Involvement of Outer Membrane Protein TolC, a Possible Member of the mar-sox Regulon, in Maintenance and Improvement of Organic Solvent Tolerance of Escherichia coli K-12

Abstract

Escherichia coli mutants with improved organic solvent tolerance levels showed high levels of outer membrane protein TolC and inner membrane protein AcrA. The TolC level was regulated positively by MarA, Rob, or SoxS. A possible mar-rob-sox box sequence was found upstream of the tolC gene. These findings suggest that tolC is a member of the mar-sox regulon responsive to stress conditions. When a defective tolC gene was transferred to n-hexane- or cyclohexane-tolerant strains by P1 transduction, the organic solvent tolerance level was lowered dramatically to the decane-tolerant and nonane-sensitive level. The tolerance level was restored by transformation of the transductants with a wild-type tolC gene. Therefore, it is evident that TolC is essential for E. coli to maintain organic solvent tolerance.

There is an increasing interest in culturing microorganisms in two-liquid-phase systems consisting of an aqueous medium and a hydrophobic organic solvent. Various organic solvents used in these systems are toxic to most microorganisms because organic solvent molecules intercalate into biological membranes and disturb the structure of microbial membranes (4, 12, 44). The toxicity of an organic solvent correlates inversely with the parameter log P_ow (18), the common logarithm of the partition coefficient (P_ow) measured for the partition equilibrium established between n-octanol and water phases (24). This empirical log P_ow toxicity rule is based upon the fact that organic solvents with lower log P_ow values bind more abundantly to viable microbial cells (3). Each bacterium has its own intrinsic level of tolerance to organic solvents. Any microorganism grows in the presence of organic solvents whose log P_ow values are equal to or higher than a certain value. The most toxic solvent that an organism tolerates is called the index solvent. The index value is the log P_ow value of the index solvent (4, 18).

We have investigated the mechanism of tolerance of Escherichia coli to organic solvents. Organic solvent tolerance levels differ considerably among species. E. coli JA300 grows in the presence of n-hexane (log P_ow, 3.9) but not cyclohexane (log P_ow, 3.4). From E. coli JA300, we have isolated a series of mutants in which organic solvent tolerance levels were improved stepwise. The most tolerant mutant grows in the presence of p-xylene (log P_ow, 3.1) (1). The cyclohexane-tolerant phenotype of one of these mutants, OST3408, was found to be due to a missense mutation in the marR gene (7). The organic solvent-tolerant mutants are also tolerant of low levels of multiple hydrophobic antibiotics (5). The organic solvent tolerance levels of E. coli strains are improved by overexpression of stress response genes marA, robA, and soxS (7, 33, 34). The overexpression of each of these genes increases multiple hydrophobic antibiotic and heavy metal ion resistance. These facts suggest that the levels of tolerance to organic solvents and hydrophobic antibiotics are controlled genetically by some common mechanism in E. coli.

The three stress response genes encode transcriptional activator proteins for mar-sox regulon genes. This regulon is known to contain more than 10 genes, including sodA (manganese-superoxide dismutase), nfo (DNA-repairing endonuclease IV), zwf (glucose-6-phosphate dehydrogenase), fumC (fumarase C), acnA (aconitase A), fpr (NADPH-ferredoxin oxidoreductase), and micF (antisense RNA to OmpF transcription) (6, 8, 10, 13, 17, 26, 27, 47). Recently, it was suggested that activity of the AcrAB efflux pump is under the control of stress response genes (29, 38). We assume that some of the mar-sox regulon genes contribute to elevation of organic solvent tolerance in E. coli.

In this study, we analyzed the outer membrane proteins of several organic solvent-tolerant mutants. We show here that at least five proteins (a 77-kDa unidentified protein, TolC, LamB, OmpF, and OmpX) are likely to be under the control of stress response genes. OmpF synthesis is known to be regulated by antisense micF RNA. Among these proteins, TolC seems to play the most important role in maintaining and improving the organic solvent tolerance level of E. coli. In addition, the mutants showed high levels of AcrA in the inner membrane, suggesting that the amount of expression of the AcrAB efflux pump system had increased.

MATERIALS AND METHODS

Organisms and plasmids.

