N-Acetyl-d-Glucosamine Induces the Expression of Multidrug Exporter Genes, mdtEF, via Catabolite Activation in Escherichia coli

Abstract

The expression of MdtEF, a multidrug exporter in Escherichia coli, is positively controlled through multiple signaling pathways, but little is known about signals that induce MdtEF expression. In this study, we investigated compounds that induce the expression of the mdtEF genes and found that out of 20 drug exporter genes in E. coli, the expression of mdtEF is greatly induced by N-acetyl-d-glucosamine (GlcNAc). The induction of mdtEF by GlcNAc is not mediated by the evgSA, ydeO, gadX, and rpoS signaling pathways that have been known to regulate mdtEF expression. On the other hand, deletion of the nagE gene, encoding the phosphotransferase (PTS) system for GlcNAc, prevented induction by GlcNAc. The induction of mdtEF by GlcNAc was also greatly inhibited by the addition of cyclic AMP (cAMP) and completely abolished upon deletion of the cAMP receptor protein gene (crp). Other PTS sugars, glucose and d-glucosamine, also induced mdtEF gene expression. These results suggest that mdtEF expression is stimulated through catabolite control.

The emergence of bacterial multidrug resistance has become an increasing problem in the treatment of infectious diseases. One of the important mechanisms underlying antibiotic resistance involves the extrusion of compounds by an efflux pump. The most intriguing mechanisms of drug extrusion are those that include a wide variety of structurally unrelated compounds as substrates for multidrug exporters. Multidrug exporters are found in a variety of bacterial species. Among them, AcrAB in Escherichia coli and MexAB in Pseudomonas aeruginosa play major roles in bacterial intrinsic tolerance against various drugs and toxic compounds (17, 26).

Previous comprehensive expression cloning studies revealed that the E. coli chromosome encodes 20 intrinsic drug exporter systems that actually contribute to drug resistance when they are overexpressed (19). Among them, MdtEF (formerly named YhiUV) confers hyperresistance against various antibiotics, antitumor reagents, pigments, and detergents, although it is hardly expressed under normal laboratory conditions (18, 19, 28). In previous studies, it was found that expression of the mdtEF genes is up-regulated by the bacterial two-component signal transduction system EvgSA (20) and the AraC/XylS acid resistance regulator YdeO (14), while their expression is repressed by a histone-like protein, H-NS (21). The multidrug resistance caused by overexpression of EvgA and YdeO is due to the induction of mdtEF gene expression (6, 7, 14, 20). On the other hand, the multidrug resistance caused by H-NS is partly dependent on mdtEF (21). We also found that mdtEF expression is highly dependent on the growth phase; that is, the expression is induced at the stationary phase (10).

Although mdtEF expression is controlled in multiple ways through these signaling pathways, little is known about the signals that induce mdtEF expression. Recently, we reported that indole greatly induces expression of the mdtEF genes. Indole signaling for mdtEF expression is independent of the EvgSA pathway but mediated by GadX, which is also an AraC/XylS regulator that cooperates with YdeO (5). These results suggest that there are at least two independent pathways for the induction of mdtEF expression, that is, the indole-driven GadX pathway and the two-component system, EvgSA.

In this study, we screened for chemical compounds that induce expression of the mdtEF genes and found that among the drug exporter genes, only the expression of mdtEF genes is greatly induced by N-acetyl-d-glucosamine (GlcNAc). This is due to “catabolite induction,” which is the third pathway for the induction of mdtEF expression and which is dependent on neither the EvgSA nor the GadX pathway.

MATERIALS AND METHODS

Bacterial strains and plasmids.

