Biolog Phenotype Microarray Is a Tool for the Identification of Multidrug Resistance Efflux Pump Inducers

Multidrug resistance efflux pumps frequently present low levels of basal expression. However, antibiotic-resistant mutants that overexpress these resistance determinants are selected during infection.

ABSTRACT

Multidrug resistance efflux pumps frequently present low levels of basal expression. However, antibiotic-resistant mutants that overexpress these resistance determinants are selected during infection. In addition, increased expression of efflux pumps can be induced by environmental signals/cues, which can lead to situations of transient antibiotic resistance. In this study, we have applied a novel high-throughput methodology in order to identify inducers able to trigger the expression of the Stenotrophomonas maltophilia SmeVWX and SmeYZ efflux pumps. To that end, bioreporters in which the expression of the yellow fluorescent protein (YFP) is linked to the activity of either smeVWX or smeYZ promoters were developed and used for the screening of potential inducers of the expression of these efflux pumps using Biolog phenotype microarrays. YFP production was also measured by flow cytometry, and the levels of expression of smeV and smeY in the presence of a set of selected compounds were also determined by real-time reverse transcription-PCR (RT-PCR). The expression of smeVWX was induced by iodoacetate, clioquinol, and selenite, while boric acid, erythromycin, chloramphenicol, and lincomycin triggered smeYZ expression. The susceptibility to antibiotics that are known substrates of the efflux pumps decreased in the presence of the inducers. However, the analyzed multidrug efflux systems did not contribute to S. maltophilia resistance to the studied inducers. To sum up, the use of fluorescent bioreporters in combination with Biolog plates is a valuable tool for identifying inducers of the expression of bacterial multidrug resistance efflux pumps, and likely of other bacterial systems whose expression is regulated in response to signals/cues.

INTRODUCTION

Multidrug resistance (MDR) efflux pumps constitute a group of antibiotic resistance determinants able to reduce the activity of antimicrobial agents through their active transport outside the cell (1–3). Expression of efflux pumps is usually tightly downregulated at several levels and frequently involves the participation of global and local transcriptional regulators (4). Depending on their basal level of expression, some efflux pumps contribute to intrinsic resistance, while the contribution of others to this phenotype is low (3, 5, 6). In addition to their contribution to intrinsic resistance to antimicrobials, MDR-overexpressing mutants presenting decreased susceptibility (cross-resistance) to different clinically useful antibiotics (7) are isolated from infected patients (8) and selected in vitro upon antibiotic selective pressure (9). Five different families of MDR efflux pumps have been described so far (6). Among them, the resistance-nodulation-division (RND) MDR efflux pumps family stands as the most relevant one in Gram-negative bacteria (10, 11).

Stenotrophomonas maltophilia is an opportunistic pathogen, with environmental origin (12), which produces nosocomial infections as well as chronic infections in cystic fibrosis patients (13, 14). S. maltophilia is considered a prototype of intrinsically resistant pathogens (15). The characteristic low susceptibility of S. maltophilia to a wide range of antibiotics is due to the reduced permeability of the cell envelope as well as to the expression of several antibiotic resistance determinants encoded in its genome, including at least eight RND efflux pumps (15–17). The expression of each S. maltophilia RND efflux pump is regulated at different levels, which makes this organism a suitable model for the investigation of these systems. Among them, SmeYZ and SmeVWX are good examples of the transcriptional regulatory diversity present among S. maltophilia RND efflux pumps. The regulation of the expression of smeYZ is modulated by the two-component system (TCS) SmeRySy (18), while smeVWX expression is regulated by the LysR-type regulator SmeRv (19). In addition, whereas smeYZ is constitutively expressed at a significant level, hence contributing to the intrinsic resistance of S. maltophilia to aminoglycosides (20, 21), the level of smeVWX expression is not high enough to contribute to the intrinsic resistance of this microorganism (19) and SmeVWX-mediated antibiotic resistance is achieved only through its overexpression, as the consequence of mutations in its regulator, SmeRv (22).

As above stated, overexpression of MDR efflux pumps (and consequently antibiotic resistance) can be achieved through mutations in the regulatory elements controlling their expression. However, it is important to recall that there are occasions in which these antibiotic resistance determinants can be overexpressed without the need for a genetic change. Indeed, efflux pumps are expected to be expressed, when needed, in response to specific signals/cues (10). Consequently, there are situations and compounds capable of triggering the expression of these MDR systems, leading bacteria to display a phenotype of transient antibiotic resistance (23, 24). It is worth mentioning that besides their function as antibiotic resistance determinants, RND efflux pumps present a range of functions (11, 25–27) that include, among others, the bacterial response to host defenses (28–31), the modulation of the quorum-sensing (QS) signaling network (32–34), and the response to general stress situations (e.g., oxidative and nitrosative stress), acting as escape valves for the accumulation of toxic compounds or stress by-products (35).

RND efflux pumps are known to extrude a wide range of structurally different compounds, some of which are well characterized, especially in the case of clinically relevant antimicrobials (36). However, less is known about the efflux pump effectors. Given the above-mentioned functions of efflux pumps, most of the RND effectors already known have been discovered through the study of specific physiological processes wherein efflux pumps participate, such as the colonization of particular niches or the extrusion of a particular toxic compound (31, 37, 38). Expanding the knowledge of the set of RND inducer molecules could thus help in the characterization of MDR efflux systems, getting novel insights on their functional and ecological roles. Besides, the characterization of these effectors might also be useful for detecting possible situations of induced antibiotic resistance in vivo, which would not be detected by classical laboratory susceptibility tests.