The organisms and plasmids used in this study are listed in Table 1. E. coli K-12 JA300 is an n-hexane-tolerant and cyclohexane-sensitive strain (F⁻ thr leuB6 trpC1117 thi rpsL20 hsdS) (21). OST3408 and OST3410 are spontaneous mutants derived independently from JA300 (1, 5). OST3408Tc was an OST3408-based zde-234::Tn10 (tetracycline-resistant [Tc^r]) transductant (7). JA300R and JA300S are zde-234::Tn10 (Tc^r) transductants constructed by P1 transduction from OST3408Tc to JA300. JA300R is cyclohexane tolerant, and JA300S is cyclohexane sensitive. JA300T and OST3408T are JA300- and OST3408-based tolC transductants, respectively. Bacillus subtilis GSY1026 (2) was used as one typical strain of gram-positive bacteria.

TABLE 1.

Bacterial strains and plasmids used

E. coli strain or plasmid	Genotype	Organic solvent tolerance	Reference or source
Strains
JA300	F−thr leuB6 trpC1117 thi rpsL20 hsdS	n-Hexane	21
OST3408	F−thr leuB6 trpC1117 thi rpsL20 hsdS marR08	Cyclohexane	1, 7
OST3410	F−thr leuB6 trpC1117 thi rpsL20 hsdS	Cyclohexane	5
OST3408Tc	F−thr leuB6 trpC1117 thi rpsL20 hsdS marR08 zde-234::Tn10	Cyclohexane	7
JA300R	F−thr leuB6 trpC1117 thi rpsL20 hsdS marR08 zde-234::Tn10	Cyclohexane	This study
JA300S	F−thr leuB6 trpC1117 thi rpsL20 hsdS zde-234::Tn10	n-Hexane	This study
JA300T	F−thr leuB6 trpC1117 thi rpsL20 hsdS tolC	Decane	This study
OST3408T	F−thr leuB6 trpC1117 thi rpsL20 hsdS marR08 tolC	Decane	This study
CH1692	topA57 argA supD74 rpsL Δ(lac-514) tolC::Tn10	NTa	11
MC4100	F−araD139 ΔlacU169 rpsL relA thi	NT	23
Plasmids
pHA105	0.4-kb AflII-PvuII fragment containing the marA gene in pBluescript II		7
pOST42BR	1.9-kb SalI-BamHI fragment containing the robA gene in pBluescript II		33
pHc3R	0.4-kb HincII fragment containing the soxS gene in pBluescript II		34
pUX	pUC19 carrying the tolC gene		M. Wachi
pMX	2.5-kb EcoRI-HindIII containing the tolC gene in pMW119		This study

^aNT, not tested. 

The high-copy-number vector pBluescript II and the low-copy-number vector pMW119 were purchased from Toyobo Biochemical Inc. (Osaka, Japan) and Nippon Gene Co. (Tokyo, Japan), respectively. pHA105 is a pBluescript II recombinant plasmid carrying the marA gene on a 0.4-kb AflII-PvuII fragment under control of the lac promoter (7). pOST42BR is a pBluescript II recombinant plasmid carrying the robA gene on a 1.9-kb SalI-BamHI fragment under control of the lac promoter (33). pHc3R is a pBluescript II recombinant plasmid carrying the soxS gene on a 0.4-kb HincII fragment under control of the lac promoter (34). pUX is a pUC19 recombinant plasmid carrying the E. coli W3110 tolC gene. pMX is a pMW119 recombinant plasmid containing a 2.5-kb EcoRI-HindIII fragment carrying the tolC gene isolated from pUX.

Culture conditions.

The organisms were grown aerobically at 37°C in Luria broth (LB medium; pH 7.0) consisting of 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 (1).

Measurement of the organic solvent tolerance level. Organic solvent tolerance levels of the bacteria were determined by measuring colony formation on LBGMg agar overlaid with a particular organic solvent (4).

P1 transduction.

Generalized transduction was done using phage P1kc by the methods described by Miller (32). OST3408Tc (cyclohexane tolerant and Tc^r) and JA300 were used as the DNA donor and as the recipient strain, respectively. Tc^r transductants were selected on LB agar containing tetracycline (20 μg/ml). From the Tc^r transductants, cyclohexane-tolerant JA300R and cyclohexane-sensitive JA300S were selected.

JA300- and OST3408-based tolC transductants were constructed using CH1692 (tolC) as the DNA donor strain. JA300 and OST3408 cells were infected with the P1kc phage grown on CH1692 and incubated in LB medium containing 0.5 M trisodium citrate for 4 h at 37°C. Then, colicin E1-resistant clones were selected. Among the colicin E1-resistant clones (27 clones from JA300 and 11 clones from OST3408), about half did not grow on LB agar containing 1% sodium dodecyl sulfate (SDS) or on MacConkey agar, respectively. The colicin E1-resistant and detergent-hypersensitive strains JA300T and OST3408T were used as tolC-defective strains.

