n-Hexane sensitivity of Escherichia coli due to low expression of imp/ostA encoding an 87 kDa minor protein associated with the outer membrane

There has been increasing interest in culturing micro-organisms in two-phase systems consisting of an aqueous medium and a hydrophobic organic solvent. Two-phase culture systems seem to be appropriate for the bioconversion of a substrate with low solubility in water. Some hydrophobic organic solvents used in these systems influx into the cells of micro-organisms and disturb the structure of microbial membranes, and are therefore regarded as harmful to micro-organisms (Favre-Bulle et al., 1991; Aono et al., 1994a; Sikkema et al., 1994). Microbial resistance to solvents is based mainly on the efflux of cellular solvent influxing from the medium. In Escherichia coli, the solvent efflux is mediated by a solvent-extruding pump system (White et al., 1997; Aono et al., 1998; Tsukagoshi & Aono, 2000); similar systems are also used by Pseudomonas aeruginosa (Li et al., 1998) and Pseudomonas putida (Isken & de Bont, 1996; Kieboom et al., 1998; Ramos et al., 1998; Mosqueda & Ramos, 2000).

In E. coli, a number of genetic determinants regulating positive expression of the AcrABTolC efflux pump contribute to its solvent resistance (Aono et al., 1995; Nakajima et al., 1995a, b; Asako et al., 1997). However, some genetic determinants seem to give E. coli additional resistance to solvents. It has been reported that the solvent resistance of Pseudomonas spp. may be modified by certain cellular components other than efflux pumps, such as elongation of O-side chains of lipopolysaccharides or cistrans isomerization of membrane fatty acids (Sikkema et al., 1995).

Previously, we have mapped a n-hexane resistance determinant at 1·2 min on the E. coli chromosome by genetic analyses of the n-hexane-resistant strain JA300 and its n-hexane-sensitive derivative (Aono et al., 1994b). At the time of that study, the gene encoding this resistance was designated ostA, denoting organic solvent tolerance. Analysis of ostA (Aono et al., 1994c) showed that a locus of this gene was identical to that of the imp gene. Two imp mutants were isolated as maltodextrin assimilators that do not require expression of the lamB gene encoding maltoporin (Sampson et al., 1989). The gene name imp refers to the increased membrane permeability of the mutants. These mutants are hypersensitive to certain hydrophobic chemicals.

In this study, we analysed the imp/ostA gene region of the n-hexane-sensitive strain OST4251 and identified the product of the imp gene. This is the first report that identifies the imp gene product as a minor outer-membrane protein. IS2 integration around the imp promoter decreased promoter activity, demonstrating that levels of the imp gene product were low in the n-hexane-sensitive strain.

Bacterial strains and plasmids.
It has been demonstrated that the degree of toxicity of an organic solvent corresponds to its log P_ow value, which is the logarithm of the partition coefficient of the organic solvent between n-octanol and water (Isken & de Bont, 1996; Inoue & Horikoshi, 1989). Organic solvents with a low log P_ow value show higher toxicity to micro-organisms. E. coli JA300 (F^- thr leuB6 trpC1117 thi rpsL20 hsdS) is resistant to solvents with log P_ow values greater than or equal to 3·9. W2252thy is sensitive to n-hexane (log P_ow 3·9) and resistant to diphenyl ether (log P_ow 4·2). OST4251 is a JA300 derivative constructed by P1 transduction transferring the n-hexane-sensitive phenotype from W2252thy (Aono et al., 1994b). E. coli BL21(DE3) was used as the host strain to produce a truncated Imp protein. E. coli DH5α [supE44 ΔlacU169(φ80 lacZΔM15) hsdR17 recA1 endA1 gyrA96 thi-1 relA1] was used as the host to construct plasmids and to assay the activity of plasmid-borne β-galactosidase.

