Cinnamaldehyde Induces Expression of Efflux Pumps and Multidrug Resistance in Pseudomonas aeruginosa

Essential oils or their components are increasingly used to fight bacterial infections. Cinnamaldehyde (CNA), the main constituent of cinnamon bark oil, has demonstrated interesting properties in vitro against various pathogens, including Pseudomonas aeruginosa. In the present study, we investigated the mechanisms and possible therapeutic consequences of P. aeruginosa adaptation to CNA.

ABSTRACT

Essential oils or their components are increasingly used to fight bacterial infections. Cinnamaldehyde (CNA), the main constituent of cinnamon bark oil, has demonstrated interesting properties in vitro against various pathogens, including Pseudomonas aeruginosa. In the present study, we investigated the mechanisms and possible therapeutic consequences of P. aeruginosa adaptation to CNA. Exposure of P. aeruginosa PA14 to subinhibitory concentrations of CNA caused a strong albeit transient increase in the expression of operons that encode the efflux systems MexAB-OprM, MexCD-OprJ, MexEF-OprN, and MexXY/OprM. This multipump activation enhanced from 2- to 8-fold the resistance (MIC) of PA14 to various antipseudomonal antibiotics, including meropenem, ceftazidime, tobramycin, and ciprofloxacin. CNA-induced production of pump MexAB-OprM was found to play a major role in the adaption of P. aeruginosa to the electrophilic biocide, through the NalC regulatory pathway. CNA was progressively transformed by bacteria into the less toxic metabolite cinnamic alcohol (CN-OH), via yet undetermined detoxifying mechanisms. In conclusion, the use of cinnamon bark oil or cinnamaldehyde as adjunctive therapy to treat P. aeruginosa infections may potentially have antagonistic effects if combined with antibiotics because of Mex pump activation.

INTRODUCTION

Pseudomonas aeruginosa is a major opportunistic pathogen involved in a wide range of acute and chronic infections (1). This versatile microorganism is notoriously known for its capacity to thrive in challenging environments thanks to a plethora of adaptive mechanisms (2, 3). Accounting for its relatively high intrinsic resistance to toxic molecules, P. aeruginosa is surrounded by a poorly permeable outer membrane that allows extrusion mechanisms, such as efflux systems, to operate efficiently (4). Some of these transmembrane nanomachineries are able to export antibiotics and, thus, to impair the interaction of drugs with their cellular targets (5). At least four efflux pumps of the resistance-nodulation-cell division (RND) family can significantly increase the resistance of P. aeruginosa to antibiotics when overproduced upon mutations, namely, MexAB-OprM, MexCD-OprJ, MexEF-OprN, and MexXY/OprM (5). These polyspecific transporters accommodate a wide range of structurally unrelated antimicrobials, including quinolones, macrolides, tetracyclines, and phenicols, but efflux of aminoglycosides and β-lactams (except cefepime) is specific to MexXY/OprM and MexAB-OprM, respectively (6).

Over the last years, the sustained emergence and spread of multiresistant strains of P. aeruginosa in hospitals have resulted in hard-to-treat infections (7, 8). When therapeutic options are limited, or to avoid the use of antibiotics, essential oils from aromatic plants are sometimes viewed as alternatives to treat non-severely ill patients. Essential oils containing an aldehyde group (e.g., citral), a phenyl group (e.g., thymol, eugenol, carvacrol, and terpineol), or both (e.g., cinnamaldehyde) indeed exhibit antimicrobial properties in vitro against various Gram-negative and Gram-positive species when used alone (9–11) or in combination (12). However, the fact that some of these oils activate bacterial efflux systems might potentially compromise the clinical efficacy of antibiotic treatments. For instance, the tea tree (Melaleuca alternifolia) oil increases expression of MexAB-OprM in P. aeruginosa (13), while the thyme (Thymus maroccanus) and pine (Pinus sp.) oils activate AcrAB-TolC in Escherichia coli (14, 15). The S-phenyl-l-cysteine sulfoxide from the tropical plant Petiveria alliacea reduces the virulence of P. aeruginosa by inhibiting kynureninase KynU, but it stimulates MexEF-OprN expression (16).

The observation that RND pumps can transiently be upregulated by environmental signals is not really surprising per se since efflux systems are a major component of adaptive responses in bacteria (17). For example, MexAB-OprM (18) and MexXY/OprM (19) are both activated by oxidative stress, MexCD-OprJ by membrane damages (20), and MexEF-OprN by nitrosative (21) or disulfide (22) stress. While it has become clear that mex genes are overexpressed under such harsh conditions (18–22), the impact of this increased efflux activity on antibiotic resistance has remained unexplored. Recently, we found that MexEF-OprN production is strongly augmented when P. aeruginosa is submitted to electrophilic stress by aldehydes, such as glyoxal, methylglyoxal, and cinnamaldehyde (CNA) (23). However, the intrinsic resistance of the microorganism to CNA appeared to be only partially dependent upon MexEF-OprN, thereby suggesting that other efflux systems also contribute to the protection against this compound (23).

