AcrB, AcrD, and MdtABC Multidrug Efflux Systems Are Involved in Enterobactin Export in Escherichia coli

Escherichia coli produces the iron-chelating compound enterobactin to enable growth under iron-limiting conditions. After biosynthesis, enterobactin is released from the cell. However, the enterobactin export system is not fully understood. Previous studies have suggested that the outer membrane channel TolC is involved in enterobactin export. There are several multidrug efflux transporters belonging to resistance-nodulation-cell division (RND) family that require interaction with TolC to function. Therefore, several RND transporters may be responsible for enterobactin export. In this study, we investigated whether RND transporters are involved in enterobactin export using deletion mutants of multidrug transporters in E. coli. Single deletions of acrB, acrD, mdtABC, acrEF, or mdtEF did not affect the ability of E. coli to excrete enterobactin, whereas deletion of tolC did affect enterobactin export. We found that multiple deletion of acrB, acrD, and mdtABC resulted in a significant decrease in enterobactin export and that plasmids carrying the acrAB, acrD, or mdtABC genes restored the decrease in enterobactin export exhibited by the ΔacrB acrD mdtABC mutant. These results indicate that AcrB, AcrD, and MdtABC are required for the secretion of enterobactin.


Introduction
Multidrug efflux transporters cause serious problems in cancer chemotherapy and in the treatment of bacterial infections. In gram-negative bacteria, transporters belonging to the resistancenodulation-cell division (RND) family are particularly effective in generating resistance because they form a tripartite complex with periplasmic proteins and an outer membrane protein channel. The RND transporters have wide substrate specificity [1]. The AcrAB-TolC system is composed of the RND transporter AcrB, membrane fusion protein AcrA, and multifunctional outer membrane channel TolC. It has been suggested that the AcrAB-TolC multidrug efflux system is capable of capturing substrates in the periplasm rather than in the membrane or the cytoplasm [2]. This claim is supported by high-resolution structures of AcrB, in which access pathways from the periplasm, but not from the cytoplasm, have been identified [3,4].
Iron is an essential element for many biological processes, such as amino acid and nucleotide synthesis, electron transport, and peroxide reduction [18]. Many bacteria excrete iron-chelating compounds called siderophores to grow under iron-limited conditions. E. coli can produce the catecholate siderophore enterobactin (also called enterochelin), which is a cyclic triester of 2, 3-dihydroxybenzoylserine (DHBS) [19,20]. Enterobactin is synthesized from chorismate in the cytoplasm [21] and exported from the cell. Extracellular iron-loaded enterobactin is taken up via the outer membrane receptor FepA and translocated to the periplasm [22,23]. Fe-enterobactin is chaperoned by FepB to the ATPase-dependent transporter FepDGC and shuttled to the cytoplasm. In the cytoplasm, enterobactin is degraded by Fes esterases to release iron [24][25][26][27].
The systems responsible for enterobactin synthesis and uptake are well characterized, as presented above. In contrast, the enterobactin export system is not fully understood. Previously, enterobactin export across the cytoplasmic membrane was shown to be dependent on the major-facilitator transporter EntS (the ybdA gene product) [28]. EntS has also been shown to be responsible for enterobactin export in Salmonella enterica serovar Typhimurium [29]. Bleuel et al. showed that the outer membrane channel TolC is involved in enterobactin export from the periplasm to the culture medium [30]; however, the RND transporters required for enterobactin export have not yet been identified. The RND transporters have wide substrate specificity and require TolC for their function. Therefore, it is possible that the RND transporters are responsible for enterobactin export. In this study, we investigated whether RND transporters are involved in enterobactin export using several deletion mutants of RND transporter genes and high-performance liquid chromatography analysis.

Bacterial strains and growth conditions
Bacterial strains and plasmids used in this study are listed in Table 1. The E. coli strains used in this study are derived from the wild-type strain MG1655 [31]. For production and detection of enterobactin, strains were grown at 37uC for 13 h with shaking in iron-restricted T medium containing the following per liter: 5.8 g NaCl, 3.7 g KCl, 0.113 g CaCl 2 , 0.1 g MgCl?6H 2 O, 1.1 g NH 4 Cl, 0.272 g KH 2 PO 4 , 0.142 g Na 2 SO 4 , and 12.2 g Tris. The pH was adjusted to 7.4 with 1 N HCl, and medium was supplemented with 33.4 ml deferrated casamino acids, 0.4% (wt/ vol) glucose, 0.005% (wt/vol) nicotinic acid, 1% (wt/vol) thiamine before inoculation, and 150 mM 2,29-dipyridyl. Deferrated casamino acids were prepared using the following methods. Twenty-five ml of 10% (wt/vol) casamino acids were mixed with 0.75 g hydroxyquinoline and 25 ml chloroform. Thereafter, the supernatant was again mixed with 25 ml chloroform and its supernatant was collected as deferrated casamino acids. Aliquots of iron-restricted T medium were inoculated with 0.1% (v/v) of overnight cultures of strains grown in Luria-Bertani (LB) broth (1% tryptone, 0.5% yeast extract, 0.5% NaCl) [28,30,32]. Ampicillin was added to the growth medium at the final concentration of 100 mg/L for plasmid maintenance.

