapeutic strategies in the treatment of chronic lung infection have not changed significantly in the last 10 years, the use of azithromycin (AZM) for infection control and inflammation modulation is one new aspect (2, 11, 38, 44, 53).

By conventional standards P. aeruginosa is insensitive to therapeutic concentrations of macrolides; however, recently macrolides have been reported to positively influence the clinical outcome in patients suffering from chronic P. aeruginosa infection in diffuse panbronchiolitis (13, 15, 21, 40, 41). Diffuse panbronchiolitis was first reported in Japan and is characterized by an inflammatory cell infiltration in the respiratory bronchioles, leading to their obstruction and dilatation (35). As disease progresses, patients typically become colonized with mucoid strains of P. aeruginosa accompanied by cystic changes of the lung and by poor clinical prognosis due to progressive deterioration of respiratory function.

The remarkable parallels between diffuse panbronchiolitis and CF led to the question of whether macrolide antibiotics would also be of benefit in patients with CF and to large-scale randomized controlled trials to elucidate the properties of macrolides for chronic P. aeruginosa infection of the lung in CF patients (30, 34, 39, 52). The majority of clinical studies report positive trends concerning the therapeutic potential of macrolide therapy (44). However, the mechanisms of action in chronic P. aeruginosa infection remain obscure (54). Immunomodulatory effects are postulated to account for some of the beneficial effects (9, 16), in addition to altered airway epithelial chloride transport and inhibition of P. aeruginosa virulence factor production by interference with interbacterial communication (33, 45).

Interbacterial communication is also referred to as quorum sensing (QS) and is a very sophisticated mechanism by which signal molecules act as autoinducers and trigger a variety of biological functions when microbial populations attain certain cell densities. QS controls not only virulence factor production but also biofilm formation in P. aeruginosa (3) and thus contributes significantly to pathogenesis and persistence of infection. The QS system in P. aeruginosa comprises two hierarchically organized systems, each consisting of an autoinducer synthetase (LasI/RhlI) and a corresponding regulator protein (LasR/RhlR). Both the las and the rhl QS systems have been shown to be transcriptionally repressed by sublethal azithromycin concentrations (45). Moreover, a QS mutant was shown to be less responsive to AZM inhibition (48).

In this study we applied a systematic approach and analyzed the transcriptome and proteome profiles of P. aeruginosa in response to sublethal concentrations of AZM. Using this global approach, we aimed to identify the influence of AZM on QS-regulated genes and proteins and thus to gain background data on therapy with macrolides for purposes other than their bactericidal properties.

MATERIALS AND METHODS

Bacterial strains and culture conditions.

For the isolation of RNA and for the preparation of cellular and extracellular protein extracts, P. aeruginosa PAO1 (DSM 1707) was grown in brain heart infusion (BHI) medium at 37°C with shaking with or without the addition of 2 μg/ml AZM (Pfizer, Germany) until early stationary phase.

RNA extraction and preparation of protein samples.

Total RNA was extracted from 10 ml of four AZM-treated and untreated PAO1 cultures each, cDNA was synthesized from the RNA pooled from two independent cultures, and subsequently two GeneChips were hybridized for each culture condition. RNA isolation, cDNA generation, fragmentation, biotinylation, and GeneChip hybridization and analysis were performed according to the Affymetrix guidelines and conform to the MIAME requirements (Minimum Information About a Microarray Experiment; experimental details are available at http://www.ncbi.nlm.nih.gov/projects/geo/submission/login/under accession number GSE2430). Three independent protein extracts from pooled supernatants and cell pellets of four 150-ml AZM-treated and untreated PAO1 cultures each were prepared and used immediately for two-dimensional gel electrophoresis. The preparation of protein extracts, two-dimensional gel electrophoresis, and matrix-assisted laser desorption ionization-time of flight (MALDI-TOF) mass spectrometry (MALDI-TOF MS) analysis were performed as described previously (51). The gels were stained with ruthenium II Tris(bathophenanthroline disulfonate) (RuBPS) (22) and differentially expressed proteins as detected in duplicate gels were quantified by ProteomWeaver 2.1 (DEFINIENS). Proteins were considered to be significantly affected when their spot intensities changed at least twofold.