In the present work, we demonstrate that CNA, the major component of cinnamon bark oil, triggers the expression of four major RND efflux pumps (MexAB-OprM, MexCD-OprJ, MexEF-OprN, and MexXY/OprM) and incidentally reinforces the resistance of P. aeruginosa to several antipseudomonal antibiotics. Results of time-kill experiments also show that MexAB-OprM is an essential mechanism in the defense of P. aeruginosa against CNA.

RESULTS

CNA induces pumps expression in P. aeruginosa.

To assess the impact of CNA on efflux systems in P. aeruginosa, wild-type reference strain PA14 was challenged with 0.25× and 0.5× MIC of CNA (256 and 512 μg/ml, respectively) in Mueller-Hinton broth. At a concentration of 256 μg/ml, the agent reduced the growth rate of PA14 for approximately 1 h, while at 512 μg/ml a transient bacteriostasis occurred that was followed by bacterial regrowth (see Fig. S1 in the supplemental material). The expression levels of genes mexB, mexC, mexE, and mexY were determined by real-time quantitative PCR (RT-qPCR) at different time points (t): t₀, t_15min, t_30min, and t_1h postexposure. As shown in Fig. 1, all these genes were activated in response to CNA, although with some differences in their temporal expression. While mexB and mexE were strongly and rapidly (t_15min) upregulated in the presence of 0.25× MIC CNA, their transcripts dropped to finally return to their uninduced baselines at t_1h. This gene activation was lower in bacteria exposed to 0.5× MIC, possibly as a result of an inverse correlation with pump MexCD-OprJ expression. Indeed, gene mexC displayed similar activation kinetics as mexB and mexE upon treatment with 0.5× MIC CNA, but the 0.25× MIC dose failed to induce its transcription. Finally, gene mexY expression gradually increased in bacteria challenged with the highest concentration of CNA, to reach a maximum at t_1h. As for mexC, no induction of mexY was visible under the 0.25× MIC condition. Compared with previously established thresholds (24–26), all the mex genes studied were found to be significantly upregulated (Fig. 1, red dotted-line) at levels close to those of in vitro mutants of PA14 overproducing MexAB-OprM (named PA14mexAB⁺), MexCD-OprJ (PA14mexCD⁺), MexEF-OprN (PA14mexEF⁺), or MexXY/OprM (PA14mexXY) (Fig. 1, solid red line). Overall, these results demonstrated that CNA has dramatic, although unequal, effects on the expression of these four pumps in P. aeruginosa.

FIG 1.

Expression levels of genes mexB, mexC, mexE, and mexY. Gene transcripts were measured over time by RT-qPCR in strain PA14 exposed to 1% DMSO (negative control, in white) and to 256 and 512 μg/ml CNA (in light and dark grey, respectively). The genes were considered activated (indicated by dotted red lines) when their expression was ≥3-fold (mexB), ≥5-fold (mexY), and ≥20-fold (mexE and mexC) that of unexposed PA14 at t₀. Indicated by solid red lines are the mRNA amounts of PA14-derived mutants PA14mexAB⁺ (mexB, 9.5-fold that of PA14), PA14mexCD⁺ (mexC, 151-fold), PA14mexEF⁺ (mexE, 411-fold), and PA14mexXY⁺ (mexY, 24-fold) that individually overexpress a different efflux pump, in the absence of CNA.

CNA induces antibiotic resistance.

Because these efflux pumps mediate antibiotic resistance when overproduced, we determined the susceptibility of CNA-challenged strain PA14 to some of their substrates, by the dilution method in Mueller-Hinton agar medium (Table 1). In concordance with CNA-induced expression of MexAB-OprM, MICs of ticarcillin, aztreonam, and ceftazidime increased up to 4-fold in the presence of 128 to 512 μg/ml biocide. In contrast, the MICs remained unchanged in bacteria lacking the mexAB genes (PA14ΔAB) or in mutant PA14mexAB+ that constitutively overexpresses operon mexAB-oprM as a result of repressor MexR alteration (Table 1). CNA-promoted resistance to meropenem, another substrate of MexAB-OprM, was not totally suppressed in PA14ΔAB because of MexT-dependent downregulation of the gene encoding the carbapenem-specific porin OprD (from 5- to 7-fold with respect to nonexposed PA14ΔAB). As demonstrated elsewhere, the LysR-type regulator MexT positively controls the expression of operon mexEF-oprN and conversely represses gene oprD activity (27).

