J.-M. Pagès, UMR-MD1, Transporteurs Membranaires, Chimiorésistance et Drug Design, Faculté de Médecine, 27 Bd Jean Moulin, 13385 Marseille Cedex 05, France. E-mail: firstname.lastname@example.org
The effects of Thymus maroccanus essential oil (EO) on the integrity of the cell membranes and the permeability of the outer membrane (OM) and inner membrane (IM) of Escherichia coli, Enterobacter aerogenes and Salmonella enterica Typhimurium were investigated.
Methods and Results
The bacterial release of intracellular proteins, cytoplasmic β-galactosidase and periplasmic β-lactamase induced by T. maroccanus EO was compared to the membranotropic activity of polymyxin B (PB) known as an effective permeabilizer of the membrane of Gram-negative bacteria. Results showed that T. maroccanus EO increased the permeability of the OM and IM of studied bacteria and induced the release of intracellular proteins into the external medium.
The effect of T. maroccanus EO on the outer membrane was comparable to that of PB, and both T. maroccanus EO and PB induce similar levels of β-lactamase release. In addition, it also promoted the release of the cytoplasmic β-galactosidase. Moreover, the lipopolysaccharide molecules and the overexpression of efflux pumps seem to play a crucial role in the level of susceptibility of studied bacteria to the permeabilizing effect of T. maroccanus EO.
Significance and Impact of Study
These results demonstrate that T. maroccanus EO can restore antibiotic activity by targeting the two bacterial membranes and would be attractive candidates for developing new adjuvants for combating resistant Gram-negative bacteria.
A noteworthy characteristic of Gram-negative bacteria is the presence of an outer membrane surrounding the bacterial cell (Benson 1998; Nikaido 2003). This complex barrier represents an efficient permeation control (Benson 1998; Nikaido 2003), able to impair the uptake of macromolecules (such as large peptides) and hydrophobic substances (e.g. hydrophobic antibiotics) (Vaara 1992). Thus, the outer membrane (OM) confers a solid protection against molecules that can normally damage the inner membrane (IM), peptidoglycan or cytoplasmic targets and also controls the uptake of various compounds including nutrients and hydrophilic antibiotics via the proteinaceous channels located through the outer membrane (Weiss et al. 1984; Vaara 1992; Pagès et al. 2008). The membrane permeability of Gram-negative bacteria is directly associated with the functional assembly of the membrane proteins and the structure of the lipopolysaccharide (LPS). The expression of narrow porin channels slows down the penetration of hydrophilic solutes (Nikaido 1996), while LPS structure controls the rate of transmembrane diffusion of lipophilic solutes (Nikaido and Vaara 1985; Plésiat and Nikaido 1992). It is important to note that many of the modifications reported in the outer membrane permeability contribute to the drastic decrease in antibiotic susceptibility (Delcour 2009). Consequently, innate antibiotic resistance is likely to reflect the synergistic action of the outer membrane acting as a permeability barrier and of the diverse and widely distributed drug efflux pumps (Delcour 2009).
Certain chemical agents classified as membrane permeabilizers (Vaara 1992) can alter the bacterial membrane integrity by releasing LPS and other components from the OM, by pore-forming effect or by intercalating activity that depolarizes the membrane. In these cases, there is a concomitant modification of permeability barrier function (Alakomi et al. 2000). Examples of these permeabilizers include EDTA, which chelates divalent cations that stabilize molecular interactions and induced LPS release (Hancock and Wong 1984), cyclic peptides such polymyxin B (PB) and PB nonapeptide, which disrupt the structure of the bacterial cell membrane by interacting with its components, causing membrane damage (Hancock and Wong 1984; Alakomi et al. 2000), and also cationic antimicrobial peptides that form channels through the membrane or create membrane discontinuities (Bishop 2008). Permeabilizers as such need not be bactericidal or bacteriostatic to Gram-negative bacteria but act mainly by enabling other compounds to penetrate and reach their final targets (Alakomi et al. 2000).
There is a continuing search for compounds that are effective in rendering MDR Gram-negative bacteria susceptible to antibiotics to which they are initially resistant (Nikaido and Vaara 1985; Nikaido 1996; Helander et al. 1998; Alakomi et al. 2000). The antimicrobial activity of plant volatile oils and extracts has been recognized for many years (Helander et al. 1998; Matasyoh et al. 2009; Bozin et al. 2006). Because of the great number of constituents, essential oils (EO) exhibit a wide spectrum of specificity (Carson et al. 2002). As typical lipophiles, they pass through the cell wall and cytoplasmic membrane of bacteria and disrupt the structure of their different layers of polysaccharides, fatty acids and phospholipids (Bakkali et al. 2008). In addition, recent studies have demonstrated that some plant compounds can effectively inhibit the efflux pumps involved in antibiotic resistance mechanisms (Kaatz 2002; Smith et al. 2007; Lorenzi et al. 2009; Kuete et al. 2010; Fadli et al. 2011, 2012).