The plasmids used in this study are summarized in Table 1. The high-copy vector pBluescript II KS⁺ was purchased from Toyobo Biochemical. The high-copy vectors pUC118 and pUC119, as well as the overexpression vector pET21d, were obtained from Takara Shuzo (Kyoto, Japan). The low-copy vectors pMW118 and pMW119 (GenBank accession nos AB005475 and AB005476) were purchased from Nippon Gene. A number of recombinant plasmids containing the imp gene were used in this study. Each of these plasmids was named so as to specify the origin of the insert DNA (JA300 or OST4251), the length of the insert (restriction enzyme sites used to clone the gene), the copy number of the vector (High or Low) and the direction in which the gene was inserted into the lac promoter (Plac) of the vector (Normal or Reverse). Using this nomenclature, in this report we renamed pBA3911, a pMW119-derived recombinant plasmid containing the OST4251 imp/ostA gene (Aono et al., 1994c), as pHB4251LN. Three plasmids (pHB4251LR, pSS4251HN and pHB4251HR), containing imp derived from OST4251, were made using the insert of pHB4251LN. Three other plasmids (pSP300HN, pSP300LN and pSP300LR), containing the gene derived from JA300, were made, as described below. pMC1403 contains a sequence downstream of the 10th codon of lacZ, but it does not contain the promoter, the ShineDalgarno sequence or the initiation codon (Casadaban et al., 1980).

Table 1. Plasmids used in this study

Culture conditions.
E. coli was grown aerobically at 37 °C in Luria broth (LB) consisting of 1 % Bacto tryptone (Difco), 0·5 % Bacto yeast extract (Difco) and 1 % NaCl. This medium was also supplemented with 0·1 % glucose and 10 mM MgSO₄ (LBGMg medium). Ampicillin (50 µg ml^-1) was added to both of these media when necessary.

Measurement of the organic solvent resistance.
This was done for each bacterium on the basis of colony formation on LBGMg agar overlaid with an appropriate organic solvent (Aono et al., 1994b). The strength of the bactericidal effect of solvent is correlated inversely with the solvent log P_ow value under the conditions tested. In this report, the solvent resistance is expressed with the lowest log P_ow value among those of solvents to which the strain is resistant. The solvent log P_ow value was calculated by the addition rule (Leo, 1993) using the log P_ow calculation software CLOGP version 4.0 (BioByte).

Southern hybridization analysis.
The 0·6 kb EcoRIPstI fragment corresponding to codons 82276 of imp was recovered from pHB4251LN (Fig. 1) and labelled with digoxigenindeoxyuridine triphosphate by means of a DNA labelling kit (Boehringer Mannheim Yamanouchi). The labelled fragment was used as a probe for Southern hybridization analysis. The probe was detected with alkaline phosphatase-conjugated rabbit anti-digoxigenin antibody (Boehringer Mannheim Yamanouchi).

(25K): Fig. 1. Physical and genetic maps of the imp region. (a) The map is drawn on the basis of the sequences deposited in GenBank (AE000115 and AE000116). Each open box shows a region encoding the gene. An arrow in the box indicates the direction of transcription. PCR primers (14) were used to analyse the impdjlA region of the E. coli chromosome. A solid bar represents the region carried by each plasmid shown under the bar. (b) The map drawn for the OST4251 chromosome. A solid bar represents the region carried by each plasmid shown under the bar. The open bar under the region represents probe 1 used to analyse the impdjlA region and to construct plasmids p80C and p80N.

PCR analysis targeting the imp region.
The region containing imp and djlA was amplified by PCR with chromosomal DNA as the template. The primers used were designed according to the sequences deposited in GenBank (accession nos AE000115 and AE000116): primer 1, 5'-CGCGGTGCTTAATGCCTGGGCGAA-3' (complementary to codons 243250 of surA); primer 2, 5'-TCTTTTTGATGGAGCTGCACTTCG-3' (complementary to codons 8693 of imp); primer 3, 5'-GCTACGGGCTTTATCAAACATATG-3' (complementary to codons 3138 of djlA); primer 4, 5'-TTAAAAGTGTAACCGGCA-3' (complementary to codons 914 of yabP). The positions of these primers are shown in Fig. 1.