Construction of plasmids
The acrAB and acrD genes were subcloned from pHSGacrAB and pHSGacrD into vector pTrc99A using BamHI and HindIII, and HindIII and SacI, respectively [7]. PCR with PrimeSTAR GXL DNA polymerase (TaKaRa Bio Inc., Otsu, Japan) and the primers mdtABC-EcoRI (CGCGAATTCAA-GAAATCCTTTCTTAGAGAA) and mdtABC-KpnI (CGCGGTACCTTACTCGGTTACCGTTTGTTT) was used to amplify mdtABC from the chromosomal DNA of E. coli MG1655. This process introduced the restriction enzyme recognition site present in the multicloning region of the pTrc99A vector. The DNA fragments were digested with restriction enzymes and then ligated into the multicloning region of pTrc99A.

Extraction of enterobactin
Enterobactin was prepared from supernatants separated from approximately 10 9 cells of each strain. Supernatants of 13 h cultures were acidified with 50 ml 12 N HCl per 10 ml and extracted twice with 5 ml ethyl acetate. Thereafter, the ethyl acetate phase was concentrated by evaporation (Buchi, Postfach, Switzerland). Dried residues were resuspended in 500 ml methanol and analyzed by reverse-phase (RP) high-performance liquid chromatography (HPLC).  The amount of enterobactin exported by DtolC mutant relative to that exported by the wild-type was calculated using the peak areas. The peak area corresponding to enterobactin released from the wild-type strain was defined as 100%. doi:10.1371/journal.pone.0108642.g001 manufacture of the standard, with monitoring at 250 nm. Peaks were identified using HPLC-grade enterobactin standards (EMC Microcollections GmbH, Tubingen, Germany). The amount of enterobactin exported was calculated from peak areas and normalized to that of the wild-type using HITACHI Model D-2000 Elite HPLC System Manager software.

TolC is required for enterobactin export
Bleuel et al. have shown that TolC is involved in the efflux of enterobactin across the outer membrane of E. coli [30]. We performed HPLC analysis to confirm whether a strain with a deletion of tolC gene differs from the wild-type in its ability to release enterobactin. The E. coli wild-type and DtolC strains were grown for 13 h and then enterobactin was extracted from supernatants of these cultures for RP HPLC analysis. Deletion of tolC resulted in a decrease in the export of enterobactin compared with the wild-type strain ( Figure 1A). When the peak area of enterobactin released from the wild-type strain was defined as 100%, enterobactin release from DtolC was decreased to 47% ( Figure 1B).

Effect of deletion of individual TolC-dependent drug efflux genes and entS on enterobactin release
To investigate the role of the TolC-dependent RND-type drug efflux systems on the release of enterobactin from E. coli, we performed RP HPLC analysis of the culture supernatants from cultures of E. coli MG1655 mutants containing single deletions of the acrB, acrD, mdtABC, acrEF, and mdtEF genes. We also investigated the effect of the entS deletion to confirm that it is Figure 2. RP HPLC analysis of enterobactin released from deletion mutants of the RND-type efflux system genes. The amounts of enterobactin exported by deletion mutants calculated using each peak area are shown. The peak area corresponding to enterobactin released from the wild-type strain was defined as 100%. The data corresponds to mean values from three independent replicates. The bars indicate standard deviations. Asterisks indicate statistically significant differences (p,0.01) according to two-tailed Student's t-tests. doi:10.1371/journal.pone.0108642.g002 Figure 3. Requirement of AcrB, AcrD, and MdtABC drug efflux systems for enterobactin export. Enterobactin was prepared from the supernatants of cultures of each multiple RND transporter mutant and analyzed by RP HPLC. The amount of enterobactin exported by each strain, calculated using each peak area, is shown. The amount of enterobactin released from the wild-type strain was defined as 100%. The data correspond to mean values from three independent replicates. The bars indicate standard deviations. Asterisks indicate statistically significant differences (p, 0.01) according to the two-tailed Student's t-tests. doi:10.1371/journal.pone.0108642.g003 Enterobactin Export by Multidrug Efflux Systems PLOS ONE | www.plosone.org involved in enterobactin export [28] Although the amounts of enterobactin released from DtolC and DentS were significantly lower than that from the wild-type strain, there was no difference among the DacrB, DacrD, DmdtABC, DacrEF, DmdtEF, and wildtype strains (Figure 2). These results are in agreement with a previous report [30].
AcrB, AcrD, and MdtABC drug efflux systems are required for enterobactin export Because individual deletions of the acrB, acrD, mdtABC, acrEF, or mdtEF gene did not affect enterobactin release from E. coli, we hypothesized that multiple efflux systems may be involved in ecterobactin export. To investigate this possibility, we performed analyses using multiple deletion mutants. The growth rates of the double mutants DacrB acrD, DacrB mdtABC, and DacrD mdtABC were almost the same as that of the wild-type strain. The mutants DacrB acrD mdtABC, DacrB acrD mdtABC mdtEF, and DacrB acrD mdtABC mdtEF acrEF grew slightly slower than the wildtype for the initial 3 h after inoculation; however. they grew to the same level as the wild-type after 13 h, when the samples were collected to measure enterobactin release. The double mutants DacrB acrD and DacrB mdtABC exported enterobactin to levels 66% and 69% of the level exported by wild type, respectively ( Figure 3). Furthermore, deletion of acrB, acrD, and mdtABC also significantly decreased enterobactin export, to only 40% of the level exported by wild type. In contrast, the double mutant DacrD mdtABC did not alter the ability of the wild-type to export enterobactin. Stepwise deletion of the mdtEF and acrEF genes from the DacrB acrD mdtABC mutant did not affect its ability to excrete enterobactin ( Figure 3). These data indicate that, among the five RND transporters of E. coli, only AcrB, AcrD, and MdtABC play a role in the excretion of enterobactin.