The aim of this study was to study the effect of Thymus maroccanus EO on the membrane of Gram-negative bacteria. We have investigated the permeabilizing properties of T. maroccanus EO on the membrane of a set of MDR Gram-negative strains and their derivatives (E. coli, Enterobacter aerogenes and Salmonella enterica Typhimurium) and compared it with the PB activity, in order to (i) elucidate the potential permeabilizing role likely played by this EO, (ii) determine the role of efflux pumps and LPS molecules in this permeabilizing effect and (iii) approach the possible mechanism of action of T. maroccanus EO on multidrug-resistant bacteria studied.
Materials and methods
Bacterial strains, media and chemicals
The Gram-negative bacteria used in this study are listed in Table 1. EAEP289 is Ent. aerogenes Kans (MDR isolate that exhibits active efflux of norfloxacin and efflux pump overproduction); EAEP294 and EAEP298 constructed from EAEP289 are respectively deleted of AcrA and TolC (Pradel and Pagès 2002). E. coli AG100 is used as control and AG100A is E. coli kanr (resistant to kanamycin), deleted of AcrAB and hypersensitive to chloramphenicol, tetracycline, ampicillin and nalidixic acid (Viveiros et al. 2005). SL696 is Salm. enterica Typhimurium; SL1069 and SL1102 are derived forms of SL696 and are respectively deleted of a small and a large part of LPS (Plésiat and Nikaido 1992). Strains were conserved at −80°C and grown on Luria–Bertani medium 24 h prior to any assay.
Table 1. Bacterial strains used
Kanr, resistance to kanamycin. Kans, susceptible to kanamycin.
Antibiotics and chemicals (ONPG, IPTG, PAβN) were obtained from Sigma (Saint Quentin Fallavier, France), and culture medium from Fisher (Illkirsh, France). Nitrocefin was obtained from Oxoid (Dardilly, France).
Plant material and EO
Thymus maroccanus EO used in this study was previously characterized (Fadli et al. 2011). Gas chromatography–mass spectroscopy (GC-MS) analysis had indicated that carvacrol was the major constituent in T. maroccanus EO (76·35%) besides other constituents exhibiting relatively low concentrations (Fadli et al. 2011).
For each test, the kinetics of bacterial growth was determined by measuring the optical density (OD) at 600 nm in the presence of T. maroccanus EO and PB at minimal inhibitory concentration (MIC) and MIC/2 in comparison with a negative control.
Permeabilizing property of T. maroccanus EO
To determine the permeabilizing properties of T. maroccanus EO, the release of intracellular proteins was studied using SDS-PAGE on 10% polyacrylamide gel. Briefly, bacteria were incubated in Mueller-Hinton broth (MHB) at 37°C with shaking until 0·25 OD unit. This suspension was divided into three portions. The first one served as control, and the others were supplemented with T. maroccanus EO at MIC and MIC/2, respectively. PB at MIC was used as control for E. coli strains.
Samples corresponding to 0·25 OD unit (A600) were harvested at 0, 30, 60, 90 and 120 min and centrifuged at 8000 g at room temperature for 10 min. The pellets were stored at −20°C, and the supernatants were placed overnight at 4°C after the addition of trichloroacetic acid (TCA) to allow the precipitation of proteins. Samples were then centrifuged at 8000 g at 4°C for 15 min, and the supernatants were discarded. Pellets were washed twice with cold acetone (90%) and were then stored at −20°C until use. Proteins were separated by 10% polyacrylamide gels in a Mini-Protean®-Tetrasystem (BIO-RAD, Mitry Maury, France) apparatus at constant voltage (180 V). AG100 treated samples were also subjected to Western blot analysis as previously described (Chollet et al. 2004). Antibodies directed against OmpA and EfTu belong to the laboratory collection and were used as previously described (Malléa et al. 1998).