Cloning of imp from JA300.
E. coli JA300 chromosomal DNA was digested with (i) PstI and (ii) a combination of StuI and EcoRI. From each digestion, (i) a 2·4 kb PstI fragment or (ii) a 1·2 kb StuIEcoRI fragment was recovered and inserted into (i) the PstI site of pUC119 or (ii) the HincII and EcoRI sites of pBluescript II KS⁺. These plasmids were tentatively named (i) p80C and (ii) p80N (Fig. 1). Then, a 1·8 kb EcoRIBamHI fragment was recovered from p80C and inserted into the BamHI and EcoRI sites of p80N. The resulting pBluescript-derived plasmid had a 3 kb StuIPstI fragment containing the JA300 imp gene oriented in the same direction as Plac. This plasmid was designated pSP300HN. An XhoIBamHI fragment (3 kb) recovered from pSP300HN was inserted into the SalI and BamHI sites of pMW118 and pMW119. The resulting plasmids were designated pSP300LR and pSP300LN, respectively (Table 1).

Cloning of djlA from JA300.
The PCR product amplified from JA300 chromosomal DNA with primers 2 and 4 was digested with combinations of restriction enzymes, (i) PstI and SphI and (ii) EcoRV and StuI. Then, (i) a 1·6 kb PstISphI fragment and (ii) a 1·2 kb EcoRVStuI fragment were inserted into (i) the PstI and SphI sites of pUC119 and (ii) the SmaI and SphI sites of pMW118. The resulting plasmids were designated p81H and p81L, according to the copy number of the vectors.

Cloning of intergenic regions between imp and djlA.
Using chromosomal DNA preparations from JA300 and OST4251 as the templates, the intergenic regions between imp and djlA were amplified with primers 5'-ATCGGAATTCCGCCATTTTACGGCT-3' (complementary to codons 3946 of djlA) and 5'-AAAAGGATCCAGGAGAGTGGGGATA-3' (complementary to codons 412 of imp). An EcoRI or a BamHI site (underlined) was introduced into the primer sequence. The amplified products were digested with EcoRI and BamHI, and inserted into the sites of pMC1403 to construct plasmids pIGR300 and pIGR4251, respectively.

Sequencing of nucleotides.
The nucleotide sequences of samples were determined using the dideoxy-terminator cycle sequencing method and an automated DNA sequencing system (model 377; Applied Biosystems).

Antiserum against a recombinant Imp protein truncated at its N and C termini.
The 1·7 kb BglIISacI fragment of pHB4251LR was ligated into the BamHI and SacI sites of pET21d. The resulting plasmid was named pET8021. Proteins produced by E. coli BL21(DE3) carrying pET8021 were separated by SDS-PAGE. Recombinant Imp protein (Imp), with a molecular mass of 67 kDa, was recovered from the gel and used to immunize a rabbit.

Detection of Imp after subcellular fractionation of the proteins.
E. coli was grown in LB at 30 °C. Cells in the exponential phase of growth were lysed by sonication. The lysate was centrifuged (100 000 g, 45 min, 4 °C) and the supernatant was used as the soluble protein fraction. The precipitate was incubated in 50 mM phosphate buffer (pH 7·2) containing 0·5 % sodium N-lauroylsarcosinate (Sarcosyl) for 30 min at room temperature (Filip et al., 1973). The suspension was then centrifuged (100 000 g, 45 min, 10 °C), and the supernatant was used as the inner-membrane protein fraction. The precipitate was used as the outer-membrane protein fraction.

Samples were electrophoresed on an SDS-polyacrylamide gel according to the method of Laemmli (1970). The samples were also electrophoresed on a high-Bis/SDS-polyacrylamide gel [7·2 % (w/v) acrylamide and 0·38 % N',N'-methylene-bis(acrylamide)]. Proteins were detected by dye-staining, silver-staining or immunoblotting. Immunoblotting was carried out with rabbit antiserum against the Imp protein, biotinylated anti-rabbit IgG(H+L) goat antibody (Wako Pure Chemicals) and a conjugate of streptavidin and alkaline phosphatase (Bio-Rad).

Measurement of protein content.
Protein content was determined by the Lowry method.

