INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Hydrophobic organic solvents with log P_OW values of 2 to 5 can inhibit the growth of microorganisms. When microorganisms are incubated in the presence of a large volume of solvent, the magnitude of growth inhibition is inversely correlated with the log P_OW of the solvent (2, 9), which is the common logarithm of the partition coefficient (P_OW) of a solvent, measured in a two-liquid phase system composed of n-octanol and water. It is known that this value is correlated with solvent hydrophobicity (16).

The solvent resistance of a microorganism is determined genetically. In Escherichia coli, the AcrAB-TolC efflux pump plays an intrinsic role in the solvent resistance mechanism (3, 30, 31). In E. coli exposed to a large volume of a particular solvent, the intracellular concentration of the solvent is maintained at a low level (30). This pump consists of three components, AcrA, AcrB, and TolC (8, 18, 19). The inner membrane protein AcrB belongs to the RND (resistance-nodulation-cell division) transporter family and is thought to be a proton antiporter (23, 28, 33). The periplasmic lipoprotein AcrA belongs to the membrane fusion protein family and is a highly asymmetric protein capable of spanning the periplasmic space (6, 32). The outer membrane protein TolC spans the membrane and the periplasm and functions as a channel tunnel (8, 14). Deletion of acrAB or tolC decreases the solvent resistance of E. coli.

The acrEF operon encodes the components of an efflux pump other than the AcrAB efflux pump. AcrE and AcrF are highly homologous to AcrA and AcrB, respectively (12, 18). On the chromosome of E. coli, the acrAB and acrEF loci are located at 10.5 and 73.5 min, respectively (5, 13, 26). The acrEF operon was cloned accidentally as envCD genes because AcrEF functionally complemented the crystal violet sensitivity of the envC mutant (27) of E. coli (11). High expression of the acrEF operon carried on a plasmid suppresses the hypersusceptible phenotype of envC or acrAB mutants to multiple drugs. On the other hand, acrEF null mutants do not show drug sensitivity (10, 20). Therefore, it is not likely that the acrEF operon contributes to the drug resistance of E. coli. The acrEF operon is considered to be expressed only weakly in E. coli under laboratory conditions (20).

In this report, we show that the acrEF operon in E. coli is expressed at a high level on integration of IS1 or IS2 into the chromosome at a site upstream of the operon, resulting in suppression of the solvent-hypersensitive phenotype caused by the acrB mutation. In addition, we show that AcrF also functions with AcrA and that AcrEF requires TolC to improve the solvent resistance of E. coli.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains and plasmids. The E. coli strains and plasmids used in this study are listed in Table 1. OST5500 and AX727 were obtained from the National Institute of Genetics (Mishima, Japan) as ME7783 and JE8495, respectively. A number of acrEF null mutants (JA300F, JA300AF, OST5500F, DK3402F, and DK4201F) were constructed by P1 transduction of kanamycin resistance with JZM221A1 (acrEF::Tn5) as the donor strain. Disruption of the acrEF genes was confirmed by PCR analysis using chromosomal DNA prepared from each strain as the template, with the combination of primers 2 and 4 described below. Low-copy-number vectors pMW118, pMW119, and pMW219 (GenBank accession no. AB005475, AB005476, and AB005478, respectively) were purchased from Nippon Gene Co. (Tokyo, Japan). These vectors are derived from pSC101, and the copy numbers are 1 to 5 per cell. pMW118 and pMW119 each contains an ampicillin resistance gene, and pMW219 contains a kanamycin resistance gene. The high-copy-number vector pBluescript KS(+) was purchased from Toyobo Biochemical, Inc. (Osaka, Japan).