The bacterial strains and plasmids used in this study are shown in Table S1 in the supplemental material. Chromosomal DNA of Escherichia coli MG1655 (8) was used as the PCR template. KAM3 (16), a derivative of TG1 (29) that lacks acrB, and gene deletion mutants of KAM3 were used to search for compounds that induce mdtEF transcription and to verify the regulatory mechanism. The construction of gene deletion mutants was performed using the plasmid pKO3 (12). The whole sequence of the targeted gene was replaced with a 33-bp linker sequence that retained the start and stop codons in the gene deletion mutants. Gene deletion was verified by PCR using the primers listed in Table S2 in the supplemental material. The plasmids for Crp expression were constructed as follows. The crp gene was cloned into the pQE30 vector (QIAGEN) by using the primers listed in Table S2 in the supplemental material. The resulting plasmid was named pQE30crp (Table S1). E. coli cells were grown in Luria-Bertani (LB) medium and supplemented with appropriate antibiotics, when necessary, under aerobic conditions at 37°C. Construction of a ptsG mutant derived from KAM3 was performed by P1 transduction with strain IT1168 (W3110 ptsG::Tn5) (9), kindly provided by Teppei Morita and Hiroji Aiba.

Reporter gene assay.

The reporter fusion plasmids (3) listed in Table S1 in the supplemental material were used to determine the effects of several compounds on the transcription of mdtEF and the effects of GlcNAc on the other genes of the 19 multidrug exporter systems of E. coli. Each bacterial strain was grown at 37°C in LB medium containing 10 μg/ml chloramphenicol until the optical density (OD) at 600 nm reached 0.3, and then compounds were added to the LB medium, and cultures were incubated for 2 hours. β-Galactosidase activity in cell lysates was monitored by the method of Miller, using o-nitrophenyl-β-galactopyranoside (ONPG) as a substrate (15).

Transcriptional analysis of drug exporter genes of E. coli responding to GlcNAc.

KAM3 cells were grown in LB medium until the optical density at 600 nm reached 0.3, and then the cells were incubated for 2 hours with or without 10 mM GlcNAc. Purification of total RNA from a KAM3 lysate was performed using an SV total-RNA isolation system (Promega). Total RNA (1.0 μg) was used as a template and converted to cDNA. The conditions for obtaining cDNA were those recommended by PE Applied Biosystems. The specific primer pairs used are listed in Table S3 in the supplemental material. Real-time reverse transcription (RT)-PCR was performed with each specific primer pair, using SYBR green PCR master mix (PE Applied Biosystems). Normalization of the cDNA loaded for each PCR was performed using the rrsA 16S rRNA gene. The real-time PCR was performed with an ABI PRISM 7000 sequence detection system (PE Applied Biosystems).

Rhodamine 6G resistance assay.

The rhodamine 6G resistance assay was carried out by the method described previously (5). Each bacterial strain was grown at 37°C in LB medium with or without 10 mM GlcNAc. After the cells were incubated with rhodamine 6G for 30 min, survival was calculated as the number of CFU ml⁻¹ remaining after rhodamine 6G treatment divided by that before drug treatment (expressed as a percentage). Each experiment was repeated at least three times.

Purification of Crp.

The His₆-Crp fusion protein was overexpressed from the plasmid pQE30crp in the E. coli M15 strain. Cells were grown at 37°C in LB broth until the absorbance at 600 nm reached 0.6. Then, IPTG (isopropyl-β-d-thiogalactopyranoside) was added (final concentration, 1 mM), and the culture was allowed to grow for an additional 2 hours at 37°C. The His₆-Crp fusion protein was purified from E. coli crude soluble lysates using Ni affinity resin (Amersham Pharmacia Biotech).

DNase I footprinting analysis.

DNase I footprinting analysis was performed by a nonradiochemical capillary electrophoresis method, as previously described (5, 33). The 6-carboxyfluorescein (6-FAM) DNA fragments were synthesized with the PCR primers listed in Table S2 in the supplemental material. The forward primer (gadE-F-6-FAM) was labeled with 6-FAM prior to PCR. A 215-bp 6-FAM-labeled fragment (from positions −1 to −215 relative to the gadE start codon), including a gadE promoter and the region upstream of it (0.45 pmol), was mixed with purified His₆-Crp (0 to 35 pmol) in a 50-μl reaction mixture consisting of 10 mM Tris-HCl (pH 8.0), 1 mM EDTA-K, 50 mM KCl, 1 mM dithiothreitol, and 100 μM cyclic AMP (cAMP). After incubation for 10 min at 37°C, 0.6 units of DNase I (Promega) was added. After incubation for 60 s at room temperature, samples were purified for GeneScan sequencing analysis (PE Applied Biosystems). The prepared samples were electrophoresed using an ABI PRISM 310 sequencer/genetic analyzer equipped with an ABI PRISM 310 GeneScan, and fragment sizes were determined with GeneScan analysis software.