Lactate dehydrogenase release assay.

The bacteria were grown in BHI medium with and without the addition of 2 μg/ml AZM and harvested in log phase (optical density at 600 nm [OD₆₀₀], ∼0.6) and stationary phase (OD₆₀₀, ∼2.7), respectively. J774.A1 cells were grown to confluence in flat-bottom 96-well plates in Dulbecco's modified Eagle's medium with 10% fetal calf serum and infected with 20 μl of the bacterial suspension in 200 μl fresh Dulbecco's modified Eagle's medium with 10% fetal calf serum. J774.A1 cell viability was assessed by the determination of lactate dehydrogenase in the supernatant fraction of six parallel wells by using a lactate dehydrogenase cytotoxicity detection kit according to the manufacturer's instructions (Roche, Mannheim, Germany). All experiments were performed in triplicate.

H₂O₂ sensitivity assay.

The H₂O₂ sensitivity disk assay was adapted from that of Hassett et al. (7). Briefly, Pseudomonas PAO1 was grown at 37°C in BHI medium for various incubation periods with and without the addition of 2 μg/ml AZM; 100 μl of the bacterial culture was suspended in 3 ml of LB soft agar at 40°C in 0.6% (wt/vol) agar, mixed, and poured on LB agar plates with 1.5% (wt/vol) agar. Sterile filter paper disks were placed on the soft solid agar, and the disks were spotted with 8 μl of 30% H₂O₂. Plates were incubated at 37°C for 24 h, and the diameter of the zone of growth inhibition was measured. All experiments were performed in triplicate.

Motility assays.

Media for swimming and twitching assays were LB containing 0.3% (wt/vol) and 1.5% (wt/vol) Bacto Agar (Difco) with and without the addition of 2 μg/ml AZM. Plates were inoculated with bacteria from an overnight LB culture grown in the presence and absence of 2 μg/ml AZM. The swimming plates were inoculated with 15 μl of the bacterial suspension and incubated at 37°C for 24 h, whereas the twitching plates were inoculated with a sterile toothpick at the bottom of the petri dish and incubated at 37°C for at least 24 h.

RESULTS AND DISCUSSION

Influence of sublethal azithromycin concentrations on the PAO1 transcriptome and proteome.

Since interference with virulence factor production in P. aeruginosa has been postulated to be responsible for the observed beneficial effect of macrolide therapy, we aimed to analyze the effect of AZM on bacteria at the early stationary growth phase, when virulence factor production is high. As observed previously (45), the addition of 2 μg/ml azithromycin (1/64th of the MIC) led to a prolonged lag phase and an only minimally affected exponential and stationary phase of growth compared to that with the PAO1 control culture (Fig. 1).

FIG. 1.

Growth of PAO1 in LB broth with (▪) and without (⧫) the addition of 2 μg/ml azithromycin.

A comparison of the gene expression profiles when the PAO1 strain cultured with and without exogenous AZM reached an OD₆₀₀ of 2.8 revealed 107 genes that were differentially expressed (78 of those genes were shown to be up-regulated and 29 genes were repressed), representing 1.9% of the entire genome (Table 1). Many of the AZM-induced genes were genes encoding ribosomal subunits. In addition, an initiation factor (infA PA2619) and an elongation factor (efp PA2851) were overexpressed in AZM-treated cultures. Since the macrolides exhibit their antibacterial activity by binding to the 50S ribosomal subunit, resulting in the blockage of transpeptidation and/or translation, this implies these genes may be overexpressed to compensate for impaired translation due to sublethal AZM concentrations. Moreover, the ribosome modulation factor rmf was strongly repressed upon AZM addition. The PAO1 rmf gene exhibits high homology to the rmf gene in Escherichia coli, in which rmf has been shown to be important for survival under stationary-phase conditions (56).

TABLE 1.