Comparison of the amounts of Imp.
After the electrophoresis of samples on an SDS-polyacrylamide gel, Imp was stained immunologically. The intensity of the Imp band was quantified by image analysis using BIO IMAGE INTELLIGENT QUANTIFIER software, version 2.1.2a (B.I. Systems).

Assay for plasmid-borne β-galactosidase activity.
E. coli DH5α (LacZ^-) was transformed with pIGR300 or pIGR4251. The transformant was grown in LB containing ampicillin at 30 °C. Cells in the exponential phase of growth were treated with a small volume of toluene and assayed for β-galactosidase activity by the method of Miller (1972).

Presence of a HindIII site in the imp/ostA region of the n-hexane-sensitive derivative strain OST4251
Previously, we have mapped the ostA locus at 1·2 min on the E. coli chromosome and cloned a 4 kb BamHIHindIII fragment containing ostA by the gene-walking method (Aono et al., 1994b, c). However, genome sequences reported for two derivatives of E. coli K-12 show that the imp/ostA gene is present in a 13·3 kb BamHI segment with no HindIII site (GenBank accession nos AE000115, AE000116 and M12544). A schematic of the sequence of the BamHI segment is shown in Fig. 1(a).

The length of the BamHI segment was also examined by Southern analysis with probe 1, as shown in Fig. 1(a). The Southern-positive BamHI fragments found in OST4251 and W2252thy were longer than that found in JA300 (Fig. 2). HindIII cleaved the Southern-positive fragments derived from OST4251 and W2252thy but not that derived from JA300. The size of the resulting Southern-positive BamHIHindIII fragment was about 4 kb. This was the same size as the BamHIHindIII insert in pHB4251LN (formerly pBA3911), indicating that plasmid pHB4251LN contained the imp/ostA gene derived from OST4251 (Aono et al., 1994c). These results showed that a ∼2 kb DNA fragment possessing a HindIII site had been integrated into the BamHI segment of the W2252thy chromosome. This integration was mediated by P1 transduction from W2252thy in the course of the construction of OST4251.

(41K): Fig. 2. Southern blot analysis targeting the imp region. Chromosomal DNA from strains JA300 (lanes 1 and 4), OST4251 (lanes 2 and 5) and W2252thy (lanes 3 and 6) was digested with BamHI (lanes 13), or BamHI and HindIII (lanes 46). The fragments were analysed by Southern blot hybridization with probe 1, shown in Fig. 1. The approximate sizes of the bands are shown.

The region of the OST4251 chromosome in which the DNA was integrated
To examine the length and range of the DNA integration, the chromosomal imp region was analysed by PCR with the primers shown in Fig. 1(a). According to the deposited sequences (GenBank accession nos AE000115 and AE000116), it was expected that PCR with a combination of primers 1 and 3 would amplify a 3527 bp fragment containing a full-length imp gene and parts of the surA and djlA genes. PCR with a combination of primers 2 and 4 would amplify a 1684 bp fragment containing a full-length djlA gene and parts of the imp and yabP genes. The sizes of the products amplified from JA300 chromosomal DNA with different combinations of the primers were as expected (Fig. 3). However, 6·0 kb (with primers 1 and 3) and 4·2 kb (with primers 2 and 4) fragments were amplified from the chromosomal DNA of OST4251 and W2252thy. These results indicated that a fragment of approximately 2·5 kb in size was integrated between codon 86 of imp and codon 31 of djlA in the cases of OST4251 and W2252thy.

(26K): Fig. 3. PCR analysis targeting the imp region. Fragments containing the imp region were amplified by PCR using chromosomal DNA from strain JA300 (lanes 1 and 4), OST4251 (lanes 2 and 5) or W2252thy (lanes 3 and 6) as template. The primers used were 1 and 3 (lanes 1, 3 and 5), or 2 and 4 (lanes 2, 4 and 6), shown in Fig. 1. M, molecular mass marker. Sizes of the marker fragments and the amplified fragments are shown to the left and right of the image, respectively.