AcrB, AcrD, and MdtABC are involved in enterobactin export
To confirm that AcrB, AcrD, and MdtABC are involved in enterobactin export, we performed a complementation analysis. The acrB, acrD and mdtABC genes were each cloned into the vector pTrc99A, and the resulting plasmids were used to investigate their ability to complement the enterobactin export defective phenotype of the DacrB acrD mdtABC mutant. All three plasmids increased enterobactin excretion from the DacrB acrD mdtABC mutant. The amounts of enterobactin released from the strains complemented with these plasmids were 72%, 80%, and 81% of the level released by the wild-type strain, respectively, which is nearly two-fold greater than that of the DacrB acrD mdtABC mutant harboring an empty control vector ( Figure 4). These data indicates that AcrB AcrD, and MdtABC are involved in enterobactin export in E. coli.

Discussion
In this study, we examined the involvement of the RND transporters in enterobactin export and the results indicate that AcrB, AcrD, and MdtABC are required for the secretion of enterobactin. The iron-chelating compound enterobactin is synthesized in the cytoplasm and exported to the growth medium to acquire iron. The MF-type transporter EntS was previously shown to be involved in enterobactin transport across the cytoplasmic membrane [28]. It has been believed that transporters in addition to EntS are involved in enterobactin export. Outer membrane channel TolC was proposed as a candidate for enterobactin export because of its wide substrate specificity [39] and ability to cooperate with multiple systems, including MF-and RND-type transporters [40,41]. Another study demonstrated that Figure 4. Complementation of enterobactin release from the DacrB acrD mdtABC mutant using plasmids carrying acrB, acrD, or mdtABC genes. The amounts of enterobactin exported by deletion mutants, calculated using peak areas, are shown. The amount of enterobactin released from the wild-type strain harboring an empty vector was defined as 100%. The data corresponds to mean values from three independent replicates. The bars indicate standard deviations. Asterisks indicate statistically significant differences (p,0.01) determined using the two-tailed Student's t-tests. doi:10.1371/journal.pone.0108642.g004 Enterobactin Export by Multidrug Efflux Systems PLOS ONE | www.plosone.org deletion of tolC leads to the abolishment of enterobactin export and suggested that TolC is involved in enterobactin export from the periplasm to the growth medium [30]. In our study, deletion of tolC resulted in a decrease in enterobactin release, but it did not result in a complete loss of the ability to excrete enterobactin; the tolC deletion mutant could still export enterobactin to some extent. We believed that this resulted, in part, from differences in strain background. We used wild-type MG1655 as the background strain, but Dfur was used in a previous study [30]. Fur encodes the global regulator of iron homeostasis and deletion of fur results in constitutive production of enterobactin [42]. Because deletion of fur can affect enterobactin production and export, a tolC deletion may show a greater effect in this genetic background.
Individual deletions of acrB, acrD, and mdtABC did not affect the ability of E. coli cells to excrete enterobactin, whereas a triple deletion of these genes resulted in a significant decrease in enterobactin export. These three genes may not be unique in their ability to mediate enterobactin excretion, but AcrB, AcrD, and MdtABC coordinately play a role in enterobactin export from the periplasm to the growth medium in E. coli ( Figure 5). Considering the results that double mutants DacrB acrD and DacrB mdtABC showed a decrease in enterobactin export, whereas DacrD mdtABC did not change its ability to export enterobactin, we speculate that AcrB plays a more pivotal role than AcrD and MdtABC.
The AcrAB-TolC system is constitutively expressed, but the expression levels of acrD and mdtABC are quite low under normal conditions [43]. We speculate the acrD and mdtABC genes may be induced when bacteria require iron to survive. Recently, our study on the identification of negative regulators for acrD and mdtABC revealed that the expression levels of these genes were affected by Fur (data not shown). A recent study by Ruiz and Levy also showed that inactivation of the enterobactin biosynthetic genes affects the expression level of acrAB [44]. These results suggest that multidrug transporters contribute to bacterial iron homeostasis, in addition to their role in multidrug resistance.