Azithromycin-regulated genes in PAO1

Genea	Gene name	Up-regulated (+) or down-regulated (−) in response to AZM (fold)	Protein
PA0044b	exoT	+13.54	Exoenzyme T
PA0263c	hcpC	+3.83	Secreted protein Hcp
PA0409	pilH	+3.24	Twitching motility protein PilH
PA0422		+6.27	Conserved hypothetical protein
PA0456		+3.38	Probable cold shock protein
PA0555	fda	+5.16	Fructose-1,6-bisphosphate aldolase
PA0649	trpG	+12.08	Anthranilate synthase component II
PA0650	trpD	+7.38	Anthranilate phosphoribosyltransferase
PA0651	trpC	+5.90	Indole-3-glycerolphosphate synthase
PA0805		+8.74	Hypothetical protein
PA0943		+3.51	Hypothetical protein
PA0996	pqsA	+3.78	Probable coenzyme A ligase
PA1440		+5.38	Hypothetical protein
PA1493d	cysP	+3.50	Sulfate-binding protein of ABC transporter
PA1494		+2.61	Conserved hypothetical protein
PA1564		+4.64	Conserved hypothetical protein
PA1696b	pscO	+8.38	Translocation protein in type III secretion
PA1706b,d	pcrV	+91.88	Type III secretion protein PcrV
PA1707b	pcrH	+14.16	Regulatory protein PcrH
PA1708b	popB	+6.77	Translocator protein PopB
PA1709b	popD	+6.25	Translocator outer membrane protein PopD precursor
PA1710b	exsC	+8.99	Exoenzyme S synthesis protein C precursor
PA1711b		+11.87	Hypothetical protein
PA1712b	exsB	+8.59	Exoenzyme S synthesis protein B
PA1714b		+4.45	Hypothetical protein
PA1715b	pscB	+16.46	Type III export apparatus protein
PA1754	cysB	+3.24	Transcriptional regulator CysB
PA2015		+2.78	Probable acyl-coenzyme A dehydrogenase
PA2016		+8.72	Probable transcriptional regulator
PA2023	galU	+4.90	UTP-glucose-l-phosphate uridylyltransferase
PA2191b	exoY	+20.12	Adenylate cyclase ExoY
PA2193e	hcnA	+2.62	Hydrogen cyanide synthase HcnA
PA2423e		+5.55	Hypothetical protein
PA2464		+3.42	Hypothetical protein
PA2619	infA	+18.96	Initiation factor
PA2747		+4.85	Hypothetical protein
PA2755	eco	+4.69	Ecotin precursor
PA2830	htpX	+6.97	Heat shock protein HtpX
PA2851	efp	+4.60	Translation elongation factor P
PA2895		+2.70	Hypothetical protein
PA2900		+5.85	Probable outer membrane protein precursor
PA2901		+14.03	Hypothetical protein
PA3244d	minD	+5.71	Cell division inhibitor MinD
PA3262		+4.72	Probable peptidyl-prolyl cis-trans isomerase, FkbP type
PA3369		+6.90	Hypothetical protein
PA3370		+6.28	Hypothetical protein
PA3371f		+11.50	Hypothetical protein
PA3531	bfrB	+4.88	Bacterioferritin B
PA3656	rpsB	+4.74	30S ribosomal protein S2
PA3686	adk	+5.59	Adenylate kinase
PA3742	rplS	+2.88	50S ribosomal protein L19
PA3841b	exoS	+9.56	Exoenzyme S
PA3842		+14.72	Probable chaperone
PA3976	thiE	+7.89	Thiamine-phosphate pyrophosphorylase
PA3988		+2.67	Hypothetical protein
PA4004		+4.36	Conserved hypothetical protein
PA4006	nadD	+4.19	Nicotinic acid mononucleotide adenylyltransferase
PA4059		+6.47	Hypothetical protein
PA4114		+8.51	Spermidine acetyltransferase
PA4235	bfrA	+4.35	Bacterioferritin A
PA4263	rplC	+3.56	50S ribosomal protein L3
PA4441		+3.18	Hypothetical protein
PA4460		+3.56	Conserved hypothetical protein
PA4495		+3.45	Hypothetical protein
PA4525	pilA	+4.89	Type 4 fimbrial precursor PilA
PA4567	rpmA	+5.05	50S ribosomal protein L27
PA4568	rplU	+9.85	50S ribosomal protein L21
PA4605		+8.72	Conserved hypothetical protein
PA4670	prs	+7.09	Ribose-phosphate pyrophosphokinase
PA4671		+3.70	Probable ribosomal protein L25
PA4751	ftsH	+3.21	Cell division protein FtsH
PA4764	fur	+3.06	Ferric uptake regulation protein
PA4765	omlA	+8.35	Outer membrane lipoprotein OmlA precursor
PA5130		+11.23	Conserved hypothetical protein
PA5191		+2.79	Hypothetical protein
PA5316	rpmB	+5.23	50S ribosomal protein L28
PA5481c		+5.46	Hypothetical protein
PA5482		+6.22	Hypothetical protein
PA0139d	ahpC	−2.97	Alkyl hydroperoxide reductase subunit C
PA0586		−4.61	Conserved hypothetical protein
PA0852d,e	cpbD	−3.31	Chitin-binding protein CbpD precursor
PA1048		−2.81	Probable outer membrane protein precursor
PA1244		−6.62	Hypothetical protein
PA1871e	lasA	−8.33	LasA protease precursor
PA2031		−3.88	Hypothetical protein
PA2146c		−6.30	Conserved hypothetical protein
PA2171c		−7.61	Hypothetical protein
PA2190c		−3.13	Conserved hypothetical protein
PA2259	ptxS	−2.93	Transcriptional regulator PtxS
PA2274		−2.73	Hypothetical protein
PA2300e	chiC	−4.38	Chitinase
PA2564e		−3.77	Hypothetical protein
PA2565		−3.23	Hypothetical protein
PA3049	rmf	−12.49	Ribosome modulation factor
PA3478e	rhlB	−4.33	Rhamnosyltransferase chain B
PA3529d		−4.86	Probable peroxidase
PA3533		−3.98	Conserved hypothetical protein
PA4078f		−18.35	Probable nonribosomal peptide synthetase
PA4205	mexG	−4.78	Hypothetical protein
PA4206	mexH	−3.53	Probable RND efflux membrane fusion protein precursor
PA4236g	katA	−10.78	Catalase KatA
PA4306e		−3.63	Hypothetical protein
PA4366d,g	sodB	−4.42	Superoxide dismutase
PA4377		−6.80	Hypothetical protein
PA4385d	groEL	−4.07	GroEL protein
PA4386	groES	−4.09	GroES protein
PA4611		−4.46	Hypothetical protein