RESULTS AND DISCUSSION

Screening of compounds that induce mdtEF gene expression.

In the first screening, we searched for compounds that induce mdtEF gene expression by means of a promoter-fused reporter gene assay. The mdtEF genes are located downstream of the gadE gene, which encodes an acid regulator. Since RT-PCR analysis revealed that mdtEF were cotranscribed with gadE (see Fig. S1 in the supplemental material), the gadE promoter-fused lacZ carried on the plasmid pNNgadE was used as a reporter for mdtEF expression in the first screening. The exporter gene expression was measured as the reporter enzyme activity in cells cultured in the presence of MdtEF substrates (antibiotics, pigments, antiseptics, and antitumor reagents) and the other compounds that are not substrates of MdtEF (bile acids, oxidation-reduction reagents, quinones, and the components of peptidoglycan). GlcNAc, which was chosen from among these compounds as a component of peptidoglycan, caused the highest induction (10-fold increase), while phosphomycin and novobiocin, which are substrates for MdtEF, caused moderate induction (two- and threefold increases, respectively) (see Fig. S2 in the supplemental material). Although the expression of mdtEF was induced on cessation of cell growth (our unpublished observation), the induction was not due to growth cessation, because these compounds at the indicated concentrations hardly affected cell growth (OD₆₀₀ = 2.4). Other compounds either did not induce the mdtEF promoter activity at all or reduced it only slightly.

GlcNAc induces mdtEF gene expression without a significant effect on the expression of other drug exporter genes.

E. coli cells carry genes for 20 drug exporter systems, including mdtEF (19). Using quantitative RT-PCR analysis and a reporter gene assay, we then investigated whether GlcNAc induces drug exporter genes other than mdtEF (Table 1). The transcriptional levels of the mdtE and mdtF genes in E. coli cells cultured with 2.2 mg/ml (10 mM) GlcNAc were about 48- and 33-fold higher, respectively, than those cultured without GlcNAc. The expression of none of the other exporter genes was affected at all by GlcNAc.

TABLE 1.

Induction of promoter activity and drug exporter gene transcripts attributed to GlcNAc^a

Gene	Induction (fold changeb)
	Reporter gene assay	Quantitative PCR
acrA	0.4	0.6
acrD	0.4	0.6
acrE	0.5	0.9
bcr	0.7	0.7
cusB	0.2	0.5
emrA	0.6	0.7
emrD	0.8	0.8
emrE	0.8	1.5
emrK	0.7	0.2
fsr	0.5	0.2
macA	0.4	0.4
mdfA	0.6	0.3
mdtA	0.7	1.4
mdtE	9.7	48.0
yceE	0.7	1.9
yceL	0.6	0.3
YidY	0.5	1.1
ydgF	1.9	0.1
ydhE	1.3	0.4
YjiO	0.5	1.8

^aAveraged results of three independent experiments are shown.

^bValues indicate changes in transcript levels and promoter activities of cells cultured with 10 mM GlcNAc compared to those of the host strain cultured without GlcNAc.

MdtEF induction by GlcNAc actually confers drug resistance.