Biolog phenotype microarrays consist of 96-well microplates developed for the determination of the bacterial response to a wide variety of chemical agents and nutrients, allowing testing nearly 2,000 phenotypes (39), which may help among several applications, including biofilm formation or the response to stressors, to decipher the phenotypes modulated by gene regulatory networks (40–48). In the current study, we have employed for the first time a different approach, based on the use of fluorescent reporters of the RND efflux pumps SmeVWX and SmeYZ, which present different levels of expression as well as different regulation patterns, to perform a wide screening of efflux pumps' inducer molecules, which could be applied as well to any other study on the regulation of gene expression. In addition, we have analyzed the effect of the induction of the expression of smeVWX and smeYZ on S. maltophilia susceptibility to the inducer compounds and to antibiotics currently used in clinical practice. Overall, our work provides new insights on the regulation of the expression and the role of the studied efflux pumps in the physiology of S. maltophilia.

RESULTS

Characterization of reporter strains PBT03 (P_EM7) and PBT10 (P_YZ).

In a previous study, we developed the reporter strain PBT02, which expresses a yellow fluorescent protein (YFP) under the control of the smeVWX promoter, in order to identify potential inducers of smeVWX expression (49). In the present work, two new YFP-based reporter strains have been developed as described in Materials and Methods: one containing the promoter of smeYZ, for measuring smeYZ expression (PBT10), and another which contains the constitutive promoter P_EM7 (50) as a control for the normalization of the results (PBT03).

In order to test the proper functioning of the reporter strains, bacterial growth and YFP levels were determined. The fluorescence levels shown by the PBT10 strain were higher than those already described for the PBT02 strain (see Fig. S1 in the supplemental material), confirming that smeYZ is constitutively expressed in the wild-type S. maltophilia strain (20), while smeVWX expression was low, as reported previously (19). The YFP level shown by the PBT03 (P_EM7) strain was also assessed, and as expected, the expression was constitutive and suitable to be used as a positive control of fluorescence as well as for normalization purposes (Fig. S1).

Screening of inducers of MDR efflux pumps expression using Biolog phenotype plates.

Biolog plates contain four different quantities (in different wells) of each tested compound. Nevertheless, information on these quantities is not available for users. The four wells for each of the tested compounds were named A, B, C, and D, with A being the well containing the highest amount of each compound. In the current work, plates PM11 to PM16 were chosen, making a total of 144 analyzed compounds (see Table S1). S. maltophilia reporter strains grew in at least one concentration for 142 of the tested compounds (Fig. S2). Those wells where S. maltophilia did not grow were discarded for further analysis.

The growth and the YFP fluorescence were recorded and analyzed in the presence of each of the tested compounds to obtain a proxy value of the transcriptional activity of the analyzed promoters. In the first step, the maximum specific rate of fluoresce production (μ) was determined as described in Materials and Methods for each promoter-compound-concentration combination (μ_VWX, μ_YZ, and μ_EM7). In the second step, the obtained values were normalized to those of the constitutive promoter and the values of normalized fluorescence production driven by each of the two analyzed promoters (μ_VWX/μ_EM7 and μ_YZ/μ_EM7) were represented for every compound concentration in a box plot (Fig. 1). From these analyses, the upper whiskers of each distribution were used to establish the thresholds defining those situations where smeVWX and smeYZ were considered to be overexpressed. These thresholds were 0.14 for μ_VWX/μ_EM7 and 0.89 for μ_YZ/μ_EM7, respectively. All the compounds and concentrations that led to the overexpression of smeVWX or smeYZ following these rules are listed in Table 1. In the case of the SmeVWX efflux pump, the value of μ_VWX/μ_EM7 for 23 compounds and 34 concentrations was found to exceed the threshold, while the value of μ_YZ/μ_EM7 was above the threshold in the case of the SmeYZ efflux pump for 22 compounds and 39 concentrations.

FIG 1.

Normalized fluorescence production. Shown is a box plot representing the values of μ_RND/μ_EM7 for all Biolog compounds and concentrations tested in plates PM11 to PM16 where there was observable growth. The upper whisker delimits the threshold above which outlier transcription activity is defined for smeVWX or smeYZ promoters and therefore efflux pump overexpression.

TABLE 1.

Normalized fluorescence production of P_VWX and P_YZ with Biolog compounds

Efflux pump, promoter analyzed	Plate	Compound	Fluorescence production of promoter normalized to that of μEM7 in wella:
			A	B	C	D
SmeVWX, μVWX	11	Chlortetracycline		0.23	0.34
		Cloxacillin				0.25
		Erythromycin				0.15
	12	d,l-Serine hydroxamate		0.17
		Penimepicycline				0.28
		5-Fluoroorotic acid				0.21
		Dodecyltrimethylammonium bromide				0.17
		Spiramycin				0.16
		Vancomycin				0.15
	13	Thallium (I) acetate	0.17
		Manganese chloride		0.19		0.65
		Potassium chromate				0.42
		Tylosin				0.30
	14	Iodoacetate		0.50	0.22
	15	Menadione		0.38	0.59
		5,7-Dichloro-8-hydroxyquinaldine			0.27	0.51
		Methyl viologen				0.18
		Nordihydroguaiaretic acid				0.18
	16	Sodium selenite	1.14	1.30	1.40	1.40
		5-Chloro-7-iodo-8-hydroxyquinoline	0.22	0.46	0.45	0.38
		Cinoxacin	0.19
		Cetylpyridinium chloride	0.16
		Protamine sulfate				0.18
SmeYz, μYZ
	11	Chloramphenicol		1.06	0.99
		Erythromycin		0.96	1.22	1.27
		Lincomycin		0.98	1.30	1.49
	12	5-Fluoroorotic acid	1.18	1.08	0.95	0.98
		d,l-Serine hydroxamate			1.02
		Penimepicycline				1.09
		Spiramycin		0.93	0.97	1.42
		Sulfadiazine				0.98
		Sulfathiazole				1.12
	13	Azlocillin				0.90
		Dequalinium chloride				0.92
		Rolitetracycline				1.03
		Tylosin			1.40	1.58
		Boric acid		0.96
	14	Chloramphenicol			1.13	1.06
		Sodium metaborate	0.99
		Fusidic acid	1.12	0.92
	15	Oleandomycin		1.11	1.04
		Puromycin				0.91
	16	Cetylpyridinium chloride	0.98
		Cinoxacin	1.46
		Protamine sulfate	1.28	1.41	1.47	1.55

^aA, B, C, and D refer to the four wells per compound, with A being the well with the highest concentration and D the well with the lowest concentration of each compound.