^aOnly open reading frames which were found in all four GeneChip pairings defined by the Affymetrix microarray suite software as having significant changes in their signal intensities and were at least twofold up- or down-regulated in each of the four pairings (the arithmetic middle of all four pairings is given) are listed. Genes identified previously as being QS regulated are in boldface.

^bThe TTSS was identified as QS repressed by Hogardt et al. (10).

^cConditional QS-induced genes; identified as QS regulated by Schuster et al. (42).

^dGenes/proteins that were shown to be regulated by both proteomics and transcriptomics.

^eGeneral QS regulon, based on the results of three microarray studies: Hentzer et al. (8), Schuster et al. (42), and Wagner et al. (49).

^fConditional QS-induced genes; identified as QS regulated by Hentzer et al. (8).

^gIdentified as QS induced by Hassett et al. (7).

An exposure-dependent bactericidal activity of macrolides has been described previously (46) and was observed in this study (after 24 h and 48 h of incubation, significantly fewer bacteria were isolated from the AZM-treated culture than from controls; paired t test, P < 0.001). In this context the AZM-dependent down-regulation of rmf might contribute to the observed impaired survival of AZM-treated PAO1 under stationary-phase conditions.

In a complementary approach to the identification of the global transcriptional pattern we analyzed the protein expression profile of AZM-treated in comparison to nontreated PAO1 cultures. Comparative analyses of the secretome and of cellular extracts of AZM-treated versus nontreated PAO1 cultures disclosed a total of 43 differentially expressed proteins (Table 2); 11 proteins were up-regulated and 15 proteins were down-regulated in the secretome of the PAO1 cultures after AZM addition (Fig. 2). Whereas the two-dimensional gels of the secretome comprised approximately 645 protein spots, 714 protein spots were detected within the cellular fraction. Six proteins were up-regulated and 11 proteins were down-regulated in the cellular fraction of the PAO1 cultures after AZM addition.