Genetic analysis. DNA manipulations, including the preparation of E. coli genomic DNA, plasmid preparation, restriction enzyme digestion and ligation, and transformation of E. coli, were carried out by standard methods. The insert DNA containing acrAB was recovered from pLKAcrAB, which was a pMW219 derivative containing acrAB (30), digested with XhoI and BamHI, and inserted in the same sites of vector pMW119 to yield pLAcrAB (Fig. 1). pLAcrAB was digested with ClaI and BamHI. The resulting 5.5-kb ClaI-BamHI DNA fragment containing the acrA region was blunted and ligated to make pLAcrA. pLKAcrAB was digested with EcoRV and BamHI to remove acrB and ligated with a SmaI-BamHI fragment obtained from pFW5 (25) to make pLKSAcrA. The SmaI-BamHI fragment contains aad9, a spectinomycin resistance marker. Thus, the resulting plasmid pLKSAcrA contains two markers, kanamycin resistance and spectinomycin resistance. pLAcrAB was digested with Bpu1102I and HindIII. The resulting 7.4-kb Bpu1102I-HindIII DNA fragment was blunted and ligated to make pLAcrB. In these plasmids, acrA and acrB were under the control of Plac. Also, the acrAB region was amplified using the chromosomal DNA from OST5500 as the template, and the presence of the mutation was confirmed by nucleotide sequencing. The primers used were designed according to the sequence deposited in GenBank (accession no. AE000152) as follows: a forward primer, 5'-AGTTATAAGCTTCATCAATAATCGACGC-3' (bp 456 to 483 upstream of the initiation codon of acrA), and a reverse primer, 5'-TAAGGATCCACTTTTTCATACTAGCACT-3' (bp 357 to 384 downstream of the stop codon of acrB).

View larger version (36K): [in this window] [in a new window]   FIG. 1.   Physical maps and construction of plasmids containing the acrAB (A) and acrEF (B) regions. Open arrows indicate the directions and extents of the acrA, acrB, acrS, acrE, and acrF genes. Arrowheads indicate the positions and directions of extension of the primers. Dotted bars represent the DNA fragments inserted into low-copy-number vectors, and a solid bar represents the DNA fragments inserted into a high-copy-number vector. Arrows indicate the direction of the lac promoter in the vectors.

Culture conditions. The organisms were grown aerobically at 37°C in Luria broth (LB medium; pH 7.0) containing 1% Bacto Tryptone (Difco Laboratories, Detroit, Mich.), 0.5% Bacto Yeast Extract (Difco), and 1% NaCl. This medium supplemented with 0.1% glucose and 10 mM MgSO₄ (LBGMg medium) was also used. When necessary, selective antibiotics were added to the medium as follows: ampicillin, 50 µg/ml, kanamycin, 50 µg/ml, or spectinomycin, 100 µg/ml.

Assay of organic solvent resistance. An overnight culture of E. coli was diluted with 0.9% NaCl (approximately 10⁷ cells/ml). A drop of the cell suspension (5 µl) was spotted on LBGMg agar medium to form a circle with a diameter of 7 to 8 mm. The surface of the agar was overlaid with an appropriate solvent to obtain a solvent depth of 3 mm. Growth was assessed after the plates were incubated at 37°C for 12 h in a sealed vessel. Confluent growth of the cells was considered to be indicative of resistance to the solvent tested. When only a few colonies (fewer than 10) were formed in the circle, the cells were considered to be sensitive to the solvent tested.

Preparation of membrane fractions. E. coli was grown in LBGMg medium. The cells were harvested during the exponential phase of growth (optical density at 660 nm, 0.6) by centrifugation (5,000 × g for 10 min at 4°C). The cells were suspended in cold 50 mM phosphate buffer (pH 7.0) and broken by sonication in an ice-water bath. Unbroken cells were removed from the lysate by centrifugation in the same manner. The supernatant was centrifuged at 100,000 × g for 45 min at 4°C. The precipitate was washed with the phosphate buffer. The insoluble matter was used as the membrane fraction. The protein content was measured by the method of Lowry et al. (17).

SDS-polyacrylamide gel electrophoresis. Samples were dissolved in a solubilization buffer containing 1% sodium dodecyl sulfate (SDS), 2.5% (vol/vol) -mercaptoethanol, 20% (vol/vol) glycerol, and 16 mM Tris-HCl (pH 6.8) and incubated in a boiling-water bath for 5 min or at 37°C for 30 min. The samples were analyzed by electrophoresis on an SDS-polyacrylamide gel mainly by the method described by Laemmli (15).

Recovery and identification of AcrE. The membranes from DK3402 cells were incubated in 0.5% N-lauroylsarcosine (sarcosyl). Sarcosyl-soluble proteins were recovered by centrifugation (100,000 × g for 30 min at 15°C) and electrophoresed on a 0.1% SDS-10% polyacrylamide gel in Tris-HCl buffer (pH 6.8) (15). Proteins with a molecular mass of approximately 42 kDa were recovered from the gel and electrophoresed on a 0.1% SDS-10% polyacrylamide gel in 30 mM NaH₂PO₄-NaOH buffer (pH 7.2) (1). The purified 42-kDa protein thereby obtained was dissolved in 70% (vol/vol) formic acid containing 1% CNBr and incubated at room temperature for 4 days. The CNBr digest was electrophoresed on a 0.1% SDS-10% polyacrylamide gel in Tris-Tricine buffer (24). The N-terminal amino acid sequence of an appropriate peptide (ca. 13 kDa) was determined by Edman degradation using a 477A protein sequencer (Applied Biosystems Inc., Foster City, Calif.).