Preparation of membrane protein.

E. coli was grown in LBGMg medium. Cells were harvested from the culture (optical density at 660 nm, 0.6) by centrifugation (5,000 × g, 10 min, 4°C), suspended in cold 50 mM phosphate buffer (pH 7.2), and broken by sonication. Unbroken cells were removed by the centrifugation. The supernatant was centrifuged at 100,000 × g for 45 min at 4°C. The precipitate was washed with phosphate buffer and incubated in phosphate buffer containing 0.5% sodium dodecyl sarcosinate (sarcosyl) for 30 min at room temperature at a protein concentration of 3 mg/ml (14). The suspension was centrifuged at 100,000 × g for 45 min at 10°C. The supernatant was recovered and used as the inner membrane protein fraction. The precipitate was washed again with the sarcosyl-phosphate buffer. The sarcosyl-insoluble fraction was used as the outer membrane protein fraction.

SDS-polyacrylamide gel electrophoresis.

Samples were dissolved in a solubilization buffer containing 1% (wt/vol) SDS, 2.5% (vol/vol) β-mercaptoethanol, 20% (vol/vol) glycerol, and 16 mM Tris-HCl (pH 6.8) and heated in a boiling water bath for 5 min. The samples were run on SDS-polyacrylamide gels as described by Laemmli (22).

Protein content.

Protein content was measured by the method of Lowry et al. (28).

N-terminal amino acid sequence.

The N-terminal amino acid sequence was determined by Edman degradation using a Protein Sequencer 477A (Applied Biosystems Inc., Foster City, Calif.).

Materials.

Plasmid pUX was obtained from M. Wachi of the Tokyo Institute of Technology. Antisera against TolC and AcrA were kind gifts from Joe A. Fralick of Texas Tech University and H. Nikaido of the University of California, Berkeley, respectively. Colicin E1 was purchased from Sigma-Aldrich Research. Organic solvents used were of the highest quality available and were purchased from Wako Pure Chemical Industries (Osaka, Japan). The log P_ow values of organic solvents and antibiotics were calculated by the addition rule (24) using the log P_ow calculation software ClogP version 1.0.3 (Bio Byte Corporation, Claremont, Calif.).

RESULTS

Composition of outer membrane-associated protein in cyclohexane-tolerant mutants.

Membrane proteins from the cyclohexane-tolerant mutants, OST3408 and OST3410, were fractionated by solubilization in sarcosyl and analyzed by SDS-polyacrylamide gel electrophoresis with Coomassie brilliant blue R-250 staining (Fig. 1). Comparisons of protein composition indicated that several outer membrane proteins (sarcosyl-insoluble membrane proteins) were altered quantitatively in these mutants as compared to the parent strain.

FIG. 1.

SDS-polyacrylamide gel electrophoresis of envelope proteins from the organic solvent-tolerant mutants. The organisms were aerobically grown in LBGMg at 37°C. Envelopes were prepared from cells in the exponential growth phase and extracted with 0.5% sarcosyl. Whole envelopes containing 40 μg of protein (A), the sarcosyl-soluble fraction containing 30 μg of protein (B), and the sarcosyl-insoluble fraction containing 20 μg of protein (C) were electrophoresed on a 0.1% SDS–12% polyacrylamide gel. Protein was stained with Coomassie brilliant blue R-250. Lanes: M, molecular mass markers; 1, JA300; 2, OST3408; 3, OST3410. Arrows indicate the proteins discussed in the text.

It was shown recently that OmpF synthesis is repressed in OST3408 and OST3410 (3). A 37-kDa protein band observed following electrophoresis was found to be composed of OmpC and OmpF, and the amounts were decreased in both mutants (Fig. 1). In addition, the levels of a 46-kDa protein were significantly decreased in OST3408 and it was not detectable in OST3410. In both mutants, the amounts of 77-, 53-, and 18-kDa proteins were increased. Thus, the quantities of at least five outer membrane proteins were altered in the mutants as compared to the parent strain. The alteration was more pronounced in OST3410 than in OST3408. On the other hand, this analysis did not show any distinctive alteration of inner membrane proteins (sarcosyl-soluble membrane proteins) between JA300 and OST3408. In OST3410, the 29-kDa protein disappeared and the 30-kDa protein increased.