Release of β-galactosidase
The release of cytoplasmic β-galactosidase from bacteria cells into the culture medium was determined using o-nitrophenyl-β-d-galactoside (ONPG) as enzymatic substrate (Witte and Lubitz 1989). Briefly, an overnight culture grown in MHB containing isopropyl-1-thio-β-d-galactopyranoside (IPTG; 1 mmol l−1) was used to inoculate MHB containing IPTG (1 mmol l−1) and incubated until 0·3 OD unit at 600 nm. T. maroccanus EO and PB at MIC and MIC/2 were separately added. Two samples of 1 ml of bacterial suspension (corresponding to A600 = 0·3) were removed at 0, 30, 60, 90 and 120 min. The first sample represents total β-galactosidase activity (internal and external activity); the second one was centrifuged at 8000 g for 10 min, and then 500 μl of the supernatant was recovered and corresponded to external activity. Chloroform (25–50 μl) and SDS 0·1% (25–50 μl) were added to intact cell suspensions and supernatants, and the resulting mix was vortexed and equilibrated for 5 min in a 37°C water bath to provide a complete solubilization. In a tube containing 0·9 ml of Z buffer (composition per 100 ml: β-mercaptoethanol, 0·27 ml; Na2HPO4 7H2O, 1·61 g; NaH2PO4 7H2O, 0·55 g; KCl, 0·075 g; MgSO4 7H2O, 0·0246 g), 0·1 ml of the total lysate was added, and the reaction was started by adding 0·1 ml of substrate, o-nitrophenyl-β-d-galactoside (ONPG; 4 mg ml−1).The hydrolysis rate was monitored at 37°C using a spectrophotometer at A420 to determine the external and total enzymatic activity. Enzymatic values are means of three independent determinations. The release activity was calculated as % of total β-galactosidase activity corresponding to the activity in supernatant/total activity.
Outer membrane permeabilization assay
Thymus maroccanus EO activity on the outer membrane was determined by using the β-lactamase assay as previously described (Chevalier et al. 2004). Briefly, an overnight culture grown in MHB containing imipenem at MIC/8 was incubated until 0·3 OD unit at 600 nm. Samples were prepared in the same way as for the β-galactosidase activity. In a tube containing 0·9 ml of phosphate buffer (100 mmol l−1, pH 7), 0·1 ml of lysate suspension (corresponding to total cells or supernatant, respectively) was added, and the reaction was started by adding 50 μl of nitrocefin solution (1 mg ml−1). The hydrolysis rate was monitored at A486 using a spectrophotometer. Enzymatic values are means of three independent determinations.
The behaviour of Gram-negative bacteria exposed to T. maroccanus EO and PB at MIC and MIC/2 was analysed (Fig. 1). The growth of bacteria decreased with the time of incubation and with the increase in T. maroccanus EO and PB concentration. At MIC/2, the effect of EO was more pronounced, compared to PB activity, whereas at MIC, the effects of both agents were similar. They reduce the average growth by about 70% after 2 h of incubation. The various MICs are presented in Table 2.
Table 2. Minimal inhibitory concentration of Thymus maroccanus essential oil and polymyxin B (PB) of tested Gram-negative bacteria
To explore the role of LPS and efflux pump systems in the permeabilizing property of T. maroccanus EO, Salm. enterica Typhimurium, Ent. aerogenes and E. coli strains were incubated with MIC and MIC/2 of T. maroccanus EO, respectively, and the release of cellular proteins was studied by SDS-PAGE.
For Salm. enterica Typhimurium strains producing various sizes of LPS molecule, the presence of intracellular proteins in the external medium was observed in the presence of T. maroccanus EO and was roughly associated with the length of LPS (Fig. 2): the release for SL1102 (deleted of a large part of LPS chain) at MIC (0·234 ml l−1) and MIC/2 (0·117 ml l−1) was more important than that observed for SL1069 strain (deleted of a small part of LPS chain). In contrast, for SL696 strain (wild-type), the release of proteins was limited even in the presence of EO at MIC (0·93 ml l−1).
For E. coli strains, the protein release in AG100 (wild-type) was observed after 30 min of incubation at MIC/2 and after 10 min at MIC (Fig. 3). In AG100A (deleted of AcrAB), the release was started at 90 min in the presence of T. maroccanus EO at MIC/2. At MIC, the release of proteins in AG100A seemed to be more important than in AG100, after 90 min of incubation. Compared to the activity of T. maroccanus EO, the PB effect was less effective in term of protein release under the conditions tested (Fig. 3). The release of intracellular proteins in Ent. aerogenes strains (Fig. 4) was observed 10 min after the addition of the EO: it was related to the concentration of the EO, and it was more important in EA294 (deleted of AcrAB) than in EA289 (parental strain).