Identification of the integrated DNA as IS2 and IS5
The impdjlA region of JA300 was sequenced using the inserts in pSP300LR and p81L. The corresponding region of OST4251 was sequenced using the insert in pHB4251LN and the PCR product amplified from OST4251 with primers 2 and 4. The nucleotide sequence determined for the impdjlA region of JA300 was identical to those of GenBank accession nos AE000115 and AE000116. Two IS elements, IS2 and IS5, had been integrated into the corresponding region of OST4251 (Fig. 4). The sequences of the IS2 and IS5 elements were the same as those deposited in GenBank (accession nos D90838 and X13668, respectively), and were 1336 and 1287 bp long, respectively. It was concluded that the increase in the length of the impdjlA-like region was due to the integration of these two IS elements in OST4251. One HindIII site is present in the IS2 sequence. A schematic of the whole of the sequence determined for the OST4251 region is shown in Fig. 1(b). This sequence has been deposited in the DDBJ database under accession number AB013134. The nucleotide sequence of OST4251 imp was identical to that of JA300, indicating that there was no mutation in the imp structural gene of OST4251.

(18K): Fig. 4. Growth of OST4251 containing the imp plasmid in the presence of n-hexane. OST4251 transformants carrying different imp plasmids were grown at 37 °C in LBGMg medium without any solvent. n-Hexane (10 % volume of the medium) was added to the culture at the time shown by the arrow. Growth was followed by measuring the OD660 value of the culture. , JA300; , OST4251; •, OST4251(pHB4251LN); , OST4251(pHB4251LR); , OST4251(pSP300LN); , OST4251(pSP300LR). Solid lines represent growth in the presence of n-hexane and dashed lines represent growth in the absence of n-hexane.

The integration of the two IS elements into OST4251 seemed to affect the expression of the neighbouring genes. The IS5 insertion divided djlA into two parts within the 26th codon and destroyed the N-terminal transmembrane domain of the product DjlA, suggesting that djlA was not functional in OST4251. DjlA belongs to the DnaJ family of molecular chaperones (Kelley & Georgopoulos, 1997). Overexpression of djlA prompts capsule synthesis (Clarke et al., 1996). IS2 had been integrated 83 bp upstream of the initiation codon of imp. It is predicted that the imp promoter is present 141172 bp upstream of the imp initiation codon (GenBank accession no. AE000116). The IS2 integration separated the imp structural gene from the predicted promoter sequence. Thus, the expression of imp might be impaired. The surA gene is arrayed downstream of imp in the same direction. This gene encodes peptidyl prolyl cistrans isomerase and is involved in the survival of E. coli in the stationary phase of growth (Tormo et al., 1990; Missiakas et al., 1996). OST4251 did not show the SurA^- phenotype during the stationary phase (results not shown), suggesting that surA expression was not impaired in OST4251. Therefore, it was presumed that the disruption of djlA and the putative low expression of imp in OST4251 were both due to IS integration.

Assignment of the mutation responsible for the n-hexane sensitivity of OST4251
Previously, we have determined a locus for the mutation resulting in the n-hexane sensitivity at 1·2 min on the E. coli chromosome by using P1 transduction analysis (Aono et al., 1994b) and observed that OST4251 acquired n-hexane resistance by introduction of pHB4251LN (Aono et al., 1994c). However, nucleotide sequence analysis showed that djlA might participate in n-hexane resistance. Thus, it might be possible that n-hexane sensitivity was under the control of imp in a genetic background of djlA disruption. To examine this possibility, OST4251 was transformed by the introduction of plasmid p81H or p81L, which differed in their copy numbers. OST4251 carrying p81H or p81L did not grow on LBGMg agar overlaid with n-hexane. Therefore, it was concluded that the n-hexane sensitivity of OST4251 did not rely upon djlA disruption. A number of plasmids containing imp were constructed by varying the origin of imp and its 5'-flanking region, the copy number of the vector and the direction in which imp was inserted into the Plac of the vector. It was shown that the growth of E. coli OST4251 was lowered upon the introduction of each high-copy-number plasmid (pSP300HN, pSP300LN, pSS4251HN and pHB4251HR) (results not shown). In particular, OST4251(pSP300HN) grew poorly, suggesting that overexpression of the imp gene was harmful to E. coli. Thus, we examined the n-hexane resistance of OST4251 carrying each low-copy-number imp plasmid. Fig. 4 shows the growth of the transformants after the addition of n-hexane. OST4251 became resistant to n-hexane upon transformation with each plasmid, irrespective of the origin of the imp gene or the insertional direction of the imp gene with respect to Plac. These results indicated that the n-hexane sensitivity was complemented not only by introducing the JA300 imp gene but also by increasing the copy number of the OST4251 imp gene. It was presumed that the OST4251 imp gene was weakly functional, and that the n-hexane sensitivity of OST4251 was probably due to the putative low expression of imp.