Identification of the outer membrane proteins showing quantitative alteration in the mutants.

The 53-kDa protein was recovered from the gel on which the outer membrane proteins of OST3410 had been electrophoresed. The N-terminal amino acid sequence of the 53-kDa protein was examined and, while the first residue could not be determined, the sequence of the next residues was found to be H₂N-XNLMQVYQQA. This was consistent with the sequence, H₂N-ENLMQVYQQA---, reported for the mature form of TolC (36), a 51,468-Da outer membrane protein of E. coli. The location, molecular mass, and N-terminal amino acid sequence of the 53-kDa protein were similar to those of the TolC protein. This 53-kDa protein was not found in JA300T and OST3408T (Fig. 2A). In addition, the 53-kDa proteins of OST3408 and OST3410 were reactive with an antiserum against TolC (Fig. 2B). These genetic and serological analyses supported identification of the 53-kDa protein as TolC protein. Therefore, it was concluded that both of these spontaneous mutants isolated independently from JA300 possessed high amounts of TolC.

FIG. 2.

SDS-polyacrylamide gel electrophoresis of outer membrane protein. The sarcosyl-insoluble envelope protein fraction was prepared as described in the legend to Fig. 1. Samples were electrophoresed on a 0.1% SDS–12.5% polyacrylamide gel. Protein was detected by staining with Coomassie brilliant blue R-250 (A) or by using antiserum against TolC (B). Lanes: M, molecular mass markers; 1, JA300; 2, OST3408; 3, OST3410; 4, JA300(pBluescript II); 5, JA300(pHA105); 6, JA300(pOST42BR); 7, JA300(pHc3R); 8, JA300T; 9, OST3408T; 10, JA300S; 11, JA300R. Arrows indicate the proteins discussed in the text.

The 18-kDa protein of OST3408 was recovered after electrophoresis in a similar manner. Its N-terminal amino acid sequence was found to be H₂N-ATSTVTGGYAQSDAQGQMNK. This sequence is consistent with the sequence H₂N-ATSTVTGGYAQSDMQGQMNK--- predicted for the putative mature form of E. coli OmpX, with the exception of the 14th residue. Mature OmpX is probably a 16,382-Da transmembrane protein (31, 45). The location, molecular mass, and N-terminal amino acid sequence of the 18-kDa protein were similar to those of the OmpX protein.

The 46-kDa protein is thought to be LamB because production of this protein was induced strongly by maltose (results not shown). Mobility of this protein corresponded to that of LamB induced in strain MC4100 by maltose. In JA300, a low level of LamB was produced even in the absence of maltose. The mature form of LamB is a 47,586-Da outer membrane protein (9).

The 77-kDa protein has not been identified.

Increase in levels of TolC in cyclohexane-tolerant marR08 transductants.

Recently, we showed that the elevated level of organic solvent tolerance in the case of OST3408 was due to a missense mutation, R73S, in marR, encoding a repressor protein for the mar operon in the E. coli chromosome. In OST3408 and OST3410, the mar-sox regulon genes, such as sodA and zwf, are expressed at high levels (results not shown). The mutant marR gene (marR08) is cotransducible with zde-234::Tn10 of OST3408Tc to JA300 by P1 transduction (7). Among the zde-234::Tn10 (Tc^r) transductants, JA300R (cyclohexane tolerant) and JA300S (cyclohexane sensitive) were used in this study. Outer membrane proteins of these strains were analyzed by gel electrophoresis (Fig. 2). The level of TolC protein in JA300R was as high as that in OST3408 and higher than that in JA300. In addition, the amounts of the 77-kDa protein and OmpX were increased and the amount of the 46-kDa protein was decreased. Levels of the four proteins were identical in JA300S and JA300. Thus, it was concluded that the increase in TolC was closely related to the marR08 mutation responsible for the cyclohexane-tolerant phenotype in OST3408, although the mutation in OST3410 has not been clearly identified.

Increase in levels of TolC in cyclohexane-tolerant E. coli cells carrying multiple copies of marA, robA, or soxS.