Confirmation of the increased expression of smeVWX and smeYZ in the presence of potential inducers.

Among those Biolog compounds that increased the YFP levels for each of the studied efflux pumps, some of those for which the induction was higher in order to validate the results were selected. As possible inducers of smeVWX, iodoacetate, clioquinol (5-chloro-7-iodo-8-hydroxyquinoline), and sodium selenite were selected; boric acid, chloramphenicol, erythromycin, and lincomycin were chosen as potential smeYZ inducers. Since the quantity of the inducing compounds present in the Biolog plate was unknown, we first determined the MICs of the potential inducers in order to establish the right concentrations to be used in the next studies. These concentrations were as follows: iodoacetate, 1 mM; clioquinol, 0.065 mM; sodium selenite, ≥200 mM; boric acid, 25 mM; erythromycin, 256 μg/ml; chloramphenicol, 4 μg/ml; and lincomycin, 4,096 μg/ml. Three concentrations below the MIC value were selected to further study the role of these compounds as inducers of the expression of the tested efflux pumps.

To carry out the experiment, the reporter strains PBT02 (P_VWX) and PBT10 (P_YZ) were incubated with their respective putative inducers for 20 h. The PBT03 (P_EM7) strain was also incubated with all the compounds and results were used as fluorescence controls. Figure 2 shows the YFP values and growth obtained for the PBT02 and PBT10 strains, using the concentration for which the highest induction was obtained and bacterial growth was less compromised in each case. These concentrations were 0.5 mM sodium selenite, 78 μM clioquinol, 0.5 mM iodoacetate, 6.25 mM boric acid, 8 μg/ml of erythromycin, 1 μg/ml of chloramphenicol, and 128 μg/ml of lincomycin. In agreement with the data obtained using the Biolog plates, iodoacetate, clioquinol, and sodium selenite increased YFP production through the induction of the smeVWX promoter in the PBT02 strain, in comparison with the fluorescence observed in control medium without inducer (Fig. 2A). Meanwhile, erythromycin, boric acid, chloramphenicol, and lincomycin increased the production of YFP through the induction of the smeYZ promoter in the PBT10 strain, compared with the strain grown in the control medium (Fig. 2B). PBT03 did not show any difference in YFP production when incubated with and without the different compounds (Fig. 2C). These results confirm that the observed fluorescence increase observed in the Biolog plates for both reporter strains PBT02 and PBT10, when grown in the presence of the above-mentioned potential effectors, is due to the induction of the promoters of the tested efflux pumps.

FIG 2.

Growth and fluorescence values produced by the reporter strains PBT02 (P_VWX), PBT10 (P_YZ), and PBT03 (P_EM7) after incubation with the selected putative inducers. (A) Expression of smeVWX during 20 h of incubation in the presence of iodoacetate (0.5 mM), clioquinol (78 μM), or sodium selenite (0.5 mM). (B) Expression of smeYZ during 20 h of incubation in the presence of erythromycin (8 μg/ml), boric acid (6.25 mM), chloramphenicol (1 μg/ml), or lincomycin (128 μg/ml). (C) P_EM7 expression during 20 h of incubation with all the compounds at the above-mentioned concentrations. Error bars show SDs from three independent replicates. FU, fluorescence units; CQ, clioquinol; IA, iodoacetate; SE, sodium selenite; BA, boric acid; CM, chloramphenicol; EM, erythromycin; LCM, lincomycin.

Single-cell analysis of the induction of the expression of MDR efflux pumps.

In order to examine the induction of smeVWX and smeYZ expression at the single-cell level, we measured by flow cytometry the YFP values of the two reporter strains PBT02 and PBT10 after 90 min of incubation with their respective inducers when cells reached exponential growth phase (optical density at 600 nm [OD₆₀₀] ≈ 0.6) (Fig. 3). Measuring induction at exponential growth phase lessens potential interferences with the fluorescence signal, growth phase, or compound degradation that could happen over time. PBT03 was also treated with the entire set of compounds as a control (Fig. 3C and D). As shown in Fig. 3A, the expression of smeVWX in the PBT02 population was higher in the presence of clioquinol, iodoacetate, or sodium selenite than in the untreated population. The same results were obtained when the PBT10 strain was incubated with boric acid, erythromycin, chloramphenicol, or lincomycin, showing that smeYZ expression is increased in the presence of these compounds (Fig. 3B). For all the compounds, a unimodal distribution of the level of YFP was seen in the population, indicating that all cells presented an increased expression of the corresponding efflux pump in the presence of the inducers. In addition, the averages of fluorescence corresponded with those obtained when induction was measured along the growing cycle (see above), showing that induction of the analyzed efflux pumps is produced at mid-exponential growth phase.

FIG 3.

Population analysis of smeVWX and smeYZ expression in the presence of the selected putative inducers. YFP production was analyzed by flow cytometry. (A) Expression profile of the smeVWX promoter in strain PBT02 treated with clioquinol (78 μM), iodoacetate (0.5 mM), or sodium selenite (0.5 mM). (B) Expression profile of the smeYZ promoter in strain PBT10 treated with erythromycin (8 μg/ml), boric acid (6.25 mM), chloramphenicol (1 μg/ml), or lincomycin (128 μg/ml). (C and D) Expression profile of the P_EM7 promoter in strain PBT03 treated with all the different compounds at the above-mentioned concentrations. Gray populations represent in all cases the basal promoters' expressions in the absence of any compound. Geometric mean (GMean) and standard deviation (StDev) were calculated for each population using Kaluza 1.5 software.