Identification of the imp gene product as a minor protein associated with the outer membrane
It was presumed that OST4251 had a low level of the Imp protein, as described above. However, Imp was not readily detected when we compared the cellular proteins of JA300 and OST4251 after electrophoresis on an SDS-polyacrylamide gel and staining with Coomassie brilliant blue R-250. The presence of Imp was tested immunologically with antiserum raised for Imp. A candidate protein of 90 kDa in the Sarcosyl-insoluble membrane protein fraction was detected (Fig. 5a).

(35K): Fig. 5. Immunological detection of the Imp protein. (a) Proteins of JA300 were fractionated into soluble protein (lane 1), Sarcosyl-soluble membrane protein (lane 2) and Sarcosyl-insoluble membrane protein (lane 3). Each sample, containing 10 µg protein, was electrophoresed on an SDS-10 % (w/v) polyacrylamide gel. Imp was stained immunologically with the antiserum against Imp. Pre-stained SDS-PAGE standards (Bio-Rad) were used as molecular mass markers (M). (b) The Sarcosyl-insoluble membrane proteins (20 µg) prepared from JA300 (lane 1) and OST4251 (lane 2) were electrophoresed. Imp was stained immunologically. (c) The Sarcosyl-insoluble proteins (80 µg) prepared from JA300 (lane 1) and OST4251 (lane 2) were electrophoresed on the high-Bis/SDS-polyacrylamide gel. Proteins were stained with Coomassie brilliant blue R-250. Solid arrows highlight the Imp band; the broken arrow highlights the band of an unidentified protein interfering with Imp detection.

Sarcosyl-insoluble membrane proteins of OST4251 were analysed in the same manner. It was shown that OST4251 produced an extremely low level of the Imp candidate protein (Fig. 5b). Quantifying the intensity of the candidate bands by image analysis showed that the level of the candidate protein produced by OST4251 was 5 % of that produced by JA300.

It has been reported that the protein profiles do not change in the envelopes of imp mutants (Sampson et al., 1989). We examined why the Imp candidate protein had not been detected by electrophoresis. When the Sarcosyl-insoluble proteins were electrophoresed on a normal-Bis/SDS-10 % polyacrylamide gel (Laemmli, 1970), it was hard to specify the Imp band readily, because the Imp candidate protein co-migrated with an unidentified protein (data not shown). However, a weak band of the Imp candidate protein was detected when a large amount of the sample was electrophoresed on the high-Bis/SDS-polyacrylamide gel (Fig. 5c). This band also reacted positively with the antiserum. The candidate band was not observed in OST4251 under these conditions. These results indicated that Imp was a remarkably minor component among the outer-membrane proteins, compared with the major outer-membrane proteins such as OmpC, OmpF and OmpA (Fig. 5c). Imp might be a protein with an unidentified catalysing activity rather than a structural protein within the outer membrane. This candidate protein was purified from JA300 outer-membrane proteins by electrophoresis. Its N-terminal sequence was shown to be ADLASQCM--.

The nucleotide sequence of the imp gene shows that its product is a protein composed of 784 aa residues. The N-terminal sequence of this protein is deduced as being MKKRIPTLLATMIATALYSQQGLAADLASQCMLGV--. This sequence has basic amino acid residues (Lys-2 to Arg-4) followed by a hydrophobic stretch and a cleavage site (Ala-24 and Ala-25). This is a typical feature of the signal peptide sequence to be excreted. The molecular mass of the possible mature protein consisting of 760 aa residues is 87 068 Da. This size corresponded to that estimated for the candidate protein detected immunologically. The sequence determined for the Imp candidate protein corresponded to Ala-25 to Met-32 of that deduced for Imp. Therefore, it was concluded that the 90 kDa candidate protein was a mature form of Imp that was excreted to the outer-membrane compartment. Also, it was concluded that the expression of Imp was indeed low in OST4251.