The organic solvent tolerance level of E. coli is improved by overexpression of marA, robA, or soxS. Most E. coli strains acquire cyclohexane tolerance by transformation with any one of these genes (7, 33, 34). We examined TolC production in these transformants (Fig. 2). Each transformant of JA300 cells carrying pHA105, pOST42BR, or pHc3R produced a higher amount of TolC than JA300 carrying or not carrying pBluescript II. In particular, TolC production was high in JA300 carrying pHA105 or pHc3R. The levels of TolC were similar in JA300 and JA300(pBluescript). These results strongly suggest that TolC production is regulated by each of the products of these genes.

Also, the other outer membrane proteins dealt with in this report were altered quantitatively in the JA300-based transformants. The amount of the 77-kDa protein was increased and the amount of LamB was decreased in each of the three transformants. The amount of OmpX was clearly increased in JA300(pHA105), compared with JA300(pBluescript II). However, the increase was not as high in JA300 carrying pOST42BR or pHc3R as in JA300(pHA105).

The quantitative alterations of the outer membrane proteins described above were not specific for JA300. MC4100(pBluescript II) did not produce a detectable amount of LamB when grown in the absence of maltose (results not shown). LamB was induced strongly in MC4100 grown in the presence of 0.2% maltose. Maltose-induced LamB production was lowered remarkably by transformation with pHA105. Amounts of the 77-kDa protein, 53-kDa TolC, and 18-kDa OmpX increased in MC4100(pHA105) compared with MC4100(pBluescript II). Therefore, it was evident that expression of the 77-kDa outer membrane protein, OmpX, and TolC was regulated positively. It is likely that the genes coding for these proteins are members of the mar-sox regulon. Production of OmpF and LamB was regulated negatively by MarA also in MC4100. OmpF production is known to be inhibited at the translation step by micF antisense RNA, which is a member of the mar-sox regulon (8), although it is not clear whether LamB production is regulated at the transcription or translation level.

Diminished levels of organic solvent tolerance due to lack of tolC.

As described above, it was shown that the amount of TolC was always higher in E. coli mutants with improved organic solvent tolerance levels, although the other phenotypic alterations were found commonly. Organic solvent tolerance levels of the tolC-defective transductants, JA300T and OST3408T, were measured on LBGMg agar (Table 2). JA300 is n-hexane (log P_ow, 3.9) tolerant, and OST3408 is cyclohexane (log P_ow, 3.4) tolerant. JA300T and OST3408T were hypersusceptible to organic solvents. Both of these transductants were decane (log P_ow, 6.0) tolerant and nonane (log P_ow, 5.5) sensitive. These tolerance levels are extremely low among various strains of E. coli (33). Both strains showed susceptibility similar to that of B. subtilis, which is highly sensitive to solvents. Thus, it is evident that TolC plays an important and indispensable role in determination and maintenance of organic solvent tolerance in E. coli.

TABLE 2.

Organic solvent tolerance levels of tolC-defective strains

Strain	Plasmid	Growth in the presence ofa:
		Decane (5.98)	Nonane (5.45)	Octane (4.93)	Diphenylether (4.24)	n-Hexane (3.87)	Cyclohexane (3.35)	p-Xylene (3.14)
E. coli
JA300	pBluescript II	+	+	+	+	+	−	−
	pUX	+	+	+	+	+	−	−
	pHA105	+	+	+	+	+	+	−
	pOST42BR	+	+	+	+	+	+	−
	pHc3R	+	+	+	+	+	+	−
OST3408	pBluescript II	+	+	+	+	+	+	−
JA300T	pBluescript II	+	−	−	−	−	−	−
	pMX	+	+	+	+	+	−	−
	pHA105	+	−	−	−	−	−	−
	pOST42BR	+	−	−	−	−	−	−
	pHc3R	+	−	−	−	−	−	−
OST3408T	pBluescript II	+	−	−	−	−	−	−
	pMX	+	+	+	+	+	+	−
	pHA105	+	−	−	−	−	−	−
	pOST42BR	+	−	−	−	−	−	−
	pHc3R	+	−	−	−	−	−	−
B. subtilis GSY1026b		+	−	−	−	−	−	−

^a+, growth; −, no growth. The log P_ow value of the organic solvent is shown in parentheses. 

^bB. subtilis GSY1026 was used as one typical example of gram-positive bacteria. 