To confirm the effect of EO on bacterial membrane and in order to characterize the mechanism of action, two marker proteins were used, OmpA, an outer membrane protein, and EfTu, the cytoplasmic elongation factor. Signals corresponding to OmpA detected in bacterial pellet and in the supernatant in the presence of EO at MIC showed no difference compared to the control (Fig. 5). The amount of OmpA released into the external medium remained quasi-similar, reflecting the presence of natural outer membrane vesicles (Deatherage et al. 2009). Gram-negative bacteria naturally produce outer membrane particles consisting of outer membrane lipids, outer membrane proteins and enclosed periplasmic proteins (Kulp and Kuehn 2010; Berlanda Scorza et al. 2012). Furthermore, no splitting up of the bacterial outer membrane induced by T. maroccanus EO was observed. A noticeable increase in EfTu release was detected in the treated sample during the incubation in comparison with the control (Fig. 5). This may suggest that the integrity of the IM was altered during the incubation with T. maroccanus EO, allowing the subsequent external release of this cytoplasmic protein.
Release of periplasmic and cytoplasmic enzymes
The β-galactosidase and β-lactamase activities of various Ent. aerogenes strains were determined in total cell lysates and in supernatants at different times in the presence of T. maroccanus EO and PB in order to compare their mode of action. PB was used as internal control as this antibacterial agent has been described to permeabilize the bacterial outer membrane (OM) of Gram-negative bacteria (Vaara 2010).
Figure 6 shows a significant amount of enzymatic activities released into the external medium, and this release was increased with the incubation time. The Ent. aerogenes strains presented a noticeable susceptibility to the permeabilizing effect of T. maroccanus EO. A significant release of β-lactamase and β-galactosidase was observed 10 min after the addition of the EO. For the mid-time incubation (60 min) with T. maroccanus EO, the percentage of released β-lactamase was about 70% for EAEP289, 60% for EAEP294 and 70% for EAEP298. A similar profile of β-lactamase release was also observed when the incubation was carried out with PB and the maximum of periplasmic enzyme detected in the external medium was about 95%, 70% and 85%, respectively, for EAEP289, EAEP294 and EAEP298 after a long-time incubation (Fig. 6).
The results were different for the β-galactosidase profile, and it is important to note the divergence in the behaviour of the periplasmic and the cytoplasmic markers for the tested strains during the incubation with PB. In the presence of this drug, the release of β-galactosidase was clearly delayed compared to the amounts obtained following the incubation of whatever the strains with EO (Fig. 6). In addition, for long times of incubation, the level of cytoplasmic enzyme detected in the supernatant corresponded only to a fraction of the activity measured with T. maroccanus EO, indicating that the cytoplasmic enzyme release was lower for PB compared to the EO. Moreover, at 120-min incubation with PB, about 1/4 to 1/3 of the total β-galactosidase activity was detected in the medium compared to the ratio observed for β-lactamase. In contrast, the curves obtained with T. maroccanus EO were roughly similar for β-lactamase and β-galactosidase activities measured in the medium. These results suggest a different mode of action between PB and T. maroccanus EO compounds on bacterial cells. These data clearly demonstrated that during the bacterial exposure to T. maroccanus EO, a noticeable fraction of periplasmic and cytoplasmic proteins was released. This characterization of periplasmic and cytoplasmic protein release explains the observation of cellular material detected at A260 in the medium after the incubation of three Gram-negative bacteria with T. maroccanus EO (Fadli et al. 2012).
Many of natural compounds exhibit a significant antibacterial activity and, consequently, permeability measurements are required to define the mode of action of the antimicrobial agents (Bolla et al. 2011). In bacterial cells, these alterations in envelope permeability can be monitored by checking the release of intracellular proteins or by following the activity of normally intracellular enzymes, for example β-lactamase or β-galactosidase, in the medium (Hancock and Wong 1984). A recent paper described that during incubation of various Gram-negative bacteria with T. maroccanus and Thymus Broussonetii, a release of cellular material absorbing at 260 nm is observed (Fadli et al. 2012). The authors hypothesize that EO can induce a membrane disruption inducing the release of bacterial compounds. However, no investigations have been carried out to demonstrate the effect of T. maroccanus EO on the outer or the inner membrane, and no identification of released protein has been made in this preliminary report (Fadli et al. 2012) to support the proposed hypothesis. To determine the membranotropic effect of T. maroccanus EO, a set of multiresistant Gram-negative strains and their derivatives were studied. First, we examined the effect of T. maroccanus EO on the release of intracellular proteins using SDS-PAGE analyses. T. maroccanus EO caused a significant release of intracellular protein in strains producing a truncated LPS molecule (Fig. 1). This demonstrates that LPS molecules are involved in the susceptibility to the permeabilizing effect of EO and contribute to a protection against EO active molecules. These results agree with those reporting the role of LPS as a barrier to antibiotics (Vaara 1993). Other studies have shown that the naturally occurring polycationic antibiotics of the polymyxin group bind avidly to LPS and disorganize the whole outer membrane (Hancock and Wong 1984; Nikaido and Vaara 1985; Vaara 1992).