Increase in the levels of Imp in OST4251 upon introduction of the cloned imp gene
Sarcosyl-insoluble membrane proteins of OST4251 carrying each imp plasmid were electrophoresed on a SDS-polyacrylamide gel. The Imp level was analysed by immunological detection and silver-staining (Fig. 6). It was shown that each low-copy-number imp plasmid increased Imp production. These results supported the assignment of the imp gene product. The level of Imp produced by OST4251(pHB4251LR) was higher than that produced by OST4251 and lower than that produced by OST4251(pSP300LR), suggesting that OST4251 imp promoter activity was not completely lost, and lower than that of JA300. OST4251(pHB4251LN) produced twofold more Imp than OST4251(pHB4251LR), suggesting that the OST4251 imp promoter, with one copy on the chromosome, was less active than that of OST4251(pHB4251LR).

(54K): Fig. 6. Increase in the Imp level in OST4251 upon the introduction of the imp gene. Sarcosyl-insoluble membrane proteins were prepared from OST4251 transformants carrying various plasmids containing imp. The proteins (20 µg) were electrophoresed on the high-Bis/SDS-polyacrylamide gel. (a) Imp was detected immunologically with the anti-Imp serum. (b) The proteins were silver-stained. Lanes: 1, JA300; 2, OST4251; 3, OST4251(pMW118); 4, OST4251(pHB4251LN); 5, OST4251(pSP300LR); 6, OST4251(pHB4251LR). Solid arrows highlight the Imp bands; broken arrows show bands of an unidentified protein interfering with Imp detection.

Evidence for the low imp promoter activity in OST4251
As described above, the IS2 integration probably decreased imp transcription in OST4251. We constructed plasmids pIGR300 and pIGR4251, both containing an implacZ fused gene. The expression of the fused gene by these plasmids depended on the imp promoter present in the intergenic regions derived from JA300 and OST4251, respectively. DH5α, harbouring one of the plasmids, was assayed for plasmid-borne β-galactosidase activity. No such activity was detected in DH5α(pMC1403). The activities of DH5α(pIGR300) and DH5α(pIGR4251) were 2670 and 264 units, respectively. These results suggested that the promoter activity was remarkably low in OST4251. The ratio between the promoter activities was similar to that of Imp production between OST4251 and JA300 (Figs 4 and 5). Thus, it was evident that a low level of Imp production was attributable to the low activity of the imp promoter in OST4251. Previously, we have mapped one of the n-hexane resistance determinants at 1·2 min, between pdxA and araD, on the E. coli chromosome by genetic analysis of a n-hexane-sensitive derivative, OST4251 (Aono et al., 1994b). This determinant was named ostA. The ostA gene was cloned as a gene suppressing the n-hexane-sensitive phenotype of OST4251 (Aono et al., 1994c), and was shown to be identical to imp. The results of the current study led to the conclusion that the n-hexane sensitivity of OST4251 was due to the low expression of imp/ostA. The product of imp/ostA was identified as a minor protein associated with the outer membrane, on the basis of its localization, molecular size, immunological interaction and N-terminal amino acid sequence. The imp gene encodes a 784 aa protein. This is in agreement with the reports of Nielsen et al. (1997), who revealed a putative signal sequence and a processing site between amino acids A24 and A25 by primary sequence analyses. The amino acid sequence of OstA showed significant similarities to the sequence of the Imp protein in the phylum Proteobacteria: 99·6 %, Shigella flexneri (GenBank accession no. AE015042-7): 92 %, Salmonella enterica serovar Typhi (AL627265-95): 67 %, Yersinia pestis (AJ414143-26): 47 %, Haemophilus influenzae (AX459898-1, AX459900-1, AX459902-1, AX459904-1 and U32756-6): 46 %, Pasteurella multocida (AE006197-9). The predicted mature form of Imp is 87 kDa and has a very high content of aromatic amino acids (13 %), typical of β-barrel proteins of the outer membrane. Mature E. coli Imp has three Cys residues, one at the N-terminal with the Cys at position 149 and the others at the C-terminal at positions 700 and 701. These conserved Cys motifs are common to the aforementioned amino acid sequences with significant similarities to E. coli Imp. Proteins with similar functions might be dispersed in micro-organisms. Furthermore, outer-membrane usher proteins such as PapC (P07110) and FimD (P30130) also contain N- and C-terminal motifs with conserved Cys residues with more than 500 aa between them. Other outer-membrane proteins such as OmpA, OmpF or TolC do not have conserved Cys residues.