When JA300T and OST3408T were transformed with pMX, the organic solvent tolerance levels of these tolC-defective strains were restored completely to n-hexane tolerance (JA300T) or cyclohexane tolerance (OST3408T) by the transformation. Overexpression of tolC alone in JA300 did not improve the level of tolerance. The tolerance levels of the tolC-defective strains were not elevated at all following transformation with pHA105, pOST42BR, or pHc3R, which provided cyclohexane tolerance to JA300. This result is interesting. It is likely that the three gene products (MarA, Rob, and SoxS) improve the level of organic solvent tolerance via a TolC function. TolC might be a key element of some putative tolerance machinery together with other essential components.

Hydrophobic antibiotic tolerance of organic solvent-tolerant derivatives.

OST3408 and OST3410 show tolerance to low levels of multiple hydrophobic antibiotics (5). OST3408T lacks this antibiotic tolerance (Table 3) and is as susceptible to hydrophobic antibiotics as JA300T. In particular, these strains were hypersusceptible to novobiocin (log P_ow, 3.8), which is the most hydrophobic among the antibiotics tested. Therefore, it seems likely that the low-level antibiotic tolerance found in the organic solvent-tolerant mutants results from the presence of tolerance machinery which has TolC as an indispensable element.

TABLE 3.

Antibiotic tolerance of organic solvent-tolerant derivatives

Antibiotic	log Pow	MIC (μg/ml) for:
		JA300	OST3408	JA300T	OST3408T
Novobiocin	3.84	400	800	6.3	6.3
Chloramphenicol	1.14	6.3	12.5	1.6	1.6
Tetracycline	−1.86	1.6	6.3	0.78	0.78
Kanamycin	−7.77	12.5	6.3	6.3	6.3

Increase in levels of AcrA in cyclohexane-tolerant mutants.

It has been shown that the AcrAB efflux pump is responsible for multiple drug resistance (38). We could not detect AcrA or AcrB in the sarcosyl-soluble membrane protein fractions of the cyclohexane-tolerant mutants by dye staining (Fig. 1). Proteins in the fractions were serologically analyzed using an antiserum against AcrA (Fig. 3). Only one protein band, with a size of 44 kDa, was found to cross-react with AcrA. The mature AcrA protein is reported to be an inner membrane-associated lipoprotein, and the molecular mass of the peptide moiety is 39,723 Da (30). The protein was concluded to be AcrA.

FIG. 3.

Detection of AcrA in the cyclohexane-tolerant mutants. Sarcosyl-soluble membrane protein fractions were electrophoresed as described in the legend to Fig. 1. Protein was detected with antiserum against AcrA protein. A block containing AcrA is shown. Lanes: 1, JA300; 2, OST3408; 3, OST3410; 4, JA300(pBluescript II); 5, JA300(pHA105); 6, JA300(pOST42BR); 7, JA300(pHc3R); 8, JA300T; 9, OST3408T; 10, JA300S; 11, JA300R.

It was shown that the amounts of AcrA in OST3408 and OST3410 were increased compared with that in JA300. The amount was higher in OST3410 than in OST3408. In JA300, AcrA production was regulated positively by each of the transcriptional activator proteins, MarA, Rob, and SoxS. The level of AcrA was high in JA300R but not in JA300S, indicating that the marR08 mutation was responsible for high-level AcrA production in OST3408. OST3408T showed a high level of AcrA and JA300T showed the same level of AcrA as JA300, indicating that productions of AcrA and TolC were independent of each other.

DISCUSSION

The cyclohexane-tolerant phenotype of OST3408 is due to the marR08 mutation (7). Although OST3410 has not been analyzed sufficiently, a preliminary analysis by P1 transduction showed that the site of the mutation in OST3410 is as near zde-234::Tn10 as marR08 (results not shown). In fact, these two mutants show very similar phenotypes (5, 33), although several phenotypic alterations are more pronounced in OST3410 than in OST3408 (3, 5, 33). Probably, marA expression is derepressed and mar-sox regulon genes are transcribed actively in OST3408 and OST3410.

We have reported that levels of OmpF are decreased in organic solvent-tolerant mutants isolated from JA300 (3). We show here the identification of the other outer membrane proteins showing quantitative alteration in the mutants (Fig. 1). Not only OmpF but also LamB was produced in decreased amounts in the mutants. Increased amounts of the 77-kDa protein, TolC, and OmpX were found. We are interested in TolC synthesis especially, among the five proteins, because the amount of TolC expressed is likely the most important in determining the cyclohexane-tolerant phenotype.