Determination of nucleotide sequences. DNA nucleotide sequences were determined by the dideoxy chain termination method using a DNA sequencing system (PRISM 377; Perkin-Elmer Applied Biosystems).

Materials. Plasmid pFW5 containing a spectinomycin resistance cassette was a kind gift from A. Podbielski of the Institute of Medical Microbiology. Antiserum against AcrA and E. coli JZM221A1 were kind gifts from H. Nikaido of the University of California, Berkeley, Calif. The organic solvents used were of the highest purity available and were purchased from Wako Pure Chemical Industries (Osaka, Japan). The log P_OW values of the solvents were calculated by the addition rule (16) using the log P_OW calculation software ClogP version 4 (Bio Byte Corp., Claremont, Calif.).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Identification of the acrB mutation as the determinant of the solvent hypersensitivity of OST5500. Most E. coli strains grow on LBGMg agar medium overlaid with diphenyl ether. However, the solvent resistance assay showed that E. coli OST5500 was resistant to n-nonane and sensitive to n-octane and diphenyl ether (Table 2). Thus, OST5500 is hypersensitive to solvents, compared to most E. coli strains. This solvent resistance is similar to that of acrAB or tolC mutants (30).

Isolation of mutants of strain OST5500 showing improved solvent resistance. From OST5500, we isolated suppressor mutants with high solvent resistance to study the solvent resistance mechanism that was not dependent on AcrAB. OST5500 was grown on LBGMg agar medium overlaid with cyclooctane, diphenyl ether, or n-hexane. These solvents are more toxic than n-nonane under the conditions employed. The mutants that grew on the agar were classified into three groups on the basis of their resistance to solvents, irrespective of the solvents used. The mutants in the first group were resistant to cyclooctane but sensitive to diphenyl ether and n-hexane, those in the second group were resistant to cyclooctane and diphenyl ether but sensitive to n-hexane, and those in the third group were resistant to all of these solvents as well as cyclohexane. Mutants belonging to the first, second, and third groups appeared at frequencies of 1.6 × 10⁹, 2.9 × 10⁹, and 4.7 × 10⁸, respectively. We analyzed the solvent resistance mechanism of two mutants, DK4201 (belonging to the second group) and DK3402 (belonging to the third group). The solvent resistance of DK4201 was similar to that of most E. coli strains, whereas that of DK3402 was greater (Table 2).

Occurrence of the 42-kDa membrane protein AcrE in the mutants. Through a survey examining changes in protein production in the mutants, we found that the mutants had a novel 42-kDa protein in the cell membranes. OST5500, DK4201, and DK3402 were grown in LBGMg. The membrane proteins were electrophoresed on a 0.1% SDS-10% polyacrylamide gel and stained with Coomassie brilliant blue R-250 (Fig. 2A). A 42-kDa protein was clearly evident in DK3402 and faintly detected in DK4201. This 42-kDa protein was not detected in OST5500. The protein was soluble in 0.5% sarcosyl (results not shown), suggesting that it was likely to be a protein associated with the inner membrane.

View larger version (74K): [in this window] [in a new window]   FIG. 2.   Occurrence of the AcrE protein in the mutants DK4201 and DK3402. Envelope preparations containing 40 µg of protein were electrophoresed on a 0.1% SDS-10% (wt/vol) polyacrylamide gel. Arrows indicate the positions of AcrA and AcrE. Proteins were detected by staining with Coomassie brilliant blue R-250 (A) or by immunostaining with antiserum against AcrA (B). Lanes: 1, OST5500; 2, DK4201; 3, DK3402.

Detection of AcrF in the membranes of the mutants. The acrE gene is known to form a transcriptional unit together with acrF. It seemed likely that acrF was expressed in the mutants. It has been reported that AcrF is an integral inner membrane protein with a molecular mass of 111.4 kDa (12). The membrane proteins were solubilized at 37°C for 30 min and electrophoresed on a 0.1% SDS-6% polyacrylamide gel. After silver staining, a 100-kDa protein was clearly evident in DK3402 and weakly detected in DK4201 (Fig. 3). This protein was also produced by OST5500 on introduction of pHAcrF into the cells. Therefore, it was concluded that the 100-kDa protein is AcrF.