TABLE 1.

Influence of CNA on MICs of antibiotics

Antibiotica	CNA (μg/ml)	MICb (μg/ml) of antibiotics for strain:
		PA14	PA14ΔAB	PA14ΔarmR	PA14mexAB+	PA14ΔXY	PA14mexXY+	PA14ΔEFN	PA14mexEFN+	PA14ΔAB/EFN	PA14ΔAB/XY/EFN/CDJ
Ticarcillin	0	16	8	16	128
	128	32	8	16	128
	256	64	8	16	128
	512	64	c	c	128
Aztreonam	0	4	0.25	4	32
	128	8	0.25	4	32
	256	16	0.25	4	32
	512	16	c	c	32
Meropenem	0	0.5	0.125	0.5	2
	128	2	0.25	0.5	2
	256	4	0.25	1	2
	512	2	c	1	2
Ceftazidime	0	1	0.25	1	4
	128	1	0.25	1	4
	256	2	0.25	1	4
	512	2	c	1	4
Gentamicin	0	1		1		0.25	4
	128	1		1		0.25	4
	256	2		2		0.25	4
	512	2		2		0.5	4
Tobramycin	0	0.5		0.5		0.25	2
	128	0.5		0.5		0.25	2
	256	1		1		0.5	2
	512	2		2		1	2
Ciprofloxacin	0	0.125	0.125	0.125	0.5	0.125	0.5	0.125	4	0.06	≤0.03
	128	0.5	0.5	0.5	1	0.5	0.5	0.5	4	0.25	≤0.03
	256	1	1	1	1	1	1	0.5	4	c	c
	512	1	c	1	1	1	1	0.5	4	c	c

^aTicarcillin, aztreonam, meropenem, and ceftazidime are substrates of MexAB-OprM; gentamicin and tobramycin are substrates of MexXY/OprM; and ciprofloxacin is a substrate of the 4 efflux pumps.

^bMIC values in boldface indicate an increase in bacterial resistance due to the presence of CNA.

^cBacterial growth was inhibited by 256 or 512 μg/ml CNA.

Seemingly concordant with CNA activating the efflux system MexXY/OprM, MICs of two of the pump substrates, namely, gentamicin and tobramycin, were increased 2- and 4-fold, respectively, in response to CNA exposure (≥256 μg/ml). However, this effect was no longer observed in the mexXY-overexpressing mutant PA14mexXY⁺ and was not abolished in mexXY deletion mutant PA14ΔXY either, thereby clearly indicating intervention of a mechanism different from MexXY/OprM in CNA-promoted resistance to aminoglycosides (Table 1).

Finally, strain PA14 susceptibility to ciprofloxacin, a fluoroquinolone actively exported by MexEF-OprN (28), and to a lesser extent by MexAB-OprM, MexCD-OprJ, and MexXY/OprM (29), was reduced up to 8-fold upon CNA challenge. This enhanced resistance could not be cancelled by single or double pump deletions but was suppressed in the quadruple mutant PA14ΔAB/XY/EFN/CDJ (Table 1). Interestingly, this highly efflux-deficient mutant was more susceptible to CNA than its wild-type parent PA14 and could not grow in the presence of CNA concentrations higher than 128 μg/ml. Furthermore, strongly suggesting a contribution of efflux systems MexAB-OprM and MexEF-OprN to the intrinsic resistance of P. aeruginosa to CNA, the double mutant PA14ΔAB/EF was unable to develop on 256 μg/ml either.

To check whether or not the antagonistic interactions observed between CNA and efflux substrates were specific to the PA14 genetic background, double-disk antagonism tests were performed with ciprofloxacin on clinical (PAO1 and LESB58) and environmental (1393, 1349, and 1429) strains (listed in Table S1 in the supplemental material). In agreement with our previous results, CNA induced a resistance to ciprofloxacin in all the five strains tested (see Fig. S2 in the supplemental material).

Efflux systems provide a protection against CNA.