TABLE 2.

Azithromycin-regulated proteins in PAO1

Group and spot no.a	Geneb	Gene name	Protein	Up- or down-regulationc
				This study	Nouwens et al. (29)	Arevalo-Ferro et al. (1)
Secretome
51, 63	PA0026		Hypothetical protein	↑	↑
65	PA0888	aotJ	Arginine/ornithine binding protein AotJ	↑	↑	↑
54	PA1065		Conserved hypothetical protein	↑
66	PA1342		Probable binding protein component of ABC transporter	↑		↑
60d	PA1493	cysP	Sulfate-binding protein of ABC transporter	↑
45	PA1673		Hypothetical protein	↑
64d	PA1706	pcrV	Type III secretion protein PcrV	↑
55	PA4175	prpL	PvdS-regulated endoprotease, lysyl class	↑	↓	↑
69	PA5489	dsbA	Thiol:disulfide interchange protein DsbA	↑	↑
59	PA4265	tufA	Elongation factor Tu, TufA	↑
	PA4277	tufB	Elongation factor Tu, TufB
41d	PA0139	ahpC	Alkyl hydroperoxide reductase subunit C	↓
1, 2, 3	PA0572		Hypothetical protein	↓	↓
29, 30, 32d	PA0852	cbpD	Chitin-binding protein CbpD precursor	↓	↓	↑
4, 5, 6, 8, 19, 20, 44	PA1086	flgK	Flagellar hook-associated protein 1 FlgK	↓	↑
18	PA1087	flgL	Flagellar hook-associated protein type 3 FlgL	↓		↑
13, 14, 15, 16, 17, 21, 22, 23, 24, 25, 46, 47, 48, 49, 50, 57, 58, 67, 68	PA1092	fliC	Flagellin type B	↓	↑	↑
26, 27, 28, 37, 38, 39, 52, 53, 56	PA1094	fliD	Flagellar capping protein FliD	↓		↑
33	PA1158		Probable two-component sensor	↓
7	PA1294	aprA	Alkaline metalloproteinase precursor AprA	↓	↓
42	PA1784		Hypothetical protein	↓
9	PA2939	pepB	Probable aminopeptidase	↓	↓	↑
31, 34, 35, 36	PA3724	lasB	Elastase LasB	↓	↓	↑
40	PA3746	ffh	Signal recognition particle protein Ffh	↓
10, 12d	PA4385	groEL	GroEL protein	↓
11	PA5192	pckA	Phosphoenolpyruvate carboxykinase PckA	↓
Cytosolic proteins
82	PA0609	trpE	Anthranilate synthase (EC 4.1.3.27) alpha chain	↑
81	PA1596	htpG	Heat shock protein HtpG	↑
86d	PA3244	minD	Cell division inhibitor MinD	↑
87	PA3397	fpr	Ferredoxin-NADP+ reductase	↑
84	PA3635	eno	Enolase	↑
85	PA4602	glyA3	Serine hydroxymethyltransferase GlyA3	↑
83	PA0139	ahpC	Alkyl hydroperoxide reductase subunit C	↓
73	PA0230	pcaB	3-Carboxy-cis,cis-muconate cycloisomerase PcaB	↓
71	PA0766	mucD	Serine protease MucD precursor	↓
76	PA0837	slyD	Peptidyl-prolyl cis-trans isomerase SlyD	↓
74	PA1337	ansB	Glutaminase asparaginase AnsB	↓
75	PA1344	yvaG	Probable short-chain dehydrogenase	↓
79	PA1584	sdhB	Succinate dehydrogenase (B subunit) SdhB	↓
80	PA1900	phzB2	Probable phenazine biosynthesis protein PhzB2	↓		↑
77d	PA3529	tsaA	Probable peroxidase	↓
78d	PA4366	sodB	Superoxide dismutase SodB	↓
72	PA5173	arcC	Carbamate kinase ArcC	↓

^aSee Fig. 2.

^bData generated from peptide mass maps were compared to the complete translated open reading frames of the PAO1 genome (www.pseudomonas.com).