Analysis of the expression of efflux pumps in the presence of inducers at the mRNA level.

For further confirmation of the results, the mRNA levels of smeV and smeY were quantified by real-time reverse transcription-PCR (RT-PCR) in the wild-type strain S. maltophilia D457 in the presence of the different inducer compounds. As shown in Fig. 4A, smeV expression increased 764-, 4,402-, and 8,972-fold in the presence of clioquinol, iodoacetate, and sodium selenite, respectively. The same effect can be observed in Fig. 4B, where smeY expression increased 6-, 47-, 71-, and 102-fold in the presence of boric acid, chloramphenicol, erythromycin, and lincomycin, respectively. These data confirm that the compounds selected from the analysis using Biolog plates, namely, clioquinol, iodoacetate, and sodium selenite, are smeVWX inducers, and boric acid and the antibiotics chloramphenicol, erythromycin, and lincomycin are able to trigger smeYZ expression.

FIG 4.

Effect of putative inducers on the mRNA levels of smeV and smeY. Expresion of smeV and smeY was measured in the presence and in the absence of their potential inducers by real-time RT-PCR. (A) smeV expression in strain D457 after 90 min of incubation with clioquinol (78 μM), iodoacetate (0.5 mM), or sodium selenite (0.5 mM). (B) smeY expression in D457 after 90 min of incubation with boric acid (6.25 mM), chloramphenicol (1 μg/ml), erythromycin (8 μg/ml), or lincomycin (128 μg/ml). Fold changes were estimated with respect to the values obtained for the untreated D457 strain. Error bars show SDs derived from three independent experiments.

Deletion of either smeW or smeZ does not alter the susceptibility of S. maltophilia to their corresponding inducers.

Since the smeVWX and smeYZ inducers described here are antimicrobials, we wanted to elucidate whether these compounds were also substrates of their respective efflux pumps. To test this, the growth of D457 and MBS704 (ΔsmeW) in the presence of clioquinol (78 μM), iodoacetate (0.5 mM), or sodium selenite (0.5 mM) was recorded. The same experiment was performed for smeYZ inducers, growing D457 and PBT100 (ΔsmeZ) strains in the presence of boric acid (6.25 mM), chloramphenicol (1 μg/ml), erythromycin (8 μg/ml), or lincomycin (128 μg/ml). If these compounds are extruded through the efflux pumps, it is expected that both smeW- and smeZ-defective mutants should have an increased deficiency in their growth in the presence of inducers compared with the wild-type strain. As shown in Fig. 5, no relevant growth differences were observed between the wild-type and the efflux pump-defective strains in the presence of the different inducer compounds, despite the fact that all the analyzed compounds impaired bacterial growth. It is generally assumed that efflux pump inducer molecules are also extruded by the efflux system; the efflux pump hence confers resistance to such effectors if they are toxic (51–54). However, the data obtained in this work suggest that efflux pumps do not always confer resistance to their inducers, likely because in this case, these compounds are not substrates of the efflux systems they induce.

FIG 5.

Effects of smeW and smeZ on the susceptibility of S. maltophilia to the inducers of these efflux pumps. The strains D457, MBS704 (ΔsmeW), and PBT100 (ΔsmeZ) were grown in LB medium as a control (H). D457 and MBS704 were grown in the presence of clioquinol (78 μM) (A), iodoacetate (0.5 mM) (B), and selenite (0.5 mM) (C). D457 and PBT100 were grown in the presence of boric acid (6.25 mM) (D), chloramphenicol (1 μg/ml) (E), erythromycin (8 μg/ml) (F), and lincomycin (128 μg/ml) (G). Represented values correspond to the means calculated from three independent replicates.

The overexpression of MDR efflux pumps promotes transient resistance to antibiotics.

It is known that overexpression of smeVWX contributes to the acquired resistance to chloramphenicol and quinolones of S. maltophilia (19), while SmeYZ is able to extrude aminoglycosides, contributing to intrinsic resistance to them (20, 21). Since new inducer compounds have been identified for both RND efflux pumps, we wondered whether the susceptibility of S. maltophilia to known antibiotic substrates of these MDR determinants would be altered in the presence of these putative effectors.

In order to test the contribution of the SmeVWX efflux pump to transient resistance, bacterial growth was measured in the presence and in the absence of ofloxacin, and with or without the inducer sodium selenite, using both D457 and MBS704 (ΔsmeW) strains. As shown in Fig. 6A and B, sodium selenite, at the tested concentration, did not compromise the growth of either strain significantly, whereas ofloxacin impaired the growth of both D457 and MBS704. However, the combination of ofloxacin together with sodium selenite did not impede growth of D457. With the aim of assessing whether SmeYZ may contribute to transient resistance during growth in the presence of its potential effectors, a similar experiment was performed in which the D457 and PBT100 (ΔsmeZ) strains were grown in the presence or absence of the efflux pump substrate amikacin, with or without the inducer lincomycin. As shown in Fig. 6C and D, lincomycin slightly reduced the growth of both strains due to its toxic effect. At the tested concentration of amikacin, the growth was compromised for both strains; however, the presence of lincomycin diminished the inhibitory effect of amikacin. These data suggest that both MDR pumps are involved in transient resistance to antibiotics when inducers are present, since sodium selenite and lincomycin are able to induce smeVWX and smeYZ, respectively, changing the susceptibility of D457 to their antibiotic substrates.

FIG 6.