The susceptibility of deficient and overproducing efflux pump mutants to CNA was first assessed by the dilution method in Mueller-Hinton agar. MexAB-OprM was the only system studied whose inactivation resulted in a modest but reproducible decrease in resistance to CNA (MIC equal to 512 μg/ml versus 1,024 μg/ml for its wild-type parent PA14) (see Table S2 in the supplemental material). Suppression of operon mexEF-oprN in the ΔmexAB background (mutant PA14ΔAB/EFN) further lowered CNA MIC from 512 to 256 μg/ml, thus confirming a role for MexEF-OprN in the intrinsic resistance to this compound only when MexAB-OprM is inactive. In contrast, deletion of mexCD-oprJ or mexXY had virtually no impact on the susceptibility of P. aeruginosa to CNA when single (PA14ΔCDJ and PA14ΔXY), double (PA14ΔXY/CDJ and PA14ΔEFN/XY), and triple mutants (PA14ΔAB/EFN/XY) were compared with their mexCD-oprJ and mexXY proficient counterparts (Table S2). Finally, the quadruple mutant PA14ΔAB/XY/EFN/CDJ showed the highest susceptibility to the electrophile (MIC equal to 128 μg/ml).

To confirm these results, we carried out time-kill experiments on bioluminescent strain PA14-lux and various efflux-deficient derivatives. CNA was added to bacterial cultures in Mueller-Hinton broth at 700 μg/ml since a concentration of 1,024 μg/ml appeared to be too bactericidal and 512 μg/ml was unable to kill PA14-lux significantly (see Fig. S3 in the supplemental material). Compared with wild-type strain PA14, inactivation of pump MexEF-OprN (mutant PA14ΔEFN), MexCD-OprJ (PA14ΔCDJ), or MexXY/OprM (PA14ΔXY) had only a modest impact on bacterial killing (Fig. 2). Two hours postexposure, a regrowth of all the mutants occurred as for PA14. The bactericidal activity of CNA was much stronger on mexAB-deleted mutants PA14ΔAB, PA14ΔAB/EFN (not shown), and PA14ΔAB/EFN/CDJ/XY, with a shift in bioluminescence of ≥4 log₁₀ relative light units (RLUs) within the first 60 min of treatment (Fig. 2). Drug-free Mueller-Hinton agar (MHA) plates inoculated with culture samples of the quadruple mutant collected 18 h after CNA addition remained sterile, thus demonstrating a complete killing of these efflux-deficient bacteria. Altogether, these data highlighted the predominant role played by MexAB-OprM in CNA tolerance, compared with the other efflux systems investigated. However, contrasting with these observations, none of the efflux pump-overproducing mutants, including PA14mexAB+, appeared to be more resistant to this toxic agent than PA14 (Table S2). As mentioned previously (Fig. 1), mexB expression levels were quite similar in the CNA-exposed PA14 strain (t_15min) and in PA14mexAB⁺, which supports the notion that MexAB-OprM contributes similarly to the bacterial resistance against the biocide, whether its dysregulation is induced or mutational. Of note, MICs of CNA were lower for PA14mexCD+ than for wild type PA14 (512 μg/ml versus 1,024 μg/ml) likely because of the impaired activity of MexAB-OprM in this nfxB mutant (26).

FIG 2.

Bactericidal activity of CNA on P. aeruginosa. Bioluminescent strain PA14-lux and its derived mutants PA14ΔAB-lux, PA14ΔEFN-lux, PA14ΔXY-lux, PA14ΔCDJ-lux, and PA14ΔAB/EFN/XY/CDJ-lux were cultured to mid-log phase and then challenged with 700 μg/ml CNA. Bioluminescence (Log RLU), used as an indicator of cell survival, was recorded in triplicate during 5 h. RLU values are mean values ± standard deviations. A bioluminescence threshold was established with sterile MHB (shaded zone).

CNA induces mexAB-oprM expression via the ArmR regulatory pathway.

In order to determine which regulatory circuitry activates MexAB-OprM production upon CNA stress, we measured the mRNA levels of genes known to control either positively (cpxR and armR) or negatively (mexR, ampR, nalC, nalD, mexT, and mdrR1) the mexAB-oprM operon (Table 2). Among this set of genes, only those of the NalC regulatory pathway increased rapidly and strongly when strain PA14 was submitted to 256 or 512 μg/ml of biocide, namely, mexR, nalC, armR, and PA3720. Fully consistent with our previous results (Fig. 1), this burst in efflux-related gene expression was transient and no longer visible at t_1h except for PA3720, a gene of unknown function. Genes PA3720 and armR form an operon whose transcription is repressed by the product of divergently oriented gene nalC (see Fig. S4 in the supplemental material) (30). Confirming the role of the ArmR/NalC pathway in the adaptive response of P. aeruginosa to CNA, gene mexB expression remained unchanged in mutant PA14ΔarmR exposed to a concentration of 256 μg/ml (1.3-fold versus 8-fold in PA14), as were the MICs of MexAB-OprM substrates ticarcillin, aztreonam, and ceftazidime (Table 1). As expected, deletion of armR had no impact on bacterial resistance to gentamicin, tobramycin, and ciprofloxacin which are preferred substrates of pumps MexCD-OprJ, MexEF-OprN, and/or MexXY/OprM and had only a partial influence on meropenem MICs because of MexT-dependent negative control of porin OprD-encoding gene (27) (Table 1).