^cIn order to detect differential protein expression, the spot intensities of the entirety of the fragments of one protein were compared between AZM-treated and nontreated PAO1 cultures. Except for protein spots that were detected only in gels of cultures with or without AZM, the differences in protein spot intensity were between 2- and 10-fold.

^dGenes shown to be regulated by both proteomics and transcriptomics.

FIG. 2.

Secretome of PAO1 cultures without (A) and with (B) the addition of 2 μg/ml AZM. Forty-three differentially expressed protein spots (1 to 43) were identified by mass spectrometry from the gels of the secretome of the PAO1 control cultures (A) and 26 protein spots (44 to 69) were identified from the gels of the AZM-treated PAO1 secretome (B).

The finding that only eight genes and proteins (ahpC, cbpD, cysP, pcrV, minD, tsaA, sodB, and groEL) were shown to be affected at both the transcriptional and the protein levels emphasizes that transcriptomics and proteomics are complementary approaches, and combining their particular strengths should give maximal relevant results. Major differences between proteome and transcriptome data have been documented in previous studies to map the P. aeruginosa QS regulon: the two proteome studies of Arevalo-Ferro et al. (1) and Nouwens et al. (29) identified 47 and 27 QS-regulated proteins, respectively, and only 11 and 8, respectively, of the corresponding genes were found to be regulated at the transcriptional level in three independent transcriptome studies analyzing QS-dependent P. aeruginosa gene expression (8, 42, 49).

Effect of azithromycin on quorum sensing in PAO1.

Eight out of 77 genes (10.4%) of the general QS regulon (identified as QS-dependent genes in all three recent microarray studies (8, 42, 49) were identified as being influenced by AZM addition (Table 1). Overall, 15 AZM- and QS-dependent genes (identified as QS dependent in at least one of the three microarray studies) were found. Ten of the 15 QS- and AZM-dependent genes were shown to be down-regulated in this study in response to AZM, but five genes were up-regulated. An up-regulation is opposite to what is expected when assuming that AZM inhibits QS. However, QS-regulated proteins have previously been found to be oppositely regulated in a comparison between two previous proteomic studies (1, 29) and it was speculated that some of the observed discrepancies are caused by differences in experimental conditions. Further work will be required to address this issue. Moreover, we found genes of the QS-dependent type III secretion system (TTSS) (10) and QS-controlled katA and sodB (7) to be AZM dependent.

In addition to the identification of a common subset between QS- and AZM-regulated genes, we compared differential protein expression due to AZM treatment with QS-dependent protein expression in the two previous proteome studies. There was a large common subset of QS- and AZM-regulated proteins: 15 proteins that were shown to be affected by AZM addition were among the 47 proteins previously identified as being QS dependent (31.9%) (Table 2). The best congruence was found in the secretome, supporting the recent findings of Wagner et al. (50), who reported the constant expression of QS-regulated virulence factors under various culture conditions.

Since chronic P. aeruginosa infection in CF is rarely eradicated despite intensive antimicrobial therapy, interference with or blocking of QS systems has been considered an attractive alternative therapeutic strategy. Recently, halogenated furanones have been shown to control P. aeruginosa infections in animal models (8, 55). The finding that QS antagonists are effective is of considerable importance, since it demonstrates that QS is a useful and promising drug target in vivo. For treatment of humans, macrolides seem to be a promising alternative to the toxic halogenated furanones that might block the QS systems within therapeutic concentration ranges. How precisely the macrolides interfere with the transcription of QS-regulated genes remains poorly defined and will be an important question to be addressed in the future.

Azithromycin enhances the expression of the type III secretion system.

Several studies have demonstrated that macrolide antibiotics suppress the expression of substances that contribute to P. aeruginosa virulence, such as exoenzymes, exopolysaccharides, and pigments (17, 25, 26, 48), and it has been hypothesized that CF patients benefit from AZM treatment due to the negative effect on virulence factor production. However, one of the major findings of this study is that AZM treatment of PAO1 led to increased expression of the TTSS. We found increased expression of 10 of 36 genes of the TTSS gene cluster (PA1690 to PA1725) in the AZM-treated cultures. Moreover, the expression of genes encoding the secreted effector proteins ExoT, ExoY, and ExoS, which are located outside the TTSS gene cluster, was enhanced. The pcrV gene, encoding a TTSS-secreted protein, exhibited the highest differential gene expression (91.9-fold), and accordingly, PcrV was overproduced in the AZM-treated cultures.