When the expression of mdtEF is induced by GlcNAc, GlcNAc-induced drug resistance is expected. Since rhodamine 6G is one of the substrates for MdtEF, we wanted to determine whether GlcNAc-treated E. coli cells actually confer rhodamine 6G resistance (19). E. coli KAM3, which is hypersusceptible to various drugs including rhodamine 6G, showed a very low survival rate (0.22%) when cells were treated with 100 mg/liter rhodamine 6G (Fig. 1). In contrast, after being cultured in 10 mM GlcNAc, E. coli KAM3 cells showed a very high survival rate (31.7%). On the other hand, the mdtEF gene deletion mutant KAM3ΔmdtEF, even when cultured with GlcNAc, showed a drastically lower survival rate (8.39%) than that of KAM3 cells cultured with rhodamine 6G (Fig. 1). The residual increase in rhodamine 6G resistance from 0.35% to 8.39% by GlcNAc in the ΔmdtEF mutant is probably due to the induction of unknown resistance factors other than drug efflux.

FIG. 1.

GlcNAc-induced rhodamine 6G resistance of E. coli KAM3 and KAM3ΔmdtEF cells. The survival rates of the cells exposed to 100 mg/ml rhodamine 6G were measured after induction of mdtEF in the absence or presence of GlcNAc as described in Materials and Methods. Solid and open bars indicate noninduced and GlcNAc-induced cells, respectively.

GlcNAc-induced mdtEF gene expression is mediated by neither the EvgSA-YdeO nor the RpoS-GadX signaling pathway.

It is known that expression of the mdtEF genes is controlled through two independent signaling pathways. One pathway is mediated by EvgSA (a two-component regulatory system)-YdeO (an AraC/XylS regulator for E. coli acid resistance) (13, 20). The other pathway is mediated by the RpoS (a stress-inducible sigma factor)-GadX (an RpoS-dependent acid inducible regulator) cascade (5, 14, 22). The latter pathway is also stimulated by indole. In order to determine which pathway mediates the induction by GlcNAc, the effect of the deletion of these genes on mdtEF induction by GlcNAc was investigated. Deletion of these genes itself did not affect mdtEF expression in the absence of GlcNAc, because these signaling systems are silent under normal laboratory conditions. As shown in Fig. 2, the deletion of neither evgSA, ydeO, gadX, nor rpoS prevented the induction of mdtEF by GlcNAc. Although the absolute promoter activity and mRNA levels differed in some of these mutants, the degree of induction by GlcNAc was not significantly changed by these mutations. This observation indicates that the GlcNAc signal is mediated by neither the EvgSA-YdeO nor the RpoS-GadX pathway.

FIG. 2.

Effects of deletion of the evgSA, ydeO, gadX, rpoS, nagE, nagC, and crp genes on GlcNAc-induced expression of the mdtEF genes. E. coli KAM3 and derivative strains were grown in LB medium with (open bars) or without (solid bars) 10 mM GlcNAc. (A) β-Galactosidase activity of the mdtEF promoter-fused reporter enzyme. (B) Amounts of mdtE mRNA. The ordinate indicates the mdtE mRNA amounts relative to the amount of rrsA 16S rRNA. The data correspond to mean values for three independent experiments. Error bars correspond to standard deviations.

mdtEF induction by GlcNAc requires the PTS system for GlcNAc.

GlcNAc is taken up into cells through the GlcNAc-specific phosphotransferase system (PTS), including NagE (23, 25, 30). In a nagE gene deletion mutant, mdtEF gene expression induced by GlcNAc was largely prevented (Fig. 2). On the other hand, the deletion of the nagC gene, which encodes a GlcNAc-6-P-binding transcriptional regulator, did not affect the induction by GlcNAc (Fig. 2). Although the possibility that the intracellular GlcNAc-6-P regulates mdtEF gene expression through unknown GlcNAc-6-P-responding regulators is not excluded, these findings strongly suggest that the PTS system for GlcNAc confers the induction of mdtEF. The reason why mdtEF genes were still slightly induced even in the nagE mutant might be that GlcNAc can be transported by another PTS system, ManXYZ, but with low efficiency (30).

mdtEF gene expression is also induced by d-glucose and d-glucosamine, and the induction is abolished by cAMP.