TABLE 2.

Expression of operon mexAB-oprM regulatory genes after exposure of strain PA14 to 256 or 512 μg/ml of CNA

Regulatory genes by typea	Transcript levels after:b			Target of the regulator	Reference
	15 min	30 min	1 h
Repressor
mexR	25 (40)	2.9 (8.4)	2.3 (5.6)	Distal promoter of mexAB-oprM	50
nalC	21 (14.7)	1.1 (3.5)	1.1 (1.2)	Promoter of PA3720-armR	51
nalD	2.2 (2.4)	0.9 (1.2)	0.9 (0.7)	Proximal promoter of mexAB-oprM	52
ampR	3 (2.7)	1 (0.6)	1 (0.5)	Promoter of mexAB-oprM	53
mexT	7.7 (1.8)	4.3 (0.4)	1 (0.6)	Promoter of mexAB-oprM	54
mdrR1	1.8 (0.8)	2 (1.2)	0.6 (1.1)	Proximal promoter of mexAB-oprM	55
Activator
cpxR	1.7 (1.1)	1 (0.6)	1 (0.8)	Distal promoter of mexAB-oprM	56
Antirepressor
armR	61 (95)	1.9 (43)	1.2 (7.5)	MexR protein	30
PA3720	206 (255)	46 (218)	20 (144)	Unknown function (in operon with armR)

^aRegulatory genes implicated in modulation of MexAB-OprM under biofilm conditions (5) were not evaluated here (rocS2-A2 and brlR). Repressors MdrR1 and MdrR2 are both required in strain PAO1 for binding to the promoter region of mexAB-oprM (55). However, only mdrR1 expression could be measured because gene mdrR2 is absent from the genome of PA14.

^bMean value expressed as a ratio to that of PA14 when grown in MHB broth containing 1% DMSO. Values in bold face highlight the increased expression of regulatory pathway nalC-PA3720/armR-mexR. Values outside and within the parentheses correspond to 256 and 512 μg/ml of CNA, respectively.

CNA is degraded by P. aeruginosa.

The observation that several Mex systems were transiently activated upon CNA exposure suggested that the biocide might be transformed into less toxic metabolites by non-efflux-based detoxifying mechanisms. The CNA molecule has two electrophilic reactive sites, namely, a conjugated double bond and an aldehyde group, that can react with nucleophilic molecules (e.g., proteins, reduced glutathione, and DNA) to form cinnamic alcohol (CN-OH), cinnamic acid, and adducts (31). Visvalingam et al. found that E. coli is able to enzymatically convert CNA into cinnamic alcohol through dehydrogenases/reductases YqhD and DkgA (32). To determine if a similar inactivation process exists in P. aeruginosa, we analyzed the composition of culture supernatants of CNA-exposed strain PA14 by thin-layer chromatography (TLC). Fig. 3 shows that CNA started to quantitatively decrease after 1 h of incubation and was no longer detectable after 3 h, while the amounts of cinnamic alcohol (CN-OH) increased. After 24 h, this metabolite could not be detected in the supernatants either (Fig. 3).

FIG 3.

TLC of cinnamaldehyde metabolites. Samples were chromatographed on thin-layer plates (Alugram Xtra SIL G UV254), developed with cyclohexane:ethyl acetate (7:2, vol:vol). (A) Separation and detection of standards (10 μg CNA, 10 μg cinnamic alcohol [CN-OH], and a mixture of 10 μg CNA + 10 μg CN-OH), and (B) detection of CNA metabolites in the supernatant of PA14 growing medium 0, 1, 3, and 24 h postexposure to CNA at 512 μg/ml.

As an indication that CN-OH is somewhat less toxic than CNA toward P. aeruginosa, the MIC of this compound was 2-fold higher than that of CNA (Table S2). Induction of mexB, mexC, mexE, and mexY expressions by 256 (0.125× MIC) or 512 μg/ml (0.25× MIC) CN-OH was assessed by RT-qPCR after 15 min of contact with bacteria. Only gene mexB was activated by CN-OH (up to 9.5-fold at 512 μg/ml) with respect to an unexposed control, which is a result that suggests that the aldehyde group of CNA (lacking in CN-OH) partially supports the toxic and efflux-inducing activities of the biocide. Upregulation of MexAB-OprM by CN-OH at 256 or 512 μg/ml was correlated with a significant augmentation (2- to 4-fold) of PA14 resistance to the pump substrates (see Table S3 in the supplemental material).