View larger version (115K): [in this window] [in a new window]   FIG. 3.   Occurrence of the AcrF protein in the mutants DK4201 and DK3402. Envelope preparations containing 15 µg of protein were electrophoresed on a 0.1% SDS-6% (wt/vol) polyacrylamide gel. Protein was detected by silver staining. The position of AcrF is indicated by an arrowhead. Lanes: 1, OST5500; 2, DK4201; 3, DK3402; 4, OST5500(pHAcrF).

IS elements integrated upstream of acrEF in the mutants. The acrSE intergenic region of the chromoosme was analyzed by PCR for strains OST5500, DK4201, and DK3402 (Fig. 4A). DNA fragments containing the acrS gene and the intergenic region were amplified using the combination of primers 4 and 5 (Fig. 1). It was expected that the length of the product would be 1,431 bp based on the sequences deposited in GenBank. A product with the same length as that expected was amplified from the chromosomal DNA of strain OST5500. The sizes of the products obtained with DK4201 and DK3402 were 2.1 and 2.8 kb, respectively. Therefore, it was evident that an insert of 0.7 or 1.4 kb was present in the intergenic region of strains DK4201 and DK3402, respectively.

View larger version (26K): [in this window] [in a new window]   FIG. 4.   Structure of the acrSE-intergenic region of OST5500 and its mutants. (A) Length of the acrSE-intergenic region. PCR products containing the intergenic region of OST5500 (lane 1), DK4201 (lane 2), and DK3402 (lane 3), as well as fragment length standards, were electrophoresed on a 0.8% agarose gel. (B) Structure of the acrSE-intergenic region of OST5500, DK4201, and DK3402. Horizontal lines indicate the acrSE intergenic region of the OST5500 chromosomal DNA. Solid bars represent the IS elements. The white arrow shown in the bar indicates the direction of the transposase gene encoded by each IS element. Open arrows indicate the directions of the acrE and acrS genes.

Effect of acrEF inactivation on the solvent resistance of E. coli. The results described above suggested that the IS integration caused high expression of acrEF, resulting in improvement of the solvent resistance of OST5500. We constructed acrEF null mutants of several E. coli strains by P1 transduction of acrEF::Tn5. Inactivation of acrEF did not cause alteration of the solvent resistance of JA300, JA300A, or OST5500 at all. These strains showed different degrees of solvent resistance. Therefore, it was concluded that acrEF did not contribute to maintenance of the high or low solvent resistance displayed by these strains.

Improvement of the solvent resistance of OST5500 by introduction of acrEF or acrF under the control of Plac into the cells. In pLAcrEF, acrEF is under the control of the lac promoter of the vector. OST5500(pLAcrEF) was found to be resistant to n-hexane (Table 3). This result indicates that overexpression of the acrEF operon improves the solvent resistance of OST5500. This transformant was resistant to cyclohexane when isopropyl--D-thiogalactopyranoside (IPTG) was present in the medium, suggesting that the resistance of this transformant depends on the levels of acrEF expression. The solvent resistance of OST5500 was also improved by introduction of pLAcrF or pHAcrF into the cells. OST5500(pLAcrF) and OST5500(pHAcrF) were resistant to n-octane and diphenyl ether (Table 3), respectively. The improvement of solvent resistance was dependent on the copy number of the plasmids. Also, the solvent resistance of OST5500F was improved by the introduction of pLAcrEF or pHAcrF into the cells, whereas that of JA300AF was not improved by introduction only of pHAcrF; cotransformation of the cells with acrA or acrE was essential to improve the solvent resistance of JA300AF. Therefore, it is concluded that a combination of AcrA and AcrF or AcrE and AcrF was functional in these cells instead of AcrAB.

Evidence of the contribution of IS integration upstream of acrEF to improvement of the solvent resistance of OST5500. A series of plasmids were constructed containing the acrSE-intergenic region of OST5500, DK4201 or DK3402 in addition to the acrEF of OST5500 (Fig. 5). The solvent resistance of OST5500 transformants carrying each of the plasmids was examined (Table 4). OST5500(pEF55) was as sensitive to solvents as was OST5500. The solvent resistance of OST5500 was improved by transformation of the cells with pEF42 or pEF34. OST5500(pEF42) and OST5500(pEF34) were as resistant to solvents as were DK4201 and DK3402, respectively. These results show that acrEF in the OST5500 chromosome has no effect on solvent resistance. When either IS1 or IS2 was integrated upstream of acrEF, the solvent resistance of OST5500 was improved through functional complementation of acrAB.