GlcNAc is known to cause catabolite control via the cAMP receptor protein (CRP) (24). Thus, the effect of crp gene deletion on mdtEF expression was examined. In the crp deletion mutant, the expression level of mdtEF was significantly higher than that in the parent KAM3 strain without GlcNAc (Fig. 2). The expression level without GlcNAc in the Δcrp mutant is almost the same as that of the wild type induced with GlcNAc. In addition, the mdtEF expression level was no longer affected by GlcNAc at all in the Δcrp mutant (Fig. 2), indicating that the induction of mdtEF gene expression by GlcNAc is mediated by a CRP-dependent catabolite control system. In this case, CRP seems to act as a repressor of the mdtEF genes.

If the induction by GlcNAc is a CRP-dependent catabolite control, the effect should be prevented by cAMP. When cAMP was added to E. coli cells, the induction of mdtEF by GlcNAc was almost cancelled (Fig. 3). In addition, mdtEF gene expression was also induced by other PTS carbohydrates, i.e., d-glucose (Glc) and d-glucosamine (GlcN) (Fig. 3). Glc-treated E. coli cells showed high-level induction of the mdtEF genes, while GlcN moderately induced mdtEF gene expression. The addition of cAMP also cancelled the mdtEF induction by Glc and GlcN.

FIG. 3.

Induction of mdtEF expression by glucose (Glc), GlcNAc, and glucosamine (GlcN) and inhibition by cAMP. The E. coli KAM3 strain was grown in LB medium with 10 mM concentrations of the indicated sugars in the presence (open bars) or absence (solid bars) of 10 mM cAMP. (A) β-Galactosidase activity of the mdtEF promoter-fused reporter enzyme. (B) Amounts of mdtE mRNA. The relative amount of mRNA indicates the expression ratio of mdtE to rrsA of the 16S rRNA. The data correspond to mean values of three independent experiments. Error bars correspond to the standard deviation.

CRP-cAMP directly binds to the mdtEF promoter region.

The binding of the CRP-cAMP complex to the upstream region of the gadE-mdtEF operon was examined by means of DNase I footprinting analysis. A 215-bp DNA fragment that included the putative promoter region (13) was used as a 6-FAM-labeled probe, and CRP binding sites were determined (Fig. 4). CRP actually bound to the sequence between positions −54 and −14 from the transcriptional start (Fig. 4). This 41-bp region contains the putative promoter of gadE (13). It has been reported that the CRP recognition motif is TGTGA(N₆ or N₈)TCACA (2, 32). There is no completely consistent motif in this region, but a homologous sequence, aaTGA(N₆)TgACA, exists between positions −27 and −12 from the transcriptional start. Our results suggest that the CRP-cAMP complex represses expression of the gadE-mdtEF operon by directly binding to its promoter region and that the catabolite induction is due to the derepression caused by a decrease in the intracellular concentration of cAMP.

FIG. 4.

DNase I footprinting analysis of CRP binding to the gadE promoter region. (A) A DNA fragment (0.45 pmol), including the gadE promoter region, was labeled with 6-FAM at the 5′ end, incubated with His₆-CRP, and then subjected to a DNase I footprinting assay using an ABI PRISM 310 Gene Scan. The fluorescence intensities (ordinate) of 6-FAM-labeled DNA fragments are plotted against the sequence lengths of the fragments (abscissa). Protein-binding sites are shown in squares. (B) The DNA sequence around the CRP binding site of gadE. Protein-binding sites are enclosed by a square. The sequence homologous to the CRP binding motif is indicated.

Effects of nagE deletion and ptsG mutation on the gene expression levels of mdtE induced by Glc, GlcNAc, and GlcN.