DISCUSSION

Cinnamon bark oil is traditionally used in anti-infective tropical medicine and tends to be a substitute for antibiotics to fight mild infections in nontropical areas. Its main component, CNA, displays a quite strong inhibitory activity against a variety of bacterial pathogens, including Listeria monocytogenes (33), Clostridium botulinum, Staphylococcus aureus, E. coli O157:H7, Salmonella enterica serovar Typhimurium (33, 34), and P. aeruginosa (35). Because of its highly bioreactive aldehyde group, CNA interacts with multiple cellular components and functions, including membrane lipids, ATPases, and cell division (36, 37). Furthermore, by antagonizing the quorum sensing regulatory systems of P. aeruginosa, CNA has potentially interesting properties to treat biofilm-associated infections (38). However, a previous study showed that its bactericidal activity against the microorganism is not potentiated by antibiotics, except colistin in a minority of clinical isolates (39).

Reminiscent of the action of some other essential oils (13, 14), we found that P. aeruginosa adapts to relatively elevated CNA concentrations by upregulating several Mex pumps, of which MexAB-OprM is essential for its survival. Such a multipump activation likely results from a global response to the stress generated by CNA (23). The energy cost of such a high efflux activity might explain the transient growth arrest observed during the first hour of CNA exposure. In agreement with this, the decline in expression of genes mexB, mexC, and mexE was concomitant with the recovery of bacterial multiplication (Fig. 1 and Fig. S1). However, the respective contribution of MexAB-OprM, MexCD-OprJ, MexEF-OprN, and MexXY/OprM to the extremely fast adaptation process to CNA remains to be clarified, as the kinetics and degree of upregulation of these systems were somewhat different. Compensation of one pump with another at a certain time point of CNA treatment cannot be excluded, as an inverse correlation between the activity of several Mex systems has been noted in single pump deletion mutants (40). Since the expression of operons mexCD-oprJ, mexEF-oprN, and mexXY was unaffected in bacteria challenged with the CNA metabolite CN-OH, one can assume that their transient activation in CNA-treated cells is due to CNA and not CN-OH. This working hypothesis would not hold for mexAB-oprM, the transcription of which was still induced by CN-OH exposure.

E. coli encodes several dehydrogenases/reductases, some of them (YqhD, DkgA, and YahK) being implicated in the biotransformation of CNA into CN-OH (32, 41). YahK, a cinnamyl alcohol dehydrogenase (CAD family), shares 67% amino acid sequence identity with a predicted protein, PA2275, from P. aeruginosa. Since the PA2275-encoding gene was highly overexpressed in bacteria exposed to CNA (23), we constructed a deletion mutant (PA14Δ2275) to better assess its physiological function. The deletion had virtually no impact on the formation of CN-OH (observed by thin-layer chromatography at t_1h postexposure; data not shown), indicating that one or several PA2275-independent pathways operate in CNA degradation.

An interesting issue is to understand how CNA can trigger the activation of so many efflux pumps in P. aeruginosa. As depicted in Fig. 3, the CNA molecule contains both an electrophilic aldehyde group and a lipophilic aromatic ring. The effectiveness of some lipophilic antimicrobial agents relies on their ability to readily dissolve into and/or to interact with bacterial membranes (42). For example, membrane-damaging agents, such as chlorhexidine (20) and benzalkonium chloride (43), are known inducers of operon mexCD-oprJ expression. As demonstrated previously in our laboratory, the aldehyde group of glyoxal or methylglyoxal strongly activates the production of MexEF-OprN and MexXY/OprM in treated bacteria (23). In agreement with these data, the present study shows that the conversion of the aldehyde group to a hydroxyl group abolishes the capacity of CNA to upregulate MexCD-OprJ, MexEF-OprN, and/or MexXY/OprM. Cinnamic acid and hydrocinnamic acid which are closely related to CNA but possess a carboxylic group in place of the aldehyde group also failed to induce the expression of these 3 pumps (data not shown).

The phenyl ring, which is shared by CNA and CN-OH molecules, likely plays a major role in the MexAB-OprM dysregulation. Indeed, phenolic acids have the capacity to induce the expression of operon acrAB-tolC (encoding a MexAB-OprM homolog) in Erwinia chrysanthemi (44) and pentachlorophenol triggers that of operon mexAB-oprM in P. aeruginosa (45). In line with the latter example, the aromatic ring of CNA and CN-OH would interact with repressor NalC, allowing antirepressor ArmR to be produced, with subsequent mexAB-oprM overexpression (45). However, it remains unclear why gene mexB expression dropped rapidly (t_1h), while CNA and CN-OH were still quantitatively present in the culture medium of challenged bacteria (i.e., the persistence of the phenyl moiety of CNA should have still stimulated mexAB-oprM transcription).