In order to test whether TTSS overproduction in P. aeruginosa has a biological effect, we determined the in vitro cytotoxicity of AZM-treated bacteria on the murine macrophage cell line J774.A1. As shown in Fig. 3 P. aeruginosa PAO1 that was cultured in medium containing 2 μg/ml AZM until the log and early stationary phases of growth exhibited increased cytotoxicity in comparison to bacteria cultured without AZM. These effects were not observed when PAO1 was grown with sublethal gentamicin or ceftazidime concentrations (data not shown). Remarkably, it has previously been demonstrated that treatment of P. aeruginosa with macrolides, including AZM, significantly enhances virulence in mice (19). An involvement of acute toxic effects rather than multiplication of the bacteria was suggested.

FIG. 3.

Cytotoxicity as determined by lactate dehydrogenase release of J774.A1 cells of AZM-treated bacteria (▴) versus the untreated PAO1 control (▪). The bacteria were harvested from the log phase of growth (A) and the stationary phase of growth (B). One representative experiment out of three is shown.

The results of our study imply that enhanced expression of the TTSS might account for these previously observed effects, as the TTSS has been shown to enhance P. aeruginosa virulence significantly (43). However, only inoculation of macrolide-treated bacteria had been shown to be associated with increased mortality in mice, whereas the administration of macrolides after inoculation of P. aeruginosa did not increase mortality (19, 28). The impact of the TTSS-induced cytotoxicity on therapeutic macrolide administration remains to be clarified in future investigations.

Azithromycin affects P. aeruginosa motility.

AZM-treated PAO1 cultures exhibited reduced expression of various proteins required for flagellum biosynthesis (Table 2) and demonstrated reduced flagellum-driven motility on swimming agar plates. The mean value of the radius of the untreated PAO1 and AZM-treated PAO1 cultures was 3.58 (±0.17) cm and 1.18 (±0.09) cm, respectively (P < 0.01). Although macrolides have been previously demonstrated to inhibit not only swimming motility (14) but also twitching motility (14, 54), we observed increased expression of pilA and pilH, involved in type IV pilus biogenesis (Table 1). No effects of sublethal AZM concentrations on type IV pili could be observed by proteomics.

Analysis of the twitching motility of AZM-treated PAO1 revealed that twitching motility was significantly reduced. The mean value of the radius of the untreated PAO1 and AZM-treated PAO1 cultures was 0.67 (±0.06) cm and 0.51 (±0.06) cm, respectively (P < 0.01). While flagellum-mediated motility has been implicated to be required to bring P. aeruginosa within proximity of a surface, type IV pili by virtue of twitching motility enable P. aeruginosa to migrate across a surface, recruit cells from adjacent monolayers, and form cell aggregates (31), thus contributing to biofilm formation. The observed impaired swimming and twitching motility of AZM-treated PAO1 could explain the previous observations that AZM delays biofilm formation, as evidenced by decreased biomass (6) and impaired alginate production (12, 27).

Azithromycin affects the oxidative stress response in PAO1.

Another major finding of this study is that AZM treatment obviously led to an impaired oxidative stress response in P. aeruginosa. The superoxide dismutase SodB, the catalase KatA, and the alkylhydroperoxide reductase AhpC, all of which contribute significantly to the stress response in P. aeruginosa (32), were shown to be repressed at the transcriptional level (sodB, ahpC, and katA) and the protein level (SodB and AhpC) upon AZM addition. We also observed an up-regulation of the fur gene in response to sublethal AZM concentrations. fur is involved in the regulation of iron uptake under iron-limiting conditions. However, fur has also been shown to be up-regulated under oxidative stress (32) and simultaneous overexpression of the bfrB gene indicates a sufficient intracellular iron storage (47).