The amount of TolC was elevated in JA300 carrying a high-copy-number plasmid coding for MarA, Rob, or SoxS (Fig. 2). These proteins are transcriptional activators for mar-sox regulon genes (6, 10, 13, 16, 20, 25, 42, 47). The results shown in this report demonstrate that TolC synthesis is regulated positively by these proteins. A consensus sequence of the sox-mar-rob box bound by the activator proteins has been proposed as AYNGCAYNRRNNRNYANNNNNWNNNNNNYW (R, A or G; Y, T or C; W, A or T) (13) or ANNGCAYNNNNNNNCWA (Y, C or T; W, A or T) (25). It is interesting that we can find the consensus box sequence ATGGCACGTAACGCCAACCTTTTGCGGTCA at a location upstream (−96 to −73) of the structural tolC gene. This sequence found upstream corresponds to both consensus box sequences proposed independently. The presence of this consensus sequence suggests that tolC expression might be regulated directly by the transcriptional activator proteins, although binding of the activator proteins to the tolC sequence must be proven experimentally. It is likely that tolC is a novel member of the mar-sox regulon.

OmpX synthesis is likely to be regulated positively by MarA, Rob, or SoxS (Fig. 2). In particular, the synthesis was high in JA300 derivatives in which MarA production was elevated rather than that of Rob and SoxS. The regulatory mechanism controlling OmpX synthesis is not clear. Genes homologous to ompX are present widely among members of the family Enterobacteriaceae. However, the functional role of OmpX in E. coli is still unclear (31, 45).

TolC is absolutely required for maintenance of the organic solvent tolerance levels displayed by most strains of E. coli (Fig. 3 and Table 2). Generally, gram-negative bacteria are more tolerant of organic solvents than gram-positive bacteria (18). However, JA300T and OST3408T are as susceptible to organic solvents as B. subtilis GSY1026. The tolC-defective strains are susceptible even to n-nonane, which scarcely binds to JA300 cells (3).

Enhancement of TolC synthesis seems essential to improve organic solvent tolerance levels of E. coli. However, an increase in TolC level alone was not effective in improving the cyclohexane tolerance level of JA300(pUX). It is known that TolC forms a channel for various compounds (35). TolC likely serves as an essential element of some machinery responsible for tolerance to various compounds, including organic solvents and hydrophobic antibiotics (Table 3). This putative tolerance machinery is invalid for hydrophilic inhibitors, such as kanamycin. With respect to the other four proteins, it seems unlikely that the quantitative alterations observed are related to improvement of organic solvent and antibiotic tolerance levels since OST3408T showed high levels of the 77-kDa protein and OmpX and low levels of LamB and OmpF as compared with JA300T. MC4100 cells grown in the presence and absence of maltose showed no difference in organic solvent tolerance levels, although the LamB levels differed greatly (results not shown). However, we cannot rule out the possibility that these proteins have some effect on tolerance or function synergistically with TolC, in particular the 77-kDa protein or OmpX.

TolC is proposed to constitute an efflux pump together with the AcrAB complex (15). The AcrAB complex is a major ΔμH⁺-dependent pump responsible for resistance to many hydrophobic agents (39). An activity of the efflux pump is suggested to be regulated by MarA (38). We have found that the amount of AcrA expressed from the chromosomal acrAB operon was elevated in OST3408, OST3410, OST3408T, JA300R, and JA300, each carrying marA, robA, and soxS (Fig. 3). We cannot detect AcrB by SDS-polyacrylamide gel electrophoresis and dye staining (Fig. 1), probably due to its high molecular mass and complex folding in the inner membrane. We suppose that AcrB production was elevated together with AcrA since the two genes consist of a common transcriptional unit. The cyclohexane-tolerant derivatives showed high levels of both TolC and AcrA. OST3408T possessing a high level of AcrA but not TolC was hypersusceptible to organic solvents, while this organism carrying pMX was cyclohexane tolerant (Table 2).

Therefore, we conclude that TolC, which is expressed in elevated amounts, contributes to improvement of organic solvent tolerance levels by serving as a component of tolerance machinery, such as the AcrAB efflux pump. After the first version of this report was submitted, it was shown that the acrAB locus was linked to n-hexane tolerance (46). Although little is known about bacterial organic solvent tolerance mechanisms (43), the toluene tolerance of Pseudomonas putida seems to be energy dependent (19, 41). Also, the organic solvent tolerance of E. coli is energy dependent (37). An efflux pump similar to the AcrAB-TolC complex (for example, the Mex pump system [40]) might be responsible for the toluene tolerance in strains of P. putida.