CNA-exposed bacteria were more resistant (up to 4-fold) to major antipseudomonal β-lactams, in part because of the higher efflux activity of MexAB-OprM. On the other hand, the mechanisms involved in CNA-dependent induction of resistance to aminoglycosides (up to 4-fold) remain to be fully elucidated. Thus, despite notable anti-infective properties, the major component of cinnamon bark oil may potentially cause therapeutic problems if used in combination with antibiotics to fight P. aeruginosa infections. Its bactericidal effects on the pathogen are transient because of defense mechanisms involving several efflux systems and a still unknown degradation pathway. Clearly, MexAB-OprM is a key element of the adaptive response to CNA in P. aeruginosa. Whether this system is able to efflux CNA or some of its metabolites warrants further investigations. The possibility that CNA selects MexAB-OprM-overproducing mutants in vivo should also be addressed to avoid the emergence of multiple resistance to antibiotics, especially in cystic fibrosis patients who use quite high doses of essential oils to control or prevent lung colonization by P. aeruginosa. From a more general perspective, a better knowledge of in vivo concentrations of the major components of essential oils would certainly be useful to establish whether these concentrations can be therapeutically relevant or, in contrast, would antagonize the efficacy of antibiotics.

MATERIALS AND METHODS

Bacterial strains, plasmids, and growth conditions.

The reference strains, mutants, and plasmids used in this study are listed in Table S1. All the bacterial cultures were incubated at 37°C in Mueller-Hinton broth (MHB) with adjusted concentrations of Ca²⁺ (from 20 to 25 μg/ml) and Mg²⁺ (from 10 to 12.5 μg/ml) (Becton, Dickinson and Company, Cockeysville, MD) or on Mueller-Hinton agar (MHA) (Bio-Rad, Marnes-la-Coquette, France) supplemented with antibiotics when required. Cinnamaldehyde (CNA), cinnamic alcohol (CN-OH), and dimethylsulfoxyde (DMSO) were obtained from Sigma-Aldrich.

Construction of PA14 efflux-defective mutants.

In order to study the role of active efflux in bacterial tolerance to CNA, overproducing and defective efflux mutants were constructed from wild-type strain PA14. Spontaneous efflux-overexpressing mutants were selected on MHA plates containing 64 μg/ml ticarcillin (MexAB-OprM overproducers), 2 μg/ml gentamicin (MexXY/OprM), and 2 μg/ml ciprofloxacin (MexCD-OprJ or MexEF-OprN). After confirmation of gene mexB, mexY, mexC, or mexE overexpression by real-time quantitative PCR (RT-qPCR) and appropriate primers (see Table S4 in the supplemental material), a clone of each selection panel was randomly chosen to identify its mutation. Sequencing experiments with primers listed in Table S4 revealed that all of the tested clones exhibited nonsynonymous mutations in known regulatory genes (5), including Δ3-bp in mexR (mutant PA14mexAB⁺), Δ620-bp in mexZ (PA14mexXY⁺), Q₁₁₆Stop in nfxB (PA14mexCD⁺), and +1-bp in mexS (PA14mexEF⁺).

Single mexAB, mexXY, and mexCD-oprJ deletion mutants were constructed from PA14 (PA14ΔAB, PA14ΔXY, and PA14ΔCDJ, respectively) using overlapping PCRs and recombination events, according to the protocol of Kaniga et al. (46) modified by Richardot et al. (25). The allelic exchanges were checked by PCR. Sequencing experiments confirmed the deletion of 5,238 bp in gene mexAB, 3,750 bp in gene mexXY, and 5,705 bp in operon mexCD-oprJ. A mutant lacking the entire operon mexEF-oprN (PA14ΔEFN) was already available in the laboratory (23). Double (PA14ΔAB/EFN, PA14ΔEFN/XY, and PA14ΔXY/CDJ), triple (PA14ΔAB/EFN/XY), and quadruple (PA14ΔAB/EFN/XY/CDJ) mutants defective in several Mex pumps were successively obtained by using the same approach as described above (Table S1). Likewise, a fragment of 289 bp was deleted from regulatory gene armR in wild-type strain PA14 (PA14ΔarmR).

Drug susceptibility testing.