View larger version (16K): [in this window] [in a new window]   FIG. 5.   Insertion of the acrSE intergenic region upstream of acrEF cloned in pEFrk. The acrSE intergenic region was amplified by PCR from chromosomal DNA of OST5500, DK4201, and DK3402. The amplified products were inserted upstream of acrE in pEFrk. The acrE gene was oriented in the direction opposite that of the lac promoter in the vector. Solid bars show the integrated IS elements. Open arrows indicate the directions and extents of acrE and acrF. Dotted bars show the 5' part of the acrS open reading frame.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

E. coli OST5500, a strain maintained in our laboratory, was found to be hypersensitive to solvents due to insertional inactivation of acrB by IS30. The two suppressor mutants characterized, which differed in terms of their solvent resistance, produced different levels of AcrE and AcrF. Ma et al. found that acrEF are dormant or expressed only at very low levels in E. coli (20). Kawamura-Sato et al. reported that disruption of acrEF resulted in decreased efflux of indole generated intracellularly although the null mutant was as resistant as the parent to various extracellular drugs (10). Therefore, acrEF might be expressed at a faint level which is hard to detect. However, we did not find AcrE or AcrF among the membrane proteins of OST5500 or OST5500(pEF55). Inactivation of acrEF did not reduce the solvent resistance of OST5500, JA300, or JA300A. These results indicated that the acrEF operon was not actively expressed and that it was not involved in solvent resistance in these strains, irrespective of the hydrophobicity of the solvent.

Nonetheless, integration of the IS element upstream of acrEF caused enhanced expression of acrEF and consequently suppressed the solvent hypersensitivity of OST5500. Such enhanced expression caused by IS2 integration has been reported for a plasmid containing acrEF (11). E. coli has several efflux pumps belonging to the RND transporter family. In recent years, we found that the solvent hypersensitivity of a acrAB mutant of E. coli was partially suppressed by overexpression of emrAB or yhiUV, encoding an efflux pump belonging to the RND family (30). The degree of solvent resistance restored by overexpression of acrEF was comparable to that restored by the introduction of acrAB.

The levels of the AcrEF proteins were lower in DK4201 than in DK3402. A difference in the AcrEF levels was also observed in the case of OST5500(pEF42) and OST5500(pEF34). These results indicate that the activity of the promoter generated by IS1 integration was weaker than that of the promoter generated by IS2 integration. Thus, the extent of improvement of solvent resistance of the mutants or transformants was correlated with the level of acrEF expression. This conclusion is confirmed by our finding that the solvent resistance of OST5500(pLAcrEF) cells grown in the presence of IPTG was greater than that of the same cells grown in the absence of IPTG.

We have reported that overproduction of each transcriptional activator of the mar-sox regulon genes, MarA, SoxS, or Rob, improves the solvent resistance of E. coli (4, 21, 22). The solvent resistance of DK3402 or OST5500(pEF34) is comparable to that of JA300, in which the mar-sox regulon genes are highly expressed. It has been reported that an E. coli strain in which the acrR gene, encoding a repressor for acrAB, was disrupted showed cyclohexane resistance (31). These observations suggest that E. coli acquires cyclohexane resistance by overproduction of either AcrAB or AcrEF in the presence of TolC. The improvement of solvent resistance by AcrEF was dependent on the tolC⁺ genotype.

Introduction of pLAcrEF or pHAcrF into OST5500 resulted in improvement of the solvent resistance. These findings indicate that the complex formed by the AcrAB proteins can be replaced functionally by a combination of the AcrEF or AcrAF proteins. This interchangeability was also found in JA300AF on cointroduction of a plasmid containing acrA and a plasmid containing acrF. AcrB has 77% amino acid sequence identity to AcrF (18).

It is likely that expression of the acrEF operon is not essential for E. coli cells, in which the acrAB operon is expressed, to survive in a milieu containing harmful compounds. It is interesting from a phylogenetic point of view that the acrEF structural genes are conserved in E. coli although they appear to be dormant or expressed only faintly. This operon might be conserved as part of a strategy which allows E. coli to survive under harmful conditions in the event that the acrAB genes become inactive.