In E. coli strains, the release of intracellular proteins was very significant and this effect was dose dependent. The release of intracellular proteins of Ent. aerogenes strains was rapid, proportional to the concentration of T. maroccanus EO. Interestingly, the effect is more important in EAEP294 (deleted of AcrAB) than in EAEP289. RND family multidrug efflux systems are prevalent in Gram-negative bacteria, and they are the most significant efflux pumps involved in resistance to multiple antimicrobial agents (Poole 2004). Their involvement in resistance to natural compounds was also demonstrated (Lorenzi et al. 2009; Kuete et al. 2010; Fadli et al. 2011). Our results show that the presence of efflux pump (AcrAB-TolC) reduced the membranotropic effect of the EO. In the same way, cationic antimicrobial polypeptides and polymyxins, membrane destabilizers, are well reported as substrates of efflux pumps, and their activity is consequently impaired in strains overproducing efflux pumps (Kraus and Peschel 2006). Consequently, the EO may act on the activity of efflux pumps causing an increase in the accumulation of active compounds or alter the membrane barrier facilitating the uptake of antibacterial agents. Owing to the great number of constituents, EOs have several cellular targets (Carson et al. 2002; Bolla et al. 2011).
Thymus maroccanus EO permeabilizes both outer and IMs of tested bacteria, but without causing detectable degradation of cellular constituents. It seems that it creates discontinuities through the membrane allowing the release of proteins such as β-lactamase, β-galactosidase or EfTu. When the membrane integrity is compromised, β-galactosidase, a cytoplasmic enzyme, could then permeate the cytoplasmic membrane and the outer ones. The ability of T. maroccanus EO to permeate the IM was compared to PB activity. For all tested strains, a conjoint increase in external enzymatic activity was observed for β-galactosidase and β-lactamase during the incubation with EO. This indicates that, in contrast to PB, T. maroccanus EO would be able to efficiently permeabilize the outer and IM. This membranotropic effect contributes to the capability of the EO for restoring the antibiotic susceptibility to various multiresistant strains of Ent. aerogenes as previously proposed (Fadli et al. 2011). Interestingly, EO and PB induced the release of periplasmic β-lactamase at a same extent and the kinetics and dose effects are roughly similar. This suggests the possible existence of a common mechanism of action on the outer membrane structure that favours the release of periplasmic components. In contrast, the data were divergent when we used cytoplasmic tools to assess their respective antibacterial activity. Hence, we may propose that the T. maroccanus EO acts by permeabilizing simultaneously the inner and the outer membrane. In contrast, PB, which is highly active against the outer membrane (Vaara 2010), showed a relatively delayed IM–permeabilizing effect.
These results clearly indicated a high membranotropic activity of T. maroccanus EO that contributed to the adjuvant efficacy previously mentioned in resistant strains (Fadli et al. 2011). Following the addition of T. maroccanus EO, the alteration in IM can induce the blocking of antibiotic efflux by perturbing the proton antiport or the functional assembly of the efflux pump component. Simultaneously to this action on IM, the permeabilization of the outer membrane can also increase the uptake of antibacterial agents.
We thank Prof. Rachid Serraj (International Centre for Agricultural Research in the Dry Areas (ICARDA), Aleppo, Syria) for taking the time to review the manuscript. A special thanks to Laila El Bouzidi, Mohamed Baz and Soraia El Baz (Université Cadi Ayyad, Marrakech, Morocco) who gave insightful help in the extraction of EO. Many thanks to Soumeya Aliouane (Aix-Marseille Université, Marseille, France) for her help during SDS-PAGE and Western blot.
This work was partially supported by the program Averroès- Erasmus Mundus, Aix-Marseille Université and IRBA.
This study did not require the ethical approval of the manuscript.