Effects of the inducer compounds sodium selenite and lincomycin in the transient resistance of S. maltophilia to antibiotics. Growth curves were performed in the presence of ofloxacin (OFX), sodium selenite (SE), and both ofloxacin and sodium selenite for strains D457 (A) and MBS704 (ΔsmeW) (B). Growth curves were also performed in the presence of amikacin (AMK), lincomycin (LCM), and both amikacin and lincomycin in strains D457 (C) and PBT100 (ΔsmeZ) (D). Growth in LB medium was used as a control. Error bars show SDs derived from three independent replicates.

DISCUSSION

The conditions that lead to the induction of the expression of MDR efflux pumps, which consequently could give rise to transient antibiotic resistance, have not been fully explored, and their detection through the use of classical susceptibility methods is usually difficult (6, 23).

In a previous screening using the PBT02 (P_VWX) strain, we determined that vitamin K₃, among 30 tested compounds, was an inducer of the SmeVWX system (49). In the current study, around 40 compounds were selected as potential candidates for being smeVWX or smeYZ inducers after screening of 144 compounds employing Biolog phenotype microarrays. Among all the potential inducer compounds derived from the Biolog data analysis, clioquinol (5-chloro-7-iodo-8-hydroxyquinoline), iodoacetate, and sodium selenite (smeVWX inducer candidates) and boric acid, chloramphenicol, erythromycin, and lincomycin (smeYZ inducer candidates) were selected for further analysis. The fluorescence data obtained through the use of Biolog plates were confirmed by testing cognate concentrations of the selected agents, as well as by flow cytometry at mid-exponential growth phase and by RT-PCR after exposure to the corresponding compounds, obtaining similar results for the expression level. Hence, the Biolog technology linked to the use of florescence reporters has allowed us to successfully identify a set of inducer molecules of the expression of RND efflux pumps.

Although the main purpose of the current work was to identify RND efflux pumps' inducers, the combination of Biolog microarrays and fluorescent reporters could also be employed for the search of inhibitors of the expression of these efflux systems, which would help to reduce the emergence and spread of antibiotic resistance. Moreover, our approach could be broadly used in the screening of effector molecules for any transcriptional regulatory system (55).

Besides contributing to antibiotic resistance, MDR efflux pumps are involved in different aspects of bacterial physiology, including bacterial interaction with hosts (human, animals, and plants) or detoxification of cellular toxic metabolites (26). MDR efflux pumps can also take part in general mechanisms of response to cellular stress that contribute to ameliorate the adverse effects caused by stressor agents or stress by-products (e.g., oxidative stress or envelope stress) (35, 56). We hypothesize that when MDR efflux systems participate in this global cellular stress response, a general common inducer mechanism of stress could be identified through the analysis of their inducer compounds, despite their structural diversity. As described here, clioquinol, iodoacetate, and sodium selenite have been identified as smeVWX inducers, in addition to the previously described vitamin K₃ (49). Although these compounds generate oxidative stress (57–60), except iodoacetate (61), we determined in our previous work that tert-butyl hydroperoxide, another oxidative stress agent, was not able to induce smeVWX expression (49). However, all of the identified inducer compounds are associated with thiol reactivity. Iodoacetate is an alkylating reagent that modifies thiol groups in proteins by S-carboxymethylation (61); selenite is known to catalyze the oxidation of thiol groups and to induce protein aggregation (62); clioquinol, as a Cu ionophore, can deliver metal ions into cells, where it exerts its activity through the interaction with thiol and amino groups (63); and, finally, vitamin K₃ (menadione) contributes to redox cycling and has alkylating properties, reacting as well with thiol groups (60). All of this evidence suggests that the smeVWX induction mechanism might be related, at least in part, to the thiol reactivity of the inducer compounds.

In the case of the SmeYZ efflux pump, boric acid, erythromycin, chloramphenicol, and lincomycin were identified as inducers. Erythromycin targets the 50S subunit of the bacterial ribosome and inhibits the nascent chain elongation (64). Chloramphenicol targets the 50S subunit of the ribosome by its binding to the peptidyl transferase center (PTC), where peptide bond formations happen (65). Both erythromycin and chloramphenicol are also known to directly inhibit the biogenesis of the ribosomal 50S subunit (66). The third identified antibiotic inducer, lincomycin, also targets the 50S subunit of the ribosome by inhibiting peptide bond formation (67). smeYZ was also found to be induced by boric acid, which is not an antibiotic but impairs the acylation of tRNAs and inhibits protein synthesis (68). All of these compounds share a mechanism of action, suggesting that SmeYZ mechanism of induction might be related to protein synthesis inhibition. Supporting this possibility, other compounds present in the Biolog plates, such as oleandomycin, spiramycin, tylosin, penimepiclyne, and fusidic acid, whose mechanism of action is also the inhibition of protein synthesis, were found to increase the YFP levels produced by the PBT10 strain (Table 1). These data reinforce the hypothesis that ribosome-stalling stress could be a signal that triggers smeYZ expression, as happens with other RND efflux pumps, such as MexXY of Pseudomonas aeruginosa (69).