The imp mutants were isolated as maltodextrin assimilators that allowed large maltodextrins to cross the outer membrane of E. coli in the absence of the LamB maltoporin (Sampson et al., 1989). These mutants are hypersensitive to erythromycin, novobiocin, deoxycholic acid and SDS. Therefore, it was proposed that the membrane permeability might be increased in the imp mutants by a functional region of Imp, such as a missing loop in the protein resulting in a large ungated pore. It has been reported that the protein profiles do not change in the envelopes of imp mutants. This is probably because the level of Imp is too low in E. coli to be detected readily. We detected Imp mainly by an immunological method. The protein was also detected by staining with Coomassie brilliant blue R-250 when a comparatively large amount of the outer-membrane protein was electrophoresed on the high-Bis/SDS-polyacrylamide gel. Alternatively, the reported mutants might be missense mutants.

E. coli has various transport systems that mediate the influx of environmental nutrients or the efflux of intracellular toxic compounds (Paulsen et al., 1996). Extracellular n-hexane is partitioned to E. coli membranes by passive diffusion. E. coli maintains resistance to n-hexane mainly by the efflux of the cellular n-hexane with the AcrABTolC pump (Aono et al., 1998; Tsukagoshi & Aono, 2000). This pump system belongs to the resistance/nodulation/cell (RND) family (Ma et al., 1995; Fralick, 1996). This system extrudes various intracellular toxic compounds and contributes to the tolerance of E. coli to such compounds, including organic solvents (Aono et al., 1995, 1998; Ma et al., 1995; Fralick, 1996; Okusu et al., 1996; White et al., 1997; Tsukagoshi & Aono, 2000). Similar efflux pump systems belonging to the RND family are involved in the solvent resistance of P. aeruginosa and P. putida (Kieboom et al., 1998; Li et al., 1998; Ramos et al., 1998; Mosqueda & Ramos, 2000).

The mutants defective in the AcrABTolC pump are hypersensitive to various solvents (Aono et al., 1998; Tsukagoshi & Aono, 2000). This pump is functional in OST4251 (results not shown). The difference in the solvent resistance between strains JA300 and OST4251 is subtle. The n-hexane sensitivity due to the low level of Imp is likely to be independent of the AcrABTolC efflux system. OST4251 is sensitive to hydrophobic compounds such as novobiocin (results not shown). It is probable that the barrier function of the outer membrane against hydrophobic compounds might be decreased in OST4251 due to low imp expression. By a decrease in the barrier function, n-hexane might enter OST4251 more rapidly than it enters JA300. Additionally, the imp/ostA depletion mutant also shows n-hexane sensitivity and filaments become visible in Imp-depleted cultures (data not shown). For example, Imp depletion may cause mislocalization of a secondary effect of an as-yet-undescribed function of Imp. Therefore, defects in the imp gene change the permeability of several compounds through the outer membrane (Sampson et al., 1989) even though the imp level is very low in the membrane. Studies using the imp/ostA depletion mutant are under way to characterize the specific role that Imp plays in envelope biogenesis.

The results described here imply that Imp, rather than inducing permeability directly, catalyses an unidentified reaction that contributes to the outer-membrane permeability.

This work was partially supported by Grants-in-Aid for Scientific Research (B), nos 10450308 and 12460042, from the Ministry of Science, Education and Culture of Japan to R. Aono.