To investigate the roles of sugar transporters in the induction of mdtEF by Glc, GlcNAc, and GlcN, we investigated the effects of nagE deletion and ptsG mutation on expression levels of mdtE. The wild-type strain, the nagE mutant, and the ptsG mutant were grown with or without sugars, and the promoter activities of mdtE (Fig. 5A) and the mRNA amounts of mdtE (Fig. 5B) were determined. In the nagE mutant, induction of mdtEF expression by GlcNAc was not observed, while the levels induced by glucose and glucosamine were almost similar to those in the wild type. In the ptsG mutant, the induction of mdtE by GlcNAc was also similar to that in the wild type, while the induction by glucose was almost nonexistent. As for glucosamine, it is mainly transported by the phosphotransferase system for mannose (PTS^Mann) (1), but it is also transported by PTS^Glc (27). The large decrease in the level of induction by glucosamine in the ptsG mutant suggests that the efficiency of the catabolite control by glucosamine via PTS^Glc may be higher than that via PTS^Mann. These results indicate that the corresponding PTS systems are required for the induction of mdtEF by sugars.

FIG. 5.

Effect of the nagE deletion and the ptsG mutation on the induction of mdtEF by sugars. (A) Effect of the nagE deletion on the gadE-mdtEF promoter activity. β-Galactosidase activities of the mdtEF promoter-fused reporter enzyme in the wild-type strain (WT) and in the ΔnagE mutant were observed in the presence of sugars as indicated. (B) The effects of the ptsG mutation on the expression level of mdtE were investigated by quantitative RT-PCR. The wild-type strain (WT) and the ptsG mutant (ptsG::Tn5) were grown with or without sugars as indicated, and the expression ratio of mdtEF to rrsA of the 16S rRNA was determined by quantitative RT-PCR analysis.

Schematic model for mdtEF expression regulation by GlcNAc.

In this study, we revealed that mdtEF expression is induced by catabolite control by PTS sugars, including GlcNAc. Figure 6 shows a schematic model of the induction of mdtEF expression by GlcNAc. Since NagE does not directly stimulate adenylate cyclase (31), it must stimulate adenylate cyclase via phosphorylation of enzyme IIA for glucose (IIA^Glc). When GlcNAc is transported by NagE, enzyme IIb for GlcNAc (IIB^Nag) is dephosphorylated. Because IIA^Nag (C-terminal region of NagE) is homologous with IIA^Glc (27), IIB^Nag can then dephosphorylate IIA^Glc (11), which in turn stops the activation of adenylate cyclase, followed by a decrease in the intracellular concentration of cAMP. The decrease in cAMP concentration causes dissociation of the cAMP-CRP complex followed by the release of CRP from the mdtEF operator region. Because, in this case, the cAMP-CRP complex acts as a negative regulator of this operon, the release causes derepression of mdtEF expression. An alternative pathway mediated by a GlcNAc-6-P-binding regulator other than NagC might be possible; however, the possibility is small because the induction by GlcNAc is strictly dependent on CRP.

FIG. 6.

Model for the regulation of mdtEF expression by GlcNAc activation. The circled plus symbol indicates activation. A cAMP-CRP complex acts as a repressor for the gadE-mdtEF operon.

Although most of the catabolite controls consist of catabolite repressions, the catabolite control of mdtEF comprises catabolite activation. In other words, the cAMP-CRP complex functions as a positive regulator for many genes controlled by catabolites, whereas, in the case of mdtEF, it is actually a negative regulator. Similar cases were reported by Gosset et al. (4), who found in their transcriptome study that some genes, including aceE, gusB, and ptsG, are activated through catabolite control.

Although the mdtEF genes are transcribed with the gadE gene, which encodes a glutamate decarboxylase-dependent acid resistance regulator, overexpression or deletion of the mdtEF genes did not affect E. coli acid resistance at all (data not shown). Because catabolite control generally has a relationship with energy metabolism (4), we suspect that MdtEF may be an exporter for waste metabolites, which have not been identified.

In this study, we revealed that the expression of the E. coli multidrug exporter MdtEF is induced through catabolite control. Our current results not only are useful for understanding the regulation mechanisms of bacterial intrinsic multidrug exporter genes but also are highly suggestive as to the physiological roles of the intrinsic drug exporter MdtEF.