It has been described that RND efflux pumps extrude several of their known inducers, such as aminoglycosides in the case of MexXY-OprM in P. aeruginosa (70), or triclosan in the case of SmeDEF in S. maltophilia (51), a protection mechanism against these toxic agents. However, in this work, we have shown that the lack of either SmeVWX or SmeYZ does not affect the growth of S. maltophilia in the presence of their toxic inducers (Fig. 5). Different circumstances might explain this apparent contradiction; one possibility is that the concentration for induction is lower than or similar to the toxic concentration of the tested compound, not allowing detection of an effect (45). Also, the bacterium can exhibit more efficient mechanisms, (e.g., other efflux pumps) able to extrude or detoxify the same compounds, in which case these mechanisms must be removed for analyzing the less proficient ones (71). In addition, the possibility that an effector is not extruded by the efflux pump it induces cannot be disregarded. In this respect, it is worth noticing that the substrates and the inducers of a given efflux pump are frequently structurally diverse (26), although they can interfere with the same cell machinery or target. Indeed, we propose two common mechanisms of stress generated by a set of RND inducers, each one inducing the expression of an efflux pump. It is important to consider that RND efflux pumps mainly extrude substrates that are located in the periplasm or in the bacterial inner membrane (72), while the regulation of these systems takes place in the cytoplasm (73). This differentiated compartmentalization may justify the possibility that RND inducers might not always be substrates of efflux pumps. This means that in some situations, MDR efflux pumps might be overexpressed despite the fact that there is no apparent advantage for bacteria. However, the presence of the inducer may imply a situation of transient resistance to other toxic compounds, which represents a benefit under some conditions. We have assessed this situation for clinically useful antibiotics that are RND efflux pumps substrates using sodium selenite and lincomycin, the strongest inducers of SmeVWX and SmeYZ efflux pumps, respectively. Both inducer molecules were able to promote transient reduced susceptibility to ofloxacin (SmeVWX substrate) or amikacin (SmeYZ substrate).

This is a fact to take into account regarding clinical situations, since some of these inducer molecules could be provided during treatment of S. maltophilia infection. For instance, sodium selenite has been recently administered during a phase I clinical trial in terminal cancer patients (74) due to its cytotoxic effect on proliferating cancer cells. Since these patients are very vulnerable to infections caused by MDR Gram-negative bacteria, such as S. maltophilia (75), the use of sodium selenite could lead to the overexpression of smeVWX and thus transient resistance to its substrates. The clinical consequences of this situation need to be further investigated.

Through the combination of Biolog phenotype microarrays and fluorescence-based reporter strains, we have developed a new high-throughput methodology to identify MDR efflux pump inducer compounds. The common mechanism of action of the detected inducer molecules has allowed us to establish possible mechanisms of induction of both smeVWX and smeYZ in S. maltophilia. smeVWX is likely induced by compounds related to thiol reactivity, while smeYZ is induced by agents that target the ribosome, suggesting a relationship between the expression of the efflux pump and the inhibition of protein synthesis. Together with already published works (76, 77), these results indicate that MDR efflux pump expression is triggered not always by a specific compound but by stress signals generated by certain molecules with similar mechanisms of action.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

All plasmids and strains derived from this study are listed in Table 2. All experiments were performed using LB medium at 37°C. The following antibiotics were added when required: streptomycin (50 μg/ml) for Escherichia coli harboring the pSEVA4413 plasmid, ampicillin (100 μg/ml) for E. coli containing the pGEM-T Easy vector and the pGEM-T-derived plasmids pPBT02 and pPBT11, and kanamycin (50 μg/ml and 500 μg/ml) for E. coli and S. maltophilia, respectively, for the selection of pSEVA237Y and the derived plasmids pPBT04, pPBT05, and pPBT08. Medium was supplemented with 0.5 mM isopropyl-β-d-thiogalactopyranoside (IPTG) and 80 μg/ml of 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside (X-Gal) for the induction and detection of β-galactosidase production.

TABLE 2.

Bacterial strains and plasmids used in this study

Strain or plasmid	Description	Source or reference
Bacterial strains
E. coli
OmniMAX	Strain used in transformation. F′ mcrA Δ(mrr-hsdRMS-mcrBC) Φ80(lacZ)ΔM15 Δ(lacZYA-argF) U169 endA1 supE44 thi-1 gyrA96 relA1 deoR tonA panD	Invitrogen, Life Technologies
CC118λpir	Donor cell in conjugation. Strain CC118 lysogenized with λpir phage (Tc) Δ(ara-leu) araD ΔlacX74 galE galK phoA20 thi-1 rpsE rpoB argE (Am) recA1	82
1047(pRK2013)	Helper cell in conjugation harboring pRK2013 (Kanr) plasmid	83
S. maltophilia
D457	Clinical strain	84
PBT02	D457 harboring the pPBT04 (PVWX) plasmid	49
PBT03	D457 harboring the pPBT05 (PEM7) plasmid	This work
PBT10	D457 harboring the pPBT08 (PYZ) plasmid	This work
MBS704	D457 ΔsmeW	49
PBT100	D457 ΔsmeZ	This work
Plasmids
pGEM-T Easy vector	Cloning vector; Ampr	Promega
pSEVA237Y	Plasmid containing YFP protein; replication origin pBBR1; Kanr	85
pSEVA4413	Plasmid containing PEM7 promoter; replication origin pRO1600/ColE1; Smr	85
pPBT02	pGEMT-derived plasmid containing smeVWX promoter region	49
pPBT11	pGEMT-derived plasmid containing the smeYZ promoter region	This work
pPBT04	pSEVA237Y-derived plasmid containing the smeVWX promoter region	49
pPBT05	pSEVA237Y-derived plasmid containing the PEM7 promoter	This work
pPBT08	pSEVA237Y-derived plasmid containing the smeYZ promoter region	This work
pEX18Tc	Gene replacement vector; sacB Tetr	81
pZAB7	pEX18Tc-derived plasmid containing the 5′ and 3′ regions of the smeZ gene	This work

Plasmid constructions.