Another important element affected by oxidative stress is sulfur, since iron-sulfur proteins have been shown to play a protective role against oxidative stress (18). The sulfate binding protein of an ABC transporter (CysP) was one of the eight proteins that were demonstrated to be up-regulated upon AZM addition both at the transcriptional level and at the protein level. PA3262, a putative peptidyl-prolyl isomerase that probably corrects misfolding caused by the damage of reactive oxygen intermediates, was up-regulated at the transcriptional level. Similar, the thiol-disulfide oxireductase (DsbA), which has been shown to be responsible for protein thiol modifications in Escherichia coli (23), was up-regulated in the secretome of AZM-treated PAO1. Thiol-disulfide interconversion plays a crucial role in the control of cellular redox potential and the prevention of oxidative damage.

PA3529, encoding a probable peroxidase, as well as the genes encoding the heat shock proteins GroEL (also shown to be differentially expressed at the protein level) and GroES were down-regulated by AZM treatment. The observed effects of AZM on several genes and proteins that are involved in the oxidative stress response implied that AZM might affect long-term survival of P. aeruginosa during chronic infection. Thus, we tested whether AZM-exposed cultures were more sensitive to H₂O₂ treatment than the untreated controls. As shown in Fig. 4 bacteria that were treated with AZM were significantly more susceptible when exposed to H₂O₂ on solid agar. Prolonged growth of PAO1 in AZM-supplemented cultures increased bacterial susceptibility to H₂O₂, whereas gentamicin or ceftazidime pretreatment had no effect (data not shown).

FIG. 4.

Growth inhibition by H₂O₂ as determined by agar diffusion assays. Prolonged cultivation (≥10 h) of PAO1 in AZM-supplemented medium significantly increased sensitivity to H₂O₂ (t test; P < 0.0017). AZM-treated PAO1 (gray bars) was compared with untreated PAO1 (white bars). Results are given as the mean ± standard deviation of three determinations.

Concluding remarks.

Both the antimicrobial and anti-inflammatory effects of AZM have been implicated as being responsible for the improvement in CF patient outcome. The results of this study clearly indicate that there is an antipseudomonal effect of AZM that is linked to reduced virulence factor production, biofilm formation, and survival under stressful conditions due to interference with QS in P. aeruginosa. We identified a large common subset of QS- and AZM-regulated genes/proteins, in particular within the secretome, comprising many virulence factors. Moreover, the TTSS, which was previously shown to be negatively regulated by QS (10), was induced upon AZM addition, whereas in accordance with the results of a study reporting QS control of genes essential for relieving oxidative stress (7), we found markedly increased sensitivity of AZM-treated PAO1 cultures to H₂O₂. Moreover, our data on P. aeruginosa motility are in accordance with the observation that AZM retards biofilm formation (6), which has been reported to be dependent on the QS systems.

Our in vitro data imply that the QS-antagonistic activity of AZM contributes to the improvement of CF patient health. Apart from the reduced expression of virulence factors, interference with the bacterial oxidative stress response might be of major relevance. One vitally adaptive response of P. aeruginosa is the ability to resist the oxidative stress that is induced during phagocytosis, when the bacteria are confronted with reactive oxygen intermediates such as H₂O₂, O₂⁻, and OH⁻ from the respiratory burst of human phagocytes. Polymorphonuclear cells are the major effector cells responsible for the clearance of P. aeruginosa from the site of infection, and the inflammatory response in the chronically infected CF lung in particular is accompanied by very high levels of reactive oxygen intermediates that the bacteria must survive to be able to persist. Thus, the impaired oxidative stress response might account for the observed beneficial effects of AZM treatment and for the significant reduction of PAO1 viability after prolonged incubation with sub-MIC concentrations of AZM that has been reported previously (46).

Macrolides inhibit protein synthesis at the ribosomal level, and it is conceivable that unidentified stress responses, bacterial regulons, or signal transduction processes are responsible for the observed effects of sublethal concentrations on gene expression. Furthermore, future studies will have to elucidate whether the observed effects of sublethal AZM concentrations are also relevant in vivo. However, the results of this in vitro study and the fact that AZM exhibits beneficial effects in the treatment of CF patients give us reason to assume that the administration of AZM for CF will have a great impact on the management of chronic infection due to its interference with P. aeruginosa QS and thus with virulence factor production, biofilm formation, and persistence during chronic infection.