The MICs of selected antibiotics were determined by the standard serial 2-fold dilution method in MHA with inoculums of 10⁴ CFU per spot, as recommended by the CLSI (47). Bacterial growth was visually assessed after 18 h of incubation at 37°C. To confirm the results, MIC experiments were performed in triplicate. To evaluate the impact of main cinnamon oil components on antibiotic resistance, drug MICs were determined by using a MHA medium supplemented with subinhibitory concentrations of CNA or CN-OH (128, 256, or 512 μg/ml in 1% DMSO). A negative control with 1% DMSO only was made in parallel. CNA-induced resistance to antibiotics was also investigated by a double-disk antagonism test on MHA, with PA14 and efflux-defective mutants. Disks loaded with 10 mg CNA were deposited onto the surface of seeded MHA plates at an appropriate distance from antibiotic disks (Bio-Rad). The flattening of inhibition zone in the direction of CNA disk was interpreted as CNA-induced resistance to the tested antibiotic.

RT-qPCR experiments.

Specific gene expression levels were measured by real-time quantitative PCR (RT-qPCR), as previously described (25). RNA was reverse transcribed with ImProm-II reverse transcriptase as specified by the manufacturer (Promega, Madison, WI). The amounts of specific cDNA were quantified on a Rotor Gene RG6000 instrument (Qiagen, Courtaboeuf, France) by using the QuantiFast SYBR green PCR kit (Qiagen) and primers annealing to the target genes (Table S4). For each strain, the mRNA levels of target genes were normalized to that of housekeeping gene rpsL and were expressed as a ratio to the transcript levels of strain PA14. Mean gene expression values were calculated from two independent bacterial cultures and each assayed in duplicate. As shown previously, transcript levels of mexB of ≥3-fold, mexY of ≥5-fold, and mexC and mexE of ≥20-fold those of PA14 were considered significantly increased because they were associated with a ≥2-fold higher resistance to the substrates of their respective pumps (25, 48).

Killing experiments with CNA.

Strain PA14 and its deletion mutants were rendered bioluminescent by using plasmid pUC18T-MiniTn7-P1-lux propagated in Escherichia coli strain XL1-Blue, as described by Damron et al. (49). Chromosomal integration of pUC18T-MiniTn7-P1-lux (Table S1) was obtained after quadripartite mating as it first requires the mobilization genes of plasmid pRK2013 (from E. coli HB101) for transfer and then a transposase encoded by plasmid pTNS3 (from E. coli DH5α) for its insertion into the P. aeruginosa genome. Overnight cultures of PA14-lux derivatives were diluted into fresh MHB up to an A_600nm of 0.1. The bacteria were cultured aerobically at 37°C for 2.5 h (A_600nm, 0.8) prior to the addition of CNA at a final concentration of 700 μg/ml. The bioluminescence of strains was monitored over a 5-h time course, in white 96-well assay plates (Corning, NY, USA), with a Synergy H1 microplate reader (Biotek Instruments, Winooski, USA) set at a gain value of 150, read height of 7 mm, and integration time of 1 s. The killing curves presented are based on 3 independent experiments. Previous experiments demonstrated a linear relationship between bioluminescence (log of RLU) and the number of cultivable bacteria between 4 and 9 log₁₀ CFU/ml, with the equation log₁₀ (CFU) = 1.202 × log₁₀ (RLU) + 1.7843 and with R² of 0.9897.

Metabolite extraction and detection using thin-layer chromatography.

Standard working solutions of CNA, cinnamic alcohol, and cinnamic acid were prepared by diluting aliquots of >98% stock solutions in methanol. Overnight bacterial cultures were diluted into 25 ml of fresh MHB and incubated with shaking (250 rpm) at 37°C. When the cultures reached an absorbance of A_600nm of 0.8, CNA was added to a final concentration of 512 μg/ml. Aliquots were removed at t_0h, t_1h, t_3h, and t_24h postexposure, and the growth medium was collected by centrifugation and filtration through two filters of 0.45- and 0.2-μm pore size. An organic extraction was repeated 5 times using 0.5 ml dichloromethane (for a total volume of 2.5 ml). The organic fractions were pooled and dried overnight in a chemical hood and were finally redissolved in 50 μl of methanol.

TLC.

Organic fractions and standards were diluted (1:1,000) in methanol and filtered on PHENEX PTFE syringe 0.2-μm-pore-size filters (Phenenomenex, Le Pecq, France). A fraction of 10 μl was sprayed as 8-mm bands on a TLC plate (Alugram Xtra SIL G UV254; Macherey-Nagel) using an automatic sampler (ATS4; Camag, Moirans, France) connected to visionCATS Camag TLC software v2.4. The TLC plate was developed in an automatic developing chamber (ADC 2; Camag) with a mobile phase containing cyclohexane:ethyl acetate (7:2) over a 70-mm developing distance. Spots were observed using UV light at 254 nm (CV-415.LS; Uvitech, England).