The PBT10 strain, which harbors the reporter plasmid pPBT08 containing the smeYZ promoter region, was obtained as described previously (49) using the primers SmeYZ_F (5′-GAATTCGGCGGGGCCGTAAG-3′; EcoRI site underlined) and SmeYZ_R (5′-AAGCTTTGCTGTGCACAATG-3′; HindIII site underlined). In order to obtain the PBT03 strain, which harbors the pPBT05 plasmid with the constitutive promoter P_EM7, the pSEVA4413 plasmid was digested with PacI and HindIII (New England BioLabs). The product corresponding to the P_EM7 promoter was purified from a 1% agarose gel using a DNA purification kit (GE Healthcare) and cloned into the pSEVA237Y plasmid, previously digested with the same restriction enzymes. The resulting plasmid, pPBT05, was introduced by transformation in E. coli CC118λpir. Then pPBT05 was introduced into S. maltophilia D457 by tripartite mating as previously described (49).

Biolog phenotype microarray assay.

The 96-well plates PM11 to PM16 from the Biolog phenotypic microarray (PM) system (Biolog), which contain a variety of chemical agents (listed in Table S1), including antimicrobials (78), were used for the screening. To carry out the experiment, 100 μl of LB medium was added to each well and plates were incubated at room temperature under constant shaking for 2 h in order to dissolve the different dried compounds in the medium before inoculating the bacterial cells. Ten microliters of an overnight cell culture was added to each well to a final OD₆₀₀ of 0.01. Plates were then incubated at 37°C for 20 h in a Tecan Spark 10M plate reader (Tecan). Growth (OD₆₀₀) and fluorescence (excitation wavelength at 508 nm and emission wavelength at 540 nm) were measured every 10 min. Plates were shaken for 5 s before each measurement.

Normalization of the results.

The obtained time courses of fluorescence determinations (see above) were fitted to a first-order differential equation, analogously to the one that represents bacterial growth, F (t) = F₀ e^{μ t} [where F (t) is fluorescence across the time, F₀ is initial fluorescence, μ is maximum specific rate of fluorescence production (per hour), and t is time (in hours)]. Using this equation, μ was obtained for every promoter-compound-concentration combination, giving rise to a set of μ_YZ, μ_VWX, and μ_EM7 values. Afterwards, the normalized values of fluorescence production (μ_VWX/μ_EM7 and μ_YZ/μ_EM7) were obtained, dividing μ_YZ and μ_VWX by μ_EM7 for each compound and concentration. This step is required to avoid any effect of the toxic compound on the fluorescence signal, which may be caused by itself or by its action on the growth rate, being inversely related to the fluorescence signal (79). To set a threshold of the normalized values of fluorescence production, which could indicate that the expression is increased (overexpression), a box plot of every normalized value of fluorescence production for all the compounds and concentrations was obtained. Then the threshold to define outlier values, and therefore values which could indicate overexpression, was set according to the formula Q₃ + 1.5 × IQR, where Q₃ is the upper quartile and IQR the interquartile range of each data set. These values were 0.14 for μ_VWX/μ_EM7 and 0.89 for μ_YZ/μ_EM.

Confirmation of the induction of gene expression.

Sodium selenite, clioquinol (5-chloro-7-iodo-8-hydroxyquinoline), and iodoacetate were selected as possible inducers of the SmeVWX efflux pump, while boric acid and the antibiotics erythromycin, chloramphenicol, and lincomycin were chosen as possible inducers of the SmeYZ efflux pump. First, the MIC was determined for each compound by microdilution using 96-well plates (Nunclon Delta Surface). Ten microliters of each overnight cell culture were added to 140 μl of medium containing different concentrations of the selected compound to a final OD₆₀₀ of 0.01. Plates were incubated at 37°C for 20 h without shaking. The MIC value was defined as the lowest concentration at which bacterial growth was not observed in the presence of the tested compound.

Three concentrations of each compound were chosen for the induction assay considering the obtained MIC values, as follows: sodium selenite, 0.25, 0.5, and 1 mM; clioquinol, 78, 156, and 312 μM; iodoacetate, 0.25, 0.5, and 1 mM; boric acid, 3.125, 6.25, and 12.5 mM; erythromycin, 4, 8, and 16 μg/ml; chloramphenicol, 0.5, 1, and 2 μg/ml; and lincomycin, 64, 128, and 256 μg/ml. Ten microliters of each bacterial culture was added to 140 μl of medium to a final OD₆₀₀ of 0.01 using Corning Costar 96-well black clear-bottom plates (Corning Incorporated). Plates were incubated at 37°C for 20 h, and growth (OD₆₀₀) and fluorescence were monitored every 10 min in the Tecan Spark 10M plate reader (Tecan). Plates were shaken for 5 s every 10 min before each measurement. An excitation wavelength at 508 nm and emission wavelength at 540 nm were set for YFP detection.

Flow cytometry assays.

S. maltophilia reporter strains PBT02 (P_VWX), PBT03 (P_EM7), and PBT10 (P_YZ) were inoculated in 100-ml Erlenmeyer flasks with 20 ml of LB medium to a final OD₆₀₀ of 0.01 and incubated at 37°C with shaking. When bacterial cultures reached mid-exponential growth phase (OD₆₀₀ ≈ 0.6), induction with the different compounds was tested using the following concentrations: 0.5 mM sodium selenite, 78 μM clioquinol, 0.5 mM iodoacetate, 6.25 mM boric acid, 8 μg/ml of erythromycin, 1 μg/ml of chloramphenicol, and 128 μg/ml of lincomycin. A bacterial culture with no compound was used as a control. Cells were then incubated with shaking for 90 min at 37°C. One milliliter of each culture was centrifuged at 13,000 rpm for 1 min at room temperature. Cells were washed with 500 μl of phosphate-buffered saline (PBS) and centrifuged as indicated above. The bacterial pellets were suspended with 300 μl of 0.4% paraformaldehyde and incubated at room temperature for 10 min. Cells were then centrifuged as indicated above and suspended in 1 ml of PBS. In order to avoid false-positive signals, all the media and buffers employed were filtered through 0.22-μm-pore filters (Corning Incorporated). With the aim of measuring YFP production at the single-cell level, samples of each reporter strain containing 20,000 cells were analyzed using a Gallios flow cytometer (Beckman Coulter). Data processing was accomplished with Kaluza 1.5 software (Beckman Coulter).

RNA preparation and real-time RT-PCR.

RNA was obtained as described previously, with some modifications (49). Briefly, a 100-ml Erlenmeyer flask with 20 ml of LB medium was inoculated with an overnight culture of S. maltophilia D457 to reach a final OD₆₀₀ of 0.01. Cell cultures were incubated at 37°C in agitation until they reached mid-exponential growth phase (OD₆₀₀ ≈ 0.6). At this point, the induction assay was performed, adding the different compounds at the concentrations required: 0.5 mM sodium selenite, 78 μM clioquinol, 0.5 mM iodoacetate, 6.25 mM boric acid, 8 μg/ml of erythromycin, 1 μg/ml of chloramphenicol, and 128 μg/ml of lincomycin. A bacterial culture with no compound was used as a control. Cultures were incubated with shaking for 90 min. Ten milliliters of each culture was taken and centrifuged at 8,000 rpm for 20 min at 4°C. RNA extraction was performed as described previously (49), and cDNA was obtained using a high-capacity cDNA reverse transcription kit (Applied Biosystems). One hundred nanograms of cDNA was used for each reaction. Real-time RT-PCR was performed in an ABI Prism 7500 real-time system (Applied Biosystem) using the Power SYBR green PCR master mix (Applied Biosystem). Primers RT-SmeV.L (5′-GTCGACTTCCTCGACAACC-3′) and RT-SmeV.R (5′-TTGCCATCCTTGTCTACCAC-3′) were used to amplify smeV, primers RT-SmeY.L (5′-CATTGGTGACCGAAGGTG-3′) and RT-SmeY.R (5′-TTGATACCGGAGAACAGCAG-3′) were used to amplify smeY, and primers RT-ftsZ.L (5′-ATGGTCAACTCGGCAGTG-3′) and RT-ftsZ.R (5′-CGGTGATGAACACCATGTC-3′) were used to amplify the housekeeping gene ftsZ. Relative changes in gene expression were determined according to the threshold cycle (2^−ΔΔCT) method (80). Mean values were obtained from three independent replicates in each experiment.

Deletion of smeZ.

A partial deletion of the smeZ gene was performed in S. maltophilia D457 through homologous recombination. To that end, 545-bp (ZA) and 549-bp (ZB) fragments from the 5′ end and 3′ end of the smeZ gene, respectively, were amplified by PCR using primers ZAF (5′-GAATTCATGGCACGTTTCTTCATCGATCGCCCGGTGTTCGC-3′; EcoRI site underlined) and ZAR (5′-ATCGACAACAACAGCAGCCATGCTCGGCACCGAACAACTG-3′) and primers ZBF (5′-CAGTTGTTCGGTGCCGAGCATGGCTGCTGTTGTTGTCGAT-3′) and ZBR (5′-GAATTCTCAACGATGTTCCGTTCCATCCACGGTTCCTCCCGGC-3′; EcoRI site underlined). An overlapping PCR was performed with ZA and ZB fragments as the template using ZAF and ZBR primers, yielding a 1,000-bp fragment (ZAB). The ZAB fragment was purified from a 1% agarose gel using a DNA purification kit (GE Healthcare) and cloned into pGEM-T Easy (Promega) following the manufacturer's protocol. The sequence was confirmed by DNA sequencing. Afterwards, this plasmid was digested using EcoRI (New England BioLabs) and the resulting ZAB fragment was cloned into the suicide vector pEX18Tc (81), obtaining the pZAB7 plasmid, which was introduced by transformation into CC118λpir. Selection was performed using tetracycline (4 μg/ml). pZAB7 was introduced by tripartite mating into S. maltophilia D457 (82), and selection was performed on LB agar plates with tetracycline (12 μg/ml) and imipenem (20 μg/ml). Tet^r colonies were streaked onto 12-μg/ml tetracycline plates and 10% sucrose plates. Tet^r and Sac^s colonies were streaked onto 10% sucrose plates and incubated at 30°C overnight. From the sucrose plates, Sac^r colonies were streaked onto 12-μg/ml tetracycline plates and 10% sucrose in order to obtain double recombinants with a partial deletion of the smeZ gene. Deletion was confirmed in the PBT100 strain using primers FragZAB_L (5′-GTGCAGAACCGGATCAAG-3′) and FragZAB_R (5′-CGAACTCGACAATGAGGAT-3′) and primers InternoSmeZ_F (5′-CGGTGTCGATCCTGTTCT-3′) and InternoSmeZ_R (5′-TGGATCGAGGTCATGAAATA-3′).

Determination of transient resistance to antibiotics.

In order to determine the contribution of SmeVWX and SmeYZ to transient resistance to antibiotics, the respective inducer molecules of each efflux pump, sodium selenite and lincomycin, were chosen to measure bacterial growth in the presence of antibiotics. D457 and MBS704 (ΔsmeW) strains were grown in ofloxacin (2 μg/ml), sodium selenite (0.5 mM), or ofloxacin and sodium selenite in combination at the same concentrations. D457 and PBT100 (ΔsmeZ) strains were grown in the presence of amikacin (16 μg/ml), lincomycin (128 μg/ml), or amikacin and lincomycin in combination at the same concentrations. Ten-microliter volumes of overnight cultures of the three S. maltophilia strains were added to 140 μl of LB medium containing the different compounds to a final OD₆₀₀ of 0.01 in 96-well plates (Nunclon Delta Surface). Plates were incubated at 37°C for 20 h using a Tecan Spark 10M plate reader (Tecan), and growth (OD₆₀₀) was recorded every 10 min. Plates were shaken for 5 s before every measurement.